X-Git-Url: http://git.onelab.eu/?a=blobdiff_plain;f=datapath%2Fflow.c;h=c52081b2f7dfa57e67d04f9c449c36a1843a5f24;hb=003ce655b7116d18c86a74c50391e54990346931;hp=eb67cf4503ec08f84647792bfb0e8b60f17c5af6;hpb=ec58547a81246c0f3f7a03ab28ba1a78f1988235;p=sliver-openvswitch.git diff --git a/datapath/flow.c b/datapath/flow.c index eb67cf450..c52081b2f 100644 --- a/datapath/flow.c +++ b/datapath/flow.c @@ -1,14 +1,24 @@ /* - * Distributed under the terms of the GNU GPL version 2. - * Copyright (c) 2007, 2008, 2009, 2010, 2011 Nicira Networks. + * Copyright (c) 2007-2014 Nicira, Inc. * - * Significant portions of this file may be copied from parts of the Linux - * kernel, by Linus Torvalds and others. + * 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 "flow.h" #include "datapath.h" -#include +#include #include #include #include @@ -22,67 +32,22 @@ #include #include #include -#include #include +#include +#include +#include #include #include #include -#include +#include +#include #include +#include +#include -static struct kmem_cache *flow_cache; -static unsigned int hash_seed __read_mostly; - -static inline bool arphdr_ok(struct sk_buff *skb) -{ - return skb->len >= skb_network_offset(skb) + sizeof(struct arp_eth_header); -} - -static inline int check_iphdr(struct sk_buff *skb) -{ - unsigned int nh_ofs = skb_network_offset(skb); - unsigned int ip_len; - - if (skb->len < nh_ofs + sizeof(struct iphdr)) - return -EINVAL; - - ip_len = ip_hdrlen(skb); - if (ip_len < sizeof(struct iphdr) || skb->len < nh_ofs + ip_len) - return -EINVAL; - - /* - * Pull enough header bytes to account for the IP header plus the - * longest transport header that we parse, currently 20 bytes for TCP. - */ - if (!pskb_may_pull(skb, min(nh_ofs + ip_len + 20, skb->len))) - return -ENOMEM; - - skb_set_transport_header(skb, nh_ofs + ip_len); - return 0; -} +#include "vlan.h" -static inline bool tcphdr_ok(struct sk_buff *skb) -{ - int th_ofs = skb_transport_offset(skb); - 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); - } - return false; -} - -static inline bool udphdr_ok(struct sk_buff *skb) -{ - return skb->len >= skb_transport_offset(skb) + sizeof(struct udphdr); -} - -static inline bool icmphdr_ok(struct sk_buff *skb) -{ - return skb->len >= skb_transport_offset(skb) + sizeof(struct icmphdr); -} - -u64 flow_used_time(unsigned long flow_jiffies) +u64 ovs_flow_used_time(unsigned long flow_jiffies) { struct timespec cur_ts; u64 cur_ms, idle_ms; @@ -95,119 +60,234 @@ u64 flow_used_time(unsigned long flow_jiffies) return cur_ms - idle_ms; } +#define TCP_FLAGS_BE16(tp) (*(__be16 *)&tcp_flag_word(tp) & htons(0x0FFF)) -#define TCP_FLAGS_OFFSET 13 -#define TCP_FLAG_MASK 0x3f +void ovs_flow_stats_update(struct sw_flow *flow, __be16 tcp_flags, + struct sk_buff *skb) +{ + struct flow_stats *stats; + int node = numa_node_id(); + + stats = rcu_dereference(flow->stats[node]); + + /* Check if already have node-specific stats. */ + if (likely(stats)) { + spin_lock(&stats->lock); + /* Mark if we write on the pre-allocated stats. */ + if (node == 0 && unlikely(flow->stats_last_writer != node)) + flow->stats_last_writer = node; + } else { + stats = rcu_dereference(flow->stats[0]); /* Pre-allocated. */ + spin_lock(&stats->lock); + + /* If the current NUMA-node is the only writer on the + * pre-allocated stats keep using them. + */ + if (unlikely(flow->stats_last_writer != node)) { + /* A previous locker may have already allocated the + * stats, so we need to check again. If node-specific + * stats were already allocated, we update the pre- + * allocated stats as we have already locked them. + */ + if (likely(flow->stats_last_writer != NUMA_NO_NODE) + && likely(!rcu_dereference(flow->stats[node]))) { + /* Try to allocate node-specific stats. */ + struct flow_stats *new_stats; + + new_stats = + kmem_cache_alloc_node(flow_stats_cache, + GFP_THISNODE | + __GFP_NOMEMALLOC, + node); + if (likely(new_stats)) { + new_stats->used = jiffies; + new_stats->packet_count = 1; + new_stats->byte_count = skb->len; + new_stats->tcp_flags = tcp_flags; + spin_lock_init(&new_stats->lock); + + rcu_assign_pointer(flow->stats[node], + new_stats); + goto unlock; + } + } + flow->stats_last_writer = node; + } + } -void flow_used(struct sw_flow *flow, struct sk_buff *skb) + stats->used = jiffies; + stats->packet_count++; + stats->byte_count += skb->len; + stats->tcp_flags |= tcp_flags; +unlock: + spin_unlock(&stats->lock); +} + +/* Must be called with rcu_read_lock or ovs_mutex. */ +void ovs_flow_stats_get(const struct sw_flow *flow, + struct ovs_flow_stats *ovs_stats, + unsigned long *used, __be16 *tcp_flags) { - u8 tcp_flags = 0; + int node; + + *used = 0; + *tcp_flags = 0; + memset(ovs_stats, 0, sizeof(*ovs_stats)); + + for_each_node(node) { + struct flow_stats *stats = rcu_dereference_ovsl(flow->stats[node]); + + if (stats) { + /* Local CPU may write on non-local stats, so we must + * block bottom-halves here. + */ + spin_lock_bh(&stats->lock); + if (!*used || time_after(stats->used, *used)) + *used = stats->used; + *tcp_flags |= stats->tcp_flags; + ovs_stats->n_packets += stats->packet_count; + ovs_stats->n_bytes += stats->byte_count; + spin_unlock_bh(&stats->lock); + } + } +} - if (flow->key.dl_type == htons(ETH_P_IP) && - flow->key.nw_proto == IPPROTO_TCP) { - u8 *tcp = (u8 *)tcp_hdr(skb); - tcp_flags = *(tcp + TCP_FLAGS_OFFSET) & TCP_FLAG_MASK; +/* Called with ovs_mutex. */ +void ovs_flow_stats_clear(struct sw_flow *flow) +{ + int node; + + for_each_node(node) { + struct flow_stats *stats = ovsl_dereference(flow->stats[node]); + + if (stats) { + spin_lock_bh(&stats->lock); + stats->used = 0; + stats->packet_count = 0; + stats->byte_count = 0; + stats->tcp_flags = 0; + spin_unlock_bh(&stats->lock); + } } +} - spin_lock_bh(&flow->lock); - flow->used = jiffies; - flow->packet_count++; - flow->byte_count += skb->len; - flow->tcp_flags |= tcp_flags; - spin_unlock_bh(&flow->lock); +static int check_header(struct sk_buff *skb, int len) +{ + if (unlikely(skb->len < len)) + return -EINVAL; + if (unlikely(!pskb_may_pull(skb, len))) + return -ENOMEM; + return 0; } -struct sw_flow_actions *flow_actions_alloc(const struct nlattr *actions) +static bool arphdr_ok(struct sk_buff *skb) { - int actions_len = nla_len(actions); - struct sw_flow_actions *sfa; - - /* 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) + actions_len, GFP_KERNEL); - if (!sfa) - return ERR_PTR(-ENOMEM); - - sfa->actions_len = actions_len; - memcpy(sfa->actions, nla_data(actions), actions_len); - return sfa; + return pskb_may_pull(skb, skb_network_offset(skb) + + sizeof(struct arp_eth_header)); } -struct sw_flow *flow_alloc(void) +static int check_iphdr(struct sk_buff *skb) { - struct sw_flow *flow; + unsigned int nh_ofs = skb_network_offset(skb); + unsigned int ip_len; + int err; - flow = kmem_cache_alloc(flow_cache, GFP_KERNEL); - if (!flow) - return ERR_PTR(-ENOMEM); + err = check_header(skb, nh_ofs + sizeof(struct iphdr)); + if (unlikely(err)) + return err; - spin_lock_init(&flow->lock); - atomic_set(&flow->refcnt, 1); - flow->dead = false; + ip_len = ip_hdrlen(skb); + if (unlikely(ip_len < sizeof(struct iphdr) || + skb->len < nh_ofs + ip_len)) + return -EINVAL; - return flow; + skb_set_transport_header(skb, nh_ofs + ip_len); + return 0; } -void flow_free_tbl(struct tbl_node *node) +static bool tcphdr_ok(struct sk_buff *skb) { - struct sw_flow *flow = flow_cast(node); + int th_ofs = skb_transport_offset(skb); + int tcp_len; + + if (unlikely(!pskb_may_pull(skb, th_ofs + sizeof(struct tcphdr)))) + return false; + + tcp_len = tcp_hdrlen(skb); + if (unlikely(tcp_len < sizeof(struct tcphdr) || + skb->len < th_ofs + tcp_len)) + return false; - flow->dead = true; - flow_put(flow); + return true; } -/* RCU callback used by flow_deferred_free. */ -static void rcu_free_flow_callback(struct rcu_head *rcu) +static bool udphdr_ok(struct sk_buff *skb) { - struct sw_flow *flow = container_of(rcu, struct sw_flow, rcu); - - flow->dead = true; - flow_put(flow); + return pskb_may_pull(skb, skb_transport_offset(skb) + + sizeof(struct udphdr)); } -/* Schedules 'flow' to be freed after the next RCU grace period. - * The caller must hold rcu_read_lock for this to be sensible. */ -void flow_deferred_free(struct sw_flow *flow) +static bool sctphdr_ok(struct sk_buff *skb) { - call_rcu(&flow->rcu, rcu_free_flow_callback); + return pskb_may_pull(skb, skb_transport_offset(skb) + + sizeof(struct sctphdr)); } -void flow_hold(struct sw_flow *flow) +static bool icmphdr_ok(struct sk_buff *skb) { - atomic_inc(&flow->refcnt); + return pskb_may_pull(skb, skb_transport_offset(skb) + + sizeof(struct icmphdr)); } -void flow_put(struct sw_flow *flow) +static int parse_ipv6hdr(struct sk_buff *skb, struct sw_flow_key *key) { - if (unlikely(!flow)) - return; + unsigned int nh_ofs = skb_network_offset(skb); + unsigned int nh_len; + int payload_ofs; + struct ipv6hdr *nh; + uint8_t nexthdr; + __be16 frag_off; + int err; + + err = check_header(skb, nh_ofs + sizeof(*nh)); + if (unlikely(err)) + return err; + + nh = ipv6_hdr(skb); + nexthdr = nh->nexthdr; + payload_ofs = (u8 *)(nh + 1) - skb->data; + + key->ip.proto = NEXTHDR_NONE; + key->ip.tos = ipv6_get_dsfield(nh); + key->ip.ttl = nh->hop_limit; + key->ipv6.label = *(__be32 *)nh & htonl(IPV6_FLOWINFO_FLOWLABEL); + key->ipv6.addr.src = nh->saddr; + key->ipv6.addr.dst = nh->daddr; + + payload_ofs = ipv6_skip_exthdr(skb, payload_ofs, &nexthdr, &frag_off); + if (unlikely(payload_ofs < 0)) + return -EINVAL; - if (atomic_dec_and_test(&flow->refcnt)) { - kfree((struct sf_flow_acts __force *)flow->sf_acts); - kmem_cache_free(flow_cache, flow); + if (frag_off) { + if (frag_off & htons(~0x7)) + key->ip.frag = OVS_FRAG_TYPE_LATER; + else + key->ip.frag = OVS_FRAG_TYPE_FIRST; } -} -/* 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, rcu); - kfree(sf_acts); + nh_len = payload_ofs - nh_ofs; + skb_set_transport_header(skb, nh_ofs + nh_len); + key->ip.proto = nexthdr; + return nh_len; } -/* Schedules 'sf_acts' to be freed after the next RCU grace period. - * The caller must hold rcu_read_lock for this to be sensible. */ -void flow_deferred_free_acts(struct sw_flow_actions *sf_acts) +static bool icmp6hdr_ok(struct sk_buff *skb) { - call_rcu(&sf_acts->rcu, rcu_free_acts_callback); + return pskb_may_pull(skb, skb_transport_offset(skb) + + sizeof(struct icmp6hdr)); } -static void parse_vlan(struct sk_buff *skb, struct sw_flow_key *key) +static int parse_vlan(struct sk_buff *skb, struct sw_flow_key *key) { struct qtag_prefix { __be16 eth_type; /* ETH_P_8021Q */ @@ -215,12 +295,18 @@ static void parse_vlan(struct sk_buff *skb, struct sw_flow_key *key) }; struct qtag_prefix *qp; - if (skb->len < sizeof(struct qtag_prefix) + sizeof(__be16)) - return; + if (unlikely(skb->len < sizeof(struct qtag_prefix) + sizeof(__be16))) + return 0; + + if (unlikely(!pskb_may_pull(skb, sizeof(struct qtag_prefix) + + sizeof(__be16)))) + return -ENOMEM; qp = (struct qtag_prefix *) skb->data; - key->dl_tci = qp->tci | htons(VLAN_TAG_PRESENT); + key->eth.tci = qp->tci | htons(VLAN_TAG_PRESENT); __skb_pull(skb, sizeof(struct qtag_prefix)); + + return 0; } static __be16 parse_ethertype(struct sk_buff *skb) @@ -238,12 +324,15 @@ static __be16 parse_ethertype(struct sk_buff *skb) proto = *(__be16 *) skb->data; __skb_pull(skb, sizeof(__be16)); - if (ntohs(proto) >= 1536) + if (ntohs(proto) >= ETH_P_802_3_MIN) return proto; - if (unlikely(skb->len < sizeof(struct llc_snap_hdr))) + if (skb->len < sizeof(struct llc_snap_hdr)) return htons(ETH_P_802_2); + if (unlikely(!pskb_may_pull(skb, sizeof(struct llc_snap_hdr)))) + return htons(0); + llc = (struct llc_snap_hdr *) skb->data; if (llc->dsap != LLC_SAP_SNAP || llc->ssap != LLC_SAP_SNAP || @@ -251,17 +340,92 @@ static __be16 parse_ethertype(struct sk_buff *skb) return htons(ETH_P_802_2); __skb_pull(skb, sizeof(struct llc_snap_hdr)); - return llc->ethertype; + + if (ntohs(llc->ethertype) >= ETH_P_802_3_MIN) + return llc->ethertype; + + return htons(ETH_P_802_2); +} + +static int parse_icmpv6(struct sk_buff *skb, struct sw_flow_key *key, + int nh_len) +{ + struct icmp6hdr *icmp = icmp6_hdr(skb); + + /* The ICMPv6 type and code fields use the 16-bit transport port + * fields, so we need to store them in 16-bit network byte order. + */ + key->tp.src = htons(icmp->icmp6_type); + key->tp.dst = htons(icmp->icmp6_code); + + if (icmp->icmp6_code == 0 && + (icmp->icmp6_type == NDISC_NEIGHBOUR_SOLICITATION || + icmp->icmp6_type == NDISC_NEIGHBOUR_ADVERTISEMENT)) { + int icmp_len = skb->len - skb_transport_offset(skb); + struct nd_msg *nd; + int offset; + + /* In order to process neighbor discovery options, we need the + * entire packet. + */ + if (unlikely(icmp_len < sizeof(*nd))) + return 0; + + if (unlikely(skb_linearize(skb))) + return -ENOMEM; + + nd = (struct nd_msg *)skb_transport_header(skb); + key->ipv6.nd.target = nd->target; + + icmp_len -= sizeof(*nd); + offset = 0; + while (icmp_len >= 8) { + struct nd_opt_hdr *nd_opt = + (struct nd_opt_hdr *)(nd->opt + offset); + int opt_len = nd_opt->nd_opt_len * 8; + + if (unlikely(!opt_len || opt_len > icmp_len)) + return 0; + + /* Store the link layer address if the appropriate + * option is provided. It is considered an error if + * the same link layer option is specified twice. + */ + if (nd_opt->nd_opt_type == ND_OPT_SOURCE_LL_ADDR + && opt_len == 8) { + if (unlikely(!is_zero_ether_addr(key->ipv6.nd.sll))) + goto invalid; + ether_addr_copy(key->ipv6.nd.sll, + &nd->opt[offset+sizeof(*nd_opt)]); + } else if (nd_opt->nd_opt_type == ND_OPT_TARGET_LL_ADDR + && opt_len == 8) { + if (unlikely(!is_zero_ether_addr(key->ipv6.nd.tll))) + goto invalid; + ether_addr_copy(key->ipv6.nd.tll, + &nd->opt[offset+sizeof(*nd_opt)]); + } + + icmp_len -= opt_len; + offset += opt_len; + } + } + + return 0; + +invalid: + memset(&key->ipv6.nd.target, 0, sizeof(key->ipv6.nd.target)); + memset(key->ipv6.nd.sll, 0, sizeof(key->ipv6.nd.sll)); + memset(key->ipv6.nd.tll, 0, sizeof(key->ipv6.nd.tll)); + + return 0; } /** - * flow_extract - extracts a flow key from an Ethernet frame. + * ovs_flow_extract - extracts a flow key from an Ethernet frame. * @skb: sk_buff that contains the frame, with skb->data pointing to the * 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. * @@ -274,105 +438,114 @@ static __be16 parse_ethertype(struct sk_buff *skb) * - 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. + * - skb->transport_header: If key->eth.type is ETH_P_IP or