X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=datapath%2Fdatapath.c;h=1ea5d1e9a7c2d850cf989b34c6a5f3e7b53d25ac;hb=6161d3fd928edf7016abae60f549a135a2f83f09;hp=a3be3254776c2b3e4b2f3cdc091fa87ecc606b6b;hpb=4edb9ae90e4092f5f56b9d914d2b88783c49860d;p=openvswitch diff --git a/datapath/datapath.c b/datapath/datapath.c index a3be3254..1ea5d1e9 100644 --- a/datapath/datapath.c +++ b/datapath/datapath.c @@ -1,13 +1,21 @@ /* - * Copyright (c) 2007, 2008, 2009, 2010, 2011 Nicira Networks. - * Distributed under the terms of the GNU GPL version 2. + * Copyright (c) 2007-2011 Nicira Networks. * - * 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 */ -/* Functions for managing the dp interface/device. */ - #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt #include @@ -33,7 +41,6 @@ #include #include #include -#include #include #include #include @@ -42,20 +49,18 @@ #include #include #include -#include #include #include "checksum.h" #include "datapath.h" -#include "actions.h" #include "flow.h" #include "vlan.h" #include "tunnel.h" #include "vport-internal_dev.h" #if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,18) || \ - LINUX_VERSION_CODE >= KERNEL_VERSION(3,2,0) -#error Kernels before 2.6.18 or after 3.1 are not supported by this version of Open vSwitch. + LINUX_VERSION_CODE > KERNEL_VERSION(3,2,0) +#error Kernels before 2.6.18 or after 3.2 are not supported by this version of Open vSwitch. #endif int (*dp_ioctl_hook)(struct net_device *dev, struct ifreq *rq, int cmd); @@ -68,8 +73,8 @@ EXPORT_SYMBOL(dp_ioctl_hook); * etc.) are protected by RTNL. * * Writes to other state (flow table modifications, set miscellaneous datapath - * parameters such as drop frags, etc.) are protected by genl_mutex. The RTNL - * lock nests inside genl_mutex. + * parameters, etc.) are protected by genl_mutex. The RTNL lock nests inside + * genl_mutex. * * Reads are protected by RCU. * @@ -84,11 +89,13 @@ EXPORT_SYMBOL(dp_ioctl_hook); static LIST_HEAD(dps); static struct vport *new_vport(const struct vport_parms *); -static int queue_userspace_packets(struct datapath *, struct sk_buff *, - const struct dp_upcall_info *); +static int queue_gso_packets(int dp_ifindex, struct sk_buff *, + const struct dp_upcall_info *); +static int queue_userspace_packet(int dp_ifindex, struct sk_buff *, + const struct dp_upcall_info *); /* Must be called with rcu_read_lock, genl_mutex, or RTNL lock. */ -struct datapath *get_dp(int dp_ifindex) +static struct datapath *get_dp(int dp_ifindex) { struct datapath *dp = NULL; struct net_device *dev; @@ -104,7 +111,6 @@ struct datapath *get_dp(int dp_ifindex) return dp; } -EXPORT_SYMBOL_GPL(get_dp); /* Must be called with genl_mutex. */ static struct flow_table *get_table_protected(struct datapath *dp) @@ -121,7 +127,8 @@ static struct vport *get_vport_protected(struct datapath *dp, u16 port_no) /* Must be called with rcu_read_lock or RTNL lock. */ const char *dp_name(const struct datapath *dp) { - return vport_get_name(rcu_dereference_rtnl(dp->ports[OVSP_LOCAL])); + struct vport *vport = rcu_dereference_rtnl(dp->ports[OVSP_LOCAL]); + return vport->ops->get_name(vport); } static int get_dpifindex(struct datapath *dp) @@ -133,7 +140,7 @@ static int get_dpifindex(struct datapath *dp) local = get_vport_protected(dp, OVSP_LOCAL); if (local) - ifindex = vport_get_ifindex(local); + ifindex = local->ops->get_ifindex(local); else ifindex = 0; @@ -142,7 +149,7 @@ static int get_dpifindex(struct datapath *dp) return ifindex; } -static inline size_t br_nlmsg_size(void) +static size_t br_nlmsg_size(void) { return NLMSG_ALIGN(sizeof(struct ifinfomsg)) + nla_total_size(IFNAMSIZ) /* IFLA_IFNAME */ @@ -158,12 +165,11 @@ static int dp_fill_ifinfo(struct sk_buff *skb, int event, unsigned int flags) { struct datapath *dp = port->dp; - int ifindex = vport_get_ifindex(port); struct ifinfomsg *hdr; struct nlmsghdr *nlh; - if (ifindex < 0) - return ifindex; + if (!port->ops->get_ifindex) + return -ENODEV; nlh = nlmsg_put(skb, 0, 0, event, sizeof(*hdr), flags); if (nlh == NULL) @@ -173,21 +179,21 @@ static int dp_fill_ifinfo(struct sk_buff *skb, hdr->ifi_family = AF_BRIDGE; hdr->__ifi_pad = 0; hdr->ifi_type = ARPHRD_ETHER; - hdr->ifi_index = ifindex; - hdr->ifi_flags = vport_get_flags(port); + hdr->ifi_index = port->ops->get_ifindex(port); + hdr->ifi_flags = port->ops->get_dev_flags(port); hdr->ifi_change = 0; - NLA_PUT_STRING(skb, IFLA_IFNAME, vport_get_name(port)); + NLA_PUT_STRING(skb, IFLA_IFNAME, port->ops->get_name(port)); NLA_PUT_U32(skb, IFLA_MASTER, get_dpifindex(dp)); - NLA_PUT_U32(skb, IFLA_MTU, vport_get_mtu(port)); + NLA_PUT_U32(skb, IFLA_MTU, port->ops->get_mtu(port)); #ifdef IFLA_OPERSTATE NLA_PUT_U8(skb, IFLA_OPERSTATE, - vport_is_running(port) - ? vport_get_operstate(port) + port->ops->is_running(port) + ? port->ops->get_operstate(port) : IF_OPER_DOWN); #endif - NLA_PUT(skb, IFLA_ADDRESS, ETH_ALEN, vport_get_addr(port)); + NLA_PUT(skb, IFLA_ADDRESS, ETH_ALEN, port->ops->get_addr(port)); return nlmsg_end(skb, nlh); @@ -200,24 +206,32 @@ nla_put_failure: static void dp_ifinfo_notify(int event, struct vport *port) { struct sk_buff *skb; - int err = -ENOBUFS; + int err; skb = nlmsg_new(br_nlmsg_size(), GFP_KERNEL); - if (skb == NULL) - goto errout; + if (!skb) { + err = -ENOBUFS; + goto err; + } err = dp_fill_ifinfo(skb, port, event, 0); if (err < 0) { - /* -EMSGSIZE implies BUG in br_nlmsg_size() */ - WARN_ON(err == -EMSGSIZE); - kfree_skb(skb); - goto errout; + if (err == -ENODEV) { + goto out; + } else { + /* -EMSGSIZE implies BUG in br_nlmsg_size() */ + WARN_ON(err == -EMSGSIZE); + goto err; + } } + rtnl_notify(skb, &init_net, 0, RTNLGRP_LINK, NULL, GFP_KERNEL); + return; -errout: - if (err < 0) - rtnl_set_sk_err(&init_net, RTNLGRP_LINK, err); +err: + rtnl_set_sk_err(&init_net, RTNLGRP_LINK, err); +out: + kfree_skb(skb); } static void release_dp(struct kobject *kobj) @@ -234,7 +248,7 @@ static void destroy_dp_rcu(struct rcu_head *rcu) { struct datapath *dp = container_of(rcu, struct datapath, rcu); - flow_tbl_destroy(dp->table); + flow_tbl_destroy((__force struct flow_table *)dp->table); free_percpu(dp->stats_percpu); kobject_put(&dp->ifobj); } @@ -284,28 +298,21 @@ void dp_process_received_packet(struct vport *p, struct sk_buff *skb) int error; stats = per_cpu_ptr(dp->stats_percpu, smp_processor_id()); - OVS_CB(skb)->vport = p; if (!OVS_CB(skb)->flow) { struct sw_flow_key key; int key_len; - bool is_frag; /* Extract flow from 'skb' into 'key'. */ - error = flow_extract(skb, p->port_no, &key, &key_len, &is_frag); + error = flow_extract(skb, p->port_no, &key, &key_len); if (unlikely(error)) { kfree_skb(skb); return; } - if (is_frag && dp->drop_frags) { - consume_skb(skb); - stats_counter = &stats->n_frags; - goto out; - } - /* Look up flow. */ - flow = flow_tbl_lookup(rcu_dereference(dp->table), &key, key_len); + flow = flow_tbl_lookup(rcu_dereference(dp->table), + &key, key_len); if (unlikely(!flow)) { struct dp_upcall_info upcall; @@ -314,7 +321,7 @@ void dp_process_received_packet(struct vport *p, struct sk_buff *skb) upcall.userdata = NULL; upcall.pid = p->upcall_pid; dp_upcall(dp, skb, &upcall); - kfree_skb(skb); + consume_skb(skb); stats_counter = &stats->n_missed; goto out; } @@ -334,34 +341,19 @@ out: write_seqcount_end(&stats->seqlock); } -static void copy_and_csum_skb(struct sk_buff *skb, void *to) -{ - u16 csum_start, csum_offset; - __wsum csum; - - get_skb_csum_pointers(skb, &csum_start, &csum_offset); - csum_start -= skb_headroom(skb); - - skb_copy_bits(skb, 0, to, csum_start); - - csum = skb_copy_and_csum_bits(skb, csum_start, to + csum_start, - skb->len - csum_start, 0); - *(__sum16 *)(to + csum_start + csum_offset) = csum_fold(csum); -} - static struct genl_family dp_packet_genl_family = { .id = GENL_ID_GENERATE, .hdrsize = sizeof(struct ovs_header), .name = OVS_PACKET_FAMILY, - .version = 1, + .version = OVS_PACKET_VERSION, .maxattr = OVS_PACKET_ATTR_MAX }; int dp_upcall(struct datapath *dp, struct sk_buff *skb, const struct dp_upcall_info *upcall_info) { - struct sk_buff *segs = NULL; struct dp_stats_percpu *stats; + int dp_ifindex; int err; if (upcall_info->pid == 0) { @@ -369,30 +361,18 @@ int dp_upcall(struct datapath *dp, struct sk_buff *skb, goto err; } - forward_ip_summed(skb, true); - - /* Break apart GSO packets into their component pieces. Otherwise - * userspace may try to stuff a 64kB packet into a 1500-byte MTU. */ - if (skb_is_gso(skb)) { - segs = skb_gso_segment(skb, NETIF_F_SG | NETIF_F_HW_CSUM); - - if (IS_ERR(segs)) { - err = PTR_ERR(segs); - goto err; - } - skb = segs; + dp_ifindex = get_dpifindex(dp); + if (!dp_ifindex) { + err = -ENODEV; + goto err; } - err = queue_userspace_packets(dp, skb, upcall_info); - if (segs) { - struct sk_buff *next; - /* Free GSO-segments */ - do { - next = segs->next; - kfree_skb(segs); - } while ((segs = next) != NULL); - } + forward_ip_summed(skb, true); + if (!skb_is_gso(skb)) + err = queue_userspace_packet(dp_ifindex, skb, upcall_info); + else + err = queue_gso_packets(dp_ifindex, skb, upcall_info); if (err) goto err; @@ -408,68 +388,111 @@ err: return err; } -/* Send each packet in the 'skb' list to userspace for 'dp' as directed by - * 'upcall_info'. There will be only one packet unless we broke up a GSO - * packet. - */ -static int queue_userspace_packets(struct datapath *dp, struct sk_buff *skb, - const struct dp_upcall_info *upcall_info) +static int queue_gso_packets(int dp_ifindex, struct sk_buff *skb, + const struct dp_upcall_info *upcall_info) { - int dp_ifindex; + struct dp_upcall_info later_info; + struct sw_flow_key later_key; + struct sk_buff *segs, *nskb; + int err; - dp_ifindex = get_dpifindex(dp); - if (!dp_ifindex) - return -ENODEV; + segs = skb_gso_segment(skb, NETIF_F_SG | NETIF_F_HW_CSUM); + if (IS_ERR(skb)) + return PTR_ERR(skb); + /* Queue all of the segments. */ + skb = segs; do { - struct ovs_header *upcall; - struct sk_buff *user_skb; /* to be queued to userspace */ - struct nlattr *nla; - unsigned int len; - int err; - - err = vlan_deaccel_tag(skb); - if (unlikely(err)) - return err; + err = queue_userspace_packet(dp_ifindex, skb, upcall_info); + if (err) + break; - if (nla_attr_size(skb->len) > USHRT_MAX) - return -EFBIG; + if (skb == segs && skb_shinfo(skb)->gso_type & SKB_GSO_UDP) { + /* The initial flow key extracted by flow_extract() in + * this case is for a first fragment, so we need to + * properly mark later fragments. + */ + later_key = *upcall_info->key; + later_key.ip.frag = OVS_FRAG_TYPE_LATER; + + later_info = *upcall_info; + later_info.key = &later_key; + upcall_info = &later_info; + } + } while ((skb = skb->next)); + + /* Free all of the segments. */ + skb = segs; + do { + nskb = skb->next; + if (err) + kfree_skb(skb); + else + consume_skb(skb); + } while ((skb = nskb)); + return err; +} - len = sizeof(struct ovs_header); - len += nla_total_size(skb->len); - len += nla_total_size(FLOW_BUFSIZE); - if (upcall_info->cmd == OVS_PACKET_CMD_ACTION) - len += nla_total_size(8); +static int queue_userspace_packet(int dp_ifindex, struct sk_buff *skb, + const struct dp_upcall_info *upcall_info) +{ + struct ovs_header *upcall; + struct sk_buff *nskb = NULL; + struct sk_buff *user_skb; /* to be queued to userspace */ + struct nlattr *nla; + unsigned int len; + int err; - user_skb = genlmsg_new(len, GFP_ATOMIC); - if (!user_skb) + if (vlan_tx_tag_present(skb)) { + nskb = skb_clone(skb, GFP_ATOMIC); + if (!nskb) return -ENOMEM; + + err = vlan_deaccel_tag(nskb); + if (err) + return err; - upcall = genlmsg_put(user_skb, 0, 0, &dp_packet_genl_family, - 0, upcall_info->cmd); - upcall->dp_ifindex = dp_ifindex; + skb = nskb; + } - nla = nla_nest_start(user_skb, OVS_PACKET_ATTR_KEY); - flow_to_nlattrs(upcall_info->key, user_skb); - nla_nest_end(user_skb, nla); + if (nla_attr_size(skb->len) > USHRT_MAX) { + err = -EFBIG; + goto out; + } - if (upcall_info->userdata) - nla_put_u64(user_skb, OVS_PACKET_ATTR_USERDATA, - nla_get_u64(upcall_info->userdata)); + len = sizeof(struct ovs_header); + len += nla_total_size(skb->len); + len += nla_total_size(FLOW_BUFSIZE); + if (upcall_info->cmd == OVS_PACKET_CMD_ACTION) + len += nla_total_size(8); - nla = __nla_reserve(user_skb, OVS_PACKET_ATTR_PACKET, skb->len); - if (skb->ip_summed == CHECKSUM_PARTIAL) - copy_and_csum_skb(skb, nla_data(nla)); - else - skb_copy_bits(skb, 0, nla_data(nla), skb->len); + user_skb = genlmsg_new(len, GFP_ATOMIC); + if (!user_skb) { + err = -ENOMEM; + goto out; + } - err = genlmsg_unicast(&init_net, user_skb, upcall_info->pid); - if (err) - return err; + upcall = genlmsg_put(user_skb, 0, 0, &dp_packet_genl_family, + 0, upcall_info->cmd); + upcall->dp_ifindex = dp_ifindex; - } while ((skb = skb->next)); + nla = nla_nest_start(user_skb, OVS_PACKET_ATTR_KEY); + flow_to_nlattrs(upcall_info->key, user_skb); + nla_nest_end(user_skb, nla); - return 0; + if (upcall_info->userdata) + nla_put_u64(user_skb, OVS_PACKET_ATTR_USERDATA, + nla_get_u64(upcall_info->userdata)); + + nla = __nla_reserve(user_skb, OVS_PACKET_ATTR_PACKET, skb->len); + + skb_copy_and_csum_dev(skb, nla_data(nla)); + + err = genlmsg_unicast(&init_net, user_skb, upcall_info->pid); + +out: + kfree_skb(nskb); + return err; } /* Called with genl_mutex. */ @@ -500,30 +523,34 @@ static int validate_actions(const struct nlattr *attr, static int validate_sample(const struct nlattr *attr, const struct sw_flow_key *key, int depth) { - static const struct nla_policy sample_policy[OVS_SAMPLE_ATTR_MAX + 1] = - { - [OVS_SAMPLE_ATTR_PROBABILITY] = {.type = NLA_U32 }, - [OVS_SAMPLE_ATTR_ACTIONS] = {.type = NLA_UNSPEC }, - }; - struct nlattr *a[OVS_SAMPLE_ATTR_MAX + 1]; - int error; - - error = nla_parse_nested(a, OVS_SAMPLE_ATTR_MAX, attr, sample_policy); - if (error) - return error; + const struct nlattr *attrs[OVS_SAMPLE_ATTR_MAX + 1]; + const struct nlattr *probability, *actions; + const struct nlattr *a; + int rem; - if (!a[OVS_SAMPLE_ATTR_PROBABILITY]) + memset(attrs, 0, sizeof(attrs)); + nla_for_each_nested(a, attr, rem) { + int type = nla_type(a); + if (!type || type > OVS_SAMPLE_ATTR_MAX || attrs[type]) + return -EINVAL; + attrs[type] = a; + } + if (rem) return -EINVAL; - if (!a[OVS_SAMPLE_ATTR_ACTIONS]) + + probability = attrs[OVS_SAMPLE_ATTR_PROBABILITY]; + if (!probability || nla_len(probability) != sizeof(u32)) return -EINVAL; - return validate_actions(a[OVS_SAMPLE_ATTR_ACTIONS], key, (depth + 1)); + actions = attrs[OVS_SAMPLE_ATTR_ACTIONS]; + if (!actions || (nla_len(actions) && nla_len(actions) < NLA_HDRLEN)) + return -EINVAL; + return validate_actions(actions, key, depth + 1); } -static int validate_action_key(const struct nlattr *a, - const struct sw_flow_key *flow_key) +static int validate_set(const struct nlattr *a, + const struct sw_flow_key *flow_key) { - int act_type = nla_type(a); const struct nlattr *ovs_key = nla_data(a); int key_type = nla_type(ovs_key); @@ -535,26 +562,15 @@ static int validate_action_key(const struct nlattr *a, nla_len(ovs_key) != ovs_key_lens[key_type]) return -EINVAL; -#define ACTION(act, key) ((act << 8) | key) - - switch(ACTION(act_type, key_type)) { + switch (key_type) { const struct ovs_key_ipv4 *ipv4_key; - const struct ovs_key_8021q *q_key; - - case ACTION(OVS_ACTION_ATTR_SET, OVS_KEY_ATTR_TUN_ID): - case ACTION(OVS_ACTION_ATTR_SET, OVS_KEY_ATTR_ETHERNET): - break; - - case ACTION(OVS_ACTION_ATTR_PUSH, OVS_KEY_ATTR_8021Q): - q_key = nla_data(ovs_key); - if (q_key->q_tpid != htons(ETH_P_8021Q)) - return -EINVAL; - if (q_key->q_tci & htons(VLAN_TAG_PRESENT)) - return -EINVAL; + case OVS_KEY_ATTR_PRIORITY: + case OVS_KEY_ATTR_TUN_ID: + case OVS_KEY_ATTR_ETHERNET: break; - case ACTION(OVS_ACTION_ATTR_SET, OVS_KEY_ATTR_IPV4): + case OVS_KEY_ATTR_IPV4: if (flow_key->eth.type != htons(ETH_P_IP)) return -EINVAL; @@ -565,11 +581,12 @@ static int validate_action_key(const struct nlattr *a, if (ipv4_key->ipv4_proto != flow_key->ip.proto) return -EINVAL; - if (ipv4_key->ipv4_tos & INET_ECN_MASK) + if (ipv4_key->ipv4_frag != flow_key->ip.frag) return -EINVAL; + break; - case ACTION(OVS_ACTION_ATTR_SET, OVS_KEY_ATTR_TCP): + case OVS_KEY_ATTR_TCP: if (flow_key->ip.proto != IPPROTO_TCP) return -EINVAL; @@ -578,7 +595,7 @@ static int validate_action_key(const struct nlattr *a, break; - case ACTION(OVS_ACTION_ATTR_SET, OVS_KEY_ATTR_UDP): + case OVS_KEY_ATTR_UDP: if (flow_key->ip.proto != IPPROTO_UDP) return -EINVAL; @@ -589,25 +606,26 @@ static int validate_action_key(const struct nlattr *a, default: return -EINVAL; } -#undef ACTION + return 0; } static int validate_userspace(const struct nlattr *attr) { - static const struct nla_policy userspace_policy[OVS_USERSPACE_ATTR_MAX + 1] = - { + static const struct nla_policy userspace_policy[OVS_USERSPACE_ATTR_MAX + 1] = { [OVS_USERSPACE_ATTR_PID] = {.type = NLA_U32 }, [OVS_USERSPACE_ATTR_USERDATA] = {.type = NLA_U64 }, }; struct nlattr *a[OVS_USERSPACE_ATTR_MAX + 1]; int error; - error = nla_parse_nested(a, OVS_USERSPACE_ATTR_MAX, attr, userspace_policy); + error = nla_parse_nested(a, OVS_USERSPACE_ATTR_MAX, + attr, userspace_policy); if (error) return error; - if (!a[OVS_USERSPACE_ATTR_PID] || !nla_get_u32(a[OVS_USERSPACE_ATTR_PID])) + if (!a[OVS_USERSPACE_ATTR_PID] || + !nla_get_u32(a[OVS_USERSPACE_ATTR_PID])) return -EINVAL; return 0; @@ -625,15 +643,14 @@ static int validate_actions(const struct nlattr *attr, nla_for_each_nested(a, attr, rem) { /* Expected argument lengths, (u32)-1 for variable length. */ static const u32 action_lens[OVS_ACTION_ATTR_MAX + 1] = { - [OVS_ACTION_ATTR_OUTPUT] = 4, + [OVS_ACTION_ATTR_OUTPUT] = sizeof(u32), [OVS_ACTION_ATTR_USERSPACE] = (u32)-1, - [OVS_ACTION_ATTR_PUSH] = (u32)-1, - [OVS_ACTION_ATTR_POP] = 2, + [OVS_ACTION_ATTR_PUSH_VLAN] = sizeof(struct ovs_action_push_vlan), + [OVS_ACTION_ATTR_POP_VLAN] = 0, [OVS_ACTION_ATTR_SET] = (u32)-1, - [OVS_ACTION_ATTR_SET_PRIORITY] = 4, - [OVS_ACTION_ATTR_POP_PRIORITY] = 0, [OVS_ACTION_ATTR_SAMPLE] = (u32)-1 }; + const struct ovs_action_push_vlan *vlan; int type = nla_type(a); if (type > OVS_ACTION_ATTR_MAX || @@ -645,11 +662,6 @@ static int validate_actions(const struct nlattr *attr, case OVS_ACTION_ATTR_UNSPEC: return -EINVAL; - case OVS_ACTION_ATTR_SET_PRIORITY: - case OVS_ACTION_ATTR_POP_PRIORITY: - /* No validation needed. */ - break; - case OVS_ACTION_ATTR_USERSPACE: err = validate_userspace(a); if (err) @@ -662,14 +674,19 @@ static int validate_actions(const struct nlattr *attr, break; - case OVS_ACTION_ATTR_POP: - if (nla_get_u16(a) != OVS_KEY_ATTR_8021Q) + case OVS_ACTION_ATTR_POP_VLAN: + break; + + case OVS_ACTION_ATTR_PUSH_VLAN: + vlan = nla_data(a); + if (vlan->vlan_tpid != htons(ETH_P_8021Q)) + return -EINVAL; + if (!(vlan->vlan_tci & htons(VLAN_TAG_PRESENT))) return -EINVAL; break; case OVS_ACTION_ATTR_SET: - case OVS_ACTION_ATTR_PUSH: - err = validate_action_key(a, key); + err = validate_set(a, key); if (err) return err; break; @@ -708,7 +725,6 @@ static int ovs_packet_cmd_execute(struct sk_buff *skb, struct genl_info *info) struct sw_flow *flow; struct datapath *dp; struct ethhdr *eth; - bool is_frag; int len; int err; int key_len; @@ -745,12 +761,13 @@ static int ovs_packet_cmd_execute(struct sk_buff *skb, struct genl_info *info) if (IS_ERR(flow)) goto err_kfree_skb; - err = flow_extract(packet, -1, &flow->key, &key_len, &is_frag); + err = flow_extract(packet, -1, &flow->key, &key_len); if (err) goto err_flow_put; - err = flow_metadata_from_nlattrs(&flow->key.eth.in_port, - &flow->key.eth.tun_id, + err = flow_metadata_from_nlattrs(&flow->key.phy.priority, + &flow->key.phy.in_port, + &flow->key.phy.tun_id, a[OVS_PACKET_ATTR_KEY]); if (err) goto err_flow_put; @@ -768,6 +785,7 @@ static int ovs_packet_cmd_execute(struct sk_buff *skb, struct genl_info *info) rcu_assign_pointer(flow->sf_acts, acts); OVS_CB(packet)->flow = flow; + packet->priority = flow->key.phy.priority; rcu_read_lock(); dp = get_dp(ovs_header->dp_ifindex); @@ -775,10 +793,6 @@ static int ovs_packet_cmd_execute(struct sk_buff *skb, struct genl_info *info) if (!dp) goto err_unlock; - if (flow->key.eth.in_port < DP_MAX_PORTS) - OVS_CB(packet)->vport = get_vport_protected(dp, - flow->key.eth.in_port); - local_bh_disable(); err = execute_actions(dp, packet); local_bh_enable(); @@ -818,7 +832,7 @@ static void get_dp_stats(struct datapath *dp, struct ovs_dp_stats *stats) stats->n_flows = flow_tbl_count(table); - stats->n_frags = stats->n_hit = stats->n_missed = stats->n_lost = 0; + stats->n_hit = stats->n_missed = stats->n_lost = 0; for_each_possible_cpu(i) { const struct dp_stats_percpu *percpu_stats; struct dp_stats_percpu local_stats; @@ -831,7 +845,6 @@ static void get_dp_stats(struct datapath *dp, struct ovs_dp_stats *stats) local_stats = *percpu_stats; } while (read_seqcount_retry(&percpu_stats->seqlock, seqcount)); - stats->n_frags += local_stats.n_frags; stats->n_hit += local_stats.n_hit; stats->n_missed += local_stats.n_missed; stats->n_lost += local_stats.n_lost; @@ -848,7 +861,7 @@ static struct genl_family dp_flow_genl_family = { .id = GENL_ID_GENERATE, .hdrsize = sizeof(struct ovs_header), .name = OVS_FLOW_FAMILY, - .version = 1, + .version = OVS_FLOW_VERSION, .maxattr = OVS_FLOW_ATTR_MAX }; @@ -858,7 +871,8 @@ static struct genl_multicast_group dp_flow_multicast_group = { /* Called with genl_lock. */ static int ovs_flow_cmd_fill_info(struct sw_flow *flow, struct datapath *dp, - struct sk_buff *skb, u32 pid, u32 seq, u32 flags, u8 cmd) + struct sk_buff *skb, u32 pid, + u32 seq, u32 flags, u8 cmd) { const int skb_orig_len = skb->len; const struct sw_flow_actions *sf_acts; @@ -897,7 +911,8 @@ static int ovs_flow_cmd_fill_info(struct sw_flow *flow, struct datapath *dp, NLA_PUT_U64(skb, OVS_FLOW_ATTR_USED, flow_used_time(used)); if (stats.n_packets) - NLA_PUT(skb, OVS_FLOW_ATTR_STATS, sizeof(struct ovs_flow_stats), &stats); + NLA_PUT(skb, OVS_FLOW_ATTR_STATS, + sizeof(struct ovs_flow_stats), &stats); if (tcp_flags) NLA_PUT_U8(skb, OVS_FLOW_ATTR_TCP_FLAGS, tcp_flags); @@ -934,15 +949,24 @@ static struct sk_buff *ovs_flow_cmd_alloc_info(struct sw_flow *flow) sf_acts = rcu_dereference_protected(flow->sf_acts, lockdep_genl_is_held()); - len = nla_total_size(FLOW_BUFSIZE); /* OVS_FLOW_ATTR_KEY */ - len += nla_total_size(sf_acts->actions_len); /* OVS_FLOW_ATTR_ACTIONS */ - len += nla_total_size(sizeof(struct ovs_flow_stats)); /* OVS_FLOW_ATTR_STATS */ - len += nla_total_size(1); /* OVS_FLOW_ATTR_TCP_FLAGS */ - len += nla_total_size(8); /* OVS_FLOW_ATTR_USED */ - return genlmsg_new(NLMSG_ALIGN(sizeof(struct ovs_header)) + len, GFP_KERNEL); + /* OVS_FLOW_ATTR_KEY */ + len = nla_total_size(FLOW_BUFSIZE); + /* OVS_FLOW_ATTR_ACTIONS */ + len += nla_total_size(sf_acts->actions_len); + /* OVS_FLOW_ATTR_STATS */ + len += nla_total_size(sizeof(struct ovs_flow_stats)); + /* OVS_FLOW_ATTR_TCP_FLAGS */ + len += nla_total_size(1); + /* OVS_FLOW_ATTR_USED */ + len += nla_total_size(8); + + len += NLMSG_ALIGN(sizeof(struct ovs_header)); + + return genlmsg_new(len, GFP_KERNEL); } -static struct sk_buff *ovs_flow_cmd_build_info(struct sw_flow *flow, struct datapath *dp, +static struct sk_buff *ovs_flow_cmd_build_info(struct sw_flow *flow, + struct datapath *dp, u32 pid, u32 seq, u8 cmd) { struct sk_buff *skb; @@ -1035,10 +1059,12 @@ static int ovs_flow_cmd_new_or_set(struct sk_buff *skb, struct genl_info *info) flow_tbl_insert(table, flow); reply = ovs_flow_cmd_build_info(flow, dp, info->snd_pid, - info->snd_seq, OVS_FLOW_CMD_NEW); + info->snd_seq, + OVS_FLOW_CMD_NEW); } else { /* We found a matching flow. */ struct sw_flow_actions *old_acts; + struct nlattr *acts_attrs; /* Bail out if we're not allowed to modify an existing flow. * We accept NLM_F_CREATE in place of the intended NLM_F_EXCL @@ -1054,13 +1080,14 @@ static int ovs_flow_cmd_new_or_set(struct sk_buff *skb, struct genl_info *info) /* Update actions. */ old_acts = rcu_dereference_protected(flow->sf_acts, lockdep_genl_is_held()); - if (a[OVS_FLOW_ATTR_ACTIONS] && - (old_acts->actions_len != nla_len(a[OVS_FLOW_ATTR_ACTIONS]) || - memcmp(old_acts->actions, nla_data(a[OVS_FLOW_ATTR_ACTIONS]), - old_acts->actions_len))) { + acts_attrs = a[OVS_FLOW_ATTR_ACTIONS]; + if (acts_attrs && + (old_acts->actions_len != nla_len(acts_attrs) || + memcmp(old_acts->actions, nla_data(acts_attrs), + old_acts->actions_len))) { struct sw_flow_actions *new_acts; - new_acts = flow_actions_alloc(a[OVS_FLOW_ATTR_ACTIONS]); + new_acts = flow_actions_alloc(acts_attrs); error = PTR_ERR(new_acts); if (IS_ERR(new_acts)) goto error; @@ -1070,7 +1097,7 @@ static int ovs_flow_cmd_new_or_set(struct sk_buff *skb, struct genl_info *info) } reply = ovs_flow_cmd_build_info(flow, dp, info->snd_pid, - info->snd_seq, OVS_FLOW_CMD_NEW); + info->snd_seq, OVS_FLOW_CMD_NEW); /* Clear stats. */ if (a[OVS_FLOW_ATTR_CLEAR]) { @@ -1082,7 +1109,7 @@ static int ovs_flow_cmd_new_or_set(struct sk_buff *skb, struct genl_info *info) if (!IS_ERR(reply)) genl_notify(reply, genl_info_net(info), info->snd_pid, - dp_flow_multicast_group.id, info->nlhdr, GFP_KERNEL); + dp_flow_multicast_group.id, info->nlhdr, GFP_KERNEL); else netlink_set_err(INIT_NET_GENL_SOCK, 0, dp_flow_multicast_group.id, PTR_ERR(reply)); @@ -1121,7 +1148,8 @@ static int ovs_flow_cmd_get(struct sk_buff *skb, struct genl_info *info) if (!flow) return -ENOENT; - reply = ovs_flow_cmd_build_info(flow, dp, info->snd_pid, info->snd_seq, OVS_FLOW_CMD_NEW); + reply = ovs_flow_cmd_build_info(flow, dp, info->snd_pid, + info->snd_seq, OVS_FLOW_CMD_NEW); if (IS_ERR(reply)) return PTR_ERR(reply); @@ -1148,7 +1176,7 @@ static int