X-Git-Url: http://git.onelab.eu/?a=blobdiff_plain;f=ofproto%2Fofproto-dpif.c;h=8ce439d9fb9f761e9704f4c01e22c8d91d37984e;hb=9c4dbc1caa7f83c16775406547f9cfba38c8d513;hp=cf6e2138329b795025c3f899c359569755b0305d;hpb=bc2d31ffe9503b64d7626d53221c70da397dcfeb;p=sliver-openvswitch.git diff --git a/ofproto/ofproto-dpif.c b/ofproto/ofproto-dpif.c index cf6e21383..8ce439d9f 100644 --- a/ofproto/ofproto-dpif.c +++ b/ofproto/ofproto-dpif.c @@ -1,5 +1,5 @@ /* - * Copyright (c) 2009, 2010, 2011, 2012, 2013 Nicira, Inc. + * Copyright (c) 2009, 2010, 2011, 2012, 2013, 2014 Nicira, Inc. * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. @@ -87,11 +87,11 @@ struct rule_dpif { * - Do include packets and bytes from datapath flows which have not * recently been processed by a revalidator. */ struct ovs_mutex stats_mutex; - uint64_t packet_count OVS_GUARDED; /* Number of packets received. */ - uint64_t byte_count OVS_GUARDED; /* Number of bytes received. */ + struct dpif_flow_stats stats OVS_GUARDED; }; -static void rule_get_stats(struct rule *, uint64_t *packets, uint64_t *bytes); +static void rule_get_stats(struct rule *, uint64_t *packets, uint64_t *bytes, + long long int *used); static struct rule_dpif *rule_dpif_cast(const struct rule *); static void rule_expire(struct rule_dpif *); @@ -256,6 +256,10 @@ struct dpif_backer { * OVS_USERSPACE_ATTR_USERDATA in OVS_ACTION_ATTR_USERSPACE actions. * False if the datapath supports only 8-byte (or shorter) userdata. */ bool variable_length_userdata; + + /* Maximum number of MPLS label stack entries that the datapath supports + * in a match */ + size_t max_mpls_depth; }; /* All existing ofproto_backer instances, indexed by ofproto->up.type. */ @@ -319,6 +323,12 @@ ofproto_dpif_cast(const struct ofproto *ofproto) return CONTAINER_OF(ofproto, struct ofproto_dpif, up); } +size_t +ofproto_dpif_get_max_mpls_depth(const struct ofproto_dpif *ofproto) +{ + return ofproto->backer->max_mpls_depth; +} + static struct ofport_dpif *get_ofp_port(const struct ofproto_dpif *ofproto, ofp_port_t ofp_port); static void ofproto_trace(struct ofproto_dpif *, const struct flow *, @@ -558,7 +568,8 @@ type_run(const char *type) ofproto->netflow, ofproto->up.frag_handling, ofproto->up.forward_bpdu, connmgr_has_in_band(ofproto->up.connmgr), - ofproto->backer->variable_length_userdata); + ofproto->backer->variable_length_userdata, + ofproto->backer->max_mpls_depth); HMAP_FOR_EACH (bundle, hmap_node, &ofproto->bundles) { xlate_bundle_set(ofproto, bundle, bundle->name, @@ -781,6 +792,7 @@ struct odp_garbage { }; static bool check_variable_length_userdata(struct dpif_backer *backer); +static size_t check_max_mpls_depth(struct dpif_backer *backer); static int open_dpif_backer(const char *type, struct dpif_backer **backerp) @@ -881,6 +893,7 @@ open_dpif_backer(const char *type, struct dpif_backer **backerp) return error; } backer->variable_length_userdata = check_variable_length_userdata(backer); + backer->max_mpls_depth = check_max_mpls_depth(backer); if (backer->recv_set_enable) { udpif_set_threads(backer->udpif, n_handlers, n_revalidators); @@ -917,7 +930,7 @@ check_variable_length_userdata(struct dpif_backer *backer) ofpbuf_init(&actions, 64); start = nl_msg_start_nested(&actions, OVS_ACTION_ATTR_USERSPACE); nl_msg_put_u32(&actions, OVS_USERSPACE_ATTR_PID, - dpif_port_get_pid(backer->dpif, ODPP_NONE)); + dpif_port_get_pid(backer->dpif, ODPP_NONE, 