ETH_P_IPV6 + * on output, then just past the IP header, if one is present and + * of a correct length, otherwise the same as skb->network_header. + * For other key->eth.type values it is left untouched. */ -int flow_extract(struct sk_buff *skb, u16 in_port, struct sw_flow_key *key, - bool *is_frag) +int ovs_flow_extract(struct sk_buff *skb, u16 in_port, struct sw_flow_key *key) { + int error; struct ethhdr *eth; memset(key, 0, sizeof(*key)); - key->tun_id = OVS_CB(skb)->tun_id; - key->in_port = in_port; - *is_frag = false; - - /* - * We would really like to pull as many bytes as we could possibly - * want to parse into the linear data area. Currently that is: - * - * 14 Ethernet header - * 4 VLAN header - * 60 max IP header with options - * 20 max TCP/UDP/ICMP header (don't care about options) - * -- - * 98 - * - * But Xen only allocates 64 or 72 bytes for the linear data area in - * netback, which means that we would reallocate and copy the skb's - * linear data on every packet if we did that. So instead just pull 64 - * bytes, which is always sufficient without IP options, and then check - * whether we need to pull more later when we look at the IP header. - */ - if (!pskb_may_pull(skb, min(skb->len, 64u))) - return -ENOMEM; + + key->phy.priority = skb->priority; + 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->mark; skb_reset_mac_header(skb); - /* Link layer. */ + /* Link layer. We are guaranteed to have at least the 14 byte Ethernet + * header in the linear data area. + */ eth = eth_hdr(skb); - memcpy(key->dl_src, eth->h_source, ETH_ALEN); - memcpy(key->dl_dst, eth->h_dest, ETH_ALEN); + ether_addr_copy(key->eth.src, eth->h_source); + ether_addr_copy(key->eth.dst, eth->h_dest); - /* dl_type, dl_vlan, dl_vlan_pcp. */ __skb_pull(skb, 2 * ETH_ALEN); - if (eth->h_proto == htons(ETH_P_8021Q)) - parse_vlan(skb, key); - key->dl_type = parse_ethertype(skb); + /* We are going to push all headers that we pull, so no need to + * update skb->csum here. */ + + if (vlan_tx_tag_present(skb)) + key->eth.tci = htons(vlan_get_tci(skb)); + else if (eth->h_proto == htons(ETH_P_8021Q)) + if (unlikely(parse_vlan(skb, key))) + return -ENOMEM; + + key->eth.type = parse_ethertype(skb); + if (unlikely(key->eth.type == htons(0))) + return -ENOMEM; + skb_reset_network_header(skb); - __skb_push(skb, skb->data - (unsigned char *)eth); + __skb_push(skb, skb->data - skb_mac_header(skb)); /* Network layer. */ - if (key->dl_type == htons(ETH_P_IP)) { + if (key->eth.type == htons(ETH_P_IP)) { struct iphdr *nh; - int error; + __be16 offset; error = check_iphdr(skb); if (unlikely(error)) { if (error == -EINVAL) { skb->transport_header = skb->network_header; - return 0; + error = 0; } return error; } nh = ip_hdr(skb); - key->nw_src = nh->saddr; - key->nw_dst = nh->daddr; - key->nw_tos = nh->tos & ~INET_ECN_MASK; - key->nw_proto = nh->protocol; + key->ipv4.addr.src = nh->saddr; + key->ipv4.addr.dst = nh->daddr; + + key->ip.proto = nh->protocol; + key->ip.tos = nh->tos; + key->ip.ttl = nh->ttl; + + offset = nh->frag_off & htons(IP_OFFSET); + if (offset) { + key->ip.frag = OVS_FRAG_TYPE_LATER; + return 0; + } + if (nh->frag_off & htons(IP_MF) || + skb_shinfo(skb)->gso_type & SKB_GSO_UDP) + key->ip.frag = OVS_FRAG_TYPE_FIRST; /* Transport layer. */ - 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 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 if (key->nw_proto == IPPROTO_ICMP) { - if (icmphdr_ok(skb)) { - struct icmphdr *icmp = icmp_hdr(skb); - /* The ICMP type and code fields use the 16-bit - * transport port fields, so we need to store them - * in 16-bit network byte