2003-12-30 11:41:43 +00:00
|
|
|
/* $NetBSD: if_gre.c,v 1.49 2003/12/11 00:22:29 itojun Exp $ */
|
2002-09-06 17:12:50 +00:00
|
|
|
/* $FreeBSD$ */
|
|
|
|
|
|
2005-01-07 01:45:51 +00:00
|
|
|
/*-
|
2002-09-06 17:12:50 +00:00
|
|
|
* Copyright (c) 1998 The NetBSD Foundation, Inc.
|
|
|
|
|
* All rights reserved.
|
|
|
|
|
*
|
|
|
|
|
* This code is derived from software contributed to The NetBSD Foundation
|
|
|
|
|
* by Heiko W.Rupp <hwr@pilhuhn.de>
|
|
|
|
|
*
|
2005-08-01 08:14:21 +00:00
|
|
|
* IPv6-over-GRE contributed by Gert Doering <gert@greenie.muc.de>
|
|
|
|
|
*
|
2002-09-06 17:12:50 +00:00
|
|
|
* 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.
|
|
|
|
|
*
|
|
|
|
|
* THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. 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 FOUNDATION 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.
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Encapsulate L3 protocols into IP
|
2005-08-01 08:14:21 +00:00
|
|
|
* See RFC 2784 (successor of RFC 1701 and 1702) for more details.
|
2002-09-06 17:12:50 +00:00
|
|
|
* If_gre is compatible with Cisco GRE tunnels, so you can
|
|
|
|
|
* have a NetBSD box as the other end of a tunnel interface of a Cisco
|
|
|
|
|
* router. See gre(4) for more details.
|
|
|
|
|
* Also supported: IP in IP encaps (proto 55) as of RFC 2004
|
|
|
|
|
*/
|
|
|
|
|
|
2002-09-16 10:10:18 +00:00
|
|
|
#include "opt_atalk.h"
|
2002-09-06 17:12:50 +00:00
|
|
|
#include "opt_inet.h"
|
2003-11-14 20:58:00 +00:00
|
|
|
#include "opt_inet6.h"
|
2002-09-06 17:12:50 +00:00
|
|
|
|
|
|
|
|
#include <sys/param.h>
|
2011-03-02 21:39:08 +00:00
|
|
|
#include <sys/jail.h>
|
2002-09-06 17:12:50 +00:00
|
|
|
#include <sys/kernel.h>
|
2011-06-18 09:34:03 +00:00
|
|
|
#include <sys/libkern.h>
|
2002-09-06 17:12:50 +00:00
|
|
|
#include <sys/malloc.h>
|
2004-05-30 20:27:19 +00:00
|
|
|
#include <sys/module.h>
|
2002-09-06 17:12:50 +00:00
|
|
|
#include <sys/mbuf.h>
|
2006-11-06 13:42:10 +00:00
|
|
|
#include <sys/priv.h>
|
Add code to allow the system to handle multiple routing tables.
This particular implementation is designed to be fully backwards compatible
and to be MFC-able to 7.x (and 6.x)
Currently the only protocol that can make use of the multiple tables is IPv4
Similar functionality exists in OpenBSD and Linux.
From my notes:
-----
One thing where FreeBSD has been falling behind, and which by chance I
have some time to work on is "policy based routing", which allows
different
packet streams to be routed by more than just the destination address.
Constraints:
------------
I want to make some form of this available in the 6.x tree
(and by extension 7.x) , but FreeBSD in general needs it so I might as
well do it in -current and back port the portions I need.
One of the ways that this can be done is to have the ability to
instantiate multiple kernel routing tables (which I will now
refer to as "Forwarding Information Bases" or "FIBs" for political
correctness reasons). Which FIB a particular packet uses to make
the next hop decision can be decided by a number of mechanisms.
The policies these mechanisms implement are the "Policies" referred
to in "Policy based routing".
One of the constraints I have if I try to back port this work to
6.x is that it must be implemented as a EXTENSION to the existing
ABIs in 6.x so that third party applications do not need to be
recompiled in timespan of the branch.
This first version will not have some of the bells and whistles that
will come with later versions. It will, for example, be limited to 16
tables in the first commit.
Implementation method, Compatible version. (part 1)
-------------------------------
For this reason I have implemented a "sufficient subset" of a
multiple routing table solution in Perforce, and back-ported it
to 6.x. (also in Perforce though not always caught up with what I
have done in -current/P4). The subset allows a number of FIBs
to be defined at compile time (8 is sufficient for my purposes in 6.x)
and implements the changes needed to allow IPV4 to use them. I have not
done the changes for ipv6 simply because I do not need it, and I do not
have enough knowledge of ipv6 (e.g. neighbor discovery) needed to do it.
Other protocol families are left untouched and should there be
users with proprietary protocol families, they should continue to work
and be oblivious to the existence of the extra FIBs.
To understand how this is done, one must know that the current FIB
code starts everything off with a single dimensional array of
pointers to FIB head structures (One per protocol family), each of
which in turn points to the trie of routes available to that family.
The basic change in the ABI compatible version of the change is to
extent that array to be a 2 dimensional array, so that
instead of protocol family X looking at rt_tables[X] for the
table it needs, it looks at rt_tables[Y][X] when for all
protocol families except ipv4 Y is always 0.
Code that is unaware of the change always just sees the first row
of the table, which of course looks just like the one dimensional
array that existed before.
The entry points rtrequest(), rtalloc(), rtalloc1(), rtalloc_ign()
are all maintained, but refer only to the first row of the array,
so that existing callers in proprietary protocols can continue to
do the "right thing".
Some new entry points are added, for the exclusive use of ipv4 code
called in_rtrequest(), in_rtalloc(), in_rtalloc1() and in_rtalloc_ign(),
which have an extra argument which refers the code to the correct row.
In addition, there are some new entry points (currently called
rtalloc_fib() and friends) that check the Address family being
looked up and call either rtalloc() (and friends) if the protocol
is not IPv4 forcing the action to row 0 or to the appropriate row
if it IS IPv4 (and that info is available). These are for calling
from code that is not specific to any particular protocol. The way
these are implemented would change in the non ABI preserving code
to be added later.
One feature of the first version of the code is that for ipv4,
the interface routes show up automatically on all the FIBs, so
that no matter what FIB you select you always have the basic
direct attached hosts available to you. (rtinit() does this
automatically).
You CAN delete an interface route from one FIB should you want
to but by default it's there. ARP information is also available
in each FIB. It's assumed that the same machine would have the
same MAC address, regardless of which FIB you are using to get
to it.
This brings us as to how the correct FIB is selected for an outgoing
IPV4 packet.
Firstly, all packets have a FIB associated with them. if nothing
has been done to change it, it will be FIB 0. The FIB is changed
in the following ways.
Packets fall into one of a number of classes.
1/ locally generated packets, coming from a socket/PCB.
Such packets select a FIB from a number associated with the
socket/PCB. This in turn is inherited from the process,
but can be changed by a socket option. The process in turn
inherits it on fork. I have written a utility call setfib
that acts a bit like nice..
setfib -3 ping target.example.com # will use fib 3 for ping.
It is an obvious extension to make it a property of a jail
but I have not done so. It can be achieved by combining the setfib and
jail commands.
2/ packets received on an interface for forwarding.
By default these packets would use table 0,
(or possibly a number settable in a sysctl(not yet)).
but prior to routing the firewall can inspect them (see below).
(possibly in the future you may be able to associate a FIB
with packets received on an interface.. An ifconfig arg, but not yet.)
3/ packets inspected by a packet classifier, which can arbitrarily
associate a fib with it on a packet by packet basis.
A fib assigned to a packet by a packet classifier
(such as ipfw) would over-ride a fib associated by
a more default source. (such as cases 1 or 2).
4/ a tcp listen socket associated with a fib will generate
accept sockets that are associated with that same fib.
5/ Packets generated in response to some other packet (e.g. reset
or icmp packets). These should use the FIB associated with the
packet being reponded to.
6/ Packets generated during encapsulation.
gif, tun and other tunnel interfaces will encapsulate using the FIB
that was in effect withthe proces that set up the tunnel.
thus setfib 1 ifconfig gif0 [tunnel instructions]
will set the fib for the tunnel to use to be fib 1.
Routing messages would be associated with their
process, and thus select one FIB or another.
messages from the kernel would be associated with the fib they
refer to and would only be received by a routing socket associated
with that fib. (not yet implemented)
In addition Netstat has been edited to be able to cope with the
fact that the array is now 2 dimensional. (It looks in system
memory using libkvm (!)). Old versions of netstat see only the first FIB.
In addition two sysctls are added to give:
a) the number of FIBs compiled in (active)
b) the default FIB of the calling process.
Early testing experience:
-------------------------
Basically our (IronPort's) appliance does this functionality already
using ipfw fwd but that method has some drawbacks.
For example,
It can't fully simulate a routing table because it can't influence the
socket's choice of local address when a connect() is done.
Testing during the generating of these changes has been
remarkably smooth so far. Multiple tables have co-existed
with no notable side effects, and packets have been routes
accordingly.
ipfw has grown 2 new keywords:
setfib N ip from anay to any
count ip from any to any fib N
In pf there seems to be a requirement to be able to give symbolic names to the
fibs but I do not have that capacity. I am not sure if it is required.
SCTP has interestingly enough built in support for this, called VRFs
in Cisco parlance. it will be interesting to see how that handles it
when it suddenly actually does something.
Where to next:
--------------------
After committing the ABI compatible version and MFCing it, I'd
like to proceed in a forward direction in -current. this will
result in some roto-tilling in the routing code.
Firstly: the current code's idea of having a separate tree per
protocol family, all of the same format, and pointed to by the
1 dimensional array is a bit silly. Especially when one considers that
there is code that makes assumptions about every protocol having the
same internal structures there. Some protocols don't WANT that
sort of structure. (for example the whole idea of a netmask is foreign
to appletalk). This needs to be made opaque to the external code.
My suggested first change is to add routing method pointers to the
'domain' structure, along with information pointing the data.
instead of having an array of pointers to uniform structures,
there would be an array pointing to the 'domain' structures
for each protocol address domain (protocol family),
and the methods this reached would be called. The methods would have
an argument that gives FIB number, but the protocol would be free
to ignore it.
When the ABI can be changed it raises the possibilty of the
addition of a fib entry into the "struct route". Currently,
the structure contains the sockaddr of the desination, and the resulting
fib entry. To make this work fully, one could add a fib number
so that given an address and a fib, one can find the third element, the
fib entry.
Interaction with the ARP layer/ LL layer would need to be
revisited as well. Qing Li has been working on this already.
This work was sponsored by Ironport Systems/Cisco
Reviewed by: several including rwatson, bz and mlair (parts each)
Obtained from: Ironport systems/Cisco
2008-05-09 23:03:00 +00:00
|
|
|
#include <sys/proc.h>
|
2002-09-06 17:12:50 +00:00
|
|
|
#include <sys/protosw.h>
|
|
|
|
|
#include <sys/socket.h>
|
|
|
|
|
#include <sys/sockio.h>
|
|
|
|
|
#include <sys/sysctl.h>
|
2002-09-15 13:54:12 +00:00
|
|
|
#include <sys/systm.h>
|
2002-09-06 17:12:50 +00:00
|
|
|
|
|
|
|
|
#include <net/ethernet.h>
|
|
|
|
|
#include <net/if.h>
|
Major overhaul of pseudo-interface cloning. Highlights include:
- Split the code out into if_clone.[ch].
- Locked struct if_clone. [1]
- Add a per-cloner match function rather then simply matching names of
the form <name><unit> and <name>.
- Use the match function to allow creation of <interface>.<tag>
vlan interfaces. The old way is preserved unchanged!
- Also the match function to allow creation of stf(4) interfaces named
stf0, stf, or 6to4. This is the only major user visible change in
that "ifconfig stf" creates the interface stf rather then stf0 and
does not print "stf0" to stdout.
- Allow destroy functions to fail so they can refuse to delete
interfaces. Currently, we forbid the deletion of interfaces which
were created in the init function, particularly lo0, pflog0, and
pfsync0. In the case of lo0 this was a panic implementation so it
does not count as a user visiable change. :-)
- Since most interfaces do not need the new functionality, an family of
wrapper functions, ifc_simple_*(), were created to wrap old style
cloner functions.
- The IF_CLONE_INITIALIZER macro is replaced with a new incompatible
IFC_CLONE_INITIALIZER and ifc_simple consumers use IFC_SIMPLE_DECLARE
instead.
