X-Git-Url: http://git.onelab.eu/?a=blobdiff_plain;f=ofproto%2Fin-band.c;h=e75d19eac340452de7a9752dfae516b2744698ec;hb=f7eaaeba017b2442edd730b929d4713ba4375b1e;hp=82e51c726897cba787bbb834d9523fc8c3a4151b;hpb=0ade584ebe8ca1a1de92fb7422566726f276918c;p=sliver-openvswitch.git diff --git a/ofproto/in-band.c b/ofproto/in-band.c index 82e51c726..e75d19eac 100644 --- a/ofproto/in-band.c +++ b/ofproto/in-band.c @@ -1,5 +1,5 @@ /* - * Copyright (c) 2008, 2009, 2010 Nicira Networks. + * Copyright (c) 2008, 2009, 2010, 2011 Nicira Networks. * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. @@ -19,33 +19,30 @@ #include #include #include +#include #include #include #include +#include "classifier.h" #include "dhcp.h" #include "dpif.h" #include "flow.h" -#include "mac-learning.h" #include "netdev.h" +#include "netlink.h" #include "odp-util.h" -#include "ofp-print.h" #include "ofproto.h" #include "ofpbuf.h" #include "openflow/openflow.h" -#include "openvswitch/datapath-protocol.h" #include "packets.h" #include "poll-loop.h" -#include "rconn.h" -#include "status.h" #include "timeval.h" -#include "vconn.h" - -#define THIS_MODULE VLM_in_band #include "vlog.h" -/* In-band control allows a single network to be used for OpenFlow - * traffic and other data traffic. Refer to ovs-vswitchd.conf(5) and - * secchan(8) for a description of configuring in-band control. +VLOG_DEFINE_THIS_MODULE(in_band); + +/* In-band control allows a single network to be used for OpenFlow traffic and + * other data traffic. See ovs-vswitchd.conf.db(5) for a description of + * configuring in-band control. * * This comment is an attempt to describe how in-band control works at a * wire- and implementation-level. Correctly implementing in-band @@ -53,34 +50,49 @@ * gone through many iterations. Please read through and understand the * reasoning behind the chosen rules before making modifications. * - * In Open vSwitch, in-band control is implemented as "hidden" flows (in - * that they are not visible through OpenFlow) and at a higher priority - * than wildcarded flows can be set up by the controller. This is done - * so that the controller cannot interfere with them and possibly break - * connectivity with its switches. It is possible to see all flows, - * including in-band ones, with the ovs-appctl "bridge/dump-flows" - * command. + * In Open vSwitch, in-band control is implemented as "hidden" flows (in that + * they are not visible through OpenFlow) and at a higher priority than + * wildcarded flows can be set up by through OpenFlow. This is done so that + * the OpenFlow controller cannot interfere with them and possibly break + * connectivity with its switches. It is possible to see all flows, including + * in-band ones, with the ovs-appctl "bridge/dump-flows" command. + * + * The Open vSwitch implementation of in-band control can hide traffic to + * arbitrary "remotes", where each remote is one TCP port on one IP address. + * Currently the remotes are automatically configured as the in-band OpenFlow + * controllers plus the OVSDB managers, if any. (The latter is a requirement + * because OVSDB managers are responsible for configuring OpenFlow controllers, + * so if the manager cannot be reached then OpenFlow cannot be reconfigured.) + * + * The following rules (with the OFPP_NORMAL action) are set up on any bridge + * that has any remotes: + * + * (a) DHCP requests sent from the local port. + * (b) ARP replies to the local port's MAC address. + * (c) ARP requests from the local port's MAC address. + * + * In-band also sets up the following rules for each unique next-hop MAC + * address for the remotes' IPs (the "next hop" is either the remote + * itself, if it is on a local subnet, or the gateway to reach the remote): + * + * (d) ARP replies to the next hop's MAC address. + * (e) ARP requests from the next hop's MAC address. + * + * In-band also sets up the following rules for each unique remote IP address: + * + * (f) ARP replies containing the remote's IP address as a target. + * (g) ARP requests containing the remote's IP address as a source. * - * The following rules are always enabled with the "normal" action by a - * switch with in-band control: + * In-band also sets up the following rules for each unique remote (IP,port) + * pair: * - * a. DHCP requests sent from the local port. - * b. ARP replies to the local port's MAC address. - * c. ARP requests from the local port's MAC address. - * d. ARP replies to the remote side's MAC address. Note that the - * remote side is either the controller or the gateway to reach - * the controller. - * e. ARP requests from the remote side's MAC address. Note that - * like (d), the MAC is either for the controller or gateway. - * f. ARP replies containing the controller's IP address as a target. - * g. ARP