mirror_frr/zebra/connected.c
Volodymyr Huti 7fb9825cf7 zebra: set ZEBRA_IFC_DOWN on connected routes for inactive interfaces
If you are in a situation where you have multiple addresses on an
interface, zebra creates one connected route for them.
The issue is that the rib entry is not created if addresses were
added before the interface was running.

We add the address to a running interface in a typical flow.
Therefore, we handle the route & rib creation within a single ADD event.
In the opposite case, we create the route entries without activating them.
These are considered to be active since ZEBRA_IFC_DOWN is not set.
On the following interface UP, we ignore the same ADDR_ADD as it overlaps
with the existing prefixes -> rib is never created.

The minimal reproducible setup:
-----------------------------------------
ip link add name dummy0 type dummy
ip addr flush dev dummy0
ip link set dummy0 down
ip addr add 192.168.1.7/24 dev dummy0
ip addr add 192.168.1.8/24 dev dummy0
ip link set dummy0 up
vtysh -c 'show ip route' | grep dummy0

Signed-off-by: Volodymyr Huti <v.huti@vyos.io>
2022-04-19 22:53:57 +03:00

640 lines
16 KiB
C

/*
* Address linked list routine.
* Copyright (C) 1997, 98 Kunihiro Ishiguro
*
* This file is part of GNU Zebra.
*
* GNU Zebra is free software; you can redistribute it and/or modify it
* under the terms of the GNU General Public License as published by the
* Free Software Foundation; either version 2, or (at your option) any
* later version.
*
* GNU Zebra is distributed in the hope that it will be useful, but
* WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* General Public License for more details.
*
* You should have received a copy of the GNU General Public License along
* with this program; see the file COPYING; if not, write to the Free Software
* Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
*/
#include <zebra.h>
#include "prefix.h"
#include "linklist.h"
#include "if.h"
#include "table.h"
#include "rib.h"
#include "table.h"
#include "log.h"
#include "memory.h"
#include "vty.h"
#include "zebra/debug.h"
#include "zebra/zserv.h"
#include "zebra/redistribute.h"
#include "zebra/interface.h"
#include "zebra/connected.h"
#include "zebra/rtadv.h"
#include "zebra/zebra_mpls.h"
#include "zebra/zebra_errors.h"
#include "zebra/zebra_router.h"
/* communicate the withdrawal of a connected address */
static void connected_withdraw(struct connected *ifc)
{
if (!ifc)
return;
/* Update interface address information to protocol daemon. */
if (CHECK_FLAG(ifc->conf, ZEBRA_IFC_REAL)) {
zebra_interface_address_delete_update(ifc->ifp, ifc);
if (ifc->address->family == AF_INET)
if_subnet_delete(ifc->ifp, ifc);
connected_down(ifc->ifp, ifc);
UNSET_FLAG(ifc->conf, ZEBRA_IFC_REAL);
}
/* The address is not in the kernel anymore, so clear the flag */
UNSET_FLAG(ifc->conf, ZEBRA_IFC_QUEUED);
if (!CHECK_FLAG(ifc->conf, ZEBRA_IFC_CONFIGURED)) {
listnode_delete(ifc->ifp->connected, ifc);
connected_free(&ifc);
}
}
static void connected_announce(struct interface *ifp, struct connected *ifc)
{
if (!ifc)
return;
if (!if_is_loopback(ifp) && ifc->address->family == AF_INET) {
if (ifc->address->prefixlen == IPV4_MAX_BITLEN)
SET_FLAG(ifc->flags, ZEBRA_IFA_UNNUMBERED);
else
UNSET_FLAG(ifc->flags, ZEBRA_IFA_UNNUMBERED);
}
listnode_add(ifp->connected, ifc);
/* Update interface address information to protocol daemon. */
