return 1;
}
-void resource_sys_init(void);
-
+
int
main(int argc, char *argv[])
{
{
event *e = SKIP_BACK(event, n, n);
- /* This is ugly hack, we want to log just events executed from the main I/O loop */
- if ((l == &global_event_list) || (l == &global_work_list))
+ if (legacy)
+ {
+ /* The legacy way of event execution */
io_log_event(e->hook, e->data);
-
- ev_run(e);
+ ev_postpone(e);
+ e->hook(e->data);
+ }
+ else
+ {
+ // io_log_event(e->hook, e->data); /* TODO: add support for event logging in other io loops */
+ ASSERT_DIE(e->list == l);
+ LOCK_DOMAIN(event, l->lock);
+ rem_node(&e->n);
+ UNLOCK_DOMAIN(event, l->lock);
+ e->hook(e->data);
+ }
+ tmp_flush();
}
- return !EMPTY_LIST(*l);
+ LOCK_DOMAIN(event, l->lock);
+ int repeat = ! EMPTY_LIST(l->events);
+ UNLOCK_DOMAIN(event, l->lock);
+ return repeat;
}
int
if (!limit)
break;
- /* This is ugly hack, we want to log just events executed from the main I/O loop */
- if ((l == &global_event_list) || (l == &global_work_list))
- io_log_event(e->hook, e->data);
+ io_log_event(e->hook, e->data);
ev_run(e);
+ tmp_flush();
limit--;
}
{
int i;
- resource_sys_init();
- resource_init();
olock_init();
- timer_init();
+ birdloop_init();
- io_init();
rt_init();
+ io_init();
if_init();
// roa_init();
config_init();
* is freed upon shutdown of the module.
*/
--struct pool {
-- resource r;
-- list inside;
- struct pool_pages *pages;
-- const char *name;
-};
-
-struct pool_pages {
- uint free;
- uint used;
- void *ptr[0];
--};
-
-#define POOL_PAGES_MAX ((page_size - sizeof(struct pool_pages)) / sizeof (void *))
--
static void pool_dump(resource *);
static void pool_free(resource *);
static resource *pool_lookup(resource *, unsigned long);
/* Generic resource manipulation */
--typedef struct pool pool;
++typedef struct pool {
++ resource r;
++ list inside;
++ const char *name;
++} pool;
++
void resource_init(void);
pool *rp_new(pool *, const char *); /* Create new pool */
#include "lib/event.h"
#include "lib/timer.h"
#include "lib/string.h"
+ #include "lib/coro.h"
#include "conf/conf.h"
-#include "nest/route.h"
+#include "nest/rt.h"
#include "nest/iface.h"
#include "nest/cli.h"
#include "filter/filter.h"
c->out_req = (struct rt_export_request) {
.name = rn,
+ .list = proto_work_list(c->proto),
+ .addr = c->out_subprefix,
+ .addr_mode = c->out_subprefix ? TE_ADDR_IN : TE_ADDR_NONE,
.trace_routes = c->debug | c->proto->debug,
.dump_req = channel_dump_export_req,
.log_state_change = channel_export_log_state_change,
/* Called by protocol to activate in_table */
void
-channel_setup_in_table(struct channel *c, int best)
-{
- int nlen = sizeof("import") + strlen(c->name) + strlen(c->proto->name) + 3;
-
- struct {
- struct channel_aux_table cat;
- struct rtable_config tab_cf;
- char name[0];
- } *cat = mb_allocz(c->proto->pool, sizeof(*cat) + nlen);
-
- bsprintf(cat->name, "%s.%s.import", c->proto->name, c->name);
-
- cat->tab_cf.name = cat->name;
- cat->tab_cf.addr_type = c->net_type;
-
- c->in_table = &cat->cat;
- c->in_table->push = (struct rt_import_request) {
- .name = cat->name,
- .trace_routes = c->debug | c->proto->debug,
- .dump_req = channel_in_push_dump_req,
- .log_state_change = channel_push_log_state_change,
- .preimport = channel_in_preimport,
- };
- c->in_table->get = (struct rt_export_request) {
- .name = cat->name,
- .list = proto_work_list(c->proto),
- .trace_routes = c->debug | c->proto->debug,
- .dump_req = channel_in_get_dump_req,
- .log_state_change = channel_get_log_state_change,
- .export_one = best ? channel_in_export_one_best : channel_in_export_one_any,
- .export_bulk = best ? channel_in_export_bulk_best : channel_in_export_bulk_any,
- };
-
- c->in_table->c = c;
- c->in_table->tab = rt_setup(c->proto->pool, &cat->tab_cf);
- self_link(&c->in_table->tab->n);
- rt_lock_table(c->in_table->tab);
-
- rt_request_import(c->in_table->tab, &c->in_table->push);
- rt_request_export(c->in_table->tab, &c->in_table->get);
-}
-
-/* Called by protocol to activate out_table */
-void
-channel_setup_out_table(struct channel *c)
+channel_setup_in_table(struct channel *c)
{
- int nlen = sizeof("export") + strlen(c->name) + strlen(c->proto->name) + 3;
-
- struct {
- struct channel_aux_table cat;
- struct rtable_config tab_cf;
- char name[0];
- } *cat = mb_allocz(c->proto->pool, sizeof(*cat) + nlen);
-
- bsprintf(cat->name, "%s.%s.export", c->proto->name, c->name);
-
- cat->tab_cf.name = cat->name;
- cat->tab_cf.addr_type = c->net_type;
-
- c->out_table = &cat->cat;
- c->out_table->push = (struct rt_import_request) {
- .name = cat->name,
- .trace_routes = c->debug | c->proto->debug,
- .dump_req = channel_out_push_dump_req,
- .log_state_change = channel_push_log_state_change,
- };
- c->out_table->get = (struct rt_export_request) {
- .name = cat->name,
+ c->reload_req = (struct rt_export_request) {
+ .name = mb_sprintf(c->proto->pool, "%s.%s.import", c->proto->name, c->name),
+ .list = proto_work_list(c->proto),
.trace_routes = c->debug | c->proto->debug,
- .dump_req = channel_out_get_dump_req,
- .log_state_change = channel_get_log_state_change,
- .export_one = (c->ra_mode == RA_ANY) ? channel_out_export_one_any : channel_out_export_one_best,
- .export_bulk = channel_out_export_bulk,
+ .export_bulk = channel_reload_export_bulk,
+ .dump_req = channel_reload_dump_req,
+ .log_state_change = channel_reload_log_state_change,
};
- c->out_table->c = c;
- c->out_table->tab = rt_setup(c->proto->pool, &cat->tab_cf);
- self_link(&c->out_table->tab->n);
- rt_lock_table(c->out_table->tab);