order. */ - key->tp_src = htons(icmp->type); - key->tp_dst = htons(icmp->code); - } + if (key->ip.proto == IPPROTO_TCP) { + if (tcphdr_ok(skb)) { + struct tcphdr *tcp = tcp_hdr(skb); + key->tp.src = tcp->source; + key->tp.dst = tcp->dest; + key->tp.flags = TCP_FLAGS_BE16(tcp); + } + } else if (key->ip.proto == IPPROTO_UDP) { + if (udphdr_ok(skb)) { + struct udphdr *udp = udp_hdr(skb); + key->tp.src = udp->source; + key->tp.dst = udp->dest; + } + } else if (key->ip.proto == IPPROTO_SCTP) { + if (sctphdr_ok(skb)) { + struct sctphdr *sctp = sctp_hdr(skb); + key->tp.src = sctp->source; + key->tp.dst = sctp->dest; } - } else - *is_frag = true; + } else if (key->ip.proto == IPPROTO_ICMP) { + if (icmphdr_ok(skb)) { + struct icmphdr *icmp = icmp_hdr(skb); + /* The ICMP type and code fields use the 16-bit + * transport port fields, so we need to store + * them in 16-bit network byte order. */ + key->tp.src = htons(icmp->type); + key->tp.dst = htons(icmp->code); + } + } - } else if (key->dl_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); @@ -384,311 +557,59 @@ int 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->nw_proto = ntohs(arp->ar_op); + key->ip.proto = ntohs(arp->ar_op); + 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)); + ether_addr_copy(key->ipv4.arp.sha, arp->ar_sha); + ether_addr_copy(key->ipv4.arp.tha, arp->ar_tha); + } + } else if (key->eth.type == htons(ETH_P_IPV6)) { + int nh_len; /* IPv6 Header + Extensions */ - 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)); + nh_len = parse_ipv6hdr(skb, key); + if (unlikely(nh_len < 0)) { + if (nh_len == -EINVAL) { + skb->transport_header = skb->network_header; + error = 0; + } else { + error = nh_len; } + return error; } - } - return 0; -} -u32 flow_hash(const struct sw_flow_key *key) -{ - return jhash2((u32*)key, sizeof(*key) / sizeof(u32), hash_seed); -} + if (key->ip.frag == OVS_FRAG_TYPE_LATER) + return 0; + if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP) + key->ip.frag = OVS_FRAG_TYPE_FIRST; -int flow_cmp(const struct tbl_node *node, void *key2_) -{ - const struct sw_flow_key *key1 = &flow_cast(node)->key; - const struct sw_flow_key *key2 = key2_; - - return !memcmp(key1, key2, sizeof(struct sw_flow_key)); -} - -/** - * flow_from_nlattrs - parses Netlink attributes into a flow key. - * @swkey: receives the extracted flow key. - * @key: Netlink attribute holding nested %ODP_KEY_ATTR_* Netlink attribute - * sequence. - * - * This state machine accepts the following forms, with [] for optional - * elements and | for alternatives: - * - * [tun_id] in_port ethernet [8021q] [ethertype [IP [TCP|UDP|ICMP] | ARP] - */ -int flow_from_nlattrs(struct sw_flow_key *swkey, const struct nlattr *attr) -{ - const struct nlattr *nla; - u16 prev_type; - int rem; - - memset(swkey, 0, sizeof(*swkey)); - swkey->dl_type = htons(ETH_P_802_2); - - prev_type = ODP_KEY_ATTR_UNSPEC; - nla_for_each_nested(nla, attr, rem) { - static const u32 key_lens[ODP_KEY_ATTR_MAX + 1] = { - [ODP_KEY_ATTR_TUN_ID] = 8, - [ODP_KEY_ATTR_IN_PORT] = 4, - [ODP_KEY_ATTR_ETHERNET] = sizeof(struct odp_key_ethernet), - [ODP_KEY_ATTR_8021Q] = sizeof(struct odp_key_8021q), - [ODP_KEY_ATTR_ETHERTYPE] = 2, - [ODP_KEY_ATTR_IPV4] = sizeof(struct odp_key_ipv4), - [ODP_KEY_ATTR_TCP] = sizeof(struct odp_key_tcp), - [ODP_KEY_ATTR_UDP] = sizeof(struct odp_key_udp), - [ODP_KEY_ATTR_ICMP] = sizeof(struct odp_key_icmp), - [ODP_KEY_ATTR_ARP] = sizeof(struct odp_key_arp), - }; - - const struct odp_key_ethernet *eth_key; - const struct odp_key_8021q *q_key; - const struct odp_key_ipv4 *ipv4_key; - const struct odp_key_tcp *tcp_key; - const