Debian: Update change log for 1.1.0~pre2.g2.ea763e0e-1 upload
[openvswitch] / datapath / flow.c
index eae703d0d4d2479cf02f27f27fb89d9a2373a889..fb339e73e581a6379a726c14fc0ca592b45ab92b 100644 (file)
 
 #include "compat.h"
 
-struct kmem_cache *flow_cache;
-static unsigned int hash_seed;
+static struct kmem_cache *flow_cache;
+static unsigned int hash_seed __read_mostly;
 
 static inline bool arphdr_ok(struct sk_buff *skb)
 {
-       int nh_ofs = skb_network_offset(skb);
-       return pskb_may_pull(skb, nh_ofs + sizeof(struct arp_eth_header));
+       return skb->len >= skb_network_offset(skb) + sizeof(struct arp_eth_header);
 }
 
 static inline int check_iphdr(struct sk_buff *skb)
@@ -66,7 +65,7 @@ static inline int check_iphdr(struct sk_buff *skb)
 static inline bool tcphdr_ok(struct sk_buff *skb)
 {
        int th_ofs = skb_transport_offset(skb);
-       if (pskb_may_pull(skb, th_ofs + sizeof(struct tcphdr))) {
+       if (skb->len >= th_ofs + sizeof(struct tcphdr)) {
                int tcp_len = tcp_hdrlen(skb);
                return (tcp_len >= sizeof(struct tcphdr)
                        && skb->len >= th_ofs + tcp_len);
@@ -76,14 +75,12 @@ static inline bool tcphdr_ok(struct sk_buff *skb)
 
 static inline bool udphdr_ok(struct sk_buff *skb)
 {
-       int th_ofs = skb_transport_offset(skb);
-       return pskb_may_pull(skb, th_ofs + sizeof(struct udphdr));
+       return skb->len >= skb_transport_offset(skb) + sizeof(struct udphdr);
 }
 
 static inline bool icmphdr_ok(struct sk_buff *skb)
 {
-       int th_ofs = skb_transport_offset(skb);
-       return pskb_may_pull(skb, th_ofs + sizeof(struct icmphdr));
+       return skb->len >= skb_transport_offset(skb) + sizeof(struct icmphdr);
 }
 
 #define TCP_FLAGS_OFFSET 13
@@ -107,43 +104,57 @@ void flow_used(struct sw_flow *flow, struct sk_buff *skb)
        spin_unlock_bh(&flow->lock);
 }
 
-struct sw_flow_actions *flow_actions_alloc(size_t n_actions)
+struct sw_flow_actions *flow_actions_alloc(u32 actions_len)
 {
        struct sw_flow_actions *sfa;
 
-       if (n_actions > (PAGE_SIZE - sizeof *sfa) / sizeof(union odp_action))
+       if (actions_len % NLA_ALIGNTO)
+               return ERR_PTR(-EINVAL);
+
+       /* At least DP_MAX_PORTS actions are required to be able to flood a
+        * packet to every port.  Factor of 2 allows for setting VLAN tags,
+        * etc. */
+       if (actions_len > 2 * DP_MAX_PORTS * nla_total_size(4))
                return ERR_PTR(-EINVAL);
 
-       sfa = kmalloc(sizeof *sfa + n_actions * sizeof(union odp_action),
-                     GFP_KERNEL);
+       sfa = kmalloc(sizeof *sfa + actions_len, GFP_KERNEL);
        if (!sfa)
                return ERR_PTR(-ENOMEM);
 
-       sfa->n_actions = n_actions;
+       sfa->actions_len = actions_len;
        return sfa;
 }
 
