freebsd-nq/usr.bin/netstat/nhops.c

481 lines
12 KiB
C
Raw Normal View History

Introduce nexthop objects and new routing KPI. This is the foundational change for the routing subsytem rearchitecture. More details and goals are available in https://reviews.freebsd.org/D24141 . This patch introduces concept of nexthop objects and new nexthop-based routing KPI. Nexthops are objects, containing all necessary information for performing the packet output decision. Output interface, mtu, flags, gw address goes there. For most of the cases, these objects will serve the same role as the struct rtentry is currently serving. Typically there will be low tens of such objects for the router even with multiple BGP full-views, as these objects will be shared between routing entries. This allows to store more information in the nexthop. New KPI: struct nhop_object *fib4_lookup(uint32_t fibnum, struct in_addr dst, uint32_t scopeid, uint32_t flags, uint32_t flowid); struct nhop_object *fib6_lookup(uint32_t fibnum, const struct in6_addr *dst6, uint32_t scopeid, uint32_t flags, uint32_t flowid); These 2 function are intended to replace all all flavours of <in_|in6_>rtalloc[1]<_ign><_fib>, mpath functions and the previous fib[46]-generation functions. Upon successful lookup, they return nexthop object which is guaranteed to exist within current NET_EPOCH. If longer lifetime is desired, one can specify NHR_REF as a flag and get a referenced version of the nexthop. Reference semantic closely resembles rtentry one, allowing sed-style conversion. Additionally, another 2 functions are introduced to support uRPF functionality inside variety of our firewalls. Their primary goal is to hide the multipath implementation details inside the routing subsystem, greatly simplifying firewalls implementation: int fib4_lookup_urpf(uint32_t fibnum, struct in_addr dst, uint32_t scopeid, uint32_t flags, const struct ifnet *src_if); int fib6_lookup_urpf(uint32_t fibnum, const struct in6_addr *dst6, uint32_t scopeid, uint32_t flags, const struct ifnet *src_if); All functions have a separate scopeid argument, paving way to eliminating IPv6 scope embedding and allowing to support IPv4 link-locals in the future. Structure changes: * rtentry gets new 'rt_nhop' pointer, slightly growing the overall size. * rib_head gets new 'rnh_preadd' callback pointer, slightly growing overall sz. Old KPI: During the transition state old and new KPI will coexists. As there are another 4-5 decent-sized conversion patches, it will probably take a couple of weeks. To support both KPIs, fields not required by the new KPI (most of rtentry) has to be kept, resulting in the temporary size increase. Once conversion is finished, rtentry will notably shrink. More details: * architectural overview: https://reviews.freebsd.org/D24141 * list of the next changes: https://reviews.freebsd.org/D24232 Reviewed by: ae,glebius(initial version) Differential Revision: https://reviews.freebsd.org/D24232
2020-04-12 14:30:00 +00:00
/*-
* SPDX-License-Identifier: BSD-3-Clause
*
* Copyright (c) 1983, 1988, 1993
* The Regents of the University of California. 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 University 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 REGENTS 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 REGENTS 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.
*
* $FreeBSD$
*/
#include <sys/cdefs.h>
#include <sys/param.h>
#include <sys/protosw.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/sysctl.h>
#include <sys/time.h>
#include <net/ethernet.h>
#include <net/if.h>
#include <net/if_dl.h>
#include <net/if_types.h>
#include <net/route.h>
#include <net/route/nhop.h>
#include <netinet/in.h>
#include <netgraph/ng_socket.h>
#include <arpa/inet.h>
#include <ifaddrs.h>
#include <libutil.h>
#include <netdb.h>
#include <stdbool.h>
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <stdbool.h>
#include <string.h>
#include <sysexits.h>
#include <unistd.h>
#include <err.h>
#include <libxo/xo.h>
#include "netstat.h"
#include "common.h"
/* column widths; each followed by one space */
#ifndef INET6
#define WID_DST_DEFAULT(af) 18 /* width of destination column */
#define WID_GW_DEFAULT(af) 18 /* width of gateway column */
#define WID_IF_DEFAULT(af) (Wflag ? 10 : 8) /* width of netif column */
#else
#define WID_DST_DEFAULT(af) \
((af) == AF_INET6 ? (numeric_addr ? 33: 18) : 18)
#define WID_GW_DEFAULT(af) \
((af) == AF_INET6 ? (numeric_addr ? 29 : 18) : 18)
#define WID_IF_DEFAULT(af) ((af) == AF_INET6 ? 8 : (Wflag ? 10 : 8))
#endif /*INET6*/
static int wid_dst;
static int wid_gw;
static int wid_flags;
static int wid_pksent;
static int wid_mtu;
static int wid_if;
static int wid_nhidx;
static int wid_nhtype;
static int wid_refcnt;
static int wid_prepend;
static struct bits nh_bits[] = {
{ NHF_REJECT, 'R', "reject" },
{ NHF_BLACKHOLE,'B', "blackhole" },
{ NHF_REDIRECT, 'r', "redirect" },
{ NHF_GATEWAY, 'G', "gateway" },
{ NHF_DEFAULT, 'd', "default" },
{ NHF_BROADCAST,'b', "broadcast" },
{ 0 , 0, NULL }
};
static char *nh_types[] = {
"empty", /* 0 */
"v4/resolve", /* 1 */
"v4/gw",
"v6/resolve",
"v6/gw"
};
struct nhop_entry {
char gw[64];
char ifname[IFNAMSIZ];
};
struct nhop_map {
struct nhop_entry *ptr;
size_t size;
};
static struct nhop_map global_nhop_map;
static struct nhop_entry *nhop_get(struct nhop_map *map, uint32_t idx);
static struct ifmap_entry *ifmap;
static size_t ifmap_size;
static void
print_sockaddr_buf(char *buf, size_t bufsize, const struct sockaddr *sa)
{
switch (sa->sa_family) {
case AF_INET:
inet_ntop(AF_INET, &((struct sockaddr_in *)sa)->sin_addr,
buf, bufsize);
break;
case AF_INET6:
inet_ntop(AF_INET6, &((struct sockaddr_in6 *)sa)->sin6_addr,
