2005-01-07 01:45:51 +00:00
|
|
|
/*-
|
1994-05-24 10:09:53 +00:00
|
|
|
* Copyright (c) 1988, 1991, 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.
|
|
|
|
* 4. 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.
|
|
|
|
*
|
2001-10-17 11:23:59 +00:00
|
|
|
* @(#)rtsock.c 8.7 (Berkeley) 10/12/95
|
1999-08-28 01:08:13 +00:00
|
|
|
* $FreeBSD$
|
1994-05-24 10:09:53 +00:00
|
|
|
*/
|
2006-11-03 15:23:16 +00:00
|
|
|
#include "opt_sctp.h"
|
This patch provides the back end support for equal-cost multi-path
(ECMP) for both IPv4 and IPv6. Previously, multipath route insertion
is disallowed. For example,
route add -net 192.103.54.0/24 10.9.44.1
route add -net 192.103.54.0/24 10.9.44.2
The second route insertion will trigger an error message of
"add net 192.103.54.0/24: gateway 10.2.5.2: route already in table"
Multiple default routes can also be inserted. Here is the netstat
output:
default 10.2.5.1 UGS 0 3074 bge0 =>
default 10.2.5.2 UGS 0 0 bge0
When multipath routes exist, the "route delete" command requires
a specific gateway to be specified or else an error message would
be displayed. For example,
route delete default
would fail and trigger the following error message:
"route: writing to routing socket: No such process"
"delete net default: not in table"
On the other hand,
route delete default 10.2.5.2
would be successful: "delete net default: gateway 10.2.5.2"
One does not have to specify a gateway if there is only a single
route for a particular destination.
I need to perform more testings on address aliases and multiple
interfaces that have the same IP prefixes. This patch as it
stands today is not yet ready for prime time. Therefore, the ECMP
code fragments are fully guarded by the RADIX_MPATH macro.
Include the "options RADIX_MPATH" in the kernel configuration
to enable this feature.
Reviewed by: robert, sam, gnn, julian, kmacy
2008-04-13 05:45:14 +00:00
|
|
|
#include "opt_mpath.h"
|
MFp4:
Bring in updated jail support from bz_jail branch.
This enhances the current jail implementation to permit multiple
addresses per jail. In addtion to IPv4, IPv6 is supported as well.
Due to updated checks it is even possible to have jails without
an IP address at all, which basically gives one a chroot with
restricted process view, no networking,..
SCTP support was updated and supports IPv6 in jails as well.
Cpuset support permits jails to be bound to specific processor
sets after creation.
Jails can have an unrestricted (no duplicate protection, etc.) name
in addition to the hostname. The jail name cannot be changed from
within a jail and is considered to be used for management purposes
or as audit-token in the future.
DDB 'show jails' command was added to aid debugging.
Proper compat support permits 32bit jail binaries to be used on 64bit
systems to manage jails. Also backward compatibility was preserved where
possible: for jail v1 syscalls, as well as with user space management
utilities.
Both jail as well as prison version were updated for the new features.
A gap was intentionally left as the intermediate versions had been
used by various patches floating around the last years.
Bump __FreeBSD_version for the afore mentioned and in kernel changes.
Special thanks to:
- Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches
and Olivier Houchard (cognet) for initial single-IPv6 patches.
- Jeff Roberson (jeff) and Randall Stewart (rrs) for their
help, ideas and review on cpuset and SCTP support.
- Robert Watson (rwatson) for lots and lots of help, discussions,
suggestions and review of most of the patch at various stages.
- John Baldwin (jhb) for his help.
- Simon L. Nielsen (simon) as early adopter testing changes
on cluster machines as well as all the testers and people
who provided feedback the last months on freebsd-jail and
other channels.
- My employer, CK Software GmbH, for the support so I could work on this.
Reviewed by: (see above)
MFC after: 3 months (this is just so that I get the mail)
X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
|
|
|
#include "opt_inet.h"
|
|
|
|
#include "opt_inet6.h"
|
This patch provides the back end support for equal-cost multi-path
(ECMP) for both IPv4 and IPv6. Previously, multipath route insertion
is disallowed. For example,
route add -net 192.103.54.0/24 10.9.44.1
route add -net 192.103.54.0/24 10.9.44.2
The second route insertion will trigger an error message of
"add net 192.103.54.0/24: gateway 10.2.5.2: route already in table"
Multiple default routes can also be inserted. Here is the netstat
output:
default 10.2.5.1 UGS 0 3074 bge0 =>
default 10.2.5.2 UGS 0 0 bge0
When multipath routes exist, the "route delete" command requires
a specific gateway to be specified or else an error message would
be displayed. For example,
route delete default
would fail and trigger the following error message:
"route: writing to routing socket: No such process"
"delete net default: not in table"
On the other hand,
route delete default 10.2.5.2
would be successful: "delete net default: gateway 10.2.5.2"
One does not have to specify a gateway if there is only a single
route for a particular destination.
I need to perform more testings on address aliases and multiple
interfaces that have the same IP prefixes. This patch as it
stands today is not yet ready for prime time. Therefore, the ECMP
code fragments are fully guarded by the RADIX_MPATH macro.
Include the "options RADIX_MPATH" in the kernel configuration
to enable this feature.
Reviewed by: robert, sam, gnn, julian, kmacy
2008-04-13 05:45:14 +00:00
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
#include <sys/param.h>
|
2002-04-30 01:54:54 +00:00
|
|
|
#include <sys/domain.h>
|
|
|
|
#include <sys/jail.h>
|
MFp4:
Bring in updated jail support from bz_jail branch.
This enhances the current jail implementation to permit multiple
addresses per jail. In addtion to IPv4, IPv6 is supported as well.
Due to updated checks it is even possible to have jails without
an IP address at all, which basically gives one a chroot with
restricted process view, no networking,..
SCTP support was updated and supports IPv6 in jails as well.
Cpuset support permits jails to be bound to specific processor
sets after creation.
Jails can have an unrestricted (no duplicate protection, etc.) name
in addition to the hostname. The jail name cannot be changed from
within a jail and is considered to be used for management purposes
or as audit-token in the future.
DDB 'show jails' command was added to aid debugging.
Proper compat support permits 32bit jail binaries to be used on 64bit
systems to manage jails. Also backward compatibility was preserved where
possible: for jail v1 syscalls, as well as with user space management
utilities.
Both jail as well as prison version were updated for the new features.
A gap was intentionally left as the intermediate versions had been
used by various patches floating around the last years.
Bump __FreeBSD_version for the afore mentioned and in kernel changes.
Special thanks to:
- Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches
and Olivier Houchard (cognet) for initial single-IPv6 patches.
- Jeff Roberson (jeff) and Randall Stewart (rrs) for their
help, ideas and review on cpuset and SCTP support.
- Robert Watson (rwatson) for lots and lots of help, discussions,
suggestions and review of most of the patch at various stages.
- John Baldwin (jhb) for his help.
- Simon L. Nielsen (simon) as early adopter testing changes
on cluster machines as well as all the testers and people
who provided feedback the last months on freebsd-jail and
other channels.
- My employer, CK Software GmbH, for the support so I could work on this.
