freebsd-dev/sys/netinet6/nd6_rtr.c

2409 lines
64 KiB
C
Raw Normal View History

/*-
* SPDX-License-Identifier: BSD-3-Clause
*
* Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. Neither the name of the project nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
2007-12-10 16:03:40 +00:00
*
* $KAME: nd6_rtr.c,v 1.111 2001/04/27 01:37:15 jinmei Exp $
*/
2007-12-10 16:03:40 +00:00
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include "opt_inet.h"
#include "opt_inet6.h"
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/malloc.h>
#include <sys/mbuf.h>
#include <sys/refcount.h>
#include <sys/socket.h>
#include <sys/sockio.h>
#include <sys/time.h>
#include <sys/kernel.h>
#include <sys/lock.h>
#include <sys/errno.h>
#include <sys/rmlock.h>
#include <sys/rwlock.h>
#include <sys/syslog.h>
#include <sys/queue.h>
#include <net/if.h>
#include <net/if_var.h>
#include <net/if_types.h>
#include <net/if_dl.h>
#include <net/route.h>
#include <net/route_var.h>
#include <net/radix.h>
#include <net/vnet.h>
#include <netinet/in.h>
#include <net/if_llatbl.h>
#include <netinet6/in6_var.h>
#include <netinet6/in6_ifattach.h>
#include <netinet/ip6.h>
#include <netinet6/ip6_var.h>
#include <netinet6/nd6.h>
#include <netinet/icmp6.h>
#include <netinet6/scope6_var.h>
2008-01-08 19:08:58 +00:00
static int rtpref(struct nd_defrouter *);
static struct nd_defrouter *defrtrlist_update(struct nd_defrouter *);
static int prelist_update(struct nd_prefixctl *, struct nd_defrouter *,
struct mbuf *, int);
static struct in6_ifaddr *in6_ifadd(struct nd_prefixctl *, int);
static struct nd_pfxrouter *pfxrtr_lookup(struct nd_prefix *,
struct nd_defrouter *);
2008-01-08 19:08:58 +00:00
static void pfxrtr_add(struct nd_prefix *, struct nd_defrouter *);
static void pfxrtr_del(struct nd_pfxrouter *);
static struct nd_pfxrouter *find_pfxlist_reachable_router(struct nd_prefix *);
2008-01-08 19:08:58 +00:00
static void defrouter_delreq(struct nd_defrouter *);
static void nd6_rtmsg(int, struct rtentry *);
2008-01-08 19:08:58 +00:00
static int in6_init_prefix_ltimes(struct nd_prefix *);
static void in6_init_address_ltimes(struct nd_prefix *,
struct in6_addrlifetime *);
static int rt6_deleteroute(const struct rtentry *, void *);
Build on Jeff Roberson's linker-set based dynamic per-CPU allocator (DPCPU), as suggested by Peter Wemm, and implement a new per-virtual network stack memory allocator. Modify vnet to use the allocator instead of monolithic global container structures (vinet, ...). This change solves many binary compatibility problems associated with VIMAGE, and restores ELF symbols for virtualized global variables. Each virtualized global variable exists as a "reference copy", and also once per virtual network stack. Virtualized global variables are tagged at compile-time, placing the in a special linker set, which is loaded into a contiguous region of kernel memory. Virtualized global variables in the base kernel are linked as normal, but those in modules are copied and relocated to a reserved portion of the kernel's vnet region with the help of a the kernel linker. Virtualized global variables exist in per-vnet memory set up when the network stack instance is created, and are initialized statically from the reference copy. Run-time access occurs via an accessor macro, which converts from the current vnet and requested symbol to a per-vnet address. When "options VIMAGE" is not compiled into the kernel, normal global ELF symbols will be used instead and indirection is avoided. This change restores static initialization for network stack global variables, restores support for non-global symbols and types, eliminates the need for many subsystem constructors, eliminates large per-subsystem structures that caused many binary compatibility issues both for monitoring applications (netstat) and kernel modules, removes the per-function INIT_VNET_*() macros throughout the stack, eliminates the need for vnet_symmap ksym(2) munging, and eliminates duplicate definitions of virtualized globals under VIMAGE_GLOBALS. Bump __FreeBSD_version and update UPDATING. Portions submitted by: bz Reviewed by: bz, zec Discussed with: gnn, jamie, jeff, jhb, julian, sam Suggested by: peter Approved by: re (kensmith)
2009-07-14 22:48:30 +00:00
VNET_DECLARE(int, nd6_recalc_reachtm_interval);
#define V_nd6_recalc_reachtm_interval VNET(nd6_recalc_reachtm_interval)
static VNET_DEFINE(struct ifnet *, nd6_defifp);
VNET_DEFINE(int, nd6_defifindex);
#define V_nd6_defifp VNET(nd6_defifp)
VNET_DEFINE(int, ip6_use_tempaddr) = 0;
Build on Jeff Roberson's linker-set based dynamic per-CPU allocator (DPCPU), as suggested by Peter Wemm, and implement a new per-virtual network stack memory allocator. Modify vnet to use the allocator instead of monolithic global container structures (vinet, ...). This change solves many binary compatibility problems associated with VIMAGE, and restores ELF symbols for virtualized global variables. Each virtualized global variable exists as a "reference copy", and also once per virtual network stack. Virtualized global variables are tagged at compile-time, placing the in a special linker set, which is loaded into a contiguous region of kernel memory. Virtualized global variables in the base kernel are linked as normal, but those in modules are copied and relocated to a reserved portion of the kernel's vnet region with the help of a the kernel linker. Virtualized global variables exist in per-vnet memory set up when the network stack instance is created, and are initialized statically from the reference copy. Run-time access occurs via an accessor macro, which converts from the current vnet and requested symbol to a per-vnet address. When "options VIMAGE" is not compiled into the kernel, normal global ELF symbols will be used instead and indirection is avoided. This change restores static initialization for network stack global variables, restores support for non-global symbols and types, eliminates the need for many subsystem constructors, eliminates large per-subsystem structures that caused many binary compatibility issues both for monitoring applications (netstat) and kernel modules, removes the per-function INIT_VNET_*() macros throughout the stack, eliminates the need for vnet_symmap ksym(2) munging, and eliminates duplicate definitions of virtualized globals under VIMAGE_GLOBALS. Bump __FreeBSD_version and update UPDATING. Portions submitted by: bz Reviewed by: bz, zec Discussed with: gnn, jamie, jeff, jhb, julian, sam Suggested by: peter Approved by: re (kensmith)
2009-07-14 22:48:30 +00:00
VNET_DEFINE(int, ip6_desync_factor);
VNET_DEFINE(u_int32_t, ip6_temp_preferred_lifetime) = DEF_TEMP_PREFERRED_LIFETIME;
VNET_DEFINE(u_int32_t, ip6_temp_valid_lifetime) = DEF_TEMP_VALID_LIFETIME;
VNET_DEFINE(int, ip6_temp_regen_advance) = TEMPADDR_REGEN_ADVANCE;
/* RTPREF_MEDIUM has to be 0! */
#define RTPREF_HIGH 1
#define RTPREF_MEDIUM 0
#define RTPREF_LOW (-1)
#define RTPREF_RESERVED (-2)
#define RTPREF_INVALID (-3) /* internal */
/*
* Receive Router Solicitation Message - just for routers.
* Router solicitation/advertisement is mostly managed by userland program
* (rtadvd) so here we have no function like nd6_ra_output().
*
* Based on RFC 2461
*/
void
nd6_rs_input(struct mbuf *m, int off, int icmp6len)
{
struct ifnet *ifp = m->m_pkthdr.rcvif;
struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
struct nd_router_solicit *nd_rs;
struct in6_addr saddr6 = ip6->ip6_src;
char *lladdr = NULL;
int lladdrlen = 0;
union nd_opts ndopts;
char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
- Accept Router Advertisement messages even when net.inet6.ip6.forwarding=1. - A new per-interface knob IFF_ND6_NO_RADR and sysctl IPV6CTL_NO_RADR. This controls if accepting a route in an RA message as the default route. The default value for each interface can be set by net.inet6.ip6.no_radr. The system wide default value is 0. - A new sysctl: net.inet6.ip6.norbit_raif. This controls if setting R-bit in NA on RA accepting interfaces. The default is 0 (R-bit is set based on net.inet6.ip6.forwarding). Background: IPv6 host/router model suggests a router sends an RA and a host accepts it for router discovery. Because of that, KAME implementation does not allow accepting RAs when net.inet6.ip6.forwarding=1. Accepting RAs on a router can make the routing table confused since it can change the default router unintentionally. However, in practice there are cases where we cannot distinguish a host from a router clearly. For example, a customer edge router often works as a host against the ISP, and as a router against the LAN at the same time. Another example is a complex network configurations like an L2TP tunnel for IPv6 connection to Internet over an Ethernet link with another native IPv6 subnet. In this case, the physical interface for the native IPv6 subnet works as a host, and the pseudo-interface for L2TP works as the default IP forwarding route. Problem: Disabling processing RA messages when net.inet6.ip6.forwarding=1 and accepting them when net.inet6.ip6.forward=0 cause the following practical issues: - A router cannot perform SLAAC. It becomes a problem if a box has multiple interfaces and you want to use SLAAC on some of them, for example. A customer edge router for IPv6 Internet access service using an IPv6-over-IPv6 tunnel sometimes needs SLAAC on the physical interface for administration purpose; updating firmware and so on (link-local addresses can be used there, but GUAs by SLAAC are often used for scalability). - When a host has multiple IPv6 interfaces and it receives multiple RAs on them, controlling the default route is difficult. Router preferences defined in RFC 4191 works only when the routers on the links are under your control. Details of Implementation Changes: Router Advertisement messages will be accepted even when net.inet6.ip6.forwarding=1. More precisely, the conditions are as follow: (ACCEPT_RTADV && !NO_RADR && !ip6.forwarding) => Normal RA processing on that interface. (as IPv6 host) (ACCEPT_RTADV && (NO_RADR || ip6.forwarding)) => Accept RA but add the router to the defroute list with rtlifetime=0 unconditionally. This effectively prevents from setting the received router address as the box's default route. (!ACCEPT_RTADV) => No RA processing on that interface. ACCEPT_RTADV and NO_RADR are per-interface knob. In short, all interface are classified as "RA-accepting" or not. An RA-accepting interface always processes RA messages regardless of ip6.forwarding. The difference caused by NO_RADR or ip6.forwarding is whether the RA source address is considered as the default router or not. R-bit in NA on the RA accepting interfaces is set based on net.inet6.ip6.forwarding. While RFC 6204 W-1 rule (for CPE case) suggests a router should disable the R-bit completely even when the box has net.inet6.ip6.forwarding=1, I believe there is no technical reason with doing so. This behavior can be set by a new sysctl net.inet6.ip6.norbit_raif (the default is 0). Usage: # ifconfig fxp0 inet6 accept_rtadv => accept RA on fxp0 # ifconfig fxp0 inet6 accept_rtadv no_radr => accept RA on fxp0 but ignore default route information in it. # sysctl net.inet6.ip6.norbit_no_radr=1 => R-bit in NAs on RA accepting interfaces will always be set to 0.
2011-06-06 02:14:23 +00:00
/*
* Accept RS only when V_ip6_forwarding=1 and the interface has
* no ND6_IFF_ACCEPT_RTADV.
*/
if (!V_ip6_forwarding || ND_IFINFO(ifp)->flags & ND6_IFF_ACCEPT_RTADV)
goto freeit;
/* RFC 6980: Nodes MUST silently ignore fragments */
if(m->m_flags & M_FRAGMENTED)
goto freeit;
/* Sanity checks */
if (ip6->ip6_hlim != 255) {
nd6log((LOG_ERR,
"nd6_rs_input: invalid hlim (%d) from %s to %s on %s\n",
ip6->ip6_hlim, ip6_sprintf(ip6bufs, &ip6->ip6_src),
ip6_sprintf(ip6bufd, &ip6->ip6_dst), if_name(ifp)));
goto bad;
}
/*
* Don't update the neighbor cache, if src = ::.
* This indicates that the src has no IP address assigned yet.
*/
if (IN6_IS_ADDR_UNSPECIFIED(&saddr6))
goto freeit;
#ifndef PULLDOWN_TEST
IP6_EXTHDR_CHECK(m, off, icmp6len,);
nd_rs = (struct nd_router_solicit *)((caddr_t)ip6 + off);
#else
IP6_EXTHDR_GET(nd_rs, struct nd_router_solicit *, m, off, icmp6len);
if (nd_rs == NULL) {
ICMP6STAT_INC(icp6s_tooshort);
return;
}
#endif
icmp6len -= sizeof(*nd_rs);
nd6_option_init(nd_rs + 1, icmp6len, &ndopts);
if (nd6_options(&ndopts) < 0) {
nd6log((LOG_INFO,
"nd6_rs_input: invalid ND option, ignored\n"));
/* nd6_options have incremented stats */
goto freeit;
}
if (ndopts.nd_opts_src_lladdr) {
lladdr = (char *)(ndopts.nd_opts_src_lladdr + 1);
lladdrlen = ndopts.nd_opts_src_lladdr->nd_opt_len << 3;
}
if (lladdr && ((ifp->if_addrlen + 2 + 7) & ~7) != lladdrlen) {
nd6log((LOG_INFO,
"nd6_rs_input: lladdrlen mismatch for %s "
"(if %d, RS packet %d)\n",
ip6_sprintf(ip6bufs, &saddr6),
ifp->if_addrlen, lladdrlen - 2));
goto bad;
}
nd6_cache_lladdr(ifp, &saddr6, lladdr, lladdrlen, ND_ROUTER_SOLICIT, 0);
freeit:
m_freem(m);
return;
bad:
ICMP6STAT_INC(icp6s_badrs);
m_freem(m);
}
/*
* Receive Router Advertisement Message.
*
* Based on RFC 2461
* TODO: on-link bit on prefix information
* TODO: ND_RA_FLAG_{OTHER,MANAGED} processing
*/
void
nd6_ra_input(struct mbuf *m, int off, int icmp6len)
{
struct ifnet *ifp = m->m_pkthdr.rcvif;
struct nd_ifinfo *ndi = ND_IFINFO(ifp);
struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
struct nd_router_advert *nd_ra;
struct in6_addr saddr6 = ip6->ip6_src;
int mcast = 0;
union nd_opts ndopts;
struct nd_defrouter *dr;
char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
dr = NULL;
/*
- Accept Router Advertisement messages even when net.inet6.ip6.forwarding=1. - A new per-interface knob IFF_ND6_NO_RADR and sysctl IPV6CTL_NO_RADR. This controls if accepting a route in an RA message as the default route. The default value for each interface can be set by net.inet6.ip6.no_radr. The system wide default value is 0. - A new sysctl: net.inet6.ip6.norbit_raif. This controls if setting R-bit in NA on RA accepting interfaces. The default is 0 (R-bit is set based on net.inet6.ip6.forwarding). Background: IPv6 host/router model suggests a router sends an RA and a host accepts it for router discovery. Because of that, KAME implementation does not allow accepting RAs when net.inet6.ip6.forwarding=1. Accepting RAs on a router can make the routing table confused since it can change the default router unintentionally. However, in practice there are cases where we cannot distinguish a host from a router clearly. For example, a customer edge router often works as a host against the ISP, and as a router against the LAN at the same time. Another example is a complex network configurations like an L2TP tunnel for IPv6 connection to Internet over an Ethernet link with another native IPv6 subnet. In this case, the physical interface for the native IPv6 subnet works as a host, and the pseudo-interface for L2TP works as the default IP forwarding route. Problem: Disabling processing RA messages when net.inet6.ip6.forwarding=1 and accepting them when net.inet6.ip6.forward=0 cause the following practical issues: - A router cannot perform SLAAC. It becomes a problem if a box has multiple interfaces and you want to use SLAAC on some of them, for example. A customer edge router for IPv6 Internet access service using an IPv6-over-IPv6 tunnel sometimes needs SLAAC on the physical interface for administration purpose; updating firmware and so on (link-local addresses can be used there, but GUAs by SLAAC are often used for scalability). - When a host has multiple IPv6 interfaces and it receives multiple RAs on them, controlling the default route is difficult. Router preferences defined in RFC 4191 works only when the routers on the links are under your control. Details of Implementation Changes: Router Advertisement messages will be accepted even when net.inet6.ip6.forwarding=1. More precisely, the conditions are as follow: (ACCEPT_RTADV && !NO_RADR && !ip6.forwarding) => Normal RA processing on that interface. (as IPv6 host) (ACCEPT_RTADV && (NO_RADR || ip6.forwarding)) => Accept RA but add the router to the defroute list with rtlifetime=0 unconditionally. This effectively prevents from setting the received router address as the box's default route. (!ACCEPT_RTADV) => No RA processing on that interface. ACCEPT_RTADV and NO_RADR are per-interface knob. In short, all interface are classified as "RA-accepting" or not. An RA-accepting interface always processes RA messages regardless of ip6.forwarding. The difference caused by NO_RADR or ip6.forwarding is whether the RA source address is considered as the default router or not. R-bit in NA on the RA accepting interfaces is set based on net.inet6.ip6.forwarding. While RFC 6204 W-1 rule (for CPE case) suggests a router should disable the R-bit completely even when the box has net.inet6.ip6.forwarding=1, I believe there is no technical reason with doing so. This behavior can be set by a new sysctl net.inet6.ip6.norbit_raif (the default is 0). Usage: # ifconfig fxp0 inet6 accept_rtadv => accept RA on fxp0 # ifconfig fxp0 inet6 accept_rtadv no_radr => accept RA on fxp0 but ignore default route information in it. # sysctl net.inet6.ip6.norbit_no_radr=1 => R-bit in NAs on RA accepting interfaces will always be set to 0.
2011-06-06 02:14:23 +00:00
* We only accept RAs only when the per-interface flag
* ND6_IFF_ACCEPT_RTADV is on the receiving interface.
*/
- Accept Router Advertisement messages even when net.inet6.ip6.forwarding=1. - A new per-interface knob IFF_ND6_NO_RADR and sysctl IPV6CTL_NO_RADR. This controls if accepting a route in an RA message as the default route. The default value for each interface can be set by net.inet6.ip6.no_radr. The system wide default value is 0. - A new sysctl: net.inet6.ip6.norbit_raif. This controls if setting R-bit in NA on RA accepting interfaces. The default is 0 (R-bit is set based on net.inet6.ip6.forwarding). Background: IPv6 host/router model suggests a router sends an RA and a host accepts it for router discovery. Because of that, KAME implementation does not allow accepting RAs when net.inet6.ip6.forwarding=1. Accepting RAs on a router can make the routing table confused since it can change the default router unintentionally. However, in practice there are cases where we cannot distinguish a host from a router clearly. For example, a customer edge router often works as a host against the ISP, and as a router against the LAN at the same time. Another example is a complex network configurations like an L2TP tunnel for IPv6 connection to Internet over an Ethernet link with another native IPv6 subnet. In this case, the physical interface for the native IPv6 subnet works as a host, and the pseudo-interface for L2TP works as the default IP forwarding route. Problem: Disabling processing RA messages when net.inet6.ip6.forwarding=1 and accepting them when net.inet6.ip6.forward=0 cause the following practical issues: - A router cannot perform SLAAC. It becomes a problem if a box has multiple interfaces and you want to use SLAAC on some of them, for example. A customer edge router for IPv6 Internet access service using an IPv6-over-IPv6 tunnel sometimes needs SLAAC on the physical interface for administration purpose; updating firmware and so on (link-local addresses can be used there, but GUAs by SLAAC are often used for scalability). - When a host has multiple IPv6 interfaces and it receives multiple RAs on them, controlling the default route is difficult. Router preferences defined in RFC 4191 works only when the routers on the links are under your control. Details of Implementation Changes: Router Advertisement messages will be accepted even when net.inet6.ip6.forwarding=1. More precisely, the conditions are as follow: (ACCEPT_RTADV && !NO_RADR && !ip6.forwarding) => Normal RA processing on that interface. (as IPv6 host) (ACCEPT_RTADV && (NO_RADR || ip6.forwarding)) => Accept RA but add the router to the defroute list with rtlifetime=0 unconditionally. This effectively prevents from setting the received router address as the box's default route. (!ACCEPT_RTADV) => No RA processing on that interface. ACCEPT_RTADV and NO_RADR are per-interface knob. In short, all interface are classified as "RA-accepting" or not. An RA-accepting interface always processes RA messages regardless of ip6.forwarding. The difference caused by NO_RADR or ip6.forwarding is whether the RA source address is considered as the default router or not. R-bit in NA on the RA accepting interfaces is set based on net.inet6.ip6.forwarding. While RFC 6204 W-1 rule (for CPE case) suggests a router should disable the R-bit completely even when the box has net.inet6.ip6.forwarding=1, I believe there is no technical reason with doing so. This behavior can be set by a new sysctl net.inet6.ip6.norbit_raif (the default is 0). Usage: # ifconfig fxp0 inet6 accept_rtadv => accept RA on fxp0 # ifconfig fxp0 inet6 accept_rtadv no_radr => accept RA on fxp0 but ignore default route information in it. # sysctl net.inet6.ip6.norbit_no_radr=1 => R-bit in NAs on RA accepting interfaces will always be set to 0.