struct odp_key_udp *udp_key; - const struct odp_key_icmp *icmp_key; - const struct odp_key_arp *arp_key; - - int type = nla_type(nla); - - if (type > ODP_KEY_ATTR_MAX || nla_len(nla) != key_lens[type]) - return -EINVAL; - -#define TRANSITION(PREV_TYPE, TYPE) (((PREV_TYPE) << 16) | (TYPE)) - switch (TRANSITION(prev_type, type)) { - case TRANSITION(ODP_KEY_ATTR_UNSPEC, ODP_KEY_ATTR_TUN_ID): - swkey->tun_id = nla_get_be64(nla); - break; - - case TRANSITION(ODP_KEY_ATTR_UNSPEC, ODP_KEY_ATTR_IN_PORT): - case TRANSITION(ODP_KEY_ATTR_TUN_ID, ODP_KEY_ATTR_IN_PORT): - if (nla_get_u32(nla) >= DP_MAX_PORTS) - return -EINVAL; - swkey->in_port = nla_get_u32(nla); - break; - - case TRANSITION(ODP_KEY_ATTR_IN_PORT, ODP_KEY_ATTR_ETHERNET): - eth_key = nla_data(nla); - memcpy(swkey->dl_src, eth_key->eth_src, ETH_ALEN); - memcpy(swkey->dl_dst, eth_key->eth_dst, ETH_ALEN); - break; - - case TRANSITION(ODP_KEY_ATTR_ETHERNET, ODP_KEY_ATTR_8021Q): - q_key = nla_data(nla); - /* Only standard 0x8100 VLANs currently supported. */ - if (q_key->q_tpid != htons(ETH_P_8021Q)) - return -EINVAL; - if (q_key->q_tci & htons(VLAN_TAG_PRESENT)) - return -EINVAL; - swkey->dl_tci = q_key->q_tci | htons(VLAN_TAG_PRESENT); - break; - - case TRANSITION(ODP_KEY_ATTR_8021Q, ODP_KEY_ATTR_ETHERTYPE): - case TRANSITION(ODP_KEY_ATTR_ETHERNET, ODP_KEY_ATTR_ETHERTYPE): - swkey->dl_type = nla_get_be16(nla); - if (ntohs(swkey->dl_type) < 1536) - return -EINVAL; - break; - - case TRANSITION(ODP_KEY_ATTR_ETHERTYPE, ODP_KEY_ATTR_IPV4): - if (swkey->dl_type != htons(ETH_P_IP)) - return -EINVAL; - ipv4_key = nla_data(nla); - swkey->nw_src = ipv4_key->ipv4_src; - swkey->nw_dst = ipv4_key->ipv4_dst; - swkey->nw_proto = ipv4_key->ipv4_proto; - swkey->nw_tos = ipv4_key->ipv4_tos; - if (swkey->nw_tos & INET_ECN_MASK) - return -EINVAL; - break; - - case TRANSITION(ODP_KEY_ATTR_IPV4, ODP_KEY_ATTR_TCP): - if (swkey->nw_proto != IPPROTO_TCP) - return -EINVAL; - tcp_key = nla_data(nla); - swkey->tp_src = tcp_key->tcp_src; - swkey->tp_dst = tcp_key->tcp_dst; - break; - - case TRANSITION(ODP_KEY_ATTR_IPV4, ODP_KEY_ATTR_UDP): - if (swkey->nw_proto != IPPROTO_UDP) - return -EINVAL; - udp_key = nla_data(nla); - swkey->tp_src = udp_key->udp_src; - swkey->tp_dst = udp_key->udp_dst; - break; - - case TRANSITION(ODP_KEY_ATTR_IPV4, ODP_KEY_ATTR_ICMP): - if (swkey->nw_proto != IPPROTO_ICMP) - return -EINVAL; - icmp_key = nla_data(nla); - swkey->tp_src = htons(icmp_key->icmp_type); - swkey->tp_dst = htons(icmp_key->icmp_code); - break; - - case TRANSITION(ODP_KEY_ATTR_ETHERTYPE, ODP_KEY_ATTR_ARP): - if (swkey->dl_type != htons(ETH_P_ARP)) - return -EINVAL; - arp_key = nla_data(nla); - swkey->nw_src = arp_key->arp_sip; - swkey->nw_dst = arp_key->arp_tip; - if (arp_key->arp_op & htons(0xff00)) - return -EINVAL; - swkey->nw_proto = ntohs(arp_key->arp_op); - break; - - default: - return -EINVAL; - } - - prev_type = type; - } - if (rem) - return -EINVAL; - - switch (prev_type) { - case ODP_KEY_ATTR_UNSPEC: - return -EINVAL; - - case ODP_KEY_ATTR_TUN_ID: - case ODP_KEY_ATTR_IN_PORT: - return -EINVAL; - - case ODP_KEY_ATTR_ETHERNET: - case ODP_KEY_ATTR_8021Q: - return 0; - - case ODP_KEY_ATTR_ETHERTYPE: - if (swkey->dl_type == htons(ETH_P_IP) || - swkey->dl_type == htons(ETH_P_ARP)) - return -EINVAL; - return 0; - - case ODP_KEY_ATTR_IPV4: - if (swkey->nw_proto == IPPROTO_TCP || - swkey->nw_proto == IPPROTO_UDP || - swkey->nw_proto == IPPROTO_ICMP) - return -EINVAL; - return 0; - - case ODP_KEY_ATTR_TCP: - case ODP_KEY_ATTR_UDP: - case ODP_KEY_ATTR_ICMP: - case ODP_KEY_ATTR_ARP: - return 0; - } - - WARN_ON_ONCE(1); - return -EINVAL; -} - -int flow_to_nlattrs(const struct sw_flow_key *swkey, struct sk_buff *skb) -{ - struct odp_key_ethernet *eth_key; - struct nlattr *nla; - - if (swkey->tun_id != cpu_to_be64(0)) - NLA_PUT_BE64(skb, ODP_KEY_ATTR_TUN_ID, swkey->tun_id); - - NLA_PUT_U32(skb, ODP_KEY_ATTR_IN_PORT, swkey->in_port); - - nla = nla_reserve(skb, ODP_KEY_ATTR_ETHERNET, sizeof(*eth_key)); - if (!nla) - goto nla_put_failure; - eth_key = nla_data(nla); - memcpy(eth_key->eth_src, swkey->dl_src, ETH_ALEN); - memcpy(eth_key->eth_dst, swkey->dl_dst, ETH_ALEN); - - if (swkey->dl_tci != htons(0)) { - struct odp_key_8021q q_key; - - q_key.q_tpid = htons(ETH_P_8021Q); - q_key.q_tci = swkey->dl_tci & ~htons(VLAN_TAG_PRESENT); - NLA_PUT(skb, ODP_KEY_ATTR_8021Q, sizeof(q_key), &q_key); - } - - if (swkey->dl_type == htons(ETH_P_802_2)) - return 0; - - NLA_PUT_BE16(skb, ODP_KEY_ATTR_ETHERTYPE, swkey->dl_type); - - if (swkey->dl_type == htons(ETH_P_IP)) { - struct odp_key_ipv4 *ipv4_key; - - nla = nla_reserve(skb, ODP_KEY_ATTR_IPV4, sizeof(*ipv4_key)); - if (!nla) - goto nla_put_failure; - ipv4_key = nla_data(nla); - ipv4_key->ipv4_src = swkey->nw_src; - ipv4_key->ipv4_dst = swkey->nw_dst; - ipv4_key->ipv4_proto = swkey->nw_proto; - ipv4_key->ipv4_tos = swkey->nw_tos; - - if (swkey->nw_proto == IPPROTO_TCP) { - struct odp_key_tcp *tcp_key; - - nla = nla_reserve(skb, ODP_KEY_ATTR_TCP, sizeof(*tcp_key)); - if (!nla) - goto nla_put_failure; - tcp_key = nla_data(nla); - tcp_key->tcp_src = swkey->tp_src; - tcp_key->tcp_dst = swkey->tp_dst; - } else if (swkey->nw_proto == IPPROTO_UDP) { - struct odp_key_udp *udp_key; - - nla = nla_reserve(skb, ODP_KEY_ATTR_UDP, sizeof(*udp_key)); - if (!nla) - goto nla_put_failure; - udp_key = nla_data(nla); - udp_key->udp_src = swkey->tp_src; - udp_key->udp_dst = swkey->tp_dst; - } else if (swkey->nw_proto == IPPROTO_ICMP) { - struct odp_key_icmp *icmp_key; - - nla = nla_reserve(skb, ODP_KEY_ATTR_ICMP, sizeof(*icmp_key)); - if (!nla) - goto nla_put_failure; - icmp_key = nla_data(nla); - icmp_key->icmp_type = ntohs(swkey->tp_src); - icmp_key->icmp_code = ntohs(swkey->tp_dst); + /* Transport layer. */ + if (key->ip.proto == NEXTHDR_TCP) { + if (tcphdr_ok(skb)) { + struct tcphdr *tcp = tcp_hdr(skb); + key->tp.src = tcp->source; + key->tp.dst = tcp->dest; + key->tp.flags = TCP_FLAGS_BE16(tcp); + } + } else if (key->ip.proto == NEXTHDR_UDP) { + if (udphdr_ok(skb)) { + struct udphdr *udp = udp_hdr(skb); + key->tp.src = udp->source; + key->tp.dst = udp->dest; + } + } else if (key->ip.proto == NEXTHDR_SCTP) { + if (sctphdr_ok(skb)) { + struct sctphdr *sctp = sctp_hdr(skb); + key->tp.src = sctp->source; + key->tp.dst = sctp->dest; + } + } else if (key->ip.proto == NEXTHDR_ICMP) { + if (icmp6hdr_ok(skb)) { + error = parse_icmpv6(skb, key, nh_len); + if (error) + return error; + } } - } else if (swkey->dl_type == htons(ETH_P_ARP)) { - struct odp_key_arp *arp_key; - - nla = nla_reserve(skb, ODP_KEY_ATTR_ARP, sizeof(*arp_key)); - if (!nla) - goto nla_put_failure; - arp_key = nla_data(nla); - arp_key->arp_sip = swkey->nw_src; - arp_key->arp_tip = swkey->nw_dst; - arp_key->arp_op = htons(swkey->nw_proto); } return 0; - -nla_put_failure: - return -EMSGSIZE; -} - -/* Initializes the flow module. - * Returns zero if successful or a negative error code. */ -int flow_init(void) -{ - flow_cache = kmem_cache_create("sw_flow", sizeof(struct sw_flow), 0, - 0, NULL); - if (flow_cache == NULL) - return -ENOMEM; - - get_random_bytes(&hash_seed, sizeof(hash_seed)); - - return 0; -} - -/* Uninitializes the flow module. */ -void flow_exit(void) -{ - kmem_cache_destroy(flow_cache); }