-
-/* Frees 'flow' immediately. */
-static void flow_free(struct sw_flow *flow)
+struct sw_flow *flow_alloc(void)
 {
-       if (unlikely(!flow))
-               return;
-       kfree(flow->sf_acts);
-       kmem_cache_free(flow_cache, flow);
+       struct sw_flow *flow;
+
+       flow = kmem_cache_alloc(flow_cache, GFP_KERNEL);
+       if (!flow)
+               return ERR_PTR(-ENOMEM);
+
+       spin_lock_init(&flow->lock);
+       atomic_set(&flow->refcnt, 1);
+       flow->dead = false;
+
+       return flow;
 }
 
 void flow_free_tbl(struct tbl_node *node)
 {
        struct sw_flow *flow = flow_cast(node);
-       flow_free(flow);
+
+       flow->dead = true;
+       flow_put(flow);
 }
 
 /* RCU callback used by flow_deferred_free. */
 static void rcu_free_flow_callback(struct rcu_head *rcu)
 {
        struct sw_flow *flow = container_of(rcu, struct sw_flow, rcu);
-       flow_free(flow);
+
+       flow->dead = true;
+       flow_put(flow);
 }
 
 /* Schedules 'flow' to be freed after the next RCU grace period.
@@ -153,10 +164,26 @@ void flow_deferred_free(struct sw_flow *flow)
        call_rcu(&flow->rcu, rcu_free_flow_callback);
 }
 
+void flow_hold(struct sw_flow *flow)
+{
+       atomic_inc(&flow->refcnt);
+}
+
+void flow_put(struct sw_flow *flow)
+{
+       if (unlikely(!flow))
+               return;
+
+       if (atomic_dec_and_test(&flow->refcnt)) {
+               kfree((struct sf_flow_acts __force *)flow->sf_acts);
+               kmem_cache_free(flow_cache, flow);
+       }
+}
+
 /* RCU callback used by flow_deferred_free_acts. */
 static void rcu_free_acts_callback(struct rcu_head *rcu)
 {
-       struct sw_flow_actions *sf_acts = container_of(rcu, 
+       struct sw_flow_actions *sf_acts = container_of(rcu,
                        struct sw_flow_actions, rcu);
        kfree(sf_acts);
 }
@@ -180,8 +207,7 @@ static void parse_vlan(struct sk_buff *skb, struct odp_flow_key *key)
                return;
 
        qp = (struct qtag_prefix *) skb->data;
-       key->dl_vlan = qp->tci & htons(VLAN_VID_MASK);
-       key->dl_vlan_pcp = (ntohs(qp->tci) & VLAN_PCP_MASK) >> VLAN_PCP_SHIFT;
+       key->dl_tci = qp->tci | htons(ODP_TCI_PRESENT);
        __skb_pull(skb, sizeof(struct qtag_prefix));
 }
 
@@ -192,7 +218,7 @@ static __be16 parse_ethertype(struct sk_buff *skb)
                u8  ssap;  /* Always 0xAA */
                u8  ctrl;
                u8  oui[3];
-               u16 ethertype;
+               __be16 ethertype;
        };
        struct llc_snap_hdr *llc;
        __be16 proto;
@@ -222,23 +248,34 @@ static __be16 parse_ethertype(struct sk_buff *skb)
  * Ethernet header
  * @in_port: port number on which @skb was received.
  * @key: output flow key
+ * @is_frag: set to 1 if @skb contains an IPv4 fragment, or to 0 if @skb does
+ * not contain an IPv4 packet or if it is not a fragment.
  *
  * The caller must ensure that skb->len >= ETH_HLEN.
  *
  * Returns 0 if successful, otherwise a negative errno value.
  *
- * Sets OVS_CB(skb)->is_frag to %true if @skb is an IPv4 fragment, otherwise to
- * %false.
+ * Initializes @skb header pointers as follows:
+ *
+ *    - skb->mac_header: the Ethernet header.
+ *
+ *    - skb->network_header: just past the Ethernet header, or just past the
+ *      VLAN header, to the first byte of the Ethernet payload.
+ *
+ *    - skb->transport_header: If key->dl_type is ETH_P_IP on output, then just
+ *      past the IPv4 header, if one is present and of a correct length,
+ *      otherwise the same as skb->network_header.  For other key->dl_type
+ *      values it is left untouched.
  */
-int flow_extract(struct sk_buff *skb, u16 in_port, struct odp_flow_key *key)
+int flow_extract(struct sk_buff *skb, u16 in_port, struct odp_flow_key *key,
+                bool *is_frag)
 {
        struct ethhdr *eth;
 