if (ifc->address->family == AF_INET)
if_subnet_add(ifp, ifc);
zebra_interface_address_add_update(ifp, ifc);
if (if_is_operative(ifp)) {
connected_up(ifp, ifc);
}
}
/* If same interface address is already exist... */
struct connected *connected_check(struct interface *ifp,
union prefixconstptr pu)
{
const struct prefix *p = pu.p;
struct connected *ifc;
struct listnode *node;
for (ALL_LIST_ELEMENTS_RO(ifp->connected, node, ifc))
if (prefix_same(ifc->address, p))
return ifc;
return NULL;
}
/* same, but with peer address */
struct connected *connected_check_ptp(struct interface *ifp,
union prefixconstptr pu,
union prefixconstptr du)
{
const struct prefix *p = pu.p;
const struct prefix *d = du.p;
struct connected *ifc;
struct listnode *node;
for (ALL_LIST_ELEMENTS_RO(ifp->connected, node, ifc)) {
if (!prefix_same(ifc->address, p))
continue;
if (!CONNECTED_PEER(ifc) && !d)
return ifc;
if (CONNECTED_PEER(ifc) && d
&& prefix_same(ifc->destination, d))
return ifc;
}
return NULL;
}
/* Check if two ifc's describe the same address in the same state */
static int connected_same(struct connected *ifc1, struct connected *ifc2)
{
if (ifc1->ifp != ifc2->ifp)
return 0;
if (ifc1->flags != ifc2->flags)
return 0;
if (ifc1->conf != ifc2->conf)
return 0;
if (ifc1->destination)
if (!ifc2->destination)
return 0;
if (ifc2->destination)
if (!ifc1->destination)
return 0;
if (ifc1->destination && ifc2->destination)
if (!prefix_same(ifc1->destination, ifc2->destination))
return 0;
return 1;
}
/* Handle changes to addresses and send the neccesary announcements
* to clients. */
static void connected_update(struct interface *ifp, struct connected *ifc)
{
struct connected *current;
/* Check same connected route. */
current = connected_check_ptp(ifp, ifc->address, ifc->destination);
if (current) {
if (CHECK_FLAG(current->conf, ZEBRA_IFC_CONFIGURED))
SET_FLAG(ifc->conf, ZEBRA_IFC_CONFIGURED);
/* Avoid spurious withdraws, this might be just the kernel
* 'reflecting'
* back an address we have already added.
*/
if (connected_same(current, ifc)) {
/* nothing to do */
connected_free(&ifc);
return;
}
/* Clear the configured flag on the old ifc, so it will be freed
* by
* connected withdraw. */
UNSET_FLAG(current->conf, ZEBRA_IFC_CONFIGURED);
connected_withdraw(
current); /* implicit withdraw - freebsd does this */
}
/* If the connected is new or has changed, announce it, if it is usable
*/
if (CHECK_FLAG(ifc->conf, ZEBRA_IFC_REAL))
connected_announce(ifp, ifc);
}
/* Called from if_up(). */
void connected_up(struct interface *ifp, struct connected *ifc)
{
afi_t afi;
struct prefix p;
struct nexthop nh = {
.type = NEXTHOP_TYPE_IFINDEX,
.ifindex = ifp->ifindex,
.vrf_id = ifp->vrf->vrf_id,
};
struct zebra_vrf *zvrf;
uint32_t metric;
uint32_t flags = 0;
uint32_t count = 0;
struct listnode *cnode;
struct connected *c;
zvrf = ifp->vrf->info;
if (!zvrf) {
flog_err(
EC_ZEBRA_VRF_NOT_FOUND,
"%s: Received Up for interface but no associated zvrf: %s(%d)",
__func__, ifp->vrf->name, ifp->vrf->vrf_id);
return;
}
if (!CHECK_FLAG(ifc->conf, ZEBRA_IFC_REAL))
return;
/* Ensure 'down' flag is cleared */
UNSET_FLAG(ifc->conf, ZEBRA_IFC_DOWN);
prefix_copy(&p, CONNECTED_PREFIX(ifc));
/* Apply mask to the network. */
apply_mask(&p);
afi = family2afi(p.family);
switch (afi) {
case AFI_IP:
/*
* In case of connected address is 0.0.0.0/0 we treat it tunnel
* address.