buf, bufsize);
break;
default:
snprintf(buf, bufsize, "unknown:%d", sa->sa_family);
break;
}
}
static int
print_addr(const char *name, const char *addr, int width)
{
char buf[128];
int protrusion;
if (width < 0) {
snprintf(buf, sizeof(buf), "{:%s/%%s} ", name);
xo_emit(buf, addr);
protrusion = 0;
} else {
if (Wflag != 0 || numeric_addr) {
snprintf(buf, sizeof(buf), "{[:%d}{:%s/%%s}{]:} ",
-width, name);
xo_emit(buf, addr);
protrusion = strlen(addr) - width;
if (protrusion < 0)
protrusion = 0;
} else {
snprintf(buf, sizeof(buf), "{[:%d}{:%s/%%-.*s}{]:} ",
-width, name);
xo_emit(buf, width, addr);
protrusion = 0;
}
}
return (protrusion);
}
static void
print_nhop_header(int af1 __unused)
{
if (Wflag) {
xo_emit("{T:/%-*.*s} {T:/%-*.*s} {T:/%-*.*s} {T:/%-*.*s} {T:/%*.*s} "
"{T:/%*.*s} {T:/%-*.*s} {T:/%*.*s} {T:/%*.*s} {T:/%*.*s} {T:/%*s}\n",
wid_nhidx, wid_nhidx, "Idx",
wid_nhtype, wid_nhtype, "Type",
wid_dst, wid_dst, "IFA",
wid_gw, wid_gw, "Gateway",
wid_flags, wid_flags, "Flags",
wid_pksent, wid_pksent, "Use",
wid_mtu, wid_mtu, "Mtu",
wid_if, wid_if, "Netif",
wid_if, wid_if, "Addrif",
wid_refcnt, wid_refcnt, "Refcnt",
wid_prepend, "Prepend");
} else {
xo_emit("{T:/%-*.*s} {T:/%-*.*s} {T:/%-*.*s} {T:/%-*.*s} {T:/%*.*s} "
" {T:/%*s}\n",
wid_nhidx, wid_nhidx, "Idx",
wid_dst, wid_dst, "IFA",
wid_gw, wid_gw, "Gateway",
wid_flags, wid_flags, "Flags",
wid_if, wid_if, "Netif",
wid_prepend, "Refcnt");
}
}
Introduce scalable route multipath. This change is based on the nexthop objects landed in D24232. The change introduces the concept of nexthop groups. Each group contains the collection of nexthops with their relative weights and a dataplane-optimized structure to enable efficient nexthop selection. Simular to the nexthops, nexthop groups are immutable. Dataplane part gets compiled during group creation and is basically an array of nexthop pointers, compiled w.r.t their weights. With this change, `rt_nhop` field of `struct rtentry` contains either nexthop or nexthop group. They are distinguished by the presense of NHF_MULTIPATH flag. All dataplane lookup functions returns pointer to the nexthop object, leaving nexhop groups details inside routing subsystem. User-visible changes: The change is intended to be backward-compatible: all non-mpath operations should work as before with ROUTE_MPATH and net.route.multipath=1. All routes now comes with weight, default weight is 1, maximum is 2^24-1. Current maximum multipath group width is statically set to 64. This will become sysctl-tunable in the followup changes. Using functionality: * Recompile kernel with ROUTE_MPATH * set net.route.multipath to 1 route add -6 2001:db8::/32 2001:db8::2 -weight 10 route add -6 2001:db8::/32 2001:db8::3 -weight 20 netstat -6On Nexthop groups data Internet6: GrpIdx NhIdx Weight Slots Gateway Netif Refcnt 1 ------- ------- ------- --------------------------------------- --------- 1 13 10 1 2001:db8::2 vlan2 14 20 2 2001:db8::3 vlan2 Next steps: * Land outbound hashing for locally-originated routes ( D26523 ). * Fix net/bird multipath (net/frr seems to work fine) * Add ROUTE_MPATH to GENERIC * Set net.route.multipath=1 by default Tested by: olivier Reviewed by: glebius Relnotes: yes Differential Revision: https://reviews.freebsd.org/D26449
2020-10-03 10:47:17 +00:00
void
Introduce nexthop objects and new routing KPI. This is the foundational change for the routing subsytem rearchitecture. More details and goals are available in https://reviews.freebsd.org/D24141 . This patch introduces concept of nexthop objects and new nexthop-based routing KPI. Nexthops are objects, containing all necessary information for performing the packet output decision. Output interface, mtu, flags, gw address goes there. For most of the cases, these objects will serve the same role as the struct rtentry is currently serving. Typically there will be low tens of such objects for the router even with multiple BGP full-views, as these objects will be shared between routing entries. This allows to store more information in the nexthop. New KPI: struct nhop_object *fib4_lookup(uint32_t fibnum, struct in_addr dst, uint32_t scopeid, uint32_t flags, uint32_t flowid); struct nhop_object *fib6_lookup(uint32_t fibnum, const struct in6_addr *dst6, uint32_t scopeid, uint32_t flags, uint32_t flowid); These 2 function are intended to replace all all flavours of <in_|in6_>rtalloc[1]<_ign><_fib>, mpath functions and the previous fib[46]-generation functions. Upon successful lookup, they return nexthop object which is guaranteed to exist within current NET_EPOCH. If longer lifetime is desired, one can specify NHR_REF as a flag and get a referenced version of the nexthop. Reference semantic closely resembles rtentry one, allowing sed-style conversion. Additionally, another 2 functions are introduced to support uRPF functionality inside variety of our firewalls. Their primary goal is to hide the multipath implementation details inside the routing subsystem, greatly simplifying firewalls implementation: int fib4_lookup_urpf(uint32_t fibnum, struct in_addr dst, uint32_t scopeid, uint32_t flags, const struct ifnet *src_if); int fib6_lookup_urpf(uint32_t fibnum, const struct in6_addr *dst6, uint32_t scopeid, uint32_t flags, const struct ifnet *src_if); All functions have a separate scopeid argument, paving way to eliminating IPv6 scope embedding and allowing to support IPv4 link-locals in the future. Structure changes: * rtentry gets new 'rt_nhop' pointer, slightly growing the overall size. * rib_head gets new 'rnh_preadd' callback pointer, slightly growing overall sz. Old KPI: During the transition state old and new KPI will coexists. As there are another 4-5 decent-sized conversion patches, it will probably take a couple of weeks. To support both KPIs, fields not required by the new KPI (most of rtentry) has to be kept, resulting in the temporary size increase. Once conversion is finished, rtentry will notably shrink. More details: * architectural overview: https://reviews.freebsd.org/D24141 * list of the next changes: https://reviews.freebsd.org/D24232 Reviewed by: ae,glebius(initial version) Differential Revision: https://reviews.freebsd.org/D24232
