942 lines
25 KiB
C
942 lines
25 KiB
C
/*
|
|
* BIRD -- OSPF
|
|
*
|
|
* (c) 1999 - 2000 Ondrej Filip <feela@network.cz>
|
|
*
|
|
* Can be freely distributed and used under the terms of the GNU GPL.
|
|
*/
|
|
|
|
/**
|
|
* DOC: Open Shortest Path First (OSPF)
|
|
*
|
|
* The OSPF protocol is quite complicated and its complex implemenation is
|
|
* split to many files. In |ospf.c|, you will find mainly the interface
|
|
* for communication with the core (e.g., reconfiguration hooks, shutdown
|
|
* and initialisation and so on). In |packet.c|, you will find various
|
|
* functions for sending and receiving generic OSPF packets. There are
|
|
* also routines for authentication and checksumming. File |iface.c| contains
|
|
* the interface state machine and functions for allocation and deallocation of OSPF's
|
|
* interface data structures. Source |neighbor.c| includes the neighbor state
|
|
* machine and functions for election of Designated Router and Backup
|
|
* Designated router. In |hello.c|, there are routines for sending
|
|
* and receiving of hello packets as well as functions for maintaining
|
|
* wait times and the inactivity timer. Files |lsreq.c|, |lsack.c|, |dbdes.c|
|
|
* contain functions for sending and receiving of link-state requests,
|
|
* link-state acknowledgements and database descriptions respectively.
|
|
* In |lsupd.c|, there are functions for sending and receiving
|
|
* of link-state updates and also the flooding algorithm. Source |topology.c| is
|
|
* a place where routines for searching LSAs in the link-state database,
|
|
* adding and deleting them reside, there also are functions for originating
|
|
* of various types of LSAs (router LSA, net LSA, external LSA). File |rt.c|
|
|
* contains routines for calculating the routing table. |lsalib.c| is a set
|
|
* of various functions for working with the LSAs (endianity conversions,
|
|
* calculation of checksum etc.).
|
|
*
|
|
* One instance of the protocol is able to hold LSA databases for
|
|
* multiple OSPF areas, to exchange routing information between
|
|
* multiple neighbors and to calculate the routing tables. The core
|
|
* structure is &proto_ospf to which multiple &ospf_area and
|
|
* &ospf_iface structures are connected. &ospf_area is also connected to
|
|
* &top_hash_graph which is a dynamic hashing structure that
|
|
* describes the link-state database. It allows fast search, addition
|
|
* and deletion. Each LSA is kept in two pieces: header and body. Both of them are
|
|
* kept in the endianity of the CPU.
|
|
*
|
|
* Every area has its own area_disp() which is
|
|
* responsible for late originating of router LSA, calculating
|
|
* of the routing table and it also ages and flushes the LSAs. This
|
|
* function is called in regular intervals.
|
|
* To every &ospf_iface, we connect one or more
|
|
* &ospf_neighbor's -- a structure containing many timers and queues
|
|
* for building adjacency and for exchange of routing messages.
|
|
*
|
|
* BIRD's OSPF implementation respects RFC2328 in every detail, but
|
|
* some of internal algorithms do differ. The RFC recommends making a snapshot
|
|
* of the link-state database when a new adjacency is forming and sending
|
|
* the database description packets based on the information in this
|
|
* snapshot. The database can be quite large in some networks, so
|
|
* rather we walk through a &slist structure which allows us to
|
|
* continue even if the actual LSA we were working with is deleted. New
|
|
* LSAs are added at the tail of this &slist.
|
|
*
|
|
* We also don't keep a separate OSPF routing table, because the core
|
|
* helps us by being able to recognize when a route is updated
|
|
* to an identical one and it suppresses the update automatically.
|
|
* Due to this, we can flush all the routes we've recalculated and
|
|
* also those we've deleted to the core's routing table and the
|
|
* core will take care of the rest. This simplifies the process
|
|
* and conserves memory.