Submitted by: Maurycy Pawlowski-Wieronski <maurycy at fouk.org> [1]
Reviewed by: andre, mlaier
Discussed on: net
2004-06-22 20:13:25 +00:00
|
|
|
#include <net/if_clone.h>
|
2002-09-06 17:12:50 +00:00
|
|
|
#include <net/if_types.h>
|
|
|
|
|
#include <net/route.h>
|
2009-08-01 19:26:27 +00:00
|
|
|
#include <net/vnet.h>
|
2002-09-06 17:12:50 +00:00
|
|
|
|
|
|
|
|
#ifdef INET
|
|
|
|
|
#include <netinet/in.h>
|
|
|
|
|
#include <netinet/in_systm.h>
|
|
|
|
|
#include <netinet/in_var.h>
|
|
|
|
|
#include <netinet/ip.h>
|
|
|
|
|
#include <netinet/ip_gre.h>
|
|
|
|
|
#include <netinet/ip_var.h>
|
|
|
|
|
#include <netinet/ip_encap.h>
|
|
|
|
|
#else
|
|
|
|
|
#error "Huh? if_gre without inet?"
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
#include <net/bpf.h>
|
|
|
|
|
|
|
|
|
|
#include <net/if_gre.h>
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* It is not easy to calculate the right value for a GRE MTU.
|
|
|
|
|
* We leave this task to the admin and use the same default that
|
|
|
|
|
* other vendors use.
|
|
|
|
|
*/
|
|
|
|
|
#define GREMTU 1476
|
|
|
|
|
|
2011-06-18 09:34:03 +00:00
|
|
|
#define MTAG_COOKIE_GRE 1307983903
|
|
|
|
|
#define MTAG_GRE_NESTING 1
|
|
|
|
|
struct mtag_gre_nesting {
|
|
|
|
|
uint16_t count;
|
|
|
|
|
uint16_t max;
|
|
|
|
|
struct ifnet *ifp[];
|
|
|
|
|
};
|
|
|
|
|
|
2004-03-22 16:04:43 +00:00
|
|
|
/*
|
|
|
|
|
* gre_mtx protects all global variables in if_gre.c.
|
|
|
|
|
* XXX: gre_softc data not protected yet.
|
|
|
|
|
*/
|
|
|
|
|
struct mtx gre_mtx;
|
2012-10-16 13:37:54 +00:00
|
|
|
static const char grename[] = "gre";
|
|
|
|
|
static MALLOC_DEFINE(M_GRE, grename, "Generic Routing Encapsulation");
|
2002-09-06 17:12:50 +00:00
|
|
|
|
|
|
|
|
struct gre_softc_head gre_softc_list;
|
|
|
|
|
|
2006-07-09 06:04:01 +00:00
|
|
|
static int gre_clone_create(struct if_clone *, int, caddr_t);
|
2002-10-16 22:26:32 +00:00
|
|
|
static void gre_clone_destroy(struct ifnet *);
|
2012-10-16 13:37:54 +00:00
|
|
|
static struct if_clone *gre_cloner;
|
|
|
|
|
|
2002-09-06 18:16:03 +00:00
|
|
|
static int gre_ioctl(struct ifnet *, u_long, caddr_t);
|
|
|
|
|
static int gre_output(struct ifnet *, struct mbuf *, struct sockaddr *,
|
2009-04-16 20:30:28 +00:00
|
|
|
struct route *ro);
|
2002-09-06 17:12:50 +00:00
|
|
|
|
2002-09-06 18:16:03 +00:00
|
|
|
static int gre_compute_route(struct gre_softc *sc);
|
2002-09-06 17:12:50 +00:00
|
|
|
|
2002-10-16 22:26:32 +00:00
|
|
|
static void greattach(void);
|
2002-09-06 17:12:50 +00:00
|
|
|
|
|
|
|
|
#ifdef INET
|
|
|
|
|
extern struct domain inetdomain;
|
2005-11-09 13:29:16 +00:00
|
|
|
static const struct protosw in_gre_protosw = {
|
|
|
|
|
.pr_type = SOCK_RAW,
|
|
|
|
|
.pr_domain = &inetdomain,
|
|
|
|
|
.pr_protocol = IPPROTO_GRE,
|
|
|
|
|
.pr_flags = PR_ATOMIC|PR_ADDR,
|
2006-01-21 10:44:34 +00:00
|
|
|
.pr_input = gre_input,
|
2005-11-09 13:29:16 +00:00
|
|
|
.pr_output = (pr_output_t *)rip_output,
|
|
|
|
|
.pr_ctlinput = rip_ctlinput,
|
|
|
|
|
.pr_ctloutput = rip_ctloutput,
|
|
|
|
|
.pr_usrreqs = &rip_usrreqs
|
2002-09-06 17:12:50 +00:00
|
|
|
};
|
2005-11-09 13:29:16 +00:00
|
|
|
static const struct protosw in_mobile_protosw = {
|
|
|
|
|
.pr_type = SOCK_RAW,
|
|
|
|
|
.pr_domain = &inetdomain,
|
|
|
|
|
.pr_protocol = IPPROTO_MOBILE,
|
|
|
|
|
.pr_flags = PR_ATOMIC|PR_ADDR,
|
2006-01-21 10:44:34 +00:00
|
|
|
.pr_input = gre_mobile_input,
|
2005-11-09 13:29:16 +00:00
|
|
|
.pr_output = (pr_output_t *)rip_output,
|
|
|
|
|
.pr_ctlinput = rip_ctlinput,
|
|
|
|
|
.pr_ctloutput = rip_ctloutput,
|
|
|
|
|
.pr_usrreqs = &rip_usrreqs
|
2002-09-06 17:12:50 +00:00
|
|
|
};
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
SYSCTL_DECL(_net_link);
|
2011-11-07 15:43:11 +00:00
|
|
|
static SYSCTL_NODE(_net_link, IFT_TUNNEL, gre, CTLFLAG_RW, 0,
|
2002-09-06 17:12:50 +00:00
|
|
|
"Generic Routing Encapsulation");
|
|
|
|
|
#ifndef MAX_GRE_NEST
|
|
|
|
|
/*
|
|
|
|
|
* This macro controls the default upper limitation on nesting of gre tunnels.
|
|
|
|
|
* Since, setting a large value to this macro with a careless configuration
|
|
|
|
|
* may introduce system crash, we don't allow any nestings by default.
|
|
|
|
|
* If you need to configure nested gre tunnels, you can define this macro
|
|
|
|
|
* in your kernel configuration file. However, if you do so, please be
|
|
|
|
|
* careful to configure the tunnels so that it won't make a loop.
|
|
|
|
|
*/
|
|
|
|
|
#define MAX_GRE_NEST 1
|
|
|
|
|
#endif
|
|
|
|
|
static int max_gre_nesting = MAX_GRE_NEST;
|
|
|
|
|
SYSCTL_INT(_net_link_gre, OID_AUTO, max_nesting, CTLFLAG_RW,
|
|
|
|
|
&max_gre_nesting, 0, "Max nested tunnels");
|
|
|
|
|
|
|
|
|
|
/* ARGSUSED */
|
2002-09-06 18:16:03 +00:00
|
|
|
static void
|
2002-09-06 17:12:50 +00:00
|
|
|
greattach(void)
|
|
|
|
|
{
|
|
|
|
|
|
2004-03-22 16:04:43 +00:00
|
|
|
mtx_init(&gre_mtx, "gre_mtx", NULL, MTX_DEF);
|
2002-09-06 17:12:50 +00:00
|
|
|
LIST_INIT(&gre_softc_list);
|
2012-10-16 13:37:54 +00:00
|
|
|
gre_cloner = if_clone_simple(grename, gre_clone_create,
|
|
|
|
|
gre_clone_destroy, 0);
|
2002-09-06 17:12:50 +00:00
|
|
|
}
|
|
|
|
|
|
2002-09-06 18:16:03 +00:00
|
|
|
static int
|
2006-07-09 06:04:01 +00:00
|
|
|
gre_clone_create(ifc, unit, params)
|
2002-09-06 17:12:50 +00:00
|
|
|
struct if_clone *ifc;
|
|
|
|
|
int unit;
|
2006-07-09 06:04:01 +00:00
|
|
|
caddr_t params;
|
2002-09-06 17:12:50 +00:00
|
|
|
{
|
|
|
|
|
struct gre_softc *sc;
|
|
|
|
|
|
2004-07-06 03:28:24 +00:00
|
|
|
sc = malloc(sizeof(struct gre_softc), M_GRE, M_WAITOK | M_ZERO);
|
2002-09-06 17:12:50 +00:00
|
|
|
|
2005-06-28 06:55:45 +00:00
|
|
|
GRE2IFP(sc) = if_alloc(IFT_TUNNEL);
|
|
|
|
|
if (GRE2IFP(sc) == NULL) {
|
|
|
|
|
free(sc, M_GRE);
|
|
|
|
|
return (ENOSPC);
|
|
|
|
|
}
|
|
|
|
|
|
2005-06-10 16:49:24 +00:00
|
|
|
GRE2IFP(sc)->if_softc = sc;
|
2012-10-16 13:37:54 +00:00
|
|
|
if_initname(GRE2IFP(sc), grename, unit);
|
2005-06-28 06:55:45 +00:00
|
|
|
|
2010-05-03 07:32:50 +00:00
|
|
|
GRE2IFP(sc)->if_snd.ifq_maxlen = ifqmaxlen;
|
2005-06-10 16:49:24 +00:00
|
|
|
GRE2IFP(sc)->if_addrlen = 0;
|
|
|
|
|
GRE2IFP(sc)->if_hdrlen = 24; /* IP + GRE */
|
|
|
|
|
GRE2IFP(sc)->if_mtu = GREMTU;
|
|
|
|
|
GRE2IFP(sc)->if_flags = IFF_POINTOPOINT|IFF_MULTICAST;
|
|
|
|
|
GRE2IFP(sc)->if_output = gre_output;
|
|
|
|
|
GRE2IFP(sc)->if_ioctl = gre_ioctl;
|
2002-09-06 17:12:50 +00:00
|
|
|
sc->g_dst.s_addr = sc->g_src.s_addr = INADDR_ANY;
|
|
|
|
|
sc->g_proto = IPPROTO_GRE;
|
2005-06-10 16:49:24 +00:00
|
|
|
GRE2IFP(sc)->if_flags |= IFF_LINK0;
|
2002-09-06 17:12:50 +00:00
|
|
|
sc->encap = NULL;
|
Add code to allow the system to handle multiple routing tables.
This particular implementation is designed to be fully backwards compatible
and to be MFC-able to 7.x (and 6.x)
Currently the only protocol that can make use of the multiple tables is IPv4
Similar functionality exists in OpenBSD and Linux.
From my notes:
-----
One thing where FreeBSD has been falling behind, and which by chance I
have some time to work on is "policy based routing", which allows
different
packet streams to be routed by more than just the destination address.
Constraints:
------------
I want to make some form of this available in the 6.x tree
(and by extension 7.x) , but FreeBSD in general needs it so I might as
well do it in -current and back port the portions I need.
One of the ways that this can be done is to have the ability to
instantiate multiple kernel routing tables (which I will now
refer to as "Forwarding Information Bases" or "FIBs" for political
correctness reasons). Which FIB a particular packet uses to make
the next hop decision can be decided by a number of mechanisms.
The policies these mechanisms implement are the "Policies" referred
to in "Policy based routing".
One of the constraints I have if I try to back port this work to
6.x is that it must be implemented as a EXTENSION to the existing
ABIs in 6.x so that third party applications do not need to be
recompiled in timespan of the branch.
This first version will not have some of the bells and whistles that
will come with later versions. It will, for example, be limited to 16
tables in the first commit.
Implementation method, Compatible version. (part 1)
-------------------------------
For this reason I have implemented a "sufficient subset" of a
multiple routing table solution in Perforce, and back-ported it
to 6.x. (also in Perforce though not always caught up with what I
have done in -current/P4). The subset allows a number of FIBs
to be defined at compile time (8 is sufficient for my purposes in 6.x)
and implements the changes needed to allow IPV4 to use them. I have not
done the changes for ipv6 simply because I do not need it, and I do not
have enough knowledge of ipv6 (e.g. neighbor discovery) needed to do it.
Other protocol families are left untouched and should there be
users with proprietary protocol families, they should continue to work
and be oblivious to the existence of the extra FIBs.
To understand how this is done, one must know that the current FIB
code starts everything off with a single dimensional array of
pointers to FIB head structures (One per protocol family), each of
which in turn points to the trie of routes available to that family.
The basic change in the ABI compatible version of the change is to
extent that array to be a 2 dimensional array, so that
instead of protocol family X looking at rt_tables[X] for the
table it needs, it looks at rt_tables[Y][X] when for all
protocol families except ipv4 Y is always 0.
Code that is unaware of the change always just sees the first row
of the table, which of course looks just like the one dimensional
array that existed before.