2020-04-12 14:30:00 +00:00
nhop_map_update(struct nhop_map *map, uint32_t idx, char *gw, char *ifname)
{
if (idx >= map->size) {
uint32_t new_size;
size_t sz;
if (map->size == 0)
new_size = 32;
else
new_size = map->size * 2;
if (new_size <= idx)
new_size = roundup(idx + 1, 32);
sz = new_size * (sizeof(struct nhop_entry));
if ((map->ptr = realloc(map->ptr, sz)) == NULL)
errx(2, "realloc(%zu) failed", sz);
Introduce nexthop objects and new routing KPI. This is the foundational change for the routing subsytem rearchitecture. More details and goals are available in https://reviews.freebsd.org/D24141 . This patch introduces concept of nexthop objects and new nexthop-based routing KPI. Nexthops are objects, containing all necessary information for performing the packet output decision. Output interface, mtu, flags, gw address goes there. For most of the cases, these objects will serve the same role as the struct rtentry is currently serving. Typically there will be low tens of such objects for the router even with multiple BGP full-views, as these objects will be shared between routing entries. This allows to store more information in the nexthop. New KPI: struct nhop_object *fib4_lookup(uint32_t fibnum, struct in_addr dst, uint32_t scopeid, uint32_t flags, uint32_t flowid); struct nhop_object *fib6_lookup(uint32_t fibnum, const struct in6_addr *dst6, uint32_t scopeid, uint32_t flags, uint32_t flowid); These 2 function are intended to replace all all flavours of <in_|in6_>rtalloc[1]<_ign><_fib>, mpath functions and the previous fib[46]-generation functions. Upon successful lookup, they return nexthop object which is guaranteed to exist within current NET_EPOCH. If longer lifetime is desired, one can specify NHR_REF as a flag and get a referenced version of the nexthop. Reference semantic closely resembles rtentry one, allowing sed-style conversion. Additionally, another 2 functions are introduced to support uRPF functionality inside variety of our firewalls. Their primary goal is to hide the multipath implementation details inside the routing subsystem, greatly simplifying firewalls implementation: int fib4_lookup_urpf(uint32_t fibnum, struct in_addr dst, uint32_t scopeid, uint32_t flags, const struct ifnet *src_if); int fib6_lookup_urpf(uint32_t fibnum, const struct in6_addr *dst6, uint32_t scopeid, uint32_t flags, const struct ifnet *src_if); All functions have a separate scopeid argument, paving way to eliminating IPv6 scope embedding and allowing to support IPv4 link-locals in the future. Structure changes: * rtentry gets new 'rt_nhop' pointer, slightly growing the overall size. * rib_head gets new 'rnh_preadd' callback pointer, slightly growing overall sz. Old KPI: During the transition state old and new KPI will coexists. As there are another 4-5 decent-sized conversion patches, it will probably take a couple of weeks. To support both KPIs, fields not required by the new KPI (most of rtentry) has to be kept, resulting in the temporary size increase. Once conversion is finished, rtentry will notably shrink. More details: * architectural overview: https://reviews.freebsd.org/D24141 * list of the next changes: https://reviews.freebsd.org/D24232 Reviewed by: ae,glebius(initial version) Differential Revision: https://reviews.freebsd.org/D24232
2020-04-12 14:30:00 +00:00
memset(&map->ptr[map->size], 0, (new_size - map->size) * sizeof(struct nhop_entry));
map->size = new_size;
}
strlcpy(map->ptr[idx].ifname, ifname, sizeof(map->ptr[idx].ifname));
strlcpy(map->ptr[idx].gw, gw, sizeof(map->ptr[idx].gw));
}
static struct nhop_entry *
nhop_get(struct nhop_map *map, uint32_t idx)
{
if (idx >= map->size)
return (NULL);
if (*map->ptr[idx].ifname == '\0')
return (NULL);
return &map->ptr[idx];
}
static void
print_nhop_entry_sysctl(const char *name, struct rt_msghdr *rtm, struct nhop_external *nh)
{
char buffer[128];
char iface_name[128];
int protrusion;
char gw_addr[64];
struct nhop_addrs *na;
struct sockaddr *sa_gw, *sa_ifa;
xo_open_instance(name);
snprintf(buffer, sizeof(buffer), "{[:-%d}{:index/%%lu}{]:} ", wid_nhidx);
//xo_emit("{t:index/%-lu} ", wid_nhidx, nh->nh_idx);
xo_emit(buffer, nh->nh_idx);
if (Wflag) {
char *cp = nh_types[nh->nh_type];
xo_emit("{t:type_str/%*s} ", wid_nhtype, cp);
}
memset(iface_name, 0, sizeof(iface_name));
if (nh->ifindex < (uint32_t)ifmap_size) {
strlcpy(iface_name, ifmap[nh->ifindex].ifname,
sizeof(iface_name));
if (*iface_name == '\0')
strlcpy(iface_name, "---", sizeof(iface_name));
}
na = (struct nhop_addrs *)((char *)nh + nh->nh_len);
//inet_ntop(nh->nh_family, &nh->nh_src, src_addr, sizeof(src_addr));
//protrusion = p_addr("ifa", src_addr, wid_dst);
sa_gw = (struct sockaddr *)((char *)na + na->gw_sa_off);
sa_ifa = (struct sockaddr *)((char *)na + na->src_sa_off);
protrusion = p_sockaddr("ifa", sa_ifa, NULL, RTF_HOST, wid_dst);
if (nh->nh_flags & NHF_GATEWAY) {
const char *cp;
cp = fmt_sockaddr(sa_gw, NULL, RTF_HOST);
strlcpy(gw_addr, cp, sizeof(gw_addr));
} else
snprintf(gw_addr, sizeof(gw_addr), "%s/resolve", iface_name);
protrusion = print_addr("gateway", gw_addr, wid_dst - protrusion);
nhop_map_update(&global_nhop_map, nh->nh_idx, gw_addr, iface_name);