|
|
*/
|
|
|
|
#include "ospf.h"
|
|
|
|
static int ospf_rte_better(struct rte *new, struct rte *old);
|
|
static int ospf_rte_same(struct rte *new, struct rte *old);
|
|
static void area_disp(timer *timer);
|
|
static void ospf_disp(timer *timer);
|
|
|
|
|
|
static int
|
|
ospf_start(struct proto *p)
|
|
{
|
|
struct proto_ospf *po = (struct proto_ospf *) p;
|
|
struct ospf_config *c = (struct ospf_config *) (p->cf);
|
|
struct ospf_area_config *ac;
|
|
struct ospf_area *oa;
|
|
struct area_net *anet, *antmp;
|
|
|
|
po->rfc1583 = c->rfc1583;
|
|
po->ebit = 0;
|
|
|
|
po->tick = c->tick;
|
|
po->disp_timer = tm_new(po->proto.pool);
|
|
po->disp_timer->data = po;
|
|
po->disp_timer->randomize = 0;
|
|
po->disp_timer->hook = ospf_disp;
|
|
po->disp_timer->recurrent = po->tick;
|
|
tm_start(po->disp_timer, 1);
|
|
|
|
fib_init(&po->efib, p->pool, sizeof(struct extfib), 16, init_efib);
|
|
init_list(&(po->iface_list));
|
|
init_list(&(po->area_list));
|
|
po->areano = 0;
|
|
if (EMPTY_LIST(c->area_list))
|
|
{
|
|
log(L_ERR "Cannot start, no OSPF areas configured!");
|
|
return PS_DOWN;
|
|
}
|
|
|
|
WALK_LIST(ac, c->area_list)
|
|
{
|
|
oa = mb_allocz(po->proto.pool, sizeof(struct ospf_area));
|
|
add_tail(&po->area_list, NODE oa);
|
|
po->areano++;
|
|
oa->stub = ac->stub;
|
|
oa->tick = ac->tick;
|
|
oa->areaid = ac->areaid;
|
|
oa->gr = ospf_top_new(po->proto.pool, po);
|
|
s_init_list(&(oa->lsal));
|
|
oa->rt = NULL;
|
|
oa->po = po;
|
|
oa->disp_timer = tm_new(po->proto.pool);
|
|
oa->disp_timer->data = oa;
|
|
oa->disp_timer->randomize = 0;
|
|
oa->disp_timer->hook = area_disp;
|
|
oa->disp_timer->recurrent = oa->tick;
|
|
tm_start(oa->disp_timer, 2);
|
|
init_list(&oa->net_list);
|
|
WALK_LIST(anet, ac->net_list)
|
|
{
|
|
antmp = mb_allocz(po->proto.pool, sizeof(struct area_net));
|
|
antmp->px.addr = anet->px.addr;
|
|
antmp->px.len = anet->px.len;
|
|
antmp->hidden = anet->hidden;
|
|
add_tail(&oa->net_list, NODE antmp);
|
|
}
|
|
fib_init(&oa->infib, po->proto.pool, sizeof(struct infib), 16,
|
|
init_infib);
|
|
}
|
|
return PS_UP;
|
|
}
|
|
|
|
static void
|
|
ospf_dump(struct proto *p)
|
|
{
|
|
struct ospf_iface *ifa;
|
|
struct ospf_neighbor *n;
|
|
struct proto_ospf *po = (struct proto_ospf *) p;
|
|
struct ospf_area *oa;
|
|
|
|
OSPF_TRACE(D_EVENTS, "Area number: %d", po->areano);
|
|
|
|
WALK_LIST(ifa, po->iface_list)
|
|
{
|
|
OSPF_TRACE(D_EVENTS, "Interface: %s", ifa->iface->name);
|
|
OSPF_TRACE(D_EVENTS, "state: %u", ifa->state);
|
|
OSPF_TRACE(D_EVENTS, "DR: %I", ifa->drid);
|
|
OSPF_TRACE(D_EVENTS, "BDR: %I", ifa->bdrid);
|
|
WALK_LIST(n, ifa->neigh_list)
|
|
{
|
|
OSPF_TRACE(D_EVENTS, " neighbor %I in state %u", n->rid, n->state);
|
|
}
|
|
}
|
|
|
|
WALK_LIST(NODE oa, po->area_list)
|
|
{
|
|
OSPF_TRACE(D_EVENTS, "LSA graph dump for area \"%I\" start:", oa->areaid);
|
|
ospf_top_dump(oa->gr, p);
|
|
OSPF_TRACE(D_EVENTS, "LSA graph dump for area \"%I\" finished",
|
|
oa->areaid);
|
|
}
|
|
neigh_dump_all();
|
|
}
|
|
|
|
static struct proto *
|
|
ospf_init(struct proto_config *c)
|
|
{
|
|
struct proto *p = proto_new(c, sizeof(struct proto_ospf));
|
|
struct proto_ospf *po = (struct proto_ospf *) p;
|
|
struct ospf_config *oc = (struct ospf_config *) c;
|
|
|
|
p->import_control = ospf_import_control;
|
|
p->make_tmp_attrs = ospf_make_tmp_attrs;
|
|
p->store_tmp_attrs = ospf_store_tmp_attrs;
|
|
p->rt_notify = ospf_rt_notify;
|
|
p->if_notify = ospf_iface_notify;
|
|
p->rte_better = ospf_rte_better;
|
|
p->rte_same = ospf_rte_same;
|
|
|
|
return p;
|
|
}
|
|
|
|
/* If new is better return 1 */
|
|
static int
|
|
ospf_rte_better(struct rte *new, struct rte *old)
|
|
{
|
|
if (new->u.ospf.metric1 == LSINFINITY)
|
|
return 0;
|
|
|
|
/* External paths are always longer that internal */
|
|
if (((new->attrs->source == RTS_OSPF)
|
|
|| (new->attrs->source == RTS_OSPF_IA))
|
|
&& (old->attrs->source == RTS_OSPF_EXT))
|
|
return 1;
|
|
if (((old->attrs->source == RTS_OSPF)
|
|
|| (old->attrs->source == RTS_OSPF_IA))
|
|
&& (new->attrs->source == RTS_OSPF_EXT))
|
|
return 0;
|
|
|
|
if (new->u.ospf.metric2 < old->u.ospf.metric2)
|
|
{
|
|
if (old->u.ospf.metric2 == LSINFINITY)
|
|
return 0; /* Old is E1, new is E2 */
|
|
return 1; /* Both are E2 */
|
|
}
|
|
|
|
if (new->u.ospf.metric2 > old->u.ospf.metric2)
|
|
{
|
|
if (new->u.ospf.metric2 == LSINFINITY)
|
|
return 1; /* New is E1, old is E2 */
|
|
return 0; /* Both are E2 */
|
|
}
|
|
|
|
/*
|
|
* E2 metrics are the same. It means that:
|
|
* 1) Paths are E2 with same metric
|
|
* 2) Paths are E1.