The entry points rtrequest(), rtalloc(), rtalloc1(), rtalloc_ign()
are all maintained, but refer only to the first row of the array,
so that existing callers in proprietary protocols can continue to
do the "right thing".
Some new entry points are added, for the exclusive use of ipv4 code
called in_rtrequest(), in_rtalloc(), in_rtalloc1() and in_rtalloc_ign(),
which have an extra argument which refers the code to the correct row.
In addition, there are some new entry points (currently called
rtalloc_fib() and friends) that check the Address family being
looked up and call either rtalloc() (and friends) if the protocol
is not IPv4 forcing the action to row 0 or to the appropriate row
if it IS IPv4 (and that info is available). These are for calling
from code that is not specific to any particular protocol. The way
these are implemented would change in the non ABI preserving code
to be added later.
One feature of the first version of the code is that for ipv4,
the interface routes show up automatically on all the FIBs, so
that no matter what FIB you select you always have the basic
direct attached hosts available to you. (rtinit() does this
automatically).
You CAN delete an interface route from one FIB should you want
to but by default it's there. ARP information is also available
in each FIB. It's assumed that the same machine would have the
same MAC address, regardless of which FIB you are using to get
to it.
This brings us as to how the correct FIB is selected for an outgoing
IPV4 packet.
Firstly, all packets have a FIB associated with them. if nothing
has been done to change it, it will be FIB 0. The FIB is changed
in the following ways.
Packets fall into one of a number of classes.
1/ locally generated packets, coming from a socket/PCB.
Such packets select a FIB from a number associated with the
socket/PCB. This in turn is inherited from the process,
but can be changed by a socket option. The process in turn
inherits it on fork. I have written a utility call setfib
that acts a bit like nice..
setfib -3 ping target.example.com # will use fib 3 for ping.
It is an obvious extension to make it a property of a jail
but I have not done so. It can be achieved by combining the setfib and
jail commands.
2/ packets received on an interface for forwarding.
By default these packets would use table 0,
(or possibly a number settable in a sysctl(not yet)).
but prior to routing the firewall can inspect them (see below).
(possibly in the future you may be able to associate a FIB
with packets received on an interface.. An ifconfig arg, but not yet.)
3/ packets inspected by a packet classifier, which can arbitrarily
associate a fib with it on a packet by packet basis.
A fib assigned to a packet by a packet classifier
(such as ipfw) would over-ride a fib associated by
a more default source. (such as cases 1 or 2).
4/ a tcp listen socket associated with a fib will generate
accept sockets that are associated with that same fib.
5/ Packets generated in response to some other packet (e.g. reset
or icmp packets). These should use the FIB associated with the
packet being reponded to.
6/ Packets generated during encapsulation.
gif, tun and other tunnel interfaces will encapsulate using the FIB
that was in effect withthe proces that set up the tunnel.
thus setfib 1 ifconfig gif0 [tunnel instructions]
will set the fib for the tunnel to use to be fib 1.
Routing messages would be associated with their
process, and thus select one FIB or another.
messages from the kernel would be associated with the fib they
refer to and would only be received by a routing socket associated
with that fib. (not yet implemented)
In addition Netstat has been edited to be able to cope with the
fact that the array is now 2 dimensional. (It looks in system
memory using libkvm (!)). Old versions of netstat see only the first FIB.
In addition two sysctls are added to give:
a) the number of FIBs compiled in (active)
b) the default FIB of the calling process.
Early testing experience:
-------------------------
Basically our (IronPort's) appliance does this functionality already
using ipfw fwd but that method has some drawbacks.
For example,
It can't fully simulate a routing table because it can't influence the
socket's choice of local address when a connect() is done.
Testing during the generating of these changes has been
remarkably smooth so far. Multiple tables have co-existed
with no notable side effects, and packets have been routes
accordingly.
ipfw has grown 2 new keywords:
setfib N ip from anay to any
count ip from any to any fib N
In pf there seems to be a requirement to be able to give symbolic names to the
fibs but I do not have that capacity. I am not sure if it is required.
SCTP has interestingly enough built in support for this, called VRFs
in Cisco parlance. it will be interesting to see how that handles it
when it suddenly actually does something.
Where to next:
--------------------
After committing the ABI compatible version and MFCing it, I'd
like to proceed in a forward direction in -current. this will
result in some roto-tilling in the routing code.
Firstly: the current code's idea of having a separate tree per
protocol family, all of the same format, and pointed to by the
1 dimensional array is a bit silly. Especially when one considers that
there is code that makes assumptions about every protocol having the
same internal structures there. Some protocols don't WANT that
sort of structure. (for example the whole idea of a netmask is foreign
to appletalk). This needs to be made opaque to the external code.
My suggested first change is to add routing method pointers to the
'domain' structure, along with information pointing the data.
instead of having an array of pointers to uniform structures,
there would be an array pointing to the 'domain' structures
for each protocol address domain (protocol family),
and the methods this reached would be called. The methods would have
an argument that gives FIB number, but the protocol would be free
to ignore it.
When the ABI can be changed it raises the possibilty of the
addition of a fib entry into the "struct route". Currently,
the structure contains the sockaddr of the desination, and the resulting
fib entry. To make this work fully, one could add a fib number
so that given an address and a fib, one can find the third element, the
fib entry.
Interaction with the ARP layer/ LL layer would need to be
revisited as well. Qing Li has been working on this already.
This work was sponsored by Ironport Systems/Cisco
Reviewed by: several including rwatson, bz and mlair (parts each)
Obtained from: Ironport systems/Cisco
2008-05-09 23:03:00 +00:00
|
|
|
sc->gre_fibnum = curthread->td_proc->p_fibnum;
|
2004-01-26 12:33:56 +00:00
|
|
|
sc->wccp_ver = WCCP_V1;
|
2008-06-20 17:26:34 +00:00
|
|
|
sc->key = 0;
|
2005-06-10 16:49:24 +00:00
|
|
|
if_attach(GRE2IFP(sc));
|
|
|
|
|
bpfattach(GRE2IFP(sc), DLT_NULL, sizeof(u_int32_t));
|
2004-03-22 16:04:43 +00:00
|
|
|
mtx_lock(&gre_mtx);
|
2002-09-06 17:12:50 +00:00
|
|
|
LIST_INSERT_HEAD(&gre_softc_list, sc, sc_list);
|
2004-03-22 16:04:43 +00:00
|
|
|
mtx_unlock(&gre_mtx);
|
2002-09-06 17:12:50 +00:00
|
|
|
return (0);
|
|
|
|
|
}
|
|
|
|
|
|
2004-03-22 16:04:43 +00:00
|
|
|
static void
|
|
|
|
|
gre_clone_destroy(ifp)
|
|
|
|
|
struct ifnet *ifp;
|
|
|
|
|
{
|
|
|
|
|
struct gre_softc *sc = ifp->if_softc;
|
|
|
|
|
|
|
|
|
|
mtx_lock(&gre_mtx);
|
|
|
|
|
LIST_REMOVE(sc, sc_list);
|
|
|
|
|
mtx_unlock(&gre_mtx);
|
2005-10-12 19:52:16 +00:00
|
|
|
|
|
|
|
|
#ifdef INET
|
|
|
|
|
if (sc->encap != NULL)
|
|
|
|
|
encap_detach(sc->encap);
|
|
|
|
|
#endif
|
|
|
|
|
bpfdetach(ifp);
|
|
|
|
|
if_detach(ifp);
|
|
|
|
|
if_free(ifp);
|
|
|
|
|
free(sc, M_GRE);
|
2004-03-22 16:04:43 +00:00
|
|
|
}
|
|
|
|
|
|
2002-09-06 17:12:50 +00:00
|
|
|
/*
|
|
|
|
|
* The output routine. Takes a packet and encapsulates it in the protocol
|
|
|
|
|
* given by sc->g_proto. See also RFC 1701 and RFC 2004
|
|
|
|
|
*/
|
2002-09-06 18:16:03 +00:00
|
|
|
static int
|
2002-09-06 17:12:50 +00:00
|
|
|
gre_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *dst,
|
2009-04-16 20:30:28 +00:00
|
|
|
struct route *ro)
|
2002-09-06 17:12:50 +00:00
|
|
|
{
|
|
|
|
|
int error = 0;
|
|
|
|
|
struct gre_softc *sc = ifp->if_softc;
|
|
|
|
|
struct greip *gh;
|
|
|
|
|
struct ip *ip;
|
2011-06-18 09:34:03 +00:00
|
|
|
struct m_tag *mtag;
|
|
|
|
|
struct mtag_gre_nesting *gt;
|
|
|
|
|
size_t len;
|
2008-06-26 22:59:49 +00:00
|
|
|
u_short gre_ip_id = 0;
|
|
|
|
|
uint8_t gre_ip_tos = 0;
|
2003-12-30 11:41:43 +00:00
|
|
|
u_int16_t etype = 0;
|
2002-09-06 17:12:50 +00:00
|
|
|
struct mobile_h mob_h;
|
2005-06-26 18:11:11 +00:00
|
|
|
u_int32_t af;
|
2011-06-18 09:34:03 +00:00
|
|
|
int extra = 0, max;
|
2002-09-06 17:12:50 +00:00
|
|
|
|
|
|
|
|
/*
|
2011-06-18 09:34:03 +00:00
|
|
|
* gre may cause infinite recursion calls when misconfigured. High
|
|
|
|
|
* nesting level may cause stack exhaustion. We'll prevent this by
|
|
|
|
|
* detecting loops and by introducing upper limit.
|
2002-09-06 17:12:50 +00:00
|
|
|
*/
|
2011-06-18 09:34:03 +00:00
|
|
|
mtag = m_tag_locate(m, MTAG_COOKIE_GRE, MTAG_GRE_NESTING, NULL);
|
|
|
|
|
if (mtag != NULL) {
|
|
|
|
|
struct ifnet **ifp2;
|
|
|
|
|
|
|
|
|
|
gt = (struct mtag_gre_nesting *)(mtag + 1);
|
|
|
|
|
gt->count++;
|
|
|
|
|
if (gt->count > min(gt->max,max_gre_nesting)) {
|
|
|
|
|
printf("%s: hit maximum recursion limit %u on %s\n",
|
|
|
|
|
__func__, gt->count - 1, ifp->if_xname);
|
|
|
|
|
m_freem(m);
|
|
|
|
|
error = EIO; /* is there better errno? */
|
|
|
|
|
goto end;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
ifp2 = gt->ifp;
|
|
|
|
|
for (max = gt->count - 1; max > 0; max--) {
|
|
|
|
|
if (*ifp2 == ifp)
|
|
|
|
|
break;
|
|
|
|
|
ifp2++;
|
|
|
|
|
}
|
|
|
|
|
if (*ifp2 == ifp) {
|
|
|
|
|
printf("%s: detected loop with nexting %u on %s\n",
|
|
|
|
|
__func__, gt->count-1, ifp->if_xname);
|
|
|
|
|
m_freem(m);
|
|
|
|
|
error = EIO; /* is there better errno? */
|
|
|
|
|
goto end;
|
|
|
|
|
}
|
|
|
|
|
*ifp2 = ifp;
|
|
|
|
|
|
|
|
|
|
} else {
|
|
|
|
|
/*
|
|
|
|
|
* Given that people should NOT increase max_gre_nesting beyond
|
|
|
|
|
* their real needs, we allocate once per packet rather than
|
|
|
|
|
* allocating an mtag once per passing through gre.
|
|
|
|
|
*
|
|
|
|
|
* Note: the sysctl does not actually check for saneness, so we
|
|
|
|
|
* limit the maximum numbers of possible recursions here.
