These are the second revision of a few patches to remove the current fragmentation code and replace it with a new version that supports more packet types and more fragments per packet, thus allowing bigger payloads.
The patches are based on current master and are meant for the next big bump in compat number. Revision 2 is changed according to Mareks review.
Martin Hundebøll (3): batman-adv: Remove old fragmentation code batman-adv: Receive fragmented packets and merge batman-adv: Fragment and send skbs larger than mtu
Makefile.kbuild | 2 +- distributed-arp-table.c | 11 +- fragmentation.c | 479 ++++++++++++++++++++++++++++++++++++++++++++++++ fragmentation.h | 50 +++++ hard-interface.c | 1 - main.c | 8 +- main.h | 9 + originator.c | 19 +- packet.h | 41 +++-- routing.c | 146 ++++++--------- routing.h | 4 +- send.c | 194 +++++++++++++++++++- send.h | 36 ++++ soft-interface.c | 15 +- types.h | 63 +++++-- unicast.c | 478 ----------------------------------------------- unicast.h | 92 ---------- 17 files changed, 930 insertions(+), 718 deletions(-) create mode 100644 fragmentation.c create mode 100644 fragmentation.h delete mode 100644 unicast.c delete mode 100644 unicast.h
Remove the existing fragmentation code before adding the new version and delete unicast.{h,c}.
batadv_unicast_send_skb() is moved to send.c and renamed to batadv_send_skb_unicast().
fragmentation entry in sysfs (bat_priv->fragmentation) is kept for use in the new fragmentation code.
BATADV_UNICAST_FRAG packet type is renamed to BATADV_FRAG for use in the new fragmentation code.
Signed-off-by: Martin Hundebøll martin@hundeboll.net ---
v2: unmangle function names with 4addr_unicast correctly indent kernel doc
Makefile.kbuild | 1 - distributed-arp-table.c | 11 +- hard-interface.c | 1 - main.c | 4 - originator.c | 9 - packet.h | 16 -- routing.c | 86 +-------- routing.h | 2 - send.c | 173 ++++++++++++++++++ send.h | 36 ++++ soft-interface.c | 4 +- types.h | 19 -- unicast.c | 478 ------------------------------------------------ unicast.h | 92 ---------- 14 files changed, 216 insertions(+), 716 deletions(-) delete mode 100644 unicast.c delete mode 100644 unicast.h
diff --git a/Makefile.kbuild b/Makefile.kbuild index 8ddbfe6..f9b465b 100644 --- a/Makefile.kbuild +++ b/Makefile.kbuild @@ -37,4 +37,3 @@ batman-adv-y += send.o batman-adv-y += soft-interface.o batman-adv-y += sysfs.o batman-adv-y += translation-table.o -batman-adv-y += unicast.o diff --git a/distributed-arp-table.c b/distributed-arp-table.c index f07ec32..99da412 100644 --- a/distributed-arp-table.c +++ b/distributed-arp-table.c @@ -29,7 +29,6 @@ #include "send.h" #include "types.h" #include "translation-table.h" -#include "unicast.h"
static void batadv_dat_purge(struct work_struct *work);
@@ -592,9 +591,9 @@ static bool batadv_dat_send_data(struct batadv_priv *bat_priv, goto free_orig;
tmp_skb = pskb_copy(skb, GFP_ATOMIC); - if (!batadv_unicast_4addr_prepare_skb(bat_priv, tmp_skb, - cand[i].orig_node, - packet_subtype)) { + if (!batadv_send_skb_prepare_unicast_4addr(bat_priv, tmp_skb, + cand[i].orig_node, + packet_subtype)) { kfree_skb(tmp_skb); goto free_neigh; } @@ -990,10 +989,10 @@ bool batadv_dat_snoop_incoming_arp_request(struct batadv_priv *bat_priv, * that a node not using the 4addr packet format doesn't support it. */ if (hdr_size == sizeof(struct batadv_unicast_4addr_packet)) - err = batadv_unicast_4addr_send_skb(bat_priv, skb_new, + err = batadv_send_skb_unicast_4addr(bat_priv, skb_new, BATADV_P_DAT_CACHE_REPLY); else - err = batadv_unicast_send_skb(bat_priv, skb_new); + err = batadv_send_skb_unicast(bat_priv, skb_new);
if (!err) { batadv_inc_counter(bat_priv, BATADV_CNT_DAT_CACHED_REPLY_TX); diff --git a/hard-interface.c b/hard-interface.c index 9746600..c343aa8 100644 --- a/hard-interface.c +++ b/hard-interface.c @@ -444,7 +444,6 @@ int batadv_hardif_enable_interface(struct batadv_hard_iface *hard_iface, hard_iface->batman_adv_ptype.dev = hard_iface->net_dev; dev_add_pack(&hard_iface->batman_adv_ptype);
- atomic_set(&hard_iface->frag_seqno, 1); batadv_info(hard_iface->soft_iface, "Adding interface: %s\n", hard_iface->net_dev->name);
diff --git a/main.c b/main.c index af89c09..3b5f78f 100644 --- a/main.c +++ b/main.c @@ -32,7 +32,6 @@ #include "gateway_client.h" #include "bridge_loop_avoidance.h" #include "distributed-arp-table.h" -#include "unicast.h" #include "gateway_common.h" #include "hash.h" #include "bat_algo.h" @@ -342,7 +341,6 @@ static void batadv_recv_handler_init(void) /* compile time checks for struct member offsets */ BUILD_BUG_ON(offsetof(struct batadv_unicast_4addr_packet, src) != 10); BUILD_BUG_ON(offsetof(struct batadv_unicast_packet, dest) != 4); - BUILD_BUG_ON(offsetof(struct batadv_unicast_frag_packet, dest) != 4); BUILD_BUG_ON(offsetof(struct batadv_unicast_tvlv_packet, dst) != 4); BUILD_BUG_ON(offsetof(struct batadv_icmp_packet, dst) != 4); BUILD_BUG_ON(offsetof(struct batadv_icmp_packet_rr, dst) != 4); @@ -355,8 +353,6 @@ static void batadv_recv_handler_init(void) batadv_rx_handler[BATADV_UNICAST_4ADDR] = batadv_recv_unicast_packet; /* unicast packet */ batadv_rx_handler[BATADV_UNICAST] = batadv_recv_unicast_packet; - /* fragmented unicast packet */ - batadv_rx_handler[BATADV_UNICAST_FRAG] = batadv_recv_ucast_frag_packet; /* unicast tvlv packet */ batadv_rx_handler[BATADV_UNICAST_TVLV] = batadv_recv_unicast_tvlv; /* batman icmp packet */ diff --git a/originator.c b/originator.c index 5d53d2f..898b0ce 100644 --- a/originator.c +++ b/originator.c @@ -25,7 +25,6 @@ #include "routing.h" #include "gateway_client.h" #include "hard-interface.h" -#include "unicast.h" #include "soft-interface.h" #include "bridge_loop_avoidance.h" #include "network-coding.h" @@ -146,7 +145,6 @@ static void batadv_orig_node_free_rcu(struct rcu_head *rcu) /* Free nc_nodes */ batadv_nc_purge_orig(orig_node->bat_priv, orig_node, NULL);
- batadv_frag_list_free(&orig_node->frag_list); batadv_tt_global_del_orig(orig_node->bat_priv, orig_node, "originator timed out");
@@ -269,9 +267,6 @@ struct batadv_orig_node *batadv_get_orig_node(struct batadv_priv *bat_priv, 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;
@@ -393,10 +388,6 @@ static void _batadv_purge_orig(struct batadv_priv *bat_priv) batadv_orig_node_free_ref(orig_node); continue; } - - if (batadv_has_timed_out(orig_node->last_frag_packet, - BATADV_FRAG_TIMEOUT)) - batadv_frag_list_free(&orig_node->frag_list); } spin_unlock_bh(list_lock); } diff --git a/packet.h b/packet.h index 7d45890..96b8e57 100644 --- a/packet.h +++ b/packet.h @@ -91,12 +91,6 @@ enum batadv_icmp_packettype { BATADV_PARAMETER_PROBLEM = 12, };
-/* fragmentation defines */ -enum batadv_unicast_frag_flags { - BATADV_UNI_FRAG_HEAD = BIT(0), - BATADV_UNI_FRAG_LARGETAIL = BIT(1), -}; - /* tt data subtypes */ #define BATADV_TT_DATA_TYPE_MASK 0x0F
@@ -255,16 +249,6 @@ struct batadv_unicast_4addr_packet { */ };
-struct batadv_unicast_frag_packet { - struct batadv_header header; - uint8_t ttvn; /* destination translation table version number */ - uint8_t dest[ETH_ALEN]; - uint8_t flags; - uint8_t align; - uint8_t orig[ETH_ALEN]; - __be16 seqno; -} __packed; - struct batadv_bcast_packet { struct batadv_header header; uint8_t reserved; diff --git a/routing.c b/routing.c index 7c372f1..a099d32 100644 --- a/routing.c +++ b/routing.c @@ -25,7 +25,6 @@ #include "icmp_socket.h" #include "translation-table.h" #include "originator.h" -#include "unicast.h" #include "bridge_loop_avoidance.h" #include "distributed-arp-table.h" #include "network-coding.h" @@ -653,11 +652,9 @@ static int batadv_route_unicast_packet(struct sk_buff *skb, { struct batadv_priv *bat_priv = netdev_priv(recv_if->soft_iface); struct batadv_orig_node *orig_node = NULL; - struct batadv_neigh_node *neigh_node = NULL; struct batadv_unicast_packet *unicast_packet; struct ethhdr *ethhdr = eth_hdr(skb); int res, ret = NET_RX_DROP; - struct sk_buff *new_skb;
unicast_packet = (struct batadv_unicast_packet *)skb->data;
@@ -674,46 +671,12 @@ static int batadv_route_unicast_packet(struct sk_buff *skb, if (!orig_node) goto out;
- /* find_router() increases neigh_nodes refcount if found. */ - neigh_node = batadv_find_router(bat_priv, orig_node, recv_if); - - if (!neigh_node) - goto out; - /* create a copy of the skb, if needed, to modify it. */ if (skb_cow(skb, ETH_HLEN) < 0) goto out;