|
|
*/
|
|
|
|
if (new->u.ospf.metric1 < old->u.ospf.metric1)
|
|
return 1;
|
|
return 0; /* Old is shorter or same */
|
|
}
|
|
|
|
static int
|
|
ospf_rte_same(struct rte *new, struct rte *old)
|
|
{
|
|
/* new->attrs == old->attrs always */
|
|
return
|
|
new->u.ospf.metric1 == old->u.ospf.metric1 &&
|
|
new->u.ospf.metric2 == old->u.ospf.metric2 &&
|
|
new->u.ospf.tag == old->u.ospf.tag;
|
|
}
|
|
|
|
static ea_list *
|
|
ospf_build_attrs(ea_list * next, struct linpool *pool, u32 m1, u32 m2,
|
|
u32 tag)
|
|
{
|
|
struct ea_list *l =
|
|
lp_alloc(pool, sizeof(struct ea_list) + 3 * sizeof(eattr));
|
|
|
|
l->next = next;
|
|
l->flags = EALF_SORTED;
|
|
l->count = 3;
|
|
l->attrs[0].id = EA_OSPF_METRIC1;
|
|
l->attrs[0].flags = 0;
|
|
l->attrs[0].type = EAF_TYPE_INT | EAF_TEMP;
|
|
l->attrs[0].u.data = m1;
|
|
l->attrs[1].id = EA_OSPF_METRIC2;
|
|
l->attrs[1].flags = 0;
|
|
l->attrs[1].type = EAF_TYPE_INT | EAF_TEMP;
|
|
l->attrs[1].u.data = m2;
|
|
l->attrs[2].id = EA_OSPF_TAG;
|
|
l->attrs[2].flags = 0;
|
|
l->attrs[2].type = EAF_TYPE_INT | EAF_TEMP;
|
|
l->attrs[2].u.data = tag;
|
|
return l;
|
|
}
|
|
|
|
void
|
|
schedule_net_lsa(struct ospf_iface *ifa)
|
|
{
|
|
ifa->orignet = 1;
|
|
}
|
|
|
|
void
|
|
schedule_rt_lsa(struct ospf_area *oa)
|
|
{
|
|
struct proto *p = &oa->po->proto;
|
|
|
|
OSPF_TRACE(D_EVENTS, "Scheduling RT lsa origination for area %I.",
|
|
oa->areaid);
|
|
oa->origrt = 1;
|
|
}
|
|
|
|
void
|
|
schedule_rtcalc(struct proto_ospf *po)
|
|
{
|
|
struct proto *p = &po->proto;
|
|
|
|
if (po->calcrt)
|
|
return;
|
|
|
|
OSPF_TRACE(D_EVENTS, "Scheduling RT calculation.");
|
|
po->calcrt = 1;
|
|
}
|
|
|
|
/**
|
|
* area_disp - invokes link-state database aging, origination of
|
|
* router LSA and routing table calculation
|
|
* @timer: it's called every @ospf_area->tick seconds
|
|
*
|
|
* It invokes aging and when @ospf_area->origrt is set to 1, start
|
|
* function for origination of router LSA and network LSAs.
|
|
*/
|
|
void
|
|
area_disp(timer * timer)
|
|
{
|
|
struct ospf_area *oa = timer->data;
|
|
struct proto_ospf *po = oa->po;
|
|
struct ospf_iface *ifa;
|
|
|
|
/* Now try to originage rt_lsa */
|
|
if (oa->origrt)
|
|
originate_rt_lsa(oa);
|
|
|
|
/* Now try to originate network LSA's */
|
|
WALK_LIST(ifa, po->iface_list)
|
|
{
|
|
if (ifa->orignet && (ifa->an == oa->areaid))
|
|
originate_net_lsa(ifa);
|
|
}
|
|
|
|
/* Age LSA DB */
|
|
ospf_age(oa);
|
|
}
|
|
|
|
void
|
|
ospf_disp(timer * timer)
|
|
{
|
|
struct proto_ospf *po = timer->data;
|
|
|
|
/* Calculate routing table */
|
|
if (po->calcrt)
|
|
ospf_rt_spf (po);
|
|
po->calcrt = 0;
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
* ospf_import_control - accept or reject new route from nest's routing table
|
|
* @p: current instance of protocol
|
|
* @new: the new route
|
|
* @attrs: list of attributes
|
|
* @pool: pool for allocation of attributes
|
|
*
|
|
* Its quite simple. It does not accept our own routes and leaves the decision on
|
|
* import to the filters.