Reviewed by: (see above)
MFC after: 3 months (this is just so that I get the mail)
X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
|
|
|
#include <sys/kernel.h>
|
2008-12-08 00:28:21 +00:00
|
|
|
#include <sys/lock.h>
|
1997-09-02 01:19:47 +00:00
|
|
|
#include <sys/malloc.h>
|
1994-05-24 10:09:53 +00:00
|
|
|
#include <sys/mbuf.h>
|
2006-11-06 13:42:10 +00:00
|
|
|
#include <sys/priv.h>
|
2002-04-30 01:54:54 +00:00
|
|
|
#include <sys/proc.h>
|
|
|
|
#include <sys/protosw.h>
|
2008-12-07 21:15:43 +00:00
|
|
|
#include <sys/rwlock.h>
|
2002-04-30 01:54:54 +00:00
|
|
|
#include <sys/signalvar.h>
|
1994-05-24 10:09:53 +00:00
|
|
|
#include <sys/socket.h>
|
|
|
|
#include <sys/socketvar.h>
|
2002-04-30 01:54:54 +00:00
|
|
|
#include <sys/sysctl.h>
|
|
|
|
#include <sys/systm.h>
|
Commit step 1 of the vimage project, (network stack)
virtualization work done by Marko Zec (zec@).
This is the first in a series of commits over the course
of the next few weeks.
Mark all uses of global variables to be virtualized
with a V_ prefix.
Use macros to map them back to their global names for
now, so this is a NOP change only.
We hope to have caught at least 85-90% of what is needed
so we do not invalidate a lot of outstanding patches again.
Obtained from: //depot/projects/vimage-commit2/...
Reviewed by: brooks, des, ed, mav, julian,
jamie, kris, rwatson, zec, ...
(various people I forgot, different versions)
md5 (with a bit of help)
Sponsored by: NLnet Foundation, The FreeBSD Foundation
X-MFC after: never
V_Commit_Message_Reviewed_By: more people than the patch
2008-08-17 23:27:27 +00:00
|
|
|
#include <sys/vimage.h>
|
1994-05-24 10:09:53 +00:00
|
|
|
|
|
|
|
#include <net/if.h>
|
2008-12-26 19:45:24 +00:00
|
|
|
#include <net/if_dl.h>
|
This main goals of this project are:
1. separating L2 tables (ARP, NDP) from the L3 routing tables
2. removing as much locking dependencies among these layers as
possible to allow for some parallelism in the search operations
3. simplify the logic in the routing code,
The most notable end result is the obsolescent of the route
cloning (RTF_CLONING) concept, which translated into code reduction
in both IPv4 ARP and IPv6 NDP related modules, and size reduction in
struct rtentry{}. The change in design obsoletes the semantics of
RTF_CLONING, RTF_WASCLONE and RTF_LLINFO routing flags. The userland
applications such as "arp" and "ndp" have been modified to reflect
those changes. The output from "netstat -r" shows only the routing
entries.
Quite a few developers have contributed to this project in the
past: Glebius Smirnoff, Luigi Rizzo, Alessandro Cerri, and
Andre Oppermann. And most recently:
- Kip Macy revised the locking code completely, thus completing
the last piece of the puzzle, Kip has also been conducting
active functional testing
- Sam Leffler has helped me improving/refactoring the code, and
provided valuable reviews
- Julian Elischer setup the perforce tree for me and has helped
me maintaining that branch before the svn conversion
2008-12-15 06:10:57 +00:00
|
|
|
#include <net/if_llatbl.h>
|
Introduce a netisr to deliver kernel-generated routing, avoiding
recursive entering of the socket code from the routing code:
- Modify rt_dispatch() to bundle up the sockaddr family, if any,
associated with a pending mbuf to dispatch to routing sockets, in
an m_tag on the mbuf.
- Allocate NETISR_ROUTE for use by routing sockets.
- Introduce rtsintrq, an ifqueue to be used by the netisr, and
introduce rts_input(), a function to unbundle the tagged sockaddr
and inject the mbuf and address into raw_input(), which previously
occurred in rt_dispatch().
- Introduce rts_init() to initialize rtsintrq, its mutex, and
register the netisr. Perform this at the same point in system
initialization as setup of the domains.
This change introduces asynchrony between the generation of a
pending routing socket message and delivery to sockets for use
by userspace. It avoids socket->routing->rtsock->socket use and
helps to avoid lock order reversals between the routing code and
socket code (in particular, raw socket control blocks), as route
locks are held over calls to rt_dispatch().
Reviewed by: "George V.Neville-Neil" <gnn@neville-neil.com>
Conceptual head nod by: sam
2004-06-09 02:48:23 +00:00
|
|
|
#include <net/netisr.h>
|
1994-05-24 10:09:53 +00:00
|
|
|
#include <net/raw_cb.h>
|
2002-04-30 01:54:54 +00:00
|
|
|
#include <net/route.h>
|
2008-12-02 21:37:28 +00:00
|
|
|
#include <net/vnet.h>
|
1994-05-24 10:09:53 +00:00
|
|
|
|
2004-04-26 19:46:52 +00:00
|
|
|
#include <netinet/in.h>
|
MFp4:
Bring in updated jail support from bz_jail branch.
This enhances the current jail implementation to permit multiple
addresses per jail. In addtion to IPv4, IPv6 is supported as well.
Due to updated checks it is even possible to have jails without
an IP address at all, which basically gives one a chroot with
restricted process view, no networking,..
SCTP support was updated and supports IPv6 in jails as well.
Cpuset support permits jails to be bound to specific processor
sets after creation.
Jails can have an unrestricted (no duplicate protection, etc.) name
in addition to the hostname. The jail name cannot be changed from
within a jail and is considered to be used for management purposes
or as audit-token in the future.
DDB 'show jails' command was added to aid debugging.
Proper compat support permits 32bit jail binaries to be used on 64bit
systems to manage jails. Also backward compatibility was preserved where
possible: for jail v1 syscalls, as well as with user space management
utilities.
Both jail as well as prison version were updated for the new features.
A gap was intentionally left as the intermediate versions had been
used by various patches floating around the last years.
Bump __FreeBSD_version for the afore mentioned and in kernel changes.
Special thanks to:
- Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches
and Olivier Houchard (cognet) for initial single-IPv6 patches.
- Jeff Roberson (jeff) and Randall Stewart (rrs) for their
help, ideas and review on cpuset and SCTP support.
- Robert Watson (rwatson) for lots and lots of help, discussions,
suggestions and review of most of the patch at various stages.
- John Baldwin (jhb) for his help.
- Simon L. Nielsen (simon) as early adopter testing changes
on cluster machines as well as all the testers and people
who provided feedback the last months on freebsd-jail and
other channels.
- My employer, CK Software GmbH, for the support so I could work on this.