- unicast_packet = (struct batadv_unicast_packet *)skb->data; - - if (unicast_packet->header.packet_type == BATADV_UNICAST && - atomic_read(&bat_priv->fragmentation) && - skb->len > neigh_node->if_incoming->net_dev->mtu) { - ret = batadv_frag_send_skb(skb, bat_priv, - neigh_node->if_incoming, - neigh_node->addr); - goto out; - } - - if (unicast_packet->header.packet_type == BATADV_UNICAST_FRAG && - batadv_frag_can_reassemble(skb, - neigh_node->if_incoming->net_dev->mtu)) { - ret = batadv_frag_reassemble_skb(skb, bat_priv, &new_skb); - - if (ret == NET_RX_DROP) - goto out; - - /* packet was buffered for late merge */ - if (!new_skb) { - ret = NET_RX_SUCCESS; - goto out; - } - - skb = new_skb; - unicast_packet = (struct batadv_unicast_packet *)skb->data; - } - /* decrement ttl */ + unicast_packet = (struct batadv_unicast_packet *)skb->data; unicast_packet->header.ttl--;
res = batadv_send_skb_to_orig(skb, orig_node, recv_if); @@ -732,8 +695,6 @@ static int batadv_route_unicast_packet(struct sk_buff *skb, }
out: - if (neigh_node) - batadv_neigh_node_free_ref(neigh_node); if (orig_node) batadv_orig_node_free_ref(orig_node); return ret; @@ -987,51 +948,6 @@ rx_success: return batadv_route_unicast_packet(skb, recv_if); }
-int batadv_recv_ucast_frag_packet(struct sk_buff *skb, - struct batadv_hard_iface *recv_if) -{ - struct batadv_priv *bat_priv = netdev_priv(recv_if->soft_iface); - struct batadv_unicast_frag_packet *unicast_packet; - int hdr_size = sizeof(*unicast_packet); - struct sk_buff *new_skb = NULL; - int ret; - - if (batadv_check_unicast_packet(bat_priv, skb, hdr_size) < 0) - return NET_RX_DROP; - - if (!batadv_check_unicast_ttvn(bat_priv, skb, hdr_size)) - return NET_RX_DROP; - - unicast_packet = (struct batadv_unicast_frag_packet *)skb->data; - - /* packet for me */ - if (batadv_is_my_mac(bat_priv, unicast_packet->dest)) { - ret = batadv_frag_reassemble_skb(skb, bat_priv, &new_skb); - - if (ret == NET_RX_DROP) - return NET_RX_DROP; - - /* packet was buffered for late merge */ - if (!new_skb) - return NET_RX_SUCCESS; - - if (batadv_dat_snoop_incoming_arp_request(bat_priv, new_skb, - hdr_size)) - goto rx_success; - if (batadv_dat_snoop_incoming_arp_reply(bat_priv, new_skb, - hdr_size)) - goto rx_success; - - batadv_interface_rx(recv_if->soft_iface, new_skb, recv_if, - sizeof(struct batadv_unicast_packet), NULL); - -rx_success: - return NET_RX_SUCCESS; - } - - return batadv_route_unicast_packet(skb, recv_if); -} - /** * batadv_recv_unicast_tvlv - receive and process unicast tvlv packets * @skb: unicast tvlv packet to process diff --git a/routing.h b/routing.h index ea15fa6..efab583 100644 --- a/routing.h +++ b/routing.h @@ -30,8 +30,6 @@ int batadv_recv_icmp_packet(struct sk_buff *skb, struct batadv_hard_iface *recv_if); int batadv_recv_unicast_packet(struct sk_buff *skb, struct batadv_hard_iface *recv_if); -int batadv_recv_ucast_frag_packet(struct sk_buff *skb, - struct batadv_hard_iface *recv_if); int batadv_recv_bcast_packet(struct sk_buff *skb, struct batadv_hard_iface *recv_if); int batadv_recv_tt_query(struct sk_buff *skb, diff --git a/send.c b/send.c index 23c7f79..e6852a1 100644 --- a/send.c +++ b/send.c @@ -25,6 +25,7 @@ #include "soft-interface.h" #include "hard-interface.h" #include "gateway_common.h" +#include "gateway_client.h" #include "originator.h" #include "network-coding.h"
@@ -128,6 +129,178 @@ int batadv_send_skb_to_orig(struct sk_buff *skb, return ret; }
+/** + * batadv_send_skb_push_fill_unicast - extends the buffer and initializes + * the common fields for unicast packets + * @skb: packet + * @hdr_size: amount of bytes to push at the beginning of the skb + * @orig_node: the destination node + * + * Returns false if the buffer extension was not possible or true otherwise + */ +static bool +batadv_send_skb_push_fill_unicast(struct sk_buff *skb, int hdr_size, + struct batadv_orig_node *orig_node) +{ + struct batadv_unicast_packet *unicast_packet; + uint8_t ttvn = (uint8_t)atomic_read(&orig_node->last_ttvn); + + if (batadv_skb_head_push(skb, hdr_size) < 0) + return false; + + unicast_packet = (struct batadv_unicast_packet *)skb->data; + unicast_packet->header.version = BATADV_COMPAT_VERSION; + /* batman packet type: unicast */ + unicast_packet->header.packet_type = BATADV_UNICAST; + /* set unicast ttl */ + unicast_packet->header.ttl = BATADV_TTL; + /* copy the destination for faster routing */ + memcpy(unicast_packet->dest, orig_node->orig, ETH_ALEN); + /* set the destination tt version number */ + unicast_packet->ttvn = ttvn; + + return true; +} + +/** + * batadv_send_skb_prepare_unicast - encapsulate an skb with a unicast header + * @skb: the skb containing the payload to encapsulate + * @orig_node: the destination node + * + * Returns false if the payload could not be encapsulated or true otherwise + */ +static bool batadv_send_skb_prepare_unicast(struct sk_buff *skb, + struct batadv_orig_node *orig_node) +{ + size_t uni_size = sizeof(struct batadv_unicast_packet); + return batadv_send_skb_push_fill_unicast(skb, uni_size, orig_node); +} + +/** + * batadv_send_skb_prepare_unicast_4addr - encapsulate an skb with a + * unicast4addr header + * @bat_priv: the bat priv with all the soft interface information + * @skb: the skb containing the payload to encapsulate + * @orig_node: the destination node + * @packet_subtype: the batman 4addr packet subtype to use + * + * Returns false if the payload could not be encapsulated or true otherwise + */ +bool batadv_send_skb_prepare_unicast_4addr(struct batadv_priv *bat_priv, + struct sk_buff *skb, + struct batadv_orig_node *orig, + int packet_subtype) +{ + struct batadv_hard_iface *primary_if; + struct batadv_unicast_4addr_packet *uc_4addr_packet; + bool ret = false; + + primary_if = batadv_primary_if_get_selected(bat_priv); + if (!primary_if) + goto out; + + /* pull the header space and fill the unicast_packet substructure. + * We can do that because the first member of the uc_4addr_packet + * is of type struct unicast_packet + */ + if (!batadv_send_skb_push_fill_unicast(skb, sizeof(*uc_4addr_packet), + orig)) + goto out; + + uc_4addr_packet = (struct batadv_unicast_4addr_packet *)skb->data; + uc_4addr_packet->u.header.packet_type = BATADV_UNICAST_4ADDR; + memcpy(uc_4addr_packet->src, primary_if->net_dev->dev_addr, ETH_ALEN); + uc_4addr_packet->subtype = packet_subtype; + uc_4addr_packet->reserved = 0; + + ret = true; +out: + if (primary_if) + batadv_hardif_free_ref(primary_if); + return ret; +} + +/** + * batadv_send_generic_unicast_skb - send an skb as unicast + * @bat_priv: the bat priv with all the soft interface information + * @skb: payload to send + * @packet_type: the batman unicast packet type to use + * @packet_subtype: the batman packet subtype. It is ignored if packet_type is + * not BATADV_UNICAT_4ADDR + * + * Returns 1 in case of error or 0 otherwise + */ +int batadv_send_skb_generic_unicast(struct batadv_priv *bat_priv, + struct sk_buff *skb, int packet_type, + int packet_subtype) +{ + struct ethhdr *ethhdr = (struct ethhdr *)skb->data; + struct batadv_unicast_packet *unicast_packet; + struct batadv_orig_node *orig_node; + struct batadv_neigh_node *neigh_node; + int ret = NET_RX_DROP; + + /* get routing information */ + if (is_multicast_ether_addr(ethhdr->h_dest)) { + orig_node = batadv_gw_get_selected_orig(bat_priv); + if (orig_node) + goto find_router; + } + + /* check for tt host - increases orig_node refcount. + * returns NULL in case of AP isolation + */ + orig_node = batadv_transtable_search(bat_priv, ethhdr->h_source, + ethhdr->h_dest); + +find_router: + /* find_router(): + * - if orig_node is NULL it returns NULL + * - increases neigh_nodes refcount if found. + */ + neigh_node = batadv_find_router(bat_priv, orig_node, NULL); + + if (!neigh_node) + goto out; + + switch (packet_type) { + case BATADV_UNICAST: + batadv_send_skb_prepare_unicast(skb, orig_node); + break; + case BATADV_UNICAST_4ADDR: + batadv_send_skb_prepare_unicast_4addr(bat_priv, skb, orig_node, + packet_subtype); + break; + default: + /* this function supports UNICAST and UNICAST_4ADDR only. It + * should never be invoked with any other packet type + */ + goto out; + } + + unicast_packet = (struct batadv_unicast_packet *)skb->data; + + /* inform the destination node that we are still missing a correct route + * for this client. The destination will receive this packet and will + * try to reroute it because the ttvn contained in the header is less + * than the current one + */ + if (batadv_tt_global_client_is_roaming(bat_priv, ethhdr->h_dest)) + unicast_packet->ttvn = unicast_packet->ttvn - 1; + + if (batadv_send_skb_to_orig(skb, orig_node, NULL) != NET_XMIT_DROP) + ret = 0; + +out: + if (neigh_node) + batadv_neigh_node_free_ref(neigh_node); + if (orig_node) + batadv_orig_node_free_ref(orig_node); + if (ret == NET_RX_DROP) + kfree_skb(skb); + return ret; +} + void batadv_schedule_bat_ogm(struct batadv_hard_iface *hard_iface) { struct batadv_priv *bat_priv = netdev_priv(hard_iface->soft_iface); diff --git a/send.h b/send.h index e7b1788..bfe65be 100644 --- a/send.h +++ b/send.h @@ -34,5 +34,41 @@ void batadv_send_outstanding_bat_ogm_packet(struct work_struct *work); void batadv_purge_outstanding_packets(struct batadv_priv *bat_priv, const struct batadv_hard_iface *hard_iface); +bool batadv_send_skb_prepare_unicast_4addr(struct batadv_priv *bat_priv, + struct sk_buff *skb, + struct batadv_orig_node *orig_node, + int packet_subtype); +int batadv_send_skb_generic_unicast(struct batadv_priv *bat_priv, + struct sk_buff *skb, int packet_type, + int packet_subtype); + + +/** + * batadv_send_unicast_skb - send the skb encapsulated in a unicast packet + * @bat_priv: the bat priv with all the soft interface information + * @skb: the payload to send + */ +static inline int batadv_send_skb_unicast(struct batadv_priv *bat_priv, + struct sk_buff *skb) +{ + return batadv_send_skb_generic_unicast(bat_priv, skb, BATADV_UNICAST, + 0); +} + +/** + * batadv_send_4addr_unicast_skb - send the skb encapsulated in a unicast4addr + * packet + * @bat_priv: the bat priv with all the soft interface information + * @skb: the payload to send + * @packet_subtype: the batman 4addr packet subtype to use + */ +static inline int batadv_send_skb_unicast_4addr(struct batadv_priv *bat_priv, + struct sk_buff *skb, + int packet_subtype) +{ + return batadv_send_skb_generic_unicast(bat_priv, skb, + BATADV_UNICAST_4ADDR, + packet_subtype); +}
