X-Git-Url: http://git.onelab.eu/?a=blobdiff_plain;f=datapath%2Fflow.c;h=fad9e1977df61827a9ed052d5eb32ad3f372be3a;hb=eb857b4824d2a62e1cc1b85c30a3da007d4942c9;hp=d07337c83109c30428b1fc24dcd7917288127c0d;hpb=1e3f34c7693bcabae8e443ac1b246680ef9b60e2;p=sliver-openvswitch.git diff --git a/datapath/flow.c b/datapath/flow.c index d07337c83..fad9e1977 100644 --- a/datapath/flow.c +++ b/datapath/flow.c @@ -200,20 +200,18 @@ void ovs_flow_used(struct sw_flow *flow, struct sk_buff *skb) spin_unlock(&flow->lock); } -struct sw_flow_actions *ovs_flow_actions_alloc(const struct nlattr *actions) +struct sw_flow_actions *ovs_flow_actions_alloc(int size) { - int actions_len = nla_len(actions); struct sw_flow_actions *sfa; - if (actions_len > MAX_ACTIONS_BUFSIZE) + if (size > MAX_ACTIONS_BUFSIZE) return ERR_PTR(-EINVAL); - sfa = kmalloc(sizeof(*sfa) + actions_len, GFP_KERNEL); + sfa = kmalloc(sizeof(*sfa) + size, GFP_KERNEL); if (!sfa) return ERR_PTR(-ENOMEM); - sfa->actions_len = actions_len; - memcpy(sfa->actions, nla_data(actions), actions_len); + sfa->actions_len = 0; return sfa; } @@ -226,9 +224,7 @@ struct sw_flow *ovs_flow_alloc(void) return ERR_PTR(-ENOMEM); spin_lock_init(&flow->lock); - atomic_set(&flow->refcnt, 1); flow->sf_acts = NULL; - flow->dead = false; return flow; } @@ -290,12 +286,6 @@ struct flow_table *ovs_flow_tbl_alloc(int new_size) return table; } -static void flow_free(struct sw_flow *flow) -{ - flow->dead = true; - ovs_flow_put(flow); -} - void ovs_flow_tbl_destroy(struct flow_table *table) { int i; @@ -314,7 +304,7 @@ void ovs_flow_tbl_destroy(struct flow_table *table) hlist_for_each_entry_safe(flow, node, n, head, hash_node[ver]) { hlist_del_rcu(&flow->hash_node[ver]); - flow_free(flow); + ovs_flow_free(flow); } } @@ -365,6 +355,14 @@ struct sw_flow *ovs_flow_tbl_next(struct flow_table *table, u32 *bucket, u32 *la return NULL; } +static void __flow_tbl_insert(struct flow_table *table, struct sw_flow *flow) +{ + struct hlist_head *head; + head = find_bucket(table, flow->hash); + hlist_add_head_rcu(&flow->hash_node[table->node_ver], head); + table->count++; +} + static void flow_table_copy_flows(struct flow_table *old, struct flow_table *new) { int old_ver; @@ -382,7 +380,7 @@ static void flow_table_copy_flows(struct flow_table *old, struct flow_table *new head = flex_array_get(old->buckets, i); hlist_for_each_entry(flow, n, head, hash_node[old_ver]) - ovs_flow_tbl_insert(new, flow); + __flow_tbl_insert(new, flow); } old->keep_flows = true; } @@ -410,13 +408,21 @@ struct flow_table *ovs_flow_tbl_expand(struct flow_table *table) return __flow_tbl_rehash(table, table->n_buckets * 2); } +void ovs_flow_free(struct sw_flow *flow) +{ + if (unlikely(!flow)) + return; + + kfree((struct sf_flow_acts __force *)flow->sf_acts); + kmem_cache_free(flow_cache, flow); +} + /* RCU callback used by ovs_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->dead = true; - ovs_flow_put(flow); + ovs_flow_free(flow); } /* Schedules 'flow' to be freed after the next RCU grace period. @@ -426,22 +432,6 @@ void ovs_flow_deferred_free(struct sw_flow *flow) call_rcu(&flow->rcu, rcu_free_flow_callback); } -void ovs_flow_hold(struct sw_flow *flow) -{ - atomic_inc(&flow->refcnt); -} - -void ovs_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 ovs_flow_deferred_free_acts. */ static void rcu_free_acts_callback(struct rcu_head *rcu) { @@ -629,8 +619,10 @@ int ovs_flow_extract(struct sk_buff *skb, u16 in_port, struct sw_flow_key *key, memset(key, 0, sizeof(*key)); key->phy.priority = skb->priority; - key->phy.tun_id = OVS_CB(skb)->tun_id; + if (OVS_CB(skb)->tun_key) + memcpy(&key->tun_key, OVS_CB(skb)->tun_key, sizeof(key->tun_key)); key->phy.in_port = in_port; + key->phy.skb_mark = skb_get_mark(skb); skb_reset_mac_header(skb); @@ -716,7 +708,8 @@ int ovs_flow_extract(struct sk_buff *skb, u16 in_port, struct sw_flow_key *key, } } - } else if (key->eth.type == htons(ETH_P_ARP) && arphdr_ok(skb)) { + } else if ((key->eth.type == htons(ETH_P_ARP) || + key->eth.type == htons(ETH_P_RARP)) && arphdr_ok(skb)) { struct arp_eth_header *arp; arp = (struct arp_eth_header *)skb_network_header(skb); @@ -729,15 +722,11 @@ int ovs_flow_extract(struct sk_buff *skb, u16 in_port, struct sw_flow_key *key, /* We only match on the lower 8 bits of the opcode. */ if (ntohs(arp->ar_op) <= 0xff) key->ip.proto = ntohs(arp->ar_op); - - if (key->ip.proto == ARPOP_REQUEST - || key->ip.proto == ARPOP_REPLY) { - memcpy(&key->ipv4.addr.src, arp->ar_sip, sizeof(key->ipv4.addr.src)); - memcpy(&key->ipv4.addr.dst, arp->ar_tip, sizeof(key->ipv4.addr.dst)); - memcpy(key->ipv4.arp.sha, arp->ar_sha, ETH_ALEN); - memcpy(key->ipv4.arp.tha, arp->ar_tha, ETH_ALEN); - key_len = SW_FLOW_KEY_OFFSET(ipv4.arp); - } + memcpy(&key->ipv4.addr.src, arp->ar_sip, sizeof(key->ipv4.addr.src)); + memcpy(&key->ipv4.addr.dst, arp->ar_tip, sizeof(key->ipv4.addr.dst)); + memcpy(key->ipv4.arp.sha, arp->ar_sha, ETH_ALEN); + memcpy(key->ipv4.arp.tha, arp->ar_tha, ETH_ALEN); + key_len = SW_FLOW_KEY_OFFSET(ipv4.arp); } } else if (key->eth.type == htons(ETH_P_IPV6)) { int nh_len; /* IPv6 Header + Extensions */ @@ -786,9 +775,18 @@ out: return error; } -u32 ovs_flow_hash(const struct sw_flow_key *key, int key_len) +static u32 ovs_flow_hash(const struct sw_flow_key *key, int key_start, int key_len) { - return jhash2((u32 *)key, DIV_ROUND_UP(key_len, sizeof(u32)), 0); + return jhash2((u32 *)((u8 *)key + key_start), + DIV_ROUND_UP(key_len - key_start, sizeof(u32)), 0); +} + +static int flow_key_start(struct sw_flow_key *key) +{ + if (key->tun_key.ipv4_dst) + return 0; + else + return offsetof(struct sw_flow_key, phy); } struct sw_flow *ovs_flow_tbl_lookup(struct flow_table *table, @@ -797,28 +795,31 @@ struct sw_flow *ovs_flow_tbl_lookup(struct flow_table *table, struct sw_flow *flow; struct hlist_node *n; struct hlist_head *head; + u8 *_key; + int key_start; u32 hash; - hash = ovs_flow_hash(key, key_len); + key_start = flow_key_start(key); + hash = ovs_flow_hash(key, key_start, key_len); + _key = (u8 *) key + key_start; head = find_bucket(table, hash); hlist_for_each_entry_rcu(flow, n, head, hash_node[table->node_ver]) { if (flow->hash == hash && - !memcmp(&flow->key, key, key_len)) { + !memcmp((u8 *)&flow->key + key_start, _key, key_len - key_start)) { return flow; } } return NULL; } -void ovs_flow_tbl_insert(struct flow_table *table, struct sw_flow *flow) +void ovs_flow_tbl_insert(struct flow_table *table, struct sw_flow *flow, + struct sw_flow_key *key, int key_len) { - struct hlist_head *head; - - head = find_bucket(table, flow->hash); - hlist_add_head_rcu(&flow->hash_node[table->node_ver], head); - table->count++; + flow->hash = ovs_flow_hash(key, flow_key_start(key), key_len); + memcpy(&flow->key, key, sizeof(flow->key)); + __flow_tbl_insert(table, flow); } void ovs_flow_tbl_remove(struct flow_table *table, struct sw_flow *flow) @@ -833,6 +834,7 @@ const int ovs_key_lens[OVS_KEY_ATTR_MAX + 1] = { [OVS_KEY_ATTR_ENCAP] = -1, [OVS_KEY_ATTR_PRIORITY] = sizeof(u32), [OVS_KEY_ATTR_IN_PORT] = sizeof(u32), + [OVS_KEY_ATTR_SKB_MARK] = sizeof(u32), [OVS_KEY_ATTR_ETHERNET] = sizeof(struct ovs_key_ethernet), [OVS_KEY_ATTR_VLAN] = sizeof(__be16), [OVS_KEY_ATTR_ETHERTYPE] = sizeof(__be16), @@ -844,6 +846,7 @@ const int ovs_key_lens[OVS_KEY_ATTR_MAX + 1] = { [OVS_KEY_ATTR_ICMPV6] = sizeof(struct ovs_key_icmpv6), [OVS_KEY_ATTR_ARP] = sizeof(struct ovs_key_arp), [OVS_KEY_ATTR_ND] = sizeof(struct ovs_key_nd), + [OVS_KEY_ATTR_TUNNEL] = -1, /* Not upstream. */ [OVS_KEY_ATTR_TUN_ID] = sizeof(__be64), @@ -984,6 +987,105 @@ static int parse_flow_nlattrs(const struct nlattr *attr, return 0; } +int ipv4_tun_from_nlattr(const struct nlattr *attr, + struct ovs_key_ipv4_tunnel *tun_key) +{ + struct nlattr *a; + int rem; + bool ttl = false; + + memset(tun_key, 0, sizeof(*tun_key)); + + nla_for_each_nested(a, attr, rem) { + int type = nla_type(a); + static const u32 ovs_tunnel_key_lens[OVS_TUNNEL_KEY_ATTR_MAX + 1] = { + [OVS_TUNNEL_KEY_ATTR_ID] = sizeof(u64), + [OVS_TUNNEL_KEY_ATTR_IPV4_SRC] = sizeof(u32), + [OVS_TUNNEL_KEY_ATTR_IPV4_DST] = sizeof(u32), + [OVS_TUNNEL_KEY_ATTR_TOS] = 1, + [OVS_TUNNEL_KEY_ATTR_TTL] = 1, + [OVS_TUNNEL_KEY_ATTR_DONT_FRAGMENT] = 0, + [OVS_TUNNEL_KEY_ATTR_CSUM] = 0, + }; + + if (type > OVS_TUNNEL_KEY_ATTR_MAX || + ovs_tunnel_key_lens[type] != nla_len(a)) + return -EINVAL; + + switch (type) { + case OVS_TUNNEL_KEY_ATTR_ID: + tun_key->tun_id = nla_get_be64(a); + tun_key->tun_flags |= OVS_TNL_F_KEY; + break; + case OVS_TUNNEL_KEY_ATTR_IPV4_SRC: + tun_key->ipv4_src = nla_get_be32(a); + break; + case OVS_TUNNEL_KEY_ATTR_IPV4_DST: + tun_key->ipv4_dst = nla_get_be32(a); + break; + case OVS_TUNNEL_KEY_ATTR_TOS: + tun_key->ipv4_tos = nla_get_u8(a); + break; + case OVS_TUNNEL_KEY_ATTR_TTL: + tun_key->ipv4_ttl = nla_get_u8(a); + ttl = true; + break; + case OVS_TUNNEL_KEY_ATTR_DONT_FRAGMENT: + tun_key->tun_flags |= OVS_TNL_F_DONT_FRAGMENT; + break; + case OVS_TUNNEL_KEY_ATTR_CSUM: + tun_key->tun_flags |= OVS_TNL_F_CSUM; + break; + default: + return -EINVAL; + + } + } + if (rem > 0) + return -EINVAL; + + if (!tun_key->ipv4_dst) + return -EINVAL; + + if (!ttl) + return -EINVAL; + + return 0; +} + +int ipv4_tun_to_nlattr(struct sk_buff *skb, + const struct ovs_key_ipv4_tunnel *tun_key) +{ + struct nlattr *nla; + + nla = nla_nest_start(skb, OVS_KEY_ATTR_TUNNEL); + if (!nla) + return -EMSGSIZE; + + if (tun_key->tun_flags & OVS_TNL_F_KEY && + nla_put_be64(skb, OVS_TUNNEL_KEY_ATTR_ID, tun_key->tun_id)) + return -EMSGSIZE; + if (tun_key->ipv4_src && + nla_put_be32(skb, OVS_TUNNEL_KEY_ATTR_IPV4_SRC, tun_key->ipv4_src)) + return -EMSGSIZE; + if (nla_put_be32(skb, OVS_TUNNEL_KEY_ATTR_IPV4_DST, tun_key->ipv4_dst)) + return -EMSGSIZE; + if (tun_key->ipv4_tos && + nla_put_u8(skb, OVS_TUNNEL_KEY_ATTR_TOS, tun_key->ipv4_tos)) + return -EMSGSIZE; + if (nla_put_u8(skb, OVS_TUNNEL_KEY_ATTR_TTL, tun_key->ipv4_ttl)) + return -EMSGSIZE; + if ((tun_key->tun_flags & OVS_TNL_F_DONT_FRAGMENT) && + nla_put_flag(skb, OVS_TUNNEL_KEY_ATTR_DONT_FRAGMENT)) + return -EMSGSIZE; + if ((tun_key->tun_flags & OVS_TNL_F_CSUM) && + nla_put_flag(skb, OVS_TUNNEL_KEY_ATTR_CSUM)) + return -EMSGSIZE; + + nla_nest_end(skb, nla); + return 0; +} + /** * ovs_flow_from_nlattrs - parses Netlink attributes into a flow key. * @swkey: receives the extracted flow key. @@ -1021,10 +1123,40 @@ int ovs_flow_from_nlattrs(struct sw_flow_key *swkey, int *key_lenp, } else { swkey->phy.in_port = DP_MAX_PORTS; } + if (attrs & (1 << OVS_KEY_ATTR_SKB_MARK)) { + uint32_t mark = nla_get_u32(a[OVS_KEY_ATTR_SKB_MARK]); +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,20) && !defined(CONFIG_NETFILTER) + if (mark != 0) + return -EINVAL; +#endif + swkey->phy.skb_mark = mark; + attrs &= ~(1 << OVS_KEY_ATTR_SKB_MARK); + } + + if (attrs & (1ULL << OVS_KEY_ATTR_TUN_ID) && + attrs & (1ULL << OVS_KEY_ATTR_TUNNEL)) { + __be64 tun_id; + + err = ipv4_tun_from_nlattr(a[OVS_KEY_ATTR_TUNNEL], &swkey->tun_key); + if (err) + return err; + + if (!(swkey->tun_key.tun_flags & OVS_TNL_F_KEY)) + return -EINVAL; + + tun_id = nla_get_be64(a[OVS_KEY_ATTR_TUN_ID]); + if (tun_id != swkey->tun_key.tun_id) + return -EINVAL; - if (attrs & (1ULL << OVS_KEY_ATTR_TUN_ID)) { - swkey->phy.tun_id = nla_get_be64(a[OVS_KEY_ATTR_TUN_ID]); attrs &= ~(1ULL << OVS_KEY_ATTR_TUN_ID); + attrs &= ~(1ULL << OVS_KEY_ATTR_TUNNEL); + } else if (attrs & (1ULL << OVS_KEY_ATTR_TUNNEL)) { + + err = ipv4_tun_from_nlattr(a[OVS_KEY_ATTR_TUNNEL], &swkey->tun_key); + if (err) + return err; + + attrs &= ~(1ULL << OVS_KEY_ATTR_TUNNEL); } /* Data attributes. */ @@ -1125,7 +1257,8 @@ int ovs_flow_from_nlattrs(struct sw_flow_key *swkey, int *key_lenp, if (err) return err; } - } else if (swkey->eth.type == htons(ETH_P_ARP)) { + } else if (swkey->eth.type == htons(ETH_P_ARP) || + swkey->eth.type == htons(ETH_P_RARP)) { const struct ovs_key_arp *arp_key; if (!(attrs & (1 << OVS_KEY_ATTR_ARP))) @@ -1162,42 +1295,89 @@ int ovs_flow_from_nlattrs(struct sw_flow_key *swkey, int *key_lenp, * get the metadata, that is, the parts of the flow key that cannot be * extracted from the packet itself. */ -int ovs_flow_metadata_from_nlattrs(u32 *priority, u16 *in_port, __be64 *tun_id, - const struct nlattr *attr) + +int ovs_flow_metadata_from_nlattrs(struct sw_flow *flow, int key_len, const struct nlattr *attr) { + struct ovs_key_ipv4_tunnel *tun_key = &flow->key.tun_key; const struct nlattr *nla; int