Reviewed by: (see above)
MFC after: 3 months (this is just so that I get the mail)
X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
|
|
|
#ifdef INET6
|
|
|
|
#include <netinet6/scope6_var.h>
|
|
|
|
#endif
|
2004-04-26 19:46:52 +00:00
|
|
|
|
2006-11-03 15:23:16 +00:00
|
|
|
#ifdef SCTP
|
|
|
|
extern void sctp_addr_change(struct ifaddr *ifa, int cmd);
|
|
|
|
#endif /* SCTP */
|
|
|
|
|
1997-10-12 20:26:33 +00:00
|
|
|
MALLOC_DEFINE(M_RTABLE, "routetbl", "routing tables");
|
|
|
|
|
2003-10-03 18:15:54 +00:00
|
|
|
/* NB: these are not modified */
|
1995-11-16 19:00:27 +00:00
|
|
|
static struct sockaddr route_src = { 2, PF_ROUTE, };
|
1997-07-18 11:44:24 +00:00
|
|
|
static struct sockaddr sa_zero = { sizeof(sa_zero), AF_INET, };
|
2003-10-03 18:15:54 +00:00
|
|
|
|
|
|
|
static struct {
|
2004-08-24 08:47:15 +00:00
|
|
|
int ip_count; /* attached w/ AF_INET */
|
2003-10-03 18:15:54 +00:00
|
|
|
int ip6_count; /* attached w/ AF_INET6 */
|
|
|
|
int ipx_count; /* attached w/ AF_IPX */
|
|
|
|
int any_count; /* total attached */
|
|
|
|
} route_cb;
|
1994-05-24 10:09:53 +00:00
|
|
|
|
2003-10-03 18:27:02 +00:00
|
|
|
struct mtx rtsock_mtx;
|
|
|
|
MTX_SYSINIT(rtsock, &rtsock_mtx, "rtsock route_cb lock", MTX_DEF);
|
|
|
|
|
|
|
|
#define RTSOCK_LOCK() mtx_lock(&rtsock_mtx)
|
|
|
|
#define RTSOCK_UNLOCK() mtx_unlock(&rtsock_mtx)
|
|
|
|
#define RTSOCK_LOCK_ASSERT() mtx_assert(&rtsock_mtx, MA_OWNED)
|
|
|
|
|
Introduce a netisr to deliver kernel-generated routing, avoiding
recursive entering of the socket code from the routing code:
- Modify rt_dispatch() to bundle up the sockaddr family, if any,
associated with a pending mbuf to dispatch to routing sockets, in
an m_tag on the mbuf.
- Allocate NETISR_ROUTE for use by routing sockets.
- Introduce rtsintrq, an ifqueue to be used by the netisr, and
introduce rts_input(), a function to unbundle the tagged sockaddr
and inject the mbuf and address into raw_input(), which previously
occurred in rt_dispatch().
- Introduce rts_init() to initialize rtsintrq, its mutex, and
register the netisr. Perform this at the same point in system
initialization as setup of the domains.
This change introduces asynchrony between the generation of a
pending routing socket message and delivery to sockets for use
by userspace. It avoids socket->routing->rtsock->socket use and
helps to avoid lock order reversals between the routing code and
socket code (in particular, raw socket control blocks), as route
locks are held over calls to rt_dispatch().
Reviewed by: "George V.Neville-Neil" <gnn@neville-neil.com>
Conceptual head nod by: sam
2004-06-09 02:48:23 +00:00
|
|
|
static struct ifqueue rtsintrq;
|
|
|
|
|
2004-08-21 21:20:06 +00:00
|
|
|
SYSCTL_NODE(_net, OID_AUTO, route, CTLFLAG_RD, 0, "");
|
|
|
|
SYSCTL_INT(_net_route, OID_AUTO, netisr_maxqlen, CTLFLAG_RW,
|
|
|
|
&rtsintrq.ifq_maxlen, 0, "maximum routing socket dispatch queue length");
|
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
struct walkarg {
|
1995-11-16 19:00:27 +00:00
|
|
|
int w_tmemsize;
|
|
|
|
int w_op, w_arg;
|
|
|
|
caddr_t w_tmem;
|
|
|
|
struct sysctl_req *w_req;
|
1994-05-24 10:09:53 +00:00
|
|
|
};
|
|
|
|
|
Introduce a netisr to deliver kernel-generated routing, avoiding
recursive entering of the socket code from the routing code:
- Modify rt_dispatch() to bundle up the sockaddr family, if any,
associated with a pending mbuf to dispatch to routing sockets, in
an m_tag on the mbuf.
- Allocate NETISR_ROUTE for use by routing sockets.
- Introduce rtsintrq, an ifqueue to be used by the netisr, and
introduce rts_input(), a function to unbundle the tagged sockaddr
and inject the mbuf and address into raw_input(), which previously
occurred in rt_dispatch().
- Introduce rts_init() to initialize rtsintrq, its mutex, and
register the netisr. Perform this at the same point in system
initialization as setup of the domains.
This change introduces asynchrony between the generation of a
pending routing socket message and delivery to sockets for use
by userspace. It avoids socket->routing->rtsock->socket use and
helps to avoid lock order reversals between the routing code and
socket code (in particular, raw socket control blocks), as route
locks are held over calls to rt_dispatch().
Reviewed by: "George V.Neville-Neil" <gnn@neville-neil.com>
Conceptual head nod by: sam
2004-06-09 02:48:23 +00:00
|
|
|
static void rts_input(struct mbuf *m);
|
2004-04-18 00:56:44 +00:00
|
|
|
static struct mbuf *rt_msg1(int type, struct rt_addrinfo *rtinfo);
|
|
|
|
static int rt_msg2(int type, struct rt_addrinfo *rtinfo,
|
|
|
|
caddr_t cp, struct walkarg *w);
|
|
|
|
static int rt_xaddrs(caddr_t cp, caddr_t cplim,
|
|
|
|
struct rt_addrinfo *rtinfo);
|
2002-03-19 21:54:18 +00:00
|
|
|
static int sysctl_dumpentry(struct radix_node *rn, void *vw);
|
|
|
|
static int sysctl_iflist(int af, struct walkarg *w);
|
2003-11-14 18:48:15 +00:00
|
|
|
static int sysctl_ifmalist(int af, struct walkarg *w);
|
2004-04-18 00:56:44 +00:00
|
|
|
static int route_output(struct mbuf *m, struct socket *so);
|
|
|
|
static void rt_setmetrics(u_long which, const struct rt_metrics *in,
|
|
|
|
struct rt_metrics_lite *out);
|
|
|
|
static void rt_getmetrics(const struct rt_metrics_lite *in,
|
|
|
|
struct rt_metrics *out);
|
|
|
|
static void rt_dispatch(struct mbuf *, const struct sockaddr *);
|
1994-05-24 10:09:53 +00:00
|
|
|
|
Introduce a netisr to deliver kernel-generated routing, avoiding
recursive entering of the socket code from the routing code:
- Modify rt_dispatch() to bundle up the sockaddr family, if any,
associated with a pending mbuf to dispatch to routing sockets, in
an m_tag on the mbuf.
- Allocate NETISR_ROUTE for use by routing sockets.
- Introduce rtsintrq, an ifqueue to be used by the netisr, and
introduce rts_input(), a function to unbundle the tagged sockaddr
and inject the mbuf and address into raw_input(), which previously
occurred in rt_dispatch().
- Introduce rts_init() to initialize rtsintrq, its mutex, and
register the netisr. Perform this at the same point in system
initialization as setup of the domains.
This change introduces asynchrony between the generation of a
pending routing socket message and delivery to sockets for use
by userspace. It avoids socket->routing->rtsock->socket use and
helps to avoid lock order reversals between the routing code and
socket code (in particular, raw socket control blocks), as route
locks are held over calls to rt_dispatch().