2011-06-06 02:14:23 +00:00
if (!(ndi->flags & ND6_IFF_ACCEPT_RTADV))
goto freeit;
/* RFC 6980: Nodes MUST silently ignore fragments */
if(m->m_flags & M_FRAGMENTED)
goto freeit;
if (ip6->ip6_hlim != 255) {
nd6log((LOG_ERR,
"nd6_ra_input: invalid hlim (%d) from %s to %s on %s\n",
ip6->ip6_hlim, ip6_sprintf(ip6bufs, &ip6->ip6_src),
ip6_sprintf(ip6bufd, &ip6->ip6_dst), if_name(ifp)));
goto bad;
}
if (!IN6_IS_ADDR_LINKLOCAL(&saddr6)) {
nd6log((LOG_ERR,
"nd6_ra_input: src %s is not link-local\n",
ip6_sprintf(ip6bufs, &saddr6)));
goto bad;
}
#ifndef PULLDOWN_TEST
IP6_EXTHDR_CHECK(m, off, icmp6len,);
nd_ra = (struct nd_router_advert *)((caddr_t)ip6 + off);
#else
IP6_EXTHDR_GET(nd_ra, struct nd_router_advert *, m, off, icmp6len);
if (nd_ra == NULL) {
ICMP6STAT_INC(icp6s_tooshort);
return;
}
#endif
icmp6len -= sizeof(*nd_ra);
nd6_option_init(nd_ra + 1, icmp6len, &ndopts);
if (nd6_options(&ndopts) < 0) {
nd6log((LOG_INFO,
"nd6_ra_input: invalid ND option, ignored\n"));
/* nd6_options have incremented stats */
goto freeit;
}
{
struct nd_defrouter dr0;
u_int32_t advreachable = nd_ra->nd_ra_reachable;
/* remember if this is a multicasted advertisement */
if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst))
mcast = 1;
bzero(&dr0, sizeof(dr0));
dr0.rtaddr = saddr6;
dr0.raflags = nd_ra->nd_ra_flags_reserved;
- Accept Router Advertisement messages even when net.inet6.ip6.forwarding=1. - A new per-interface knob IFF_ND6_NO_RADR and sysctl IPV6CTL_NO_RADR. This controls if accepting a route in an RA message as the default route. The default value for each interface can be set by net.inet6.ip6.no_radr. The system wide default value is 0. - A new sysctl: net.inet6.ip6.norbit_raif. This controls if setting R-bit in NA on RA accepting interfaces. The default is 0 (R-bit is set based on net.inet6.ip6.forwarding). Background: IPv6 host/router model suggests a router sends an RA and a host accepts it for router discovery. Because of that, KAME implementation does not allow accepting RAs when net.inet6.ip6.forwarding=1. Accepting RAs on a router can make the routing table confused since it can change the default router unintentionally. However, in practice there are cases where we cannot distinguish a host from a router clearly. For example, a customer edge router often works as a host against the ISP, and as a router against the LAN at the same time. Another example is a complex network configurations like an L2TP tunnel for IPv6 connection to Internet over an Ethernet link with another native IPv6 subnet. In this case, the physical interface for the native IPv6 subnet works as a host, and the pseudo-interface for L2TP works as the default IP forwarding route. Problem: Disabling processing RA messages when net.inet6.ip6.forwarding=1 and accepting them when net.inet6.ip6.forward=0 cause the following practical issues: - A router cannot perform SLAAC. It becomes a problem if a box has multiple interfaces and you want to use SLAAC on some of them, for example. A customer edge router for IPv6 Internet access service using an IPv6-over-IPv6 tunnel sometimes needs SLAAC on the physical interface for administration purpose; updating firmware and so on (link-local addresses can be used there, but GUAs by SLAAC are often used for scalability). - When a host has multiple IPv6 interfaces and it receives multiple RAs on them, controlling the default route is difficult. Router preferences defined in RFC 4191 works only when the routers on the links are under your control. Details of Implementation Changes: Router Advertisement messages will be accepted even when net.inet6.ip6.forwarding=1. More precisely, the conditions are as follow: (ACCEPT_RTADV && !NO_RADR && !ip6.forwarding) => Normal RA processing on that interface. (as IPv6 host) (ACCEPT_RTADV && (NO_RADR || ip6.forwarding)) => Accept RA but add the router to the defroute list with rtlifetime=0 unconditionally. This effectively prevents from setting the received router address as the box's default route. (!ACCEPT_RTADV) => No RA processing on that interface. ACCEPT_RTADV and NO_RADR are per-interface knob. In short, all interface are classified as "RA-accepting" or not. An RA-accepting interface always processes RA messages regardless of ip6.forwarding. The difference caused by NO_RADR or ip6.forwarding is whether the RA source address is considered as the default router or not. R-bit in NA on the RA accepting interfaces is set based on net.inet6.ip6.forwarding. While RFC 6204 W-1 rule (for CPE case) suggests a router should disable the R-bit completely even when the box has net.inet6.ip6.forwarding=1, I believe there is no technical reason with doing so. This behavior can be set by a new sysctl net.inet6.ip6.norbit_raif (the default is 0). Usage: # ifconfig fxp0 inet6 accept_rtadv => accept RA on fxp0 # ifconfig fxp0 inet6 accept_rtadv no_radr => accept RA on fxp0 but ignore default route information in it. # sysctl net.inet6.ip6.norbit_no_radr=1 => R-bit in NAs on RA accepting interfaces will always be set to 0.
2011-06-06 02:14:23 +00:00
/*
* Effectively-disable routes from RA messages when
* ND6_IFF_NO_RADR enabled on the receiving interface or
* (ip6.forwarding == 1 && ip6.rfc6204w3 != 1).
- Accept Router Advertisement messages even when net.inet6.ip6.forwarding=1. - A new per-interface knob IFF_ND6_NO_RADR and sysctl IPV6CTL_NO_RADR. This controls if accepting a route in an RA message as the default route. The default value for each interface can be set by net.inet6.ip6.no_radr. The system wide default value is 0. - A new sysctl: net.inet6.ip6.norbit_raif. This controls if setting R-bit in NA on RA accepting interfaces. The default is 0 (R-bit is set based on net.inet6.ip6.forwarding). Background: IPv6 host/router model suggests a router sends an RA and a host accepts it for router discovery. Because of that, KAME implementation does not allow accepting RAs when net.inet6.ip6.forwarding=1. Accepting RAs on a router can make the routing table confused since it can change the default router unintentionally. However, in practice there are cases where we cannot distinguish a host from a router clearly. For example, a customer edge router often works as a host against the ISP, and as a router against the LAN at the same time. Another example is a complex network configurations like an L2TP tunnel for IPv6 connection to Internet over an Ethernet link with another native IPv6 subnet. In this case, the physical interface for the native IPv6 subnet works as a host, and the pseudo-interface for L2TP works as the default IP forwarding route. Problem: Disabling processing RA messages when net.inet6.ip6.forwarding=1 and accepting them when net.inet6.ip6.forward=0 cause the following practical issues: - A router cannot perform SLAAC. It becomes a problem if a box has multiple interfaces and you want to use SLAAC on some of them, for example. A customer edge router for IPv6 Internet access service using an IPv6-over-IPv6 tunnel sometimes needs SLAAC on the physical interface for administration purpose; updating firmware and so on (link-local addresses can be used there, but GUAs by SLAAC are often used for scalability). - When a host has multiple IPv6 interfaces and it receives multiple RAs on them, controlling the default route is difficult. Router preferences defined in RFC 4191 works only when the routers on the links are under your control. Details of Implementation Changes: Router Advertisement messages will be accepted even when net.inet6.ip6.forwarding=1. More precisely, the conditions are as follow: (ACCEPT_RTADV && !NO_RADR && !ip6.forwarding) => Normal RA processing on that interface. (as IPv6 host) (ACCEPT_RTADV && (NO_RADR || ip6.forwarding)) => Accept RA but add the router to the defroute list with rtlifetime=0 unconditionally. This effectively prevents from setting the received router address as the box's default route. (!ACCEPT_RTADV) => No RA processing on that interface. ACCEPT_RTADV and NO_RADR are per-interface knob. In short, all interface are classified as "RA-accepting" or not. An RA-accepting interface always processes RA messages regardless of ip6.forwarding. The difference caused by NO_RADR or ip6.forwarding is whether the RA source address is considered as the default router or not. R-bit in NA on the RA accepting interfaces is set based on net.inet6.ip6.forwarding. While RFC 6204 W-1 rule (for CPE case) suggests a router should disable the R-bit completely even when the box has net.inet6.ip6.forwarding=1, I believe there is no technical reason with doing so. This behavior can be set by a new sysctl net.inet6.ip6.norbit_raif (the default is 0). Usage: # ifconfig fxp0 inet6 accept_rtadv => accept RA on fxp0 # ifconfig fxp0 inet6 accept_rtadv no_radr => accept RA on fxp0 but ignore default route information in it. # sysctl net.inet6.ip6.norbit_no_radr=1 => R-bit in NAs on RA accepting interfaces will always be set to 0.
2011-06-06 02:14:23 +00:00
*/
if (ndi->flags & ND6_IFF_NO_RADR)
dr0.rtlifetime = 0;
else if (V_ip6_forwarding && !V_ip6_rfc6204w3)
- Accept Router Advertisement messages even when net.inet6.ip6.forwarding=1. - A new per-interface knob IFF_ND6_NO_RADR and sysctl IPV6CTL_NO_RADR. This controls if accepting a route in an RA message as the default route. The default value for each interface can be set by net.inet6.ip6.no_radr. The system wide default value is 0. - A new sysctl: net.inet6.ip6.norbit_raif. This controls if setting R-bit in NA on RA accepting interfaces. The default is 0 (R-bit is set based on net.inet6.ip6.forwarding). Background: IPv6 host/router model suggests a router sends an RA and a host accepts it for router discovery. Because of that, KAME implementation does not allow accepting RAs when net.inet6.ip6.forwarding=1. Accepting RAs on a router can make the routing table confused since it can change the default router unintentionally. However, in practice there are cases where we cannot distinguish a host from a router clearly. For example, a customer edge router often works as a host against the ISP, and as a router against the LAN at the same time. Another example is a complex network configurations like an L2TP tunnel for IPv6 connection to Internet over an Ethernet link with another native IPv6 subnet. In this case, the physical interface for the native IPv6 subnet works as a host, and the pseudo-interface for L2TP works as the default IP forwarding route. Problem: Disabling processing RA messages when net.inet6.ip6.forwarding=1 and accepting them when net.inet6.ip6.forward=0 cause the following practical issues: - A router cannot perform SLAAC. It becomes a problem if a box has multiple interfaces and you want to use SLAAC on some of them, for example. A customer edge router for IPv6 Internet access service using an IPv6-over-IPv6 tunnel sometimes needs SLAAC on the physical interface for administration purpose; updating firmware and so on (link-local addresses can be used there, but GUAs by SLAAC are often used for scalability). - When a host has multiple IPv6 interfaces and it receives multiple RAs on them, controlling the default route is difficult. Router preferences defined in RFC 4191 works only when the routers on the links are under your control. Details of Implementation Changes: Router Advertisement messages will be accepted even when net.inet6.ip6.forwarding=1. More precisely, the conditions are as follow: (ACCEPT_RTADV && !NO_RADR && !ip6.forwarding) => Normal RA processing on that interface. (as IPv6 host) (ACCEPT_RTADV && (NO_RADR || ip6.forwarding)) => Accept RA but add the router to the defroute list with rtlifetime=0 unconditionally. This effectively prevents from setting the received router address as the box's default route. (!ACCEPT_RTADV) => No RA processing on that interface. ACCEPT_RTADV and NO_RADR are per-interface knob. In short, all interface are classified as "RA-accepting" or not. An RA-accepting interface always processes RA messages regardless of ip6.forwarding. The difference caused by NO_RADR or ip6.forwarding is whether the RA source address is considered as the default router or not. R-bit in NA on the RA accepting interfaces is set based on net.inet6.ip6.forwarding. While RFC 6204 W-1 rule (for CPE case) suggests a router should disable the R-bit completely even when the box has net.inet6.ip6.forwarding=1, I believe there is no technical reason with doing so. This behavior can be set by a new sysctl net.inet6.ip6.norbit_raif (the default is 0). Usage: # ifconfig fxp0 inet6 accept_rtadv => accept RA on fxp0 # ifconfig fxp0 inet6 accept_rtadv no_radr => accept RA on fxp0 but ignore default route information in it. # sysctl net.inet6.ip6.norbit_no_radr=1 => R-bit in NAs on RA accepting interfaces will always be set to 0.
2011-06-06 02:14:23 +00:00
dr0.rtlifetime = 0;
else
dr0.rtlifetime = ntohs(nd_ra->nd_ra_router_lifetime);
dr0.expire = time_uptime + dr0.rtlifetime;
dr0.ifp = ifp;
/* unspecified or not? (RFC 2461 6.3.4) */
if (advreachable) {
advreachable = ntohl(advreachable);
if (advreachable <= MAX_REACHABLE_TIME &&
ndi->basereachable != advreachable) {
ndi->basereachable = advreachable;
ndi->reachable = ND_COMPUTE_RTIME(ndi->basereachable);
ndi->recalctm = V_nd6_recalc_reachtm_interval; /* reset */
}
}
if (nd_ra->nd_ra_retransmit)
ndi->retrans = ntohl(nd_ra->nd_ra_retransmit);
if (nd_ra->nd_ra_curhoplimit) {
if (ndi->chlim < nd_ra->nd_ra_curhoplimit)
ndi->chlim = nd_ra->nd_ra_curhoplimit;
else if (ndi->chlim != nd_ra->nd_ra_curhoplimit) {
log(LOG_ERR, "RA with a lower CurHopLimit sent from "
"%s on %s (current = %d, received = %d). "
"Ignored.\n", ip6_sprintf(ip6bufs, &ip6->ip6_src),
if_name(ifp), ndi->chlim, nd_ra->nd_ra_curhoplimit);
}
}
dr = defrtrlist_update(&dr0);
}
/*
* prefix
*/
if (ndopts.nd_opts_pi) {
struct nd_opt_hdr *pt;
struct nd_opt_prefix_info *pi = NULL;
struct nd_prefixctl pr;
for (pt = (struct nd_opt_hdr *)ndopts.nd_opts_pi;
pt <= (struct nd_opt_hdr *)ndopts.nd_opts_pi_end;
pt = (struct nd_opt_hdr *)((caddr_t)pt +
(pt->nd_opt_len << 3))) {
if (pt->nd_opt_type != ND_OPT_PREFIX_INFORMATION)
continue;
pi = (struct nd_opt_prefix_info *)pt;
if (pi->nd_opt_pi_len != 4) {
nd6log((LOG_INFO,
"nd6_ra_input: invalid option "
"len %d for prefix information option, "
"ignored\n", pi->nd_opt_pi_len));
continue;
}
if (128 < pi->nd_opt_pi_prefix_len) {
nd6log((LOG_INFO,
"nd6_ra_input: invalid prefix "
"len %d for prefix information option, "
"ignored\n", pi->nd_opt_pi_prefix_len));
continue;
}
if (IN6_IS_ADDR_MULTICAST(&pi->nd_opt_pi_prefix)
|| IN6_IS_ADDR_LINKLOCAL(&pi->nd_opt_pi_prefix)) {
nd6log((LOG_INFO,
"nd6_ra_input: invalid prefix "
"%s, ignored\n",
ip6_sprintf(ip6bufs,
&pi->nd_opt_pi_prefix)));
continue;
}
bzero(&pr, sizeof(pr));
pr.ndpr_prefix.sin6_family = AF_INET6;
pr.ndpr_prefix.sin6_len = sizeof(pr.ndpr_prefix);
pr.ndpr_prefix.sin6_addr = pi->nd_opt_pi_prefix;
pr.ndpr_ifp = (struct ifnet *)m->m_pkthdr.rcvif;
pr.ndpr_raf_onlink = (pi->nd_opt_pi_flags_reserved &
ND_OPT_PI_FLAG_ONLINK) ? 1 : 0;
pr.ndpr_raf_auto = (pi->nd_opt_pi_flags_reserved &
ND_OPT_PI_FLAG_AUTO) ? 1 : 0;
pr.ndpr_plen = pi->nd_opt_pi_prefix_len;
pr.ndpr_vltime = ntohl(pi->nd_opt_pi_valid_time);
pr.ndpr_pltime = ntohl(pi->nd_opt_pi_preferred_time);
(void)prelist_update(&pr, dr, m, mcast);
}
}
if (dr != NULL) {
defrouter_rele(dr);
dr = NULL;
}
/*
* MTU
*/
if (ndopts.nd_opts_mtu && ndopts.nd_opts_mtu->nd_opt_mtu_len == 1) {
u_long mtu;
u_long maxmtu;
mtu = (u_long)ntohl(ndopts.nd_opts_mtu->nd_opt_mtu_mtu);
/* lower bound */
if (mtu < IPV6_MMTU) {
nd6log((LOG_INFO, "nd6_ra_input: bogus mtu option "
"mtu=%lu sent from %s, ignoring\n",
mtu, ip6_sprintf(ip6bufs, &ip6->ip6_src)));
goto skip;
}
/* upper bound */
maxmtu = (ndi->maxmtu && ndi->maxmtu < ifp->if_mtu)
? ndi->maxmtu : ifp->if_mtu;
if (mtu <= maxmtu) {
int change = (ndi->linkmtu != mtu);
ndi->linkmtu = mtu;
if (change) {
/* in6_maxmtu may change */
in6_setmaxmtu();
rt_updatemtu(ifp);
}
} else {
nd6log((LOG_INFO, "nd6_ra_input: bogus mtu "
"mtu=%lu sent from %s; "
"exceeds maxmtu %lu, ignoring\n",
mtu, ip6_sprintf(ip6bufs, &ip6->ip6_src), maxmtu));
}
}
skip:
/*
* Source link layer address
*/
{
char *lladdr = NULL;
int lladdrlen = 0;
if (ndopts.nd_opts_src_lladdr) {
lladdr = (char *)(ndopts.nd_opts_src_lladdr + 1);
lladdrlen = ndopts.nd_opts_src_lladdr->nd_opt_len << 3;
}
if (lladdr && ((ifp->if_addrlen + 2 + 7) & ~7) != lladdrlen) {
nd6log((LOG_INFO,
"nd6_ra_input: lladdrlen mismatch for %s "
"(if %d, RA packet %d)\n", ip6_sprintf(ip6bufs, &saddr6),
ifp->if_addrlen, lladdrlen - 2));
goto bad;
}
nd6_cache_lladdr(ifp, &saddr6, lladdr,
lladdrlen, ND_ROUTER_ADVERT, 0);
/*
* Installing a link-layer address might change the state of the
* router's neighbor cache, which might also affect our on-link
* detection of adveritsed prefixes.