ovs_flow_cmd_del(struct sk_buff *skb, struct genl_info *info) dp = get_dp(ovs_header->dp_ifindex); if (!dp) - return -ENODEV; + return -ENODEV; table = get_table_protected(dp); flow = flow_tbl_lookup(table, &key, key_len); @@ -1191,7 +1219,8 @@ static int ovs_flow_cmd_dump(struct sk_buff *skb, struct netlink_callback *cb) if (!flow) break; - if (ovs_flow_cmd_fill_info(flow, dp, skb, NETLINK_CB(cb->skb).pid, + if (ovs_flow_cmd_fill_info(flow, dp, skb, + NETLINK_CB(cb->skb).pid, cb->nlh->nlmsg_seq, NLM_F_MULTI, OVS_FLOW_CMD_NEW) < 0) break; @@ -1231,14 +1260,13 @@ static const struct nla_policy datapath_policy[OVS_DP_ATTR_MAX + 1] = { [OVS_DP_ATTR_NAME] = { .type = NLA_NUL_STRING, .len = IFNAMSIZ - 1 }, #endif [OVS_DP_ATTR_UPCALL_PID] = { .type = NLA_U32 }, - [OVS_DP_ATTR_IPV4_FRAGS] = { .type = NLA_U32 }, }; static struct genl_family dp_datapath_genl_family = { .id = GENL_ID_GENERATE, .hdrsize = sizeof(struct ovs_header), .name = OVS_DATAPATH_FAMILY, - .version = 1, + .version = OVS_DATAPATH_VERSION, .maxattr = OVS_DP_ATTR_MAX }; @@ -1250,7 +1278,7 @@ static int ovs_dp_cmd_fill_info(struct datapath *dp, struct sk_buff *skb, u32 pid, u32 seq, u32 flags, u8 cmd) { struct ovs_header *ovs_header; - struct nlattr *nla; + struct ovs_dp_stats dp_stats; int err; ovs_header = genlmsg_put(skb, pid, seq, &dp_datapath_genl_family, @@ -1266,13 +1294,8 @@ static int ovs_dp_cmd_fill_info(struct datapath *dp, struct sk_buff *skb, if (err) goto nla_put_failure; - nla = nla_reserve(skb, OVS_DP_ATTR_STATS, sizeof(struct ovs_dp_stats)); - if (!nla) - goto nla_put_failure; - get_dp_stats(dp, nla_data(nla)); - - NLA_PUT_U32(skb, OVS_DP_ATTR_IPV4_FRAGS, - dp->drop_frags ? OVS_DP_FRAG_DROP : OVS_DP_FRAG_ZERO); + get_dp_stats(dp, &dp_stats); + NLA_PUT(skb, OVS_DP_ATTR_STATS, sizeof(struct ovs_dp_stats), &dp_stats); return genlmsg_end(skb, ovs_header); @@ -1302,18 +1325,12 @@ static struct sk_buff *ovs_dp_cmd_build_info(struct datapath *dp, u32 pid, static int ovs_dp_cmd_validate(struct nlattr *a[OVS_DP_ATTR_MAX + 1]) { - if (a[OVS_DP_ATTR_IPV4_FRAGS]) { - u32 frags = nla_get_u32(a[OVS_DP_ATTR_IPV4_FRAGS]); - - if (frags != OVS_DP_FRAG_ZERO && frags != OVS_DP_FRAG_DROP) - return -EINVAL; - } - return CHECK_NUL_STRING(a[OVS_DP_ATTR_NAME], IFNAMSIZ - 1); } /* Called with genl_mutex and optionally with RTNL lock also. */ -static struct datapath *lookup_datapath(struct ovs_header *ovs_header, struct nlattr *a[OVS_DP_ATTR_MAX + 1]) +static struct datapath *lookup_datapath(struct ovs_header *ovs_header, + struct nlattr *a[OVS_DP_ATTR_MAX + 1]) { struct datapath *dp; @@ -1330,13 +1347,6 @@ static struct datapath *lookup_datapath(struct ovs_header *ovs_header, struct nl return dp ? dp : ERR_PTR(-ENODEV); } -/* Called with genl_mutex. */ -static void change_datapath(struct datapath *dp, struct nlattr *a[OVS_DP_ATTR_MAX + 1]) -{ - if (a[OVS_DP_ATTR_IPV4_FRAGS]) - dp->drop_frags = nla_get_u32(a[OVS_DP_ATTR_IPV4_FRAGS]) == OVS_DP_FRAG_DROP; -} - static int ovs_dp_cmd_new(struct sk_buff *skb, struct genl_info *info) { struct nlattr **a = info->attrs; @@ -1376,15 +1386,12 @@ static int ovs_dp_cmd_new(struct sk_buff *skb, struct genl_info *info) if (!dp->table) goto err_free_dp; - dp->drop_frags = 0; dp->stats_percpu = alloc_percpu(struct dp_stats_percpu); if (!dp->stats_percpu) { err = -ENOMEM; goto err_destroy_table; } - change_datapath(dp, a); - /* Set up our datapath device. */ parms.name = nla_data(a[OVS_DP_ATTR_NAME]); parms.type = OVS_VPORT_TYPE_INTERNAL; @@ -1402,7 +1409,8 @@ static int ovs_dp_cmd_new(struct sk_buff *skb, struct genl_info *info) goto err_destroy_percpu; } - reply = ovs_dp_cmd_build_info(dp, info->snd_pid, info->snd_seq, OVS_DP_CMD_NEW); + reply = ovs_dp_cmd_build_info(dp, info->snd_pid, + info->snd_seq, OVS_DP_CMD_NEW); err = PTR_ERR(reply); if (IS_ERR(reply)) goto err_destroy_local_port; @@ -1449,12 +1457,13 @@ static int ovs_dp_cmd_del(struct sk_buff *skb, struct genl_info *info) if (IS_ERR(dp)) goto exit_unlock; - reply = ovs_dp_cmd_build_info(dp, info->snd_pid, info->snd_seq, OVS_DP_CMD_DEL); + reply = ovs_dp_cmd_build_info(dp, info->snd_pid, + info->snd_seq, OVS_DP_CMD_DEL); err = PTR_ERR(reply); if (IS_ERR(reply)) goto exit_unlock; - list_for_each_entry_safe (vport, next_vport, &dp->port_list, node) + list_for_each_entry_safe(vport, next_vport, &dp->port_list, node) if (vport->port_no != OVSP_LOCAL) dp_detach_port(vport); @@ -1497,9 +1506,8 @@ static int ovs_dp_cmd_set(struct sk_buff *skb, struct genl_info *info) if (IS_ERR(dp)) return PTR_ERR(dp); - change_datapath(dp, info->attrs); - - reply = ovs_dp_cmd_build_info(dp, info->snd_pid, info->snd_seq, OVS_DP_CMD_NEW); + reply = ovs_dp_cmd_build_info(dp, info->snd_pid, + info->snd_seq, OVS_DP_CMD_NEW); if (IS_ERR(reply)) { err = PTR_ERR(reply); netlink_set_err(INIT_NET_GENL_SOCK, 0, @@ -1526,7 +1534,8 @@ static int ovs_dp_cmd_get(struct sk_buff *skb, struct genl_info *info) if (IS_ERR(dp)) return PTR_ERR(dp); - reply = ovs_dp_cmd_build_info(dp, info->snd_pid, info->snd_seq, OVS_DP_CMD_NEW); + reply = ovs_dp_cmd_build_info(dp, info->snd_pid, + info->snd_seq, OVS_DP_CMD_NEW); if (IS_ERR(reply)) return PTR_ERR(reply); @@ -1539,7 +1548,7 @@ static int ovs_dp_cmd_dump(struct sk_buff *skb, struct netlink_callback *cb) int skip = cb->args[0]; int i = 0; - list_for_each_entry (dp, &dps, list_node) { + list_for_each_entry(dp, &dps, list_node) { if (i < skip) continue; if (ovs_dp_cmd_fill_info(dp, skb, NETLINK_CB(cb->skb).pid, @@ -1597,7 +1606,7 @@ static struct genl_family dp_vport_genl_family = { .id = GENL_ID_GENERATE, .hdrsize = sizeof(struct ovs_header), .name = OVS_VPORT_FAMILY, - .version = 1, + .version = OVS_VPORT_VERSION, .maxattr = OVS_VPORT_ATTR_MAX }; @@ -1610,7 +1619,7 @@ static int ovs_vport_cmd_fill_info(struct vport *vport, struct sk_buff *skb, u32 pid, u32 seq, u32 flags, u8 cmd) { struct ovs_header *ovs_header; - struct nlattr *nla; + struct ovs_vport_stats vport_stats; int err; ovs_header = genlmsg_put(skb, pid, seq, &dp_vport_genl_family, @@ -1621,17 +1630,16 @@ static int ovs_vport_cmd_fill_info(struct vport *vport, struct sk_buff *skb, ovs_header->dp_ifindex = get_dpifindex(vport->dp); NLA_PUT_U32(skb, OVS_VPORT_ATTR_PORT_NO, vport->port_no); - NLA_PUT_U32(skb, OVS_VPORT_ATTR_TYPE, vport_get_type(vport)); - NLA_PUT_STRING(skb, OVS_VPORT_ATTR_NAME, vport_get_name(vport)); + NLA_PUT_U32(skb, OVS_VPORT_ATTR_TYPE, vport->ops->type); + NLA_PUT_STRING(skb, OVS_VPORT_ATTR_NAME, vport->ops->get_name(vport)); NLA_PUT_U32(skb, OVS_VPORT_ATTR_UPCALL_PID, vport->upcall_pid); - nla = nla_reserve(skb, OVS_VPORT_ATTR_STATS, sizeof(struct ovs_vport_stats)); - if (!nla) - goto nla_put_failure; + vport_get_stats(vport, &vport_stats); + NLA_PUT(skb, OVS_VPORT_ATTR_STATS, sizeof(struct ovs_vport_stats), + &vport_stats); - vport_get_stats(vport, nla_data(nla)); - - NLA_PUT(skb, OVS_VPORT_ATTR_ADDRESS, ETH_ALEN, vport_get_addr(vport)); + NLA_PUT(skb, OVS_VPORT_ATTR_ADDRESS, ETH_ALEN, + vport->ops->get_addr(vport)); err = vport_get_options(vport, skb); if (err == -EMSGSIZE) @@ -1701,7 +1709,8 @@ static struct vport *lookup_vport(struct ovs_header *ovs_header, } /* Called with RTNL lock. */ -static int change_vport(struct vport *vport, struct nlattr *a[OVS_VPORT_ATTR_MAX + 1]) +static int change_vport(struct vport *vport, + struct nlattr *a[OVS_VPORT_ATTR_MAX + 1]) { int err = 0; @@ -1775,12 +1784,13 @@ static int ovs_vport_cmd_new(struct sk_buff *skb, struct genl_info *info) if (IS_ERR(vport)) goto exit_unlock; - dp_sysfs_add_if(vport); + dp_sysfs_add_if(vport); err = change_vport(vport, a); if (!err) { reply = ovs_vport_cmd_build_info(vport, info->snd_pid, - info->snd_seq, OVS_VPORT_CMD_NEW); + info->snd_seq, + OVS_VPORT_CMD_NEW); if (IS_ERR(reply)) err = PTR_ERR(reply); } @@ -1816,8 +1826,10 @@ static int ovs_vport_cmd_set(struct sk_buff *skb, struct genl_info *info) goto exit_unlock; err = 0; - if (a[OVS_VPORT_ATTR_TYPE] && nla_get_u32(a[OVS_VPORT_ATTR_TYPE]) != vport_get_type(vport)) + if (a[OVS_VPORT_ATTR_TYPE] && + nla_get_u32(a[OVS_VPORT_ATTR_TYPE]) != vport->ops->type) err = -EINVAL; + if (!err && a[OVS_VPORT_ATTR_OPTIONS]) err = vport_set_options(vport, a[OVS_VPORT_ATTR_OPTIONS]); if (!err) @@ -2039,7 +2051,8 @@ static int __init dp_init(void) BUILD_BUG_ON(sizeof(struct ovs_skb_cb) > sizeof(dummy_skb->cb)); - printk("Open vSwitch %s, built "__DATE__" "__TIME__"\n", VERSION BUILDNR); + pr_info("Open vSwitch switching datapath %s, built "__DATE__" "__TIME__"\n", + VERSION BUILDNR); err = tnl_init(); if (err)