Reviewed by: "George V.Neville-Neil" <gnn@neville-neil.com>
Conceptual head nod by: sam
2004-06-09 02:48:23 +00:00
|
|
|
static void
|
|
|
|
rts_init(void)
|
|
|
|
{
|
2004-08-21 21:45:40 +00:00
|
|
|
int tmp;
|
Introduce a netisr to deliver kernel-generated routing, avoiding
recursive entering of the socket code from the routing code:
- Modify rt_dispatch() to bundle up the sockaddr family, if any,
associated with a pending mbuf to dispatch to routing sockets, in
an m_tag on the mbuf.
- Allocate NETISR_ROUTE for use by routing sockets.
- Introduce rtsintrq, an ifqueue to be used by the netisr, and
introduce rts_input(), a function to unbundle the tagged sockaddr
and inject the mbuf and address into raw_input(), which previously
occurred in rt_dispatch().
- Introduce rts_init() to initialize rtsintrq, its mutex, and
register the netisr. Perform this at the same point in system
initialization as setup of the domains.
This change introduces asynchrony between the generation of a
pending routing socket message and delivery to sockets for use
by userspace. It avoids socket->routing->rtsock->socket use and
helps to avoid lock order reversals between the routing code and
socket code (in particular, raw socket control blocks), as route
locks are held over calls to rt_dispatch().
Reviewed by: "George V.Neville-Neil" <gnn@neville-neil.com>
Conceptual head nod by: sam
2004-06-09 02:48:23 +00:00
|
|
|
|
2004-08-21 21:20:06 +00:00
|
|
|
rtsintrq.ifq_maxlen = 256;
|
2004-08-21 21:45:40 +00:00
|
|
|
if (TUNABLE_INT_FETCH("net.route.netisr_maxqlen", &tmp))
|
|
|
|
rtsintrq.ifq_maxlen = tmp;
|
Introduce a netisr to deliver kernel-generated routing, avoiding
recursive entering of the socket code from the routing code:
- Modify rt_dispatch() to bundle up the sockaddr family, if any,
associated with a pending mbuf to dispatch to routing sockets, in
an m_tag on the mbuf.
- Allocate NETISR_ROUTE for use by routing sockets.
- Introduce rtsintrq, an ifqueue to be used by the netisr, and
introduce rts_input(), a function to unbundle the tagged sockaddr
and inject the mbuf and address into raw_input(), which previously
occurred in rt_dispatch().
- Introduce rts_init() to initialize rtsintrq, its mutex, and
register the netisr. Perform this at the same point in system
initialization as setup of the domains.
This change introduces asynchrony between the generation of a
pending routing socket message and delivery to sockets for use
by userspace. It avoids socket->routing->rtsock->socket use and
helps to avoid lock order reversals between the routing code and
socket code (in particular, raw socket control blocks), as route
locks are held over calls to rt_dispatch().
Reviewed by: "George V.Neville-Neil" <gnn@neville-neil.com>
Conceptual head nod by: sam
2004-06-09 02:48:23 +00:00
|
|
|
mtx_init(&rtsintrq.ifq_mtx, "rts_inq", NULL, MTX_DEF);
|
2008-07-04 00:21:38 +00:00
|
|
|
netisr_register(NETISR_ROUTE, rts_input, &rtsintrq, 0);
|
Introduce a netisr to deliver kernel-generated routing, avoiding
recursive entering of the socket code from the routing code:
- Modify rt_dispatch() to bundle up the sockaddr family, if any,
associated with a pending mbuf to dispatch to routing sockets, in
an m_tag on the mbuf.
- Allocate NETISR_ROUTE for use by routing sockets.
- Introduce rtsintrq, an ifqueue to be used by the netisr, and
introduce rts_input(), a function to unbundle the tagged sockaddr
and inject the mbuf and address into raw_input(), which previously
occurred in rt_dispatch().
- Introduce rts_init() to initialize rtsintrq, its mutex, and
register the netisr. Perform this at the same point in system
initialization as setup of the domains.
This change introduces asynchrony between the generation of a
pending routing socket message and delivery to sockets for use
by userspace. It avoids socket->routing->rtsock->socket use and
helps to avoid lock order reversals between the routing code and
socket code (in particular, raw socket control blocks), as route
locks are held over calls to rt_dispatch().
Reviewed by: "George V.Neville-Neil" <gnn@neville-neil.com>
Conceptual head nod by: sam
2004-06-09 02:48:23 +00:00
|
|
|
}
|
2008-03-16 10:58:09 +00:00
|
|
|
SYSINIT(rtsock, SI_SUB_PROTO_DOMAIN, SI_ORDER_THIRD, rts_init, 0);
|
Introduce a netisr to deliver kernel-generated routing, avoiding
recursive entering of the socket code from the routing code:
- Modify rt_dispatch() to bundle up the sockaddr family, if any,
associated with a pending mbuf to dispatch to routing sockets, in
an m_tag on the mbuf.
- Allocate NETISR_ROUTE for use by routing sockets.
- Introduce rtsintrq, an ifqueue to be used by the netisr, and
introduce rts_input(), a function to unbundle the tagged sockaddr
and inject the mbuf and address into raw_input(), which previously
occurred in rt_dispatch().
- Introduce rts_init() to initialize rtsintrq, its mutex, and
register the netisr. Perform this at the same point in system
initialization as setup of the domains.
This change introduces asynchrony between the generation of a
pending routing socket message and delivery to sockets for use
by userspace. It avoids socket->routing->rtsock->socket use and
helps to avoid lock order reversals between the routing code and
socket code (in particular, raw socket control blocks), as route
locks are held over calls to rt_dispatch().
Reviewed by: "George V.Neville-Neil" <gnn@neville-neil.com>
Conceptual head nod by: sam
2004-06-09 02:48:23 +00:00
|
|
|
|
|
|
|
static void
|
|
|
|
rts_input(struct mbuf *m)
|
|
|
|
{
|
|
|
|
struct sockproto route_proto;
|
|
|
|
unsigned short *family;
|
|
|
|
struct m_tag *tag;
|
|
|
|
|
|
|
|
route_proto.sp_family = PF_ROUTE;
|
|
|
|
tag = m_tag_find(m, PACKET_TAG_RTSOCKFAM, NULL);
|
|
|
|
if (tag != NULL) {
|
|
|
|
family = (unsigned short *)(tag + 1);
|
|
|
|
route_proto.sp_protocol = *family;
|
|
|
|
m_tag_delete(m, tag);
|
|
|
|
} else
|
|
|
|
route_proto.sp_protocol = 0;
|
|
|
|
|
Remove unused support for local and foreign addresses in generic raw
socket support. These utility routines are used only for routing and
pfkey sockets, neither of which have a notion of address, so were
required to mock up fake socket addresses to avoid connection
requirements for applications that did not specify their own fake
addresses (most of them).
Quite a bit of the removed code is #ifdef notdef, since raw sockets
don't support bind() or connect() in practice. Removing this
simplifies the raw socket implementation, and removes two (commented
out) uses of dtom(9).
Fake addresses passed to sendto(2) by applications are ignored for
compatibility reasons, but this is now done in a more consistent way
(and with a comment). Possibly, EINVAL could be returned here in
the future if it is determined that no applications depend on the
semantic inconsistency of specifying a destination address for a
protocol without address support, but this will require some amount
of careful surveying.