*/
if (prefix_ipv4_any((struct prefix_ipv4 *)&p))
return;
break;
case AFI_IP6:
#ifndef GNU_LINUX
/* XXX: It is already done by rib_bogus_ipv6 within rib_add */
if (IN6_IS_ADDR_UNSPECIFIED(&p.u.prefix6))
return;
#endif
break;
default:
flog_warn(EC_ZEBRA_CONNECTED_AFI_UNKNOWN,
"Received unknown AFI: %s", afi2str(afi));
return;
break;
}
metric = (ifc->metric < (uint32_t)METRIC_MAX) ?
ifc->metric : ifp->metric;
/*
* Since we are hand creating the connected routes
* in our main routing table, *if* we are working
* in an offloaded environment then we need to
* pretend like the route is offloaded so everything
* else will work
*/
if (zrouter.asic_offloaded)
flags |= ZEBRA_FLAG_OFFLOADED;
/*
* It's possible to add the same network and mask
* to an interface over and over. This would
* result in an equivalent number of connected
* routes. Just add one connected route in
* for all the addresses on an interface that
* resolve to the same network and mask
*/
for (ALL_LIST_ELEMENTS_RO(ifp->connected, cnode, c)) {
struct prefix cp;
prefix_copy(&cp, CONNECTED_PREFIX(c));
apply_mask(&cp);
if (prefix_same(&cp, &p) &&
!CHECK_FLAG(c->conf, ZEBRA_IFC_DOWN))
count++;
if (count >= 2)
return;
}
rib_add(afi, SAFI_UNICAST, zvrf->vrf->vrf_id, ZEBRA_ROUTE_CONNECT, 0,
flags, &p, NULL, &nh, 0, zvrf->table_id, metric, 0, 0, 0,
false);
rib_add(afi, SAFI_MULTICAST, zvrf->vrf->vrf_id, ZEBRA_ROUTE_CONNECT, 0,
flags, &p, NULL, &nh, 0, zvrf->table_id, metric, 0, 0, 0,
false);
/* Schedule LSP forwarding entries for processing, if appropriate. */
if (zvrf->vrf->vrf_id == VRF_DEFAULT) {
if (IS_ZEBRA_DEBUG_MPLS)
zlog_debug(
"%u: IF %s IP %pFX address add/up, scheduling MPLS processing",
zvrf->vrf->vrf_id, ifp->name, &p);
mpls_mark_lsps_for_processing(zvrf, &p);
}
}
/* Add connected IPv4 route to the interface. */
void connected_add_ipv4(struct interface *ifp, int flags,
const struct in_addr *addr, uint16_t prefixlen,
const struct in_addr *dest, const char *label,
uint32_t metric)
{
struct prefix_ipv4 *p;
struct connected *ifc;
if (ipv4_martian(addr))
return;
/* Make connected structure. */
ifc = connected_new();
ifc->ifp = ifp;
ifc->flags = flags;
ifc->metric = metric;
/* If we get a notification from the kernel,
* we can safely assume the address is known to the kernel */
SET_FLAG(ifc->conf, ZEBRA_IFC_QUEUED);
if (!if_is_operative(ifp))
SET_FLAG(ifc->conf, ZEBRA_IFC_DOWN);
/* Allocate new connected address. */
p = prefix_ipv4_new();
p->family = AF_INET;
p->prefix = *addr;
p->prefixlen =
CHECK_FLAG(flags, ZEBRA_IFA_PEER) ? IPV4_MAX_BITLEN : prefixlen;
ifc->address = (struct prefix *)p;
/* If there is a peer address. */
if (CONNECTED_PEER(ifc)) {
/* validate the destination address */
if (dest) {
p = prefix_ipv4_new();
p->family = AF_INET;
p->prefix = *dest;
p->prefixlen = prefixlen;
ifc->destination = (struct prefix *)p;
if (IPV4_ADDR_SAME(addr, dest))
flog_warn(
EC_ZEBRA_IFACE_SAME_LOCAL_AS_PEER,
"interface %s has same local and peer address %pI4, routing protocols may malfunction",
ifp->name, addr);
} else {
zlog_debug(
"%s called for interface %s with peer flag set, but no peer address supplied",
__func__, ifp->name);
UNSET_FLAG(ifc->flags, ZEBRA_IFA_PEER);
}
}
/* no destination address was supplied */
if (!dest && (prefixlen == IPV4_MAX_BITLEN) && if_is_pointopoint(ifp))
zlog_debug(
"PtP interface %s with addr %pI4/%d needs a peer address",
ifp->name, addr, prefixlen);
/* Label of this address. */
if (label)
ifc->label = XSTRDUP(MTYPE_CONNECTED_LABEL, label);
/* For all that I know an IPv4 address is always ready when we receive
* the notification. So it should be safe to set the REAL flag here. */
SET_FLAG(ifc->conf, ZEBRA_IFC_REAL);
connected_update(ifp, ifc);
}
void connected_down(struct interface *ifp, struct connected *ifc)
{
afi_t afi;
struct prefix p;
struct nexthop nh = {
.type = NEXTHOP_TYPE_IFINDEX,
.ifindex = ifp->ifindex,
.vrf_id = ifp->vrf->vrf_id,
};
struct zebra_vrf *zvrf;
uint32_t count = 0;
struct listnode *cnode;
struct connected *c;
zvrf = ifp->vrf->info;
if (!zvrf) {
flog_err(
EC_ZEBRA_VRF_NOT_FOUND,
"%s: Received Down for interface but no associated zvrf: %s(%d)",
__func__, ifp->vrf->name, ifp->vrf->vrf_id);
return;
}
if (!CHECK_FLAG(ifc->conf, ZEBRA_IFC_REAL))
return;
/* Skip if we've already done this; this can happen if we have a
* config change that takes an interface down, then we receive kernel
* notifications about the downed interface and its addresses.