|
|
*/
|
|
|
|
int
|
|
ospf_import_control(struct proto *p, rte ** new, ea_list ** attrs,
|
|
struct linpool *pool)
|
|
{
|
|
rte *e = *new;
|
|
|
|
if (p == e->attrs->proto)
|
|
return -1; /* Reject our own routes */
|
|
*attrs = ospf_build_attrs(*attrs, pool, LSINFINITY, 10000, 0);
|
|
return 0; /* Leave decision to the filters */
|
|
}
|
|
|
|
struct ea_list *
|
|
ospf_make_tmp_attrs(struct rte *rt, struct linpool *pool)
|
|
{
|
|
return ospf_build_attrs(NULL, pool, rt->u.ospf.metric1, rt->u.ospf.metric2,
|
|
rt->u.ospf.tag);
|
|
}
|
|
|
|
void
|
|
ospf_store_tmp_attrs(struct rte *rt, struct ea_list *attrs)
|
|
{
|
|
rt->u.ospf.metric1 = ea_get_int(attrs, EA_OSPF_METRIC1, LSINFINITY);
|
|
rt->u.ospf.metric2 = ea_get_int(attrs, EA_OSPF_METRIC2, 10000);
|
|
rt->u.ospf.tag = ea_get_int(attrs, EA_OSPF_TAG, 0);
|
|
}
|
|
|
|
/**
|
|
* ospf_shutdown - Finish of OSPF instance
|
|
* @p: current instance of protocol
|
|
*
|
|
* RFC does not define any action that should be taken before router
|
|
* shutdown. To make my neighbors react as fast as possible, I send
|
|
* them hello packet with empty neighbor list. They should start
|
|
* their neighbor state machine with event %NEIGHBOR_1WAY.
|
|
*/
|
|
|
|
static int
|
|
ospf_shutdown(struct proto *p)
|
|
{
|
|
struct proto_ospf *po = (struct proto_ospf *) p;
|
|
struct ospf_iface *ifa;
|
|
OSPF_TRACE(D_EVENTS, "Shutdown requested");
|
|
|
|
/* And send to all my neighbors 1WAY */
|
|
WALK_LIST(ifa, po->iface_list) ospf_iface_shutdown(ifa);
|
|
|
|
return PS_DOWN;
|
|
}
|
|
|
|
void
|
|
ospf_rt_notify(struct proto *p, net * n, rte * new, rte * old UNUSED,
|
|
ea_list * attrs)
|
|
{
|
|
struct proto_ospf *po = (struct proto_ospf *) p;
|
|
|
|
/* Temporarily down write anything
|
|
OSPF_TRACE(D_EVENTS, "Got route %I/%d %s", p->name, n->n.prefix,
|
|
n->n.pxlen, new ? "up" : "down");
|
|
*/
|
|
|
|
if (new) /* Got some new route */
|
|
{
|
|
originate_ext_lsa(n, new, po, attrs);
|
|
}
|
|
else
|
|
{
|
|
u32 rtid = po->proto.cf->global->router_id;
|
|
struct ospf_area *oa;
|
|
struct top_hash_entry *en;
|
|
u32 pr = ipa_to_u32(n->n.prefix);
|
|
struct ospf_lsa_ext *ext;
|
|
int i;
|
|
int max = max_ext_lsa(n->n.pxlen);
|
|
|
|
/* Flush old external LSA */
|
|
WALK_LIST(oa, po->area_list)
|
|
{
|
|
for (i = 0; i < max; i++, pr++)
|
|
{
|
|
if (en = ospf_hash_find(oa->gr, pr, rtid, LSA_T_EXT))
|
|
{
|
|
ext = en->lsa_body;
|
|
if (ipa_compare(ext->netmask, ipa_mkmask(n->n.pxlen)) == 0)
|
|
{
|
|
ospf_lsupd_flush_nlsa(en, oa);
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
static void
|
|
ospf_get_status(struct proto *p, byte * buf)
|
|
{
|
|
struct proto_ospf *po = (struct proto_ospf *) p;
|
|
|
|
if (p->proto_state == PS_DOWN)
|
|
buf[0] = 0;
|
|
else
|
|
{
|
|
struct ospf_iface *ifa;
|
|
struct ospf_neighbor *n;
|
|
int adj = 0;
|
|
|
|
WALK_LIST(ifa, po->iface_list)
|
|
WALK_LIST(n, ifa->neigh_list) if (n->state == NEIGHBOR_FULL)
|
|
adj = 1;
|
|
|
|
if (adj == 0)
|
|
strcpy(buf, "Alone");
|
|
else
|
|
strcpy(buf, "Running");
|
|
}
|