#endif /* _NET_BATMAN_ADV_SEND_H_ */ diff --git a/soft-interface.c b/soft-interface.c index a6ab9e4..2f3810d 100644 --- a/soft-interface.c +++ b/soft-interface.c @@ -34,8 +34,6 @@ #include <linux/ethtool.h> #include <linux/etherdevice.h> #include <linux/if_vlan.h> -#include <linux/if_ether.h> -#include "unicast.h" #include "bridge_loop_avoidance.h" #include "network-coding.h"
@@ -276,7 +274,7 @@ static int batadv_interface_tx(struct sk_buff *skb,
batadv_dat_snoop_outgoing_arp_reply(bat_priv, skb);
- ret = batadv_unicast_send_skb(bat_priv, skb); + ret = batadv_send_skb_unicast(bat_priv, skb); if (ret != 0) goto dropped_freed; } diff --git a/types.h b/types.h index c84f5cc..6cde9c9 100644 --- a/types.h +++ b/types.h @@ -60,7 +60,6 @@ struct batadv_hard_iface_bat_iv { * @if_num: identificator of the interface * @if_status: status of the interface for batman-adv * @net_dev: pointer to the net_device - * @frag_seqno: last fragment sequence number sent by this interface * @num_bcasts: number of payload re-broadcasts on this interface (ARQ) * @hardif_obj: kobject of the per interface sysfs "mesh" directory * @refcount: number of contexts the object is used @@ -76,7 +75,6 @@ struct batadv_hard_iface { int16_t if_num; char if_status; struct net_device *net_dev; - atomic_t frag_seqno; uint8_t num_bcasts; struct kobject *hardif_obj; atomic_t refcount; @@ -116,9 +114,6 @@ struct batadv_hard_iface { * last_bcast_seqno) * @last_bcast_seqno: last broadcast sequence number received by this host * @neigh_list: list of potential next hop neighbor towards this orig node - * @frag_list: fragmentation buffer list for fragment re-assembly - * @last_frag_packet: time when last fragmented packet from this node was - * received * @neigh_list_lock: lock protecting neigh_list, router and bonding_list * @hash_entry: hlist node for batadv_priv::orig_hash * @bat_priv: pointer to soft_iface this orig node belongs to @@ -159,8 +154,6 @@ struct batadv_orig_node { DECLARE_BITMAP(bcast_bits, BATADV_TQ_LOCAL_WINDOW_SIZE); uint32_t last_bcast_seqno; struct hlist_head neigh_list; - struct list_head frag_list; - unsigned long last_frag_packet; /* neigh_list_lock protects: neigh_list, router & bonding_list */ spinlock_t neigh_list_lock; struct hlist_node hash_entry; @@ -874,18 +867,6 @@ struct batadv_forw_packet { };
/** - * struct batadv_frag_packet_list_entry - storage for fragment packet - * @list: list node for orig_node::frag_list - * @seqno: sequence number of the fragment - * @skb: fragment's skb buffer - */ -struct batadv_frag_packet_list_entry { - struct list_head list; - uint16_t seqno; - struct sk_buff *skb; -}; - -/** * struct batadv_algo_ops - mesh algorithm callbacks * @list: list node for the batadv_algo_list * @name: name of the algorithm diff --git a/unicast.c b/unicast.c deleted file mode 100644 index dc8b5d4..0000000 --- a/unicast.c +++ /dev/null @@ -1,478 +0,0 @@ -/* Copyright (C) 2010-2013 B.A.T.M.A.N. contributors: - * - * Andreas Langer - * - * 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 - */ - -#include "main.h" -#include "unicast.h" -#include "send.h" -#include "soft-interface.h" -#include "gateway_client.h" -#include "originator.h" -#include "hash.h" -#include "translation-table.h" -#include "routing.h" -#include "hard-interface.h" - - -static struct sk_buff * -batadv_frag_merge_packet(struct list_head *head, - struct batadv_frag_packet_list_entry *tfp, - struct sk_buff *skb) -{ - struct batadv_unicast_frag_packet *up; - struct sk_buff *tmp_skb; - struct batadv_unicast_packet *unicast_packet; - int hdr_len = sizeof(*unicast_packet); - int uni_diff = sizeof(*up) - hdr_len; - uint8_t *packet_pos; - - up = (struct batadv_unicast_frag_packet *)skb->data; - /* set skb to the first part and tmp_skb to the second part */ - if (up->flags & BATADV_UNI_FRAG_HEAD) { - tmp_skb = tfp->skb; - } else { - tmp_skb = skb; - skb = tfp->skb; - } - - if (skb_linearize(skb) < 0 || skb_linearize(tmp_skb) < 0) - goto err; - - skb_pull(tmp_skb, sizeof(*up)); - if (pskb_expand_head(skb, 0, tmp_skb->len, GFP_ATOMIC) < 0) - goto err; - - /* move free entry to end */ - tfp->skb = NULL; - tfp->seqno = 0; - list_move_tail(&tfp->list, head); - - memcpy(skb_put(skb, tmp_skb->len), tmp_skb->data, tmp_skb->len); - kfree_skb(tmp_skb); - - memmove(skb->data + uni_diff, skb->data, hdr_len); - packet_pos = skb_pull(skb, uni_diff); - unicast_packet = (struct batadv_unicast_packet *)packet_pos; - unicast_packet->header.packet_type = BATADV_UNICAST; - - return skb; - -err: - /* free buffered skb, skb will be freed later */ - kfree_skb(tfp->skb); - return NULL; -} - -static void batadv_frag_create_entry(struct list_head *head, - struct sk_buff *skb) -{ - struct batadv_frag_packet_list_entry *tfp; - struct batadv_unicast_frag_packet *up; - - up = (struct batadv_unicast_frag_packet *)skb->data; - - /* free and oldest packets stand at the end */ - tfp = list_entry((head)->prev, typeof(*tfp), list); - kfree_skb(tfp->skb); - - tfp->seqno = ntohs(up->seqno); - tfp->skb = skb; - list_move(&tfp->list, head); - return; -} - -static int batadv_frag_create_buffer(struct list_head *head) -{ - int i; - struct batadv_frag_packet_list_entry *tfp; - - for (i = 0; i < BATADV_FRAG_BUFFER_SIZE; i++) { - tfp = kmalloc(sizeof(*tfp), GFP_ATOMIC); - if (!tfp) { - batadv_frag_list_free(head); - return -ENOMEM; - } - tfp->skb = NULL; - tfp->seqno = 0; - INIT_LIST_HEAD(&tfp->list); - list_add(&tfp->list, head); - } - - return 0; -} - -static struct batadv_frag_packet_list_entry * -batadv_frag_search_packet(struct list_head *head, - const struct batadv_unicast_frag_packet *up) -{ - struct batadv_frag_packet_list_entry *tfp; - struct batadv_unicast_frag_packet *tmp_up = NULL; - bool is_head_tmp, is_head; - uint16_t search_seqno; - - if (up->flags & BATADV_UNI_FRAG_HEAD) - search_seqno = ntohs(up->seqno)+1; - else - search_seqno = ntohs(up->seqno)-1; - - is_head = up->flags & BATADV_UNI_FRAG_HEAD; - - list_for_each_entry(tfp, head, list) { - if (!tfp->skb) - continue; - - if (tfp->seqno == ntohs(up->seqno)) - goto mov_tail; - - tmp_up = (struct batadv_unicast_frag_packet *)tfp->skb->data; - - if (tfp->seqno == search_seqno) { - is_head_tmp = tmp_up->flags & BATADV_UNI_FRAG_HEAD; - if (is_head_tmp != is_head) - return tfp; - else - goto mov_tail; - } - } - return NULL; - -mov_tail: - list_move_tail(&tfp->list, head); - return NULL; -} - -void batadv_frag_list_free(struct list_head *head) -{ - struct batadv_frag_packet_list_entry *pf, *tmp_pf; - - if (!list_empty(head)) { - list_for_each_entry_safe(pf, tmp_pf, head, list) { - kfree_skb(pf->skb); - list_del(&pf->list); - kfree(pf); - } - } - return; -} - -/* frag_reassemble_skb(): - * returns NET_RX_DROP if the operation failed - skb is left intact - * returns NET_RX_SUCCESS if the fragment was buffered (skb_new will be NULL) - * or the skb could be reassembled (skb_new will point to the new packet and - * skb was freed) - */ -int batadv_frag_reassemble_skb(struct sk_buff *skb, - struct batadv_priv *bat_priv, - struct sk_buff **new_skb) -{ - struct batadv_orig_node *orig_node; - struct batadv_frag_packet_list_entry *tmp_frag_entry; - int ret = NET_RX_DROP; - struct batadv_unicast_frag_packet *unicast_packet; - - unicast_packet = (struct batadv_unicast_frag_packet *)skb->data; - *new_skb = NULL; - - orig_node = batadv_orig_hash_find(bat_priv, unicast_packet->orig); - if (!orig_node) - goto out; - - orig_node->last_frag_packet = jiffies; - - if (list_empty(&orig_node->frag_list) && - batadv_frag_create_buffer(&orig_node->frag_list)) { - pr_debug("couldn't