-
- rt_request_import(c->out_table->tab, &c->out_table->push);
- rt_request_export(c->out_table->tab, &c->out_table->get);
+ c->in_keep |= RIK_PREFILTER;
}
-static void
-channel_aux_request_refeed(struct channel_aux_table *cat)
-{
- cat->refeed_pending = 1;
- rt_stop_export(&cat->get, channel_aux_export_stopped);
-}
static void
channel_do_start(struct channel *c)
static void
proto_start(struct proto *p)
{
- /* Here we cannot use p->cf->name since it won't survive reconfiguration */
- p->pool = rp_new(proto_pool, p->proto->name);
+ DBG("Kicking %s up\n", p->name);
+ PD(p, "Starting");
+
- int ns = strlen("Protocol ") + strlen(p->cf->name) + 1;
- void *nb = mb_alloc(proto_pool, ns);
- ASSERT_DIE(ns - 1 == bsnprintf(nb, ns, "Protocol %s", p->cf->name));
-
- p->pool = rp_new(proto_pool, nb);
++ p->pool = rp_newf(proto_pool, "Protocol %s", p->cf->name);
if (graceful_restart_state == GRS_INIT)
p->gr_recovery = 1;
- p->loop = birdloop_new(p->pool, p->cf->loop_order, nb);
+
+ if (p->cf->loop_order != DOMAIN_ORDER(the_bird))
- mb_move(nb, p->pool);
-
++ p->loop = birdloop_new(p->pool, p->cf->loop_order, p->pool->name);
+
+ p->event->list = proto_event_list(p);
+
+ PROTO_LOCKED_FROM_MAIN(p)
+ proto_notify_state(p, (p->proto->start ? p->proto->start(p) : PS_UP));
}
WALK_LIST(p, proto_list)
{
#define DPF(x) (p->x ? " " #x : "")
- debug(" protocol %s (%p) state %s with %d active channels flags: %s%s%s%s\n",
+ debug(" protocol %s (%p) state %s with %d active channels flags: %s%s%s%s%s\n",
p->name, p, p_states[p->proto_state], p->active_channels,
- DPF(disabled), DPF(active), DPF(do_start), DPF(do_stop), DPF(reconfiguring));
+ DPF(disabled), DPF(active), DPF(do_stop), DPF(reconfiguring));
#undef DPF
struct channel *c;
if (d->running_on_config && (d->running_on_config != config))
{
cli_printf(c, 8004, "Stopped due to reconfiguration");
- goto done;
+ return rt_show_done(d);
}
- if (!d->table_open)
- {
- FIB_ITERATE_INIT(&d->fit, &d->tab->table->fib);
- d->table_open = 1;
- d->table_counter++;
- d->kernel = rt_show_get_kernel(d);
+ d->req = (struct rt_export_request) {
+ .addr = d->addr,
+ .name = "CLI Show Route",
++ .list = &global_work_list,
+ .export_bulk = rt_show_net_export_bulk,
+ .dump_req = rt_show_dump_req,
+ .log_state_change = rt_show_log_state_change,
+ .addr_mode = d->addr_mode,
+ };
- d->show_counter_last = d->show_counter;
- d->rt_counter_last = d->rt_counter;
- d->net_counter_last = d->net_counter;
+ d->table_counter++;
- if (d->tables_defined_by & RSD_TDB_SET)
- rt_show_table(c, d);
- }
+ d->show_counter_last = d->show_counter;
+ d->rt_counter_last = d->rt_counter;
+ d->net_counter_last = d->net_counter;
- FIB_ITERATE_START(fib, it, net, n)
- {
- if (!max--)
- {
- FIB_ITERATE_PUT(it);
- return;
- }
- rt_show_net(c, n, d);
- }
- FIB_ITERATE_END;
+ if (d->tables_defined_by & RSD_TDB_SET)
+ rt_show_table(d);
+
+ rt_request_export(d->tab->table, &d->req);
+}
+
+static void
+rt_show_export_stopped(struct rt_export_request *req)
+{
+ struct rt_show_data *d = SKIP_BACK(struct rt_show_data, req, req);
+
+ /* The hook is now invalid */
+ req->hook = NULL;
if (d->stats)
{
/* No, therefore we must process the table's first pending export */
else
- return tab->first_export;
+ return tab->first;
}
+ static inline void
+ rt_send_export_event(struct rt_export_hook *hook)
+ {
+ ev_send(hook->req->list, hook->event);
+ }
+
static void
rt_announce_exports(timer *tm)
{
atomic_store_explicit(&hook->last_export, rpe, memory_order_relaxed);
hook->n = (node) {};
- add_tail(&tab->exports, &hook->n);
+ add_tail(&re->hooks, &hook->n);
- FIB_ITERATE_INIT(&hook->feed_fit, &tab->fib);
+ /* Regular export */
+ rt_set_export_state(hook, TES_FEEDING);
- ev_schedule_work(hook->event);
++ rt_send_export_event(hook);
+}
- DBG("New export hook %p req %p in table %s uc=%u\n", hook, req, tab->name, tab->use_count);
+static void
+rt_table_export_stop(struct rt_export_hook *hook)
+{
+ rtable *tab = SKIP_BACK(rtable, exporter, hook->table);
- hook->event = ev_new_init(p, rt_feed_channel, hook);
- rt_send_export_event(hook);
+ if (atomic_load_explicit(&hook->export_state, memory_order_relaxed) != TES_FEEDING)
+ return;
- rt_set_export_state(hook, TES_FEEDING);
+ switch (hook->req->addr_mode)
+ {
+ case TE_ADDR_IN:
+ if (hook->walk_lock)
+ {
+ rt_unlock_trie(tab, hook->walk_lock);
+ hook->walk_lock = NULL;
+ mb_free(hook->walk_state);
+ hook->walk_state = NULL;
+ break;
+ }
+ /* fall through */
+ case TE_ADDR_NONE:
+ fit_get(&tab->fib, &hook->feed_fit);
+ break;
+ }
}
void
hook->event->hook = rt_export_stopped;
hook->stopped = stopped;
- rt_send_export_event(hook);
-
+ /* Update export state */
rt_set_export_state(hook, TES_STOP);
-}
- ev_schedule(hook->event);
+ /* Run the stopped event */
++ rt_send_export_event(hook);
+}
/**
* rt_refresh_begin - start a refresh cycle
DBG("\tdone\n");
}
- ev_schedule_work(c->event);
+static void
+rt_feed_done(struct rt_export_hook *c)
+{
+ c->event->hook = rt_export_hook;
+
+ rt_set_export_state(c, TES_READY);
+
++ rt_send_export_event(c);
+}
+
/**
- * rt_feed_channel - advertise all routes to a channel
+ * rt_feed_by_fib - advertise all routes to a channel by walking a fib
* @c: channel to be fed
*
* This function performs one pass of advertisement of routes to a channel that
--- /dev/null
+/*
+ * BIRD Internet Routing Daemon -- Routing Table
+ *
+ * (c) 1998--2000 Martin Mares <mj@ucw.cz>
+ * (c) 2019--2021 Maria Matejka <mq@jmq.cz>
+ *
+ * Can be freely distributed and used under the terms of the GNU GPL.