        memset(key, 0, sizeof *key);
        key->tun_id = OVS_CB(skb)->tun_id;
        key->in_port = in_port;
-       key->dl_vlan = htons(ODP_VLAN_NONE);
-       OVS_CB(skb)->is_frag = false;
+       *is_frag = false;
 
        /*
         * We would really like to pull as many bytes as we could possibly
@@ -296,28 +333,19 @@ int flow_extract(struct sk_buff *skb, u16 in_port, struct odp_flow_key *key)
                key->nw_proto = nh->protocol;
 
                /* Transport layer. */
-               if (!(nh->frag_off & htons(IP_MF | IP_OFFSET))) {
+               if (!(nh->frag_off & htons(IP_MF | IP_OFFSET)) &&
+                   !(skb_shinfo(skb)->gso_type & SKB_GSO_UDP)) {
                        if (key->nw_proto == IPPROTO_TCP) {
                                if (tcphdr_ok(skb)) {
                                        struct tcphdr *tcp = tcp_hdr(skb);
                                        key->tp_src = tcp->source;
                                        key->tp_dst = tcp->dest;
-                               } else {
-                                       /* Avoid tricking other code into
-                                        * thinking that this packet has an L4
-                                        * header. */
-                                       key->nw_proto = 0;
                                }
                        } else if (key->nw_proto == IPPROTO_UDP) {
                                if (udphdr_ok(skb)) {
                                        struct udphdr *udp = udp_hdr(skb);
                                        key->tp_src = udp->source;
                                        key->tp_dst = udp->dest;
-                               } else {
-                                       /* Avoid tricking other code into
-                                        * thinking that this packet has an L4
-                                        * header. */
-                                       key->nw_proto = 0;
                                }
                        } else if (key->nw_proto == IPPROTO_ICMP) {
                                if (icmphdr_ok(skb)) {
@@ -327,16 +355,11 @@ int flow_extract(struct sk_buff *skb, u16 in_port, struct odp_flow_key *key)
                                         * in 16-bit network byte order. */
                                        key->tp_src = htons(icmp->type);
                                        key->tp_dst = htons(icmp->code);
-                               } else {
-                                       /* Avoid tricking other code into
-                                        * thinking that this packet has an L4
-                                        * header. */
-                                       key->nw_proto = 0;
                                }
                        }
-               } else {
-                       OVS_CB(skb)->is_frag = true;
-               }
+               } else
+                       *is_frag = true;
+
        } else if (key->dl_type == htons(ETH_P_ARP) && arphdr_ok(skb)) {
                struct arp_eth_header *arp;
 
@@ -348,18 +371,15 @@ int flow_extract(struct sk_buff *skb, u16 in_port, struct odp_flow_key *key)
                                && arp->ar_pln == 4) {
 
                        /* We only match on the lower 8 bits of the opcode. */
-                       if (ntohs(arp->ar_op) <= 0xff) {
+                       if (ntohs(arp->ar_op) <= 0xff)
                                key->nw_proto = ntohs(arp->ar_op);
-                       }
 
-                       if (key->nw_proto == ARPOP_REQUEST 
+                       if (key->nw_proto == ARPOP_REQUEST
                                        || key->nw_proto == ARPOP_REPLY) {
                                memcpy(&key->nw_src, arp->ar_sip, sizeof(key->nw_src));
                                memcpy(&key->nw_dst, arp->ar_tip, sizeof(key->nw_dst));
                        }
                }
-       } else {
-               skb_reset_transport_header(skb);
        }
        return 0;
 }