*/
pfxlist_onlink_check();
}
freeit:
m_freem(m);
return;
bad:
ICMP6STAT_INC(icp6s_badra);
m_freem(m);
}
/* tell the change to user processes watching the routing socket. */
static void
nd6_rtmsg(int cmd, struct rtentry *rt)
{
struct rt_addrinfo info;
struct ifnet *ifp;
struct ifaddr *ifa;
bzero((caddr_t)&info, sizeof(info));
info.rti_info[RTAX_DST] = rt_key(rt);
info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
info.rti_info[RTAX_NETMASK] = rt_mask(rt);
ifp = rt->rt_ifp;
if (ifp != NULL) {
IF_ADDR_RLOCK(ifp);
ifnet: Replace if_addr_lock rwlock with epoch + mutex Run on LLNW canaries and tested by pho@ gallatin: Using a 14-core, 28-HTT single socket E5-2697 v3 with a 40GbE MLX5 based ConnectX 4-LX NIC, I see an almost 12% improvement in received packet rate, and a larger improvement in bytes delivered all the way to userspace. When the host receiving 64 streams of netperf -H $DUT -t UDP_STREAM -- -m 1, I see, using nstat -I mce0 1 before the patch: InMpps OMpps InGbs OGbs err TCP Est %CPU syscalls csw irq GBfree 4.98 0.00 4.42 0.00 4235592 33 83.80 4720653 2149771 1235 247.32 4.73 0.00 4.20 0.00 4025260 33 82.99 4724900 2139833 1204 247.32 4.72 0.00 4.20 0.00 4035252 33 82.14 4719162 2132023 1264 247.32 4.71 0.00 4.21 0.00 4073206 33 83.68 4744973 2123317 1347 247.32 4.72 0.00 4.21 0.00 4061118 33 80.82 4713615 2188091 1490 247.32 4.72 0.00 4.21 0.00 4051675 33 85.29 4727399 2109011 1205 247.32 4.73 0.00 4.21 0.00 4039056 33 84.65 4724735 2102603 1053 247.32 After the patch InMpps OMpps InGbs OGbs err TCP Est %CPU syscalls csw irq GBfree 5.43 0.00 4.20 0.00 3313143 33 84.96 5434214 1900162 2656 245.51 5.43 0.00 4.20 0.00 3308527 33 85.24 5439695 1809382 2521 245.51 5.42 0.00 4.19 0.00 3316778 33 87.54 5416028 1805835 2256 245.51 5.42 0.00 4.19 0.00 3317673 33 90.44 5426044 1763056 2332 245.51 5.42 0.00 4.19 0.00 3314839 33 88.11 5435732 1792218 2499 245.52 5.44 0.00 4.19 0.00 3293228 33 91.84 5426301 1668597 2121 245.52 Similarly, netperf reports 230Mb/s before the patch, and 270Mb/s after the patch Reviewed by: gallatin Sponsored by: Limelight Networks Differential Revision: https://reviews.freebsd.org/D15366
2018-05-18 20:13:34 +00:00
ifa = CK_STAILQ_FIRST(&ifp->if_addrhead);
info.rti_info[RTAX_IFP] = ifa->ifa_addr;
ifa_ref(ifa);
IF_ADDR_RUNLOCK(ifp);
info.rti_info[RTAX_IFA] = rt->rt_ifa->ifa_addr;
} else
ifa = NULL;
rt_missmsg_fib(cmd, &info, rt->rt_flags, 0, rt->rt_fibnum);
if (ifa != NULL)
ifa_free(ifa);
}
/*
* default router list processing sub routines
*/
static void
defrouter_addreq(struct nd_defrouter *new)
{
struct sockaddr_in6 def, mask, gate;
struct rtentry *newrt = NULL;
int error;
bzero(&def, sizeof(def));
bzero(&mask, sizeof(mask));
bzero(&gate, sizeof(gate));
def.sin6_len = mask.sin6_len = gate.sin6_len =
sizeof(struct sockaddr_in6);
def.sin6_family = gate.sin6_family = AF_INET6;
gate.sin6_addr = new->rtaddr;
error = in6_rtrequest(RTM_ADD, (struct sockaddr *)&def,
(struct sockaddr *)&gate, (struct sockaddr *)&mask,
Constrain IPv6 routes to single FIBs when net.add_addr_allfibs=0 sys/netinet6/icmp6.c Use the interface's FIB for source address selection in ICMPv6 error responses. sys/netinet6/in6.c In in6_newaddrmsg, announce arrival of local addresses on the interface's FIB only. In in6_lltable_rtcheck, use a per-fib ND6 cache instead of a single cache. sys/netinet6/in6_src.c In in6_selectsrc, use the caller's fib instead of the default fib. In in6_selectsrc_socket, remove a superfluous check. sys/netinet6/nd6.c In nd6_lle_event, use the interface's fib for routing socket messages. In nd6_is_new_addr_neighbor, check all FIBs when trying to determine whether an address is a neighbor. Also, simplify the code for point to point interfaces. sys/netinet6/nd6.h sys/netinet6/nd6.c sys/netinet6/nd6_rtr.c Make defrouter_select fib-aware, and make all of its callers pass in the interface fib. sys/netinet6/nd6_nbr.c When inputting a Neighbor Solicitation packet, consider the interface fib instead of the default fib for DAD. Output NS and Neighbor Advertisement packets on the correct fib. sys/netinet6/nd6_rtr.c Allow installing the same host route on different interfaces in different FIBs. If rt_add_addr_allfibs=0, only install or delete the prefix route on the interface fib. tests/sys/netinet/fibs_test.sh Clear some expected failures, but add a skip for the newly revealed BUG217871. PR: 196361 Submitted by: Erick Turnquist <jhujhiti@adjectivism.org> Reported by: Jason Healy <jhealy@logn.net> Reviewed by: asomers MFC after: 3 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D9451
2017-03-17 16:50:37 +00:00
RTF_GATEWAY, &newrt, new->ifp->if_fib);
if (newrt) {
nd6_rtmsg(RTM_ADD, newrt); /* tell user process */
RTFREE(newrt);
}
if (error == 0)
new->installed = 1;
}
struct nd_defrouter *
defrouter_lookup_locked(struct in6_addr *addr, struct ifnet *ifp)
{
struct nd_defrouter *dr;
ND6_LOCK_ASSERT();
TAILQ_FOREACH(dr, &V_nd_defrouter, dr_entry)
if (dr->ifp == ifp && IN6_ARE_ADDR_EQUAL(addr, &dr->rtaddr)) {
defrouter_ref(dr);
return (dr);
}
return (NULL);
}
struct nd_defrouter *
defrouter_lookup(struct in6_addr *addr, struct ifnet *ifp)
{
struct nd_defrouter *dr;
ND6_RLOCK();
dr = defrouter_lookup_locked(addr, ifp);
ND6_RUNLOCK();
return (dr);
}
void
defrouter_ref(struct nd_defrouter *dr)
{
refcount_acquire(&dr->refcnt);
}
void
defrouter_rele(struct nd_defrouter *dr)
{
if (refcount_release(&dr->refcnt))
free(dr, M_IP6NDP);
}
/*
* Remove the default route for a given router.
Constrain IPv6 routes to single FIBs when net.add_addr_allfibs=0 sys/netinet6/icmp6.c Use the interface's FIB for source address selection in ICMPv6 error responses. sys/netinet6/in6.c In in6_newaddrmsg, announce arrival of local addresses on the interface's FIB only. In in6_lltable_rtcheck, use a per-fib ND6 cache instead of a single cache. sys/netinet6/in6_src.c In in6_selectsrc, use the caller's fib instead of the default fib. In in6_selectsrc_socket, remove a superfluous check. sys/netinet6/nd6.c In nd6_lle_event, use the interface's fib for routing socket messages. In nd6_is_new_addr_neighbor, check all FIBs when trying to determine whether an address is a neighbor. Also, simplify the code for point to point interfaces. sys/netinet6/nd6.h sys/netinet6/nd6.c sys/netinet6/nd6_rtr.c Make defrouter_select fib-aware, and make all of its callers pass in the interface fib. sys/netinet6/nd6_nbr.c When inputting a Neighbor Solicitation packet, consider the interface fib instead of the default fib for DAD. Output NS and Neighbor Advertisement packets on the correct fib. sys/netinet6/nd6_rtr.c Allow installing the same host route on different interfaces in different FIBs. If rt_add_addr_allfibs=0, only install or delete the prefix route on the interface fib. tests/sys/netinet/fibs_test.sh Clear some expected failures, but add a skip for the newly revealed BUG217871. PR: 196361 Submitted by: Erick Turnquist <jhujhiti@adjectivism.org> Reported by: Jason Healy <jhealy@logn.net> Reviewed by: asomers MFC after: 3 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D9451
2017-03-17 16:50:37 +00:00
* This is just a subroutine function for defrouter_select_fib(), and
* should not be called from anywhere else.
*/
static void
defrouter_delreq(struct nd_defrouter *dr)
{
struct sockaddr_in6 def, mask, gate;
struct rtentry *oldrt = NULL;
bzero(&def, sizeof(def));
bzero(&mask, sizeof(mask));
bzero(&gate, sizeof(gate));
def.sin6_len = mask.sin6_len = gate.sin6_len =
sizeof(struct sockaddr_in6);
def.sin6_family = gate.sin6_family = AF_INET6;
gate.sin6_addr = dr->rtaddr;
in6_rtrequest(RTM_DELETE, (struct sockaddr *)&def,
(struct sockaddr *)&gate,
Constrain IPv6 routes to single FIBs when net.add_addr_allfibs=0 sys/netinet6/icmp6.c Use the interface's FIB for source address selection in ICMPv6 error responses. sys/netinet6/in6.c In in6_newaddrmsg, announce arrival of local addresses on the interface's FIB only. In in6_lltable_rtcheck, use a per-fib ND6 cache instead of a single cache. sys/netinet6/in6_src.c In in6_selectsrc, use the caller's fib instead of the default fib. In in6_selectsrc_socket, remove a superfluous check. sys/netinet6/nd6.c In nd6_lle_event, use the interface's fib for routing socket messages. In nd6_is_new_addr_neighbor, check all FIBs when trying to determine whether an address is a neighbor. Also, simplify the code for point to point interfaces. sys/netinet6/nd6.h sys/netinet6/nd6.c sys/netinet6/nd6_rtr.c Make defrouter_select fib-aware, and make all of its callers pass in the interface fib. sys/netinet6/nd6_nbr.c When inputting a Neighbor Solicitation packet, consider the interface fib instead of the default fib for DAD. Output NS and Neighbor Advertisement packets on the correct fib. sys/netinet6/nd6_rtr.c Allow installing the same host route on different interfaces in different FIBs. If rt_add_addr_allfibs=0, only install or delete the prefix route on the interface fib. tests/sys/netinet/fibs_test.sh Clear some expected failures, but add a skip for the newly revealed BUG217871. PR: 196361 Submitted by: Erick Turnquist <jhujhiti@adjectivism.org> Reported by: Jason Healy <jhealy@logn.net> Reviewed by: asomers MFC after: 3 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D9451
2017-03-17 16:50:37 +00:00
(struct sockaddr *)&mask, RTF_GATEWAY, &oldrt, dr->ifp->if_fib);
if (oldrt) {
nd6_rtmsg(RTM_DELETE, oldrt);
RTFREE(oldrt);
}
dr->installed = 0;
}
/*
* Remove all default routes from default router list.
*/
void
defrouter_reset(void)
{
struct nd_defrouter *dr, **dra;
int count, i;
count = i = 0;
/*
* We can't delete routes with the ND lock held, so make a copy of the
* current default router list and use that when deleting routes.
*/
ND6_RLOCK();
TAILQ_FOREACH(dr, &V_nd_defrouter, dr_entry)
count++;
ND6_RUNLOCK();
dra = malloc(count * sizeof(*dra), M_TEMP, M_WAITOK | M_ZERO);
ND6_RLOCK();
TAILQ_FOREACH(dr, &V_nd_defrouter, dr_entry) {
if (i == count)
break;
defrouter_ref(dr);
dra[i++] = dr;
}
ND6_RUNLOCK();
for (i = 0; i < count && dra[i] != NULL; i++) {
defrouter_delreq(dra[i]);
defrouter_rele(dra[i]);
}
free(dra, M_TEMP);
/*
* XXX should we also nuke any default routers in the kernel, by
* going through them by rtalloc1()?
*/
}
/*
* Look up a matching default router list entry and remove it. Returns true if a
* matching entry was found, false otherwise.
*/
bool
defrouter_remove(struct in6_addr *addr, struct ifnet *ifp)
{
struct nd_defrouter *dr;
ND6_WLOCK();
dr = defrouter_lookup_locked(addr, ifp);
if (dr == NULL) {
ND6_WUNLOCK();
return (false);
}
defrouter_unlink(dr, NULL);
ND6_WUNLOCK();
defrouter_del(dr);
defrouter_rele(dr);
return (true);
}
/*
* Remove a router from the global list and optionally stash it in a
* caller-supplied queue.
*
* The ND lock must be held.
*/
void
defrouter_unlink(struct nd_defrouter *dr, struct nd_drhead *drq)
{
ND6_WLOCK_ASSERT();
TAILQ_REMOVE(&V_nd_defrouter, dr, dr_entry);
V_nd6_list_genid++;
if (drq != NULL)
TAILQ_INSERT_TAIL(drq, dr, dr_entry);
}
void
defrouter_del(struct nd_defrouter *dr)
{
struct nd_defrouter *deldr = NULL;
struct nd_prefix *pr;
struct nd_pfxrouter *pfxrtr;
ND6_UNLOCK_ASSERT();
/*
* Flush all the routing table entries that use the router
* as a next hop.
*/
- Accept Router Advertisement messages even when net.inet6.ip6.forwarding=1. - A new per-interface knob IFF_ND6_NO_RADR and sysctl IPV6CTL_NO_RADR. This controls if accepting a route in an RA message as the default route. The default value for each interface can be set by net.inet6.ip6.no_radr. The system wide default value is 0. - A new sysctl: net.inet6.ip6.norbit_raif. This controls if setting R-bit in NA on RA accepting interfaces. The default is 0 (R-bit is set based on net.inet6.ip6.forwarding). Background: IPv6 host/router model suggests a router sends an RA and a host accepts it for router discovery. Because of that, KAME implementation does not allow accepting RAs when net.inet6.ip6.forwarding=1. Accepting RAs on a router can make the routing table confused since it can change the default router unintentionally. However, in practice there are cases where we cannot distinguish a host from a router clearly. For example, a customer edge router often works as a host against the ISP, and as a router against the LAN at the same time. Another example is a complex network configurations like an L2TP tunnel for IPv6 connection to Internet over an Ethernet link with another native IPv6 subnet. In this case, the physical interface for the native IPv6 subnet works as a host, and the pseudo-interface for L2TP works as the default IP forwarding route. Problem: Disabling processing RA messages when net.inet6.ip6.forwarding=1 and accepting them when net.inet6.ip6.forward=0 cause the following practical issues: - A router cannot perform SLAAC. It becomes a problem if a box has multiple interfaces and you want to use SLAAC on some of them, for example. A customer edge router for IPv6 Internet access service using an IPv6-over-IPv6 tunnel sometimes needs SLAAC on the physical interface for administration purpose; updating firmware and so on (link-local addresses can be used there, but GUAs by SLAAC are often used for scalability). - When a host has multiple IPv6 interfaces and it receives multiple RAs on them, controlling the default route is difficult. Router preferences defined in RFC 4191 works only when the routers on the links are under your control. Details of Implementation Changes: Router Advertisement messages will be accepted even when net.inet6.ip6.forwarding=1. More precisely, the conditions are as follow: (ACCEPT_RTADV && !NO_RADR && !ip6.forwarding) => Normal RA processing on that interface. (as IPv6 host) (ACCEPT_RTADV && (NO_RADR || ip6.forwarding)) => Accept RA but add the router to the defroute list with rtlifetime=0 unconditionally. This effectively prevents from setting the received router address as the box's default route. (!ACCEPT_RTADV) => No RA processing on that interface. ACCEPT_RTADV and NO_RADR are per-interface knob. In short, all interface are classified as "RA-accepting" or not. An RA-accepting interface always processes RA messages regardless of ip6.forwarding. The difference caused by NO_RADR or ip6.forwarding is whether the RA source address is considered as the default router or not. R-bit in NA on the RA accepting interfaces is set based on net.inet6.ip6.forwarding. While RFC 6204 W-1 rule (for CPE case) suggests a router should disable the R-bit completely even when the box has net.inet6.ip6.forwarding=1, I believe there is no technical reason with doing so. This behavior can be set by a new sysctl net.inet6.ip6.norbit_raif (the default is 0). Usage: # ifconfig fxp0 inet6 accept_rtadv => accept RA on fxp0 # ifconfig fxp0 inet6 accept_rtadv no_radr => accept RA on fxp0 but ignore default route information in it. # sysctl net.inet6.ip6.norbit_no_radr=1 => R-bit in NAs on RA accepting interfaces will always be set to 0.
2011-06-06 02:14:23 +00:00
if (ND_IFINFO(dr->ifp)->flags & ND6_IFF_ACCEPT_RTADV)
rt6_flush(&dr->rtaddr, dr->ifp);
if (dr->installed) {
deldr = dr;
defrouter_delreq(dr);
}
/*
* Also delete all the pointers to the router in each prefix lists.
*/
ND6_WLOCK();
LIST_FOREACH(pr, &V_nd_prefix, ndpr_entry) {
if ((pfxrtr = pfxrtr_lookup(pr, dr)) != NULL)
pfxrtr_del(pfxrtr);
}
ND6_WUNLOCK();
pfxlist_onlink_check();
/*
* If the router is the primary one, choose a new one.
Constrain IPv6 routes to single FIBs when net.add_addr_allfibs=0 sys/netinet6/icmp6.c Use the interface's FIB for source address selection in ICMPv6 error responses. sys/netinet6/in6.c In in6_newaddrmsg, announce arrival of local addresses on the interface's FIB only. In in6_lltable_rtcheck, use a per-fib ND6 cache instead of a single cache. sys/netinet6/in6_src.c In in6_selectsrc, use the caller's fib instead of the default fib. In in6_selectsrc_socket, remove a superfluous check. sys/netinet6/nd6.c In nd6_lle_event, use the interface's fib for routing socket messages. In nd6_is_new_addr_neighbor, check all FIBs when trying to determine whether an address is a neighbor. Also, simplify the code for point to point interfaces. sys/netinet6/nd6.h sys/netinet6/nd6.c sys/netinet6/nd6_rtr.c Make defrouter_select fib-aware, and make all of its callers pass in the interface fib. sys/netinet6/nd6_nbr.c When inputting a Neighbor Solicitation packet, consider the interface fib instead of the default fib for DAD. Output NS and Neighbor Advertisement packets on the correct fib. sys/netinet6/nd6_rtr.c Allow installing the same host route on different interfaces in different FIBs. If rt_add_addr_allfibs=0, only install or delete the prefix route on the interface fib. tests/sys/netinet/fibs_test.sh Clear some expected failures, but add a skip for the newly revealed BUG217871. PR: 196361 Submitted by: Erick Turnquist <jhujhiti@adjectivism.org> Reported by: Jason Healy <jhealy@logn.net> Reviewed by: asomers MFC after: 3 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D9451
2017-03-17 16:50:37 +00:00
* Note that defrouter_select_fib() will remove the current
* gateway from the routing table.
*/
if (deldr)
Constrain IPv6 routes to single FIBs when net.add_addr_allfibs=0 sys/netinet6/icmp6.c Use the interface's FIB for source address selection in ICMPv6 error responses. sys/netinet6/in6.c In in6_newaddrmsg, announce arrival of local addresses on the interface's FIB only. In in6_lltable_rtcheck, use a per-fib ND6 cache instead of a single cache. sys/netinet6/in6_src.c In in6_selectsrc, use the caller's fib instead of the default fib. In in6_selectsrc_socket, remove a superfluous check. sys/netinet6/nd6.c In nd6_lle_event, use the interface's fib for routing socket messages. In nd6_is_new_addr_neighbor, check all FIBs when trying to determine whether an address is a neighbor. Also, simplify the code for point to point interfaces. sys/netinet6/nd6.h sys/netinet6/nd6.c sys/netinet6/nd6_rtr.c Make defrouter_select fib-aware, and make all of its callers pass in the interface fib. sys/netinet6/nd6_nbr.c When inputting a Neighbor Solicitation packet, consider the interface fib instead of the default fib for DAD. Output NS and Neighbor Advertisement packets on the correct fib. sys/netinet6/nd6_rtr.c Allow installing the same host route on different interfaces in different FIBs. If rt_add_addr_allfibs=0, only install or delete the prefix route on the interface fib. tests/sys/netinet/fibs_test.sh Clear some expected failures, but add a skip for the newly revealed BUG217871. PR: 196361 Submitted by: Erick Turnquist <jhujhiti@adjectivism.org> Reported by: Jason Healy <jhealy@logn.net> Reviewed by: asomers MFC after: 3 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D9451
2017-03-17 16:50:37 +00:00
defrouter_select_fib(deldr->ifp->if_fib);
/*
* Release the list reference.
*/
defrouter_rele(dr);
}
/*
* Default Router Selection according to Section 6.3.6 of RFC 2461 and
* draft-ietf-ipngwg-router-selection:
* 1) Routers that are reachable or probably reachable should be preferred.
* If we have more than one (probably) reachable router, prefer ones
* with the highest router preference.