NB: This does not affect netinet, netinet6, or other wire protocol
raw sockets, which provide their own independent infrastructure with
control block address support specific to the protocol.
MFC after: 3 weeks
Reviewed by: bz
2008-07-09 15:48:16 +00:00
|
|
|
raw_input(m, &route_proto, &route_src);
|
Introduce a netisr to deliver kernel-generated routing, avoiding
recursive entering of the socket code from the routing code:
- Modify rt_dispatch() to bundle up the sockaddr family, if any,
associated with a pending mbuf to dispatch to routing sockets, in
an m_tag on the mbuf.
- Allocate NETISR_ROUTE for use by routing sockets.
- Introduce rtsintrq, an ifqueue to be used by the netisr, and
introduce rts_input(), a function to unbundle the tagged sockaddr
and inject the mbuf and address into raw_input(), which previously
occurred in rt_dispatch().
- Introduce rts_init() to initialize rtsintrq, its mutex, and
register the netisr. Perform this at the same point in system
initialization as setup of the domains.
This change introduces asynchrony between the generation of a
pending routing socket message and delivery to sockets for use
by userspace. It avoids socket->routing->rtsock->socket use and
helps to avoid lock order reversals between the routing code and
socket code (in particular, raw socket control blocks), as route
locks are held over calls to rt_dispatch().
Reviewed by: "George V.Neville-Neil" <gnn@neville-neil.com>
Conceptual head nod by: sam
2004-06-09 02:48:23 +00:00
|
|
|
}
|
|
|
|
|
1997-04-27 20:01:29 +00:00
|
|
|
/*
|
|
|
|
* It really doesn't make any sense at all for this code to share much
|
|
|
|
* with raw_usrreq.c, since its functionality is so restricted. XXX
|
|
|
|
*/
|
2006-04-01 15:15:05 +00:00
|
|
|
static void
|
1997-04-27 20:01:29 +00:00
|
|
|
rts_abort(struct socket *so)
|
|
|
|
{
|
2005-08-25 13:30:04 +00:00
|
|
|
|
2006-04-01 15:15:05 +00:00
|
|
|
raw_usrreqs.pru_abort(so);
|
1997-04-27 20:01:29 +00:00
|
|
|
}
|
|
|
|
|
2006-07-21 17:11:15 +00:00
|
|
|
static void
|
|
|
|
rts_close(struct socket *so)
|
|
|
|
{
|
|
|
|
|
|
|
|
raw_usrreqs.pru_close(so);
|
|
|
|
}
|
|
|
|
|
1997-04-27 20:01:29 +00:00
|
|
|
/* pru_accept is EOPNOTSUPP */
|
|
|
|
|
1995-11-16 19:00:27 +00:00
|
|
|
static int
|
2001-09-12 08:38:13 +00:00
|
|
|
rts_attach(struct socket *so, int proto, struct thread *td)
|
1994-05-24 10:09:53 +00:00
|
|
|
{
|
1997-04-27 20:01:29 +00:00
|
|
|
struct rawcb *rp;
|
|
|
|
int s, error;
|
|
|
|
|
Chance protocol switch method pru_detach() so that it returns void
rather than an error. Detaches do not "fail", they other occur or
the protocol flags SS_PROTOREF to take ownership of the socket.
soclose() no longer looks at so_pcb to see if it's NULL, relying
entirely on the protocol to decide whether it's time to free the
socket or not using SS_PROTOREF. so_pcb is now entirely owned and
managed by the protocol code. Likewise, no longer test so_pcb in
other socket functions, such as soreceive(), which have no business
digging into protocol internals.
Protocol detach routines no longer try to free the socket on detach,
this is performed in the socket code if the protocol permits it.
In rts_detach(), no longer test for rp != NULL in detach, and
likewise in other protocols that don't permit a NULL so_pcb, reduce
the incidence of testing for it during detach.
netinet and netinet6 are not fully updated to this change, which
will be in an upcoming commit. In their current state they may leak
memory or panic.
MFC after: 3 months
2006-04-01 15:42:02 +00:00
|
|
|
KASSERT(so->so_pcb == NULL, ("rts_attach: so_pcb != NULL"));
|
|
|
|
|
2000-12-08 21:51:06 +00:00
|
|
|
/* XXX */
|
2008-10-23 15:53:51 +00:00
|
|
|
rp = malloc(sizeof *rp, M_PCB, M_WAITOK | M_ZERO);
|
2004-04-18 00:56:44 +00:00
|
|
|
if (rp == NULL)
|
1997-04-27 20:01:29 +00:00
|
|
|
return ENOBUFS;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* The splnet() is necessary to block protocols from sending
|
|
|
|
* error notifications (like RTM_REDIRECT or RTM_LOSING) while
|
|
|
|
* this PCB is extant but incompletely initialized.
|
|
|
|
* Probably we should try to do more of this work beforehand and
|
|
|
|
* eliminate the spl.
|
|
|
|
*/
|
|
|
|
s = splnet();
|
|
|
|
so->so_pcb = (caddr_t)rp;
|
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
|
|
|
so->so_fibnum = td->td_proc->p_fibnum;
|
2001-08-31 12:31:09 +00:00
|
|
|
error = raw_attach(so, proto);
|
1997-04-27 20:01:29 +00:00
|
|
|
rp = sotorawcb(so);
|
|
|
|
if (error) {
|
|
|
|
splx(s);
|
2001-08-02 19:56:29 +00:00
|
|
|
so->so_pcb = NULL;
|
1997-04-27 20:01:29 +00:00
|
|
|
free(rp, M_PCB);
|
|
|
|
return error;
|
|
|
|
}
|
2003-10-03 18:27:02 +00:00
|
|
|
RTSOCK_LOCK();
|
1997-04-27 20:01:29 +00:00
|
|
|
switch(rp->rcb_proto.sp_protocol) {
|
|
|
|
case AF_INET:
|
|
|
|
route_cb.ip_count++;
|
|
|
|
break;
|
2000-01-28 20:10:51 +00:00
|
|
|
case AF_INET6:
|
|
|
|
route_cb.ip6_count++;
|
|
|
|
break;
|
1997-04-27 20:01:29 +00:00
|
|
|
case AF_IPX:
|
|
|
|
route_cb.ipx_count++;
|
|
|
|
break;
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
1997-04-27 20:01:29 +00:00
|
|
|
route_cb.any_count++;
|
2003-10-03 18:27:02 +00:00
|
|
|
RTSOCK_UNLOCK();
|
2002-06-18 07:42:02 +00:00
|
|
|
soisconnected(so);
|
1997-04-27 20:01:29 +00:00
|
|
|
so->so_options |= SO_USELOOPBACK;
|
|
|
|
splx(s);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
2001-09-12 08:38:13 +00:00
|
|
|
rts_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
|
1997-04-27 20:01:29 +00:00
|
|
|
{
|
2005-08-25 13:30:04 +00:00
|
|
|
|
|
|
|
return (raw_usrreqs.pru_bind(so, nam, td)); /* xxx just EINVAL */
|
1997-04-27 20:01:29 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
2001-09-12 08:38:13 +00:00
|
|
|
rts_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
|
1997-04-27 20:01:29 +00:00
|
|
|
{
|
2005-08-25 13:30:04 +00:00
|
|
|
|
|
|
|
return (raw_usrreqs.pru_connect(so, nam, td)); /* XXX just EINVAL */
|
1997-04-27 20:01:29 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/* pru_connect2 is EOPNOTSUPP */
|
|
|
|
/* pru_control is EOPNOTSUPP */
|
|
|
|
|
Chance protocol switch method pru_detach() so that it returns void
rather than an error. Detaches do not "fail", they other occur or
the protocol flags SS_PROTOREF to take ownership of the socket.
soclose() no longer looks at so_pcb to see if it's NULL, relying
entirely on the protocol to decide whether it's time to free the
socket or not using SS_PROTOREF. so_pcb is now entirely owned and
managed by the protocol code. Likewise, no longer test so_pcb in
other socket functions, such as soreceive(), which have no business
digging into protocol internals.