requests containing the controller's IP address as a source. - * h. OpenFlow (6633/tcp) traffic to the controller's IP. - * i. OpenFlow (6633/tcp) traffic from the controller's IP. + * (h) TCP traffic to the remote's IP and port. + * (i) TCP traffic from the remote's IP and port. * * The goal of these rules is to be as narrow as possible to allow a - * switch to join a network and be able to communicate with a - * controller. As mentioned earlier, these rules have higher priority - * than the controller's rules, so if they are too broad, they may + * switch to join a network and be able to communicate with the + * remotes. As mentioned earlier, these rules have higher priority + * than the controller's rules, so if they are too broad, they may * prevent the controller from implementing its policy. As such, * in-band actively monitors some aspects of flow and packet processing * so that the rules can be made more precise. @@ -90,106 +102,101 @@ * match entries, so in-band control is able to be very precise about * the flows it prevents. Flows that miss in the datapath are sent to * userspace to be processed, so preventing these flows from being - * cached in the "fast path" does not affect correctness. The only type - * of flow that is currently prevented is one that would prevent DHCP - * replies from being seen by the local port. For example, a rule that - * forwarded all DHCP traffic to the controller would not be allowed, + * cached in the "fast path" does not affect correctness. The only type + * of flow that is currently prevented is one that would prevent DHCP + * replies from being seen by the local port. For example, a rule that + * forwarded all DHCP traffic to the controller would not be allowed, * but one that forwarded to all ports (including the local port) would. * * As mentioned earlier, packets that miss in the datapath are sent to * the userspace for processing. The userspace has its own flow table, - * the "classifier", so in-band checks whether any special processing - * is needed before the classifier is consulted. If a packet is a DHCP - * response to a request from the local port, the packet is forwarded to - * the local port, regardless of the flow table. Note that this requires - * L7 processing of DHCP replies to determine whether the 'chaddr' field + * the "classifier", so in-band checks whether any special processing + * is needed before the classifier is consulted. If a packet is a DHCP + * response to a request from the local port, the packet is forwarded to + * the local port, regardless of the flow table. Note that this requires + * L7 processing of DHCP replies to determine whether the 'chaddr' field * matches the MAC address of the local port. * * It is interesting to note that for an L3-based in-band control - * mechanism, the majority of rules are devoted to ARP traffic. At first - * glance, some of these rules appear redundant. However, each serves an - * important role. First, in order to determine the MAC address of the - * remote side (controller or gateway) for other ARP rules, we must allow - * ARP traffic for our local port with rules (b) and (c). If we are - * between a switch and its connection to the controller, we have to - * allow the other switch's ARP traffic to through. This is done with + * mechanism, the majority of rules are devoted to ARP traffic. At first + * glance, some of these rules appear redundant. However, each serves an + * important role. First, in order to determine the MAC address of the + * remote side (controller or gateway) for other ARP rules, we must allow + * ARP traffic for our local port with rules (b) and (c). If we are + * between a switch and its connection to the remote, we have to + * allow the other switch's ARP traffic to through. This is done with * rules (d) and (e), since we do not know the addresses of the other - * switches a priori, but do know the controller's or gateway's. Finally, - * if the controller is running in a local guest VM that is not reached - * through the local port, the switch that is connected to the VM must - * allow ARP traffic based on the controller's IP address, since it will - * not know the MAC address of the local port that is sending the traffic - * or the MAC address of the controller in the guest VM. + * switches a priori, but do know the remote's or gateway's. Finally, + * if the remote is running in a local guest VM that is not reached + * through the local port, the switch that is connected to the VM must + * allow ARP traffic based on the remote's IP address, since it will + * not know the MAC address of the local port that is sending the traffic + * or the MAC address of the remote in the guest VM. * * With a few notable exceptions below, in-band should work in most * network