*/
if (CHECK_FLAG(ifc->conf, ZEBRA_IFC_DOWN)) {
if (IS_ZEBRA_DEBUG_RIB)
zlog_debug("%s: ifc %p, %pFX already DOWN",
__func__, ifc, ifc->address);
return;
}
prefix_copy(&p, CONNECTED_PREFIX(ifc));
/* Apply mask to the network. */
apply_mask(&p);
afi = family2afi(p.family);
switch (afi) {
case AFI_IP:
/*
* In case of connected address is 0.0.0.0/0 we treat it tunnel
* address.
*/
if (prefix_ipv4_any((struct prefix_ipv4 *)&p))
return;
break;
case AFI_IP6:
if (IN6_IS_ADDR_UNSPECIFIED(&p.u.prefix6))
return;
break;
default:
zlog_warn("Unknown AFI: %s", afi2str(afi));
break;
}
/* Mark the address as 'down' */
SET_FLAG(ifc->conf, ZEBRA_IFC_DOWN);
/*
* It's possible to have X number of addresses
* on a interface that all resolve to the same
* network and mask. Find them and just
* allow the deletion when are removing the last
* one.
*/
for (ALL_LIST_ELEMENTS_RO(ifp->connected, cnode, c)) {
struct prefix cp;
prefix_copy(&cp, CONNECTED_PREFIX(c));
apply_mask(&cp);
if (prefix_same(&p, &cp) &&
!CHECK_FLAG(c->conf, ZEBRA_IFC_DOWN))
count++;
if (count >= 1)
return;
}
/*
* Same logic as for connected_up(): push the changes into the
* head.
*/
rib_delete(afi, SAFI_UNICAST, zvrf->vrf->vrf_id, ZEBRA_ROUTE_CONNECT, 0,
0, &p, NULL, &nh, 0, zvrf->table_id, 0, 0, false);
rib_delete(afi, SAFI_MULTICAST, zvrf->vrf->vrf_id, ZEBRA_ROUTE_CONNECT,
0, 0, &p, NULL, &nh, 0, zvrf->table_id, 0, 0, false);
/* Schedule LSP forwarding entries for processing, if appropriate. */
if (zvrf->vrf->vrf_id == VRF_DEFAULT) {
if (IS_ZEBRA_DEBUG_MPLS)
zlog_debug(
"%u: IF %s IP %pFX address down, scheduling MPLS processing",
zvrf->vrf->vrf_id, ifp->name, &p);
mpls_mark_lsps_for_processing(zvrf, &p);
}
}
static void connected_delete_helper(struct connected *ifc, struct prefix *p)
{
struct interface *ifp;
if (!ifc)
return;
ifp = ifc->ifp;
connected_withdraw(ifc);
/* Schedule LSP forwarding entries for processing, if appropriate. */
if (ifp->vrf->vrf_id == VRF_DEFAULT) {
if (IS_ZEBRA_DEBUG_MPLS)
zlog_debug(
"%u: IF %s IP %pFX address delete, scheduling MPLS processing",
ifp->vrf->vrf_id, ifp->name, p);
mpls_mark_lsps_for_processing(ifp->vrf->info, p);
}
}
/* Delete connected IPv4 route to the interface. */
void connected_delete_ipv4(struct interface *ifp, int flags,
const struct in_addr *addr, uint16_t prefixlen,
const struct in_addr *dest)
{
struct prefix p, d;
struct connected *ifc;
memset(&p, 0, sizeof(struct prefix));
p.family = AF_INET;
p.u.prefix4 = *addr;
p.prefixlen =
CHECK_FLAG(flags, ZEBRA_IFA_PEER) ? IPV4_MAX_BITLEN : prefixlen;
if (dest) {
memset(&d, 0, sizeof(struct prefix));
d.family = AF_INET;
d.u.prefix4 = *dest;
d.prefixlen = prefixlen;
ifc = connected_check_ptp(ifp, &p, &d);
} else
ifc = connected_check_ptp(ifp, &p, NULL);
connected_delete_helper(ifc, &p);