Protocol detach routines no longer try to free the socket on detach,
this is performed in the socket code if the protocol permits it.
In rts_detach(), no longer test for rp != NULL in detach, and
likewise in other protocols that don't permit a NULL so_pcb, reduce
the incidence of testing for it during detach.
netinet and netinet6 are not fully updated to this change, which
will be in an upcoming commit. In their current state they may leak
memory or panic.
MFC after: 3 months
2006-04-01 15:42:02 +00:00
|
|
|
static void
|
1997-04-27 20:01:29 +00:00
|
|
|
rts_detach(struct socket *so)
|
|
|
|
{
|
|
|
|
struct rawcb *rp = sotorawcb(so);
|
|
|
|
|
Chance protocol switch method pru_detach() so that it returns void
rather than an error. Detaches do not "fail", they other occur or
the protocol flags SS_PROTOREF to take ownership of the socket.
soclose() no longer looks at so_pcb to see if it's NULL, relying
entirely on the protocol to decide whether it's time to free the
socket or not using SS_PROTOREF. so_pcb is now entirely owned and
managed by the protocol code. Likewise, no longer test so_pcb in
other socket functions, such as soreceive(), which have no business
digging into protocol internals.
Protocol detach routines no longer try to free the socket on detach,
this is performed in the socket code if the protocol permits it.
In rts_detach(), no longer test for rp != NULL in detach, and
likewise in other protocols that don't permit a NULL so_pcb, reduce
the incidence of testing for it during detach.
netinet and netinet6 are not fully updated to this change, which
will be in an upcoming commit. In their current state they may leak
memory or panic.
MFC after: 3 months
2006-04-01 15:42:02 +00:00
|
|
|
KASSERT(rp != NULL, ("rts_detach: rp == NULL"));
|
|
|
|
|
|
|
|
RTSOCK_LOCK();
|
|
|
|
switch(rp->rcb_proto.sp_protocol) {
|
|
|
|
case AF_INET:
|
|
|
|
route_cb.ip_count--;
|
|
|
|
break;
|
|
|
|
case AF_INET6:
|
|
|
|
route_cb.ip6_count--;
|
|
|
|
break;
|
|
|
|
case AF_IPX:
|
|
|
|
route_cb.ipx_count--;
|
|
|
|
break;
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
Chance protocol switch method pru_detach() so that it returns void
rather than an error. Detaches do not "fail", they other occur or
the protocol flags SS_PROTOREF to take ownership of the socket.
soclose() no longer looks at so_pcb to see if it's NULL, relying
entirely on the protocol to decide whether it's time to free the
socket or not using SS_PROTOREF. so_pcb is now entirely owned and
managed by the protocol code. Likewise, no longer test so_pcb in
other socket functions, such as soreceive(), which have no business
digging into protocol internals.
Protocol detach routines no longer try to free the socket on detach,
this is performed in the socket code if the protocol permits it.
In rts_detach(), no longer test for rp != NULL in detach, and
likewise in other protocols that don't permit a NULL so_pcb, reduce
the incidence of testing for it during detach.
netinet and netinet6 are not fully updated to this change, which
will be in an upcoming commit. In their current state they may leak
memory or panic.
MFC after: 3 months
2006-04-01 15:42:02 +00:00
|
|
|
route_cb.any_count--;
|
|
|
|
RTSOCK_UNLOCK();
|
|
|
|
raw_usrreqs.pru_detach(so);
|
1997-04-27 20:01:29 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
|
|
|
rts_disconnect(struct socket *so)
|
|
|
|
{
|
2005-08-25 13:30:04 +00:00
|
|
|
|
|
|
|
return (raw_usrreqs.pru_disconnect(so));
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
|
|
|
|
1997-04-27 20:01:29 +00:00
|
|
|
/* pru_listen is EOPNOTSUPP */
|
|
|
|
|
|
|
|
static int
|
1997-08-16 19:16:27 +00:00
|
|
|
rts_peeraddr(struct socket *so, struct sockaddr **nam)
|
1997-04-27 20:01:29 +00:00
|
|
|
{
|
2005-08-25 13:30:04 +00:00
|
|
|
|
|
|
|
return (raw_usrreqs.pru_peeraddr(so, nam));
|
1997-04-27 20:01:29 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/* pru_rcvd is EOPNOTSUPP */
|
|
|
|
/* pru_rcvoob is EOPNOTSUPP */
|
|
|
|
|
|
|
|
static int
|
1997-08-16 19:16:27 +00:00
|
|
|
rts_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
|
2001-09-12 08:38:13 +00:00
|
|
|
struct mbuf *control, struct thread *td)
|
1997-04-27 20:01:29 +00:00
|
|
|
{
|
2005-08-25 13:30:04 +00:00
|
|
|
|
|
|
|
return (raw_usrreqs.pru_send(so, flags, m, nam, control, td));
|
1997-04-27 20:01:29 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/* pru_sense is null */
|
|
|
|
|
|
|
|
static int
|
|
|
|
rts_shutdown(struct socket *so)
|
|
|
|
{
|
2005-08-25 13:30:04 +00:00
|
|
|
|
|
|
|
return (raw_usrreqs.pru_shutdown(so));
|
1997-04-27 20:01:29 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
1997-08-16 19:16:27 +00:00
|
|
|
rts_sockaddr(struct socket *so, struct sockaddr **nam)
|
1997-04-27 20:01:29 +00:00
|
|
|
{
|
2005-08-25 13:30:04 +00:00
|
|
|
|
|
|
|
return (raw_usrreqs.pru_sockaddr(so, nam));
|
1997-04-27 20:01:29 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static struct pr_usrreqs route_usrreqs = {
|
2004-11-08 14:44:54 +00:00
|
|
|
.pru_abort = rts_abort,
|
|
|
|
.pru_attach = rts_attach,
|
|
|
|
.pru_bind = rts_bind,
|
|
|
|
.pru_connect = rts_connect,
|
|
|
|
.pru_detach = rts_detach,
|
|
|
|
.pru_disconnect = rts_disconnect,
|
|
|
|
.pru_peeraddr = rts_peeraddr,
|
|
|
|
.pru_send = rts_send,
|
|
|
|
.pru_shutdown = rts_shutdown,
|
|
|
|
.pru_sockaddr = rts_sockaddr,
|
2006-07-21 17:11:15 +00:00
|
|
|
.pru_close = rts_close,
|
1997-04-27 20:01:29 +00:00
|
|
|
};
|
|
|
|
|
MFp4:
Bring in updated jail support from bz_jail branch.