create frag buffer\n"); - goto out; - } - - tmp_frag_entry = batadv_frag_search_packet(&orig_node->frag_list, - unicast_packet); - - if (!tmp_frag_entry) { - batadv_frag_create_entry(&orig_node->frag_list, skb); - ret = NET_RX_SUCCESS; - goto out; - } - - *new_skb = batadv_frag_merge_packet(&orig_node->frag_list, - tmp_frag_entry, skb); - /* if not, merge failed */ - if (*new_skb) - ret = NET_RX_SUCCESS; - -out: - if (orig_node) - batadv_orig_node_free_ref(orig_node); - return ret; -} - -int batadv_frag_send_skb(struct sk_buff *skb, struct batadv_priv *bat_priv, - struct batadv_hard_iface *hard_iface, - const uint8_t dstaddr[]) -{ - struct batadv_unicast_packet tmp_uc, *unicast_packet; - struct batadv_hard_iface *primary_if; - struct sk_buff *frag_skb; - struct batadv_unicast_frag_packet *frag1, *frag2; - int uc_hdr_len = sizeof(*unicast_packet); - int ucf_hdr_len = sizeof(*frag1); - int data_len = skb->len - uc_hdr_len; - int large_tail = 0, ret = NET_RX_DROP; - uint16_t seqno; - - primary_if = batadv_primary_if_get_selected(bat_priv); - if (!primary_if) - goto dropped; - - frag_skb = dev_alloc_skb(data_len - (data_len / 2) + ucf_hdr_len); - if (!frag_skb) - goto dropped; - skb_reserve(frag_skb, ucf_hdr_len); - - unicast_packet = (struct batadv_unicast_packet *)skb->data; - memcpy(&tmp_uc, unicast_packet, uc_hdr_len); - skb_split(skb, frag_skb, data_len / 2 + uc_hdr_len); - - if (batadv_skb_head_push(skb, ucf_hdr_len - uc_hdr_len) < 0 || - batadv_skb_head_push(frag_skb, ucf_hdr_len) < 0) - goto drop_frag; - - frag1 = (struct batadv_unicast_frag_packet *)skb->data; - frag2 = (struct batadv_unicast_frag_packet *)frag_skb->data; - - memcpy(frag1, &tmp_uc, sizeof(tmp_uc)); - - frag1->header.ttl--; - frag1->header.version = BATADV_COMPAT_VERSION; - frag1->header.packet_type = BATADV_UNICAST_FRAG; - - memcpy(frag1->orig, primary_if->net_dev->dev_addr, ETH_ALEN); - memcpy(frag2, frag1, sizeof(*frag2)); - - if (data_len & 1) - large_tail = BATADV_UNI_FRAG_LARGETAIL; - - frag1->flags = BATADV_UNI_FRAG_HEAD | large_tail; - frag2->flags = large_tail; - - seqno = atomic_add_return(2, &hard_iface->frag_seqno); - frag1->seqno = htons(seqno - 1); - frag2->seqno = htons(seqno); - - batadv_send_skb_packet(skb, hard_iface, dstaddr); - batadv_send_skb_packet(frag_skb, hard_iface, dstaddr); - ret = NET_RX_SUCCESS; - goto out; - -drop_frag: - kfree_skb(frag_skb); -dropped: - kfree_skb(skb); -out: - if (primary_if) - batadv_hardif_free_ref(primary_if); - return ret; -} - -/** - * batadv_unicast_push_and_fill_skb - extends the buffer and initializes the - * common fields for unicast packets - * @skb: packet - * @hdr_size: amount of bytes to push at the beginning of the skb - * @orig_node: the destination node - * - * Returns false if the buffer extension was not possible or true otherwise - */ -static bool batadv_unicast_push_and_fill_skb(struct sk_buff *skb, int hdr_size, - struct batadv_orig_node *orig_node) -{ - struct batadv_unicast_packet *unicast_packet; - uint8_t ttvn = (uint8_t)atomic_read(&orig_node->last_ttvn); - - if (batadv_skb_head_push(skb, hdr_size) < 0) - return false; - - unicast_packet = (struct batadv_unicast_packet *)skb->data; - unicast_packet->header.version = BATADV_COMPAT_VERSION; - /* batman packet type: unicast */ - unicast_packet->header.packet_type = BATADV_UNICAST; - /* set unicast ttl */ - unicast_packet->header.ttl = BATADV_TTL; - /* copy the destination for faster routing */ - memcpy(unicast_packet->dest, orig_node->orig, ETH_ALEN); - /* set the destination tt version number */ - unicast_packet->ttvn = ttvn; - - return true; -} - -/** - * batadv_unicast_prepare_skb - encapsulate an skb with a unicast header - * @skb: the skb containing the payload to encapsulate - * @orig_node: the destination node - * - * Returns false if the payload could not be encapsulated or true otherwise - */ -static bool batadv_unicast_prepare_skb(struct sk_buff *skb, - struct batadv_orig_node *orig_node) -{ - size_t uni_size = sizeof(struct batadv_unicast_packet); - return batadv_unicast_push_and_fill_skb(skb, uni_size, orig_node); -} - -/** - * batadv_unicast_4addr_prepare_skb - encapsulate an skb with a unicast4addr - * header - * @bat_priv: the bat priv with all the soft interface information - * @skb: the skb containing the payload to encapsulate - * @orig_node: the destination node - * @packet_subtype: the batman 4addr packet subtype to use - * - * Returns false if the payload could not be encapsulated or true otherwise - */ -bool batadv_unicast_4addr_prepare_skb(struct batadv_priv *bat_priv, - struct sk_buff *skb, - struct batadv_orig_node *orig, - int packet_subtype) -{ - struct batadv_hard_iface *primary_if; - struct batadv_unicast_4addr_packet *unicast_4addr_packet; - bool ret = false; - - primary_if = batadv_primary_if_get_selected(bat_priv); - if (!primary_if) - goto out; - - /* pull the header space and fill the unicast_packet substructure. - * We can do that because the first member of the unicast_4addr_packet - * is of type struct unicast_packet - */ - if (!batadv_unicast_push_and_fill_skb(skb, - sizeof(*unicast_4addr_packet), - orig)) - goto out; - - unicast_4addr_packet = (struct batadv_unicast_4addr_packet *)skb->data; - unicast_4addr_packet->u.header.packet_type = BATADV_UNICAST_4ADDR; - memcpy(unicast_4addr_packet->src, primary_if->net_dev->dev_addr, - ETH_ALEN); - unicast_4addr_packet->subtype = packet_subtype; - unicast_4addr_packet->reserved = 0; - - ret = true; -out: - if (primary_if) - batadv_hardif_free_ref(primary_if); - return ret; -} - -/** - * batadv_unicast_generic_send_skb - send an skb as unicast - * @bat_priv: the bat priv with all the soft interface information - * @skb: payload to send - * @packet_type: the batman unicast packet type to use - * @packet_subtype: the batman packet subtype. It is ignored if packet_type is - * not BATADV_UNICAT_4ADDR - * - * Returns 1 in case of error or 0 otherwise - */ -int batadv_unicast_generic_send_skb(struct batadv_priv *bat_priv, - struct sk_buff *skb, int packet_type, - int packet_subtype) -{ - struct ethhdr *ethhdr = (struct ethhdr *)skb->data; - struct batadv_unicast_packet *unicast_packet; - struct batadv_orig_node *orig_node; - struct batadv_neigh_node *neigh_node; - int data_len = skb->len; - int ret = NET_RX_DROP; - unsigned int dev_mtu; - - /* get routing information */ - if (is_multicast_ether_addr(ethhdr->h_dest)) { - orig_node = batadv_gw_get_selected_orig(bat_priv); - if (orig_node) - goto find_router; - } - - /* check for tt host - increases orig_node refcount. - * returns NULL in case of AP isolation - */ - orig_node = batadv_transtable_search(bat_priv, ethhdr->h_source, - ethhdr->h_dest); - -find_router: - /* find_router(): - * - if orig_node is NULL it returns NULL - * - increases neigh_nodes refcount if found. - */ - neigh_node = batadv_find_router(bat_priv, orig_node, NULL); - - if (!neigh_node) - goto out; - - switch (packet_type) { - case BATADV_UNICAST: - batadv_unicast_prepare_skb(skb, orig_node); - break; - case BATADV_UNICAST_4ADDR: - batadv_unicast_4addr_prepare_skb(bat_priv, skb, orig_node, - packet_subtype); - break; - default: - /* this function supports UNICAST and UNICAST_4ADDR only. It - * should never be invoked with any other packet type - */ - goto out; - } - - unicast_packet = (struct batadv_unicast_packet *)skb->data; - - /* inform the destination node that we are still missing a correct route - * for this client. The destination will receive this packet and will - * try to reroute it because the ttvn contained in the header is less - * than the current one - */ - if (batadv_tt_global_client_is_roaming(bat_priv, ethhdr->h_dest)) - unicast_packet->ttvn = unicast_packet->ttvn - 1; - - dev_mtu = neigh_node->if_incoming->net_dev->mtu; - /* fragmentation mechanism only works for UNICAST (now) */ - if (packet_type == BATADV_UNICAST && - atomic_read(&bat_priv->fragmentation) && - data_len + sizeof(*unicast_packet) > dev_mtu) { - /* send frag skb decreases ttl */ - unicast_packet->header.ttl++; - ret = batadv_frag_send_skb(skb, bat_priv, - neigh_node->if_incoming, - neigh_node->addr); - goto out; - } - - if (batadv_send_skb_to_orig(skb, orig_node, NULL) != NET_XMIT_DROP) - ret = 0; - -out: - if (neigh_node) - batadv_neigh_node_free_ref(neigh_node); - if (orig_node) - batadv_orig_node_free_ref(orig_node); - if (ret == NET_RX_DROP) - kfree_skb(skb); - return ret; -} diff --git a/unicast.h b/unicast.h deleted file mode 100644 index 429cf8a..0000000 --- a/unicast.h +++ /dev/null @@ -1,92 +0,0 @@ -/* Copyright (C) 2010-2013 B.A.T.M.A.N. contributors: - * - * Andreas Langer - * - * 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_UNICAST_H_ -#define _NET_BATMAN_ADV_UNICAST_H_ - -#include "packet.h" - -#define BATADV_FRAG_TIMEOUT 10000 /* purge frag list entries after time in ms */ -#define BATADV_FRAG_BUFFER_SIZE 6 /* number of list elements in buffer */ - -int batadv_frag_reassemble_skb(struct sk_buff *skb, - struct batadv_priv *bat_priv, - struct sk_buff **new_skb); -void batadv_frag_list_free(struct list_head *head); -int batadv_frag_send_skb(struct sk_buff *skb, struct batadv_priv *bat_priv, - struct batadv_hard_iface *hard_iface, - const uint8_t dstaddr[]); -bool batadv_unicast_4addr_prepare_skb(struct batadv_priv *bat_priv, - struct sk_buff *skb, - struct batadv_orig_node *orig_node, - int packet_subtype); -int batadv_unicast_generic_send_skb(struct batadv_priv *bat_priv, - struct sk_buff *skb, int packet_type, - int packet_subtype); - - -/** - * batadv_unicast_send_skb - send the skb encapsulated in a unicast packet - * @bat_priv: the bat priv with all the soft interface information - * @skb: the payload to send - */ -static inline int batadv_unicast_send_skb(struct batadv_priv *bat_priv, - struct sk_buff *skb) -{ - return batadv_unicast_generic_send_skb(bat_priv, skb, BATADV_UNICAST, - 0); -} - -/** - * batadv_unicast_send_skb - send the skb encapsulated in a unicast4addr packet - * @bat_priv: the bat priv with all the soft interface information - * @skb: the payload to send - * @packet_subtype: the batman 4addr packet subtype to use - */ -static inline int batadv_unicast_4addr_send_skb(struct batadv_priv *bat_priv, - struct sk_buff *skb, - int packet_subtype) -{ - return batadv_unicast_generic_send_skb(bat_priv, skb, - BATADV_UNICAST_4ADDR, - packet_subtype); -} - -static inline int batadv_frag_can_reassemble(const struct sk_buff *skb, int mtu) -{ - const struct batadv_unicast_frag_packet *unicast_packet; - int uneven_correction = 0; - unsigned int merged_size; - - unicast_packet = (struct batadv_unicast_frag_packet *)skb->data; - - if (unicast_packet->flags & BATADV_UNI_FRAG_LARGETAIL) { - if (unicast_packet->flags & BATADV_UNI_FRAG_HEAD) - uneven_correction = 1; - else - uneven_correction = -1; - } - - merged_size = (skb->len - sizeof(*unicast_packet)) * 2; - merged_size += sizeof(struct batadv_unicast_packet) + uneven_correction; - - return merged_size <= mtu; -} - -#endif /* _NET_BATMAN_ADV_UNICAST_H_ */
Hi,
On Mon, May 13, 2013 at 10:23:00PM +0200, Martin Hundebøll wrote:
[...]
+/**
- batadv_send_skb_prepare_unicast - encapsulate an skb with a unicast header
- @skb: the skb containing the payload to encapsulate
- @orig_node: the destination node
- Returns false if the payload could not be encapsulated or true otherwise
- */
+static bool batadv_send_skb_prepare_unicast(struct sk_buff *skb,
struct batadv_orig_node *orig_node)
+{
- size_t uni_size = sizeof(struct batadv_unicast_packet);
a blank line here is needed
- return batadv_send_skb_push_fill_unicast(skb, uni_size, orig_node);
+}
Cheers,
Fragments arriving at their destination are buffered for later merge. Merged packets are passed to the main receive function as had they never been fragmented.
Fragments are forwarded without merging if the MTU of the outgoing interface is smaller than the size of the merged packet.
Signed-off-by: Martin Hundebøll martin@hundeboll.net ---
v2: update copyright years remove () and uppercase beginnings from kernel doc use hlist_for_entry_safe() in batadv_frag_clear_list() merge batadv_frag_clear_orig() and batadv_frag_check_orig() update some kernel doc with return values remove "inline" from static function in .c file rename lists of fragments to "chain" remove type in kernel doc remove initialization from "curr" and "new" in insert_packet() rename "curr" and "new" to "frag_entry_{curr,new}" rename packet to frag_packet change return description in kernel doc for batadv_frag_merge_packets() fix return value on failure in pskb_expand_head() in merge_packets() add kernel doc for counters in types.h
Makefile.kbuild | 1 + fragmentation.c | 360 +++++++++++++++++++++++++++++++++++++++++++++++++++++++ fragmentation.h | 47 ++++++++ main.c | 4 + main.h | 9 ++ originator.c | 14 ++- packet.h | 27 +++++ routing.c | 60 ++++++++++ routing.h | 2 + soft-interface.c | 4 + types.h | 38 ++++++ 11 files changed, 565 insertions(+), 1 deletion(-) create mode 100644 fragmentation.c create mode 100644 fragmentation.h
diff --git a/Makefile.kbuild b/Makefile.kbuild index f9b465b..4f4aabb 100644 --- a/Makefile.kbuild +++ b/Makefile.kbuild @@ -24,6 +24,7 @@ batman-adv-y += bitarray.o batman-adv-$(CONFIG_BATMAN_ADV_BLA) += bridge_loop_avoidance.o batman-adv-y += debugfs.o batman-adv-$(CONFIG_BATMAN_ADV_DAT) += distributed-arp-table.o +batman-adv-y += fragmentation.o batman-adv-y += gateway_client.o batman-adv-y += gateway_common.o batman-adv-y += hard-interface.o diff --git a/fragmentation.c b/fragmentation.c new file mode 100644 index 0000000..5a891d7 --- /dev/null +++ b/fragmentation.c @@ -0,0 +1,360 @@ +/* Copyright (C) 2013 B.A.T.M.A.N. contributors: + * + * Martin Hundebøll martin@hundeboll.net + * + * 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 + */ + +#include "main.h" +#include "fragmentation.h" +#include "send.h" +#include "originator.h" +#include "routing.h" +#include "hard-interface.h" +#include "soft-interface.h" + + +/** + * batadv_frag_clear_chain - delete entries in the fragment buffer chain. + * @head: head of chain with entries. + * + * Frees fragments in the passed hlist. Should be called with appropriate lock. + */ +static void batadv_frag_clear_chain(struct hlist_head *head) +{ + struct batadv_frag_list_entry *entry; + struct hlist_node *node; + + hlist_for_each_entry_safe(entry, node, head, list) { + hlist_del(&entry->list); + kfree_skb(entry->skb); + kfree(entry); + } +} + +/** + * batadv_frag_purge_orig - free fragments associated to an orig. + * @orig_node: Originator to free fragments from. + * @check_cb: Optional function to tell if an entry should be purged. + */ +void batadv_frag_purge_orig(struct batadv_orig_node *orig_node, + bool (*check_cb)(struct batadv_frag_table_entry *)) +{ + struct batadv_frag_table_entry *chain; + uint8_t i; + + for (i = 0; i < BATADV_FRAG_BUFFER_COUNT; i++) { + chain = &orig_node->fragments[i]; + spin_lock_bh(&orig_node->fragments[i].lock); + + if (!check_cb || check_cb(chain)) { + batadv_frag_clear_chain(&orig_node->fragments[i].head); + orig_node->fragments[i].size = 0; + } + + spin_unlock_bh(&orig_node->fragments[i].lock); + } +} + +/** + * batadv_frag_size_limit - maximum possible size of packet to be fragmented. + * + * Returns the maximum size of payload that can be fragmented. + */ +static int batadv_frag_size_limit(void) +{ + int limit = BATADV_FRAG_MAX_FRAG_SIZE; + limit -= sizeof(struct batadv_frag_packet); + limit *= BATADV_FRAG_MAX_FRAGMENTS; + return limit; +} + +/** + * batadv_frag_init_chain - check and prepare fragment chain for new fragment. + * @chain: Chain in fragments table to init. + * @seqno: Sequence number of the received fragment. + * + * Make chain ready for a fragment with sequence number "seqno". Delete existing + * entries if they have an "old" sequence number. + * + * Returns true if chain is empty and caller can just insert the new fragment + * without searching for the right position. + * + * Caller must hold chain->lock. + */ +static bool batadv_frag_init_chain(struct batadv_frag_table_entry *chain, + uint16_t seqno) +{ + if (chain->seqno == seqno) + return false; + + if (!hlist_empty(&chain->head)) + batadv_frag_clear_chain(&chain->head); + + chain->size = 0; + chain->seqno = seqno; + + return true; +} + +/** + * batadv_frag_insert_packet - insert a fragment into a fragment chain. + * @orig_node: Originator that the fragment was received from. + * @skb: skb to insert. + * @head: head to attach complete chains of fragments to. + * + * Insert a new fragment into the reverse ordered chain in the right table + * entry. The hash table entry is cleared if "old" fragments exist in it. + * + * Returns true if skb is buffered, false on error. If the chain has all the + * fragments needed to merge the packet, the chain is moved to the passed head + * to avoid locking the chain in the table. + */ +static bool batadv_frag_insert_packet(struct batadv_orig_node *orig_node, + struct sk_buff *skb, + struct hlist_head *chain_out) +{ + struct batadv_frag_table_entry *chain; + struct batadv_frag_list_entry *frag_entry_new, *frag_entry_curr; + struct batadv_frag_packet *frag_packet; + uint8_t bucket; + uint16_t seqno, hdr_size = sizeof(struct batadv_frag_packet); + + if (skb_linearize(skb) < 0) + goto err; + + frag_packet = (struct batadv_frag_packet *)skb->data; + seqno = ntohs(frag_packet->seqno); + bucket = seqno % BATADV_FRAG_BUFFER_COUNT; + + frag_entry_new = kmalloc(sizeof(*frag_entry_new), GFP_ATOMIC); + if (!frag_entry_new) + goto