setups. The following are considered "supported' in the - * current implementation: + * current implementation: * - * - Locally Connected. The switch and controller are on the same + * - Locally Connected. The switch and remote are on the same * subnet. This uses rules (a), (b), (c), (h), and (i). * - * - Reached through Gateway. The switch and controller are on + * - Reached through Gateway. The switch and remote are on * different subnets and must go through a gateway. This uses * rules (a), (b), (c), (h), and (i). * - * - Between Switch and Controller. This switch is between another - * switch and the controller, and we want to allow the other + * - Between Switch and Remote. This switch is between another + * switch and the remote, and we want to allow the other * switch's traffic through. This uses rules (d), (e), (h), and * (i). It uses (b) and (c) indirectly in order to know the MAC * address for rules (d) and (e). Note that DHCP for the other - * switch will not work unless the controller explicitly lets this + * switch will not work unless an OpenFlow controller explicitly lets this * switch pass the traffic. * * - Between Switch and Gateway. This switch is between another * switch and the gateway, and we want to allow the other switch's * traffic through. This uses the same rules and logic as the - * "Between Switch and Controller" configuration described earlier. + * "Between Switch and Remote" configuration described earlier. * - * - Controller on Local VM. The controller is a guest VM on the - * system running in-band control. This uses rules (a), (b), (c), + * - Remote on Local VM. The remote is a guest VM on the + * system running in-band control. This uses rules (a), (b), (c), * (h), and (i). * - * - Controller on Local VM with Different Networks. The controller + * - Remote on Local VM with Different Networks. The remote * is a guest VM on the system running in-band control, but the - * local port is not used to connect to the controller. For + * local port is not used to connect to the remote. For * example, an IP address is configured on eth0 of the switch. The - * controller's VM is connected through eth1 of the switch, but an + * remote's VM is connected through eth1 of the switch, but an * IP address has not been configured for that port on the switch. - * As such, the switch will use eth0 to connect to the controller, + * As such, the switch will use eth0 to connect to the remote, * and eth1's rules about the local port will not work. In the - * example, the switch attached to eth0 would use rules (a), (b), - * (c), (h), and (i) on eth0. The switch attached to eth1 would use + * example, the switch attached to eth0 would use rules (a), (b), + * (c), (h), and (i) on eth0. The switch attached to eth1 would use * rules (f), (g), (h), and (i). * * The following are explicitly *not* supported by in-band control: * - * - Specify Controller by Name. Currently, the controller must be + * - Specify Remote by Name. Currently, the remote must be * identified by IP address. A naive approach would be to permit * all DNS traffic. Unfortunately, this would prevent the * controller from defining any policy over DNS. Since switches - * that are located behind us need to connect to the controller, + * that are located behind us need to connect to the remote, * in-band cannot simply add a rule that allows DNS traffic from * the local port. The "correct" way to support this is to parse * DNS requests to allow all traffic related to a request for the - * controller's name through. Due to the potential security + * remote's name through. Due to the potential security * problems and amount of processing, we decided to hold off for * the time-being. * - * - Multiple Controllers. There is nothing intrinsic in the high- - * level design that prevents using multiple (known) controllers, - * however, the current implementation's data structures assume - * only one. - * - * - Differing Controllers for Switches. All switches must know - * the L3 addresses for all the controllers that other switches + * - Differing Remotes for Switches. All switches must know + * the L3 addresses for all the remotes that other switches * may use, since rules need to be set up to allow traffic related - * to those controllers through. See rules (f), (g), (h), and (i). + * to those remotes through. See rules (f), (g), (h), and (i). * - * - Differing Routes for Switches. In order for the switch to - * allow other switches to connect to a controller through a + * - Differing Routes for Switches. In order for the switch to + * allow other switches to connect to a remote through a * gateway, it allows the gateway's traffic through with rules (d) - * and (e). If the routes to the controller differ for the two - * switches, we will not