|
}
|
|
|
|
static void
|
|
ospf_get_route_info(rte * rte, byte * buf, ea_list * attrs UNUSED)
|
|
{
|
|
char met = ' ';
|
|
char type = ' ';
|
|
|
|
if (rte->attrs->source == RTS_OSPF_EXT)
|
|
{
|
|
met = '1';
|
|
type = 'E';
|
|
|
|
}
|
|
if (rte->u.ospf.metric2 != LSINFINITY)
|
|
met = '2';
|
|
if (rte->attrs->source == RTS_OSPF_IA)
|
|
type = 'A';
|
|
if (rte->attrs->source == RTS_OSPF)
|
|
type = 'I';
|
|
buf += bsprintf(buf, " %c", type);
|
|
if (met != ' ')
|
|
buf += bsprintf(buf, "%c", met);
|
|
buf += bsprintf(buf, " (%d/%d", rte->pref, rte->u.ospf.metric1);
|
|
if (rte->u.ospf.metric2 != LSINFINITY)
|
|
buf += bsprintf(buf, "/%d", rte->u.ospf.metric2);
|
|
buf += bsprintf(buf, ")");
|
|
if (rte->attrs->source == RTS_OSPF_EXT && rte->u.ospf.tag)
|
|
{
|
|
buf += bsprintf(buf, " [%x]", rte->u.ospf.tag);
|
|
}
|
|
}
|
|
|
|
static int
|
|
ospf_get_attr(eattr * a, byte * buf)
|
|
{
|
|
switch (a->id)
|
|
{
|
|
case EA_OSPF_METRIC1:
|
|
bsprintf(buf, "metric1");
|
|
return GA_NAME;
|
|
case EA_OSPF_METRIC2:
|
|
bsprintf(buf, "metric2");
|
|
return GA_NAME;
|
|
case EA_OSPF_TAG:
|
|
bsprintf(buf, "tag: %08x", a->u.data);
|
|
return GA_FULL;
|
|
default:
|
|
return GA_UNKNOWN;
|
|
}
|
|
}
|
|
|
|
static int
|
|
ospf_patt_compare(struct ospf_iface_patt *a, struct ospf_iface_patt *b)
|
|
{
|
|
return ((a->type == b->type) && (a->priority == b->priority));
|
|
}
|
|
|
|
/**
|
|
* ospf_reconfigure - reconfiguration hook
|
|
* @p: current instance of protocol (with old configuration)
|
|
* @c: new configuration requested by user
|
|
*
|
|
* This hook tries to be a little bit intelligent. Instance of OSPF
|
|
* will survive change of many constants like hello interval,
|
|
* password change, addition or deletion of some neighbor on
|
|
* nonbroadcast network, cost of interface, etc.
|
|
*/
|
|
static int
|
|
ospf_reconfigure(struct proto *p, struct proto_config *c)
|
|
{
|
|
struct ospf_config *old = (struct ospf_config *) (p->cf);
|
|
struct ospf_config *new = (struct ospf_config *) c;
|
|
struct ospf_area_config *ac1, *ac2;
|
|
struct proto_ospf *po = (struct proto_ospf *) p;
|
|
struct ospf_iface_patt *ip1, *ip2;
|
|
struct ospf_iface *ifa;
|
|
struct nbma_node *nb1, *nb2, *nbnx;
|
|
struct ospf_area *oa = NULL;
|
|
struct area_net *anet, *antmp;
|
|
int found;
|
|
|
|
po->rfc1583 = new->rfc1583;
|
|
schedule_rtcalc(po);
|
|
|
|
ac1 = HEAD(old->area_list);
|
|
ac2 = HEAD(new->area_list);
|
|
|
|
/* I should get it in same order */
|
|
|
|
while (((NODE(ac1))->next != NULL) && ((NODE(ac2))->next != NULL))
|
|
{
|
|
if (ac1->areaid != ac2->areaid)
|
|
return 0;
|
|
if (ac1->stub != ac2->stub)
|
|
return 0; /* FIXME: non zero values can change */
|
|
|
|
WALK_LIST(oa, po->area_list) if (oa->areaid == ac2->areaid)
|
|
break;
|
|
|
|
if (!oa)
|
|
return 0;
|
|
|
|
if (ac1->tick != ac2->tick)
|
|
{
|
|
if (oa->areaid == ac2->areaid)
|
|
{
|
|
oa->tick = ac2->tick;
|
|
tm_start(oa->disp_timer, oa->tick);
|
|
OSPF_TRACE(D_EVENTS,
|
|
"Changing tick interval on area %I from %d to %d",
|