err; + + frag_entry_new->skb = skb; + frag_entry_new->no = frag_packet->no; + + /* Select entry in the "chain table" and delete any prior fragments + * with another sequence number. batadv_frag_init_chain() returns true, + * if the list is empty at return. + */ + chain = &orig_node->fragments[bucket]; + spin_lock_bh(&chain->lock); + if (batadv_frag_init_chain(chain, seqno)) { + hlist_add_head(&frag_entry_new->list, &chain->head); + chain->size = skb->len - hdr_size; + chain->timestamp = jiffies; + goto out; + } + + /* Find the position for the new fragment. */ + hlist_for_each_entry(frag_entry_curr, &chain->head, list) { + /* Drop packet if fragment already exists. */ + if (frag_entry_curr->no == frag_entry_new->no) + goto err_unlock; + + /* Order fragments from highest to lowest. */ + if (frag_entry_curr->no < frag_entry_new->no) { + hlist_add_before(&frag_entry_new->list, + &frag_entry_curr->list); + chain->size += skb->len - hdr_size; + chain->timestamp = jiffies; + goto out; + } + } + + /* We reached the end of the list, so insert after 'frag_entry_curr'. */ + if (likely(frag_entry_curr)) { + hlist_add_after(&frag_entry_curr->list, &frag_entry_new->list); + chain->size += skb->len - hdr_size; + chain->timestamp = jiffies; + } + +out: + if (chain->size > batadv_frag_size_limit() || + ntohs(frag_packet->total_size) > batadv_frag_size_limit()) { + /* Clear chain if total size of either the list or the packet + * exceeds the maximum size of one merged packet. + */ + batadv_frag_clear_chain(&chain->head); + chain->size = 0; + } else if (ntohs(frag_packet->total_size) == chain->size) { + /* All fragments received. Hand over chain to caller. */ + hlist_move_list(&chain->head, chain_out); + chain->size = 0; + } + + spin_unlock_bh(&chain->lock); + return true; + +err_unlock: + spin_unlock_bh(&chain->lock); + kfree(frag_entry_new); +err: + return false; +} + +/** + * batadv_frag_merge_packets - merge a chain of fragments. + * @head: head of chain with fragments. + * @skb: Packet with total size of skb after merging. + * + * Expands the first skb in the chain and copies the content of the remaining + * skb's into the expanded one. After doing so, it clears the chain. + * + * Returns the merged skb or NULL on error. + */ +static struct sk_buff * +batadv_frag_merge_packets(struct hlist_head *chain, struct sk_buff *skb) +{ + struct batadv_frag_packet *packet; + struct batadv_frag_list_entry *entry; + struct sk_buff *skb_out = NULL; + int size, hdr_size = sizeof(struct batadv_frag_packet); + + /* Make sure incoming skb has non-bogus data. */ + packet = (struct batadv_frag_packet *)skb->data; + size = ntohs(packet->total_size); + if (size > batadv_frag_size_limit()) + goto free; + + /* Remove first entry, as this is the destination for the rest of the + * fragments. + */ + entry = hlist_entry(chain->first, struct batadv_frag_list_entry, list); + hlist_del(&entry->list); + skb_out = entry->skb; + kfree(entry); + + /* Make room for the rest of the fragments. */ + if (pskb_expand_head(skb_out, 0, size - skb->len, GFP_ATOMIC) < 0) { + kfree_skb(skb_out); + skb_out = NULL; + goto free; + } + + /* Move the existing MAC header to just before the payload. (Override + * the fragment header.) + */ + skb_pull_rcsum(skb_out, hdr_size); + memmove(skb_out->data - ETH_HLEN, skb_mac_header(skb_out), ETH_HLEN); + skb_set_mac_header(skb_out, -ETH_HLEN); + skb_reset_network_header(skb_out); + skb_reset_transport_header(skb_out); + + /* Copy the payload of the each fragment into the last skb */ + hlist_for_each_entry(entry, chain, list) { + size = entry->skb->len - hdr_size; + memcpy(skb_put(skb_out, size), entry->skb->data + hdr_size, + size); + } + +free: + /* Locking is not needed, because 'chain' is not part of any orig. */ + batadv_frag_clear_chain(chain); + return skb_out; +} + +/** + * batadv_frag_skb_buffer - buffer fragment for later merge. + * @skb: skb to buffer. + * @orig_node_src: Originator that the skb is received from. + * + * Add fragment to buffer and merge fragments if possible. + * + * There are three possible outcomes: 1) Packet is merged: Return true and + * set *skb to merged packet; 2) Packet is buffered: Return true and set *skb + * to NULL; 3) Error: Return false and leave skb as is. + */ +bool batadv_frag_skb_buffer(struct sk_buff **skb, + struct batadv_orig_node *orig_node_src) +{ + struct sk_buff *skb_out = NULL; + struct hlist_head head = HLIST_HEAD_INIT; + bool ret = false; + + /* Add packet to buffer and table entry if merge is possible. */ + if (!batadv_frag_insert_packet(orig_node_src, *skb, &head)) + goto out_err; + + /* Leave if more fragments are needed to merge. */ + if (hlist_empty(&head)) + goto out; + + skb_out = batadv_frag_merge_packets(&head, *skb); + if (!skb_out) + goto out_err; + +out: + *skb = skb_out; + ret = true; +out_err: + return ret; +} + +/** + * batadv_frag_skb_fwd - forward fragments that would exceed MTU when merged. + * @skb: skb to forward. + * @recv_if: Interface that the skb is received on. + * @orig_node_src: Originator that the skb is received from. + * + * Look up the next-hop of the fragments payload and check if the merged packet + * will exceed the MTU towards the next-hop. If so, the fragment is forwarded + * without merging it. + * + * Returns true if the fragment is consumed/forwarded, false otherwise. + */ +bool batadv_frag_skb_fwd(struct sk_buff *skb, + struct batadv_hard_iface *recv_if, + struct batadv_orig_node *orig_node_src) +{ + struct batadv_priv *bat_priv = netdev_priv(recv_if->soft_iface); + struct batadv_orig_node *orig_node_dst = NULL; + struct batadv_neigh_node *neigh_node = NULL; + struct batadv_frag_packet *packet; + uint16_t total_size; + bool ret = false; + + packet = (struct batadv_frag_packet *)skb->data; + orig_node_dst = batadv_orig_hash_find(bat_priv, packet->dest); + if (!orig_node_dst) + goto out; + + neigh_node = batadv_find_router(bat_priv, orig_node_dst, recv_if); + if (!neigh_node) + goto out; + + /* Forward the fragment, if the merged packet would be too big to + * be assembled. + */ + total_size = ntohs(packet->total_size) + ETH_HLEN; + if (total_size > neigh_node->if_incoming->net_dev->mtu) { + batadv_inc_counter(bat_priv, BATADV_CNT_FRAG_FWD); + batadv_add_counter(bat_priv, BATADV_CNT_FRAG_FWD_BYTES, + skb->len + ETH_HLEN); + + packet->header.ttl--; + batadv_send_skb_packet(skb, neigh_node->if_incoming, + neigh_node->addr); + ret = true; + } + +out: + if (orig_node_dst) + batadv_orig_node_free_ref(orig_node_dst); + if (neigh_node) + batadv_neigh_node_free_ref(neigh_node); + return ret; +} diff --git a/fragmentation.h b/fragmentation.h new file mode 100644 index 0000000..2af5df5 --- /dev/null +++ b/fragmentation.h @@ -0,0 +1,47 @@ +/* Copyright (C) 2013 B.A.T.M.A.N. contributors: + * + * Martin Hundebøll martin@hundeboll.net + * + * 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_FRAGMENTATION_H_ +#define _NET_BATMAN_ADV_FRAGMENTATION_H_ + +void batadv_frag_purge_orig(struct batadv_orig_node *orig, + bool (*check_cb)(struct batadv_frag_table_entry *)); +bool batadv_frag_skb_fwd(struct sk_buff *skb, + struct batadv_hard_iface *recv_if, + struct batadv_orig_node *orig_node_src); +bool batadv_frag_skb_buffer(struct sk_buff **skb, + struct batadv_orig_node *orig_node); + +/** + * batadv_frag_check_entry - check if a list of fragments has timed out. + * @frags_entry: table entry to check + * + * Returns true if the frags entry has timed out; false otherwise + */ +static inline bool +batadv_frag_check_entry(struct batadv_frag_table_entry *frags_entry) +{ + if (!hlist_empty(&frags_entry->head) && + batadv_has_timed_out(frags_entry->timestamp, BATADV_FRAG_TIMEOUT)) + return true; + else + return false; +} + +#endif /* _NET_BATMAN_ADV_FRAGMENTATION_H_ */ diff --git a/main.c b/main.c index 3b5f78f..7188beb 100644 --- a/main.c +++ b/main.c @@ -36,6 +36,7 @@ #include "hash.h" #include "bat_algo.h" #include "network-coding.h" +#include "fragmentation.h"
/* List manipulations on hardif_list have to be rtnl_lock()'ed, @@ -342,6 +343,7 @@ static void batadv_recv_handler_init(void) BUILD_BUG_ON(offsetof(struct batadv_unicast_4addr_packet, src) != 10); BUILD_BUG_ON(offsetof(struct batadv_unicast_packet, dest) != 4); BUILD_BUG_ON(offsetof(struct batadv_unicast_tvlv_packet, dst) != 4); + BUILD_BUG_ON(offsetof(struct batadv_frag_packet, dest) != 4); BUILD_BUG_ON(offsetof(struct batadv_icmp_packet, dst) != 4); BUILD_BUG_ON(offsetof(struct batadv_icmp_packet_rr, dst) != 4);
@@ -357,6 +359,8 @@ static void batadv_recv_handler_init(void) batadv_rx_handler[BATADV_UNICAST_TVLV] = batadv_recv_unicast_tvlv; /* batman icmp packet */ batadv_rx_handler[BATADV_ICMP] = batadv_recv_icmp_packet; + /* Fragmented packets */ + batadv_rx_handler[BATADV_UNICAST_FRAG] = batadv_recv_frag_packet; }