know the MAC address of the alternate + * and (e). If the routes to the remote differ for the two + * switches, we will not know the MAC address of the alternate * gateway. */ @@ -199,27 +206,27 @@ * by in-band control have the same action. The only reason to use more than * one priority is to make the kind of flow easier to see during debugging. */ enum { + /* One set per bridge. */ IBR_FROM_LOCAL_DHCP = 180000, /* (a) From local port, DHCP. */ IBR_TO_LOCAL_ARP, /* (b) To local port, ARP. */ IBR_FROM_LOCAL_ARP, /* (c) From local port, ARP. */ - IBR_TO_REMOTE_ARP, /* (d) To remote MAC, ARP. */ - IBR_FROM_REMOTE_ARP, /* (e) From remote MAC, ARP. */ - IBR_TO_CTL_ARP, /* (f) To controller IP, ARP. */ - IBR_FROM_CTL_ARP, /* (g) From controller IP, ARP. */ - IBR_TO_CTL_OFP, /* (h) To controller, OpenFlow port. */ - IBR_FROM_CTL_OFP /* (i) From controller, OpenFlow port. */ -}; -struct in_band_rule { - flow_t flow; - uint32_t wildcards; - unsigned int priority; + /* One set per unique next-hop MAC. */ + IBR_TO_NEXT_HOP_ARP, /* (d) To remote MAC, ARP. */ + IBR_FROM_NEXT_HOP_ARP, /* (e) From remote MAC, ARP. */ + + /* One set per unique remote IP address. */ + IBR_TO_REMOTE_ARP, /* (f) To remote IP, ARP. */ + IBR_FROM_REMOTE_ARP, /* (g) From remote IP, ARP. */ + + /* One set per unique remote (IP,port) pair. */ + IBR_TO_REMOTE_TCP, /* (h) To remote IP, TCP port. */ + IBR_FROM_REMOTE_TCP /* (i) From remote IP, TCP port. */ }; /* Track one remote IP and next hop information. */ struct in_band_remote { - struct rconn *rconn; /* Connection to remote. */ - uint32_t remote_ip; /* Remote IP, 0 if unknown. */ + struct sockaddr_in remote_addr; /* IP address, in network byte order. */ uint8_t remote_mac[ETH_ADDR_LEN]; /* Next-hop MAC, all-zeros if unknown. */ uint8_t last_remote_mac[ETH_ADDR_LEN]; /* Previous nonzero next-hop MAC. */ struct netdev *remote_netdev; /* Device to send to next-hop MAC. */ @@ -227,7 +234,7 @@ struct in_band_remote { struct in_band { struct ofproto *ofproto; - struct status_category *ss_cat; + int queue_id, prev_queue_id; /* Remote information. */ time_t next_remote_refresh; /* Refresh timer. */ @@ -241,8 +248,8 @@ struct in_band { /* Local and remote addresses that are installed as flows. */ uint8_t installed_local_mac[ETH_ADDR_LEN]; - uint32_t *remote_ips; - uint32_t n_remote_ips; + struct sockaddr_in *remote_addrs; + size_t n_remote_addrs; uint8_t *remote_macs; size_t n_remote_macs; }; @@ -252,29 +259,24 @@ static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(60, 60); static int refresh_remote(struct in_band *ib, struct in_band_remote *r) { - struct in_addr remote_inaddr; struct in_addr next_hop_inaddr; char *next_hop_dev; int retval; - /* Get remote IP address. */ - r->remote_ip = rconn_get_remote_ip(r->rconn); - /* Find the next-hop IP address. */ - remote_inaddr.s_addr = r->remote_ip; memset(r->remote_mac, 0, sizeof r->remote_mac); - retval = netdev_get_next_hop(ib->local_netdev, &remote_inaddr, + retval = netdev_get_next_hop(ib->local_netdev, &r->remote_addr.sin_addr, &next_hop_inaddr, &next_hop_dev); if (retval) { VLOG_WARN("cannot find route for controller ("IP_FMT"): %s", - IP_ARGS(&r->remote_ip), strerror(retval)); + IP_ARGS(&r->remote_addr.sin_addr), strerror(retval)); return 1; } if (!next_hop_inaddr.s_addr) { - next_hop_inaddr.s_addr = remote_inaddr.s_addr; + next_hop_inaddr = r->remote_addr.sin_addr; } - /* Get the next-hop IP and network device. */ + /* Open the next-hop network device. */ if (!r->remote_netdev || strcmp(netdev_get_name(r->remote_netdev), next_hop_dev)) { @@ -284,7 +286,7 @@ refresh_remote(struct in_band *ib, struct in_band_remote *r) if (retval) { VLOG_WARN_RL(&rl, "cannot open netdev %s (next hop " "to controller "IP_FMT"): %s", - next_hop_dev, IP_ARGS(&r->remote_ip), + next_hop_dev, IP_ARGS(&r->remote_addr.sin_addr), strerror(retval)); free(next_hop_dev); return 1; @@ -300,10 +302,10 @@ refresh_remote(struct in_band *ib, struct in_band_remote *r) IP_ARGS(&next_hop_inaddr.s_addr), strerror(retval)); } - /* If we have an IP address but not a MAC address, then refresh quickly, - * since we probably will get a MAC address soon (via ARP). Otherwise, we - * can afford to wait a little while. */ - return r->remote_ip && eth_addr_is_zero(r->remote_mac) ? 1 : 10; + /* If we don't have a MAC address, then refresh quickly, since we probably + * will get a MAC address soon (via ARP). Otherwise, we can afford to wait + * a little while. */ + return eth_addr_is_zero(r->remote_mac) ? 