This enhances the current jail implementation to permit multiple
addresses per jail. In addtion to IPv4, IPv6 is supported as well.
Due to updated checks it is even possible to have jails without
an IP address at all, which basically gives one a chroot with
restricted process view, no networking,..
SCTP support was updated and supports IPv6 in jails as well.
Cpuset support permits jails to be bound to specific processor
sets after creation.
Jails can have an unrestricted (no duplicate protection, etc.) name
in addition to the hostname. The jail name cannot be changed from
within a jail and is considered to be used for management purposes
or as audit-token in the future.
DDB 'show jails' command was added to aid debugging.
Proper compat support permits 32bit jail binaries to be used on 64bit
systems to manage jails. Also backward compatibility was preserved where
possible: for jail v1 syscalls, as well as with user space management
utilities.
Both jail as well as prison version were updated for the new features.
A gap was intentionally left as the intermediate versions had been
used by various patches floating around the last years.
Bump __FreeBSD_version for the afore mentioned and in kernel changes.
Special thanks to:
- Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches
and Olivier Houchard (cognet) for initial single-IPv6 patches.
- Jeff Roberson (jeff) and Randall Stewart (rrs) for their
help, ideas and review on cpuset and SCTP support.
- Robert Watson (rwatson) for lots and lots of help, discussions,
suggestions and review of most of the patch at various stages.
- John Baldwin (jhb) for his help.
- Simon L. Nielsen (simon) as early adopter testing changes
on cluster machines as well as all the testers and people
who provided feedback the last months on freebsd-jail and
other channels.
- My employer, CK Software GmbH, for the support so I could work on this.
Reviewed by: (see above)
MFC after: 3 months (this is just so that I get the mail)
X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
|
|
|
#ifndef _SOCKADDR_UNION_DEFINED
|
|
|
|
#define _SOCKADDR_UNION_DEFINED
|
|
|
|
/*
|
|
|
|
* The union of all possible address formats we handle.
|
|
|
|
*/
|
|
|
|
union sockaddr_union {
|
|
|
|
struct sockaddr sa;
|
|
|
|
struct sockaddr_in sin;
|
|
|
|
struct sockaddr_in6 sin6;
|
|
|
|
};
|
|
|
|
#endif /* _SOCKADDR_UNION_DEFINED */
|
|
|
|
|
|
|
|
static int
|
|
|
|
rtm_get_jailed(struct rt_addrinfo *info, struct ifnet *ifp,
|
|
|
|
struct rtentry *rt, union sockaddr_union *saun, struct ucred *cred)
|
|
|
|
{
|
|
|
|
|
|
|
|
switch (info->rti_info[RTAX_DST]->sa_family) {
|
|
|
|
#ifdef INET
|
|
|
|
case AF_INET:
|
|
|
|
{
|
|
|
|
struct in_addr ia;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* 1. Check if the returned address is part of the jail.
|
|
|
|
*/
|
|
|
|
ia = ((struct sockaddr_in *)rt->rt_ifa->ifa_addr)->sin_addr;
|
|
|
|
if (prison_check_ip4(cred, &ia) != 0) {
|
|
|
|
info->rti_info[RTAX_IFA] = rt->rt_ifa->ifa_addr;
|
|
|
|
|
|
|
|
} else {
|
|
|
|
struct ifaddr *ifa;
|
|
|
|
int found;
|
|
|
|
|
|
|
|
found = 0;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* 2. Try to find an address on the given outgoing
|
|
|
|
* interface that belongs to the jail.
|
|
|
|
*/
|
|
|
|
TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
|
|
|
|
struct sockaddr *sa;
|
|
|
|
sa = ifa->ifa_addr;
|
|
|
|
if (sa->sa_family != AF_INET)
|
|
|
|
continue;
|
|
|
|
ia = ((struct sockaddr_in *)sa)->sin_addr;
|
|
|
|
if (prison_check_ip4(cred, &ia) != 0) {
|
|
|
|
found = 1;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
if (!found) {
|
|
|
|
/*
|
|
|
|
* 3. As a last resort return the 'default'
|
|
|
|
* jail address.
|
|
|
|
*/
|
|
|
|
if (prison_getip4(cred, &ia) != 0)
|
|
|
|
return (ESRCH);
|
|
|
|
}
|
|
|
|
bzero(&saun->sin, sizeof(struct sockaddr_in));
|
|
|
|
saun->sin.sin_len = sizeof(struct sockaddr_in);
|
|
|
|
saun->sin.sin_family = AF_INET;
|
|
|
|
saun->sin.sin_addr.s_addr = ia.s_addr;
|
|
|
|
info->rti_info[RTAX_IFA] =
|
|
|
|
(struct sockaddr *)&saun->sin;
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
#ifdef INET6
|
|
|
|
case AF_INET6:
|
|
|
|
{
|
|
|
|
struct in6_addr ia6;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* 1. Check if the returned address is part of the jail.
|
|
|
|
*/
|
|
|
|
bcopy(&((struct sockaddr_in6 *)rt->rt_ifa->ifa_addr)->sin6_addr,
|
|
|
|
&ia6, sizeof(struct in6_addr));
|
|
|
|
if (prison_check_ip6(cred, &ia6) != 0) {
|
|
|
|
info->rti_info[RTAX_IFA] = rt->rt_ifa->ifa_addr;
|
|
|
|
} else {
|
|
|
|
struct ifaddr *ifa;
|
|
|
|
int found;
|
|
|
|
|
|
|
|
found = 0;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* 2. Try to find an address on the given outgoing
|
|
|
|
* interface that belongs to the jail.
|
|
|
|
*/
|
|
|
|
TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
|
|
|
|
struct sockaddr *sa;
|
|
|
|
sa = ifa->ifa_addr;
|
|
|
|
if (sa->sa_family != AF_INET6)
|
|
|
|
continue;
|
|
|
|
bcopy(&((struct sockaddr_in6 *)sa)->sin6_addr,
|
|
|
|
&ia6, sizeof(struct in6_addr));
|
|
|
|
if (prison_check_ip6(cred, &ia6) != 0) {
|
|
|
|
found = 1;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
if (!found) {
|
|
|
|
/*
|
|
|
|
* 3. As a last resort return the 'default'
|
|
|
|
* jail address.
|
|
|
|
*/
|
|
|
|
if (prison_getip6(cred, &ia6) != 0)
|
|
|
|
return (ESRCH);
|
|
|
|
}
|
|
|
|
bzero(&saun->sin6, sizeof(struct sockaddr_in6));
|
|
|
|
saun->sin6.sin6_len = sizeof(struct sockaddr_in6);
|
|
|
|
saun->sin6.sin6_family = AF_INET6;
|
|
|
|
bcopy(&ia6, &saun->sin6.sin6_addr,
|
|
|
|
sizeof(struct in6_addr));
|
|
|
|
if (sa6_recoverscope(&saun->sin6) != 0)
|
|
|
|
return (ESRCH);
|
|
|
|
info->rti_info[RTAX_IFA] =
|
|
|
|
(struct sockaddr *)&saun->sin6;
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
default:
|
|
|
|
return (ESRCH);
|
|
|
|
}
|
|
|
|
return (0);
|
|
|
|
}
|
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
/*ARGSUSED*/
|
1995-11-16 19:00:27 +00:00
|
|
|
static int
|
2004-04-18 00:56:44 +00:00
|
|
|
route_output(struct mbuf *m, struct socket *so)
|
1994-05-24 10:09:53 +00:00
|
|
|
{
|
2003-10-03 18:15:54 +00:00
|
|
|
#define sa_equal(a1, a2) (bcmp((a1), (a2), (a1)->sa_len) == 0)
|
Step 1.5 of importing the network stack virtualization infrastructure
from the vimage project, as per plan established at devsummit 08/08:
http://wiki.freebsd.org/Image/Notes200808DevSummit
Introduce INIT_VNET_*() initializer macros, VNET_FOREACH() iterator
macros, and CURVNET_SET() context setting macros, all currently
resolving to NOPs.