rem; + __be64 tun_id = 0; - *in_port = DP_MAX_PORTS; - *tun_id = 0; - *priority = 0; + flow->key.phy.in_port = DP_MAX_PORTS; + flow->key.phy.priority = 0; + flow->key.phy.skb_mark = 0; + memset(tun_key, 0, sizeof(flow->key.tun_key)); nla_for_each_nested(nla, attr, rem) { int type = nla_type(nla); if (type <= OVS_KEY_ATTR_MAX && ovs_key_lens[type] > 0) { + int err; + if (nla_len(nla) != ovs_key_lens[type]) return -EINVAL; switch (type) { case OVS_KEY_ATTR_PRIORITY: - *priority = nla_get_u32(nla); + flow->key.phy.priority = nla_get_u32(nla); break; case OVS_KEY_ATTR_TUN_ID: - *tun_id = nla_get_be64(nla); + tun_id = nla_get_be64(nla); + + if (tun_key->ipv4_dst) { + if (!(tun_key->tun_flags & OVS_TNL_F_KEY)) + return -EINVAL; + if (tun_key->tun_id != tun_id) + return -EINVAL; + break; + } + tun_key->tun_id = tun_id; + tun_key->tun_flags |= OVS_TNL_F_KEY; + + break; + + case OVS_KEY_ATTR_TUNNEL: + if (tun_key->tun_flags & OVS_TNL_F_KEY) { + tun_id = tun_key->tun_id; + err = ipv4_tun_from_nlattr(nla, tun_key); + if (err) + return err; + + if (!(tun_key->tun_flags & OVS_TNL_F_KEY)) + return -EINVAL; + + if (tun_key->tun_id != tun_id) + return -EINVAL; + } else { + err = ipv4_tun_from_nlattr(nla, tun_key); + if (err) + return err; + } break; case OVS_KEY_ATTR_IN_PORT: if (nla_get_u32(nla) >= DP_MAX_PORTS) return -EINVAL; - *in_port = nla_get_u32(nla); + flow->key.phy.in_port = nla_get_u32(nla); + break; + + case OVS_KEY_ATTR_SKB_MARK: +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,20) && !defined(CONFIG_NETFILTER) + if (nla_get_u32(nla) != 0) + return -EINVAL; +#endif + flow->key.phy.skb_mark = nla_get_u32(nla); break; } } } if (rem) return -EINVAL; + + flow->hash = ovs_flow_hash(&flow->key, + flow_key_start(&flow->key), key_len); + return 0; } @@ -1210,14 +1390,22 @@ int ovs_flow_to_nlattrs(const struct sw_flow_key *swkey, struct sk_buff *skb) nla_put_u32(skb, OVS_KEY_ATTR_PRIORITY, swkey->phy.priority)) goto nla_put_failure; - if (swkey->phy.tun_id != cpu_to_be64(0) && - nla_put_be64(skb, OVS_KEY_ATTR_TUN_ID, swkey->phy.tun_id)) + if (swkey->tun_key.ipv4_dst && + ipv4_tun_to_nlattr(skb, &swkey->tun_key)) + goto nla_put_failure; + + if ((swkey->tun_key.tun_flags & OVS_TNL_F_KEY) && + nla_put_be64(skb, OVS_KEY_ATTR_TUN_ID, swkey->tun_key.tun_id)) goto nla_put_failure; if (swkey->phy.in_port != DP_MAX_PORTS && nla_put_u32(skb, OVS_KEY_ATTR_IN_PORT, swkey->phy.in_port)) goto nla_put_failure; + if (swkey->phy.skb_mark && + nla_put_u32(skb, OVS_KEY_ATTR_SKB_MARK, swkey->phy.skb_mark)) + goto nla_put_failure; + nla = nla_reserve(skb, OVS_KEY_ATTR_ETHERNET, sizeof(*eth_key)); if (!nla) goto nla_put_failure; @@ -1271,7 +1459,8 @@ int ovs_flow_to_nlattrs(const struct sw_flow_key *swkey, struct sk_buff *skb) ipv6_key->ipv6_tclass = swkey->ip.tos; ipv6_key->ipv6_hlimit = swkey->ip.ttl; ipv6_key->ipv6_frag = swkey->ip.frag; - } else if (swkey->eth.type == htons(ETH_P_ARP)) { + } else if (swkey->eth.type == htons(ETH_P_ARP) || + swkey->eth.type == htons(ETH_P_RARP)) { struct ovs_key_arp *arp_key; nla = nla_reserve(skb, OVS_KEY_ATTR_ARP, sizeof(*arp_key));