1 : 10; } static bool @@ -311,35 +313,25 @@ refresh_remotes(struct in_band *ib) { struct in_band_remote *r; bool any_changes; - int min_refresh; if (time_now() < ib->next_remote_refresh) { return false; } any_changes = false; - min_refresh = 10; + ib->next_remote_refresh = TIME_MAX; for (r = ib->remotes; r < &ib->remotes[ib->n_remotes]; r++) { uint8_t old_remote_mac[ETH_ADDR_LEN]; - uint32_t old_remote_ip; - int refresh_interval; + time_t next_refresh; - /* Save old remote information. */ - old_remote_ip = r->remote_ip; + /* Save old MAC. */ memcpy(old_remote_mac, r->remote_mac, ETH_ADDR_LEN); /* Refresh remote information. */ - refresh_interval = refresh_remote(ib, r); - min_refresh = MIN(min_refresh, refresh_interval); + next_refresh = refresh_remote(ib, r) + time_now(); + ib->next_remote_refresh = MIN(ib->next_remote_refresh, next_refresh); - /* If anything changed, log the changes. */ - if (old_remote_ip != r->remote_ip) { - any_changes = true; - if (r->remote_ip) { - VLOG_DBG("remote IP address changed from "IP_FMT" to "IP_FMT, - IP_ARGS(&old_remote_ip), IP_ARGS(&r->remote_ip)); - } - } + /* If the MAC changed, log the changes. */ if (!eth_addr_equals(r->remote_mac, old_remote_mac)) { any_changes = true; if (!eth_addr_is_zero(r->remote_mac) @@ -352,7 +344,6 @@ refresh_remotes(struct in_band *ib) } } } - ib->next_remote_refresh = time_now() + min_refresh; return any_changes; } @@ -380,37 +371,16 @@ refresh_local(struct in_band *ib) return true; } -static void -in_band_status_cb(struct status_reply *sr, void *in_band_) -{ - struct in_band *in_band = in_band_; - - if (!eth_addr_is_zero(in_band->local_mac)) { - status_reply_put(sr, "local-mac="ETH_ADDR_FMT, - ETH_ADDR_ARGS(in_band->local_mac)); - } - - if (in_band->n_remotes - && !eth_addr_is_zero(in_band->remotes[0].remote_mac)) { - status_reply_put(sr, "remote-mac="ETH_ADDR_FMT, - ETH_ADDR_ARGS(in_band->remotes[0].remote_mac)); - } -} - /* Returns true if 'packet' should be sent to the local port regardless - * of the flow table. */ + * of the flow table. */ bool -in_band_msg_in_hook(struct in_band *in_band, const flow_t *flow, +in_band_msg_in_hook(struct in_band *in_band, const struct flow *flow, const struct ofpbuf *packet) { - if (!in_band) { - return false; - } - /* Regardless of how the flow table is configured, we want to be * able to see replies to our DHCP requests. */ if (flow->dl_type == htons(ETH_TYPE_IP) - && flow->nw_proto == IP_TYPE_UDP + && flow->nw_proto == IPPROTO_UDP && flow->tp_src == htons(DHCP_SERVER_PORT) && flow->tp_dst == htons(DHCP_CLIENT_PORT) && packet->l7) { @@ -432,29 +402,26 @@ in_band_msg_in_hook(struct in_band *in_band, const flow_t *flow, return false; } -/* Returns true if the rule that would match 'flow' with 'actions' is +/* Returns true if the rule that would match 'flow' with 'actions' is * allowed to be set up in the datapath. */ bool -in_band_rule_check(struct in_band *in_band, const flow_t *flow, - const struct odp_actions *actions) +in_band_rule_check(const struct flow *flow, + const struct nlattr *actions, size_t actions_len) { - if (!in_band) { - return true; - } - /* Don't allow flows that would prevent DHCP replies from being seen * by the local port. */ if (flow->dl_type == htons(ETH_TYPE_IP) - && flow->nw_proto == IP_TYPE_UDP - && flow->tp_src == htons(DHCP_SERVER_PORT) + && flow->nw_proto == IPPROTO_UDP + && flow->tp_src == htons(DHCP_SERVER_PORT) && flow->tp_dst == htons(DHCP_CLIENT_PORT)) { - int i; + const struct nlattr *a; + unsigned int left; - for (i=0; in_actions; i++) { - if (actions->actions[i].output.type == ODPAT_OUTPUT - && actions->actions[i].output.port == ODPP_LOCAL) { + NL_ATTR_FOR_EACH_UNSAFE (a, left, actions, actions_len) { + if (nl_attr_type(a) == ODP_ACTION_ATTR_OUTPUT + && nl_attr_get_u32(a) == ODPP_LOCAL) { return true; - } + } } return false; } @@ -462,109 +429,36 @@ in_band_rule_check(struct in_band *in_band, const flow_t *flow, return true; } -static void -init_rule(struct in_band_rule *rule, unsigned int priority) -{ - rule->wildcards = OVSFW_ALL; - rule->priority = priority; - - /* Not strictly necessary but seems cleaner. */ - memset(&rule->flow, 0, sizeof rule->flow); -} - -static void -set_in_port(struct in_band_rule *rule, uint16_t odp_port) -{ - rule->wildcards &= ~OFPFW_IN_PORT; - rule->flow.in_port = odp_port; -} - -static void -set_dl_type(struct in_band_rule *rule, uint16_t dl_type) -{ - rule->wildcards &= ~OFPFW_DL_TYPE; - rule->flow.dl_type = htons(dl_type); -} - -static void -set_dl_src(struct in_band_rule *rule, const uint8_t dl_src[ETH_ADDR_LEN]) -{ - rule->wildcards &= ~OFPFW_DL_SRC; - memcpy(rule->flow.dl_src, dl_src, ETH_ADDR_LEN); -} - -static void -set_dl_dst(struct in_band_rule *rule, const uint8_t dl_dst[ETH_ADDR_LEN]) -{ - rule->wildcards &= ~OFPFW_DL_DST; - memcpy(rule->flow.dl_dst, dl_dst, ETH_ADDR_LEN); -} - -static void -set_tp_src(struct in_band_rule *rule, uint16_t tp_src) -{ - rule->wildcards &= ~OFPFW_TP_SRC; - rule->flow.tp_src = htons(tp_src); -} - -static void -set_tp_dst(struct in_band_rule *rule, uint16_t tp_dst) -{ - rule->wildcards &= ~OFPFW_TP_DST; - rule->flow.tp_dst = htons(tp_dst); -} - -static void -set_nw_proto(struct in_band_rule *rule, uint8_t nw_proto) -{ - rule->wildcards &= ~OFPFW_NW_PROTO; - rule->flow.nw_proto = nw_proto; -} - -static void -set_nw_src(struct in_band_rule *rule, uint32_t nw_src) -{ - rule->wildcards &= ~OFPFW_NW_SRC_MASK; - rule->flow.nw_src = nw_src; -} - -static void -set_nw_dst(struct in_band_rule *rule, uint32_t nw_dst) -{ - rule->wildcards &= ~OFPFW_NW_DST_MASK; - rule->flow.nw_dst = nw_dst; -} - static void make_rules(struct in_band *ib, - void (*cb)(struct in_band *, const struct in_band_rule *)) + void (*cb)(struct in_band *, const struct cls_rule *)) { - struct in_band_rule rule; + struct cls_rule rule; size_t i; if (!eth_addr_is_zero(ib->installed_local_mac)) { - /* Allow DHCP requests to be sent from the local port. */ - init_rule(&rule, IBR_FROM_LOCAL_DHCP); - set_in_port(&rule, ODPP_LOCAL); - set_dl_type(&rule, ETH_TYPE_IP); - set_dl_src(&rule, ib->installed_local_mac); - set_nw_proto(&rule, IP_TYPE_UDP); - set_tp_src(&rule, DHCP_CLIENT_PORT); - set_tp_dst(&rule, DHCP_SERVER_PORT); + /* (a) Allow DHCP requests sent from the local port. */ + cls_rule_init_catchall(&rule, IBR_FROM_LOCAL_DHCP); + cls_rule_set_in_port(&rule, ODPP_LOCAL); + cls_rule_set_dl_type(&rule, htons(ETH_TYPE_IP)); + cls_rule_set_dl_src(&rule, ib->installed_local_mac); + cls_rule_set_nw_proto(&rule, IPPROTO_UDP); + cls_rule_set_tp_src(&rule, htons(DHCP_CLIENT_PORT)); + cls_rule_set_tp_dst(&rule, htons(DHCP_SERVER_PORT)); cb(ib, &rule); - /* Allow the connection's interface to receive directed ARP traffic. */ - init_rule(&rule, IBR_TO_LOCAL_ARP); - set_dl_type(&rule, ETH_TYPE_ARP); - set_dl_dst(&rule, ib->installed_local_mac); - set_nw_proto(&rule, ARP_OP_REPLY); + /* (b) Allow ARP replies to the local port's MAC address. */ + cls_rule_init_catchall(&rule, IBR_TO_LOCAL_ARP); + cls_rule_set_dl_type(&rule, htons(ETH_TYPE_ARP)); + cls_rule_set_dl_dst(&rule, ib->installed_local_mac); + cls_rule_set_nw_proto(&rule, ARP_OP_REPLY); cb(ib, &rule); - /* Allow the connection's interface to be the source of ARP traffic. */ - init_rule(&rule, IBR_FROM_LOCAL_ARP); - set_dl_type(&rule, ETH_TYPE_ARP); - set_dl_src(&rule, ib->installed_local_mac); - set_nw_proto(&rule, ARP_OP_REQUEST); + /* (c) Allow ARP requests from the local port's MAC address. */ + cls_rule_init_catchall(&rule, IBR_FROM_LOCAL_ARP); + cls_rule_set_dl_type(&rule, htons(ETH_TYPE_ARP)); + cls_rule_set_dl_src(&rule, ib->installed_local_mac); + cls_rule_set_nw_proto(&rule, ARP_OP_REQUEST); cb(ib, &rule); } @@ -579,69 +473,85 @@ make_rules(struct in_band *ib, } } - /* Allow ARP replies to the remote side's MAC. */ - init_rule(&rule, IBR_TO_REMOTE_ARP); - set_dl_type(&rule, ETH_TYPE_ARP); - set_dl_dst(&rule, remote_mac); - set_nw_proto(&rule, ARP_OP_REPLY); + /* (d) Allow ARP replies to the next hop's MAC address. */ + cls_rule_init_catchall(&rule, IBR_TO_NEXT_HOP_ARP); + cls_rule_set_dl_type(&rule, htons(ETH_TYPE_ARP)); + cls_rule_set_dl_dst(&rule, remote_mac); + cls_rule_set_nw_proto(&rule, ARP_OP_REPLY); cb(ib, &rule); - /* Allow ARP requests from the remote side's MAC. */ - init_rule(&rule, IBR_FROM_REMOTE_ARP); - set_dl_type(&rule, ETH_TYPE_ARP); - set_dl_src(&rule, remote_mac); - set_nw_proto(&rule, ARP_OP_REQUEST); + /* (e) Allow ARP requests from the next hop's MAC address. */ + cls_rule_init_catchall(&rule, IBR_FROM_NEXT_HOP_ARP); + cls_rule_set_dl_type(&rule, htons(ETH_TYPE_ARP)); + cls_rule_set_dl_src(&rule, remote_mac); + cls_rule_set_nw_proto(&rule, ARP_OP_REQUEST); cb(ib, &rule); } - for (i = 0; i < ib->n_remote_ips; i++) { - uint32_t remote_ip = ib->remote_ips[i]; - - if (i > 0 && ib->remote_ips[i - 1] == remote_ip) { - /* Skip duplicates. */ - continue; + for (i = 0; i < ib->n_remote_addrs; i++) { + const struct sockaddr_in *a = &ib->remote_addrs[i]; + + if (!i || a->sin_addr.s_addr != a[-1].sin_addr.s_addr) { + /* (f) Allow ARP replies containing the remote's IP address as a + * target. */ + cls_rule_init_catchall(&rule, IBR_TO_REMOTE_ARP); + cls_rule_set_dl_type(&rule, htons(ETH_TYPE_ARP)); + cls_rule_set_nw_proto(&rule, ARP_OP_REPLY); + cls_rule_set_nw_dst(&rule, a->sin_addr.s_addr); + cb(ib, &rule); + + /* (g) Allow ARP requests containing the remote's IP address as a + * source. */ + cls_rule_init_catchall(&rule, IBR_FROM_REMOTE_ARP); + cls_rule_set_dl_type(&rule, htons(ETH_TYPE_ARP)); + cls_rule_set_nw_proto(&rule, ARP_OP_REQUEST); + cls_rule_set_nw_src(&rule, a->sin_addr.s_addr); + cb(ib, &rule); } - /* Allow ARP replies to the controller's IP. */ - init_rule(&rule, IBR_TO_CTL_ARP); - set_dl_type(&rule, ETH_TYPE_ARP); - set_nw_proto(&rule, ARP_OP_REPLY); - set_nw_dst(&rule, remote_ip); - cb(ib, &rule); - - /* Allow ARP requests from the