+ */
+
+#ifndef _BIRD_NEST_RT_H_
+#define _BIRD_NEST_RT_H_
+
+#include "lib/lists.h"
+#include "lib/bitmap.h"
+#include "lib/resource.h"
+#include "lib/net.h"
+#include "lib/type.h"
+#include "lib/fib.h"
+#include "lib/route.h"
++#include "lib/event.h"
+
+#include <stdatomic.h>
+
+struct ea_list;
+struct protocol;
+struct proto;
+struct channel;
+struct rte_src;
+struct symbol;
+struct timer;
+struct filter;
+struct f_trie;
+struct f_trie_walk_state;
+struct cli;
+
+/*
+ * Master Routing Tables. Generally speaking, each of them contains a FIB
+ * with each entry pointing to a list of route entries representing routes
+ * to given network (with the selected one at the head).
+ *
+ * Each of the RTE's contains variable data (the preference and protocol-dependent
+ * metrics) and a pointer to a route attribute block common for many routes).
+ *
+ * It's guaranteed that there is at most one RTE for every (prefix,proto) pair.
+ */
+
+struct rtable_config {
+ node n;
+ char *name;
+ struct rtable *table;
+ struct proto_config *krt_attached; /* Kernel syncer attached to this table */
+ uint addr_type; /* Type of address data stored in table (NET_*) */
+ uint gc_threshold; /* Maximum number of operations before GC is run */
+ uint gc_period; /* Approximate time between two consecutive GC runs */
+ byte sorted; /* Routes of network are sorted according to rte_better() */
+ byte trie_used; /* Rtable has attached trie */
+ btime min_settle_time; /* Minimum settle time for notifications */
+ btime max_settle_time; /* Maximum settle time for notifications */
+ btime export_settle_time; /* Delay before exports are announced */
+};
+
+struct rt_export_hook;
+struct rt_export_request;
+
+struct rt_exporter {
+ list hooks; /* Registered route export hooks */
+ uint addr_type; /* Type of address data exported (NET_*) */
+
+ struct rt_export_hook *(*start)(struct rt_exporter *, struct rt_export_request *);
+ void (*stop)(struct rt_export_hook *);
+ void (*done)(struct rt_export_hook *);
+ void (*used)(struct rt_exporter *);
+
+ list pending; /* List of packed struct rt_pending_export */
+ struct timer *export_timer;
+
+ struct rt_pending_export *first; /* First export to announce */
+ u64 next_seq; /* The next export will have this ID */
+};
+
+typedef struct rtable {
+ resource r;
+ node n; /* Node in list of all tables */
+ pool *rp; /* Resource pool to allocate everything from, including itself */
+ struct slab *rte_slab; /* Slab to allocate route objects */
+ struct fib fib;
+ struct f_trie *trie; /* Trie of prefixes defined in fib */
+ char *name; /* Name of this table */
+ uint addr_type; /* Type of address data stored in table (NET_*) */
+ int use_count; /* Number of protocols using this table */
+ u32 rt_count; /* Number of routes in the table */
+
+ list imports; /* Registered route importers */
+ struct rt_exporter exporter; /* Exporter API structure */
+
+ struct hmap id_map;
+ struct hostcache *hostcache;
+ struct rtable_config *config; /* Configuration of this table */
+ struct config *deleted; /* Table doesn't exist in current configuration,
+ * delete as soon as use_count becomes 0 and remove
+ * obstacle from this routing table.