snprintf(buffer, sizeof(buffer), "{[:-%d}{:flags/%%s}{]:} ",
wid_flags - protrusion);
//p_nhflags(nh->nh_flags, buffer);
print_flags_generic(rtm->rtm_flags, rt_bits, buffer, "rt_flags_pretty");
if (Wflag) {
xo_emit("{t:use/%*lu} ", wid_pksent, nh->nh_pksent);
xo_emit("{t:mtu/%*lu} ", wid_mtu, nh->nh_mtu);
}
//printf("IDX: %d IFACE: %s FAMILY: %d TYPE: %d FLAGS: %X GW \n");
if (Wflag)
xo_emit("{t:interface-name/%*s}", wid_if, iface_name);
else
xo_emit("{t:interface-name/%*.*s}", wid_if, wid_if, iface_name);
memset(iface_name, 0, sizeof(iface_name));
if (nh->aifindex < (uint32_t)ifmap_size && nh->ifindex != nh->aifindex) {
strlcpy(iface_name, ifmap[nh->aifindex].ifname,
sizeof(iface_name));
if (*iface_name == '\0')
strlcpy(iface_name, "---", sizeof(iface_name));
}
if (Wflag)
xo_emit("{t:address-interface-name/%*s}", wid_if, iface_name);
xo_emit("{t:refcount/%*lu} ", wid_refcnt, nh->nh_refcount);
if (Wflag && nh->prepend_len) {
char *prepend_hex = "AABBCCDDEE";
xo_emit(" {:nhop-prepend/%*s}", wid_prepend, prepend_hex);
}
xo_emit("\n");
xo_close_instance(name);
}
static int
cmp_nh_idx(const void *_a, const void *_b)
{
const struct nhops_map *a, *b;
a = _a;
b = _b;
if (a->idx > b->idx)
return (1);
else if (a->idx < b->idx)
return (-1);
return (0);
}
Introduce scalable route multipath. This change is based on the nexthop objects landed in D24232. The change introduces the concept of nexthop groups. Each group contains the collection of nexthops with their relative weights and a dataplane-optimized structure to enable efficient nexthop selection. Simular to the nexthops, nexthop groups are immutable. Dataplane part gets compiled during group creation and is basically an array of nexthop pointers, compiled w.r.t their weights. With this change, `rt_nhop` field of `struct rtentry` contains either nexthop or nexthop group. They are distinguished by the presense of NHF_MULTIPATH flag. All dataplane lookup functions returns pointer to the nexthop object, leaving nexhop groups details inside routing subsystem. User-visible changes: The change is intended to be backward-compatible: all non-mpath operations should work as before with ROUTE_MPATH and net.route.multipath=1. All routes now comes with weight, default weight is 1, maximum is 2^24-1. Current maximum multipath group width is statically set to 64. This will become sysctl-tunable in the followup changes. Using functionality: * Recompile kernel with ROUTE_MPATH * set net.route.multipath to 1 route add -6 2001:db8::/32 2001:db8::2 -weight 10 route add -6 2001:db8::/32 2001:db8::3 -weight 20 netstat -6On Nexthop groups data Internet6: GrpIdx NhIdx Weight Slots Gateway Netif Refcnt 1 ------- ------- ------- --------------------------------------- --------- 1 13 10 1 2001:db8::2 vlan2 14 20 2 2001:db8::3 vlan2 Next steps: * Land outbound hashing for locally-originated routes ( D26523 ). * Fix net/bird multipath (net/frr seems to work fine) * Add ROUTE_MPATH to GENERIC * Set net.route.multipath=1 by default Tested by: olivier Reviewed by: glebius Relnotes: yes Differential Revision: https://reviews.freebsd.org/D26449
2020-10-03 10:47:17 +00:00
void
dump_nhops_sysctl(int fibnum, int af, struct nhops_dump *nd)
Introduce nexthop objects and new routing KPI. This is the foundational change for the routing subsytem rearchitecture. More details and goals are available in https://reviews.freebsd.org/D24141 . This patch introduces concept of nexthop objects and new nexthop-based routing KPI. Nexthops are objects, containing all necessary information for performing the packet output decision. Output interface, mtu, flags, gw address goes there. For most of the cases, these objects will serve the same role as the struct rtentry is currently serving. Typically there will be low tens of such objects for the router even with multiple BGP full-views, as these objects will be shared between routing entries. This allows to store more information in the nexthop. New KPI: struct nhop_object *fib4_lookup(uint32_t fibnum, struct in_addr dst, uint32_t scopeid, uint32_t flags, uint32_t flowid); struct nhop_object *fib6_lookup(uint32_t fibnum, const struct in6_addr *dst6, uint32_t scopeid, uint32_t flags, uint32_t flowid); These 2 function are intended to replace all all flavours of <in_|in6_>rtalloc[1]<_ign><_fib>, mpath functions and the previous fib[46]-generation functions. Upon successful lookup, they return nexthop object which is guaranteed to exist within current NET_EPOCH. If longer lifetime is desired, one can specify NHR_REF as a flag and get a referenced version of the nexthop. Reference semantic closely resembles rtentry one, allowing sed-style conversion. Additionally, another 2 functions are introduced to support uRPF functionality inside variety of our firewalls. Their primary goal is to hide the multipath implementation details inside the routing subsystem, greatly simplifying firewalls implementation: int fib4_lookup_urpf(uint32_t fibnum, struct in_addr dst, uint32_t scopeid, uint32_t flags, const struct ifnet *src_if); int fib6_lookup_urpf(uint32_t fibnum, const struct in6_addr *dst6, uint32_t scopeid, uint32_t flags, const struct ifnet *src_if); All functions have a separate scopeid argument, paving way to eliminating IPv6 scope embedding and allowing to support IPv4 link-locals in the future. Structure changes: * rtentry gets new 'rt_nhop' pointer, slightly growing the overall size. * rib_head gets new 'rnh_preadd' callback pointer, slightly growing overall sz. Old KPI: During the transition state old and new KPI will coexists. As there are another 4-5 decent-sized conversion patches, it will probably take a couple of weeks. To support both KPIs, fields not required by the new KPI (most of rtentry) has to be kept, resulting in the temporary size increase. Once conversion is finished, rtentry will notably shrink. More details: * architectural overview: https://reviews.freebsd.org/D24141 * list of the next changes: https://reviews.freebsd.org/D24232 Reviewed by: ae,glebius(initial version) Differential Revision: https://reviews.freebsd.org/D24232