|
|
|
|
|
*/
|
|
|
|
|
max = imin(max_gre_nesting, 256);
|
|
|
|
|
/* If someone sets the sysctl <= 0, we want at least 1. */
|
|
|
|
|
max = imax(max, 1);
|
|
|
|
|
len = sizeof(struct mtag_gre_nesting) +
|
|
|
|
|
max * sizeof(struct ifnet *);
|
|
|
|
|
mtag = m_tag_alloc(MTAG_COOKIE_GRE, MTAG_GRE_NESTING, len,
|
|
|
|
|
M_NOWAIT);
|
|
|
|
|
if (mtag == NULL) {
|
|
|
|
|
m_freem(m);
|
|
|
|
|
error = ENOMEM;
|
|
|
|
|
goto end;
|
|
|
|
|
}
|
|
|
|
|
gt = (struct mtag_gre_nesting *)(mtag + 1);
|
|
|
|
|
bzero(gt, len);
|
|
|
|
|
gt->count = 1;
|
|
|
|
|
gt->max = max;
|
|
|
|
|
*gt->ifp = ifp;
|
|
|
|
|
m_tag_prepend(m, mtag);
|
2002-09-06 17:12:50 +00:00
|
|
|
}
|
|
|
|
|
|
2005-08-09 10:20:02 +00:00
|
|
|
if (!((ifp->if_flags & IFF_UP) &&
|
|
|
|
|
(ifp->if_drv_flags & IFF_DRV_RUNNING)) ||
|
2002-09-06 17:12:50 +00:00
|
|
|
sc->g_src.s_addr == INADDR_ANY || sc->g_dst.s_addr == INADDR_ANY) {
|
|
|
|
|
m_freem(m);
|
|
|
|
|
error = ENETDOWN;
|
|
|
|
|
goto end;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
gh = NULL;
|
|
|
|
|
ip = NULL;
|
|
|
|
|
|
2005-06-26 18:11:11 +00:00
|
|
|
/* BPF writes need to be handled specially. */
|
|
|
|
|
if (dst->sa_family == AF_UNSPEC) {
|
|
|
|
|
bcopy(dst->sa_data, &af, sizeof(af));
|
|
|
|
|
dst->sa_family = af;
|
|
|
|
|
}
|
|
|
|
|
|
2006-06-02 19:59:33 +00:00
|
|
|
if (bpf_peers_present(ifp->if_bpf)) {
|
2005-06-26 18:11:11 +00:00
|
|
|
af = dst->sa_family;
|
2003-12-28 03:56:00 +00:00
|
|
|
bpf_mtap2(ifp->if_bpf, &af, sizeof(af), m);
|
2002-09-06 17:12:50 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
m->m_flags &= ~(M_BCAST|M_MCAST);
|
|
|
|
|
|
|
|
|
|
if (sc->g_proto == IPPROTO_MOBILE) {
|
|
|
|
|
if (dst->sa_family == AF_INET) {
|
|
|
|
|
struct mbuf *m0;
|
|
|
|
|
int msiz;
|
|
|
|
|
|
|
|
|
|
ip = mtod(m, struct ip *);
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* RFC2004 specifies that fragmented diagrams shouldn't
|
|
|
|
|
* be encapsulated.
|
|
|
|
|
*/
|
2006-05-11 00:53:43 +00:00
|
|
|
if (ip->ip_off & (IP_MF | IP_OFFMASK)) {
|
2002-09-06 17:12:50 +00:00
|
|
|
_IF_DROP(&ifp->if_snd);
|
|
|
|
|
m_freem(m);
|
|
|
|
|
error = EINVAL; /* is there better errno? */
|
|
|
|
|
goto end;
|
|
|
|
|
}
|
|
|
|
|
memset(&mob_h, 0, MOB_H_SIZ_L);
|
|
|
|
|
mob_h.proto = (ip->ip_p) << 8;
|
|
|
|
|
mob_h.odst = ip->ip_dst.s_addr;
|
|
|
|
|
ip->ip_dst.s_addr = sc->g_dst.s_addr;
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* If the packet comes from our host, we only change
|
|
|
|
|
* the destination address in the IP header.
|
|
|
|
|
* Else we also need to save and change the source
|
|
|
|
|
*/
|
|
|
|
|
if (in_hosteq(ip->ip_src, sc->g_src)) {
|
|
|
|
|
msiz = MOB_H_SIZ_S;
|
|
|
|
|
} else {
|
|
|
|
|
mob_h.proto |= MOB_H_SBIT;
|
|
|
|
|
mob_h.osrc = ip->ip_src.s_addr;
|
|
|
|
|
ip->ip_src.s_addr = sc->g_src.s_addr;
|
|
|
|
|
msiz = MOB_H_SIZ_L;
|
|
|
|
|
}
|
|
|
|
|
mob_h.proto = htons(mob_h.proto);
|
2003-12-30 11:41:43 +00:00
|
|
|
mob_h.hcrc = gre_in_cksum((u_int16_t *)&mob_h, msiz);
|
2002-09-06 17:12:50 +00:00
|
|
|
|
|
|
|
|
if ((m->m_data - msiz) < m->m_pktdat) {
|
|
|
|
|
/* need new mbuf */
|
2005-11-02 13:46:32 +00:00
|
|
|
MGETHDR(m0, M_DONTWAIT, MT_DATA);
|
2002-09-06 17:12:50 +00:00
|
|
|
if (m0 == NULL) {
|
|
|
|
|
_IF_DROP(&ifp->if_snd);
|
|
|
|
|
m_freem(m);
|
|
|
|
|
error = ENOBUFS;
|
|
|
|
|
goto end;
|
|
|
|
|
}
|
|
|
|
|
m0->m_next = m;
|
|
|
|
|
m->m_data += sizeof(struct ip);
|
|
|
|
|
m->m_len -= sizeof(struct ip);
|
|
|
|
|
m0->m_pkthdr.len = m->m_pkthdr.len + msiz;
|
|
|
|
|
m0->m_len = msiz + sizeof(struct ip);
|
|
|
|
|
m0->m_data += max_linkhdr;
|
|
|
|
|
memcpy(mtod(m0, caddr_t), (caddr_t)ip,
|
|
|
|
|
sizeof(struct ip));
|
|
|
|
|
m = m0;
|
|
|
|
|
} else { /* we have some space left in the old one */
|
|
|
|
|
m->m_data -= msiz;
|
|
|
|
|
m->m_len += msiz;
|
|
|
|
|
m->m_pkthdr.len += msiz;
|
|
|
|
|
bcopy(ip, mtod(m, caddr_t),
|
|
|
|
|
sizeof(struct ip));
|
|
|
|
|
}
|
|
|
|
|
ip = mtod(m, struct ip *);
|
|
|
|
|
memcpy((caddr_t)(ip + 1), &mob_h, (unsigned)msiz);
|
|
|
|
|
ip->ip_len = ntohs(ip->ip_len) + msiz;
|
|
|
|
|
} else { /* AF_INET */
|
|
|
|
|
_IF_DROP(&ifp->if_snd);
|
|
|
|
|
m_freem(m);
|
|
|
|
|
error = EINVAL;
|
|
|
|
|
goto end;
|
|
|
|
|
}
|
|
|
|
|
} else if (sc->g_proto == IPPROTO_GRE) {
|
|
|
|
|
switch (dst->sa_family) {
|
|
|
|
|
case AF_INET:
|
|
|
|
|
ip = mtod(m, struct ip *);
|
2008-06-26 22:59:49 +00:00
|
|
|
gre_ip_tos = ip->ip_tos;
|
|
|
|
|
gre_ip_id = ip->ip_id;
|
2008-07-20 21:45:15 +00:00
|
|
|
if (sc->wccp_ver == WCCP_V2) {
|
|
|
|
|
extra = sizeof(uint32_t);
|
|
|
|
|
etype = WCCP_PROTOCOL_TYPE;
|
|
|
|
|
} else {
|
|
|
|
|
etype = ETHERTYPE_IP;
|
|
|
|
|
}
|
2002-09-06 17:12:50 +00:00
|
|
|
break;
|
2005-08-01 08:14:21 +00:00
|
|
|
#ifdef INET6
|
|
|
|
|
case AF_INET6:
|
2008-06-26 22:59:49 +00:00
|
|
|
gre_ip_id = ip_newid();
|
2005-08-01 08:14:21 +00:00
|
|
|
etype = ETHERTYPE_IPV6;
|
|
|
|
|
break;
|
|
|
|
|
#endif
|
2002-09-06 17:12:50 +00:00
|
|
|
#ifdef NETATALK
|
|
|
|
|
case AF_APPLETALK:
|
|
|
|
|
etype = ETHERTYPE_ATALK;
|
|
|
|
|
break;
|
|
|
|
|
#endif
|
|
|
|
|
default:
|
|
|
|
|
_IF_DROP(&ifp->if_snd);
|
|
|
|
|
m_freem(m);
|
|
|
|
|
error = EAFNOSUPPORT;
|
|
|
|
|
goto end;
|
|
|
|
|
}
|
2008-06-20 17:26:34 +00:00
|
|
|
|
|
|
|
|
/* Reserve space for GRE header + optional GRE key */
|
2008-07-20 21:45:15 +00:00
|
|
|
int hdrlen = sizeof(struct greip) + extra;
|
2008-06-20 17:26:34 +00:00
|
|
|
if (sc->key)
|
|
|
|
|
hdrlen += sizeof(uint32_t);
|
|
|
|
|
M_PREPEND(m, hdrlen, M_DONTWAIT);
|
2002-09-06 17:12:50 +00:00
|
|
|
} else {
|
|
|
|
|
_IF_DROP(&ifp->if_snd);
|
|
|
|
|
m_freem(m);
|
|
|
|
|
error = EINVAL;
|
|
|
|
|
goto end;
|
|
|
|
|
}
|
|
|
|
|
|
2004-04-23 16:10:23 +00:00
|
|
|
if (m == NULL) { /* mbuf allocation failed */
|
2002-09-06 17:12:50 +00:00
|
|
|
_IF_DROP(&ifp->if_snd);
|
|
|
|
|
error = ENOBUFS;
|
|
|
|
|
goto end;
|
|
|
|
|
}
|
|
|
|
|
|
Add code to allow the system to handle multiple routing tables.
This particular implementation is designed to be fully backwards compatible
and to be MFC-able to 7.x (and 6.x)
Currently the only protocol that can make use of the multiple tables is IPv4
Similar functionality exists in OpenBSD and Linux.
From my notes:
-----
One thing where FreeBSD has been falling behind, and which by chance I
have some time to work on is "policy based routing", which allows
different
packet streams to be routed by more than just the destination address.
Constraints:
------------
I want to make some form of this available in the 6.x tree
(and by extension 7.x) , but FreeBSD in general needs it so I might as
well do it in -current and back port the portions I need.
One of the ways that this can be done is to have the ability to
instantiate multiple kernel routing tables (which I will now
refer to as "Forwarding Information Bases" or "FIBs" for political
correctness reasons). Which FIB a particular packet uses to make
the next hop decision can be decided by a number of mechanisms.
The policies these mechanisms implement are the "Policies" referred
to in "Policy based routing".
One of the constraints I have if I try to back port this work to
6.x is that it must be implemented as a EXTENSION to the existing
ABIs in 6.x so that third party applications do not need to be
recompiled in timespan of the branch.
This first version will not have some of the bells and whistles that
will come with later versions. It will, for example, be limited to 16
tables in the first commit.
Implementation method, Compatible version. (part 1)
-------------------------------
For this reason I have implemented a "sufficient subset" of a
multiple routing table solution in Perforce, and back-ported it
to 6.x. (also in Perforce though not always caught up with what I
have done in -current/P4). The subset allows a number of FIBs
to be defined at compile time (8 is sufficient for my purposes in 6.x)
and implements the changes needed to allow IPV4 to use them. I have not
done the changes for ipv6 simply because I do not need it, and I do not
have enough knowledge of ipv6 (e.g. neighbor discovery) needed to do it.
Other protocol families are left untouched and should there be
users with proprietary protocol families, they should continue to work
and be oblivious to the existence of the extra FIBs.
To understand how this is done, one must know that the current FIB
code starts everything off with a single dimensional array of
pointers to FIB head structures (One per protocol family), each of
which in turn points to the trie of routes available to that family.
The basic change in the ABI compatible version of the change is to
extent that array to be a 2 dimensional array, so that
instead of protocol family X looking at rt_tables[X] for the
table it needs, it looks at rt_tables[Y][X] when for all
protocol families except ipv4 Y is always 0.
Code that is unaware of the change always just sees the first row
of the table, which of course looks just like the one dimensional
array that existed before.
The entry points rtrequest(), rtalloc(), rtalloc1(), rtalloc_ign()
are all maintained, but refer only to the first row of the array,
so that existing callers in proprietary protocols can continue to
do the "right thing".
Some new entry points are added, for the exclusive use of ipv4 code
called in_rtrequest(), in_rtalloc(), in_rtalloc1() and in_rtalloc_ign(),
which have an extra argument which refers the code to the correct row.
In addition, there are some new entry points (currently called
rtalloc_fib() and friends) that check the Address family being
looked up and call either rtalloc() (and friends) if the protocol
is not IPv4 forcing the action to row 0 or to the appropriate row
if it IS IPv4 (and that info is available). These are for calling
from code that is not specific to any particular protocol. The way
these are implemented would change in the non ABI preserving code
to be added later.
One feature of the first version of the code is that for ipv4,
the interface routes show up automatically on all the FIBs, so
that no matter what FIB you select you always have the basic
direct attached hosts available to you. (rtinit() does this
automatically).
You CAN delete an interface route from one FIB should you want
to but by default it's there. ARP information is also available
in each FIB. It's assumed that the same machine would have the
same MAC address, regardless of which FIB you are using to get
to it.
This brings us as to how the correct FIB is selected for an outgoing
IPV4 packet.
Firstly, all packets have a FIB associated with them. if nothing
has been done to change it, it will be FIB 0. The FIB is changed
in the following ways.