+ */
+ struct event *rt_event; /* Routing table event */
+ struct timer *prune_timer; /* Timer for periodic pruning / GC */
+ btime last_rt_change; /* Last time when route changed */
+ btime base_settle_time; /* Start time of rtable settling interval */
+ btime gc_time; /* Time of last GC */
+ uint gc_counter; /* Number of operations since last GC */
+ byte prune_state; /* Table prune state, 1 -> scheduled, 2-> running */
+ byte prune_trie; /* Prune prefix trie during next table prune */
+ byte hcu_scheduled; /* Hostcache update is scheduled */
+ byte nhu_state; /* Next Hop Update state */
+ byte export_used; /* Pending Export pruning is scheduled */
+ struct fib_iterator prune_fit; /* Rtable prune FIB iterator */
+ struct fib_iterator nhu_fit; /* Next Hop Update FIB iterator */
+ struct f_trie *trie_new; /* New prefix trie defined during pruning */
+ struct f_trie *trie_old; /* Old prefix trie waiting to be freed */
+ u32 trie_lock_count; /* Prefix trie locked by walks */
+ u32 trie_old_lock_count; /* Old prefix trie locked by walks */
+ struct tbf rl_pipe; /* Rate limiting token buffer for pipe collisions */
+
+ list subscribers; /* Subscribers for notifications */
+ struct timer *settle_timer; /* Settle time for notifications */
+ list flowspec_links; /* List of flowspec links, src for NET_IPx and dst for NET_FLOWx */
+ struct f_trie *flowspec_trie; /* Trie for evaluation of flowspec notifications */
+} rtable;
+
+struct rt_subscription {
+ node n;
+ rtable *tab;
+ void (*hook)(struct rt_subscription *b);
+ void *data;
+};
+
+struct rt_flowspec_link {
+ node n;
+ rtable *src;
+ rtable *dst;
+ u32 uc;
+};
+
+#define NHU_CLEAN 0
+#define NHU_SCHEDULED 1
+#define NHU_RUNNING 2
+#define NHU_DIRTY 3
+
+typedef struct network {
+ struct rte_storage *routes; /* Available routes for this network */
+ struct rt_pending_export *first, *last;
+ struct fib_node n; /* FIB flags reserved for kernel syncer */
+} net;
+
+struct hostcache {
+ slab *slab; /* Slab holding all hostentries */
+ struct hostentry **hash_table; /* Hash table for hostentries */
+ unsigned hash_order, hash_shift;
+ unsigned hash_max, hash_min;
+ unsigned hash_items;
+ linpool *lp; /* Linpool for trie */
+ struct f_trie *trie; /* Trie of prefixes that might affect hostentries */
+ list hostentries; /* List of all hostentries */
+ byte update_hostcache;
+};
+
+struct hostentry {
+ node ln;
+ ip_addr addr; /* IP address of host, part of key */
+ ip_addr link; /* (link-local) IP address of host, used as gw
+ if host is directly attached */
+ struct rtable *tab; /* Dependent table, part of key */
+ struct hostentry *next; /* Next in hash chain */
+ unsigned hash_key; /* Hash key */
+ unsigned uc; /* Use count */
+ ea_list *src; /* Source attributes */
+ byte nexthop_linkable; /* Nexthop list is completely non-device */
+ u32 igp_metric; /* Chosen route IGP metric */
+};
+
+struct rte_storage {
+ struct rte_storage *next; /* Next in chain */
+ struct rte rte; /* Route data */
+};
+
+#define RTE_COPY(r) ((r) ? (r)->rte : (rte) {})
+#define RTE_COPY_VALID(r) (((r) && (rte_is_valid(&(r)->rte))) ? (r)->rte : (rte) {})
+#define RTE_OR_NULL(r) ((r) ? &((r)->rte) : NULL)
+#define RTE_VALID_OR_NULL(r) (((r) && (rte_is_valid(&(r)->rte))) ? &((r)->rte) : NULL)
+
+/* Table-channel connections */
+
+struct rt_import_request {
+ struct rt_import_hook *hook; /* The table part of importer */
+ char *name;
+ u8 trace_routes;
+
+ void (*dump_req)(struct rt_import_request *req);
+ void (*log_state_change)(struct rt_import_request *req, u8 state);
+ /* Preimport is called when the @new route is just-to-be inserted, replacing @old.
+ * Return a route (may be different or modified in-place) to continue or NULL to withdraw. */
+ int (*preimport)(struct rt_import_request *req, struct rte *new, struct rte *old);
+};
+
+struct rt_import_hook {
+ node n;
+ rtable *table; /* The connected table */
+ struct rt_import_request *req; /* The requestor */
+
+ struct rt_import_stats {
+ /* Import - from protocol to core */
+ u32 pref; /* Number of routes selected as best in the (adjacent) routing table */
+ u32 updates_ignored; /* Number of route updates rejected as already in route table */
+ u32 updates_accepted; /* Number of route updates accepted and imported */
+ u32 withdraws_ignored; /* Number of route withdraws rejected as already not in route table */
+ u32 withdraws_accepted; /* Number of route withdraws accepted and processed */
+ } stats;
+
+ u64 flush_seq; /* Table export seq when the channel announced flushing */
+ btime last_state_change; /* Time of last state transition */
+
+ u8 import_state; /* IS_* */
+ u8 stale_set; /* Set this stale_cycle to imported routes */
+ u8 stale_valid; /* Routes with this stale_cycle and bigger are considered valid */
+ u8 stale_pruned; /* Last prune finished when this value was set at stale_valid */
+ u8 stale_pruning; /* Last prune started when this value was set at stale_valid */
+
+ void (*stopped)(struct rt_import_request *); /* Stored callback when import is stopped */
+};
+
+struct rt_pending_export {
+ struct rt_pending_export * _Atomic next; /* Next export for the same destination */
+ struct rte_storage *new, *new_best, *old, *old_best;
+ u64 seq; /* Sequential ID (table-local) of the pending export */
+};
+
+struct rt_export_request {
+ struct rt_export_hook *hook; /* Table part of the export */
+ char *name;
+ const net_addr *addr; /* Network prefilter address */
+ u8 trace_routes;
+ u8 addr_mode; /* Network prefilter mode (TE_ADDR_*) */
+
++ event_list *list; /* Where to schedule export events */
++
+ /* There are two methods of export. You can either request feeding every single change
+ * or feeding the whole route feed. In case of regular export, &export_one is preferred.
+ * Anyway, when feeding, &export_bulk is preferred, falling back to &export_one.
