0
0
mirror of https://gitlab.nic.cz/labs/bird.git synced 2024-11-19 01:28:44 +00:00
bird/nest/neighbor.c
Ondrej Zajicek (work) b8bbbbaf56 Nest: Fix neighbor handling for colliding ranges
Resolve neighbors using longest prefix match. Although interface ranges
should not generally collide, it may happen for unnumbered links.

Thanks to Kenth Eriksson for the bugreport.
2020-05-11 04:29:36 +02:00

570 lines
15 KiB
C

/*
* BIRD -- Neighbor Cache
*
* (c) 1998--2000 Martin Mares <mj@ucw.cz>
* (c) 2008--2018 Ondrej Zajicek <santiago@crfreenet.org>
* (c) 2008--2018 CZ.NIC z.s.p.o.
*
* Can be freely distributed and used under the terms of the GNU GPL.
*/
/**
* DOC: Neighbor cache
*
* Most routing protocols need to associate their internal state data with
* neighboring routers, check whether an address given as the next hop attribute
* of a route is really an address of a directly connected host and which
* interface is it connected through. Also, they often need to be notified when
* a neighbor ceases to exist or when their long awaited neighbor becomes
* connected. The neighbor cache is there to solve all these problems.
*
* The neighbor cache maintains a collection of neighbor entries. Each entry
* represents one IP address corresponding to either our directly connected
* neighbor or our own end of the link (when the scope of the address is set to
* %SCOPE_HOST) together with per-neighbor data belonging to a single protocol.
* A neighbor entry may be bound to a specific interface, which is required for
* link-local IP addresses and optional for global IP addresses.
*
* Neighbor cache entries are stored in a hash table, which is indexed by triple
* (protocol, IP, requested-iface), so if both regular and iface-bound neighbors
* are requested, they are represented by two neighbor cache entries. Active
* entries are also linked in per-interface list (allowing quick processing of
* interface change events). Inactive entries exist only when the protocol has
* explicitly requested it via the %NEF_STICKY flag because it wishes to be
* notified when the node will again become a neighbor. Such entries are instead
* linked in a special list, which is walked whenever an interface changes its
* state to up. Neighbor entry VRF association is implied by respective
* protocol.
*
* Besides the already mentioned %NEF_STICKY flag, there is also %NEF_ONLINK,
* which specifies that neighbor should be considered reachable on given iface
* regardless of associated address ranges, and %NEF_IFACE, which represents
* pseudo-neighbor entry for whole interface (and uses %IPA_NONE IP address).
*
* When a neighbor event occurs (a neighbor gets disconnected or a sticky
* inactive neighbor becomes connected), the protocol hook neigh_notify() is
* called to advertise the change.
*/
#undef LOCAL_DEBUG
#include "nest/bird.h"
#include "nest/iface.h"
#include "nest/protocol.h"
#include "lib/hash.h"
#include "lib/resource.h"
#define NEIGH_HASH_SIZE 256
#define NEIGH_HASH_OFFSET 24
static slab *neigh_slab;
static list neigh_hash_table[NEIGH_HASH_SIZE], sticky_neigh_list;
static inline uint
neigh_hash(struct proto *p, ip_addr a, struct iface *i)
{
return (p->hash_key ^ ipa_hash(a) ^ ptr_hash(i)) >> NEIGH_HASH_OFFSET;
}
static inline int
ifa_better(struct ifa *a, struct ifa *b)
{
return a && (!b || (a->prefix.pxlen > b->prefix.pxlen));
}
static inline int
scope_better(int sa, int sb)
{
/* Order per preference: -1 unknown, 0 for remote, 1 for local */
sa = (sa < 0) ? sa : !sa;
sb = (sb < 0) ? sb : !sb;
return sa > sb;
}
static inline int
scope_remote(int sa, int sb)
{
return (sa > SCOPE_HOST) && (sb > SCOPE_HOST);
}
static int
if_connected(ip_addr a, struct iface *i, struct ifa **ap, uint flags)
{
struct ifa *b, *addr = NULL;
/* Handle iface pseudo-neighbors */
if (flags & NEF_IFACE)
return *ap = NULL, (i->flags & IF_UP) ? SCOPE_HOST : -1;
/* Host addresses match even if iface is down */
WALK_LIST(b, i->addrs)
if (ipa_equal(a, b->ip))