* 2) When no routers on the list are known to be reachable or
* probably reachable, routers SHOULD be selected in a round-robin
* fashion, regardless of router preference values.
* 3) If the Default Router List is empty, assume that all
* destinations are on-link.
*
* We assume nd_defrouter is sorted by router preference value.
* Since the code below covers both with and without router preference cases,
* we do not need to classify the cases by ifdef.
*
* At this moment, we do not try to install more than one default router,
* even when the multipath routing is available, because we're not sure about
* the benefits for stub hosts comparing to the risk of making the code
* complicated and the possibility of introducing bugs.
Constrain IPv6 routes to single FIBs when net.add_addr_allfibs=0 sys/netinet6/icmp6.c Use the interface's FIB for source address selection in ICMPv6 error responses. sys/netinet6/in6.c In in6_newaddrmsg, announce arrival of local addresses on the interface's FIB only. In in6_lltable_rtcheck, use a per-fib ND6 cache instead of a single cache. sys/netinet6/in6_src.c In in6_selectsrc, use the caller's fib instead of the default fib. In in6_selectsrc_socket, remove a superfluous check. sys/netinet6/nd6.c In nd6_lle_event, use the interface's fib for routing socket messages. In nd6_is_new_addr_neighbor, check all FIBs when trying to determine whether an address is a neighbor. Also, simplify the code for point to point interfaces. sys/netinet6/nd6.h sys/netinet6/nd6.c sys/netinet6/nd6_rtr.c Make defrouter_select fib-aware, and make all of its callers pass in the interface fib. sys/netinet6/nd6_nbr.c When inputting a Neighbor Solicitation packet, consider the interface fib instead of the default fib for DAD. Output NS and Neighbor Advertisement packets on the correct fib. sys/netinet6/nd6_rtr.c Allow installing the same host route on different interfaces in different FIBs. If rt_add_addr_allfibs=0, only install or delete the prefix route on the interface fib. tests/sys/netinet/fibs_test.sh Clear some expected failures, but add a skip for the newly revealed BUG217871. PR: 196361 Submitted by: Erick Turnquist <jhujhiti@adjectivism.org> Reported by: Jason Healy <jhealy@logn.net> Reviewed by: asomers MFC after: 3 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D9451
2017-03-17 16:50:37 +00:00
*
* We maintain a single list of routers for multiple FIBs, only considering one
* at a time based on the receiving interface's FIB. If @fibnum is RT_ALL_FIBS,
* we do the whole thing multiple times.
*/
void
Constrain IPv6 routes to single FIBs when net.add_addr_allfibs=0 sys/netinet6/icmp6.c Use the interface's FIB for source address selection in ICMPv6 error responses. sys/netinet6/in6.c In in6_newaddrmsg, announce arrival of local addresses on the interface's FIB only. In in6_lltable_rtcheck, use a per-fib ND6 cache instead of a single cache. sys/netinet6/in6_src.c In in6_selectsrc, use the caller's fib instead of the default fib. In in6_selectsrc_socket, remove a superfluous check. sys/netinet6/nd6.c In nd6_lle_event, use the interface's fib for routing socket messages. In nd6_is_new_addr_neighbor, check all FIBs when trying to determine whether an address is a neighbor. Also, simplify the code for point to point interfaces. sys/netinet6/nd6.h sys/netinet6/nd6.c sys/netinet6/nd6_rtr.c Make defrouter_select fib-aware, and make all of its callers pass in the interface fib. sys/netinet6/nd6_nbr.c When inputting a Neighbor Solicitation packet, consider the interface fib instead of the default fib for DAD. Output NS and Neighbor Advertisement packets on the correct fib. sys/netinet6/nd6_rtr.c Allow installing the same host route on different interfaces in different FIBs. If rt_add_addr_allfibs=0, only install or delete the prefix route on the interface fib. tests/sys/netinet/fibs_test.sh Clear some expected failures, but add a skip for the newly revealed BUG217871. PR: 196361 Submitted by: Erick Turnquist <jhujhiti@adjectivism.org> Reported by: Jason Healy <jhealy@logn.net> Reviewed by: asomers MFC after: 3 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D9451
2017-03-17 16:50:37 +00:00
defrouter_select_fib(int fibnum)
{
struct nd_defrouter *dr, *selected_dr, *installed_dr;
struct llentry *ln = NULL;
Constrain IPv6 routes to single FIBs when net.add_addr_allfibs=0 sys/netinet6/icmp6.c Use the interface's FIB for source address selection in ICMPv6 error responses. sys/netinet6/in6.c In in6_newaddrmsg, announce arrival of local addresses on the interface's FIB only. In in6_lltable_rtcheck, use a per-fib ND6 cache instead of a single cache. sys/netinet6/in6_src.c In in6_selectsrc, use the caller's fib instead of the default fib. In in6_selectsrc_socket, remove a superfluous check. sys/netinet6/nd6.c In nd6_lle_event, use the interface's fib for routing socket messages. In nd6_is_new_addr_neighbor, check all FIBs when trying to determine whether an address is a neighbor. Also, simplify the code for point to point interfaces. sys/netinet6/nd6.h sys/netinet6/nd6.c sys/netinet6/nd6_rtr.c Make defrouter_select fib-aware, and make all of its callers pass in the interface fib. sys/netinet6/nd6_nbr.c When inputting a Neighbor Solicitation packet, consider the interface fib instead of the default fib for DAD. Output NS and Neighbor Advertisement packets on the correct fib. sys/netinet6/nd6_rtr.c Allow installing the same host route on different interfaces in different FIBs. If rt_add_addr_allfibs=0, only install or delete the prefix route on the interface fib. tests/sys/netinet/fibs_test.sh Clear some expected failures, but add a skip for the newly revealed BUG217871. PR: 196361 Submitted by: Erick Turnquist <jhujhiti@adjectivism.org> Reported by: Jason Healy <jhealy@logn.net> Reviewed by: asomers MFC after: 3 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D9451
2017-03-17 16:50:37 +00:00
if (fibnum == RT_ALL_FIBS) {
for (fibnum = 0; fibnum < rt_numfibs; fibnum++) {
defrouter_select_fib(fibnum);
}
}
ND6_RLOCK();
/*
* Let's handle easy case (3) first:
* If default router list is empty, there's nothing to be done.
*/
if (TAILQ_EMPTY(&V_nd_defrouter)) {
ND6_RUNLOCK();
return;
}
/*
* Search for a (probably) reachable router from the list.
* We just pick up the first reachable one (if any), assuming that
* the ordering rule of the list described in defrtrlist_update().
*/
selected_dr = installed_dr = NULL;
TAILQ_FOREACH(dr, &V_nd_defrouter, dr_entry) {
IF_AFDATA_RLOCK(dr->ifp);
Constrain IPv6 routes to single FIBs when net.add_addr_allfibs=0 sys/netinet6/icmp6.c Use the interface's FIB for source address selection in ICMPv6 error responses. sys/netinet6/in6.c In in6_newaddrmsg, announce arrival of local addresses on the interface's FIB only. In in6_lltable_rtcheck, use a per-fib ND6 cache instead of a single cache. sys/netinet6/in6_src.c In in6_selectsrc, use the caller's fib instead of the default fib. In in6_selectsrc_socket, remove a superfluous check. sys/netinet6/nd6.c In nd6_lle_event, use the interface's fib for routing socket messages. In nd6_is_new_addr_neighbor, check all FIBs when trying to determine whether an address is a neighbor. Also, simplify the code for point to point interfaces. sys/netinet6/nd6.h sys/netinet6/nd6.c sys/netinet6/nd6_rtr.c Make defrouter_select fib-aware, and make all of its callers pass in the interface fib. sys/netinet6/nd6_nbr.c When inputting a Neighbor Solicitation packet, consider the interface fib instead of the default fib for DAD. Output NS and Neighbor Advertisement packets on the correct fib. sys/netinet6/nd6_rtr.c Allow installing the same host route on different interfaces in different FIBs. If rt_add_addr_allfibs=0, only install or delete the prefix route on the interface fib. tests/sys/netinet/fibs_test.sh Clear some expected failures, but add a skip for the newly revealed BUG217871. PR: 196361 Submitted by: Erick Turnquist <jhujhiti@adjectivism.org> Reported by: Jason Healy <jhealy@logn.net> Reviewed by: asomers MFC after: 3 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D9451
2017-03-17 16:50:37 +00:00
if (selected_dr == NULL && dr->ifp->if_fib == fibnum &&
(ln = nd6_lookup(&dr->rtaddr, 0, dr->ifp)) &&
ND6_IS_LLINFO_PROBREACH(ln)) {
selected_dr = dr;
defrouter_ref(selected_dr);
}
IF_AFDATA_RUNLOCK(dr->ifp);
if (ln != NULL) {
LLE_RUNLOCK(ln);
ln = NULL;
}
Constrain IPv6 routes to single FIBs when net.add_addr_allfibs=0 sys/netinet6/icmp6.c Use the interface's FIB for source address selection in ICMPv6 error responses. sys/netinet6/in6.c In in6_newaddrmsg, announce arrival of local addresses on the interface's FIB only. In in6_lltable_rtcheck, use a per-fib ND6 cache instead of a single cache. sys/netinet6/in6_src.c In in6_selectsrc, use the caller's fib instead of the default fib. In in6_selectsrc_socket, remove a superfluous check. sys/netinet6/nd6.c In nd6_lle_event, use the interface's fib for routing socket messages. In nd6_is_new_addr_neighbor, check all FIBs when trying to determine whether an address is a neighbor. Also, simplify the code for point to point interfaces. sys/netinet6/nd6.h sys/netinet6/nd6.c sys/netinet6/nd6_rtr.c Make defrouter_select fib-aware, and make all of its callers pass in the interface fib. sys/netinet6/nd6_nbr.c When inputting a Neighbor Solicitation packet, consider the interface fib instead of the default fib for DAD. Output NS and Neighbor Advertisement packets on the correct fib. sys/netinet6/nd6_rtr.c Allow installing the same host route on different interfaces in different FIBs. If rt_add_addr_allfibs=0, only install or delete the prefix route on the interface fib. tests/sys/netinet/fibs_test.sh Clear some expected failures, but add a skip for the newly revealed BUG217871. PR: 196361 Submitted by: Erick Turnquist <jhujhiti@adjectivism.org> Reported by: Jason Healy <jhealy@logn.net> Reviewed by: asomers MFC after: 3 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D9451
2017-03-17 16:50:37 +00:00
if (dr->installed && dr->ifp->if_fib == fibnum) {
if (installed_dr == NULL) {
installed_dr = dr;
defrouter_ref(installed_dr);
} else {
Constrain IPv6 routes to single FIBs when net.add_addr_allfibs=0 sys/netinet6/icmp6.c Use the interface's FIB for source address selection in ICMPv6 error responses. sys/netinet6/in6.c In in6_newaddrmsg, announce arrival of local addresses on the interface's FIB only. In in6_lltable_rtcheck, use a per-fib ND6 cache instead of a single cache. sys/netinet6/in6_src.c In in6_selectsrc, use the caller's fib instead of the default fib. In in6_selectsrc_socket, remove a superfluous check. sys/netinet6/nd6.c In nd6_lle_event, use the interface's fib for routing socket messages. In nd6_is_new_addr_neighbor, check all FIBs when trying to determine whether an address is a neighbor. Also, simplify the code for point to point interfaces. sys/netinet6/nd6.h sys/netinet6/nd6.c sys/netinet6/nd6_rtr.c Make defrouter_select fib-aware, and make all of its callers pass in the interface fib. sys/netinet6/nd6_nbr.c When inputting a Neighbor Solicitation packet, consider the interface fib instead of the default fib for DAD. Output NS and Neighbor Advertisement packets on the correct fib. sys/netinet6/nd6_rtr.c Allow installing the same host route on different interfaces in different FIBs. If rt_add_addr_allfibs=0, only install or delete the prefix route on the interface fib. tests/sys/netinet/fibs_test.sh Clear some expected failures, but add a skip for the newly revealed BUG217871. PR: 196361 Submitted by: Erick Turnquist <jhujhiti@adjectivism.org> Reported by: Jason Healy <jhealy@logn.net> Reviewed by: asomers MFC after: 3 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D9451
2017-03-17 16:50:37 +00:00
/*
* this should not happen.
* warn for diagnosis.
*/
log(LOG_ERR, "defrouter_select_fib: more than "
"one router is installed\n");
}
}
}
/*
* If none of the default routers was found to be reachable,
* round-robin the list regardless of preference.
* Otherwise, if we have an installed router, check if the selected
* (reachable) router should really be preferred to the installed one.
* We only prefer the new router when the old one is not reachable
* or when the new one has a really higher preference value.
*/
if (selected_dr == NULL) {
if (installed_dr == NULL ||
TAILQ_NEXT(installed_dr, dr_entry) == NULL)
Constrain IPv6 routes to single FIBs when net.add_addr_allfibs=0 sys/netinet6/icmp6.c Use the interface's FIB for source address selection in ICMPv6 error responses. sys/netinet6/in6.c In in6_newaddrmsg, announce arrival of local addresses on the interface's FIB only. In in6_lltable_rtcheck, use a per-fib ND6 cache instead of a single cache. sys/netinet6/in6_src.c In in6_selectsrc, use the caller's fib instead of the default fib. In in6_selectsrc_socket, remove a superfluous check. sys/netinet6/nd6.c In nd6_lle_event, use the interface's fib for routing socket messages. In nd6_is_new_addr_neighbor, check all FIBs when trying to determine whether an address is a neighbor. Also, simplify the code for point to point interfaces. sys/netinet6/nd6.h sys/netinet6/nd6.c sys/netinet6/nd6_rtr.c Make defrouter_select fib-aware, and make all of its callers pass in the interface fib. sys/netinet6/nd6_nbr.c When inputting a Neighbor Solicitation packet, consider the interface fib instead of the default fib for DAD. Output NS and Neighbor Advertisement packets on the correct fib. sys/netinet6/nd6_rtr.c Allow installing the same host route on different interfaces in different FIBs. If rt_add_addr_allfibs=0, only install or delete the prefix route on the interface fib. tests/sys/netinet/fibs_test.sh Clear some expected failures, but add a skip for the newly revealed BUG217871. PR: 196361 Submitted by: Erick Turnquist <jhujhiti@adjectivism.org> Reported by: Jason Healy <jhealy@logn.net> Reviewed by: asomers MFC after: 3 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D9451
2017-03-17 16:50:37 +00:00
dr = TAILQ_FIRST(&V_nd_defrouter);
else
Constrain IPv6 routes to single FIBs when net.add_addr_allfibs=0 sys/netinet6/icmp6.c Use the interface's FIB for source address selection in ICMPv6 error responses. sys/netinet6/in6.c In in6_newaddrmsg, announce arrival of local addresses on the interface's FIB only. In in6_lltable_rtcheck, use a per-fib ND6 cache instead of a single cache. sys/netinet6/in6_src.c In in6_selectsrc, use the caller's fib instead of the default fib. In in6_selectsrc_socket, remove a superfluous check. sys/netinet6/nd6.c In nd6_lle_event, use the interface's fib for routing socket messages. In nd6_is_new_addr_neighbor, check all FIBs when trying to determine whether an address is a neighbor. Also, simplify the code for point to point interfaces. sys/netinet6/nd6.h sys/netinet6/nd6.c sys/netinet6/nd6_rtr.c Make defrouter_select fib-aware, and make all of its callers pass in the interface fib. sys/netinet6/nd6_nbr.c When inputting a Neighbor Solicitation packet, consider the interface fib instead of the default fib for DAD. Output NS and Neighbor Advertisement packets on the correct fib. sys/netinet6/nd6_rtr.c Allow installing the same host route on different interfaces in different FIBs. If rt_add_addr_allfibs=0, only install or delete the prefix route on the interface fib. tests/sys/netinet/fibs_test.sh Clear some expected failures, but add a skip for the newly revealed BUG217871. PR: 196361 Submitted by: Erick Turnquist <jhujhiti@adjectivism.org> Reported by: Jason Healy <jhealy@logn.net> Reviewed by: asomers MFC after: 3 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D9451
2017-03-17 16:50:37 +00:00
dr = TAILQ_NEXT(installed_dr, dr_entry);
/* Ensure we select a router for this FIB. */
TAILQ_FOREACH_FROM(dr, &V_nd_defrouter, dr_entry) {
if (dr->ifp->if_fib == fibnum) {
selected_dr = dr;
defrouter_ref(selected_dr);
break;
}
}
} else if (installed_dr != NULL) {
IF_AFDATA_RLOCK(installed_dr->ifp);
Constrain IPv6 routes to single FIBs when net.add_addr_allfibs=0 sys/netinet6/icmp6.c Use the interface's FIB for source address selection in ICMPv6 error responses. sys/netinet6/in6.c In in6_newaddrmsg, announce arrival of local addresses on the interface's FIB only. In in6_lltable_rtcheck, use a per-fib ND6 cache instead of a single cache. sys/netinet6/in6_src.c In in6_selectsrc, use the caller's fib instead of the default fib. In in6_selectsrc_socket, remove a superfluous check. sys/netinet6/nd6.c In nd6_lle_event, use the interface's fib for routing socket messages. In nd6_is_new_addr_neighbor, check all FIBs when trying to determine whether an address is a neighbor. Also, simplify the code for point to point interfaces. sys/netinet6/nd6.h sys/netinet6/nd6.c sys/netinet6/nd6_rtr.c Make defrouter_select fib-aware, and make all of its callers pass in the interface fib. sys/netinet6/nd6_nbr.c When inputting a Neighbor Solicitation packet, consider the interface fib instead of the default fib for DAD. Output NS and Neighbor Advertisement packets on the correct fib. sys/netinet6/nd6_rtr.c Allow installing the same host route on different interfaces in different FIBs. If rt_add_addr_allfibs=0, only install or delete the prefix route on the interface fib. tests/sys/netinet/fibs_test.sh Clear some expected failures, but add a skip for the newly revealed BUG217871. PR: 196361 Submitted by: Erick Turnquist <jhujhiti@adjectivism.org> Reported by: Jason Healy <jhealy@logn.net> Reviewed by: asomers MFC after: 3 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D9451
2017-03-17 16:50:37 +00:00
if ((ln = nd6_lookup(&installed_dr->rtaddr, 0,
installed_dr->ifp)) &&
ND6_IS_LLINFO_PROBREACH(ln) &&
Constrain IPv6 routes to single FIBs when net.add_addr_allfibs=0 sys/netinet6/icmp6.c Use the interface's FIB for source address selection in ICMPv6 error responses. sys/netinet6/in6.c In in6_newaddrmsg, announce arrival of local addresses on the interface's FIB only. In in6_lltable_rtcheck, use a per-fib ND6 cache instead of a single cache. sys/netinet6/in6_src.c In in6_selectsrc, use the caller's fib instead of the default fib. In in6_selectsrc_socket, remove a superfluous check. sys/netinet6/nd6.c In nd6_lle_event, use the interface's fib for routing socket messages. In nd6_is_new_addr_neighbor, check all FIBs when trying to determine whether an address is a neighbor. Also, simplify the code for point to point interfaces. sys/netinet6/nd6.h sys/netinet6/nd6.c sys/netinet6/nd6_rtr.c Make defrouter_select fib-aware, and make all of its callers pass in the interface fib. sys/netinet6/nd6_nbr.c When inputting a Neighbor Solicitation packet, consider the interface fib instead of the default fib for DAD. Output NS and Neighbor Advertisement packets on the correct fib. sys/netinet6/nd6_rtr.c Allow installing the same host route on different interfaces in different FIBs. If rt_add_addr_allfibs=0, only install or delete the prefix route on the interface fib. tests/sys/netinet/fibs_test.sh Clear some expected failures, but add a skip for the newly revealed BUG217871. PR: 196361 Submitted by: Erick Turnquist <jhujhiti@adjectivism.org> Reported by: Jason Healy <jhealy@logn.net> Reviewed by: asomers MFC after: 3 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D9451
2017-03-17 16:50:37 +00:00
installed_dr->ifp->if_fib == fibnum &&
rtpref(selected_dr) <= rtpref(installed_dr)) {
defrouter_rele(selected_dr);
selected_dr = installed_dr;
}
IF_AFDATA_RUNLOCK(installed_dr->ifp);
if (ln != NULL)
LLE_RUNLOCK(ln);
}
ND6_RUNLOCK();
/*
Constrain IPv6 routes to single FIBs when net.add_addr_allfibs=0 sys/netinet6/icmp6.c Use the interface's FIB for source address selection in ICMPv6 error responses. sys/netinet6/in6.c In in6_newaddrmsg, announce arrival of local addresses on the interface's FIB only. In in6_lltable_rtcheck, use a per-fib ND6 cache instead of a single cache. sys/netinet6/in6_src.c In in6_selectsrc, use the caller's fib instead of the default fib. In in6_selectsrc_socket, remove a superfluous check. sys/netinet6/nd6.c In nd6_lle_event, use the interface's fib for routing socket messages. In nd6_is_new_addr_neighbor, check all FIBs when trying to determine whether an address is a neighbor. Also, simplify the code for point to point interfaces. sys/netinet6/nd6.h sys/netinet6/nd6.c sys/netinet6/nd6_rtr.c Make defrouter_select fib-aware, and make all of its callers pass in the interface fib. sys/netinet6/nd6_nbr.c When inputting a Neighbor Solicitation packet, consider the interface fib instead of the default fib for DAD. Output NS and Neighbor Advertisement packets on the correct fib. sys/netinet6/nd6_rtr.c Allow installing the same host route on different interfaces in different FIBs. If rt_add_addr_allfibs=0, only install or delete the prefix route on the interface fib. tests/sys/netinet/fibs_test.sh Clear some expected failures, but add a skip for the newly revealed BUG217871. PR: 196361 Submitted by: Erick Turnquist <jhujhiti@adjectivism.org> Reported by: Jason Healy <jhealy@logn.net> Reviewed by: asomers MFC after: 3 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D9451
2017-03-17 16:50:37 +00:00
* If we selected a router for this FIB and it's different
* than the installed one, remove the installed router and
* install the selected one in its place.