|
oa->areaid, ac1->tick, ac2->tick);
|
|
break;
|
|
}
|
|
}
|
|
|
|
/* Change net_list */
|
|
WALK_LIST_DELSAFE(anet, antmp, oa->net_list)
|
|
{
|
|
rem_node(NODE anet);
|
|
mb_free(anet);
|
|
}
|
|
WALK_LIST(anet, ac2->net_list)
|
|
{
|
|
antmp = mb_alloc(p->pool, sizeof(struct area_net));
|
|
antmp->px.addr = anet->px.addr;
|
|
antmp->px.len = anet->px.len;
|
|
antmp->hidden = anet->hidden;
|
|
add_tail(&oa->net_list, NODE antmp);
|
|
}
|
|
|
|
if (!iface_patts_equal(&ac1->patt_list, &ac2->patt_list,
|
|
(void *) ospf_patt_compare))
|
|
return 0;
|
|
|
|
WALK_LIST(ifa, po->iface_list)
|
|
{
|
|
if (ip1 = (struct ospf_iface_patt *)
|
|
iface_patt_match(&ac1->patt_list, ifa->iface))
|
|
{
|
|
/* Now reconfigure interface */
|
|
if (!(ip2 = (struct ospf_iface_patt *)
|
|
iface_patt_match(&ac2->patt_list, ifa->iface)))
|
|
return 0;
|
|
|
|
/* HELLO TIMER */
|
|
if (ip1->helloint != ip2->helloint)
|
|
{
|
|
ifa->helloint = ip2->helloint;
|
|
ifa->hello_timer->recurrent = ifa->helloint;
|
|
tm_start(ifa->hello_timer, ifa->helloint);
|
|
OSPF_TRACE(D_EVENTS,
|
|
"Changing hello interval on interface %s from %d to %d",
|
|
ifa->iface->name, ip1->helloint, ip2->helloint);
|
|
}
|
|
|
|
/* POLL TIMER */
|
|
if (ip1->pollint != ip2->pollint)
|
|
{
|
|
ifa->pollint = ip2->helloint;
|
|
ifa->poll_timer->recurrent = ifa->pollint;
|
|
tm_start(ifa->poll_timer, ifa->pollint);
|
|
OSPF_TRACE(D_EVENTS,
|
|
"Changing poll interval on interface %s from %d to %d",
|
|
ifa->iface->name, ip1->pollint, ip2->pollint);
|
|
}
|
|
|
|
/* COST */
|
|
if (ip1->cost != ip2->cost)
|
|
{
|
|
ifa->cost = ip2->cost;
|
|
OSPF_TRACE(D_EVENTS,
|
|
"Changing cost interface %s from %d to %d",
|
|
ifa->iface->name, ip1->cost, ip2->cost);
|
|
schedule_rt_lsa(ifa->oa);
|
|
}
|
|
|
|
/* strict nbma */
|
|
if ((ip1->strictnbma == 0) && (ip2->strictnbma != 0))
|
|
{
|
|
ifa->strictnbma = ip2->strictnbma;
|
|
OSPF_TRACE(D_EVENTS,
|
|
"Interface %s is now strict NBMA.", ifa->iface->name);
|
|
}
|
|
if ((ip1->strictnbma != 0) && (ip2->strictnbma == 0))
|
|
{
|
|
ifa->strictnbma = ip2->strictnbma;
|
|
OSPF_TRACE(D_EVENTS,
|
|
"Interface %s is no longer strict NBMA.",
|
|
ifa->iface->name);
|
|
}
|
|
|
|
/* stub */
|
|
if ((ip1->stub == 0) && (ip2->stub != 0))
|
|
{
|
|
ifa->stub = ip2->stub;
|
|
OSPF_TRACE(D_EVENTS, "Interface %s is now stub.", ifa->iface->name);
|
|
}
|
|
if ((ip1->stub != 0) && (ip2->stub == 0) &&
|
|
((ifa->ioprob & OSPF_I_IP) == 0) &&
|
|
(((ifa->ioprob & OSPF_I_MC) == 0) || (ifa->type == OSPF_IT_NBMA)))
|
|
{
|
|
ifa->stub = ip2->stub;
|
|
OSPF_TRACE(D_EVENTS,
|
|
"Interface %s is no longer stub.", ifa->iface->name);
|
|
}
|
|
|
|
/* AUTHENTICATION */
|
|
if (ip1->autype != ip2->autype)
|
|
{
|
|
ifa->autype = ip2->autype;
|
|
OSPF_TRACE(D_EVENTS,
|
|
"Changing authentication type on interface %s",
|
|
ifa->iface->name);
|
|
}
|
|
if (strncmp(ip1->password, ip2->password, 8) != 0)
|
|
{
|
|
memcpy(ifa->aukey, ip2->password, 8);
|
|
OSPF_TRACE(D_EVENTS,
|
|
"Changing password on interface %s", ifa->iface->name);