0)); nl_msg_put_unspec_zero(&actions, OVS_USERSPACE_ATTR_USERDATA, 4); nl_msg_end_nested(&actions, start); @@ -965,6 +978,53 @@ check_variable_length_userdata(struct dpif_backer *backer) } } +/* Tests the MPLS label stack depth supported by 'backer''s datapath. + * + * Returns the number of elements in a struct flow's mpls_lse field + * if the datapath supports at least that many entries in an + * MPLS label stack. + * Otherwise returns the number of MPLS push actions supported by + * the datapath. */ +static size_t +check_max_mpls_depth(struct dpif_backer *backer) +{ + struct flow flow; + int n; + + for (n = 0; n < FLOW_MAX_MPLS_LABELS; n++) { + struct odputil_keybuf keybuf; + struct ofpbuf key; + int error; + + memset(&flow, 0, sizeof flow); + flow.dl_type = htons(ETH_TYPE_MPLS); + flow_set_mpls_bos(&flow, n, 1); + + ofpbuf_use_stack(&key, &keybuf, sizeof keybuf); + odp_flow_key_from_flow(&key, &flow, 0); + + error = dpif_flow_put(backer->dpif, DPIF_FP_CREATE | DPIF_FP_MODIFY, + key.data, key.size, NULL, 0, NULL, 0, NULL); + if (error && error != EEXIST) { + if (error != EINVAL) { + VLOG_WARN("%s: MPLS stack length feature probe failed (%s)", + dpif_name(backer->dpif), ovs_strerror(error)); + } + break; + } + + error = dpif_flow_del(backer->dpif, key.data, key.size, NULL); + if (error) { + VLOG_WARN("%s: failed to delete MPLS feature probe flow", + dpif_name(backer->dpif)); + } + } + + VLOG_INFO("%s: MPLS label stack length probed as %d", + dpif_name(backer->dpif), n); + return n; +} + static int construct(struct ofproto *ofproto_) { @@ -987,7 +1047,7 @@ construct(struct ofproto *ofproto_) ofproto->mbridge = mbridge_create(); ofproto->has_bonded_bundles = false; ofproto->lacp_enabled = false; - ovs_mutex_init(&ofproto->stats_mutex); + ovs_mutex_init_adaptive(&ofproto->stats_mutex); ovs_mutex_init(&ofproto->vsp_mutex); guarded_list_init(&ofproto->pins); @@ -1035,6 +1095,7 @@ add_internal_flow(struct ofproto_dpif *ofproto, int id, const struct ofpbuf *ofpacts, struct rule_dpif **rulep) { struct ofputil_flow_mod fm; + struct classifier *cls; int error; match_init_catchall(&fm.match); @@ -1061,12 +1122,12 @@ add_internal_flow(struct ofproto_dpif *ofproto, int id, return error; } - if (rule_dpif_lookup_in_table(ofproto, &fm.match.flow, NULL, TBL_INTERNAL, - rulep)) { - rule_dpif_unref(*rulep); - } else { - OVS_NOT_REACHED(); - } + cls = &ofproto->up.tables[TBL_INTERNAL].cls; + fat_rwlock_rdlock(&cls->rwlock); + *rulep = rule_dpif_cast(rule_from_cls_rule( + classifier_lookup(cls, &fm.match.flow, NULL))); + ovs_assert(*rulep != NULL); + fat_rwlock_unlock(&cls->rwlock); return 0; } @@ -1120,18 +1181,18 @@ destruct(struct ofproto *ofproto_) xlate_remove_ofproto(ofproto); ovs_rwlock_unlock(&xlate_rwlock); - /* Discard any flow_miss_batches queued up for 'ofproto', avoiding a - * use-after-free error. */ - udpif_revalidate(ofproto->backer->udpif); + /* Ensure that the upcall processing threads have no remaining references + * to the ofproto or anything in it. */ + udpif_synchronize(ofproto->backer->udpif); hmap_remove(&all_ofproto_dpifs, &ofproto->all_ofproto_dpifs_node); OFPROTO_FOR_EACH_TABLE (table, &ofproto->up) { struct cls_cursor cursor; - ovs_rwlock_rdlock(&table->cls.rwlock); + fat_rwlock_rdlock(&table->cls.rwlock); cls_cursor_init(&cursor, &table->cls, NULL); - ovs_rwlock_unlock(&table->cls.rwlock); + fat_rwlock_unlock(&table->cls.rwlock); CLS_CURSOR_FOR_EACH_SAFE (rule, next_rule, up.cr, &cursor) { ofproto_rule_delete(&ofproto->up, &rule->up); } @@ -1311,9 +1372,14 @@ type_get_memory_usage(const char *type, struct simap *usage) } static void -flush(struct ofproto *ofproto OVS_UNUSED) +flush(struct ofproto *ofproto_) { - udpif_flush(); + struct ofproto_dpif *ofproto = ofproto_dpif_cast(ofproto_); + struct dpif_backer *backer = ofproto->backer; + + if (backer) { + udpif_flush(backer->udpif); + } } static void @@ -1342,14 +1408,16 @@ get_tables(struct ofproto *ofproto_, struct ofp12_table_stats *ots) struct dpif_dp_stats s; uint64_t n_miss, n_no_pkt_in, n_bytes, n_dropped_frags; uint64_t n_lookup; + long long int used; strcpy(ots->name, "classifier"); dpif_get_dp_stats(ofproto->backer->dpif, &s); - rule_get_stats(&ofproto->miss_rule->up, &n_miss, &n_bytes); - rule_get_stats(&ofproto->no_packet_in_rule->up, &n_no_pkt_in, &n_bytes); - rule_get_stats(&ofproto->drop_frags_rule->up, &n_dropped_frags, &n_bytes); - + rule_get_stats(&ofproto->miss_rule->up, &n_miss, &n_bytes, &used); + rule_get_stats(&ofproto->no_packet_in_rule->up, &n_no_pkt_in, &n_bytes, + &used); + rule_get_stats(&ofproto->drop_frags_rule->up, &n_dropped_frags, &n_bytes, + &used); n_lookup = s.n_hit + s.n_missed - n_dropped_frags; ots->lookup_count = htonll(n_lookup); ots->matched_count = htonll(n_lookup - n_miss - n_no_pkt_in); @@ -1486,6 +1554,9 @@ port_destruct(struct ofport *port_) bundle_remove(port_); set_cfm(port_, NULL); set_bfd(port_, NULL); + if (port->stp_port) { + stp_port_disable(port->stp_port); + } if (ofproto->sflow) { dpif_sflow_del_port(ofproto->sflow, port->odp_port); } @@ -2851,24 +2922,39 @@ static void rule_expire(struct rule_dpif *rule) OVS_REQUIRES(ofproto_mutex) { - uint16_t idle_timeout, hard_timeout; + uint16_t hard_timeout, idle_timeout; long long int now = time_msec(); - int reason; + int reason = -1; ovs_assert(!rule->up.pending); - /* Has 'rule' expired? */ - ovs_mutex_lock(&rule->up.mutex); hard_timeout = rule->up.hard_timeout; idle_timeout = rule->up.idle_timeout; - if (hard_timeout && now > rule->up.modified + hard_timeout * 1000) { - reason = OFPRR_HARD_TIMEOUT; - } else if (idle_timeout && now > rule->up.used + idle_timeout * 1000) { - reason = OFPRR_IDLE_TIMEOUT; - } else { - reason = -1; + + /* Has 'rule' expired? */ + if (hard_timeout) { + long long int modified; + + ovs_mutex_lock(&rule->up.mutex); + modified = rule->up.modified; + ovs_mutex_unlock(&rule->up.mutex); + + if (now > modified + hard_timeout * 1000) { + reason = OFPRR_HARD_TIMEOUT; + } + } + + if (reason < 0 && idle_timeout) { + long long int used; + + ovs_mutex_lock(&rule->stats_mutex); + used = rule->stats.used; + ovs_mutex_unlock(&rule->stats_mutex); + + if (now > used + idle_timeout * 1000) { + reason = OFPRR_IDLE_TIMEOUT; + } } - ovs_mutex_unlock(&rule->up.mutex); if (reason >= 0) { COVERAGE_INC(ofproto_dpif_expired); @@ -2915,7 +3001,7 @@ ofproto_dpif_execute_actions(struct ofproto_dpif *ofproto, execute.md.tunnel = flow->tunnel; execute.md.skb_priority = flow->skb_priority; execute.md.pkt_mark = flow->pkt_mark; - execute.md.in_port = ofp_port_to_odp_port(ofproto, in_port); + execute.md.in_port.odp_port = ofp_port_to_odp_port(ofproto, in_port); execute.needs_help = (xout.slow & SLOW_ACTION) != 0; error = dpif_execute(ofproto->backer->dpif, &execute); @@ -2930,9 +3016,9 @@ rule_dpif_credit_stats(struct rule_dpif *rule, const struct dpif_flow_stats *stats) { ovs_mutex_lock(&rule->stats_mutex); - rule->packet_count += stats->n_packets; - rule->byte_count += stats->n_bytes; - rule->up.used = MAX(rule->up.used, stats->used); + rule->stats.n_packets += stats->n_packets; + rule->stats.n_bytes += stats->n_bytes; + rule->stats.used = MAX(rule->stats.used, stats->used); ovs_mutex_unlock(&rule->stats_mutex); } @@ -2948,6 +3034,12 @@ rule_dpif_is_table_miss(const struct rule_dpif *rule) return rule_is_table_miss(&rule->up); } +bool +rule_dpif_is_internal(const struct rule_dpif *rule) +{ + return rule_is_internal(&rule->up); +} + ovs_be64 rule_dpif_get_flow_cookie(const struct rule_dpif *rule) OVS_REQUIRES(rule->up.mutex) @@ -2971,67 +3063,151 @@ rule_dpif_get_actions(const struct rule_dpif *rule) return rule_get_actions(&rule->up); } -/* Lookup 'flow' in 'ofproto''s classifier. If 'wc' is non-null, sets - * the fields that were relevant as part of the lookup. */ -void +/* Lookup 'flow' in table 0 of 'ofproto''s classifier. + * If 'wc' is non-null, sets the fields that were relevant as part of + * the lookup. Returns the table_id where a match or miss occurred. + * + * The return value will be zero unless there was a miss and + * OFPTC_TABLE_MISS_CONTINUE is in effect for the sequence of tables + * where misses occur. */ +uint8_t rule_dpif_lookup(struct ofproto_dpif *ofproto, const struct flow *flow, struct flow_wildcards *wc, struct rule_dpif **rule) { - struct ofport_dpif *port; + enum rule_dpif_lookup_verdict verdict; + enum ofputil_port_config config = 0; + uint8_t table_id = 0; - if (rule_dpif_lookup_in_table(ofproto, flow, wc, 0, rule)) { - return; + verdict = rule_dpif_lookup_from_table(ofproto, flow, wc, true, + &table_id, rule); + + switch (verdict) { + case RULE_DPIF_LOOKUP_VERDICT_MATCH: + return table_id; + case RULE_DPIF_LOOKUP_VERDICT_CONTROLLER: { + struct ofport_dpif *port; + + port = get_ofp_port(ofproto, flow->in_port.ofp_port); + if (!port) { + VLOG_WARN_RL(&rl, "packet-in on unknown OpenFlow port %"PRIu16, + flow->in_port.ofp_port); + } + config = port ? port->up.pp.config : 0; + break; } - port = get_ofp_port(ofproto, flow->in_port.ofp_port); - if (!port) { - VLOG_WARN_RL(&rl, "packet-in on unknown OpenFlow port %"PRIu16, - flow->in_port.ofp_port); + case RULE_DPIF_LOOKUP_VERDICT_DROP: + config = OFPUTIL_PC_NO_PACKET_IN; + break; + default: + OVS_NOT_REACHED(); } - choose_miss_rule(port ? port->up.pp.config : 0, ofproto->miss_rule, + choose_miss_rule(config, ofproto->miss_rule, ofproto->no_packet_in_rule, rule); + return table_id; } -bool -rule_dpif_lookup_in_table(struct ofproto_dpif *ofproto, - const struct flow *flow, struct flow_wildcards *wc, - uint8_t table_id, struct rule_dpif **rule) +static struct rule_dpif * +rule_dpif_lookup_in_table(struct ofproto_dpif *ofproto, uint8_t table_id, + const struct flow *flow, struct flow_wildcards *wc) { + struct classifier *cls = &ofproto->up.tables[table_id].cls; const struct cls_rule *cls_rule; - struct classifier *cls; - bool frag; - - *rule = NULL; - if (table_id >= N_TABLES) { - return false; - } + struct rule_dpif *rule; - if (wc) { - memset(&wc->masks.dl_type, 0xff, sizeof wc->masks.dl_type); - wc->masks.nw_frag |= FLOW_NW_FRAG_MASK; - } + fat_rwlock_rdlock(&cls->rwlock); + if (ofproto->up.frag_handling != OFPC_FRAG_NX_MATCH) { + if (wc) { + memset(&wc->masks.dl_type, 