int diff --git a/main.h b/main.h index 71440df..3db6c2e 100644 --- a/main.h +++ b/main.h @@ -131,6 +131,15 @@ enum batadv_uev_type {
#define BATADV_GW_THRESHOLD 50
+/* Number of fragment chains for each orig_node */ +#define BATADV_FRAG_BUFFER_COUNT 8 +/* Maximum number of fragments for one packet */ +#define BATADV_FRAG_MAX_FRAGMENTS 16 +/* Maxumim size of each fragment */ +#define BATADV_FRAG_MAX_FRAG_SIZE 1400 +/* Time to keep fragments while waiting for rest of the fragments */ +#define BATADV_FRAG_TIMEOUT 10000 + #define BATADV_DAT_CANDIDATE_NOT_FOUND 0 #define BATADV_DAT_CANDIDATE_ORIG 1
diff --git a/originator.c b/originator.c index 898b0ce..a591dc5 100644 --- a/originator.c +++ b/originator.c @@ -28,6 +28,7 @@ #include "soft-interface.h" #include "bridge_loop_avoidance.h" #include "network-coding.h" +#include "fragmentation.h"
/* hash class keys */ static struct lock_class_key batadv_orig_hash_lock_class_key; @@ -145,6 +146,8 @@ static void batadv_orig_node_free_rcu(struct rcu_head *rcu) /* Free nc_nodes */ batadv_nc_purge_orig(orig_node->bat_priv, orig_node, NULL);
+ batadv_frag_purge_orig(orig_node, NULL); + batadv_tt_global_del_orig(orig_node->bat_priv, orig_node, "originator timed out");
@@ -215,7 +218,7 @@ struct batadv_orig_node *batadv_get_orig_node(struct batadv_priv *bat_priv, const uint8_t *addr) { struct batadv_orig_node *orig_node; - int size; + int size, i; int hash_added; unsigned long reset_time;
@@ -267,6 +270,12 @@ struct batadv_orig_node *batadv_get_orig_node(struct batadv_priv *bat_priv, size = bat_priv->num_ifaces * sizeof(uint8_t); orig_node->bcast_own_sum = kzalloc(size, GFP_ATOMIC);
+ for (i = 0; i < BATADV_FRAG_BUFFER_COUNT; i++) { + INIT_HLIST_HEAD(&orig_node->fragments[i].head); + spin_lock_init(&orig_node->fragments[i].lock); + orig_node->fragments[i].size = 0; + } + if (!orig_node->bcast_own_sum) goto free_bcast_own;
@@ -388,6 +397,9 @@ static void _batadv_purge_orig(struct batadv_priv *bat_priv) batadv_orig_node_free_ref(orig_node); continue; } + + batadv_frag_purge_orig(orig_node, + batadv_frag_check_entry); } spin_unlock_bh(list_lock); } diff --git a/packet.h b/packet.h index 96b8e57..ae38769 100644 --- a/packet.h +++ b/packet.h @@ -249,6 +249,33 @@ struct batadv_unicast_4addr_packet { */ };
+/** + * struct batadv_frag_packet - Network structure for fragments. + * @header: Common batman-adv header with type, compatibility version, and ttl. + * @dest: Final destination used when routing fragments. + * @orig: Originator of the fragment used when merging the packet. + * @no: Fragment number within this sequence. + * @reserved: Unused + * @seqno: Sequence identification. + * @total_size: Size of the merged packet. + */ +struct batadv_frag_packet { + struct batadv_header header; +#if defined(__BIG_ENDIAN_BITFIELD) + uint8_t no:4; + uint8_t reserved:4; +#elif defined(__LITTLE_ENDIAN_BITFIELD) + uint8_t reserved:4; + uint8_t no:4; +#else +#error "unknown bitfield endianess" +#endif + uint8_t dest[ETH_ALEN]; + uint8_t orig[ETH_ALEN]; + __be16 seqno; + __be16 total_size; +}; + struct batadv_bcast_packet { struct batadv_header header; uint8_t reserved; diff --git a/routing.c b/routing.c index a099d32..b8239f1 100644 --- a/routing.c +++ b/routing.c @@ -28,6 +28,7 @@ #include "bridge_loop_avoidance.h" #include "distributed-arp-table.h" #include "network-coding.h" +#include "fragmentation.h"
static int batadv_route_unicast_packet(struct sk_buff *skb, struct batadv_hard_iface *recv_if); @@ -997,6 +998,65 @@ int batadv_recv_unicast_tvlv(struct sk_buff *skb, return ret; }
+/* + * batadv_recv_frag_packet() - Process received fragment. + * @skb: The received fragment. + * @recv_if: Interface that the skb is received on. + * + * This function does one of the three following things: 1) Forward fragment, if + * the assembled packet will exceed our MTU; 2) Buffer fragment, if we till + * lack further fragments; 3) Merge fragments, if we have all needed parts. + * + * Returns NET_RX_DROP if the skb is not consumed, NET_RX_SUCCESS otherwise. + */ +int batadv_recv_frag_packet(struct sk_buff *skb, + struct batadv_hard_iface *recv_if) +{ + struct batadv_priv *bat_priv = netdev_priv(recv_if->soft_iface); + struct batadv_orig_node *orig_node_src = NULL; + struct batadv_frag_packet *frag_packet; + int ret = NET_RX_DROP; + + if (batadv_check_unicast_packet(bat_priv, skb, + sizeof(*frag_packet)) < 0) + goto out; + + frag_packet = (struct batadv_frag_packet *)skb->data; + orig_node_src = batadv_orig_hash_find(bat_priv, frag_packet->orig); + if (!orig_node_src) + goto out; + + /* Route the fragment if it is not for us and too big to be merged. */ + if (!batadv_is_my_mac(bat_priv, frag_packet->dest) && + batadv_frag_skb_fwd(skb, recv_if, orig_node_src)) { + ret = NET_RX_SUCCESS; + goto out; + } + + batadv_inc_counter(bat_priv, BATADV_CNT_FRAG_RX); + batadv_add_counter(bat_priv, BATADV_CNT_FRAG_RX_BYTES, + skb->len + ETH_HLEN); + + /* Add fragment to buffer and merge if possible. */ + if (!batadv_frag_skb_buffer(&skb, orig_node_src)) + goto out; + + /* Deliver merged packet to the appropriate handler, if it was + * merged + */ + if (skb) + batadv_batman_skb_recv(skb, recv_if->net_dev, + &recv_if->batman_adv_ptype, NULL); + + ret = NET_RX_SUCCESS; + +out: + if (orig_node_src) + batadv_orig_node_free_ref(orig_node_src); + + return ret; +} + int batadv_recv_bcast_packet(struct sk_buff *skb, struct batadv_hard_iface *recv_if) { diff --git a/routing.h b/routing.h index efab583..55d637a 100644 --- a/routing.h +++ b/routing.h @@ -30,6 +30,8 @@ int batadv_recv_icmp_packet(struct sk_buff *skb, struct batadv_hard_iface *recv_if); int batadv_recv_unicast_packet(struct sk_buff *skb, struct batadv_hard_iface *recv_if); +int batadv_recv_frag_packet(struct sk_buff *skb, + struct batadv_hard_iface *iface); int batadv_recv_bcast_packet(struct sk_buff *skb, struct batadv_hard_iface *recv_if); int batadv_recv_tt_query(struct sk_buff *skb, diff --git a/soft-interface.c b/soft-interface.c index 2f3810d..481642e 100644 --- a/soft-interface.c +++ b/soft-interface.c @@ -747,6 +747,10 @@ static const struct { { "mgmt_tx_bytes" }, { "mgmt_rx" }, { "mgmt_rx_bytes" }, + { "frag_rx" }, + { "frag_rx_bytes" }, + { "frag_fwd" }, + { "frag_fwd_bytes" }, { "tt_request_tx" }, { "tt_request_rx" }, { "tt_response_tx" }, diff --git a/types.h b/types.h index 6cde9c9..dd3f505 100644 --- a/types.h +++ b/types.h @@ -86,6 +86,34 @@ struct batadv_hard_iface { };
/** + * struct batadv_frag_table_entry - Head in the fragment buffer table. + * @head: Head of list with fragments. + * @lock: Lock to protect the list of fragments. + * @timestamp: Time (jiffie) of last received fragment. + * @seqno: Sequence number of the fragments in the list. + * @size: Accumulated size of packets in list. + */ +struct batadv_frag_table_entry { + struct hlist_head head; + spinlock_t lock; /* Protects head */ + unsigned long timestamp; + uint16_t seqno; + uint16_t size; +}; + +/** + * struct batadv_frag_list_entry - Entry in a list of fragments. + * @list: List node information + * @skb: Fragment + * @no: Fragment number in the set. + */ +struct batadv_frag_list_entry { + struct hlist_node list; + struct sk_buff *skb; + uint8_t no; +}; + +/** * struct batadv_orig_node - structure for orig_list maintaining nodes of mesh * @orig: originator ethernet address * @primary_addr: hosts primary interface address @@ -128,6 +156,7 @@ struct batadv_hard_iface { * @out_coding_list: list of nodes that can hear this orig * @in_coding_list_lock: protects in_coding_list * @out_coding_list_lock: protects out_coding_list + * @fragments: array with heads for fragment chains */ struct batadv_orig_node { uint8_t orig[ETH_ALEN]; @@ -174,6 +203,7 @@ struct batadv_orig_node { spinlock_t in_coding_list_lock; /* Protects in_coding_list */ spinlock_t out_coding_list_lock; /* Protects out_coding_list */ #endif + struct batadv_frag_table_entry fragments[BATADV_FRAG_BUFFER_COUNT]; };
/** @@ -270,6 +300,10 @@ 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_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 + * @BATADV_CNT_FRAG_FWD_BYTES: forwarded fragment traffic bytes counter * @BATADV_CNT_TT_REQUEST_TX: transmitted tt req traffic packet counter * @BATADV_CNT_TT_REQUEST_RX: received tt req traffic packet counter * @BATADV_CNT_TT_RESPONSE_TX: transmitted tt resp traffic packet counter @@ -307,6 +341,10 @@ enum batadv_counters { BATADV_CNT_MGMT_TX_BYTES, BATADV_CNT_MGMT_RX, BATADV_CNT_MGMT_RX_BYTES, + BATADV_CNT_FRAG_RX, + BATADV_CNT_FRAG_RX_BYTES, + BATADV_CNT_FRAG_FWD, + BATADV_CNT_FRAG_FWD_BYTES, BATADV_CNT_TT_REQUEST_TX, BATADV_CNT_TT_REQUEST_RX, BATADV_CNT_TT_RESPONSE_TX,
Hi,
On Mon, May 13, 2013 at 10:23:01PM +0200, Martin Hundebøll wrote:
[...]