2020-04-12 14:30:00 +00:00
{
size_t needed;
int mib[7];
char *buf, *next, *lim;
struct rt_msghdr *rtm;
struct nhop_external *nh;
struct nhops_map *nh_map;
size_t nh_count, nh_size;
mib[0] = CTL_NET;
mib[1] = PF_ROUTE;
mib[2] = 0;
mib[3] = af;
mib[4] = NET_RT_NHOP;
mib[5] = 0;
mib[6] = fibnum;
if (sysctl(mib, nitems(mib), NULL, &needed, NULL, 0) < 0)
err(EX_OSERR, "sysctl: net.route.0.%d.nhdump.%d estimate", af,
fibnum);
if ((buf = malloc(needed)) == NULL)
errx(2, "malloc(%lu)", (unsigned long)needed);
if (sysctl(mib, nitems(mib), buf, &needed, NULL, 0) < 0)
err(1, "sysctl: net.route.0.%d.nhdump.%d", af, fibnum);
lim = buf + needed;
/*
* nexhops are received unsorted. Collect everything first, sort and then display
* sorted.
*/
nh_count = 0;
nh_size = 16;
nh_map = calloc(nh_size, sizeof(struct nhops_map));
for (next = buf; next < lim; next += rtm->rtm_msglen) {
rtm = (struct rt_msghdr *)next;
if (rtm->rtm_version != RTM_VERSION)
continue;
if (nh_count >= nh_size) {
nh_size *= 2;
nh_map = realloc(nh_map, nh_size * sizeof(struct nhops_map));
}
nh = (struct nhop_external *)(rtm + 1);
nh_map[nh_count].idx = nh->nh_idx;
nh_map[nh_count].rtm = rtm;
nh_count++;
}
Introduce scalable route multipath. This change is based on the nexthop objects landed in D24232. The change introduces the concept of nexthop groups. Each group contains the collection of nexthops with their relative weights and a dataplane-optimized structure to enable efficient nexthop selection. Simular to the nexthops, nexthop groups are immutable. Dataplane part gets compiled during group creation and is basically an array of nexthop pointers, compiled w.r.t their weights. With this change, `rt_nhop` field of `struct rtentry` contains either nexthop or nexthop group. They are distinguished by the presense of NHF_MULTIPATH flag. All dataplane lookup functions returns pointer to the nexthop object, leaving nexhop groups details inside routing subsystem. User-visible changes: The change is intended to be backward-compatible: all non-mpath operations should work as before with ROUTE_MPATH and net.route.multipath=1. All routes now comes with weight, default weight is 1, maximum is 2^24-1. Current maximum multipath group width is statically set to 64. This will become sysctl-tunable in the followup changes. Using functionality: * Recompile kernel with ROUTE_MPATH * set net.route.multipath to 1 route add -6 2001:db8::/32 2001:db8::2 -weight 10 route add -6 2001:db8::/32 2001:db8::3 -weight 20 netstat -6On Nexthop groups data Internet6: GrpIdx NhIdx Weight Slots Gateway Netif Refcnt 1 ------- ------- ------- --------------------------------------- --------- 1 13 10 1 2001:db8::2 vlan2 14 20 2 2001:db8::3 vlan2 Next steps: * Land outbound hashing for locally-originated routes ( D26523 ). * Fix net/bird multipath (net/frr seems to work fine) * Add ROUTE_MPATH to GENERIC * Set net.route.multipath=1 by default Tested by: olivier Reviewed by: glebius Relnotes: yes Differential Revision: https://reviews.freebsd.org/D26449
2020-10-03 10:47:17 +00:00
if (nh_count > 0)
Introduce nexthop objects and new routing KPI. This is the foundational change for the routing subsytem rearchitecture. More details and goals are available in https://reviews.freebsd.org/D24141 . This patch introduces concept of nexthop objects and new nexthop-based routing KPI. Nexthops are objects, containing all necessary information for performing the packet output decision. Output interface, mtu, flags, gw address goes there. For most of the cases, these objects will serve the same role as the struct rtentry is currently serving. Typically there will be low tens of such objects for the router even with multiple BGP full-views, as these objects will be shared between routing entries. This allows to store more information in the nexthop. New KPI: struct nhop_object *fib4_lookup(uint32_t fibnum, struct in_addr dst, uint32_t scopeid, uint32_t flags, uint32_t flowid); struct nhop_object *fib6_lookup(uint32_t fibnum, const struct in6_addr *dst6, uint32_t scopeid, uint32_t flags, uint32_t flowid); These 2 function are intended to replace all all flavours of <in_|in6_>rtalloc[1]<_ign><_fib>, mpath functions and the previous fib[46]-generation functions. Upon successful lookup, they return nexthop object which is guaranteed to exist within current NET_EPOCH. If longer lifetime is desired, one can specify NHR_REF as a flag and get a referenced version of the nexthop. Reference semantic closely resembles rtentry one, allowing sed-style conversion. Additionally, another 2 functions are introduced to support uRPF functionality inside variety of our firewalls. Their primary goal is to hide the multipath implementation details inside the routing subsystem, greatly simplifying firewalls implementation: int fib4_lookup_urpf(uint32_t fibnum, struct in_addr dst, uint32_t scopeid, uint32_t flags, const struct ifnet *src_if); int fib6_lookup_urpf(uint32_t fibnum, const struct in6_addr *dst6, uint32_t scopeid, uint32_t flags, const struct ifnet *src_if); All functions have a separate scopeid argument, paving way to eliminating IPv6 scope embedding and allowing to support IPv4 link-locals in the future. Structure changes: * rtentry gets new 'rt_nhop' pointer, slightly growing the overall size. * rib_head gets new 'rnh_preadd' callback pointer, slightly growing overall sz. Old KPI: During the transition state old and new KPI will coexists. As there are another 4-5 decent-sized conversion patches, it will probably take a couple of weeks. To support both KPIs, fields not required by the new KPI (most of rtentry) has to be kept, resulting in the temporary size increase. Once conversion is finished, rtentry will notably shrink. More details: * architectural overview: https://reviews.freebsd.org/D24141 * list of the next changes: https://reviews.freebsd.org/D24232 Reviewed by: ae,glebius(initial version) Differential Revision: https://reviews.freebsd.org/D24232