0xff, sizeof wc->masks.dl_type); + if (is_ip_any(flow)) { + wc->masks.nw_frag |= FLOW_NW_FRAG_MASK; + } + } - cls = &ofproto->up.tables[table_id].cls; - ovs_rwlock_rdlock(&cls->rwlock); - frag = (flow->nw_frag & FLOW_NW_FRAG_ANY) != 0; - if (frag && ofproto->up.frag_handling == OFPC_FRAG_NORMAL) { - /* We must pretend that transport ports are unavailable. */ - struct flow ofpc_normal_flow = *flow; - ofpc_normal_flow.tp_src = htons(0); - ofpc_normal_flow.tp_dst = htons(0); - cls_rule = classifier_lookup(cls, &ofpc_normal_flow, wc); - } else if (frag && ofproto->up.frag_handling == OFPC_FRAG_DROP) { - cls_rule = &ofproto->drop_frags_rule->up.cr; - /* Frag mask in wc already set above. */ + if (flow->nw_frag & FLOW_NW_FRAG_ANY) { + if (ofproto->up.frag_handling == OFPC_FRAG_NORMAL) { + /* We must pretend that transport ports are unavailable. */ + struct flow ofpc_normal_flow = *flow; + ofpc_normal_flow.tp_src = htons(0); + ofpc_normal_flow.tp_dst = htons(0); + cls_rule = classifier_lookup(cls, &ofpc_normal_flow, wc); + } else { + /* Must be OFPC_FRAG_DROP (we don't have OFPC_FRAG_REASM). */ + cls_rule = &ofproto->drop_frags_rule->up.cr; + } + } else { + cls_rule = classifier_lookup(cls, flow, wc); + } } else { cls_rule = classifier_lookup(cls, flow, wc); } - *rule = rule_dpif_cast(rule_from_cls_rule(cls_rule)); - rule_dpif_ref(*rule); - ovs_rwlock_unlock(&cls->rwlock); + rule = rule_dpif_cast(rule_from_cls_rule(cls_rule)); + rule_dpif_ref(rule); + fat_rwlock_unlock(&cls->rwlock); - return *rule != NULL; + return rule; +} + +/* Look up 'flow' in 'ofproto''s classifier starting from table '*table_id'. + * Stores the rule that was found in '*rule', or NULL if none was found. + * Updates 'wc', if nonnull, to reflect the fields that were used during the + * lookup. + * + * If 'honor_table_miss' is true, the first lookup occurs in '*table_id', but + * if none is found then the table miss configuration for that table is + * honored, which can result in additional lookups in other OpenFlow tables. + * In this case the function updates '*table_id' to reflect the final OpenFlow + * table that was searched. + * + * If 'honor_table_miss' is false, then only one table lookup occurs, in + * '*table_id'. + * + * Returns: + * + * - RULE_DPIF_LOOKUP_VERDICT_MATCH if a rule (in '*rule') was found. + * + * - RULE_DPIF_LOOKUP_VERDICT_DROP if no rule was found and a table miss + * configuration specified that the packet should be dropped in this + * case. (This occurs only if 'honor_table_miss' is true, because only in + * this case does the table miss configuration matter.) + * + * - RULE_DPIF_LOOKUP_VERDICT_CONTROLLER if no rule was found otherwise. */ +enum rule_dpif_lookup_verdict +rule_dpif_lookup_from_table(struct ofproto_dpif *ofproto, + const struct flow *flow, + struct flow_wildcards *wc, + bool honor_table_miss, + uint8_t *table_id, struct rule_dpif **rule) +{ + uint8_t next_id; + + for (next_id = *table_id; + next_id < ofproto->up.n_tables; + next_id++, next_id += (next_id == TBL_INTERNAL)) + { + *table_id = next_id; + *rule = rule_dpif_lookup_in_table(ofproto, *table_id, flow, wc); + if (*rule) { + return RULE_DPIF_LOOKUP_VERDICT_MATCH; + } else if (!honor_table_miss) { + return RULE_DPIF_LOOKUP_VERDICT_CONTROLLER; + } else { + switch (table_get_config(&ofproto->up, *table_id) + & OFPTC11_TABLE_MISS_MASK) { + case OFPTC11_TABLE_MISS_CONTINUE: + break; + + case OFPTC11_TABLE_MISS_CONTROLLER: + return RULE_DPIF_LOOKUP_VERDICT_CONTROLLER; + + case OFPTC11_TABLE_MISS_DROP: + return RULE_DPIF_LOOKUP_VERDICT_DROP; + } + } + } + + return RULE_DPIF_LOOKUP_VERDICT_CONTROLLER; } /* Given a port configuration (specified as zero if there's no port), chooses @@ -3092,13 +3268,13 @@ rule_dealloc(struct rule *rule_) static enum ofperr rule_construct(struct rule *rule_) + OVS_NO_THREAD_SAFETY_ANALYSIS { struct rule_dpif *rule = rule_dpif_cast(rule_); - ovs_mutex_init(&rule->stats_mutex); - ovs_mutex_lock(&rule->stats_mutex); - rule->packet_count = 0; - rule->byte_count = 0; - ovs_mutex_unlock(&rule->stats_mutex); + ovs_mutex_init_adaptive(&rule->stats_mutex); + rule->stats.n_packets = 0; + rule->stats.n_bytes = 0; + rule->stats.used = rule->up.modified; return 0; } @@ -3126,13 +3302,15 @@ rule_destruct(struct rule *rule_) } static void -rule_get_stats(struct rule *rule_, uint64_t *packets, uint64_t *bytes) +rule_get_stats(struct rule *rule_, uint64_t *packets, uint64_t *bytes, + long long int *used) { struct rule_dpif *rule = rule_dpif_cast(rule_); ovs_mutex_lock(&rule->stats_mutex); - *packets = rule->packet_count; - *bytes = rule->byte_count; + *packets = rule->stats.n_packets; + *bytes = rule->stats.n_bytes; + *used = rule->stats.used; ovs_mutex_unlock(&rule->stats_mutex); } @@ -3162,8 +3340,8 @@ rule_modify_actions(struct rule *rule_, bool reset_counters) if (reset_counters) { ovs_mutex_lock(&rule->stats_mutex); - rule->packet_count = 0; - rule->byte_count = 0; + rule->stats.n_packets = 0; + rule->stats.n_bytes = 0; ovs_mutex_unlock(&rule->stats_mutex); } @@ -3208,7 +3386,21 @@ static enum ofperr group_construct(struct ofgroup *group_) { struct group_dpif *group = group_dpif_cast(group_); - ovs_mutex_init(&group->stats_mutex); + const struct ofputil_bucket *bucket; + + /* Prevent group chaining because our locking structure makes it hard to + * implement deadlock-free. (See xlate_group_resource_check().) */ + LIST_FOR_EACH (bucket, list_node, &group->up.buckets) { + const struct ofpact *a; + + OFPACT_FOR_EACH (a, bucket->ofpacts, bucket->ofpacts_len) { + if (a->type == OFPACT_GROUP) { + return OFPERR_OFPGMFC_CHAINING_UNSUPPORTED; + } + } + } + + ovs_mutex_init_adaptive(&group->stats_mutex); ovs_mutex_lock(&group->stats_mutex); group_construct_stats(group); ovs_mutex_unlock(&group->stats_mutex); @@ -3236,6 +3428,7 @@ group_destruct(struct ofgroup *group_) static enum ofperr group_modify(struct ofgroup *group_, struct ofgroup *victim_) { + struct ofproto_dpif *ofproto = ofproto_dpif_cast(group_->ofproto); struct group_dpif *group = group_dpif_cast(group_); struct group_dpif *victim = group_dpif_cast(victim_); @@ -3246,6 +3439,8 @@ group_modify(struct ofgroup *group_, struct ofgroup *victim_) group_construct_stats(group); ovs_mutex_unlock(&group->stats_mutex); + ofproto->backer->need_revalidate = REV_FLOW_TABLE; + return 0; } @@ -3456,6 +3651,7 @@ struct trace_ctx { struct xlate_out xout; struct xlate_in xin; struct flow flow; + struct flow_wildcards wc; struct ds *result; }; @@ -3486,8 +3682,6 @@ trace_format_rule(struct ds *result, int level, const struct rule_dpif *rule) ds_put_cstr(result, "OpenFlow actions="); ofpacts_format(actions->ofpacts, actions->ofpacts_len, result); ds_put_char(result, '\n'); - - rule_actions_unref(actions); } static void @@ -3531,6 +3725,20 @@ trace_format_odp(struct ds *result, int level, const char *title, ds_put_char(result, '\n'); } +static void +trace_format_megaflow(struct ds *result, int level, const char *title, + struct trace_ctx *trace) +{ + struct match match; + + ds_put_char_multiple(result, '\t', level); + ds_put_format(result, "%s: ", title); + flow_wildcards_or(&trace->wc, &trace->xout.wc, &trace->wc); + match_init(&match, &trace->flow, &trace->wc); + match_format(&match, result, OFP_DEFAULT_PRIORITY); + ds_put_char(result, '\n'); +} + static void trace_resubmit(struct xlate_in *xin, struct rule_dpif *rule, int recurse) { @@ -3541,6 +3749,7 @@ trace_resubmit(struct xlate_in *xin, struct rule_dpif *rule, int recurse) trace_format_flow(result, recurse + 1, "Resubmitted flow", trace); trace_format_regs(result, recurse + 1, "Resubmitted regs", trace); trace_format_odp(result, recurse + 1, "Resubmitted odp", trace); + trace_format_megaflow(result, recurse + 1, "Resubmitted megaflow", trace); trace_format_rule(result, recurse + 1, rule); } @@ -3637,7 +3846,7 @@ parse_flow_and_packet(int argc, const char *argv[], } if (xlate_receive(backer, NULL, odp_key.data, odp_key.size, flow, - NULL, ofprotop, NULL, NULL, NULL, NULL)) { + ofprotop, NULL, NULL, NULL, NULL)) { error = "Invalid datapath flow"; goto exit; } @@ -3667,12 +3876,11 @@ parse_flow_and_packet(int argc, const char *argv[], if (!packet->size) { flow_compose(packet, flow); } else { - union flow_in_port in_port = flow->in_port; + struct pkt_metadata md = PKT_METADATA_INITIALIZER_FLOW(flow); /* Use the metadata from the flow and the packet argument * to reconstruct the flow. */ - flow_extract(packet, flow->skb_priority, flow->pkt_mark, NULL, - &in_port, flow); + flow_extract(packet, &md, flow); } } @@ -3819,18 +4027,18 @@ ofproto_trace(struct ofproto_dpif *ofproto, const struct flow *flow, struct ds *ds) { struct rule_dpif *rule; - struct flow_wildcards wc; + struct trace_ctx trace; ds_put_format(ds, "Bridge: %s\n", ofproto->up.name); ds_put_cstr(ds, "Flow: "); flow_format(ds, flow); ds_put_char(ds, '\n'); - flow_wildcards_init_catchall(&wc); + flow_wildcards_init_catchall(&trace.wc); if (ofpacts) { rule = NULL; } else { - rule_dpif_lookup(ofproto, flow, &wc, &rule); + rule_dpif_lookup(ofproto, flow, &trace.wc, &rule); trace_format_rule(ds, 0, rule); if (rule == ofproto->miss_rule) { @@ -3845,14 +4053,10 @@ ofproto_trace(struct ofproto_dpif *ofproto, const struct flow *flow, } if (rule || ofpacts) { - struct trace_ctx trace; - struct match match; - uint16_t tcp_flags; - - tcp_flags = packet ? packet_get_tcp_flags(packet, flow) : 0; trace.result = ds; trace.flow = *flow; - xlate_in_init(&trace.xin, ofproto, flow, rule, tcp_flags, packet); + xlate_in_init(&trace.xin, ofproto, flow, rule, ntohs(flow->tcp_flags), + packet); if (ofpacts) { trace.xin.ofpacts = ofpacts; trace.xin.ofpacts_len = ofpacts_len; @@ -3861,15 +4065,10 @@ ofproto_trace(struct ofproto_dpif *ofproto, const struct flow *flow, trace.xin.report_hook = trace_report; xlate_actions(&trace.xin, &trace.xout); - flow_wildcards_or(&trace.xout.wc, &trace.xout.wc, &wc); ds_put_char(ds, '\n'); trace_format_flow(ds, 0, "Final flow", &trace); - - match_init(&match, &trace.flow, &trace.xout.wc); - ds_put_cstr(ds, "Relevant fields: "); - match_format(&match, ds, OFP_DEFAULT_PRIORITY); - ds_put_char(ds, '\n'); + trace_format_megaflow(ds, 0, "Megaflow", &trace); ds_put_cstr(ds, "Datapath