controller's IP. */ - init_rule(&rule, IBR_FROM_CTL_ARP); - set_dl_type(&rule, ETH_TYPE_ARP); - set_nw_proto(&rule, ARP_OP_REQUEST); - set_nw_src(&rule, remote_ip); - cb(ib, &rule); - - /* OpenFlow traffic to the controller. */ - init_rule(&rule, IBR_TO_CTL_OFP); - set_dl_type(&rule, ETH_TYPE_IP); - set_nw_proto(&rule, IP_TYPE_TCP); - set_nw_dst(&rule, remote_ip); - set_tp_dst(&rule, OFP_TCP_PORT); - cb(ib, &rule); - - /* OpenFlow traffic from the controller. */ - init_rule(&rule, IBR_FROM_CTL_OFP); - set_dl_type(&rule, ETH_TYPE_IP); - set_nw_proto(&rule, IP_TYPE_TCP); - set_nw_src(&rule, remote_ip); - set_tp_src(&rule, OFP_TCP_PORT); - cb(ib, &rule); + if (!i + || a->sin_addr.s_addr != a[-1].sin_addr.s_addr + || a->sin_port != a[-1].sin_port) { + /* (h) Allow TCP traffic to the remote's IP and port. */ + cls_rule_init_catchall(&rule, IBR_TO_REMOTE_TCP); + cls_rule_set_dl_type(&rule, htons(ETH_TYPE_IP)); + cls_rule_set_nw_proto(&rule, IPPROTO_TCP); + cls_rule_set_nw_dst(&rule, a->sin_addr.s_addr); + cls_rule_set_tp_dst(&rule, a->sin_port); + cb(ib, &rule); + + /* (i) Allow TCP traffic from the remote's IP and port. */ + cls_rule_init_catchall(&rule, IBR_FROM_REMOTE_TCP); + cls_rule_set_dl_type(&rule, htons(ETH_TYPE_IP)); + cls_rule_set_nw_proto(&rule, IPPROTO_TCP); + cls_rule_set_nw_src(&rule, a->sin_addr.s_addr); + cls_rule_set_tp_src(&rule, a->sin_port); + cb(ib, &rule); + } } } static void -clear_rules(struct in_band *ib) +drop_rule(struct in_band *ib, const struct cls_rule *rule) +{ + ofproto_delete_flow(ib->ofproto, rule); +} + +/* Drops from the flow table all of the flows set up by 'ib', then clears out + * the information about the installed flows so that they can be filled in + * again if necessary. */ +static void +drop_rules(struct in_band *ib) { + /* Drop rules. */ + make_rules(ib, drop_rule); + + /* Clear out state. */ memset(ib->installed_local_mac, 0, sizeof ib->installed_local_mac); - free(ib->remote_ips); - ib->remote_ips = NULL; - ib->n_remote_ips = 0; + free(ib->remote_addrs); + ib->remote_addrs = NULL; + ib->n_remote_addrs = 0; free(ib->remote_macs); ib->remote_macs = NULL; @@ -649,32 +559,36 @@ clear_rules(struct in_band *ib) } static void -drop_rule(struct in_band *ib, const struct in_band_rule *rule) +add_rule(struct in_band *ib, const struct cls_rule *rule) { - ofproto_delete_flow(ib->ofproto, &rule->flow, - rule->wildcards, rule->priority); -} + struct { + struct nx_action_set_queue nxsq; + struct ofp_action_output oao; + } actions; -static void -drop_rules(struct in_band *ib) -{ - make_rules(ib, drop_rule); - clear_rules(ib); -} + memset(&actions, 0, sizeof actions); -static void -add_rule(struct in_band *ib, const struct in_band_rule *rule) -{ - union ofp_action action; - - action.type = htons(OFPAT_OUTPUT); - action.output.len = htons(sizeof action); - action.output.port = htons(OFPP_NORMAL); - action.output.max_len = htons(0); - ofproto_add_flow(ib->ofproto, &rule->flow, rule->wildcards, - rule->priority, &action, 1, 0); + actions.oao.type = htons(OFPAT_OUTPUT); + actions.oao.len = htons(sizeof actions.oao); + actions.oao.port = htons(OFPP_NORMAL); + actions.oao.max_len = htons(0); + + if (ib->queue_id < 0) { + ofproto_add_flow(ib->ofproto, rule, + (union ofp_action *) &actions.oao, 1); + } else { + actions.nxsq.type = htons(OFPAT_VENDOR); + actions.nxsq.len = htons(sizeof actions.nxsq); + actions.nxsq.vendor = htonl(NX_VENDOR_ID); + actions.nxsq.subtype = htons(NXAST_SET_QUEUE); + actions.nxsq.queue_id = htonl(ib->queue_id); + + ofproto_add_flow(ib->ofproto, rule, (union ofp_action *) &actions, + sizeof actions / sizeof(union ofp_action)); + } } +/* Inserts flows into the flow table for the current state of 'ib'. */ static void add_rules(struct in_band *ib) { @@ -682,40 +596,53 @@ add_rules(struct in_band *ib) } static int -compare_ips(const void *a, const void *b) +compare_addrs(const void *a_, const void *b_) { - return memcmp(a, b, sizeof(uint32_t)); + const struct sockaddr_in *a = a_; + const struct sockaddr_in *b = b_; + int cmp; + + cmp = memcmp(&a->sin_addr.s_addr, + &b->sin_addr.s_addr, + sizeof a->sin_addr.s_addr); + if (cmp) { + return cmp; + } + return memcmp(&a->sin_port, &b->sin_port, sizeof a->sin_port); } static int compare_macs(const void *a, const void *b) { - return memcmp(a, b, ETH_ADDR_LEN); + return eth_addr_compare_3way(a, b); } void in_band_run(struct in_band *ib) { + bool local_change, remote_change, queue_id_change; struct in_band_remote *r; - if (!refresh_local(ib) && !refresh_remotes(ib)) { + local_change = refresh_local(ib); + remote_change = refresh_remotes(ib); + queue_id_change = ib->queue_id != ib->prev_queue_id; + if (!local_change && !remote_change && !queue_id_change) { /* Nothing changed, nothing to do. */ return; } + ib->prev_queue_id = ib->queue_id; /* Drop old rules. */ drop_rules(ib); /* Figure out new rules. */ memcpy(ib->installed_local_mac, ib->local_mac, ETH_ADDR_LEN); - ib->remote_ips = xmalloc(ib->n_remotes * sizeof *ib->remote_ips); - ib->n_remote_ips = 0; + ib->remote_addrs = xmalloc(ib->n_remotes * sizeof *ib->remote_addrs); + ib->n_remote_addrs = 0; ib->remote_macs = xmalloc(ib->n_remotes * ETH_ADDR_LEN); ib->n_remote_macs = 0; for (r = ib->remotes; r < &ib->remotes[ib->n_remotes]; r++) { - if (r->remote_ip) { - ib->remote_ips[ib->n_remote_ips++] = r->remote_ip; - } + ib->remote_addrs[ib->n_remote_addrs++] = r->remote_addr; if (!eth_addr_is_zero(r->remote_mac)) { memcpy(&ib->remote_macs[ib->n_remote_macs * ETH_ADDR_LEN], r->remote_mac, ETH_ADDR_LEN); @@ -724,8 +651,8 @@ in_band_run(struct in_band *ib) } /* Sort, to allow make_rules() to easily skip duplicates. */ - qsort(ib->remote_ips, ib->n_remote_ips, sizeof *ib->remote_ips, - compare_ips); + qsort(ib->remote_addrs, ib->n_remote_addrs, sizeof *ib->remote_addrs, + compare_addrs); qsort(ib->remote_macs, ib->n_remote_macs, ETH_ADDR_LEN, compare_macs); /* Add new rules. */ @@ -735,39 +662,28 @@ in_band_run(struct in_band *ib) void in_band_wait(struct in_band *in_band) { - time_t now = time_now(); - time_t wakeup + long long int wakeup = MIN(in_band->next_remote_refresh, in_band->next_local_refresh); - if (wakeup > now) { - poll_timer_wait((wakeup - now) * 1000); - } else { - poll_immediate_wake(); - } + poll_timer_wait_until(wakeup * 1000); } +/* ofproto has flushed all flows from the flow table and it is calling us back + * to allow us to reinstall the ones that are important to us. */ void in_band_flushed(struct in_band *in_band) { - clear_rules(in_band); + add_rules(in_band); } int -in_band_create(struct ofproto *ofproto, struct dpif *dpif, - struct switch_status *ss, struct in_band **in_bandp) +in_band_create(struct ofproto *ofproto, const char *local_name, + struct in_band **in_bandp) { struct in_band *in_band; - char local_name[IF_NAMESIZE]; struct netdev *local_netdev; int error; - error = dpif_port_get_name(dpif, ODPP_LOCAL, - local_name, sizeof local_name); - if (error) { - VLOG_ERR("failed to initialize in-band control: cannot get name " - "of datapath local port (%s)", strerror(error)); - return error; - } - + *in_bandp = NULL; error = netdev_open_default(local_name, &local_netdev); if (error) { VLOG_ERR("failed to initialize in-band control: cannot open " @@ -777,8 +693,7 @@ in_band_create(struct ofproto *ofproto, struct dpif *dpif, in_band = xzalloc(sizeof *in_band); in_band->ofproto = ofproto; - in_band->ss_cat = switch_status_register(ss, "in-band", - in_band_status_cb, in_band); + in_band->queue_id = in_band->prev_queue_id = -1; in_band->next_remote_refresh = TIME_MIN; in_band->next_local_refresh = TIME_MIN; in_band->local_netdev = local_netdev; @@ -794,39 +709,67 @@ in_band_destroy(struct in_band *ib) if (ib) { drop_rules(ib); in_band_set_remotes(ib, NULL, 0); - switch_status_unregister(ib->ss_cat); netdev_close(ib->local_netdev); free(ib); } } -void -in_band_set_remotes(struct in_band *ib, struct rconn **remotes, size_t n) +static bool +any_addresses_changed(struct in_band *ib, + const struct sockaddr_in *addresses, size_t n) { size_t i; - /* Optimize the case where the rconns are the same as last time. */ - if (n == ib->n_remotes) { - for (i = 0; i < n; i++) { - if (ib->remotes[i].rconn != remotes[i]) { - goto different; - } + if (n != ib->n_remotes) { + return true; + } + + for (i = 0; i < n; i++) { + const struct sockaddr_in *old = &ib->remotes[i].remote_addr; + const struct sockaddr_in *new = &addresses[i]; + + if (old->sin_addr.s_addr != new->sin_addr.s_addr || + old->sin_port != new->sin_port) { + return true; } - return; + } - different:; + return false; +} + +void +in_band_set_remotes(struct in_band *ib, + const struct sockaddr_in *addresses, size_t n) +{ + size_t i; + + if (!any_addresses_changed(ib, addresses, n)) { + return; } + /* Clear old remotes. */ for (i = 0; i < ib->n_remotes; i++) { - /* We don't own the rconn. */ netdev_close(ib->remotes[i].remote_netdev); } free(ib->remotes); - ib->next_remote_refresh = TIME_MIN; - ib->remotes = n ? xzalloc(n * sizeof *ib->remotes) : 0; + /* Set up new remotes. */ + ib->remotes = n ? xzalloc(n * sizeof *ib->remotes) : NULL; ib->n_remotes = n; for (i = 0; i < n; i++) { - ib->remotes[i].rconn = remotes[i]; + ib->remotes[i].remote_addr = addresses[i]; } + + /* Force refresh in next call to in_band_run(). */ + ib->next_remote_refresh = TIME_MIN; +} + +/* Sets the OpenFlow queue used by flows set up by 'ib' to 'queue_id'. If + * 'queue_id' is negative, 'ib' will not set any queue (which is also the + * default). */ +void +in_band_set_queue(struct in_band *ib, int queue_id) +{ + ib->queue_id = queue_id; } +