*/
if (installed_dr != selected_dr) {
if (installed_dr != NULL) {
defrouter_delreq(installed_dr);
defrouter_rele(installed_dr);
}
Constrain IPv6 routes to single FIBs when net.add_addr_allfibs=0 sys/netinet6/icmp6.c Use the interface's FIB for source address selection in ICMPv6 error responses. sys/netinet6/in6.c In in6_newaddrmsg, announce arrival of local addresses on the interface's FIB only. In in6_lltable_rtcheck, use a per-fib ND6 cache instead of a single cache. sys/netinet6/in6_src.c In in6_selectsrc, use the caller's fib instead of the default fib. In in6_selectsrc_socket, remove a superfluous check. sys/netinet6/nd6.c In nd6_lle_event, use the interface's fib for routing socket messages. In nd6_is_new_addr_neighbor, check all FIBs when trying to determine whether an address is a neighbor. Also, simplify the code for point to point interfaces. sys/netinet6/nd6.h sys/netinet6/nd6.c sys/netinet6/nd6_rtr.c Make defrouter_select fib-aware, and make all of its callers pass in the interface fib. sys/netinet6/nd6_nbr.c When inputting a Neighbor Solicitation packet, consider the interface fib instead of the default fib for DAD. Output NS and Neighbor Advertisement packets on the correct fib. sys/netinet6/nd6_rtr.c Allow installing the same host route on different interfaces in different FIBs. If rt_add_addr_allfibs=0, only install or delete the prefix route on the interface fib. tests/sys/netinet/fibs_test.sh Clear some expected failures, but add a skip for the newly revealed BUG217871. PR: 196361 Submitted by: Erick Turnquist <jhujhiti@adjectivism.org> Reported by: Jason Healy <jhealy@logn.net> Reviewed by: asomers MFC after: 3 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D9451
2017-03-17 16:50:37 +00:00
if (selected_dr != NULL)
defrouter_addreq(selected_dr);
}
Constrain IPv6 routes to single FIBs when net.add_addr_allfibs=0 sys/netinet6/icmp6.c Use the interface's FIB for source address selection in ICMPv6 error responses. sys/netinet6/in6.c In in6_newaddrmsg, announce arrival of local addresses on the interface's FIB only. In in6_lltable_rtcheck, use a per-fib ND6 cache instead of a single cache. sys/netinet6/in6_src.c In in6_selectsrc, use the caller's fib instead of the default fib. In in6_selectsrc_socket, remove a superfluous check. sys/netinet6/nd6.c In nd6_lle_event, use the interface's fib for routing socket messages. In nd6_is_new_addr_neighbor, check all FIBs when trying to determine whether an address is a neighbor. Also, simplify the code for point to point interfaces. sys/netinet6/nd6.h sys/netinet6/nd6.c sys/netinet6/nd6_rtr.c Make defrouter_select fib-aware, and make all of its callers pass in the interface fib. sys/netinet6/nd6_nbr.c When inputting a Neighbor Solicitation packet, consider the interface fib instead of the default fib for DAD. Output NS and Neighbor Advertisement packets on the correct fib. sys/netinet6/nd6_rtr.c Allow installing the same host route on different interfaces in different FIBs. If rt_add_addr_allfibs=0, only install or delete the prefix route on the interface fib. tests/sys/netinet/fibs_test.sh Clear some expected failures, but add a skip for the newly revealed BUG217871. PR: 196361 Submitted by: Erick Turnquist <jhujhiti@adjectivism.org> Reported by: Jason Healy <jhealy@logn.net> Reviewed by: asomers MFC after: 3 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D9451
2017-03-17 16:50:37 +00:00
if (selected_dr != NULL)
defrouter_rele(selected_dr);
}
/*
* Maintain old KPI for default router selection.
* If unspecified, we can re-select routers for all FIBs.
*/
void
defrouter_select(void)
{
defrouter_select_fib(RT_ALL_FIBS);
}
/*
* for default router selection
* regards router-preference field as a 2-bit signed integer
*/
static int
rtpref(struct nd_defrouter *dr)
{
switch (dr->raflags & ND_RA_FLAG_RTPREF_MASK) {
case ND_RA_FLAG_RTPREF_HIGH:
return (RTPREF_HIGH);
case ND_RA_FLAG_RTPREF_MEDIUM:
case ND_RA_FLAG_RTPREF_RSV:
return (RTPREF_MEDIUM);
case ND_RA_FLAG_RTPREF_LOW:
return (RTPREF_LOW);
default:
/*
* This case should never happen. If it did, it would mean a
* serious bug of kernel internal. We thus always bark here.
* Or, can we even panic?
*/
log(LOG_ERR, "rtpref: impossible RA flag %x\n", dr->raflags);
return (RTPREF_INVALID);
}
/* NOTREACHED */
}
static struct nd_defrouter *
defrtrlist_update(struct nd_defrouter *new)
{
struct nd_defrouter *dr, *n;
uint64_t genid;
int oldpref;
bool writelocked;
if (new->rtlifetime == 0) {
defrouter_remove(&new->rtaddr, new->ifp);
return (NULL);
}
ND6_RLOCK();
writelocked = false;
restart:
dr = defrouter_lookup_locked(&new->rtaddr, new->ifp);
if (dr != NULL) {
oldpref = rtpref(dr);
/* override */
dr->raflags = new->raflags; /* XXX flag check */
dr->rtlifetime = new->rtlifetime;
dr->expire = new->expire;
/*
* If the preference does not change, there's no need
* to sort the entries. Also make sure the selected
* router is still installed in the kernel.
*/
if (dr->installed && rtpref(new) == oldpref) {
if (writelocked)
ND6_WUNLOCK();
else
ND6_RUNLOCK();
return (dr);
}
}
/*
* The router needs to be reinserted into the default router
* list, so upgrade to a write lock. If that fails and the list
* has potentially changed while the lock was dropped, we'll
* redo the lookup with the write lock held.
*/
if (!writelocked) {
writelocked = true;
if (!ND6_TRY_UPGRADE()) {
genid = V_nd6_list_genid;
ND6_RUNLOCK();
ND6_WLOCK();
if (genid != V_nd6_list_genid)
goto restart;
}
}
if (dr != NULL) {
/*
* The preferred router may have changed, so relocate this
* router.
*/
TAILQ_REMOVE(&V_nd_defrouter, dr, dr_entry);
n = dr;
} else {
n = malloc(sizeof(*n), M_IP6NDP, M_NOWAIT | M_ZERO);
if (n == NULL) {
ND6_WUNLOCK();
return (NULL);
}
memcpy(n, new, sizeof(*n));
/* Initialize with an extra reference for the caller. */
refcount_init(&n->refcnt, 2);
}
/*
* Insert the new router in the Default Router List;
* The Default Router List should be in the descending order
* of router-preferece. Routers with the same preference are
* sorted in the arriving time order.
*/
/* insert at the end of the group */
TAILQ_FOREACH(dr, &V_nd_defrouter, dr_entry) {
if (rtpref(n) > rtpref(dr))
break;
}
if (dr != NULL)
TAILQ_INSERT_BEFORE(dr, n, dr_entry);
else
TAILQ_INSERT_TAIL(&V_nd_defrouter, n, dr_entry);
V_nd6_list_genid++;
ND6_WUNLOCK();
Constrain IPv6 routes to single FIBs when net.add_addr_allfibs=0 sys/netinet6/icmp6.c Use the interface's FIB for source address selection in ICMPv6 error responses. sys/netinet6/in6.c In in6_newaddrmsg, announce arrival of local addresses on the interface's FIB only. In in6_lltable_rtcheck, use a per-fib ND6 cache instead of a single cache. sys/netinet6/in6_src.c In in6_selectsrc, use the caller's fib instead of the default fib. In in6_selectsrc_socket, remove a superfluous check. sys/netinet6/nd6.c In nd6_lle_event, use the interface's fib for routing socket messages. In nd6_is_new_addr_neighbor, check all FIBs when trying to determine whether an address is a neighbor. Also, simplify the code for point to point interfaces. sys/netinet6/nd6.h sys/netinet6/nd6.c sys/netinet6/nd6_rtr.c Make defrouter_select fib-aware, and make all of its callers pass in the interface fib. sys/netinet6/nd6_nbr.c When inputting a Neighbor Solicitation packet, consider the interface fib instead of the default fib for DAD. Output NS and Neighbor Advertisement packets on the correct fib. sys/netinet6/nd6_rtr.c Allow installing the same host route on different interfaces in different FIBs. If rt_add_addr_allfibs=0, only install or delete the prefix route on the interface fib. tests/sys/netinet/fibs_test.sh Clear some expected failures, but add a skip for the newly revealed BUG217871. PR: 196361 Submitted by: Erick Turnquist <jhujhiti@adjectivism.org> Reported by: Jason Healy <jhealy@logn.net> Reviewed by: asomers MFC after: 3 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D9451
2017-03-17 16:50:37 +00:00
defrouter_select_fib(new->ifp->if_fib);
return (n);
}
static struct nd_pfxrouter *
pfxrtr_lookup(struct nd_prefix *pr, struct nd_defrouter *dr)
{
struct nd_pfxrouter *search;
ND6_LOCK_ASSERT();
LIST_FOREACH(search, &pr->ndpr_advrtrs, pfr_entry) {
if (search->router == dr)
break;
}
return (search);
}
static void
pfxrtr_add(struct nd_prefix *pr, struct nd_defrouter *dr)
{
struct nd_pfxrouter *new;
bool update;
ND6_UNLOCK_ASSERT();
ND6_RLOCK();
if (pfxrtr_lookup(pr, dr) != NULL) {
ND6_RUNLOCK();
return;
}
ND6_RUNLOCK();
new = malloc(sizeof(*new), M_IP6NDP, M_NOWAIT | M_ZERO);
if (new == NULL)
return;
defrouter_ref(dr);
new->router = dr;
ND6_WLOCK();
if (pfxrtr_lookup(pr, dr) == NULL) {
LIST_INSERT_HEAD(&pr->ndpr_advrtrs, new, pfr_entry);
update = true;
} else {
/* We lost a race to add the reference. */
defrouter_rele(dr);
free(new, M_IP6NDP);
update = false;
}
ND6_WUNLOCK();
if (update)
pfxlist_onlink_check();
}
static void
pfxrtr_del(struct nd_pfxrouter *pfr)
{
ND6_WLOCK_ASSERT();
LIST_REMOVE(pfr, pfr_entry);
defrouter_rele(pfr->router);
free(pfr, M_IP6NDP);
}
static struct nd_prefix *
nd6_prefix_lookup_locked(struct nd_prefixctl *key)
{
struct nd_prefix *search;
ND6_LOCK_ASSERT();
LIST_FOREACH(search, &V_nd_prefix, ndpr_entry) {
if (key->ndpr_ifp == search->ndpr_ifp &&
key->ndpr_plen == search->ndpr_plen &&
in6_are_prefix_equal(&key->ndpr_prefix.sin6_addr,
&search->ndpr_prefix.sin6_addr, key->ndpr_plen)) {
nd6_prefix_ref(search);
break;
}
}
return (search);
}
struct nd_prefix *
nd6_prefix_lookup(struct nd_prefixctl *key)
{
struct nd_prefix *search;
ND6_RLOCK();
search = nd6_prefix_lookup_locked(key);
ND6_RUNLOCK();
return (search);
}
void
nd6_prefix_ref(struct nd_prefix *pr)
{
refcount_acquire(&pr->ndpr_refcnt);
}
void
nd6_prefix_rele(struct nd_prefix *pr)
{
if (refcount_release(&pr->ndpr_refcnt)) {
KASSERT(LIST_EMPTY(&pr->ndpr_advrtrs),
("prefix %p has advertising routers", pr));
free(pr, M_IP6NDP);
}
}
int
nd6_prelist_add(struct nd_prefixctl *pr, struct nd_defrouter *dr,
struct nd_prefix **newp)
{
struct nd_prefix *new;
char ip6buf[INET6_ADDRSTRLEN];
int error;
new = malloc(sizeof(*new), M_IP6NDP, M_NOWAIT | M_ZERO);
if (new == NULL)
return (ENOMEM);
refcount_init(&new->ndpr_refcnt, newp != NULL ? 2 : 1);
new->ndpr_ifp = pr->ndpr_ifp;
new->ndpr_prefix = pr->ndpr_prefix;
new->ndpr_plen = pr->ndpr_plen;
new->ndpr_vltime = pr->ndpr_vltime;
new->ndpr_pltime = pr->ndpr_pltime;
new->ndpr_flags = pr->ndpr_flags;
if ((error = in6_init_prefix_ltimes(new)) != 0) {
free(new, M_IP6NDP);
return (error);
}
new->ndpr_lastupdate = time_uptime;
/* initialization */
LIST_INIT(&new->ndpr_advrtrs);
in6_prefixlen2mask(&new->ndpr_mask, new->ndpr_plen);
/* make prefix in the canonical form */
IN6_MASK_ADDR(&new->ndpr_prefix.sin6_addr, &new->ndpr_mask);
ND6_WLOCK();
LIST_INSERT_HEAD(&V_nd_prefix, new, ndpr_entry);
V_nd6_list_genid++;
ND6_WUNLOCK();
/* ND_OPT_PI_FLAG_ONLINK processing */
if (new->ndpr_raf_onlink) {
ND6_ONLINK_LOCK();
if ((error = nd6_prefix_onlink(new)) != 0) {
nd6log((LOG_ERR, "nd6_prelist_add: failed to make "
"the prefix %s/%d on-link on %s (errno=%d)\n",
ip6_sprintf(ip6buf, &pr->ndpr_prefix.sin6_addr),
pr->ndpr_plen, if_name(pr->ndpr_ifp), error));
/* proceed anyway. XXX: is it correct? */
}
ND6_ONLINK_UNLOCK();
}
if (dr != NULL)
pfxrtr_add(new, dr);
if (newp != NULL)
*newp = new;
return (0);
}
/*
* Remove a prefix from the prefix list and optionally stash it in a
* caller-provided list.
*
* The ND6 lock must be held.
*/
void
nd6_prefix_unlink(struct nd_prefix *pr, struct nd_prhead *list)
{
ND6_WLOCK_ASSERT();
LIST_REMOVE(pr, ndpr_entry);
V_nd6_list_genid++;
if (list != NULL)
LIST_INSERT_HEAD(list, pr, ndpr_entry);
}
/*
* Free an unlinked prefix, first marking it off-link if necessary.
*/
void
nd6_prefix_del(struct nd_prefix *pr)
{
struct nd_pfxrouter *pfr, *next;
int e;
char ip6buf[INET6_ADDRSTRLEN];
KASSERT(pr->ndpr_addrcnt == 0,
("prefix %p has referencing addresses", pr));
ND6_UNLOCK_ASSERT();
/*
* Though these flags are now meaningless, we'd rather keep the value
* of pr->ndpr_raf_onlink and pr->ndpr_raf_auto not to confuse users
* when executing "ndp -p".
*/
if ((pr->ndpr_stateflags & NDPRF_ONLINK) != 0) {
ND6_ONLINK_LOCK();
if ((e = nd6_prefix_offlink(pr)) != 0) {
nd6log((LOG_ERR,
"nd6_prefix_del: failed to make %s/%d offlink "
"on %s, errno=%d\n",
ip6_sprintf(ip6buf, &pr->ndpr_prefix.sin6_addr),
pr->ndpr_plen, if_name(pr->ndpr_ifp), e));
/* what should we do? */
}
ND6_ONLINK_UNLOCK();
}
/* Release references to routers that have advertised this prefix. */
ND6_WLOCK();
LIST_FOREACH_SAFE(pfr, &pr->ndpr_advrtrs, pfr_entry, next)
pfxrtr_del(pfr);
ND6_WUNLOCK();
nd6_prefix_rele(pr);
pfxlist_onlink_check();
}
static int
prelist_update(struct nd_prefixctl *new, struct nd_defrouter *dr,
struct mbuf *m, int mcast)
{
struct in6_ifaddr *ia6 = NULL, *ia6_match = NULL;
struct ifaddr *ifa;
struct ifnet *ifp = new->ndpr_ifp;
struct nd_prefix *pr;
int error = 0;
int auth;
struct in6_addrlifetime lt6_tmp;
char ip6buf[INET6_ADDRSTRLEN];
auth = 0;
if (m) {
/*
* Authenticity for NA consists authentication for
* both IP header and IP datagrams, doesn't it ?
*/
#if defined(M_AUTHIPHDR) && defined(M_AUTHIPDGM)
auth = ((m->m_flags & M_AUTHIPHDR) &&
(m->m_flags & M_AUTHIPDGM));
#endif
}
if ((pr = nd6_prefix_lookup(new)) != NULL) {
/*
* nd6_prefix_lookup() ensures that pr and new have the same
* prefix on a same interface.
*/
/*
* Update prefix information. Note that the on-link (L) bit
* and the autonomous (A) bit should NOT be changed from 1
* to 0.
*/
if (new->ndpr_raf_onlink == 1)
pr->ndpr_raf_onlink = 1;
if (new->ndpr_raf_auto == 1)
pr->ndpr_raf_auto = 1;
if (new->ndpr_raf_onlink) {
pr->ndpr_vltime = new->ndpr_vltime;
pr->ndpr_pltime = new->ndpr_pltime;
(void)in6_init_prefix_ltimes(pr); /* XXX error case? */
pr->ndpr_lastupdate = time_uptime;
}
if (new->ndpr_raf_onlink &&
(pr->ndpr_stateflags & NDPRF_ONLINK) == 0) {
ND6_ONLINK_LOCK();
if ((error = nd6_prefix_onlink(pr)) != 0) {
nd6log((LOG_ERR,
"prelist_update: failed to make "
"the prefix %s/%d on-link on %s "
"(errno=%d)\n",
ip6_sprintf(ip6buf,
&pr->ndpr_prefix.sin6_addr),
pr->ndpr_plen, if_name(pr->ndpr_ifp),
error));
/* proceed anyway. XXX: is it correct? */
}
ND6_ONLINK_UNLOCK();
}
if (dr != NULL)
pfxrtr_add(pr, dr);
} else {
if (new->ndpr_vltime == 0)
goto end;
if (new->ndpr_raf_onlink == 0 && new->ndpr_raf_auto == 0)
goto end;
error = nd6_prelist_add(new, dr, &pr);
if (error != 0) {
nd6log((LOG_NOTICE, "prelist_update: "
"nd6_prelist_add failed for %s/%d on %s errno=%d\n",
ip6_sprintf(ip6buf, &new->ndpr_prefix.sin6_addr),
new->ndpr_plen, if_name(new->ndpr_ifp), error));
goto end; /* we should just give up in this case. */
}
/*
* XXX: from the ND point of view, we can ignore a prefix
* with the on-link bit being zero. However, we need a
* prefix structure for references from autoconfigured
* addresses. Thus, we explicitly make sure that the prefix
* itself expires now.