+ * Thus, for RA_OPTIMAL, &export_one is only set,
+ * for RA_MERGED and RA_ACCEPTED, &export_bulk is only set
+ * and for RA_ANY, both are set to accomodate for feeding all routes but receiving single changes
+ */
+ void (*export_one)(struct rt_export_request *req, const net_addr *net, struct rt_pending_export *rpe);
+ void (*export_bulk)(struct rt_export_request *req, const net_addr *net, struct rt_pending_export *rpe, rte **feed, uint count);
+
+ void (*dump_req)(struct rt_export_request *req);
+ void (*log_state_change)(struct rt_export_request *req, u8);
+};
+
+struct rt_export_hook {
+ node n;
+ struct rt_exporter *table; /* The connected table */
+
+ pool *pool;
+
+ struct rt_export_request *req; /* The requestor */
+
+ struct rt_export_stats {
+ /* Export - from core to protocol */
+ u32 updates_received; /* Number of route updates received */
+ u32 withdraws_received; /* Number of route withdraws received */
+ } stats;
+
+ union {
+ struct fib_iterator feed_fit; /* Routing table iterator used during feeding */
+ struct {
+ struct f_trie_walk_state *walk_state; /* Iterator over networks in trie */
+ struct f_trie *walk_lock; /* Locked trie for walking */
+ };
+ u32 hash_iter; /* Iterator over hash */
+ };
+
+ struct bmap seq_map; /* Keep track which exports were already procesed */
+
+ struct rt_pending_export * _Atomic last_export;/* Last export processed */
+ struct rt_pending_export *rpe_next; /* Next pending export to process */
+
+ btime last_state_change; /* Time of last state transition */
+
+ u8 refeed_pending; /* Refeeding and another refeed is scheduled */
+ _Atomic u8 export_state; /* Route export state (TES_*, see below) */
+ u8 feed_type; /* Which feeding method is used (TFT_*, see below) */
+
+ struct event *event; /* Event running all the export operations */
+
+ void (*stopped)(struct rt_export_request *); /* Stored callback when export is stopped */
+};
+
+#define TIS_DOWN 0
+#define TIS_UP 1
+#define TIS_STOP 2
+#define TIS_FLUSHING 3
+#define TIS_WAITING 4
+#define TIS_CLEARED 5
+#define TIS_MAX 6
+
+#define TES_DOWN 0
+#define TES_FEEDING 2
+#define TES_READY 3
+#define TES_STOP 4
+#define TES_MAX 5
+
+/* Value of addr_mode */
+#define TE_ADDR_NONE 0 /* No address matching */
+#define TE_ADDR_EQUAL 1 /* Exact query - show route <addr> */
+#define TE_ADDR_FOR 2 /* Longest prefix match - show route for <addr> */
+#define TE_ADDR_IN 3 /* Interval query - show route in <addr> */
+
+
+#define TFT_FIB 1
+#define TFT_TRIE 2
+#define TFT_HASH 3
+
+void rt_request_import(rtable *tab, struct rt_import_request *req);
+void rt_request_export(struct rt_exporter *tab, struct rt_export_request *req);
+
+void rt_export_once(struct rt_exporter *tab, struct rt_export_request *req);
+
+void rt_stop_import(struct rt_import_request *, void (*stopped)(struct rt_import_request *));
+void rt_stop_export(struct rt_export_request *, void (*stopped)(struct rt_export_request *));
+
+const char *rt_import_state_name(u8 state);
+const char *rt_export_state_name(u8 state);
+
+static inline u8 rt_import_get_state(struct rt_import_hook *ih) { return ih ? ih->import_state : TIS_DOWN; }
+static inline u8 rt_export_get_state(struct rt_export_hook *eh) { return eh ? eh->export_state : TES_DOWN; }
+
+void rt_set_export_state(struct rt_export_hook *hook, u8 state);
+
+void rte_import(struct rt_import_request *req, const net_addr *net, rte *new, struct rte_src *src);
+
+/* Get next rpe. If src is given, it must match. */
+struct rt_pending_export *rpe_next(struct rt_pending_export *rpe, struct rte_src *src);
+
+/* Mark the pending export processed */
+void rpe_mark_seen(struct rt_export_hook *hook, struct rt_pending_export *rpe);
+
+/* Get pending export seen status */
+int rpe_get_seen(struct rt_export_hook *hook, struct rt_pending_export *rpe);
+
+/* Types of route announcement, also used as flags */
+#define RA_UNDEF 0 /* Undefined RA type */
+#define RA_OPTIMAL 1 /* Announcement of optimal route change */
+#define RA_ACCEPTED 2 /* Announcement of first accepted route */
+#define RA_ANY 3 /* Announcement of any route change */
+#define RA_MERGED 4 /* Announcement of optimal route merged with next ones */
+
+/* Return value of preexport() callback */
+#define RIC_ACCEPT 1 /* Accepted by protocol */
+#define RIC_PROCESS 0 /* Process it through import filter */
+#define RIC_REJECT -1 /* Rejected by protocol */
+#define RIC_DROP -2 /* Silently dropped by protocol */
+
+#define rte_update channel_rte_import
+/**
+ * rte_update - enter a new update to a routing table
+ * @c: channel doing the update
+ * @net: network address
+ * @rte: a &rte representing the new route
+ * @src: old route source identifier
+ *
+ * This function imports a new route to the appropriate table (via the channel).
+ * Table keys are @net (obligatory) and @rte->attrs->src.
+ * Both the @net and @rte pointers can be local.
+ *
+ * The route attributes (@rte->attrs) are obligatory. They can be also allocated
+ * locally. Anyway, if you use an already-cached attribute object, you shall
+ * call rta_clone() on that object yourself. (This semantics may change in future.)
+ *
+ * If the route attributes are local, you may set @rte->attrs->src to NULL, then
+ * the protocol's default route source will be supplied.
+ *
+ * When rte_update() gets a route, it automatically validates it. This includes
+ * checking for validity of the given network and next hop addresses and also
+ * checking for host-scope or link-scope routes. Then the import filters are
+ * processed and if accepted, the route is passed to route table recalculation.
+ *
+ * The accepted routes are then inserted into the table, replacing the old route
+ * for the same @net identified by @src. Then the route is announced
+ * to all the channels connected to the table using the standard export mechanism.
+ * Setting @rte to NULL makes this a withdraw, otherwise @rte->src must be the same
+ * as @src.
+ *
+ * All memory used for temporary allocations is taken from a special linpool
+ * @rte_update_pool and freed when rte_update() finishes.