return *ap = b, SCOPE_HOST;
/* Rest do not match if iface is down */
if (!(i->flags & IF_UP))
return *ap = NULL, -1;
/* Regular neighbors */
WALK_LIST(b, i->addrs)
{
if (b->flags & IA_PEER)
{
if (ipa_equal(a, b->opposite) && ifa_better(b, addr))
addr = b;
}
else
{
if (ipa_in_netX(a, &b->prefix) && ifa_better(b, addr))
{
/* Do not allow IPv4 network and broadcast addresses */
if (ipa_is_ip4(a) &&
(net_pxlen(&b->prefix) < (IP4_MAX_PREFIX_LENGTH - 1)) &&
(ipa_equal(a, net_prefix(&b->prefix)) || /* Network address */
ipa_equal(a, b->brd))) /* Broadcast */
return *ap = NULL, -1;
addr = b;
}
}
}
/* Return found address */
if (addr)
return *ap = addr, addr->scope;
/* Handle ONLINK flag */
if (flags & NEF_ONLINK)
return *ap = NULL, ipa_classify(a) & IADDR_SCOPE_MASK;
return *ap = NULL, -1;
}
static inline int
if_connected_any(ip_addr a, struct iface *vrf, uint vrf_set, struct iface **iface, struct ifa **addr, uint flags)
{
struct iface *i;
struct ifa *b;
int s, scope = -1;
*iface = NULL;
*addr = NULL;
/* Prefer SCOPE_HOST or longer prefix */
WALK_LIST(i, iface_list)
if ((!vrf_set || vrf == i->master) && ((s = if_connected(a, i, &b, flags)) >= 0))
if (scope_better(s, scope) || (scope_remote(s, scope) && ifa_better(b, *addr)))
{
*iface = i;
*addr = b;
scope = s;
}
return scope;
}
/* Is ifa @a subnet of any ifa on iface @ib ? */
static inline int
ifa_intersect(struct ifa *a, struct iface *ib)
{
struct ifa *b;
WALK_LIST(b, ib->addrs)
if (net_in_netX(&a->prefix, &b->prefix))
return 1;
return 0;
}
/* Is any ifa of iface @ia subnet of any ifa on iface @ib ? */
static inline int
if_intersect(struct iface *ia, struct iface *ib)
{
struct ifa *a, *b;
WALK_LIST(a, ia->addrs)
WALK_LIST(b, ib->addrs)
if (net_in_netX(&a->prefix, &b->prefix))
return 1;
return 0;
}
/**
* neigh_find - find or create a neighbor entry
* @p: protocol which asks for the entry
* @a: IP address of the node to be searched for
* @iface: optionally bound neighbor to this iface (may be NULL)
* @flags: %NEF_STICKY for sticky entry, %NEF_ONLINK for onlink entry
*
* Search the neighbor cache for a node with given IP address. Iface can be
* specified for link-local addresses or for cases, where neighbor is expected
* on given interface. If it is found, a pointer to the neighbor entry is
* returned. If no such entry exists and the node is directly connected on one
* of our active interfaces, a new entry is created and returned to the caller
* with protocol-dependent fields initialized to zero. If the node is not
* connected directly or *@a is not a valid unicast IP address, neigh_find()
* returns %NULL.
*/
neighbor *
neigh_find(struct proto *p, ip_addr a, struct iface *iface, uint flags)
{
neighbor *n;
int class, scope = -1;
uint h = neigh_hash(p, a, iface);
struct iface *ifreq = iface;
struct ifa *addr = NULL;
WALK_LIST(n, neigh_hash_table[h]) /* Search the cache */
if ((n->proto == p) && ipa_equal(n->addr, a) && (n->ifreq == iface))
return n;
if (flags & NEF_IFACE)
{
if (ipa_nonzero(a) || !iface)
return NULL;
}
else
{
class = ipa_classify(a);
if (class < 0) /* Invalid address */
return NULL;
if (((class & IADDR_SCOPE_MASK) == SCOPE_HOST) ||
(((class & IADDR_SCOPE_MASK) == SCOPE_LINK) && !iface) ||
!(class & IADDR_HOST))
return NULL; /* Bad scope or a somecast */
}
if ((flags & NEF_ONLINK) && !iface)
return NULL;
if (iface)
{
scope = if_connected(a, iface, &addr, flags);
iface = (scope < 0) ? NULL : iface;
}
else
scope = if_connected_any(a, p->vrf, p->vrf_set, &iface, &addr, flags);
/* scope < 0 means i don't know neighbor */
/* scope >= 0 <=> iface != NULL */
if ((scope < 0) && !(flags & NEF_STICKY))
return NULL;
n = sl_alloc(neigh_slab);
memset(n, 0, sizeof(neighbor));
add_tail(&neigh_hash_table[h], &n->n);
add_tail((scope >= 0) ? &iface->neighbors : &sticky_neigh_list, &n->if_n);
n->addr = a;
n->ifa = addr;
n->iface = iface;
n->ifreq = ifreq;
n->proto = p;
n->flags = flags;
n->scope = scope;
return n;
}
/**
* neigh_dump - dump specified neighbor entry.