actions: "); format_odp_actions(ds, trace.xout.odp_actions.data, @@ -4029,11 +4228,10 @@ static bool ofproto_dpif_contains_flow(const struct ofproto_dpif *ofproto, const struct nlattr *key, size_t key_len) { - enum odp_key_fitness fitness; struct ofproto_dpif *ofp; struct flow flow; - xlate_receive(ofproto->backer, NULL, key, key_len, &flow, &fitness, &ofp, + xlate_receive(ofproto->backer, NULL, key, key_len, &flow, &ofp, NULL, NULL, NULL, NULL); return ofp == ofproto; } @@ -4057,6 +4255,8 @@ ofproto_unixctl_dpif_dump_flows(struct unixctl_conn *conn, struct dpif_port dpif_port; struct dpif_port_dump port_dump; struct hmap portno_names; + void *state = NULL; + int error; ofproto = ofproto_dpif_lookup(argv[argc - 1]); if (!ofproto) { @@ -4074,9 +4274,14 @@ ofproto_unixctl_dpif_dump_flows(struct unixctl_conn *conn, } ds_init(&ds); - dpif_flow_dump_start(&flow_dump, ofproto->backer->dpif); - while (dpif_flow_dump_next(&flow_dump, &key, &key_len, &mask, &mask_len, - &actions, &actions_len, &stats)) { + error = dpif_flow_dump_start(&flow_dump, ofproto->backer->dpif); + if (error) { + goto exit; + } + dpif_flow_dump_state_init(ofproto->backer->dpif, &state); + while (dpif_flow_dump_next(&flow_dump, state, &key, &key_len, + &mask, &mask_len, &actions, &actions_len, + &stats)) { if (!ofproto_dpif_contains_flow(ofproto, key, key_len)) { continue; } @@ -4089,8 +4294,11 @@ ofproto_unixctl_dpif_dump_flows(struct unixctl_conn *conn, format_odp_actions(&ds, actions, actions_len); ds_put_char(&ds, '\n'); } + dpif_flow_dump_state_uninit(ofproto->backer->dpif, state); + error = dpif_flow_dump_done(&flow_dump); - if (dpif_flow_dump_done(&flow_dump)) { +exit: + if (error) { ds_clear(&ds); ds_put_format(&ds, "dpif/dump_flows failed: %s", ovs_strerror(errno)); unixctl_command_reply_error(conn, ds_cstr(&ds)); @@ -4130,6 +4338,14 @@ ofproto_dpif_unixctl_init(void) unixctl_command_register("dpif/dump-flows", "[-m] bridge", 1, 2, ofproto_unixctl_dpif_dump_flows, NULL); } + + +/* Returns true if 'rule' is an internal rule, false otherwise. */ +bool +rule_is_internal(const struct rule *rule) +{ + return rule->table_id == TBL_INTERNAL; +} /* Linux VLAN device support (e.g. "eth0.10" for VLAN 10.) * @@ -4180,12 +4396,8 @@ bool ofproto_has_vlan_splinters(const struct ofproto_dpif *ofproto) OVS_EXCLUDED(ofproto->vsp_mutex) { - bool ret; - - ovs_mutex_lock(&ofproto->vsp_mutex); - ret = !hmap_is_empty(&ofproto->realdev_vid_map); - ovs_mutex_unlock(&ofproto->vsp_mutex); - return ret; + /* hmap_is_empty is thread safe. */ + return !hmap_is_empty(&ofproto->realdev_vid_map); } static ofp_port_t @@ -4223,6 +4435,10 @@ vsp_realdev_to_vlandev(const struct ofproto_dpif *ofproto, { ofp_port_t ret; + /* hmap_is_empty is thread safe, see if we can return immediately. */ + if (hmap_is_empty(&ofproto->realdev_vid_map)) { + return realdev_ofp_port; + } ovs_mutex_lock(&ofproto->vsp_mutex); ret = vsp_realdev_to_vlandev__(ofproto, realdev_ofp_port, vlan_tci); ovs_mutex_unlock(&ofproto->vsp_mutex); @@ -4286,6 +4502,11 @@ vsp_adjust_flow(const struct ofproto_dpif *ofproto, struct flow *flow) ofp_port_t realdev; int vid; + /* hmap_is_empty is thread safe. */ + if (hmap_is_empty(&ofproto->vlandev_map)) { + return false; + } + ovs_mutex_lock(&ofproto->vsp_mutex); realdev = vsp_vlandev_to_realdev(ofproto, flow->in_port.ofp_port, &vid); ovs_mutex_unlock(&ofproto->vsp_mutex);