Packets fall into one of a number of classes.
1/ locally generated packets, coming from a socket/PCB.
Such packets select a FIB from a number associated with the
socket/PCB. This in turn is inherited from the process,
but can be changed by a socket option. The process in turn
inherits it on fork. I have written a utility call setfib
that acts a bit like nice..
setfib -3 ping target.example.com # will use fib 3 for ping.
It is an obvious extension to make it a property of a jail
but I have not done so. It can be achieved by combining the setfib and
jail commands.
2/ packets received on an interface for forwarding.
By default these packets would use table 0,
(or possibly a number settable in a sysctl(not yet)).
but prior to routing the firewall can inspect them (see below).
(possibly in the future you may be able to associate a FIB
with packets received on an interface.. An ifconfig arg, but not yet.)
3/ packets inspected by a packet classifier, which can arbitrarily
associate a fib with it on a packet by packet basis.
A fib assigned to a packet by a packet classifier
(such as ipfw) would over-ride a fib associated by
a more default source. (such as cases 1 or 2).
4/ a tcp listen socket associated with a fib will generate
accept sockets that are associated with that same fib.
5/ Packets generated in response to some other packet (e.g. reset
or icmp packets). These should use the FIB associated with the
packet being reponded to.
6/ Packets generated during encapsulation.
gif, tun and other tunnel interfaces will encapsulate using the FIB
that was in effect withthe proces that set up the tunnel.
thus setfib 1 ifconfig gif0 [tunnel instructions]
will set the fib for the tunnel to use to be fib 1.
Routing messages would be associated with their
process, and thus select one FIB or another.
messages from the kernel would be associated with the fib they
refer to and would only be received by a routing socket associated
with that fib. (not yet implemented)
In addition Netstat has been edited to be able to cope with the
fact that the array is now 2 dimensional. (It looks in system
memory using libkvm (!)). Old versions of netstat see only the first FIB.
In addition two sysctls are added to give:
a) the number of FIBs compiled in (active)
b) the default FIB of the calling process.
Early testing experience:
-------------------------
Basically our (IronPort's) appliance does this functionality already
using ipfw fwd but that method has some drawbacks.
For example,
It can't fully simulate a routing table because it can't influence the
socket's choice of local address when a connect() is done.
Testing during the generating of these changes has been
remarkably smooth so far. Multiple tables have co-existed
with no notable side effects, and packets have been routes
accordingly.
ipfw has grown 2 new keywords:
setfib N ip from anay to any
count ip from any to any fib N
In pf there seems to be a requirement to be able to give symbolic names to the
fibs but I do not have that capacity. I am not sure if it is required.
SCTP has interestingly enough built in support for this, called VRFs
in Cisco parlance. it will be interesting to see how that handles it
when it suddenly actually does something.
Where to next:
--------------------
After committing the ABI compatible version and MFCing it, I'd
like to proceed in a forward direction in -current. this will
result in some roto-tilling in the routing code.
Firstly: the current code's idea of having a separate tree per
protocol family, all of the same format, and pointed to by the
1 dimensional array is a bit silly. Especially when one considers that
there is code that makes assumptions about every protocol having the
same internal structures there. Some protocols don't WANT that
sort of structure. (for example the whole idea of a netmask is foreign
to appletalk). This needs to be made opaque to the external code.
My suggested first change is to add routing method pointers to the
'domain' structure, along with information pointing the data.
instead of having an array of pointers to uniform structures,
there would be an array pointing to the 'domain' structures
for each protocol address domain (protocol family),
and the methods this reached would be called. The methods would have
an argument that gives FIB number, but the protocol would be free
to ignore it.
When the ABI can be changed it raises the possibilty of the
addition of a fib entry into the "struct route". Currently,
the structure contains the sockaddr of the desination, and the resulting
fib entry. To make this work fully, one could add a fib number
so that given an address and a fib, one can find the third element, the
fib entry.
Interaction with the ARP layer/ LL layer would need to be
revisited as well. Qing Li has been working on this already.
This work was sponsored by Ironport Systems/Cisco
Reviewed by: several including rwatson, bz and mlair (parts each)
Obtained from: Ironport systems/Cisco
2008-05-09 23:03:00 +00:00
|
|
|
M_SETFIB(m, sc->gre_fibnum); /* The envelope may use a different FIB */
|
|
|
|
|
|
2002-09-06 17:12:50 +00:00
|
|
|
gh = mtod(m, struct greip *);
|
|
|
|
|
if (sc->g_proto == IPPROTO_GRE) {
|
2008-06-20 17:26:34 +00:00
|
|
|
uint32_t *options = gh->gi_options;
|
|
|
|
|
|
2008-07-20 21:45:15 +00:00
|
|
|
memset((void *)gh, 0, sizeof(struct greip) + extra);
|
2002-09-06 17:12:50 +00:00
|
|
|
gh->gi_ptype = htons(etype);
|
2008-06-20 17:26:34 +00:00
|
|
|
gh->gi_flags = 0;
|
|
|
|
|
|
|
|
|
|
/* Add key option */
|
|
|
|
|
if (sc->key)
|
|
|
|
|
{
|
|
|
|
|
gh->gi_flags |= htons(GRE_KP);
|
|
|
|
|
*(options++) = htonl(sc->key);
|
|
|
|
|
}
|
2002-09-06 17:12:50 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
gh->gi_pr = sc->g_proto;
|
|
|
|
|
if (sc->g_proto != IPPROTO_MOBILE) {
|
|
|
|
|
gh->gi_src = sc->g_src;
|
|
|
|
|
gh->gi_dst = sc->g_dst;
|
2004-08-05 08:12:46 +00:00
|
|
|
((struct ip*)gh)->ip_v = IPPROTO_IPV4;
|
2002-09-06 17:12:50 +00:00
|
|
|
((struct ip*)gh)->ip_hl = (sizeof(struct ip)) >> 2;
|
2002-09-06 18:16:03 +00:00
|
|
|
((struct ip*)gh)->ip_ttl = GRE_TTL;
|
2008-06-26 22:59:49 +00:00
|
|
|
((struct ip*)gh)->ip_tos = gre_ip_tos;
|
|
|
|
|
((struct ip*)gh)->ip_id = gre_ip_id;
|
2004-01-30 09:03:01 +00:00
|
|
|
gh->gi_len = m->m_pkthdr.len;
|
2002-09-06 17:12:50 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
ifp->if_opackets++;
|
|
|
|
|
ifp->if_obytes += m->m_pkthdr.len;
|
2004-04-23 16:57:43 +00:00
|
|
|
/*
|
|
|
|
|
* Send it off and with IP_FORWARD flag to prevent it from
|
|
|
|
|
* overwriting the ip_id again. ip_id is already set to the
|
|
|
|
|
* ip_id of the encapsulated packet.
|
|
|
|
|
*/
|
2004-04-23 16:10:23 +00:00
|
|
|
error = ip_output(m, NULL, &sc->route, IP_FORWARDING,
|
2003-12-30 11:41:43 +00:00
|
|
|
(struct ip_moptions *)NULL, (struct inpcb *)NULL);
|
2002-09-06 17:12:50 +00:00
|
|
|
end:
|
|
|
|
|
if (error)
|
|
|
|
|
ifp->if_oerrors++;
|
|
|
|
|
return (error);
|
|
|
|
|
}
|
|
|
|
|
|
2002-09-06 18:16:03 +00:00
|
|
|
static int
|
2002-09-06 17:12:50 +00:00
|
|
|
gre_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
|
|
|
|
|
{
|
|
|
|
|
struct ifreq *ifr = (struct ifreq *)data;
|
|
|
|
|
struct if_laddrreq *lifr = (struct if_laddrreq *)data;
|
|
|
|
|
struct in_aliasreq *aifr = (struct in_aliasreq *)data;
|
|
|
|
|
struct gre_softc *sc = ifp->if_softc;
|
|
|
|
|
int s;
|
|
|
|
|
struct sockaddr_in si;
|
|
|
|
|
struct sockaddr *sa = NULL;
|
2008-06-20 17:26:34 +00:00
|
|
|
int error, adj;
|
2002-09-06 17:12:50 +00:00
|
|
|
struct sockaddr_in sp, sm, dp, dm;
|
2008-06-20 17:26:34 +00:00
|
|
|
uint32_t key;
|
2002-09-06 17:12:50 +00:00
|
|
|
|
|
|
|
|
error = 0;
|
2008-06-20 17:26:34 +00:00
|
|
|
adj = 0;
|
2002-09-06 17:12:50 +00:00
|
|
|
|
|
|
|
|
s = splnet();
|
|
|
|
|
switch (cmd) {
|
|
|
|
|
case SIOCSIFADDR:
|
|
|
|
|
ifp->if_flags |= IFF_UP;
|
|
|
|
|
break;
|
2004-01-26 12:21:59 +00:00
|
|
|
case SIOCSIFDSTADDR:
|
2002-09-06 17:12:50 +00:00
|
|
|
break;
|
|
|
|
|
case SIOCSIFFLAGS:
|
2006-11-06 13:42:10 +00:00
|
|
|
/*
|
2007-06-26 23:01:01 +00:00
|
|
|
* XXXRW: Isn't this priv_check() redundant to the ifnet
|
|
|
|
|
* layer check?
|
2006-11-06 13:42:10 +00:00
|
|
|
*/
|
|
|
|
|
if ((error = priv_check(curthread, PRIV_NET_SETIFFLAGS)) != 0)
|
2002-09-06 17:12:50 +00:00
|
|
|
break;
|
|
|
|
|
if ((ifr->ifr_flags & IFF_LINK0) != 0)
|
|
|
|
|
sc->g_proto = IPPROTO_GRE;
|
|
|
|
|
else
|
|
|
|
|
sc->g_proto = IPPROTO_MOBILE;
|
2004-01-26 12:33:56 +00:00
|
|
|
if ((ifr->ifr_flags & IFF_LINK2) != 0)
|
|
|
|
|
sc->wccp_ver = WCCP_V2;
|
|
|
|
|
else
|
|
|
|
|
sc->wccp_ver = WCCP_V1;
|
2002-09-06 17:12:50 +00:00
|
|
|
goto recompute;
|
|
|
|
|
case SIOCSIFMTU:
|
2006-11-06 13:42:10 +00:00
|
|
|
/*
|
2007-06-26 23:01:01 +00:00
|
|
|
* XXXRW: Isn't this priv_check() redundant to the ifnet
|
|
|
|
|
* layer check?
|
2006-11-06 13:42:10 +00:00
|
|
|
*/
|
|
|
|
|
if ((error = priv_check(curthread, PRIV_NET_SETIFMTU)) != 0)
|
2002-09-06 17:12:50 +00:00
|
|
|
break;
|
|
|
|
|
if (ifr->ifr_mtu < 576) {
|
|
|
|
|
error = EINVAL;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
ifp->if_mtu = ifr->ifr_mtu;
|
|
|
|
|
break;
|
|
|
|
|
case SIOCGIFMTU:
|
2005-06-10 16:49:24 +00:00
|
|
|
ifr->ifr_mtu = GRE2IFP(sc)->if_mtu;
|
2002-09-06 17:12:50 +00:00
|
|
|
break;
|
|
|
|
|
case SIOCADDMULTI:
|
2006-11-06 13:42:10 +00:00
|
|
|
/*
|
2007-06-26 23:01:01 +00:00
|
|
|
* XXXRW: Isn't this priv_checkr() redundant to the ifnet
|
|
|
|
|
* layer check?
|
2006-11-06 13:42:10 +00:00
|
|
|
*/
|
|
|
|
|
if ((error = priv_check(curthread, PRIV_NET_ADDMULTI)) != 0)
|
|
|
|
|
break;
|
|
|
|
|
if (ifr == 0) {
|
|
|
|
|
error = EAFNOSUPPORT;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
switch (ifr->ifr_addr.sa_family) {
|
|
|
|
|
#ifdef INET
|
|
|
|
|
case AF_INET:
|
|
|
|
|
break;
|
|
|
|
|
#endif
|
|
|
|
|
#ifdef INET6
|
|
|
|
|
case AF_INET6:
|
|
|
|
|
break;
|
|
|
|
|
#endif
|
|
|
|
|
default:
|
|
|
|
|
error = EAFNOSUPPORT;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
break;
|
2002-09-06 17:12:50 +00:00
|
|
|
case SIOCDELMULTI:
|
2006-11-06 13:42:10 +00:00
|
|
|
/*
|
2007-06-26 23:01:01 +00:00
|
|
|
* XXXRW: Isn't this priv_check() redundant to the ifnet
|
|
|
|
|
* layer check?