* @n: the entry to dump
*
* This functions dumps the contents of a given neighbor entry to debug output.
*/
void
neigh_dump(neighbor *n)
{
debug("%p %I %s %s ", n, n->addr,
n->iface ? n->iface->name : "[]",
n->ifreq ? n->ifreq->name : "[]");
debug("%s %p %08x scope %s", n->proto->name, n->data, n->aux, ip_scope_text(n->scope));
if (n->flags & NEF_STICKY)
debug(" STICKY");
if (n->flags & NEF_ONLINK)
debug(" ONLINK");
debug("\n");
}
/**
* neigh_dump_all - dump all neighbor entries.
*
* This function dumps the contents of the neighbor cache to debug output.
*/
void
neigh_dump_all(void)
{
neighbor *n;
int i;
debug("Known neighbors:\n");
for(i=0; i<NEIGH_HASH_SIZE; i++)
WALK_LIST(n, neigh_hash_table[i])
neigh_dump(n);
debug("\n");
}
static inline void
neigh_notify(neighbor *n)
{
if (n->proto->neigh_notify && (n->proto->proto_state != PS_STOP))
n->proto->neigh_notify(n);
}
static void
neigh_up(neighbor *n, struct iface *i, struct ifa *a, int scope)
{
DBG("Waking up sticky neighbor %I\n", n->addr);
n->iface = i;
n->ifa = a;
n->scope = scope;
rem_node(&n->if_n);
add_tail(&i->neighbors, &n->if_n);
neigh_notify(n);
}
static void
neigh_down(neighbor *n)
{
DBG("Flushing neighbor %I on %s\n", n->addr, n->iface->name);
n->iface = NULL;
n->ifa = NULL;
n->scope = -1;
rem_node(&n->if_n);
add_tail(&sticky_neigh_list, &n->if_n);
neigh_notify(n);
}
static inline void
neigh_free(neighbor *n)
{
rem_node(&n->n);
rem_node(&n->if_n);
sl_free(neigh_slab, n);
}
/**
* neigh_update: update neighbor entry w.r.t. change on specific iface
* @n: neighbor to update
* @iface: changed iface
*
* The function recalculates state of the neighbor entry @n assuming that only
* the interface @iface may changed its state or addresses. Then, appropriate
* actions are executed (the neighbor goes up, down, up-down, or just notified).
*/
void
neigh_update(neighbor *n, struct iface *iface)
{
struct proto *p = n->proto;
struct ifa *ifa = NULL;
int scope = -1;
/* Iface-bound neighbors ignore other ifaces */
if (n->ifreq && (n->ifreq != iface))
return;
/* VRF-bound neighbors ignore changes in other VRFs */
if (p->vrf_set && (p->vrf != iface->master))
return;
scope = if_connected(n->addr, iface, &ifa, n->flags);
/* Update about already assigned iface, or some other iface */
if (iface == n->iface)
{
/* When neighbor is going down, try to respawn it on other ifaces */
if ((scope < 0) && (n->scope >= 0) && !n->ifreq && (n->flags & NEF_STICKY))
scope = if_connected_any(n->addr, p->vrf, p->vrf_set, &iface, &ifa, n->flags);
}
else
{
/* Continue only if the new variant is better than the existing one */
if (! (scope_better(scope, n->scope) ||
(scope_remote(scope, n->scope) && ifa_better(ifa, n->ifa))))
return;
}
/* No change or minor change - ignore or notify */
if ((scope == n->scope) && (iface == n->iface))
{
if (ifa != n->ifa)
{
n->ifa = ifa;
neigh_notify(n);
}
return;
}
/* Major change - going down and/or going up */
if (n->scope >= 0)
neigh_down(n);
if ((n->scope < 0) && !(n->flags & NEF_STICKY))
{
neigh_free(n);
return;
}
if (scope >= 0)
neigh_up(n, iface, ifa, scope);
}
/**
* neigh_if_up: notify neighbor cache about interface up event
* @i: interface in question
*
* Tell the neighbor cache that a new interface became up.