+ */
+void rte_update(struct channel *c, const net_addr *net, struct rte *rte, struct rte_src *src);
+
+extern list routing_tables;
+struct config;
+
+void rt_init(void);
+void rt_preconfig(struct config *);
+void rt_postconfig(struct config *);
+void rt_commit(struct config *new, struct config *old);
+void rt_lock_table(rtable *);
+void rt_unlock_table(rtable *);
+struct f_trie * rt_lock_trie(rtable *tab);
+void rt_unlock_trie(rtable *tab, struct f_trie *trie);
+void rt_subscribe(rtable *tab, struct rt_subscription *s);
+void rt_unsubscribe(struct rt_subscription *s);
+void rt_flowspec_link(rtable *src, rtable *dst);
+void rt_flowspec_unlink(rtable *src, rtable *dst);
+rtable *rt_setup(pool *, struct rtable_config *);
+static inline void rt_shutdown(rtable *r) { rfree(r->rp); }
+
+static inline net *net_find(rtable *tab, const net_addr *addr) { return (net *) fib_find(&tab->fib, addr); }
+static inline net *net_find_valid(rtable *tab, const net_addr *addr)
+{ net *n = net_find(tab, addr); return (n && n->routes && rte_is_valid(&n->routes->rte)) ? n : NULL; }
+static inline net *net_get(rtable *tab, const net_addr *addr) { return (net *) fib_get(&tab->fib, addr); }
+net *net_get(rtable *tab, const net_addr *addr);
+net *net_route(rtable *tab, const net_addr *n);
+int rt_examine(rtable *t, net_addr *a, struct channel *c, const struct filter *filter);
+rte *rt_export_merged(struct channel *c, rte ** feed, uint count, linpool *pool, int silent);
+void rt_refresh_begin(struct rt_import_request *);
+void rt_refresh_end(struct rt_import_request *);
+void rt_modify_stale(rtable *t, struct rt_import_request *);
+void rt_schedule_prune(rtable *t);
+void rte_dump(struct rte_storage *);
+void rte_free(struct rte_storage *);
+struct rte_storage *rte_store(const rte *, net *net, rtable *);
+void rt_dump(rtable *);
+void rt_dump_all(void);
+void rt_dump_hooks(rtable *);
+void rt_dump_hooks_all(void);
+int rt_reload_channel(struct channel *c);
+void rt_reload_channel_abort(struct channel *c);
+void rt_refeed_channel(struct channel *c);
+void rt_prune_sync(rtable *t, int all);
+struct rtable_config *rt_new_table(struct symbol *s, uint addr_type);
+
+static inline int rt_is_ip(rtable *tab)
+{ return (tab->addr_type == NET_IP4) || (tab->addr_type == NET_IP6); }
+
+static inline int rt_is_vpn(rtable *tab)
+{ return (tab->addr_type == NET_VPN4) || (tab->addr_type == NET_VPN6); }
+
+static inline int rt_is_roa(rtable *tab)
+{ return (tab->addr_type == NET_ROA4) || (tab->addr_type == NET_ROA6); }
+
+static inline int rt_is_flow(rtable *tab)
+{ return (tab->addr_type == NET_FLOW4) || (tab->addr_type == NET_FLOW6); }
+
+
+/* Default limit for ECMP next hops, defined in sysdep code */
+extern const int rt_default_ecmp;
+
+struct rt_show_data_rtable {
+ node n;
+ const char *name;
+ struct rt_exporter *table;
+ struct channel *export_channel;
+ struct channel *prefilter;
+ struct krt_proto *kernel;
+};
+
+struct rt_show_data {
+ struct cli *cli; /* Pointer back to the CLI */
+ net_addr *addr;
+ list tables;
+ struct rt_show_data_rtable *tab; /* Iterator over table list */
+ struct rt_show_data_rtable *last_table; /* Last table in output */
+ struct rt_export_request req; /* Export request in use */
+ int verbose, tables_defined_by;
+ const struct filter *filter;
+ struct proto *show_protocol;
+ struct proto *export_protocol;
+ struct channel *export_channel;
+ struct config *running_on_config;
+ struct rt_export_hook *kernel_export_hook;
+ int export_mode, addr_mode, primary_only, filtered, stats;
+
+ int net_counter, rt_counter, show_counter, table_counter;
+ int net_counter_last, rt_counter_last, show_counter_last;
+ int show_counter_last_flush;
+};
+
+void rt_show(struct rt_show_data *);
+struct rt_show_data_rtable * rt_show_add_exporter(struct rt_show_data *d, struct rt_exporter *t, const char *name);
+struct rt_show_data_rtable * rt_show_add_table(struct rt_show_data *d, struct rtable *t);
+
+/* Value of table definition mode in struct rt_show_data */
+#define RSD_TDB_DEFAULT 0 /* no table specified */
+#define RSD_TDB_INDIRECT 0 /* show route ... protocol P ... */
+#define RSD_TDB_ALL RSD_TDB_SET /* show route ... table all ... */
+#define RSD_TDB_DIRECT RSD_TDB_SET | RSD_TDB_NMN /* show route ... table X table Y ... */
+
+#define RSD_TDB_SET 0x1 /* internal: show empty tables */
+#define RSD_TDB_NMN 0x2 /* internal: need matching net */
+
+/* Value of export_mode in struct rt_show_data */
+#define RSEM_NONE 0 /* Export mode not used */
+#define RSEM_PREEXPORT 1 /* Routes ready for export, before filtering */
+#define RSEM_EXPORT 2 /* Routes accepted by export filter */
+#define RSEM_NOEXPORT 3 /* Routes rejected by export filter */
+#define RSEM_EXPORTED 4 /* Routes marked in export map */
+
+/* Host entry: Resolve hook for recursive nexthops */