2020-04-12 14:30:00 +00:00
qsort(nh_map, nh_count, sizeof(struct nhops_map), cmp_nh_idx);
Introduce scalable route multipath. This change is based on the nexthop objects landed in D24232. The change introduces the concept of nexthop groups. Each group contains the collection of nexthops with their relative weights and a dataplane-optimized structure to enable efficient nexthop selection. Simular to the nexthops, nexthop groups are immutable. Dataplane part gets compiled during group creation and is basically an array of nexthop pointers, compiled w.r.t their weights. With this change, `rt_nhop` field of `struct rtentry` contains either nexthop or nexthop group. They are distinguished by the presense of NHF_MULTIPATH flag. All dataplane lookup functions returns pointer to the nexthop object, leaving nexhop groups details inside routing subsystem. User-visible changes: The change is intended to be backward-compatible: all non-mpath operations should work as before with ROUTE_MPATH and net.route.multipath=1. All routes now comes with weight, default weight is 1, maximum is 2^24-1. Current maximum multipath group width is statically set to 64. This will become sysctl-tunable in the followup changes. Using functionality: * Recompile kernel with ROUTE_MPATH * set net.route.multipath to 1 route add -6 2001:db8::/32 2001:db8::2 -weight 10 route add -6 2001:db8::/32 2001:db8::3 -weight 20 netstat -6On Nexthop groups data Internet6: GrpIdx NhIdx Weight Slots Gateway Netif Refcnt 1 ------- ------- ------- --------------------------------------- --------- 1 13 10 1 2001:db8::2 vlan2 14 20 2 2001:db8::3 vlan2 Next steps: * Land outbound hashing for locally-originated routes ( D26523 ). * Fix net/bird multipath (net/frr seems to work fine) * Add ROUTE_MPATH to GENERIC * Set net.route.multipath=1 by default Tested by: olivier Reviewed by: glebius Relnotes: yes Differential Revision: https://reviews.freebsd.org/D26449
2020-10-03 10:47:17 +00:00
nd->nh_buf = buf;
nd->nh_count = nh_count;
nd->nh_map = nh_map;
}
static void
print_nhops_sysctl(int fibnum, int af)
{
struct nhops_dump nd;
struct nhop_external *nh;
int fam;
struct rt_msghdr *rtm;
dump_nhops_sysctl(fibnum, af, &nd);
xo_open_container("nhop-table");
xo_open_list("rt-family");
if (nd.nh_count > 0) {
nh = (struct nhop_external *)(nd.nh_map[0].rtm + 1);
Introduce nexthop objects and new routing KPI. This is the foundational change for the routing subsytem rearchitecture. More details and goals are available in https://reviews.freebsd.org/D24141 . This patch introduces concept of nexthop objects and new nexthop-based routing KPI. Nexthops are objects, containing all necessary information for performing the packet output decision. Output interface, mtu, flags, gw address goes there. For most of the cases, these objects will serve the same role as the struct rtentry is currently serving. Typically there will be low tens of such objects for the router even with multiple BGP full-views, as these objects will be shared between routing entries. This allows to store more information in the nexthop. New KPI: struct nhop_object *fib4_lookup(uint32_t fibnum, struct in_addr dst, uint32_t scopeid, uint32_t flags, uint32_t flowid); struct nhop_object *fib6_lookup(uint32_t fibnum, const struct in6_addr *dst6, uint32_t scopeid, uint32_t flags, uint32_t flowid); These 2 function are intended to replace all all flavours of <in_|in6_>rtalloc[1]<_ign><_fib>, mpath functions and the previous fib[46]-generation functions. Upon successful lookup, they return nexthop object which is guaranteed to exist within current NET_EPOCH. If longer lifetime is desired, one can specify NHR_REF as a flag and get a referenced version of the nexthop. Reference semantic closely resembles rtentry one, allowing sed-style conversion. Additionally, another 2 functions are introduced to support uRPF functionality inside variety of our firewalls. Their primary goal is to hide the multipath implementation details inside the routing subsystem, greatly simplifying firewalls implementation: int fib4_lookup_urpf(uint32_t fibnum, struct in_addr dst, uint32_t scopeid, uint32_t flags, const struct ifnet *src_if); int fib6_lookup_urpf(uint32_t fibnum, const struct in6_addr *dst6, uint32_t scopeid, uint32_t flags, const struct ifnet *src_if); All functions have a separate scopeid argument, paving way to eliminating IPv6 scope embedding and allowing to support IPv4 link-locals in the future. Structure changes: * rtentry gets new 'rt_nhop' pointer, slightly growing the overall size. * rib_head gets new 'rnh_preadd' callback pointer, slightly growing overall sz. Old KPI: During the transition state old and new KPI will coexists. As there are another 4-5 decent-sized conversion patches, it will probably take a couple of weeks. To support both KPIs, fields not required by the new KPI (most of rtentry) has to be kept, resulting in the temporary size increase. Once conversion is finished, rtentry will notably shrink. More details: * architectural overview: https://reviews.freebsd.org/D24141 * list of the next changes: https://reviews.freebsd.org/D24232 Reviewed by: ae,glebius(initial version) Differential Revision: https://reviews.freebsd.org/D24232