*
* The neighbor cache wakes up all inactive sticky neighbors with
* addresses belonging to prefixes of the interface @i.
*/
void
neigh_if_up(struct iface *i)
{
struct iface *ii;
neighbor *n;
node *x, *y;
/* Update neighbors that might be better off with the new iface */
WALK_LIST(ii, iface_list)
if (!EMPTY_LIST(ii->neighbors) && (ii != i) && if_intersect(i, ii))
WALK_LIST2_DELSAFE(n, x, y, ii->neighbors, if_n)
neigh_update(n, i);
WALK_LIST2_DELSAFE(n, x, y, sticky_neigh_list, if_n)
neigh_update(n, i);
}
/**
* neigh_if_down - notify neighbor cache about interface down event
* @i: the interface in question
*
* Notify the neighbor cache that an interface has ceased to exist.
*
* It causes all neighbors connected to this interface to be updated or removed.
*/
void
neigh_if_down(struct iface *i)
{
neighbor *n;
node *x, *y;
WALK_LIST2_DELSAFE(n, x, y, i->neighbors, if_n)
neigh_update(n, i);
}
/**
* neigh_if_link - notify neighbor cache about interface link change
* @i: the interface in question
*
* Notify the neighbor cache that an interface changed link state. All owners of
* neighbor entries connected to this interface are notified.
*/
void
neigh_if_link(struct iface *i)
{
neighbor *n;
node *x, *y;
WALK_LIST2_DELSAFE(n, x, y, i->neighbors, if_n)
neigh_notify(n);
}
/**
* neigh_ifa_update: notify neighbor cache about interface address add or remove event
* @a: interface address in question
*
* Tell the neighbor cache that an address was added or removed.
*
* The neighbor cache wakes up all inactive sticky neighbors with
* addresses belonging to prefixes of the interface belonging to @ifa
* and causes all unreachable neighbors to be flushed.
*/
void
neigh_ifa_up(struct ifa *a)
{
struct iface *i = a->iface, *ii;
neighbor *n;
node *x, *y;
/* Update neighbors that might be better off with the new ifa */
WALK_LIST(ii, iface_list)
if (!EMPTY_LIST(ii->neighbors) && ifa_intersect(a, ii))
WALK_LIST2_DELSAFE(n, x, y, ii->neighbors, if_n)
neigh_update(n, i);
/* Wake up all sticky neighbors that are reachable now */
WALK_LIST2_DELSAFE(n, x, y, sticky_neigh_list, if_n)
neigh_update(n, i);
}
void
neigh_ifa_down(struct ifa *a)
{
struct iface *i = a->iface;
neighbor *n;
node *x, *y;
/* Update all neighbors whose scope has changed */
WALK_LIST2_DELSAFE(n, x, y, i->neighbors, if_n)
if (n->ifa == a)
neigh_update(n, i);
}
static inline void
neigh_prune_one(neighbor *n)
{
if (n->proto->proto_state != PS_DOWN)
return;
neigh_free(n);
}
/**
* neigh_prune - prune neighbor cache
*
* neigh_prune() examines all neighbor entries cached and removes those
* corresponding to inactive protocols. It's called whenever a protocol
* is shut down to get rid of all its heritage.
*/
void
neigh_prune(void)
{
neighbor *n;
node *m;
int i;
DBG("Pruning neighbors\n");
for(i=0; i<NEIGH_HASH_SIZE; i++)
WALK_LIST_DELSAFE(n, m, neigh_hash_table[i])
neigh_prune_one(n);
}
/**
* neigh_init - initialize the neighbor cache.
* @if_pool: resource pool to be used for neighbor entries.
*
* This function is called during BIRD startup to initialize
* the neighbor cache module.
*/
void
neigh_init(pool *if_pool)
{
neigh_slab = sl_new(if_pool, sizeof(neighbor));
for(int i = 0; i < NEIGH_HASH_SIZE; i++)
init_list(&neigh_hash_table[i]);
init_list(&sticky_neigh_list);
}