|
|
}
|
|
|
|
/* RXMT */
|
|
if (ip1->rxmtint != ip2->rxmtint)
|
|
{
|
|
ifa->rxmtint = ip2->rxmtint;
|
|
OSPF_TRACE(D_EVENTS,
|
|
"Changing retransmit interval on interface %s from %d to %d",
|
|
ifa->iface->name, ip1->rxmtint, ip2->rxmtint);
|
|
}
|
|
|
|
/* WAIT */
|
|
if (ip1->waitint != ip2->waitint)
|
|
{
|
|
ifa->waitint = ip2->waitint;
|
|
if (ifa->wait_timer->expires != 0)
|
|
tm_start(ifa->wait_timer, ifa->waitint);
|
|
OSPF_TRACE(D_EVENTS,
|
|
"Changing wait interval on interface %s from %d to %d",
|
|
ifa->iface->name, ip1->waitint, ip2->waitint);
|
|
}
|
|
|
|
/* INFTRANS */
|
|
if (ip1->inftransdelay != ip2->inftransdelay)
|
|
{
|
|
ifa->inftransdelay = ip2->inftransdelay;
|
|
OSPF_TRACE(D_EVENTS,
|
|
"Changing transmit delay on interface %s from %d to %d",
|
|
ifa->iface->name, ip1->inftransdelay,
|
|
ip2->inftransdelay);
|
|
}
|
|
|
|
/* DEAD COUNT */
|
|
if (ip1->deadc != ip2->deadc)
|
|
{
|
|
ifa->deadc = ip2->deadc;
|
|
OSPF_TRACE(D_EVENTS,
|
|
"Changing dead count on interface %s from %d to %d",
|
|
ifa->iface->name, ip1->deadc, ip2->deadc);
|
|
}
|
|
|
|
/* NBMA LIST */
|
|
/* First remove old */
|
|
WALK_LIST_DELSAFE(nb1, nbnx, ifa->nbma_list)
|
|
{
|
|
found = 0;
|
|
WALK_LIST(nb2, ip2->nbma_list)
|
|
if (ipa_compare(nb1->ip, nb2->ip) == 0)
|
|
{
|
|
found = 1;
|
|
if (nb1->eligible != nb2->eligible)
|
|
OSPF_TRACE(D_EVENTS,
|
|
"Changing neighbor eligibility %I on interface %s",
|
|
nb1->ip, ifa->iface->name);
|
|
break;
|
|
}
|
|
|
|
if (!found)
|
|
{
|
|
OSPF_TRACE(D_EVENTS,
|
|
"Removing NBMA neighbor %I on interface %s",
|
|
nb1->ip, ifa->iface->name);
|
|
rem_node(NODE nb1);
|
|
mb_free(nb1);
|
|
}
|
|
}
|
|
/* And then add new */
|
|
WALK_LIST(nb2, ip2->nbma_list)
|
|
{
|
|
found = 0;
|
|
WALK_LIST(nb1, ifa->nbma_list)
|
|
if (ipa_compare(nb1->ip, nb2->ip) == 0)
|
|
{
|
|
found = 1;
|
|
break;
|
|
}
|
|
if (!found)
|
|
{
|
|
nb1 = mb_alloc(p->pool, sizeof(struct nbma_node));
|
|
nb1->ip = nb2->ip;
|
|
nb1->eligible = nb2->eligible;
|
|
add_tail(&ifa->nbma_list, NODE nb1);
|
|
OSPF_TRACE(D_EVENTS,
|
|
"Adding NBMA neighbor %I on interface %s",
|
|
nb1->ip, ifa->iface->name);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
NODE ac1 = (NODE(ac1))->next;
|
|
NODE ac2 = (NODE(ac2))->next;
|
|
}
|
|
|
|
if (((NODE(ac1))->next) != ((NODE(ac2))->next))
|
|
return 0; /* One is not null */
|
|
|
|
return 1; /* Everything OK :-) */
|
|
}
|
|
|
|
void
|
|
ospf_sh_neigh(struct proto *p, char *iff)
|
|
{
|
|
struct ospf_iface *ifa = NULL, *f;
|
|
struct ospf_neighbor *n;
|
|
struct proto_ospf *po = (struct proto_ospf *) p;
|
|
|
|
if (p->proto_state != PS_UP)
|
|
{
|
|
cli_msg(-1013, "%s: is not up", p->name);
|
|
cli_msg(0, "");
|
|
return;
|
|
}
|
|
|
|
if (iff != NULL)
|
|
{
|
|
WALK_LIST(f, po->iface_list)
|
|
{
|
|
if (strcmp(iff, f->iface->name) == 0)
|
|
{
|
|
ifa = f;
|
|
break;
|
|
}
|
|
}
|
|
if (ifa == NULL)
|
|
{
|
|
cli_msg(0, "");
|
|
return;
|
|
}
|
|
cli_msg(-1013, "%s:", p->name);
|
|
cli_msg(-1013, "%-12s\t%3s\t%-15s\t%-5s\t%-12s\t%-10s", "Router ID",