+/**
- batadv_frag_size_limit - maximum possible size of packet to be fragmented.
- Returns the maximum size of payload that can be fragmented.
- */
+static int batadv_frag_size_limit(void) +{
- int limit = BATADV_FRAG_MAX_FRAG_SIZE;
a blank line should go here
- limit -= sizeof(struct batadv_frag_packet);
- limit *= BATADV_FRAG_MAX_FRAGMENTS;
- return limit;
+}
+/**
- batadv_frag_init_chain - check and prepare fragment chain for new fragment.
- @chain: Chain in fragments table to init.
- @seqno: Sequence number of the received fragment.
- Make chain ready for a fragment with sequence number "seqno". Delete existing
- entries if they have an "old" sequence number.
- Returns true if chain is empty and caller can just insert the new fragment
- without searching for the right position.
- Caller must hold chain->lock.
- */
+static bool batadv_frag_init_chain(struct batadv_frag_table_entry *chain,
uint16_t seqno)
+{
- if (chain->seqno == seqno)
return false;
- if (!hlist_empty(&chain->head))
batadv_frag_clear_chain(&chain->head);
- chain->size = 0;
- chain->seqno = seqno;
- return true;
+}
+/**
- batadv_frag_insert_packet - insert a fragment into a fragment chain.
- @orig_node: Originator that the fragment was received from.
- @skb: skb to insert.
- @head: head to attach complete chains of fragments to.
- Insert a new fragment into the reverse ordered chain in the right table
- entry. The hash table entry is cleared if "old" fragments exist in it.
- Returns true if skb is buffered, false on error. If the chain has all the
- fragments needed to merge the packet, the chain is moved to the passed head
- to avoid locking the chain in the table.
- */
+static bool batadv_frag_insert_packet(struct batadv_orig_node *orig_node,
struct sk_buff *skb,
struct hlist_head *chain_out)
+{
- struct batadv_frag_table_entry *chain;
- struct batadv_frag_list_entry *frag_entry_new, *frag_entry_curr;
- struct batadv_frag_packet *frag_packet;
- uint8_t bucket;
- uint16_t seqno, hdr_size = sizeof(struct batadv_frag_packet);
- if (skb_linearize(skb) < 0)
goto err;
I guess the skb is linearised here so that, later, during the merge, the skbs can be copied by means of memcpy. I think you should put a comment here to explain why this skb_linearise is done, otherwise it seems to be unrelated to everything since the skb data is not accessed in this function (I guess we do it here and not in the merge function so that we do one linearisation at a time instead of linearising all the fragments at once).
Moreover, I'll add a todo saying that we may want to optimise fragmented skbs manipulation in the future and avoid this linearisation at all.
- frag_packet = (struct batadv_frag_packet *)skb->data;
- seqno = ntohs(frag_packet->seqno);
- bucket = seqno % BATADV_FRAG_BUFFER_COUNT;
- frag_entry_new = kmalloc(sizeof(*frag_entry_new), GFP_ATOMIC);
- if (!frag_entry_new)
goto err;
- frag_entry_new->skb = skb;
- frag_entry_new->no = frag_packet->no;
- /* Select entry in the "chain table" and delete any prior fragments
* with another sequence number. batadv_frag_init_chain() returns true,
* if the list is empty at return.
*/
- chain = &orig_node->fragments[bucket];
- spin_lock_bh(&chain->lock);
- if (batadv_frag_init_chain(chain, seqno)) {
hlist_add_head(&frag_entry_new->list, &chain->head);
chain->size = skb->len - hdr_size;
chain->timestamp = jiffies;
goto out;
- }
- /* Find the position for the new fragment. */
- hlist_for_each_entry(frag_entry_curr, &chain->head, list) {
/* Drop packet if fragment already exists. */
if (frag_entry_curr->no == frag_entry_new->no)
goto err_unlock;
/* Order fragments from highest to lowest. */
if (frag_entry_curr->no < frag_entry_new->no) {
hlist_add_before(&frag_entry_new->list,
&frag_entry_curr->list);
chain->size += skb->len - hdr_size;
chain->timestamp = jiffies;
goto out;
}
- }
- /* We reached the end of the list, so insert after 'frag_entry_curr'. */
- if (likely(frag_entry_curr)) {
hlist_add_after(&frag_entry_curr->list, &frag_entry_new->list);
chain->size += skb->len - hdr_size;
chain->timestamp = jiffies;
- }
+out:
- if (chain->size > batadv_frag_size_limit() ||
ntohs(frag_packet->total_size) > batadv_frag_size_limit()) {
/* Clear chain if total size of either the list or the packet
* exceeds the maximum size of one merged packet.
*/
batadv_frag_clear_chain(&chain->head);
chain->size = 0;
- } else if (ntohs(frag_packet->total_size) == chain->size) {
/* All fragments received. Hand over chain to caller. */
hlist_move_list(&chain->head, chain_out);
chain->size = 0;
- }
- spin_unlock_bh(&chain->lock);
- return true;
+err_unlock:
- spin_unlock_bh(&chain->lock);
- kfree(frag_entry_new);
+err:
- return false;
+}
[...]
+/*
- batadv_recv_frag_packet() - Process received fragment.
- @skb: The received fragment.
- @recv_if: Interface that the skb is received on.
- This function does one of the three following things: 1) Forward fragment, if
- the assembled packet will exceed our MTU; 2) Buffer fragment, if we till
- lack further fragments; 3) Merge fragments, if we have all needed parts.
- Returns NET_RX_DROP if the skb is not consumed, NET_RX_SUCCESS otherwise.
- */
+int batadv_recv_frag_packet(struct sk_buff *skb,
struct batadv_hard_iface *recv_if)
+{
- struct batadv_priv *bat_priv = netdev_priv(recv_if->soft_iface);
- struct batadv_orig_node *orig_node_src = NULL;
- struct batadv_frag_packet *frag_packet;
- int ret = NET_RX_DROP;
- if (batadv_check_unicast_packet(bat_priv, skb,
sizeof(*frag_packet)) < 0)
goto out;
- frag_packet = (struct batadv_frag_packet *)skb->data;
- orig_node_src = batadv_orig_hash_find(bat_priv, frag_packet->orig);
- if (!orig_node_src)
goto out;
- /* Route the fragment if it is not for us and too big to be merged. */
- if (!batadv_is_my_mac(bat_priv, frag_packet->dest) &&
batadv_frag_skb_fwd(skb, recv_if, orig_node_src)) {
ret = NET_RX_SUCCESS;
goto out;
- }
- batadv_inc_counter(bat_priv, BATADV_CNT_FRAG_RX);
- batadv_add_counter(bat_priv, BATADV_CNT_FRAG_RX_BYTES,
skb->len + ETH_HLEN);
why adding ETH_HLEN ? skb->len already counts the outer Ethernet header, no?
- /* Add fragment to buffer and merge if possible. */
- if (!batadv_frag_skb_buffer(&skb, orig_node_src))
goto out;
- /* Deliver merged packet to the appropriate handler, if it was
* merged
*/
- if (skb)
batadv_batman_skb_recv(skb, recv_if->net_dev,
&recv_if->batman_adv_ptype, NULL);
- ret = NET_RX_SUCCESS;
+out:
- if (orig_node_src)
batadv_orig_node_free_ref(orig_node_src);
- return ret;
+}
Cheers,
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 ---
v2: change commit message to explain the reverse order of packets remove parentheses and capital letters from kernel doc merge return statement for batadv_frag_create() add spaces around *-multiplication add kernel doc for FRAG_TX counters and frag_seqno
fragmentation.c | 119 +++++++++++++++++++++++++++++++++++++++++++++++++++++++ fragmentation.h | 3 ++ send.c | 21 ++++++++-- soft-interface.c | 7 ++++ types.h | 6 +++ 5 files changed, 152 insertions(+), 4 deletions(-)
diff --git a/fragmentation.c b/fragmentation.c index 5a891d7..ba02b3c 100644 --- a/fragmentation.c +++ b/fragmentation.c @@ -358,3 +358,122 @@ 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 - ETH_HLEN; + + skb_fragment = dev_alloc_skb(mtu); + if (!skb_fragment) + goto err; + + /* Eat the last mtu-bytes of the skb */ + skb_reserve(skb_fragment, header_size); + 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(int, 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/fragmentation.h b/fragmentation.h index 2af5df5..a457245 100644 --- a/fragmentation.h +++ b/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/send.c b/send.c index e6852a1..4a35139 100644 --- a/send.c +++ b/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);
@@ -110,7 +109,19 @@ 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 + ETH_HLEN > 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 @@ -124,7 +135,9 @@ 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/soft-interface.c b/soft-interface.c index 481642e..e157463 100644 --- a/soft-interface.c +++ b/soft-interface.c @@ -431,6 +431,7 @@ 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;
@@ -480,6 +481,10 @@ 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;
@@ -747,6 +752,8 @@ 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/types.h b/types.h index dd3f505..525e014 100644 --- a/types.h +++ b/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;
Hi³, :)
On Mon, May 13, 2013 at 10:23:02PM +0200, Martin Hundebøll wrote:
[..]
+/**
- 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 - ETH_HLEN;
- skb_fragment = dev_alloc_skb(mtu);
If I am not wrong, the external Ethernet header is not counted in the MTU of an interface. Therefore if the MTU of eth0 is X, it should mean that eth0 will send a packet of 1500 of payload + the external Ethernet encapsulation. isn't it?
- if (!skb_fragment)
goto err;
- /* Eat the last mtu-bytes of the skb */
- skb_reserve(skb_fragment, header_size);
why not reserving header_size + ETH_HLEN (ETH_HLEN would be the space for the external Ethernet header).
- 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;
+}
[...]
@@ -110,7 +109,19 @@ 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 + ETH_HLEN > neigh_node->if_incoming->net_dev->mtu) {
same as before: the external Ethernet header should not be taken into consideration while checking the MTU (no?).
My suggestion is to do:
"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
Thank you for your work Martin! :)
Cheers,
b.a.t.m.a.n@lists.open-mesh.org