+extern struct ea_class ea_gen_hostentry;
+struct hostentry_adata {
+ adata ad;
+ struct hostentry *he;
+ u32 labels[0];
+};
+
+void
+ea_set_hostentry(ea_list **to, struct rtable *dep, struct rtable *tab, ip_addr gw, ip_addr ll, u32 lnum, u32 labels[lnum]);
+
+void ea_show_hostentry(const struct adata *ad, byte *buf, uint size);
+void ea_show_nexthop_list(struct cli *c, struct nexthop_adata *nhad);
+
+/*
+ * Default protocol preferences
+ */
+
+#define DEF_PREF_DIRECT 240 /* Directly connected */
+#define DEF_PREF_STATIC 200 /* Static route */
+#define DEF_PREF_OSPF 150 /* OSPF intra-area, inter-area and type 1 external routes */
+#define DEF_PREF_BABEL 130 /* Babel */
+#define DEF_PREF_RIP 120 /* RIP */
+#define DEF_PREF_BGP 100 /* BGP */
+#define DEF_PREF_RPKI 100 /* RPKI */
+#define DEF_PREF_INHERITED 10 /* Routes inherited from other routing daemons */
+#define DEF_PREF_UNKNOWN 0 /* Routes with no preference set */
+
+/*
+ * Route Origin Authorization
+ */
+
+#define ROA_UNKNOWN 0
+#define ROA_VALID 1
+#define ROA_INVALID 2
+
+int net_roa_check(rtable *tab, const net_addr *n, u32 asn);
+
+#endif
HASH_REMOVE(p->session_hash_id, HASH_ID, s);
HASH_REMOVE(p->session_hash_ip, HASH_IP, s);
- sl_free(s);
+ TRACE(D_EVENTS, "Session to %I removed", s->addr);
- TRACE(D_EVENTS, "Session to %I removed", ip);
- sl_free(p->session_slab, s);
++ sl_free(s);
+ }
- birdloop_leave(p->loop);
+ static void
+ bfd_remove_session(struct bfd_proto *p, struct bfd_session *s)
+ {
+ birdloop_enter(p->p.loop);
+ bfd_remove_session_locked(p, s);
+ birdloop_leave(p->p.loop);
}
static void
init_list(&p->notify_list);
bfd_notify_init(p);
- add_tail(&bfd_proto_list, &p->bfd_node);
-
- birdloop_enter(p->loop);
+ add_tail(&bfd_global.proto_list, &p->bfd_node);
- if (cf->accept_ipv4 && cf->accept_direct)
- p->rx4_1 = bfd_open_rx_sk(p, 0, SK_IPV4);
+ if (!cf->strict_bind)
+ {
+ if (cf->accept_ipv4 && cf->accept_direct)
+ p->rx4_1 = bfd_open_rx_sk(p, 0, SK_IPV4);
- if (cf->accept_ipv4 && cf->accept_multihop)
- p->rx4_m = bfd_open_rx_sk(p, 1, SK_IPV4);
+ if (cf->accept_ipv4 && cf->accept_multihop)
+ p->rx4_m = bfd_open_rx_sk(p, 1, SK_IPV4);
- if (cf->accept_ipv6 && cf->accept_direct)
- p->rx6_1 = bfd_open_rx_sk(p, 0, SK_IPV6);
+ if (cf->accept_ipv6 && cf->accept_direct)
+ p->rx6_1 = bfd_open_rx_sk(p, 0, SK_IPV6);
- if (cf->accept_ipv6 && cf->accept_multihop)
- p->rx6_m = bfd_open_rx_sk(p, 1, SK_IPV6);
+ if (cf->accept_ipv6 && cf->accept_multihop)
+ p->rx6_m = bfd_open_rx_sk(p, 1, SK_IPV6);
+ }
- birdloop_leave(p->loop);
-
bfd_take_requests(p);
struct bfd_neighbor *n;
if ((new->accept_ipv4 != old->accept_ipv4) ||
(new->accept_ipv6 != old->accept_ipv6) ||
(new->accept_direct != old->accept_direct) ||
- (new->accept_multihop != old->accept_multihop))
+ (new->accept_multihop != old->accept_multihop) ||
+ (new->strict_bind != old->strict_bind))
return 0;
- birdloop_mask_wakeups(p->loop);
+ birdloop_mask_wakeups(p->p.loop);
WALK_LIST(ifa, p->iface_list)
bfd_reconfigure_iface(p, ifa, new);
.reconfigure = bfd_reconfigure,
.copy_config = bfd_copy_config,
};
+
+void
+bfd_build(void)
+{
+ proto_build(&proto_bfd);
++
++ bfd_global.lock = DOMAIN_NEW(rtable, "BFD Global");
++ init_list(&bfd_global.wait_list);
++ init_list(&bfd_global.pickup_list);
++ init_list(&bfd_global.proto_list);
+}
#endif
long page_size = 0;
-_Bool alloc_multipage = 0;
#ifdef HAVE_MMAP
+#define KEEP_PAGES_MAIN_MAX 256
+#define KEEP_PAGES_MAIN_MIN 8
+#define CLEANUP_PAGES_BULK 256
+
+STATIC_ASSERT(KEEP_PAGES_MAIN_MIN * 4 < KEEP_PAGES_MAIN_MAX);
+
static _Bool use_fake = 0;
+
+#if DEBUGGING
+struct free_page {
+ node unused[42];
+ node n;
+};
#else
-static _Bool use_fake = 1;
+struct free_page {
+ node n;
+};
+#endif
+
+struct free_pages {
+ list pages;
+ u16 min, max; /* Minimal and maximal number of free pages kept */
+ uint cnt; /* Number of empty pages */
+ event cleanup;
+};
+
+static void global_free_pages_cleanup_event(void *);
+
+static struct free_pages global_free_pages = {
+ .min = KEEP_PAGES_MAIN_MIN,
+ .max = KEEP_PAGES_MAIN_MAX,
+ .cleanup = { .hook = global_free_pages_cleanup_event },
+};
+
+uint *pages_kept = &global_free_pages.cnt;
+
+static void *
+alloc_sys_page(void)
+{
+ void *ptr = mmap(NULL, page_size, PROT_WRITE | PROT_READ, MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
+
+ if (ptr == MAP_FAILED)
+ bug("mmap(%lu) failed: %m", page_size);
+
+ return ptr;
+}
+
+extern int shutting_down; /* Shutdown requested. */
+
+#else // ! HAVE_MMAP
+#define use_fake 1
#endif
-void resource_sys_init(void)
+void *
+alloc_page(void)
{
+ if (use_fake)
+ {
+ void *ptr = NULL;
+ int err = posix_memalign(&ptr, page_size, page_size);
+
+ if (err || !ptr)
+ bug("posix_memalign(%lu) failed", (long unsigned int) page_size);
+
+ return ptr;
+ }
+