|
2006-11-06 13:42:10 +00:00
|
|
|
*/
|
|
|
|
|
if ((error = priv_check(curthread, PRIV_NET_DELIFGROUP)) != 0)
|
2002-09-06 17:12:50 +00:00
|
|
|
break;
|
|
|
|
|
if (ifr == 0) {
|
|
|
|
|
error = EAFNOSUPPORT;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
switch (ifr->ifr_addr.sa_family) {
|
|
|
|
|
#ifdef INET
|
|
|
|
|
case AF_INET:
|
|
|
|
|
break;
|
2005-08-01 08:14:21 +00:00
|
|
|
#endif
|
|
|
|
|
#ifdef INET6
|
|
|
|
|
case AF_INET6:
|
|
|
|
|
break;
|
2002-09-06 17:12:50 +00:00
|
|
|
#endif
|
|
|
|
|
default:
|
|
|
|
|
error = EAFNOSUPPORT;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
break;
|
|
|
|
|
case GRESPROTO:
|
2006-11-06 13:42:10 +00:00
|
|
|
/*
|
2007-06-26 23:01:01 +00:00
|
|
|
* XXXRW: Isn't this priv_check() redundant to the ifnet
|
|
|
|
|
* layer check?
|
2006-11-06 13:42:10 +00:00
|
|
|
*/
|
|
|
|
|
if ((error = priv_check(curthread, PRIV_NET_GRE)) != 0)
|
2002-09-06 17:12:50 +00:00
|
|
|
break;
|
|
|
|
|
sc->g_proto = ifr->ifr_flags;
|
|
|
|
|
switch (sc->g_proto) {
|
|
|
|
|
case IPPROTO_GRE:
|
|
|
|
|
ifp->if_flags |= IFF_LINK0;
|
|
|
|
|
break;
|
|
|
|
|
case IPPROTO_MOBILE:
|
|
|
|
|
ifp->if_flags &= ~IFF_LINK0;
|
|
|
|
|
break;
|
|
|
|
|
default:
|
|
|
|
|
error = EPROTONOSUPPORT;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
goto recompute;
|
|
|
|
|
case GREGPROTO:
|
|
|
|
|
ifr->ifr_flags = sc->g_proto;
|
|
|
|
|
break;
|
|
|
|
|
case GRESADDRS:
|
|
|
|
|
case GRESADDRD:
|
2006-11-06 13:42:10 +00:00
|
|
|
error = priv_check(curthread, PRIV_NET_GRE);
|
|
|
|
|
if (error)
|
|
|
|
|
return (error);
|
2002-09-06 17:12:50 +00:00
|
|
|
/*
|
|
|
|
|
* set tunnel endpoints, compute a less specific route
|
|
|
|
|
* to the remote end and mark if as up
|
|
|
|
|
*/
|
|
|
|
|
sa = &ifr->ifr_addr;
|
|
|
|
|
if (cmd == GRESADDRS)
|
|
|
|
|
sc->g_src = (satosin(sa))->sin_addr;
|
|
|
|
|
if (cmd == GRESADDRD)
|
|
|
|
|
sc->g_dst = (satosin(sa))->sin_addr;
|
|
|
|
|
recompute:
|
|
|
|
|
#ifdef INET
|
|
|
|
|
if (sc->encap != NULL) {
|
|
|
|
|
encap_detach(sc->encap);
|
|
|
|
|
sc->encap = NULL;
|
|
|
|
|
}
|
|
|
|
|
#endif
|
|
|
|
|
if ((sc->g_src.s_addr != INADDR_ANY) &&
|
|
|
|
|
(sc->g_dst.s_addr != INADDR_ANY)) {
|
|
|
|
|
bzero(&sp, sizeof(sp));
|
|
|
|
|
bzero(&sm, sizeof(sm));
|
|
|
|
|
bzero(&dp, sizeof(dp));
|
|
|
|
|
bzero(&dm, sizeof(dm));
|
|
|
|
|
sp.sin_len = sm.sin_len = dp.sin_len = dm.sin_len =
|
|
|
|
|
sizeof(struct sockaddr_in);
|
|
|
|
|
sp.sin_family = sm.sin_family = dp.sin_family =
|
|
|
|
|
dm.sin_family = AF_INET;
|
|
|
|
|
sp.sin_addr = sc->g_src;
|
|
|
|
|
dp.sin_addr = sc->g_dst;
|
2004-01-26 12:21:59 +00:00
|
|
|
sm.sin_addr.s_addr = dm.sin_addr.s_addr =
|
2002-09-06 17:12:50 +00:00
|
|
|
INADDR_BROADCAST;
|
|
|
|
|
#ifdef INET
|
|
|
|
|
sc->encap = encap_attach(AF_INET, sc->g_proto,
|
|
|
|
|
sintosa(&sp), sintosa(&sm), sintosa(&dp),
|
|
|
|
|
sintosa(&dm), (sc->g_proto == IPPROTO_GRE) ?
|
|
|
|
|
&in_gre_protosw : &in_mobile_protosw, sc);
|
|
|
|
|
if (sc->encap == NULL)
|
|
|
|
|
printf("%s: unable to attach encap\n",
|
2005-06-10 16:49:24 +00:00
|
|
|
if_name(GRE2IFP(sc)));
|
2002-09-06 17:12:50 +00:00
|
|
|
#endif
|
|
|
|
|
if (sc->route.ro_rt != 0) /* free old route */
|
|
|
|
|
RTFREE(sc->route.ro_rt);
|
|
|
|
|
if (gre_compute_route(sc) == 0)
|
2005-08-09 10:20:02 +00:00
|
|
|
ifp->if_drv_flags |= IFF_DRV_RUNNING;
|
2002-09-06 17:12:50 +00:00
|
|
|
else
|
2005-08-09 10:20:02 +00:00
|
|
|
ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
|
2002-09-06 17:12:50 +00:00
|
|
|
}
|
|
|
|
|
break;
|
|
|
|
|
case GREGADDRS:
|
|
|
|
|
memset(&si, 0, sizeof(si));
|
|
|
|
|
si.sin_family = AF_INET;
|
|
|
|
|
si.sin_len = sizeof(struct sockaddr_in);
|
|
|
|
|
si.sin_addr.s_addr = sc->g_src.s_addr;
|
|
|
|
|
sa = sintosa(&si);
|
2011-03-02 21:39:08 +00:00
|
|
|
error = prison_if(curthread->td_ucred, sa);
|
|
|
|
|
if (error != 0)
|
|
|
|
|
break;
|
2002-09-06 17:12:50 +00:00
|
|
|
ifr->ifr_addr = *sa;
|
|
|
|
|
break;
|
|
|
|
|
case GREGADDRD:
|
|
|
|
|
memset(&si, 0, sizeof(si));
|
|
|
|
|
si.sin_family = AF_INET;
|
|
|
|
|
si.sin_len = sizeof(struct sockaddr_in);
|
|
|
|
|
si.sin_addr.s_addr = sc->g_dst.s_addr;
|
|
|
|
|
sa = sintosa(&si);
|
2011-03-02 21:39:08 +00:00
|
|
|
error = prison_if(curthread->td_ucred, sa);
|
|
|
|
|
if (error != 0)
|
|
|
|
|
break;
|
2002-09-06 17:12:50 +00:00
|
|
|
ifr->ifr_addr = *sa;
|
|
|
|
|
break;
|
|
|
|
|
case SIOCSIFPHYADDR:
|
2006-11-06 13:42:10 +00:00
|
|
|
/*
|
2007-06-26 23:01:01 +00:00
|
|
|
* XXXRW: Isn't this priv_check() redundant to the ifnet
|
|
|
|
|
* layer check?
|
2006-11-06 13:42:10 +00:00
|
|
|
*/
|
|
|
|
|
if ((error = priv_check(curthread, PRIV_NET_SETIFPHYS)) != 0)
|
2002-09-06 17:12:50 +00:00
|
|
|
break;
|
|
|
|
|
if (aifr->ifra_addr.sin_family != AF_INET ||
|
|
|
|
|
aifr->ifra_dstaddr.sin_family != AF_INET) {
|
|
|
|
|
error = EAFNOSUPPORT;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
if (aifr->ifra_addr.sin_len != sizeof(si) ||
|
|
|
|
|
aifr->ifra_dstaddr.sin_len != sizeof(si)) {
|
|
|
|
|
error = EINVAL;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
sc->g_src = aifr->ifra_addr.sin_addr;
|
|
|
|
|
sc->g_dst = aifr->ifra_dstaddr.sin_addr;
|
|
|
|
|
goto recompute;
|
|
|
|
|
case SIOCSLIFPHYADDR:
|
2006-11-06 13:42:10 +00:00
|
|
|
/*
|
2007-06-26 23:01:01 +00:00
|
|
|
* XXXRW: Isn't this priv_check() redundant to the ifnet
|
|
|
|
|
* layer check?
|
2006-11-06 13:42:10 +00:00
|
|
|
*/
|
|
|
|
|
if ((error = priv_check(curthread, PRIV_NET_SETIFPHYS)) != 0)
|
2002-09-06 17:12:50 +00:00
|
|
|
break;
|
|
|
|
|
if (lifr->addr.ss_family != AF_INET ||
|
|
|
|
|
lifr->dstaddr.ss_family != AF_INET) {
|
|
|
|
|
error = EAFNOSUPPORT;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
if (lifr->addr.ss_len != sizeof(si) ||
|
|
|
|
|
lifr->dstaddr.ss_len != sizeof(si)) {
|
|
|
|
|
error = EINVAL;
|
|
|
|
|
break;
|
|
|
|
|
}
|
2006-02-07 20:09:02 +00:00
|
|
|
sc->g_src = (satosin(&lifr->addr))->sin_addr;
|
2002-09-06 17:12:50 +00:00
|
|
|
sc->g_dst =
|
2006-02-07 20:09:02 +00:00
|
|
|
(satosin(&lifr->dstaddr))->sin_addr;
|
2002-09-06 17:12:50 +00:00
|
|
|
goto recompute;
|
|
|
|
|
case SIOCDIFPHYADDR:
|
2006-11-06 13:42:10 +00:00
|
|
|
/*
|
2007-06-26 23:01:01 +00:00
|
|
|
* XXXRW: Isn't this priv_check() redundant to the ifnet
|
|
|
|
|
* layer check?