2020-04-12 14:30:00 +00:00
fam = nh->nh_family;
wid_dst = WID_GW_DEFAULT(fam);
wid_gw = WID_GW_DEFAULT(fam);
wid_nhidx = 5;
wid_nhtype = 12;
wid_refcnt = 6;
wid_flags = 6;
wid_pksent = 8;
wid_mtu = 6;
wid_if = WID_IF_DEFAULT(fam);
xo_open_instance("rt-family");
pr_family(fam);
xo_open_list("nh-entry");
print_nhop_header(fam);
Introduce scalable route multipath. This change is based on the nexthop objects landed in D24232. The change introduces the concept of nexthop groups. Each group contains the collection of nexthops with their relative weights and a dataplane-optimized structure to enable efficient nexthop selection. Simular to the nexthops, nexthop groups are immutable. Dataplane part gets compiled during group creation and is basically an array of nexthop pointers, compiled w.r.t their weights. With this change, `rt_nhop` field of `struct rtentry` contains either nexthop or nexthop group. They are distinguished by the presense of NHF_MULTIPATH flag. All dataplane lookup functions returns pointer to the nexthop object, leaving nexhop groups details inside routing subsystem. User-visible changes: The change is intended to be backward-compatible: all non-mpath operations should work as before with ROUTE_MPATH and net.route.multipath=1. All routes now comes with weight, default weight is 1, maximum is 2^24-1. Current maximum multipath group width is statically set to 64. This will become sysctl-tunable in the followup changes. Using functionality: * Recompile kernel with ROUTE_MPATH * set net.route.multipath to 1 route add -6 2001:db8::/32 2001:db8::2 -weight 10 route add -6 2001:db8::/32 2001:db8::3 -weight 20 netstat -6On Nexthop groups data Internet6: GrpIdx NhIdx Weight Slots Gateway Netif Refcnt 1 ------- ------- ------- --------------------------------------- --------- 1 13 10 1 2001:db8::2 vlan2 14 20 2 2001:db8::3 vlan2 Next steps: * Land outbound hashing for locally-originated routes ( D26523 ). * Fix net/bird multipath (net/frr seems to work fine) * Add ROUTE_MPATH to GENERIC * Set net.route.multipath=1 by default Tested by: olivier Reviewed by: glebius Relnotes: yes Differential Revision: https://reviews.freebsd.org/D26449
2020-10-03 10:47:17 +00:00
for (size_t i = 0; i < nd.nh_count; i++) {
rtm = nd.nh_map[i].rtm;
Introduce nexthop objects and new routing KPI. This is the foundational change for the routing subsytem rearchitecture. More details and goals are available in https://reviews.freebsd.org/D24141 . This patch introduces concept of nexthop objects and new nexthop-based routing KPI. Nexthops are objects, containing all necessary information for performing the packet output decision. Output interface, mtu, flags, gw address goes there. For most of the cases, these objects will serve the same role as the struct rtentry is currently serving. Typically there will be low tens of such objects for the router even with multiple BGP full-views, as these objects will be shared between routing entries. This allows to store more information in the nexthop. New KPI: struct nhop_object *fib4_lookup(uint32_t fibnum, struct in_addr dst, uint32_t scopeid, uint32_t flags, uint32_t flowid); struct nhop_object *fib6_lookup(uint32_t fibnum, const struct in6_addr *dst6, uint32_t scopeid, uint32_t flags, uint32_t flowid); These 2 function are intended to replace all all flavours of <in_|in6_>rtalloc[1]<_ign><_fib>, mpath functions and the previous fib[46]-generation functions. Upon successful lookup, they return nexthop object which is guaranteed to exist within current NET_EPOCH. If longer lifetime is desired, one can specify NHR_REF as a flag and get a referenced version of the nexthop. Reference semantic closely resembles rtentry one, allowing sed-style conversion. Additionally, another 2 functions are introduced to support uRPF functionality inside variety of our firewalls. Their primary goal is to hide the multipath implementation details inside the routing subsystem, greatly simplifying firewalls implementation: int fib4_lookup_urpf(uint32_t fibnum, struct in_addr dst, uint32_t scopeid, uint32_t flags, const struct ifnet *src_if); int fib6_lookup_urpf(uint32_t fibnum, const struct in6_addr *dst6, uint32_t scopeid, uint32_t flags, const struct ifnet *src_if); All functions have a separate scopeid argument, paving way to eliminating IPv6 scope embedding and allowing to support IPv4 link-locals in the future. Structure changes: * rtentry gets new 'rt_nhop' pointer, slightly growing the overall size. * rib_head gets new 'rnh_preadd' callback pointer, slightly growing overall sz. Old KPI: During the transition state old and new KPI will coexists. As there are another 4-5 decent-sized conversion patches, it will probably take a couple of weeks. To support both KPIs, fields not required by the new KPI (most of rtentry) has to be kept, resulting in the temporary size increase. Once conversion is finished, rtentry will notably shrink. More details: * architectural overview: https://reviews.freebsd.org/D24141 * list of the next changes: https://reviews.freebsd.org/D24232 Reviewed by: ae,glebius(initial version) Differential Revision: https://reviews.freebsd.org/D24232
2020-04-12 14:30:00 +00:00
nh = (struct nhop_external *)(rtm + 1);
print_nhop_entry_sysctl("nh-entry", rtm, nh);
}
xo_close_list("nh-entry");
xo_close_instance("rt-family");
}
xo_close_list("rt-family");
xo_close_container("nhop-table");