*/
if (pr->ndpr_raf_onlink == 0) {
pr->ndpr_vltime = 0;
pr->ndpr_pltime = 0;
in6_init_prefix_ltimes(pr);
}
}
/*
* Address autoconfiguration based on Section 5.5.3 of RFC 2462.
* Note that pr must be non NULL at this point.
*/
/* 5.5.3 (a). Ignore the prefix without the A bit set. */
if (!new->ndpr_raf_auto)
goto end;
/*
* 5.5.3 (b). the link-local prefix should have been ignored in
* nd6_ra_input.
*/
/* 5.5.3 (c). Consistency check on lifetimes: pltime <= vltime. */
if (new->ndpr_pltime > new->ndpr_vltime) {
error = EINVAL; /* XXX: won't be used */
goto end;
}
/*
* 5.5.3 (d). If the prefix advertised is not equal to the prefix of
* an address configured by stateless autoconfiguration already in the
* list of addresses associated with the interface, and the Valid
* Lifetime is not 0, form an address. We first check if we have
* a matching prefix.
* Note: we apply a clarification in rfc2462bis-02 here. We only
* consider autoconfigured addresses while RFC2462 simply said
* "address".
*/
IF_ADDR_RLOCK(ifp);
ifnet: Replace if_addr_lock rwlock with epoch + mutex Run on LLNW canaries and tested by pho@ gallatin: Using a 14-core, 28-HTT single socket E5-2697 v3 with a 40GbE MLX5 based ConnectX 4-LX NIC, I see an almost 12% improvement in received packet rate, and a larger improvement in bytes delivered all the way to userspace. When the host receiving 64 streams of netperf -H $DUT -t UDP_STREAM -- -m 1, I see, using nstat -I mce0 1 before the patch: InMpps OMpps InGbs OGbs err TCP Est %CPU syscalls csw irq GBfree 4.98 0.00 4.42 0.00 4235592 33 83.80 4720653 2149771 1235 247.32 4.73 0.00 4.20 0.00 4025260 33 82.99 4724900 2139833 1204 247.32 4.72 0.00 4.20 0.00 4035252 33 82.14 4719162 2132023 1264 247.32 4.71 0.00 4.21 0.00 4073206 33 83.68 4744973 2123317 1347 247.32 4.72 0.00 4.21 0.00 4061118 33 80.82 4713615 2188091 1490 247.32 4.72 0.00 4.21 0.00 4051675 33 85.29 4727399 2109011 1205 247.32 4.73 0.00 4.21 0.00 4039056 33 84.65 4724735 2102603 1053 247.32 After the patch InMpps OMpps InGbs OGbs err TCP Est %CPU syscalls csw irq GBfree 5.43 0.00 4.20 0.00 3313143 33 84.96 5434214 1900162 2656 245.51 5.43 0.00 4.20 0.00 3308527 33 85.24 5439695 1809382 2521 245.51 5.42 0.00 4.19 0.00 3316778 33 87.54 5416028 1805835 2256 245.51 5.42 0.00 4.19 0.00 3317673 33 90.44 5426044 1763056 2332 245.51 5.42 0.00 4.19 0.00 3314839 33 88.11 5435732 1792218 2499 245.52 5.44 0.00 4.19 0.00 3293228 33 91.84 5426301 1668597 2121 245.52 Similarly, netperf reports 230Mb/s before the patch, and 270Mb/s after the patch Reviewed by: gallatin Sponsored by: Limelight Networks Differential Revision: https://reviews.freebsd.org/D15366
2018-05-18 20:13:34 +00:00
CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
struct in6_ifaddr *ifa6;
u_int32_t remaininglifetime;
if (ifa->ifa_addr->sa_family != AF_INET6)
continue;
ifa6 = (struct in6_ifaddr *)ifa;
/*
* We only consider autoconfigured addresses as per rfc2462bis.
*/
if (!(ifa6->ia6_flags & IN6_IFF_AUTOCONF))
continue;
/*
* Spec is not clear here, but I believe we should concentrate
* on unicast (i.e. not anycast) addresses.
* XXX: other ia6_flags? detached or duplicated?
*/
if ((ifa6->ia6_flags & IN6_IFF_ANYCAST) != 0)
continue;
/*
* Ignore the address if it is not associated with a prefix
* or is associated with a prefix that is different from this
* one. (pr is never NULL here)
*/
if (ifa6->ia6_ndpr != pr)
continue;
if (ia6_match == NULL) /* remember the first one */
ia6_match = ifa6;
/*
* An already autoconfigured address matched. Now that we
* are sure there is at least one matched address, we can
* proceed to 5.5.3. (e): update the lifetimes according to the
* "two hours" rule and the privacy extension.
* We apply some clarifications in rfc2462bis:
* - use remaininglifetime instead of storedlifetime as a
* variable name
* - remove the dead code in the "two-hour" rule
*/
#define TWOHOUR (120*60)
lt6_tmp = ifa6->ia6_lifetime;
if (lt6_tmp.ia6t_vltime == ND6_INFINITE_LIFETIME)
remaininglifetime = ND6_INFINITE_LIFETIME;
else if (time_uptime - ifa6->ia6_updatetime >
lt6_tmp.ia6t_vltime) {
/*
* The case of "invalid" address. We should usually
* not see this case.
*/
remaininglifetime = 0;
} else
remaininglifetime = lt6_tmp.ia6t_vltime -
(time_uptime - ifa6->ia6_updatetime);
/* when not updating, keep the current stored lifetime. */
lt6_tmp.ia6t_vltime = remaininglifetime;
if (TWOHOUR < new->ndpr_vltime ||
remaininglifetime < new->ndpr_vltime) {
lt6_tmp.ia6t_vltime = new->ndpr_vltime;
} else if (remaininglifetime <= TWOHOUR) {
if (auth) {
lt6_tmp.ia6t_vltime = new->ndpr_vltime;
}
} else {
/*
* new->ndpr_vltime <= TWOHOUR &&
* TWOHOUR < remaininglifetime
*/
lt6_tmp.ia6t_vltime = TWOHOUR;
}
/* The 2 hour rule is not imposed for preferred lifetime. */
lt6_tmp.ia6t_pltime = new->ndpr_pltime;
in6_init_address_ltimes(pr, &lt6_tmp);
/*
* We need to treat lifetimes for temporary addresses
* differently, according to
* draft-ietf-ipv6-privacy-addrs-v2-01.txt 3.3 (1);
* we only update the lifetimes when they are in the maximum
* intervals.
*/
if ((ifa6->ia6_flags & IN6_IFF_TEMPORARY) != 0) {
u_int32_t maxvltime, maxpltime;
if (V_ip6_temp_valid_lifetime >
(u_int32_t)((time_uptime - ifa6->ia6_createtime) +
V_ip6_desync_factor)) {
maxvltime = V_ip6_temp_valid_lifetime -
(time_uptime - ifa6->ia6_createtime) -
V_ip6_desync_factor;
} else
maxvltime = 0;
if (V_ip6_temp_preferred_lifetime >
(u_int32_t)((time_uptime - ifa6->ia6_createtime) +
V_ip6_desync_factor)) {
maxpltime = V_ip6_temp_preferred_lifetime -
(time_uptime - ifa6->ia6_createtime) -
V_ip6_desync_factor;
} else
maxpltime = 0;
if (lt6_tmp.ia6t_vltime == ND6_INFINITE_LIFETIME ||
lt6_tmp.ia6t_vltime > maxvltime) {
lt6_tmp.ia6t_vltime = maxvltime;
}
if (lt6_tmp.ia6t_pltime == ND6_INFINITE_LIFETIME ||
lt6_tmp.ia6t_pltime > maxpltime) {
lt6_tmp.ia6t_pltime = maxpltime;
}
}
ifa6->ia6_lifetime = lt6_tmp;
ifa6->ia6_updatetime = time_uptime;
}
IF_ADDR_RUNLOCK(ifp);
if (ia6_match == NULL && new->ndpr_vltime) {
int ifidlen;
/*
* 5.5.3 (d) (continued)
* No address matched and the valid lifetime is non-zero.
* Create a new address.
*/
/*
* Prefix Length check:
* If the sum of the prefix length and interface identifier
* length does not equal 128 bits, the Prefix Information
* option MUST be ignored. The length of the interface
* identifier is defined in a separate link-type specific
* document.
*/
ifidlen = in6_if2idlen(ifp);
if (ifidlen < 0) {
/* this should not happen, so we always log it. */
log(LOG_ERR, "prelist_update: IFID undefined (%s)\n",
if_name(ifp));
goto end;
}
if (ifidlen + pr->ndpr_plen != 128) {
nd6log((LOG_INFO,
"prelist_update: invalid prefixlen "
"%d for %s, ignored\n",
pr->ndpr_plen, if_name(ifp)));
goto end;
}
if ((ia6 = in6_ifadd(new, mcast)) != NULL) {
/*
* note that we should use pr (not new) for reference.
*/
pr->ndpr_addrcnt++;
ia6->ia6_ndpr = pr;
/*
* RFC 3041 3.3 (2).
* When a new public address is created as described
* in RFC2462, also create a new temporary address.
*
* RFC 3041 3.5.
* When an interface connects to a new link, a new
* randomized interface identifier should be generated
* immediately together with a new set of temporary
* addresses. Thus, we specifiy 1 as the 2nd arg of
* in6_tmpifadd().
*/
if (V_ip6_use_tempaddr) {
int e;
if ((e = in6_tmpifadd(ia6, 1, 1)) != 0) {
nd6log((LOG_NOTICE, "prelist_update: "
"failed to create a temporary "
"address, errno=%d\n",
e));
}
}
ifa_free(&ia6->ia_ifa);
/*
* A newly added address might affect the status
* of other addresses, so we check and update it.
* XXX: what if address duplication happens?
*/
pfxlist_onlink_check();
} else {
/* just set an error. do not bark here. */
error = EADDRNOTAVAIL; /* XXX: might be unused. */
}
}
end:
if (pr != NULL)
nd6_prefix_rele(pr);
return (error);
}
/*
* A supplement function used in the on-link detection below;
* detect if a given prefix has a (probably) reachable advertising router.
* XXX: lengthy function name...
*/
static struct nd_pfxrouter *
find_pfxlist_reachable_router(struct nd_prefix *pr)
{
struct nd_pfxrouter *pfxrtr;
struct llentry *ln;
int canreach;
ND6_LOCK_ASSERT();
LIST_FOREACH(pfxrtr, &pr->ndpr_advrtrs, pfr_entry) {
IF_AFDATA_RLOCK(pfxrtr->router->ifp);
ln = nd6_lookup(&pfxrtr->router->rtaddr, 0, pfxrtr->router->ifp);
IF_AFDATA_RUNLOCK(pfxrtr->router->ifp);
if (ln == NULL)
continue;
canreach = ND6_IS_LLINFO_PROBREACH(ln);
LLE_RUNLOCK(ln);
if (canreach)
break;
}
return (pfxrtr);
}
/*
* Check if each prefix in the prefix list has at least one available router
* that advertised the prefix (a router is "available" if its neighbor cache
* entry is reachable or probably reachable).
* If the check fails, the prefix may be off-link, because, for example,
* we have moved from the network but the lifetime of the prefix has not
* expired yet. So we should not use the prefix if there is another prefix
* that has an available router.
2016-10-07 00:35:28 +00:00
* But, if there is no prefix that has an available router, we still regard
* all the prefixes as on-link. This is because we can't tell if all the
* routers are simply dead or if we really moved from the network and there
* is no router around us.
*/
void
pfxlist_onlink_check(void)
{
struct nd_prefix *pr;
struct in6_ifaddr *ifa;
struct nd_defrouter *dr;
struct nd_pfxrouter *pfxrtr = NULL;
struct rm_priotracker in6_ifa_tracker;
uint64_t genid;
uint32_t flags;
ND6_ONLINK_LOCK();
ND6_RLOCK();
/*
* Check if there is a prefix that has a reachable advertising
* router.
*/
LIST_FOREACH(pr, &V_nd_prefix, ndpr_entry) {
if (pr->ndpr_raf_onlink && find_pfxlist_reachable_router(pr))
break;
}
/*
* If we have no such prefix, check whether we still have a router
* that does not advertise any prefixes.
*/
if (pr == NULL) {
TAILQ_FOREACH(dr, &V_nd_defrouter, dr_entry) {
struct nd_prefix *pr0;
LIST_FOREACH(pr0, &V_nd_prefix, ndpr_entry) {
if ((pfxrtr = pfxrtr_lookup(pr0, dr)) != NULL)
break;
}
if (pfxrtr != NULL)
break;
}
}
if (pr != NULL || (!TAILQ_EMPTY(&V_nd_defrouter) && pfxrtr == NULL)) {
/*
* There is at least one prefix that has a reachable router,
* or at least a router which probably does not advertise
* any prefixes. The latter would be the case when we move
* to a new link where we have a router that does not provide
* prefixes and we configure an address by hand.
* Detach prefixes which have no reachable advertising
* router, and attach other prefixes.
*/
LIST_FOREACH(pr, &V_nd_prefix, ndpr_entry) {
/* XXX: a link-local prefix should never be detached */
if (IN6_IS_ADDR_LINKLOCAL(&pr->ndpr_prefix.sin6_addr) ||
pr->ndpr_raf_onlink == 0 ||
pr->ndpr_raf_auto == 0)
continue;
if ((pr->ndpr_stateflags & NDPRF_DETACHED) == 0 &&
find_pfxlist_reachable_router(pr) == NULL)
pr->ndpr_stateflags |= NDPRF_DETACHED;
else if ((pr->ndpr_stateflags & NDPRF_DETACHED) != 0 &&
find_pfxlist_reachable_router(pr) != NULL)
pr->ndpr_stateflags &= ~NDPRF_DETACHED;
}
} else {
/* there is no prefix that has a reachable router */
LIST_FOREACH(pr, &V_nd_prefix, ndpr_entry) {
if (IN6_IS_ADDR_LINKLOCAL(&pr->ndpr_prefix.sin6_addr) ||
pr->ndpr_raf_onlink == 0 ||
pr->ndpr_raf_auto == 0)
continue;
pr->ndpr_stateflags &= ~NDPRF_DETACHED;
}
}
/*
* Remove each interface route associated with a (just) detached
* prefix, and reinstall the interface route for a (just) attached
* prefix. Note that all attempt of reinstallation does not
* necessarily success, when a same prefix is shared among multiple
* interfaces. Such cases will be handled in nd6_prefix_onlink,
* so we don't have to care about them.
*/
restart:
LIST_FOREACH(pr, &V_nd_prefix, ndpr_entry) {
char ip6buf[INET6_ADDRSTRLEN];
int e;
if (IN6_IS_ADDR_LINKLOCAL(&pr->ndpr_prefix.sin6_addr) ||
pr->ndpr_raf_onlink == 0 ||
pr->ndpr_raf_auto == 0)
continue;
flags = pr->ndpr_stateflags & (NDPRF_DETACHED | NDPRF_ONLINK);
if (flags == 0 || flags == (NDPRF_DETACHED | NDPRF_ONLINK)) {
genid = V_nd6_list_genid;
ND6_RUNLOCK();
if ((flags & NDPRF_ONLINK) != 0 &&
(e = nd6_prefix_offlink(pr)) != 0) {
nd6log((LOG_ERR,
"pfxlist_onlink_check: failed to "
"make %s/%d offlink, errno=%d\n",
ip6_sprintf(ip6buf,
&pr->ndpr_prefix.sin6_addr),
pr->ndpr_plen, e));
} else if ((flags & NDPRF_ONLINK) == 0 &&
(e = nd6_prefix_onlink(pr)) != 0) {
nd6log((LOG_ERR,
"pfxlist_onlink_check: failed to "
"make %s/%d onlink, errno=%d\n",
ip6_sprintf(ip6buf,
&pr->ndpr_prefix.sin6_addr),
pr->ndpr_plen, e));
}
ND6_RLOCK();
if (genid != V_nd6_list_genid)
goto restart;
}
}
/*
* Changes on the prefix status might affect address status as well.
* Make sure that all addresses derived from an attached prefix are
* attached, and that all addresses derived from a detached prefix are
* detached. Note, however, that a manually configured address should
* always be attached.
* The precise detection logic is same as the one for prefixes.
*/
IN6_IFADDR_RLOCK(&in6_ifa_tracker);
ifnet: Replace if_addr_lock rwlock with epoch + mutex Run on LLNW canaries and tested by pho@ gallatin: Using a 14-core, 28-HTT single socket E5-2697 v3 with a 40GbE MLX5 based ConnectX 4-LX NIC, I see an almost 12% improvement in received packet rate, and a larger improvement in bytes delivered all the way to userspace. When the host receiving 64 streams of netperf -H $DUT -t UDP_STREAM -- -m 1, I see, using nstat -I mce0 1 before the patch: InMpps OMpps InGbs OGbs err TCP Est %CPU syscalls csw irq GBfree 4.98 0.00 4.42 0.00 4235592 33 83.80 4720653 2149771 1235 247.32 4.73 0.00 4.20 0.00 4025260 33 82.99 4724900 2139833 1204 247.32 4.72 0.00 4.20 0.00 4035252 33 82.14 4719162 2132023 1264 247.32 4.71 0.00 4.21 0.00 4073206 33 83.68 4744973 2123317 1347 247.32 4.72 0.00 4.21 0.00 4061118 33 80.82 4713615 2188091 1490 247.32 4.72 0.00 4.21 0.00 4051675 33 85.29 4727399 2109011 1205 247.32 4.73 0.00 4.21 0.00 4039056 33 84.65 4724735 2102603 1053 247.32 After the patch InMpps OMpps InGbs OGbs err TCP Est %CPU syscalls csw irq GBfree 5.43 0.00 4.20 0.00 3313143 33 84.96 5434214 1900162 2656 245.51 5.43 0.00 4.20 0.00 3308527 33 85.24 5439695 1809382 2521 245.51 5.42 0.00 4.19 0.00 3316778 33 87.54 5416028 1805835 2256 245.51 5.42 0.00 4.19 0.00 3317673 33 90.44 5426044 1763056 2332 245.51 5.42 0.00 4.19 0.00 3314839 33 88.11 5435732 1792218 2499 245.52 5.44 0.00 4.19 0.00 3293228 33 91.84 5426301 1668597 2121 245.52 Similarly, netperf reports 230Mb/s before the patch, and 270Mb/s after the patch Reviewed by: gallatin Sponsored by: Limelight Networks Differential Revision: https://reviews.freebsd.org/D15366
2018-05-18 20:13:34 +00:00
CK_STAILQ_FOREACH(ifa, &V_in6_ifaddrhead, ia_link) {
if (!(ifa->ia6_flags & IN6_IFF_AUTOCONF))
continue;
if (ifa->ia6_ndpr == NULL) {
/*
* This can happen when we first configure the address
* (i.e. the address exists, but the prefix does not).
* XXX: complicated relationships...