|
2006-11-06 13:42:10 +00:00
|
|
|
*/
|
|
|
|
|
if ((error = priv_check(curthread, PRIV_NET_SETIFPHYS)) != 0)
|
2002-09-06 17:12:50 +00:00
|
|
|
break;
|
|
|
|
|
sc->g_src.s_addr = INADDR_ANY;
|
|
|
|
|
sc->g_dst.s_addr = INADDR_ANY;
|
|
|
|
|
goto recompute;
|
|
|
|
|
case SIOCGLIFPHYADDR:
|
|
|
|
|
if (sc->g_src.s_addr == INADDR_ANY ||
|
|
|
|
|
sc->g_dst.s_addr == INADDR_ANY) {
|
|
|
|
|
error = EADDRNOTAVAIL;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
memset(&si, 0, sizeof(si));
|
|
|
|
|
si.sin_family = AF_INET;
|
|
|
|
|
si.sin_len = sizeof(struct sockaddr_in);
|
|
|
|
|
si.sin_addr.s_addr = sc->g_src.s_addr;
|
2011-03-02 21:39:08 +00:00
|
|
|
error = prison_if(curthread->td_ucred, (struct sockaddr *)&si);
|
|
|
|
|
if (error != 0)
|
|
|
|
|
break;
|
2002-09-06 17:12:50 +00:00
|
|
|
memcpy(&lifr->addr, &si, sizeof(si));
|
|
|
|
|
si.sin_addr.s_addr = sc->g_dst.s_addr;
|
2011-03-02 21:39:08 +00:00
|
|
|
error = prison_if(curthread->td_ucred, (struct sockaddr *)&si);
|
|
|
|
|
if (error != 0)
|
|
|
|
|
break;
|
2002-09-06 17:12:50 +00:00
|
|
|
memcpy(&lifr->dstaddr, &si, sizeof(si));
|
|
|
|
|
break;
|
|
|
|
|
case SIOCGIFPSRCADDR:
|
2003-11-14 20:58:00 +00:00
|
|
|
#ifdef INET6
|
|
|
|
|
case SIOCGIFPSRCADDR_IN6:
|
|
|
|
|
#endif
|
2002-09-06 17:12:50 +00:00
|
|
|
if (sc->g_src.s_addr == INADDR_ANY) {
|
|
|
|
|
error = EADDRNOTAVAIL;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
memset(&si, 0, sizeof(si));
|
|
|
|
|
si.sin_family = AF_INET;
|
|
|
|
|
si.sin_len = sizeof(struct sockaddr_in);
|
|
|
|
|
si.sin_addr.s_addr = sc->g_src.s_addr;
|
2011-03-02 21:39:08 +00:00
|
|
|
error = prison_if(curthread->td_ucred, (struct sockaddr *)&si);
|
|
|
|
|
if (error != 0)
|
|
|
|
|
break;
|
2002-09-06 17:12:50 +00:00
|
|
|
bcopy(&si, &ifr->ifr_addr, sizeof(ifr->ifr_addr));
|
|
|
|
|
break;
|
|
|
|
|
case SIOCGIFPDSTADDR:
|
2003-11-14 20:58:00 +00:00
|
|
|
#ifdef INET6
|
|
|
|
|
case SIOCGIFPDSTADDR_IN6:
|
|
|
|
|
#endif
|
2002-09-06 17:12:50 +00:00
|
|
|
if (sc->g_dst.s_addr == INADDR_ANY) {
|
|
|
|
|
error = EADDRNOTAVAIL;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
memset(&si, 0, sizeof(si));
|
|
|
|
|
si.sin_family = AF_INET;
|
|
|
|
|
si.sin_len = sizeof(struct sockaddr_in);
|
|
|
|
|
si.sin_addr.s_addr = sc->g_dst.s_addr;
|
2011-03-02 21:39:08 +00:00
|
|
|
error = prison_if(curthread->td_ucred, (struct sockaddr *)&si);
|
|
|
|
|
if (error != 0)
|
|
|
|
|
break;
|
2002-09-06 17:12:50 +00:00
|
|
|
bcopy(&si, &ifr->ifr_addr, sizeof(ifr->ifr_addr));
|
|
|
|
|
break;
|
2008-06-20 17:26:34 +00:00
|
|
|
case GRESKEY:
|
|
|
|
|
error = priv_check(curthread, PRIV_NET_GRE);
|
|
|
|
|
if (error)
|
|
|
|
|
break;
|
|
|
|
|
error = copyin(ifr->ifr_data, &key, sizeof(key));
|
|
|
|
|
if (error)
|
|
|
|
|
break;
|
|
|
|
|
/* adjust MTU for option header */
|
|
|
|
|
if (key == 0 && sc->key != 0) /* clear */
|
|
|
|
|
adj += sizeof(key);
|
|
|
|
|
else if (key != 0 && sc->key == 0) /* set */
|
|
|
|
|
adj -= sizeof(key);
|
|
|
|
|
|
|
|
|
|
if (ifp->if_mtu + adj < 576) {
|
|
|
|
|
error = EINVAL;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
ifp->if_mtu += adj;
|
|
|
|
|
sc->key = key;
|
|
|
|
|
break;
|
|
|
|
|
case GREGKEY:
|
|
|
|
|
error = copyout(&sc->key, ifr->ifr_data, sizeof(sc->key));
|
|
|
|
|
break;
|
|
|
|
|
|
2002-09-06 17:12:50 +00:00
|
|
|
default:
|
|
|
|
|
error = EINVAL;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
splx(s);
|
|
|
|
|
return (error);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* computes a route to our destination that is not the one
|
|
|
|
|
* which would be taken by ip_output(), as this one will loop back to
|
|
|
|
|
* us. If the interface is p2p as a--->b, then a routing entry exists
|
|
|
|
|
* If we now send a packet to b (e.g. ping b), this will come down here
|
2003-12-30 11:41:43 +00:00
|
|
|
* gets src=a, dst=b tacked on and would from ip_output() sent back to
|
2002-09-06 17:12:50 +00:00
|
|
|
* if_gre.
|
|
|
|
|
* Goal here is to compute a route to b that is less specific than
|
|
|
|
|
* a-->b. We know that this one exists as in normal operation we have
|
|
|
|
|
* at least a default route which matches.
|
|
|
|
|
*/
|
2002-09-06 18:16:03 +00:00
|
|
|
static int
|
2002-09-06 17:12:50 +00:00
|
|
|
gre_compute_route(struct gre_softc *sc)
|
|
|
|
|
{
|
|
|
|
|
struct route *ro;
|
|
|
|
|
|
|
|
|
|
ro = &sc->route;
|
|
|
|
|
|
|
|
|
|
memset(ro, 0, sizeof(struct route));
|
|
|
|
|
((struct sockaddr_in *)&ro->ro_dst)->sin_addr = sc->g_dst;
|
|
|
|
|
ro->ro_dst.sa_family = AF_INET;
|
|
|
|
|
ro->ro_dst.sa_len = sizeof(ro->ro_dst);
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* toggle last bit, so our interface is not found, but a less
|
|
|
|
|
* specific route. I'd rather like to specify a shorter mask,
|
|
|
|
|
* but this is not possible. Should work though. XXX
|
Add code to allow the system to handle multiple routing tables.
This particular implementation is designed to be fully backwards compatible
and to be MFC-able to 7.x (and 6.x)
Currently the only protocol that can make use of the multiple tables is IPv4
Similar functionality exists in OpenBSD and Linux.
From my notes:
-----
One thing where FreeBSD has been falling behind, and which by chance I
have some time to work on is "policy based routing", which allows
different
packet streams to be routed by more than just the destination address.
Constraints:
------------
I want to make some form of this available in the 6.x tree
(and by extension 7.x) , but FreeBSD in general needs it so I might as
well do it in -current and back port the portions I need.
One of the ways that this can be done is to have the ability to
instantiate multiple kernel routing tables (which I will now
refer to as "Forwarding Information Bases" or "FIBs" for political
correctness reasons). Which FIB a particular packet uses to make
the next hop decision can be decided by a number of mechanisms.
The policies these mechanisms implement are the "Policies" referred
to in "Policy based routing".
One of the constraints I have if I try to back port this work to
6.x is that it must be implemented as a EXTENSION to the existing
ABIs in 6.x so that third party applications do not need to be
recompiled in timespan of the branch.
This first version will not have some of the bells and whistles that
will come with later versions. It will, for example, be limited to 16
tables in the first commit.
Implementation method, Compatible version. (part 1)
-------------------------------
For this reason I have implemented a "sufficient subset" of a
multiple routing table solution in Perforce, and back-ported it
to 6.x. (also in Perforce though not always caught up with what I
have done in -current/P4). The subset allows a number of FIBs
to be defined at compile time (8 is sufficient for my purposes in 6.x)
and implements the changes needed to allow IPV4 to use them. I have not
done the changes for ipv6 simply because I do not need it, and I do not
have enough knowledge of ipv6 (e.g. neighbor discovery) needed to do it.
Other protocol families are left untouched and should there be
users with proprietary protocol families, they should continue to work
and be oblivious to the existence of the extra FIBs.
To understand how this is done, one must know that the current FIB
code starts everything off with a single dimensional array of
pointers to FIB head structures (One per protocol family), each of
which in turn points to the trie of routes available to that family.
The basic change in the ABI compatible version of the change is to
extent that array to be a 2 dimensional array, so that
instead of protocol family X looking at rt_tables[X] for the
table it needs, it looks at rt_tables[Y][X] when for all
protocol families except ipv4 Y is always 0.
Code that is unaware of the change always just sees the first row
of the table, which of course looks just like the one dimensional
array that existed before.
The entry points rtrequest(), rtalloc(), rtalloc1(), rtalloc_ign()
are all maintained, but refer only to the first row of the array,
so that existing callers in proprietary protocols can continue to
do the "right thing".
Some new entry points are added, for the exclusive use of ipv4 code
called in_rtrequest(), in_rtalloc(), in_rtalloc1() and in_rtalloc_ign(),
which have an extra argument which refers the code to the correct row.
In addition, there are some new entry points (currently called
rtalloc_fib() and friends) that check the Address family being
looked up and call either rtalloc() (and friends) if the protocol
is not IPv4 forcing the action to row 0 or to the appropriate row
if it IS IPv4 (and that info is available). These are for calling
from code that is not specific to any particular protocol. The way
these are implemented would change in the non ABI preserving code
to be added later.
One feature of the first version of the code is that for ipv4,
the interface routes show up automatically on all the FIBs, so
that no matter what FIB you select you always have the basic
direct attached hosts available to you. (rtinit() does this
automatically).
You CAN delete an interface route from one FIB should you want
to but by default it's there. ARP information is also available
in each FIB. It's assumed that the same machine would have the
same MAC address, regardless of which FIB you are using to get
to it.
This brings us as to how the correct FIB is selected for an outgoing
IPV4 packet.
Firstly, all packets have a FIB associated with them. if nothing
has been done to change it, it will be FIB 0. The FIB is changed
in the following ways.
Packets fall into one of a number of classes.
1/ locally generated packets, coming from a socket/PCB.
Such packets select a FIB from a number associated with the
socket/PCB. This in turn is inherited from the process,
but can be changed by a socket option. The process in turn
inherits it on fork. I have written a utility call setfib
that acts a bit like nice..
setfib -3 ping target.example.com # will use fib 3 for ping.
It is an obvious extension to make it a property of a jail
but I have not done so. It can be achieved by combining the setfib and
jail commands.
2/ packets received on an interface for forwarding.
By default these packets would use table 0,
(or possibly a number settable in a sysctl(not yet)).
but prior to routing the firewall can inspect them (see below).
(possibly in the future you may be able to associate a FIB
with packets received on an interface.. An ifconfig arg, but not yet.)
3/ packets inspected by a packet classifier, which can arbitrarily
associate a fib with it on a packet by packet basis.
A fib assigned to a packet by a packet classifier
(such as ipfw) would over-ride a fib associated by
a more default source. (such as cases 1 or 2).
4/ a tcp listen socket associated with a fib will generate
accept sockets that are associated with that same fib.
5/ Packets generated in response to some other packet (e.g. reset
or icmp packets). These should use the FIB associated with the
packet being reponded to.
6/ Packets generated during encapsulation.
gif, tun and other tunnel interfaces will encapsulate using the FIB
that was in effect withthe proces that set up the tunnel.
thus setfib 1 ifconfig gif0 [tunnel instructions]
will set the fib for the tunnel to use to be fib 1.
Routing messages would be associated with their
process, and thus select one FIB or another.
messages from the kernel would be associated with the fib they
refer to and would only be received by a routing socket associated
with that fib. (not yet implemented)
In addition Netstat has been edited to be able to cope with the
fact that the array is now 2 dimensional. (It looks in system
memory using libkvm (!)). Old versions of netstat see only the first FIB.
In addition two sysctls are added to give:
a) the number of FIBs compiled in (active)
b) the default FIB of the calling process.
Early testing experience:
-------------------------
Basically our (IronPort's) appliance does this functionality already
using ipfw fwd but that method has some drawbacks.
For example,
It can't fully simulate a routing table because it can't influence the
socket's choice of local address when a connect() is done.
Testing during the generating of these changes has been
remarkably smooth so far. Multiple tables have co-existed
with no notable side effects, and packets have been routes
accordingly.
ipfw has grown 2 new keywords:
setfib N ip from anay to any
count ip from any to any fib N
In pf there seems to be a requirement to be able to give symbolic names to the
fibs but I do not have that capacity. I am not sure if it is required.
SCTP has interestingly enough built in support for this, called VRFs
in Cisco parlance. it will be interesting to see how that handles it
when it suddenly actually does something.
Where to next:
--------------------
After committing the ABI compatible version and MFCing it, I'd
like to proceed in a forward direction in -current. this will
result in some roto-tilling in the routing code.
Firstly: the current code's idea of having a separate tree per
protocol family, all of the same format, and pointed to by the
1 dimensional array is a bit silly. Especially when one considers that
there is code that makes assumptions about every protocol having the
same internal structures there. Some protocols don't WANT that
sort of structure. (for example the whole idea of a netmask is foreign
to appletalk). This needs to be made opaque to the external code.
My suggested first change is to add routing method pointers to the
'domain' structure, along with information pointing the data.
instead of having an array of pointers to uniform structures,
there would be an array pointing to the 'domain' structures
for each protocol address domain (protocol family),
and the methods this reached would be called. The methods would have
an argument that gives FIB number, but the protocol would be free
to ignore it.
When the ABI can be changed it raises the possibilty of the
addition of a fib entry into the "struct route". Currently,
the structure contains the sockaddr of the desination, and the resulting
fib entry. To make this work fully, one could add a fib number
so that given an address and a fib, one can find the third element, the
fib entry.
Interaction with the ARP layer/ LL layer would need to be
revisited as well. Qing Li has been working on this already.