Introduce scalable route multipath. This change is based on the nexthop objects landed in D24232. The change introduces the concept of nexthop groups. Each group contains the collection of nexthops with their relative weights and a dataplane-optimized structure to enable efficient nexthop selection. Simular to the nexthops, nexthop groups are immutable. Dataplane part gets compiled during group creation and is basically an array of nexthop pointers, compiled w.r.t their weights. With this change, `rt_nhop` field of `struct rtentry` contains either nexthop or nexthop group. They are distinguished by the presense of NHF_MULTIPATH flag. All dataplane lookup functions returns pointer to the nexthop object, leaving nexhop groups details inside routing subsystem. User-visible changes: The change is intended to be backward-compatible: all non-mpath operations should work as before with ROUTE_MPATH and net.route.multipath=1. All routes now comes with weight, default weight is 1, maximum is 2^24-1. Current maximum multipath group width is statically set to 64. This will become sysctl-tunable in the followup changes. Using functionality: * Recompile kernel with ROUTE_MPATH * set net.route.multipath to 1 route add -6 2001:db8::/32 2001:db8::2 -weight 10 route add -6 2001:db8::/32 2001:db8::3 -weight 20 netstat -6On Nexthop groups data Internet6: GrpIdx NhIdx Weight Slots Gateway Netif Refcnt 1 ------- ------- ------- --------------------------------------- --------- 1 13 10 1 2001:db8::2 vlan2 14 20 2 2001:db8::3 vlan2 Next steps: * Land outbound hashing for locally-originated routes ( D26523 ). * Fix net/bird multipath (net/frr seems to work fine) * Add ROUTE_MPATH to GENERIC * Set net.route.multipath=1 by default Tested by: olivier Reviewed by: glebius Relnotes: yes Differential Revision: https://reviews.freebsd.org/D26449
2020-10-03 10:47:17 +00:00
free(nd.nh_buf);
Introduce nexthop objects and new routing KPI. This is the foundational change for the routing subsytem rearchitecture. More details and goals are available in https://reviews.freebsd.org/D24141 . This patch introduces concept of nexthop objects and new nexthop-based routing KPI. Nexthops are objects, containing all necessary information for performing the packet output decision. Output interface, mtu, flags, gw address goes there. For most of the cases, these objects will serve the same role as the struct rtentry is currently serving. Typically there will be low tens of such objects for the router even with multiple BGP full-views, as these objects will be shared between routing entries. This allows to store more information in the nexthop. New KPI: struct nhop_object *fib4_lookup(uint32_t fibnum, struct in_addr dst, uint32_t scopeid, uint32_t flags, uint32_t flowid); struct nhop_object *fib6_lookup(uint32_t fibnum, const struct in6_addr *dst6, uint32_t scopeid, uint32_t flags, uint32_t flowid); These 2 function are intended to replace all all flavours of <in_|in6_>rtalloc[1]<_ign><_fib>, mpath functions and the previous fib[46]-generation functions. Upon successful lookup, they return nexthop object which is guaranteed to exist within current NET_EPOCH. If longer lifetime is desired, one can specify NHR_REF as a flag and get a referenced version of the nexthop. Reference semantic closely resembles rtentry one, allowing sed-style conversion. Additionally, another 2 functions are introduced to support uRPF functionality inside variety of our firewalls. Their primary goal is to hide the multipath implementation details inside the routing subsystem, greatly simplifying firewalls implementation: int fib4_lookup_urpf(uint32_t fibnum, struct in_addr dst, uint32_t scopeid, uint32_t flags, const struct ifnet *src_if); int fib6_lookup_urpf(uint32_t fibnum, const struct in6_addr *dst6, uint32_t scopeid, uint32_t flags, const struct ifnet *src_if); All functions have a separate scopeid argument, paving way to eliminating IPv6 scope embedding and allowing to support IPv4 link-locals in the future. Structure changes: * rtentry gets new 'rt_nhop' pointer, slightly growing the overall size. * rib_head gets new 'rnh_preadd' callback pointer, slightly growing overall sz. Old KPI: During the transition state old and new KPI will coexists. As there are another 4-5 decent-sized conversion patches, it will probably take a couple of weeks. To support both KPIs, fields not required by the new KPI (most of rtentry) has to be kept, resulting in the temporary size increase. Once conversion is finished, rtentry will notably shrink. More details: * architectural overview: https://reviews.freebsd.org/D24141 * list of the next changes: https://reviews.freebsd.org/D24232 Reviewed by: ae,glebius(initial version) Differential Revision: https://reviews.freebsd.org/D24232
2020-04-12 14:30:00 +00:00
}
static void
p_nhflags(int f, const char *format)
{
struct bits *p;
char *pretty_name = "nh_flags_pretty";
xo_emit(format, fmt_flags(nh_bits, f));
xo_open_list(pretty_name);
for (p = nh_bits; p->b_mask; p++)
if (p->b_mask & f)
xo_emit("{le:nh_flags_pretty/%s}", p->b_name);
xo_close_list(pretty_name);
}
void
nhops_print(int fibnum, int af)
{
size_t intsize;
int numfibs;
intsize = sizeof(int);
if (fibnum == -1 &&
sysctlbyname("net.my_fibnum", &fibnum, &intsize, NULL, 0) == -1)
fibnum = 0;
if (sysctlbyname("net.fibs", &numfibs, &intsize, NULL, 0) == -1)
numfibs = 1;
if (fibnum < 0 || fibnum > numfibs - 1)
errx(EX_USAGE, "%d: invalid fib", fibnum);
ifmap = prepare_ifmap(&ifmap_size);
xo_open_container("route-nhop-information");
xo_emit("{T:Nexthop data}");
if (fibnum)
xo_emit(" ({L:fib}: {:fib/%d})", fibnum);
xo_emit("\n");
print_nhops_sysctl(fibnum, af);
xo_close_container("route-nhop-information");
}