}
/* Add connected IPv6 route to the interface. */
void connected_add_ipv6(struct interface *ifp, int flags,
const struct in6_addr *addr,
const struct in6_addr *dest, uint16_t prefixlen,
const char *label, uint32_t metric)
{
struct prefix_ipv6 *p;
struct connected *ifc;
if (ipv6_martian(addr))
return;
/* Make connected structure. */
ifc = connected_new();
ifc->ifp = ifp;
ifc->flags = flags;
ifc->metric = metric;
/* If we get a notification from the kernel,
* we can safely assume the address is known to the kernel */
SET_FLAG(ifc->conf, ZEBRA_IFC_QUEUED);
if (!if_is_operative(ifp))
SET_FLAG(ifc->conf, ZEBRA_IFC_DOWN);
/* Allocate new connected address. */
p = prefix_ipv6_new();
p->family = AF_INET6;
IPV6_ADDR_COPY(&p->prefix, addr);
p->prefixlen = prefixlen;
ifc->address = (struct prefix *)p;
/* Add global ipv6 address to the RA prefix list */
if (!IN6_IS_ADDR_LINKLOCAL(&p->prefix))
rtadv_add_prefix(ifp->info, p);
if (dest) {
p = prefix_ipv6_new();
p->family = AF_INET6;
IPV6_ADDR_COPY(&p->prefix, dest);
p->prefixlen = prefixlen;
ifc->destination = (struct prefix *)p;
} else {
if (CHECK_FLAG(ifc->flags, ZEBRA_IFA_PEER)) {
zlog_debug(
"%s called for interface %s with peer flag set, but no peer address supplied",
__func__, ifp->name);
UNSET_FLAG(ifc->flags, ZEBRA_IFA_PEER);
}
}
/* Label of this address. */
if (label)
ifc->label = XSTRDUP(MTYPE_CONNECTED_LABEL, label);
/* On Linux, we only get here when DAD is complete, therefore we can set
* ZEBRA_IFC_REAL.
*
* On BSD, there currently doesn't seem to be a way to check for
* completion of
* DAD, so we replicate the old behaviour and set ZEBRA_IFC_REAL,
* although DAD
* might still be running.
*/
SET_FLAG(ifc->conf, ZEBRA_IFC_REAL);
connected_update(ifp, ifc);
}
void connected_delete_ipv6(struct interface *ifp,
const struct in6_addr *address,
const struct in6_addr *dest, uint16_t prefixlen)
{
struct prefix p, d;
struct connected *ifc;
memset(&p, 0, sizeof(struct prefix));
p.family = AF_INET6;
memcpy(&p.u.prefix6, address, sizeof(struct in6_addr));
p.prefixlen = prefixlen;
/* Delete global ipv6 address from RA prefix list */
if (!IN6_IS_ADDR_LINKLOCAL(&p.u.prefix6))
rtadv_delete_prefix(ifp->info, &p);
if (dest) {
memset(&d, 0, sizeof(struct prefix));
d.family = AF_INET6;
IPV6_ADDR_COPY(&d.u.prefix6, dest);
d.prefixlen = prefixlen;
ifc = connected_check_ptp(ifp, &p, &d);
} else
ifc = connected_check_ptp(ifp, &p, NULL);
connected_delete_helper(ifc, &p);
}
int connected_is_unnumbered(struct interface *ifp)
{
struct connected *connected;
struct listnode *node;
for (ALL_LIST_ELEMENTS_RO(ifp->connected, node, connected)) {
if (CHECK_FLAG(connected->conf, ZEBRA_IFC_REAL)
&& connected->address->family == AF_INET)
return CHECK_FLAG(connected->flags,
ZEBRA_IFA_UNNUMBERED);
}
return 0;
}