This work was sponsored by Ironport Systems/Cisco
Reviewed by: several including rwatson, bz and mlair (parts each)
Obtained from: Ironport systems/Cisco
2008-05-09 23:03:00 +00:00
|
|
|
* XXX MRT Use a different FIB for the tunnel to solve this problem.
|
2002-09-06 17:12:50 +00:00
|
|
|
*/
|
2005-06-10 16:49:24 +00:00
|
|
|
if ((GRE2IFP(sc)->if_flags & IFF_LINK1) == 0) {
|
2008-03-19 22:29:11 +00:00
|
|
|
((struct sockaddr_in *)&ro->ro_dst)->sin_addr.s_addr ^=
|
|
|
|
|
htonl(0x01);
|
2002-09-06 17:12:50 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
#ifdef DIAGNOSTIC
|
2005-06-10 16:49:24 +00:00
|
|
|
printf("%s: searching for a route to %s", if_name(GRE2IFP(sc)),
|
2002-09-06 17:12:50 +00:00
|
|
|
inet_ntoa(((struct sockaddr_in *)&ro->ro_dst)->sin_addr));
|
|
|
|
|
#endif
|
|
|
|
|
|
Add code to allow the system to handle multiple routing tables.
This particular implementation is designed to be fully backwards compatible
and to be MFC-able to 7.x (and 6.x)
Currently the only protocol that can make use of the multiple tables is IPv4
Similar functionality exists in OpenBSD and Linux.
From my notes:
-----
One thing where FreeBSD has been falling behind, and which by chance I
have some time to work on is "policy based routing", which allows
different
packet streams to be routed by more than just the destination address.
Constraints:
------------
I want to make some form of this available in the 6.x tree
(and by extension 7.x) , but FreeBSD in general needs it so I might as
well do it in -current and back port the portions I need.
One of the ways that this can be done is to have the ability to
instantiate multiple kernel routing tables (which I will now
refer to as "Forwarding Information Bases" or "FIBs" for political
correctness reasons). Which FIB a particular packet uses to make
the next hop decision can be decided by a number of mechanisms.
The policies these mechanisms implement are the "Policies" referred
to in "Policy based routing".
One of the constraints I have if I try to back port this work to
6.x is that it must be implemented as a EXTENSION to the existing
ABIs in 6.x so that third party applications do not need to be
recompiled in timespan of the branch.
This first version will not have some of the bells and whistles that
will come with later versions. It will, for example, be limited to 16
tables in the first commit.
Implementation method, Compatible version. (part 1)
-------------------------------
For this reason I have implemented a "sufficient subset" of a
multiple routing table solution in Perforce, and back-ported it
to 6.x. (also in Perforce though not always caught up with what I
have done in -current/P4). The subset allows a number of FIBs
to be defined at compile time (8 is sufficient for my purposes in 6.x)
and implements the changes needed to allow IPV4 to use them. I have not
done the changes for ipv6 simply because I do not need it, and I do not
have enough knowledge of ipv6 (e.g. neighbor discovery) needed to do it.
Other protocol families are left untouched and should there be
users with proprietary protocol families, they should continue to work
and be oblivious to the existence of the extra FIBs.
To understand how this is done, one must know that the current FIB
code starts everything off with a single dimensional array of
pointers to FIB head structures (One per protocol family), each of
which in turn points to the trie of routes available to that family.
The basic change in the ABI compatible version of the change is to
extent that array to be a 2 dimensional array, so that
instead of protocol family X looking at rt_tables[X] for the
table it needs, it looks at rt_tables[Y][X] when for all
protocol families except ipv4 Y is always 0.
Code that is unaware of the change always just sees the first row
of the table, which of course looks just like the one dimensional
array that existed before.
The entry points rtrequest(), rtalloc(), rtalloc1(), rtalloc_ign()
are all maintained, but refer only to the first row of the array,
so that existing callers in proprietary protocols can continue to
do the "right thing".
Some new entry points are added, for the exclusive use of ipv4 code
called in_rtrequest(), in_rtalloc(), in_rtalloc1() and in_rtalloc_ign(),
which have an extra argument which refers the code to the correct row.
In addition, there are some new entry points (currently called
rtalloc_fib() and friends) that check the Address family being
looked up and call either rtalloc() (and friends) if the protocol
is not IPv4 forcing the action to row 0 or to the appropriate row
if it IS IPv4 (and that info is available). These are for calling
from code that is not specific to any particular protocol. The way
these are implemented would change in the non ABI preserving code
to be added later.
One feature of the first version of the code is that for ipv4,
the interface routes show up automatically on all the FIBs, so
that no matter what FIB you select you always have the basic
direct attached hosts available to you. (rtinit() does this
automatically).
You CAN delete an interface route from one FIB should you want
to but by default it's there. ARP information is also available
in each FIB. It's assumed that the same machine would have the
same MAC address, regardless of which FIB you are using to get
to it.
This brings us as to how the correct FIB is selected for an outgoing
IPV4 packet.
Firstly, all packets have a FIB associated with them. if nothing
has been done to change it, it will be FIB 0. The FIB is changed
in the following ways.
Packets fall into one of a number of classes.
1/ locally generated packets, coming from a socket/PCB.
Such packets select a FIB from a number associated with the
socket/PCB. This in turn is inherited from the process,
but can be changed by a socket option. The process in turn
inherits it on fork. I have written a utility call setfib
that acts a bit like nice..
setfib -3 ping target.example.com # will use fib 3 for ping.
It is an obvious extension to make it a property of a jail
but I have not done so. It can be achieved by combining the setfib and
jail commands.
2/ packets received on an interface for forwarding.
By default these packets would use table 0,
(or possibly a number settable in a sysctl(not yet)).
but prior to routing the firewall can inspect them (see below).
(possibly in the future you may be able to associate a FIB
with packets received on an interface.. An ifconfig arg, but not yet.)
3/ packets inspected by a packet classifier, which can arbitrarily
associate a fib with it on a packet by packet basis.
A fib assigned to a packet by a packet classifier
(such as ipfw) would over-ride a fib associated by
a more default source. (such as cases 1 or 2).
4/ a tcp listen socket associated with a fib will generate
accept sockets that are associated with that same fib.
5/ Packets generated in response to some other packet (e.g. reset
or icmp packets). These should use the FIB associated with the
packet being reponded to.
6/ Packets generated during encapsulation.
gif, tun and other tunnel interfaces will encapsulate using the FIB
that was in effect withthe proces that set up the tunnel.
thus setfib 1 ifconfig gif0 [tunnel instructions]
will set the fib for the tunnel to use to be fib 1.
Routing messages would be associated with their
process, and thus select one FIB or another.
messages from the kernel would be associated with the fib they
refer to and would only be received by a routing socket associated
with that fib. (not yet implemented)
In addition Netstat has been edited to be able to cope with the
fact that the array is now 2 dimensional. (It looks in system
memory using libkvm (!)). Old versions of netstat see only the first FIB.
In addition two sysctls are added to give:
a) the number of FIBs compiled in (active)
b) the default FIB of the calling process.
Early testing experience:
-------------------------
Basically our (IronPort's) appliance does this functionality already
using ipfw fwd but that method has some drawbacks.
For example,
It can't fully simulate a routing table because it can't influence the
socket's choice of local address when a connect() is done.
Testing during the generating of these changes has been
remarkably smooth so far. Multiple tables have co-existed
with no notable side effects, and packets have been routes
accordingly.
ipfw has grown 2 new keywords:
setfib N ip from anay to any
count ip from any to any fib N
In pf there seems to be a requirement to be able to give symbolic names to the
fibs but I do not have that capacity. I am not sure if it is required.
SCTP has interestingly enough built in support for this, called VRFs
in Cisco parlance. it will be interesting to see how that handles it
when it suddenly actually does something.
Where to next:
--------------------
After committing the ABI compatible version and MFCing it, I'd
like to proceed in a forward direction in -current. this will
result in some roto-tilling in the routing code.
Firstly: the current code's idea of having a separate tree per
protocol family, all of the same format, and pointed to by the
1 dimensional array is a bit silly. Especially when one considers that
there is code that makes assumptions about every protocol having the
same internal structures there. Some protocols don't WANT that
sort of structure. (for example the whole idea of a netmask is foreign
to appletalk). This needs to be made opaque to the external code.
My suggested first change is to add routing method pointers to the
'domain' structure, along with information pointing the data.
instead of having an array of pointers to uniform structures,
there would be an array pointing to the 'domain' structures
for each protocol address domain (protocol family),
and the methods this reached would be called. The methods would have
an argument that gives FIB number, but the protocol would be free
to ignore it.
When the ABI can be changed it raises the possibilty of the
addition of a fib entry into the "struct route". Currently,
the structure contains the sockaddr of the desination, and the resulting
fib entry. To make this work fully, one could add a fib number
so that given an address and a fib, one can find the third element, the
fib entry.
Interaction with the ARP layer/ LL layer would need to be
revisited as well. Qing Li has been working on this already.
This work was sponsored by Ironport Systems/Cisco
Reviewed by: several including rwatson, bz and mlair (parts each)
Obtained from: Ironport systems/Cisco
2008-05-09 23:03:00 +00:00
|
|
|
rtalloc_fib(ro, sc->gre_fibnum);
|
2002-09-06 17:12:50 +00:00
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* check if this returned a route at all and this route is no
|
|
|
|
|
* recursion to ourself
|
|
|
|
|
*/
|
|
|
|
|
if (ro->ro_rt == NULL || ro->ro_rt->rt_ifp->if_softc == sc) {
|
|
|
|
|
#ifdef DIAGNOSTIC
|
|
|
|
|
if (ro->ro_rt == NULL)
|
|
|
|
|
printf(" - no route found!\n");
|
|
|
|
|
else
|
|
|
|
|
printf(" - route loops back to ourself!\n");
|
|
|
|
|
#endif
|
|
|
|
|
return EADDRNOTAVAIL;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* now change it back - else ip_output will just drop
|
|
|
|
|
* the route and search one to this interface ...
|
|
|
|
|
*/
|
2005-06-10 16:49:24 +00:00
|
|
|
if ((GRE2IFP(sc)->if_flags & IFF_LINK1) == 0)
|
2002-09-06 17:12:50 +00:00
|
|
|
((struct sockaddr_in *)&ro->ro_dst)->sin_addr = sc->g_dst;
|
|
|
|
|
|
|
|
|
|
#ifdef DIAGNOSTIC
|
|
|
|
|
printf(", choosing %s with gateway %s", if_name(ro->ro_rt->rt_ifp),
|
|
|
|
|
inet_ntoa(((struct sockaddr_in *)(ro->ro_rt->rt_gateway))->sin_addr));
|
|
|
|
|
printf("\n");
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* do a checksum of a buffer - much like in_cksum, which operates on
|
|
|
|
|
* mbufs.
|
|
|
|
|
*/
|
2003-12-30 11:41:43 +00:00
|
|
|
u_int16_t
|
|
|
|
|
gre_in_cksum(u_int16_t *p, u_int len)
|
2002-09-06 17:12:50 +00:00
|
|
|
{
|
2003-12-30 11:41:43 +00:00
|
|
|
u_int32_t sum = 0;
|
2002-09-06 17:12:50 +00:00
|
|
|
int nwords = len >> 1;
|
|
|
|
|
|
|
|
|
|
while (nwords-- != 0)
|
|
|
|
|
sum += *p++;
|
|
|
|
|
|
|
|
|
|
if (len & 1) {
|
|
|
|
|
union {
|
|
|
|
|
u_short w;
|
|
|
|
|
u_char c[2];
|
|
|
|
|
} u;
|
|
|
|
|
u.c[0] = *(u_char *)p;
|
|
|
|
|
u.c[1] = 0;
|
|
|
|
|
sum += u.w;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* end-around-carry */
|
|
|
|
|
sum = (sum >> 16) + (sum & 0xffff);
|
|
|
|
|
sum += (sum >> 16);
|
|
|
|
|
return (~sum);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static int
|
|
|
|
|
gremodevent(module_t mod, int type, void *data)
|
|
|
|
|
{
|
|
|
|
|
|
|
|
|
|
switch (type) {
|
|
|
|
|
case MOD_LOAD:
|
|
|
|
|
greattach();
|
|
|
|
|
break;
|
|
|
|
|
case MOD_UNLOAD:
|
2012-10-16 13:37:54 +00:00
|
|
|
if_clone_detach(gre_cloner);
|
2004-03-22 16:04:43 +00:00
|
|
|
mtx_destroy(&gre_mtx);
|
2002-09-06 17:12:50 +00:00
|
|
|
break;
|
2004-07-15 08:26:07 +00:00
|
|
|
default:
|
|
|
|
|
return EOPNOTSUPP;
|
2002-09-06 17:12:50 +00:00
|
|
|
}
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static moduledata_t gre_mod = {
|
|
|
|
|
"if_gre",
|
|
|
|
|
gremodevent,
|
2012-10-10 08:36:38 +00:00
|
|
|
0
|
2002-09-06 17:12:50 +00:00
|
|
|
};
|
|
|
|
|
|
|
|
|
|
DECLARE_MODULE(if_gre, gre_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
|
|
|
|
|
MODULE_VERSION(if_gre, 1);
|