|
|
"Pri", " State", "DTime", "Router IP", "Interface");
|
|
WALK_LIST(n, ifa->neigh_list) ospf_sh_neigh_info(n);
|
|
cli_msg(0, "");
|
|
return;
|
|
}
|
|
|
|
cli_msg(-1013, "%s:", p->name);
|
|
cli_msg(-1013, "%-12s\t%3s\t%-15s\t%-5s\t%-12s\t%-10s", "Router ID", "Pri",
|
|
" State", "DTime", "Router IP", "Interface");
|
|
WALK_LIST(ifa, po->iface_list)
|
|
WALK_LIST(n, ifa->neigh_list) ospf_sh_neigh_info(n);
|
|
cli_msg(0, "");
|
|
}
|
|
|
|
void
|
|
ospf_sh(struct proto *p)
|
|
{
|
|
struct ospf_area *oa;
|
|
struct proto_ospf *po = (struct proto_ospf *) p;
|
|
struct ospf_iface *ifa;
|
|
struct ospf_neighbor *n;
|
|
int ifano;
|
|
int nno;
|
|
int adjno;
|
|
|
|
if (p->proto_state != PS_UP)
|
|
{
|
|
cli_msg(-1014, "%s: is not up", p->name);
|
|
cli_msg(0, "");
|
|
return;
|
|
}
|
|
|
|
cli_msg(-1014, "%s:", p->name);
|
|
cli_msg(-1014, "Number of areas: %u", po->areano);
|
|
|
|
WALK_LIST(oa, po->area_list)
|
|
{
|
|
cli_msg(-1014, "\tArea: %I (%u) %s", oa->areaid, oa->areaid,
|
|
oa->areaid == 0 ? "[BACKBONE]" : "");
|
|
ifano = 0;
|
|
nno = 0;
|
|
adjno = 0;
|
|
WALK_LIST(ifa, po->iface_list)
|
|
{
|
|
if (oa == ifa->oa)
|
|
ifano++;
|
|
WALK_LIST(n, ifa->neigh_list)
|
|
{
|
|
nno++;
|
|
if (n->state == NEIGHBOR_FULL)
|
|
adjno++;
|
|
}
|
|
}
|
|
cli_msg(-1014, "\t\tStub:\t%s", oa->stub ? "Yes" : "No");
|
|
cli_msg(-1014, "\t\tRT scheduler tick:\t%u", oa->tick);
|
|
cli_msg(-1014, "\t\tNumber of interfaces:\t%u", ifano);
|
|
cli_msg(-1014, "\t\tNumber of LSAs in DB:\t%u", oa->gr->hash_entries);
|
|
cli_msg(-1014, "\t\tNumber of neighbors:\t%u", nno);
|
|
cli_msg(-1014, "\t\tNumber of adjacent neighbors:\t%u", adjno);
|
|
if (!EMPTY_LIST(oa->net_list))
|
|
{
|
|
struct area_net *anet;
|
|
cli_msg(-1014, "\t\tArea networks:");
|
|
WALK_LIST(anet, oa->net_list)
|
|
{
|
|
cli_msg(-1014, "\t\t\t%1I/%u\t%s", anet->px.addr, anet->px.len,
|
|
anet->hidden ? "Hidden" : "Advertise");
|
|
}
|
|
}
|
|
}
|
|
cli_msg(0, "");
|
|
}
|
|
|
|
void
|
|
ospf_sh_iface(struct proto *p, char *iff)
|
|
{
|
|
struct proto_ospf *po = (struct proto_ospf *) p;
|
|
struct ospf_iface *ifa = NULL, *f;
|
|
|
|
if (p->proto_state != PS_UP)
|
|
{
|
|
cli_msg(-1015, "%s: is not up", p->name);
|
|
cli_msg(0, "");
|
|
return;
|
|
}
|
|
|
|
if (iff != NULL)
|
|
{
|
|
WALK_LIST(f, po->iface_list)
|
|
{
|
|
if (strcmp(iff, f->iface->name) == 0)
|
|
{
|
|
ifa = f;
|
|
break;
|
|
}
|
|
}
|
|
|
|
if (ifa == NULL)
|
|
{
|
|
cli_msg(0, "");
|
|
return;
|
|
}
|
|
cli_msg(-1015, "%s:", p->name);
|
|
ospf_iface_info(ifa);
|
|
cli_msg(0, "");
|
|
return;
|
|
}
|
|
cli_msg(-1015, "%s:", p->name);
|
|
WALK_LIST(ifa, po->iface_list) ospf_iface_info(ifa);
|
|
cli_msg(0, "");
|
|
}
|
|
|
|
struct protocol proto_ospf = {
|
|
name:"OSPF",
|
|
template:"ospf%d",
|
|
attr_class:EAP_OSPF,
|
|
init:ospf_init,
|
|
dump:ospf_dump,
|
|
start:ospf_start,
|
|
shutdown:ospf_shutdown,
|
|
get_route_info:ospf_get_route_info,
|
|
get_attr:ospf_get_attr,
|
|
get_status:ospf_get_status,
|
|
reconfigure:ospf_reconfigure
|
|
};
|