#ifdef HAVE_MMAP
- if (!(page_size = sysconf(_SC_PAGESIZE)))
- die("System page size must be non-zero");
+ struct free_pages *fps = &global_free_pages;
- if ((u64_popcount(page_size) > 1) || (page_size > 16384))
+ if (fps->cnt)
{
-#endif
- /* Too big or strange page, use the aligned allocator instead */
- page_size = 4096;
- use_fake = 1;
+ struct free_page *fp = SKIP_BACK(struct free_page, n, HEAD(fps->pages));
+ rem_node(&fp->n);
+ if ((--fps->cnt < fps->min) && !shutting_down)
- ev_schedule(&fps->cleanup);
++ ev_send(&global_work_list, &fps->cleanup);
+
+ bzero(fp, page_size);
+ return fp;
}
+
+ return alloc_sys_page();
+#endif
}
-void *
-alloc_sys_page(void)
+void
+free_page(void *ptr)
{
-#ifdef HAVE_MMAP
- if (!use_fake)
+ if (use_fake)
{
- if (alloc_multipage)
- {
- void *big = mmap(NULL, page_size * 2, PROT_WRITE | PROT_READ, MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
- if (big == MAP_FAILED)
- bug("mmap(%lu) failed: %m", page_size);
-
- uintptr_t offset = ((uintptr_t) big) % page_size;
- if (offset)
- {
- void *ret = big + page_size - offset;
- munmap(big, page_size - offset);
- munmap(ret + page_size, offset);
- return ret;
- }
- else
- {
- munmap(big + page_size, page_size);
- return big;
- }
- }
-
- void *ret = mmap(NULL, page_size, PROT_WRITE | PROT_READ, MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
- if (ret == MAP_FAILED)
- bug("mmap(%lu) failed: %m", page_size);
-
- return ret;
+ free(ptr);
+ return;
}
- else
+
+#ifdef HAVE_MMAP
+ struct free_pages *fps = &global_free_pages;
+ struct free_page *fp = ptr;
+
+ fp->n = (node) {};
+ add_tail(&fps->pages, &fp->n);
+
+ if ((++fps->cnt > fps->max) && !shutting_down)
- ev_schedule(&fps->cleanup);
++ ev_send(&global_work_list, &fps->cleanup);
#endif
+}
+
+#ifdef HAVE_MMAP
+static void
+global_free_pages_cleanup_event(void *data UNUSED)
+{
+ if (shutting_down)
+ return;
+
+ struct free_pages *fps = &global_free_pages;
+
+ while (fps->cnt / 2 < fps->min)
{
- void *ret = aligned_alloc(page_size, page_size);
- if (!ret)
- bug("aligned_alloc(%lu) failed", page_size);
- return ret;
+ struct free_page *fp = alloc_sys_page();
+ fp->n = (node) {};
+ add_tail(&fps->pages, &fp->n);
+ fps->cnt++;
+ }
+
+ for (uint seen = 0; (seen < CLEANUP_PAGES_BULK) && (fps->cnt > fps->max / 2); seen++)
+ {
+ struct free_page *fp = SKIP_BACK(struct free_page, n, TAIL(fps->pages));
+ rem_node(&fp->n);
+
+ if (munmap(fp, page_size) == 0)
+ fps->cnt--;
+ else if (errno == ENOMEM)
+ add_head(&fps->pages, &fp->n);
+ else
+ bug("munmap(%p) failed: %m", fp);
}
}
+#endif
void
-free_sys_page(void *ptr)
+resource_sys_init(void)
{
#ifdef HAVE_MMAP
- if (!use_fake)
+ ASSERT_DIE(global_free_pages.cnt == 0);
+
+ if (!(page_size = sysconf(_SC_PAGESIZE)))
+ die("System page size must be non-zero");
+
+ if (u64_popcount(page_size) == 1)
{
- if (munmap(ptr, page_size) < 0)
- bug("munmap(%p) failed: %m", ptr);
+ struct free_pages *fps = &global_free_pages;
+
+ init_list(&fps->pages);
+ global_free_pages_cleanup_event(NULL);
+ return;
}
- else
+
+ /* Too big or strange page, use the aligned allocator instead */
+ log(L_WARN "Got strange memory page size (%lu), using the aligned allocator instead", page_size);
+ use_fake = 1;
#endif
- free(ptr);
+
+ page_size = 4096;
}
timer *t;
int rv, timeout;
- birdloop_set_current(loop);
+ btime loop_begin = current_time();
- pthread_mutex_lock(&loop->mutex);
+ tmp_init(loop->pool);
+
+ birdloop_enter(loop);
while (1)
{
- events_fire(loop);
- timers_fire(&loop->time);
-
- times_update();
- if (events_waiting(loop))
+ timers_fire(&loop->time, 0);
+ if (ev_run_list(&loop->event_list))
timeout = 0;
else if (t = timers_first(&loop->time))
timeout = (tm_remains(t) TO_MS) + 1;
random_init();
net_init();
resource_init();
- timer_init();
+ birdloop_init();
olock_init();
- io_init();
rt_init();
+ io_init();
if_init();
// roa_init();
config_init();
#include "test/birdtest.h"
#include "lib/string.h"
+#include "lib/event.h"
++#include "lib/io-loop.h"
#ifdef HAVE_EXECINFO_H
#include <execinfo.h>
clock_gettime(CLOCK_MONOTONIC, &bt_begin);
bt_suite_case_begin = bt_suite_begin = bt_begin;
- ev_init_list(&global_event_list);
-
+ resource_init();
++ ev_init_list(&global_event_list, &main_birdloop, "Global event list in unit tests");
++ ev_init_list(&global_work_list, &main_birdloop, "Global work list in unit tests");
return;
usage:
log_init_debug("");
log_switch(bt_verbose != 0, NULL, NULL);
- resource_init();
+ the_bird_lock();
olock_init();
- timer_init();
+ birdloop_init();
- io_init();
rt_init();
+ io_init();
if_init();
config_init();
void bt_bird_cleanup(void)
{
- for (int i = 0; i < PROTOCOL__MAX; i++)
- class_to_protocol[i] = NULL;
-
config = new_config = NULL;
+ the_bird_unlock();
}
static char *
projects / thirdparty / bird.git / commitdiff