*/
continue;
}
if (find_pfxlist_reachable_router(ifa->ia6_ndpr))
break;
}
if (ifa) {
ifnet: Replace if_addr_lock rwlock with epoch + mutex Run on LLNW canaries and tested by pho@ gallatin: Using a 14-core, 28-HTT single socket E5-2697 v3 with a 40GbE MLX5 based ConnectX 4-LX NIC, I see an almost 12% improvement in received packet rate, and a larger improvement in bytes delivered all the way to userspace. When the host receiving 64 streams of netperf -H $DUT -t UDP_STREAM -- -m 1, I see, using nstat -I mce0 1 before the patch: InMpps OMpps InGbs OGbs err TCP Est %CPU syscalls csw irq GBfree 4.98 0.00 4.42 0.00 4235592 33 83.80 4720653 2149771 1235 247.32 4.73 0.00 4.20 0.00 4025260 33 82.99 4724900 2139833 1204 247.32 4.72 0.00 4.20 0.00 4035252 33 82.14 4719162 2132023 1264 247.32 4.71 0.00 4.21 0.00 4073206 33 83.68 4744973 2123317 1347 247.32 4.72 0.00 4.21 0.00 4061118 33 80.82 4713615 2188091 1490 247.32 4.72 0.00 4.21 0.00 4051675 33 85.29 4727399 2109011 1205 247.32 4.73 0.00 4.21 0.00 4039056 33 84.65 4724735 2102603 1053 247.32 After the patch InMpps OMpps InGbs OGbs err TCP Est %CPU syscalls csw irq GBfree 5.43 0.00 4.20 0.00 3313143 33 84.96 5434214 1900162 2656 245.51 5.43 0.00 4.20 0.00 3308527 33 85.24 5439695 1809382 2521 245.51 5.42 0.00 4.19 0.00 3316778 33 87.54 5416028 1805835 2256 245.51 5.42 0.00 4.19 0.00 3317673 33 90.44 5426044 1763056 2332 245.51 5.42 0.00 4.19 0.00 3314839 33 88.11 5435732 1792218 2499 245.52 5.44 0.00 4.19 0.00 3293228 33 91.84 5426301 1668597 2121 245.52 Similarly, netperf reports 230Mb/s before the patch, and 270Mb/s after the patch Reviewed by: gallatin Sponsored by: Limelight Networks Differential Revision: https://reviews.freebsd.org/D15366
2018-05-18 20:13:34 +00:00
CK_STAILQ_FOREACH(ifa, &V_in6_ifaddrhead, ia_link) {
if ((ifa->ia6_flags & IN6_IFF_AUTOCONF) == 0)
continue;
if (ifa->ia6_ndpr == NULL) /* XXX: see above. */
continue;
if (find_pfxlist_reachable_router(ifa->ia6_ndpr)) {
if (ifa->ia6_flags & IN6_IFF_DETACHED) {
ifa->ia6_flags &= ~IN6_IFF_DETACHED;
ifa->ia6_flags |= IN6_IFF_TENTATIVE;
nd6_dad_start((struct ifaddr *)ifa, 0);
}
} else {
ifa->ia6_flags |= IN6_IFF_DETACHED;
}
}
} else {
ifnet: Replace if_addr_lock rwlock with epoch + mutex Run on LLNW canaries and tested by pho@ gallatin: Using a 14-core, 28-HTT single socket E5-2697 v3 with a 40GbE MLX5 based ConnectX 4-LX NIC, I see an almost 12% improvement in received packet rate, and a larger improvement in bytes delivered all the way to userspace. When the host receiving 64 streams of netperf -H $DUT -t UDP_STREAM -- -m 1, I see, using nstat -I mce0 1 before the patch: InMpps OMpps InGbs OGbs err TCP Est %CPU syscalls csw irq GBfree 4.98 0.00 4.42 0.00 4235592 33 83.80 4720653 2149771 1235 247.32 4.73 0.00 4.20 0.00 4025260 33 82.99 4724900 2139833 1204 247.32 4.72 0.00 4.20 0.00 4035252 33 82.14 4719162 2132023 1264 247.32 4.71 0.00 4.21 0.00 4073206 33 83.68 4744973 2123317 1347 247.32 4.72 0.00 4.21 0.00 4061118 33 80.82 4713615 2188091 1490 247.32 4.72 0.00 4.21 0.00 4051675 33 85.29 4727399 2109011 1205 247.32 4.73 0.00 4.21 0.00 4039056 33 84.65 4724735 2102603 1053 247.32 After the patch InMpps OMpps InGbs OGbs err TCP Est %CPU syscalls csw irq GBfree 5.43 0.00 4.20 0.00 3313143 33 84.96 5434214 1900162 2656 245.51 5.43 0.00 4.20 0.00 3308527 33 85.24 5439695 1809382 2521 245.51 5.42 0.00 4.19 0.00 3316778 33 87.54 5416028 1805835 2256 245.51 5.42 0.00 4.19 0.00 3317673 33 90.44 5426044 1763056 2332 245.51 5.42 0.00 4.19 0.00 3314839 33 88.11 5435732 1792218 2499 245.52 5.44 0.00 4.19 0.00 3293228 33 91.84 5426301 1668597 2121 245.52 Similarly, netperf reports 230Mb/s before the patch, and 270Mb/s after the patch Reviewed by: gallatin Sponsored by: Limelight Networks Differential Revision: https://reviews.freebsd.org/D15366
2018-05-18 20:13:34 +00:00
CK_STAILQ_FOREACH(ifa, &V_in6_ifaddrhead, ia_link) {
if ((ifa->ia6_flags & IN6_IFF_AUTOCONF) == 0)
continue;
if (ifa->ia6_flags & IN6_IFF_DETACHED) {
ifa->ia6_flags &= ~IN6_IFF_DETACHED;
ifa->ia6_flags |= IN6_IFF_TENTATIVE;
/* Do we need a delay in this case? */
nd6_dad_start((struct ifaddr *)ifa, 0);
}
}
}
IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
ND6_RUNLOCK();
ND6_ONLINK_UNLOCK();
}
static int
nd6_prefix_onlink_rtrequest(struct nd_prefix *pr, struct ifaddr *ifa)
{
static struct sockaddr_dl null_sdl = {sizeof(null_sdl), AF_LINK};
struct rib_head *rnh;
struct rtentry *rt;
struct sockaddr_in6 mask6;
u_long rtflags;
Constrain IPv6 routes to single FIBs when net.add_addr_allfibs=0 sys/netinet6/icmp6.c Use the interface's FIB for source address selection in ICMPv6 error responses. sys/netinet6/in6.c In in6_newaddrmsg, announce arrival of local addresses on the interface's FIB only. In in6_lltable_rtcheck, use a per-fib ND6 cache instead of a single cache. sys/netinet6/in6_src.c In in6_selectsrc, use the caller's fib instead of the default fib. In in6_selectsrc_socket, remove a superfluous check. sys/netinet6/nd6.c In nd6_lle_event, use the interface's fib for routing socket messages. In nd6_is_new_addr_neighbor, check all FIBs when trying to determine whether an address is a neighbor. Also, simplify the code for point to point interfaces. sys/netinet6/nd6.h sys/netinet6/nd6.c sys/netinet6/nd6_rtr.c Make defrouter_select fib-aware, and make all of its callers pass in the interface fib. sys/netinet6/nd6_nbr.c When inputting a Neighbor Solicitation packet, consider the interface fib instead of the default fib for DAD. Output NS and Neighbor Advertisement packets on the correct fib. sys/netinet6/nd6_rtr.c Allow installing the same host route on different interfaces in different FIBs. If rt_add_addr_allfibs=0, only install or delete the prefix route on the interface fib. tests/sys/netinet/fibs_test.sh Clear some expected failures, but add a skip for the newly revealed BUG217871. PR: 196361 Submitted by: Erick Turnquist <jhujhiti@adjectivism.org> Reported by: Jason Healy <jhealy@logn.net> Reviewed by: asomers MFC after: 3 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D9451
2017-03-17 16:50:37 +00:00
int error, a_failure, fibnum, maxfib;
/*
* in6_ifinit() sets nd6_rtrequest to ifa_rtrequest for all ifaddrs.
* ifa->ifa_rtrequest = nd6_rtrequest;
*/
bzero(&mask6, sizeof(mask6));
mask6.sin6_len = sizeof(mask6);
mask6.sin6_addr = pr->ndpr_mask;
rtflags = (ifa->ifa_flags & ~IFA_RTSELF) | RTF_UP;
Constrain IPv6 routes to single FIBs when net.add_addr_allfibs=0 sys/netinet6/icmp6.c Use the interface's FIB for source address selection in ICMPv6 error responses. sys/netinet6/in6.c In in6_newaddrmsg, announce arrival of local addresses on the interface's FIB only. In in6_lltable_rtcheck, use a per-fib ND6 cache instead of a single cache. sys/netinet6/in6_src.c In in6_selectsrc, use the caller's fib instead of the default fib. In in6_selectsrc_socket, remove a superfluous check. sys/netinet6/nd6.c In nd6_lle_event, use the interface's fib for routing socket messages. In nd6_is_new_addr_neighbor, check all FIBs when trying to determine whether an address is a neighbor. Also, simplify the code for point to point interfaces. sys/netinet6/nd6.h sys/netinet6/nd6.c sys/netinet6/nd6_rtr.c Make defrouter_select fib-aware, and make all of its callers pass in the interface fib. sys/netinet6/nd6_nbr.c When inputting a Neighbor Solicitation packet, consider the interface fib instead of the default fib for DAD. Output NS and Neighbor Advertisement packets on the correct fib. sys/netinet6/nd6_rtr.c Allow installing the same host route on different interfaces in different FIBs. If rt_add_addr_allfibs=0, only install or delete the prefix route on the interface fib. tests/sys/netinet/fibs_test.sh Clear some expected failures, but add a skip for the newly revealed BUG217871. PR: 196361 Submitted by: Erick Turnquist <jhujhiti@adjectivism.org> Reported by: Jason Healy <jhealy@logn.net> Reviewed by: asomers MFC after: 3 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D9451
2017-03-17 16:50:37 +00:00
if(V_rt_add_addr_allfibs) {
fibnum = 0;
maxfib = rt_numfibs;
} else {
fibnum = ifa->ifa_ifp->if_fib;
maxfib = fibnum + 1;
}
a_failure = 0;
Constrain IPv6 routes to single FIBs when net.add_addr_allfibs=0 sys/netinet6/icmp6.c Use the interface's FIB for source address selection in ICMPv6 error responses. sys/netinet6/in6.c In in6_newaddrmsg, announce arrival of local addresses on the interface's FIB only. In in6_lltable_rtcheck, use a per-fib ND6 cache instead of a single cache. sys/netinet6/in6_src.c In in6_selectsrc, use the caller's fib instead of the default fib. In in6_selectsrc_socket, remove a superfluous check. sys/netinet6/nd6.c In nd6_lle_event, use the interface's fib for routing socket messages. In nd6_is_new_addr_neighbor, check all FIBs when trying to determine whether an address is a neighbor. Also, simplify the code for point to point interfaces. sys/netinet6/nd6.h sys/netinet6/nd6.c sys/netinet6/nd6_rtr.c Make defrouter_select fib-aware, and make all of its callers pass in the interface fib. sys/netinet6/nd6_nbr.c When inputting a Neighbor Solicitation packet, consider the interface fib instead of the default fib for DAD. Output NS and Neighbor Advertisement packets on the correct fib. sys/netinet6/nd6_rtr.c Allow installing the same host route on different interfaces in different FIBs. If rt_add_addr_allfibs=0, only install or delete the prefix route on the interface fib. tests/sys/netinet/fibs_test.sh Clear some expected failures, but add a skip for the newly revealed BUG217871. PR: 196361 Submitted by: Erick Turnquist <jhujhiti@adjectivism.org> Reported by: Jason Healy <jhealy@logn.net> Reviewed by: asomers MFC after: 3 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D9451
2017-03-17 16:50:37 +00:00
for (; fibnum < maxfib; fibnum++) {
rt = NULL;
error = in6_rtrequest(RTM_ADD,
(struct sockaddr *)&pr->ndpr_prefix, ifa->ifa_addr,
(struct sockaddr *)&mask6, rtflags, &rt, fibnum);
if (error == 0) {
KASSERT(rt != NULL, ("%s: in6_rtrequest return no "
"error(%d) but rt is NULL, pr=%p, ifa=%p", __func__,
error, pr, ifa));
rnh = rt_tables_get_rnh(rt->rt_fibnum, AF_INET6);
/* XXX what if rhn == NULL? */
RIB_WLOCK(rnh);
RT_LOCK(rt);
if (rt_setgate(rt, rt_key(rt),
(struct sockaddr *)&null_sdl) == 0) {
struct sockaddr_dl *dl;
dl = (struct sockaddr_dl *)rt->rt_gateway;
dl->sdl_type = rt->rt_ifp->if_type;
dl->sdl_index = rt->rt_ifp->if_index;
}
RIB_WUNLOCK(rnh);
nd6_rtmsg(RTM_ADD, rt);
RT_UNLOCK(rt);
pr->ndpr_stateflags |= NDPRF_ONLINK;
} else {
char ip6buf[INET6_ADDRSTRLEN];
char ip6bufg[INET6_ADDRSTRLEN];
char ip6bufm[INET6_ADDRSTRLEN];
struct sockaddr_in6 *sin6;
sin6 = (struct sockaddr_in6 *)ifa->ifa_addr;
nd6log((LOG_ERR, "nd6_prefix_onlink: failed to add "
"route for a prefix (%s/%d) on %s, gw=%s, mask=%s, "
"flags=%lx errno = %d\n",
ip6_sprintf(ip6buf, &pr->ndpr_prefix.sin6_addr),
pr->ndpr_plen, if_name(pr->ndpr_ifp),
ip6_sprintf(ip6bufg, &sin6->sin6_addr),
ip6_sprintf(ip6bufm, &mask6.sin6_addr),
rtflags, error));
/* Save last error to return, see rtinit(). */
a_failure = error;
}
if (rt != NULL) {
RT_LOCK(rt);
RT_REMREF(rt);
RT_UNLOCK(rt);
}
}
/* Return the last error we got. */
return (a_failure);
}
int
nd6_prefix_onlink(struct nd_prefix *pr)
{
struct ifaddr *ifa;
struct ifnet *ifp = pr->ndpr_ifp;
struct nd_prefix *opr;
char ip6buf[INET6_ADDRSTRLEN];
int error;
ND6_ONLINK_LOCK_ASSERT();
ND6_UNLOCK_ASSERT();
if ((pr->ndpr_stateflags & NDPRF_ONLINK) != 0)
return (EEXIST);
/*
* Add the interface route associated with the prefix. Before
* installing the route, check if there's the same prefix on another
* interface, and the prefix has already installed the interface route.
* Although such a configuration is expected to be rare, we explicitly
* allow it.
*/
ND6_RLOCK();
LIST_FOREACH(opr, &V_nd_prefix, ndpr_entry) {
if (opr == pr)
continue;
if ((opr->ndpr_stateflags & NDPRF_ONLINK) == 0)
continue;
Constrain IPv6 routes to single FIBs when net.add_addr_allfibs=0 sys/netinet6/icmp6.c Use the interface's FIB for source address selection in ICMPv6 error responses. sys/netinet6/in6.c In in6_newaddrmsg, announce arrival of local addresses on the interface's FIB only. In in6_lltable_rtcheck, use a per-fib ND6 cache instead of a single cache. sys/netinet6/in6_src.c In in6_selectsrc, use the caller's fib instead of the default fib. In in6_selectsrc_socket, remove a superfluous check. sys/netinet6/nd6.c In nd6_lle_event, use the interface's fib for routing socket messages. In nd6_is_new_addr_neighbor, check all FIBs when trying to determine whether an address is a neighbor. Also, simplify the code for point to point interfaces. sys/netinet6/nd6.h sys/netinet6/nd6.c sys/netinet6/nd6_rtr.c Make defrouter_select fib-aware, and make all of its callers pass in the interface fib. sys/netinet6/nd6_nbr.c When inputting a Neighbor Solicitation packet, consider the interface fib instead of the default fib for DAD. Output NS and Neighbor Advertisement packets on the correct fib. sys/netinet6/nd6_rtr.c Allow installing the same host route on different interfaces in different FIBs. If rt_add_addr_allfibs=0, only install or delete the prefix route on the interface fib. tests/sys/netinet/fibs_test.sh Clear some expected failures, but add a skip for the newly revealed BUG217871. PR: 196361 Submitted by: Erick Turnquist <jhujhiti@adjectivism.org> Reported by: Jason Healy <jhealy@logn.net> Reviewed by: asomers MFC after: 3 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D9451
2017-03-17 16:50:37 +00:00
if (!V_rt_add_addr_allfibs &&
opr->ndpr_ifp->if_fib != pr->ndpr_ifp->if_fib)
continue;
if (opr->ndpr_plen == pr->ndpr_plen &&
in6_are_prefix_equal(&pr->ndpr_prefix.sin6_addr,
&opr->ndpr_prefix.sin6_addr, pr->ndpr_plen)) {
ND6_RUNLOCK();
return (0);
}
}
ND6_RUNLOCK();
/*
* We prefer link-local addresses as the associated interface address.
*/
/* search for a link-local addr */
ifa = (struct ifaddr *)in6ifa_ifpforlinklocal(ifp,
IN6_IFF_NOTREADY | IN6_IFF_ANYCAST);
if (ifa == NULL) {
/* XXX: freebsd does not have ifa_ifwithaf */
IF_ADDR_RLOCK(ifp);
ifnet: Replace if_addr_lock rwlock with epoch + mutex Run on LLNW canaries and tested by pho@ gallatin: Using a 14-core, 28-HTT single socket E5-2697 v3 with a 40GbE MLX5 based ConnectX 4-LX NIC, I see an almost 12% improvement in received packet rate, and a larger improvement in bytes delivered all the way to userspace. When the host receiving 64 streams of netperf -H $DUT -t UDP_STREAM -- -m 1, I see, using nstat -I mce0 1 before the patch: InMpps OMpps InGbs OGbs err TCP Est %CPU syscalls csw irq GBfree 4.98 0.00 4.42 0.00 4235592 33 83.80 4720653 2149771 1235 247.32 4.73 0.00 4.20 0.00 4025260 33 82.99 4724900 2139833 1204 247.32 4.72 0.00 4.20 0.00 4035252 33 82.14 4719162 2132023 1264 247.32 4.71 0.00 4.21 0.00 4073206 33 83.68 4744973 2123317 1347 247.32 4.72 0.00 4.21 0.00 4061118 33 80.82 4713615 2188091 1490 247.32 4.72 0.00 4.21 0.00 4051675 33 85.29 4727399 2109011 1205 247.32 4.73 0.00 4.21 0.00 4039056 33 84.65 4724735 2102603 1053 247.32 After the patch InMpps OMpps InGbs OGbs err TCP Est %CPU syscalls csw irq GBfree 5.43 0.00 4.20 0.00 3313143 33 84.96 5434214 1900162 2656 245.51 5.43 0.00 4.20 0.00 3308527 33 85.24 5439695 1809382 2521 245.51 5.42 0.00 4.19 0.00 3316778 33 87.54 5416028 1805835 2256 245.51 5.42 0.00 4.19 0.00 3317673 33 90.44 5426044 1763056 2332 245.51 5.42 0.00 4.19 0.00 3314839 33 88.11 5435732 1792218 2499 245.52 5.44 0.00 4.19 0.00 3293228 33 91.84 5426301 1668597 2121 245.52 Similarly, netperf reports 230Mb/s before the patch, and 270Mb/s after the patch Reviewed by: gallatin Sponsored by: Limelight Networks Differential Revision: https://reviews.freebsd.org/D15366
2018-05-18 20:13:34 +00:00
CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
if (ifa->ifa_addr->sa_family == AF_INET6) {
ifa_ref(ifa);
break;
}
}
IF_ADDR_RUNLOCK(ifp);
/* should we care about ia6_flags? */
}
if (ifa == NULL) {
/*
* This can still happen, when, for example, we receive an RA
* containing a prefix with the L bit set and the A bit clear,
* after removing all IPv6 addresses on the receiving
* interface. This should, of course, be rare though.
*/
nd6log((LOG_NOTICE,
"nd6_prefix_onlink: failed to find any ifaddr"
" to add route for a prefix(%s/%d) on %s\n",
ip6_sprintf(ip6buf, &pr->ndpr_prefix.sin6_addr),
pr->ndpr_plen, if_name(ifp)));
return (0);
}
error = nd6_prefix_onlink_rtrequest(pr, ifa);
if (ifa != NULL)
ifa_free(ifa);
return (error);
}
int
nd6_prefix_offlink(struct nd_prefix *pr)
{
int error = 0;
struct ifnet *ifp = pr->ndpr_ifp;
struct nd_prefix *opr;
struct sockaddr_in6 sa6, mask6;
struct rtentry *rt;
char ip6buf[INET6_ADDRSTRLEN];
uint64_t genid;
Constrain IPv6 routes to single FIBs when net.add_addr_allfibs=0 sys/netinet6/icmp6.c Use the interface's FIB for source address selection in ICMPv6 error responses. sys/netinet6/in6.c In in6_newaddrmsg, announce arrival of local addresses on the interface's FIB only. In in6_lltable_rtcheck, use a per-fib ND6 cache instead of a single cache. sys/netinet6/in6_src.c In in6_selectsrc, use the caller's fib instead of the default fib. In in6_selectsrc_socket, remove a superfluous check. sys/netinet6/nd6.c In nd6_lle_event, use the interface's fib for routing socket messages. In nd6_is_new_addr_neighbor, check all FIBs when trying to determine whether an address is a neighbor. Also, simplify the code for point to point interfaces. sys/netinet6/nd6.h sys/netinet6/nd6.c sys/netinet6/nd6_rtr.c Make defrouter_select fib-aware, and make all of its callers pass in the interface fib. sys/netinet6/nd6_nbr.c When inputting a Neighbor Solicitation packet, consider the interface fib instead of the default fib for DAD. Output NS and Neighbor Advertisement packets on the correct fib. sys/netinet6/nd6_rtr.c Allow installing the same host route on different interfaces in different FIBs. If rt_add_addr_allfibs=0, only install or delete the prefix route on the interface fib. tests/sys/netinet/fibs_test.sh Clear some expected failures, but add a skip for the newly revealed BUG217871. PR: 196361 Submitted by: Erick Turnquist <jhujhiti@adjectivism.org> Reported by: Jason Healy <jhealy@logn.net> Reviewed by: asomers MFC after: 3 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D9451
2017-03-17 16:50:37 +00:00
int fibnum, maxfib, a_failure;
ND6_ONLINK_LOCK_ASSERT();
ND6_UNLOCK_ASSERT();
if ((pr->ndpr_stateflags & NDPRF_ONLINK) == 0)
return (EEXIST);
bzero(&sa6, sizeof(sa6));
sa6.sin6_family = AF_INET6;
sa6.sin6_len = sizeof(sa6);
bcopy(&pr->ndpr_prefix.sin6_addr, &sa6.sin6_addr,
sizeof(struct in6_addr));
bzero(&mask6, sizeof(mask6));
mask6.sin6_family = AF_INET6;
mask6.sin6_len = sizeof(sa6);
bcopy(&pr->ndpr_mask, &mask6.sin6_addr, sizeof(struct in6_addr));
Constrain IPv6 routes to single FIBs when net.add_addr_allfibs=0 sys/netinet6/icmp6.c Use the interface's FIB for source address selection in ICMPv6 error responses. sys/netinet6/in6.c In in6_newaddrmsg, announce arrival of local addresses on the interface's FIB only. In in6_lltable_rtcheck, use a per-fib ND6 cache instead of a single cache. sys/netinet6/in6_src.c In in6_selectsrc, use the caller's fib instead of the default fib. In in6_selectsrc_socket, remove a superfluous check. sys/netinet6/nd6.c In nd6_lle_event, use the interface's fib for routing socket messages. In nd6_is_new_addr_neighbor, check all FIBs when trying to determine whether an address is a neighbor. Also, simplify the code for point to point interfaces. sys/netinet6/nd6.h sys/netinet6/nd6.c sys/netinet6/nd6_rtr.c Make defrouter_select fib-aware, and make all of its callers pass in the interface fib. sys/netinet6/nd6_nbr.c When inputting a Neighbor Solicitation packet, consider the interface fib instead of the default fib for DAD. Output NS and Neighbor Advertisement packets on the correct fib. sys/netinet6/nd6_rtr.c Allow installing the same host route on different interfaces in different FIBs. If rt_add_addr_allfibs=0, only install or delete the prefix route on the interface fib. tests/sys/netinet/fibs_test.sh Clear some expected failures, but add a skip for the newly revealed BUG217871. PR: 196361 Submitted by: Erick Turnquist <jhujhiti@adjectivism.org> Reported by: Jason Healy <jhealy@logn.net> Reviewed by: asomers MFC after: 3 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D9451
2017-03-17 16:50:37 +00:00
if (V_rt_add_addr_allfibs) {
fibnum = 0;
maxfib = rt_numfibs;
} else {
fibnum = ifp->if_fib;
maxfib = fibnum + 1;
}
a_failure = 0;
Constrain IPv6 routes to single FIBs when net.add_addr_allfibs=0 sys/netinet6/icmp6.c Use the interface's FIB for source address selection in ICMPv6 error responses. sys/netinet6/in6.c In in6_newaddrmsg, announce arrival of local addresses on the interface's FIB only. In in6_lltable_rtcheck, use a per-fib ND6 cache instead of a single cache. sys/netinet6/in6_src.c In in6_selectsrc, use the caller's fib instead of the default fib. In in6_selectsrc_socket, remove a superfluous check. sys/netinet6/nd6.c In nd6_lle_event, use the interface's fib for routing socket messages. In nd6_is_new_addr_neighbor, check all FIBs when trying to determine whether an address is a neighbor. Also, simplify the code for point to point interfaces. sys/netinet6/nd6.h sys/netinet6/nd6.c sys/netinet6/nd6_rtr.c Make defrouter_select fib-aware, and make all of its callers pass in the interface fib. sys/netinet6/nd6_nbr.c When inputting a Neighbor Solicitation packet, consider the interface fib instead of the default fib for DAD. Output NS and Neighbor Advertisement packets on the correct fib. sys/netinet6/nd6_rtr.c Allow installing the same host route on different interfaces in different FIBs. If rt_add_addr_allfibs=0, only install or delete the prefix route on the interface fib. tests/sys/netinet/fibs_test.sh Clear some expected failures, but add a skip for the newly revealed BUG217871. PR: 196361 Submitted by: Erick Turnquist <jhujhiti@adjectivism.org> Reported by: Jason Healy <jhealy@logn.net> Reviewed by: asomers MFC after: 3 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D9451
2017-03-17 16:50:37 +00:00
for (; fibnum < maxfib; fibnum++) {
rt = NULL;
error = in6_rtrequest(RTM_DELETE, (struct sockaddr *)&sa6, NULL,
(struct sockaddr *)&mask6, 0, &rt, fibnum);
if (error == 0) {
/* report the route deletion to the routing socket. */
if (rt != NULL)
nd6_rtmsg(RTM_DELETE, rt);
} else {
/* Save last error to return, see rtinit(). */
a_failure = error;
}
if (rt != NULL) {
RTFREE(rt);
}
}
error = a_failure;
a_failure = 1;
if (error == 0) {
pr->ndpr_stateflags &= ~NDPRF_ONLINK;
/*
* There might be the same prefix on another interface,
* the prefix which could not be on-link just because we have
* the interface route (see comments in nd6_prefix_onlink).
* If there's one, try to make the prefix on-link on the
* interface.
*/
ND6_RLOCK();
restart:
LIST_FOREACH(opr, &V_nd_prefix, ndpr_entry) {
/*
* KAME specific: detached prefixes should not be
* on-link.
*/
if (opr == pr || (opr->ndpr_stateflags &
(NDPRF_ONLINK | NDPRF_DETACHED)) != 0)
continue;
if (opr->ndpr_plen == pr->ndpr_plen &&
in6_are_prefix_equal(&pr->ndpr_prefix.sin6_addr,
&opr->ndpr_prefix.sin6_addr, pr->ndpr_plen)) {
int e;
genid = V_nd6_list_genid;
ND6_RUNLOCK();
if ((e = nd6_prefix_onlink(opr)) != 0) {
nd6log((LOG_ERR,
"nd6_prefix_offlink: failed to "
"recover a prefix %s/%d from %s "
"to %s (errno = %d)\n",
ip6_sprintf(ip6buf,
&opr->ndpr_prefix.sin6_addr),
opr->ndpr_plen, if_name(ifp),
if_name(opr->ndpr_ifp), e));
} else
a_failure = 0;
ND6_RLOCK();
if (genid != V_nd6_list_genid)
goto restart;
}
}
ND6_RUNLOCK();
} else {
/* XXX: can we still set the NDPRF_ONLINK flag? */
nd6log((LOG_ERR,
"nd6_prefix_offlink: failed to delete route: "
"%s/%d on %s (errno = %d)\n",
ip6_sprintf(ip6buf, &sa6.sin6_addr), pr->ndpr_plen,
if_name(ifp), error));
}
if (a_failure)
lltable_prefix_free(AF_INET6, (struct sockaddr *)&sa6,
(struct sockaddr *)&mask6, LLE_STATIC);
return (error);
}
static struct in6_ifaddr *
in6_ifadd(struct nd_prefixctl *pr, int mcast)
{
struct ifnet *ifp = pr->ndpr_ifp;
struct ifaddr *ifa;
struct in6_aliasreq ifra;
struct in6_ifaddr *ia, *ib;
int error, plen0;
struct in6_addr mask;
int prefixlen = pr->ndpr_plen;
int updateflags;
char ip6buf[INET6_ADDRSTRLEN];
in6_prefixlen2mask(&mask, prefixlen);
/*
* find a link-local address (will be interface ID).
* Is it really mandatory? Theoretically, a global or a site-local
* address can be configured without a link-local address, if we
* have a unique interface identifier...
*
* it is not mandatory to have a link-local address, we can generate
* interface identifier on the fly. we do this because:
* (1) it should be the easiest way to find interface identifier.
* (2) RFC2462 5.4 suggesting the use of the same interface identifier
* for multiple addresses on a single interface, and possible shortcut
* of DAD. we omitted DAD for this reason in the past.
* (3) a user can prevent autoconfiguration of global address
* by removing link-local address by hand (this is partly because we
* don't have other way to control the use of IPv6 on an interface.
* this has been our design choice - cf. NRL's "ifconfig auto").
* (4) it is easier to manage when an interface has addresses
* with the same interface identifier, than to have multiple addresses
* with different interface identifiers.
*/
ifa = (struct ifaddr *)in6ifa_ifpforlinklocal(ifp, 0); /* 0 is OK? */
if (ifa)
ib = (struct in6_ifaddr *)ifa;
else
return NULL;
/* prefixlen + ifidlen must be equal to 128 */
plen0 = in6_mask2len(&ib->ia_prefixmask.sin6_addr, NULL);
if (prefixlen != plen0) {
ifa_free(ifa);
nd6log((LOG_INFO, "in6_ifadd: wrong prefixlen for %s "
"(prefix=%d ifid=%d)\n",
if_name(ifp), prefixlen, 128 - plen0));
return NULL;
}
/* make ifaddr */
in6_prepare_ifra(&ifra, &pr->ndpr_prefix.sin6_addr, &mask);
IN6_MASK_ADDR(&ifra.ifra_addr.sin6_addr, &mask);
/* interface ID */
ifra.ifra_addr.sin6_addr.s6_addr32[0] |=
(ib->ia_addr.sin6_addr.s6_addr32[0] & ~mask.s6_addr32[0]);
ifra.ifra_addr.sin6_addr.s6_addr32[1] |=
(ib->ia_addr.sin6_addr.s6_addr32[1] & ~mask.s6_addr32[1]);
ifra.ifra_addr.sin6_addr.s6_addr32[2] |=
(ib->ia_addr.sin6_addr.s6_addr32[2] & ~mask.s6_addr32[2]);
ifra.ifra_addr.sin6_addr.s6_addr32[3] |=
(ib->ia_addr.sin6_addr.s6_addr32[3] & ~mask.s6_addr32[3]);
ifa_free(ifa);
/* lifetimes. */
ifra.ifra_lifetime.ia6t_vltime = pr->ndpr_vltime;
ifra.ifra_lifetime.ia6t_pltime = pr->ndpr_pltime;
/* XXX: scope zone ID? */
ifra.ifra_flags |= IN6_IFF_AUTOCONF; /* obey autoconf */
/*
* Make sure that we do not have this address already. This should
* usually not happen, but we can still see this case, e.g., if we
* have manually configured the exact address to be configured.
*/
ifa = (struct ifaddr *)in6ifa_ifpwithaddr(ifp,
&ifra.ifra_addr.sin6_addr);
if (ifa != NULL) {
ifa_free(ifa);
/* this should be rare enough to make an explicit log */
log(LOG_INFO, "in6_ifadd: %s is already configured\n",
ip6_sprintf(ip6buf, &ifra.ifra_addr.sin6_addr));
return (NULL);
}
/*
* Allocate ifaddr structure, link into chain, etc.
* If we are going to create a new address upon receiving a multicasted
* RA, we need to impose a random delay before starting DAD.
* [draft-ietf-ipv6-rfc2462bis-02.txt, Section 5.4.2]
*/
updateflags = 0;
if (mcast)
updateflags |= IN6_IFAUPDATE_DADDELAY;
if ((error = in6_update_ifa(ifp, &ifra, NULL, updateflags)) != 0) {
nd6log((LOG_ERR,
"in6_ifadd: failed to make ifaddr %s on %s (errno=%d)\n",
ip6_sprintf(ip6buf, &ifra.ifra_addr.sin6_addr),
if_name(ifp), error));
return (NULL); /* ifaddr must not have been allocated. */
}
ia = in6ifa_ifpwithaddr(ifp, &ifra.ifra_addr.sin6_addr);
/*
* XXXRW: Assumption of non-NULLness here might not be true with
* fine-grained locking -- should we validate it? Or just return
* earlier ifa rather than looking it up again?
*/
return (ia); /* this is always non-NULL and referenced. */
}
/*
* ia0 - corresponding public address
*/
int
in6_tmpifadd(const struct in6_ifaddr *ia0, int forcegen, int delay)
{
struct ifnet *ifp = ia0->ia_ifa.ifa_ifp;
struct in6_ifaddr *newia;
struct in6_aliasreq ifra;
int error;
int trylimit = 3; /* XXX: adhoc value */
int updateflags;
u_int32_t randid[2];
time_t vltime0, pltime0;
in6_prepare_ifra(&ifra, &ia0->ia_addr.sin6_addr,
&ia0->ia_prefixmask.sin6_addr);
ifra.ifra_addr = ia0->ia_addr; /* XXX: do we need this ? */
/* clear the old IFID */
IN6_MASK_ADDR(&ifra.ifra_addr.sin6_addr,
&ifra.ifra_prefixmask.sin6_addr);
again:
if (in6_get_tmpifid(ifp, (u_int8_t *)randid,
(const u_int8_t *)&ia0->ia_addr.sin6_addr.s6_addr[8], forcegen)) {
nd6log((LOG_NOTICE, "in6_tmpifadd: failed to find a good "
"random IFID\n"));
return (EINVAL);
}
ifra.ifra_addr.sin6_addr.s6_addr32[2] |=
(randid[0] & ~(ifra.ifra_prefixmask.sin6_addr.s6_addr32[2]));
ifra.ifra_addr.sin6_addr.s6_addr32[3] |=
(randid[1] & ~(ifra.ifra_prefixmask.sin6_addr.s6_addr32[3]));
/*
* in6_get_tmpifid() quite likely provided a unique interface ID.
* However, we may still have a chance to see collision, because
* there may be a time lag between generation of the ID and generation
* of the address. So, we'll do one more sanity check.
*/
if (in6_localip(&ifra.ifra_addr.sin6_addr) != 0) {
if (trylimit-- > 0) {
forcegen = 1;
goto again;
}
/* Give up. Something strange should have happened. */
nd6log((LOG_NOTICE, "in6_tmpifadd: failed to "
"find a unique random IFID\n"));
return (EEXIST);
}
/*
* The Valid Lifetime is the lower of the Valid Lifetime of the
* public address or TEMP_VALID_LIFETIME.
* The Preferred Lifetime is the lower of the Preferred Lifetime
* of the public address or TEMP_PREFERRED_LIFETIME -
* DESYNC_FACTOR.
*/
if (ia0->ia6_lifetime.ia6t_vltime != ND6_INFINITE_LIFETIME) {
vltime0 = IFA6_IS_INVALID(ia0) ? 0 :
(ia0->ia6_lifetime.ia6t_vltime -
(time_uptime - ia0->ia6_updatetime));
if (vltime0 > V_ip6_temp_valid_lifetime)
vltime0 = V_ip6_temp_valid_lifetime;
} else
vltime0 = V_ip6_temp_valid_lifetime;
if (ia0->ia6_lifetime.ia6t_pltime != ND6_INFINITE_LIFETIME) {
pltime0 = IFA6_IS_DEPRECATED(ia0) ? 0 :
(ia0->ia6_lifetime.ia6t_pltime -
(time_uptime - ia0->ia6_updatetime));
if (pltime0 > V_ip6_temp_preferred_lifetime - V_ip6_desync_factor){
pltime0 = V_ip6_temp_preferred_lifetime -
V_ip6_desync_factor;
}
} else
pltime0 = V_ip6_temp_preferred_lifetime - V_ip6_desync_factor;
ifra.ifra_lifetime.ia6t_vltime = vltime0;
ifra.ifra_lifetime.ia6t_pltime = pltime0;
/*
* A temporary address is created only if this calculated Preferred
* Lifetime is greater than REGEN_ADVANCE time units.
*/
if (ifra.ifra_lifetime.ia6t_pltime <= V_ip6_temp_regen_advance)
return (0);
/* XXX: scope zone ID? */
ifra.ifra_flags |= (IN6_IFF_AUTOCONF|IN6_IFF_TEMPORARY);
/* allocate ifaddr structure, link into chain, etc. */
updateflags = 0;
if (delay)
updateflags |= IN6_IFAUPDATE_DADDELAY;
if ((error = in6_update_ifa(ifp, &ifra, NULL, updateflags)) != 0)
return (error);
newia = in6ifa_ifpwithaddr(ifp, &ifra.ifra_addr.sin6_addr);
if (newia == NULL) { /* XXX: can it happen? */
nd6log((LOG_ERR,
"in6_tmpifadd: ifa update succeeded, but we got "
"no ifaddr\n"));
return (EINVAL); /* XXX */
}
newia->ia6_ndpr = ia0->ia6_ndpr;
newia->ia6_ndpr->ndpr_addrcnt++;
ifa_free(&newia->ia_ifa);
/*
* A newly added address might affect the status of other addresses.
* XXX: when the temporary address is generated with a new public
* address, the onlink check is redundant. However, it would be safe
* to do the check explicitly everywhere a new address is generated,
* and, in fact, we surely need the check when we create a new
* temporary address due to deprecation of an old temporary address.
*/
pfxlist_onlink_check();
return (0);
}
static int
in6_init_prefix_ltimes(struct nd_prefix *ndpr)
{
if (ndpr->ndpr_pltime == ND6_INFINITE_LIFETIME)
ndpr->ndpr_preferred = 0;
else
ndpr->ndpr_preferred = time_uptime + ndpr->ndpr_pltime;
if (ndpr->ndpr_vltime == ND6_INFINITE_LIFETIME)
ndpr->ndpr_expire = 0;
else
ndpr->ndpr_expire = time_uptime + ndpr->ndpr_vltime;
return 0;
}
static void
in6_init_address_ltimes(struct nd_prefix *new, struct in6_addrlifetime *lt6)
{
/* init ia6t_expire */
if (lt6->ia6t_vltime == ND6_INFINITE_LIFETIME)
lt6->ia6t_expire = 0;
else {
lt6->ia6t_expire = time_uptime;
lt6->ia6t_expire += lt6->ia6t_vltime;
}
/* init ia6t_preferred */
if (lt6->ia6t_pltime == ND6_INFINITE_LIFETIME)
lt6->ia6t_preferred = 0;
else {
lt6->ia6t_preferred = time_uptime;
lt6->ia6t_preferred += lt6->ia6t_pltime;
}
}
/*
* Delete all the routing table entries that use the specified gateway.
* XXX: this function causes search through all entries of routing table, so
* it shouldn't be called when acting as a router.
*/
void
rt6_flush(struct in6_addr *gateway, struct ifnet *ifp)
{
/* We'll care only link-local addresses */
if (!IN6_IS_ADDR_LINKLOCAL(gateway))
return;
/* XXX Do we really need to walk any but the default FIB? */
rt_foreach_fib_walk_del(AF_INET6, rt6_deleteroute, (void *)gateway);
}
static int
rt6_deleteroute(const struct rtentry *rt, void *arg)
{
#define SIN6(s) ((struct sockaddr_in6 *)s)
struct in6_addr *gate = (struct in6_addr *)arg;
if (rt->rt_gateway == NULL || rt->rt_gateway->sa_family != AF_INET6)
return (0);
if (!IN6_ARE_ADDR_EQUAL(gate, &SIN6(rt->rt_gateway)->sin6_addr)) {
return (0);
}
/*
* Do not delete a static route.
* XXX: this seems to be a bit ad-hoc. Should we consider the
* 'cloned' bit instead?
*/
if ((rt->rt_flags & RTF_STATIC) != 0)
return (0);
/*
* We delete only host route. This means, in particular, we don't
* delete default route.
*/
if ((rt->rt_flags & RTF_HOST) == 0)
return (0);
return (1);
#undef SIN6
}
int
nd6_setdefaultiface(int ifindex)
{
int error = 0;
if (ifindex < 0 || V_if_index < ifindex)
return (EINVAL);
if (ifindex != 0 && !ifnet_byindex(ifindex))
return (EINVAL);
if (V_nd6_defifindex != ifindex) {
V_nd6_defifindex = ifindex;
if (V_nd6_defifindex > 0)
V_nd6_defifp = ifnet_byindex(V_nd6_defifindex);
else
V_nd6_defifp = NULL;
/*
* Our current implementation assumes one-to-one maping between
* interfaces and links, so it would be natural to use the
* default interface as the default link.
*/
scope6_setdefault(V_nd6_defifp);
}
return (error);
}