Prepare for virtualization of selected SYSCTL objects by introducing a
family of SYSCTL_V_*() macros, currently resolving to their global
counterparts, i.e. SYSCTL_V_INT() == SYSCTL_INT().
Move selected #defines from sys/sys/vimage.h to newly introduced header
files specific to virtualized subsystems (sys/net/vnet.h,
sys/netinet/vinet.h etc.).
All the changes are verified to have zero functional impact at this
point in time by doing MD5 comparision between pre- and post-change
object files(*).
(*) netipsec/keysock.c did not validate depending on compile time options.
Implemented by: julian, bz, brooks, zec
Reviewed by: julian, bz, brooks, kris, rwatson, ...
Approved by: julian (mentor)
Obtained from: //depot/projects/vimage-commit2/...
X-MFC after: never
Sponsored by: NLnet Foundation, The FreeBSD Foundation
2008-10-02 15:37:58 +00:00
|
|
|
INIT_VNET_NET(so->so_vnet);
|
2004-04-18 00:56:44 +00:00
|
|
|
struct rt_msghdr *rtm = NULL;
|
|
|
|
struct rtentry *rt = NULL;
|
1995-05-10 16:48:52 +00:00
|
|
|
struct radix_node_head *rnh;
|
1994-05-24 10:09:53 +00:00
|
|
|
struct rt_addrinfo info;
|
|
|
|
int len, error = 0;
|
2004-04-18 00:56:44 +00:00
|
|
|
struct ifnet *ifp = NULL;
|
MFp4:
Bring in updated jail support from bz_jail branch.
This enhances the current jail implementation to permit multiple
addresses per jail. In addtion to IPv4, IPv6 is supported as well.
Due to updated checks it is even possible to have jails without
an IP address at all, which basically gives one a chroot with
restricted process view, no networking,..
SCTP support was updated and supports IPv6 in jails as well.
Cpuset support permits jails to be bound to specific processor
sets after creation.
Jails can have an unrestricted (no duplicate protection, etc.) name
in addition to the hostname. The jail name cannot be changed from
within a jail and is considered to be used for management purposes
or as audit-token in the future.
DDB 'show jails' command was added to aid debugging.
Proper compat support permits 32bit jail binaries to be used on 64bit
systems to manage jails. Also backward compatibility was preserved where
possible: for jail v1 syscalls, as well as with user space management
utilities.
Both jail as well as prison version were updated for the new features.
A gap was intentionally left as the intermediate versions had been
used by various patches floating around the last years.
Bump __FreeBSD_version for the afore mentioned and in kernel changes.
Special thanks to:
- Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches
and Olivier Houchard (cognet) for initial single-IPv6 patches.
- Jeff Roberson (jeff) and Randall Stewart (rrs) for their
help, ideas and review on cpuset and SCTP support.
- Robert Watson (rwatson) for lots and lots of help, discussions,
suggestions and review of most of the patch at various stages.
- John Baldwin (jhb) for his help.
- Simon L. Nielsen (simon) as early adopter testing changes
on cluster machines as well as all the testers and people
who provided feedback the last months on freebsd-jail and
other channels.
- My employer, CK Software GmbH, for the support so I could work on this.
Reviewed by: (see above)
MFC after: 3 months (this is just so that I get the mail)
X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
|
|
|
union sockaddr_union saun;
|
1994-05-24 10:09:53 +00:00
|
|
|
|
|
|
|
#define senderr(e) { error = e; goto flush;}
|
2004-04-18 00:56:44 +00:00
|
|
|
if (m == NULL || ((m->m_len < sizeof(long)) &&
|
|
|
|
(m = m_pullup(m, sizeof(long))) == NULL))
|
1994-05-24 10:09:53 +00:00
|
|
|
return (ENOBUFS);
|
|
|
|
if ((m->m_flags & M_PKTHDR) == 0)
|
|
|
|
panic("route_output");
|
|
|
|
len = m->m_pkthdr.len;
|
|
|
|
if (len < sizeof(*rtm) ||
|
|
|
|
len != mtod(m, struct rt_msghdr *)->rtm_msglen) {
|
2004-04-18 00:56:44 +00:00
|
|
|
info.rti_info[RTAX_DST] = NULL;
|
1994-05-24 10:09:53 +00:00
|
|
|
senderr(EINVAL);
|
|
|
|
}
|
|
|
|
R_Malloc(rtm, struct rt_msghdr *, len);
|
2004-04-18 00:56:44 +00:00
|
|
|
if (rtm == NULL) {
|
|
|
|
info.rti_info[RTAX_DST] = NULL;
|
1994-05-24 10:09:53 +00:00
|
|
|
senderr(ENOBUFS);
|
|
|
|
}
|
|
|
|
m_copydata(m, 0, len, (caddr_t)rtm);
|
|
|
|
if (rtm->rtm_version != RTM_VERSION) {
|
2004-04-18 00:56:44 +00:00
|
|
|
info.rti_info[RTAX_DST] = NULL;
|
1994-05-24 10:09:53 +00:00
|
|
|
senderr(EPROTONOSUPPORT);
|
|
|
|
}
|
|
|
|
rtm->rtm_pid = curproc->p_pid;
|
2001-10-17 18:07:05 +00:00
|
|
|
bzero(&info, sizeof(info));
|
1994-05-24 10:09:53 +00:00
|
|
|
info.rti_addrs = rtm->rtm_addrs;
|
1997-07-18 11:44:24 +00:00
|
|
|
if (rt_xaddrs((caddr_t)(rtm + 1), len + (caddr_t)rtm, &info)) {
|
2004-04-18 00:56:44 +00:00
|
|
|
info.rti_info[RTAX_DST] = NULL;
|
1997-07-18 11:44:24 +00:00
|
|
|
senderr(EINVAL);
|
|
|
|
}
|
2001-10-17 18:07:05 +00:00
|
|
|
info.rti_flags = rtm->rtm_flags;
|
2004-04-18 00:56:44 +00:00
|
|
|
if (info.rti_info[RTAX_DST] == NULL ||
|
2003-10-03 18:15:54 +00:00
|
|
|
info.rti_info[RTAX_DST]->sa_family >= AF_MAX ||
|
2004-04-18 00:56:44 +00:00
|
|
|
(info.rti_info[RTAX_GATEWAY] != NULL &&
|
2003-10-03 18:15:54 +00:00
|
|
|
info.rti_info[RTAX_GATEWAY]->sa_family >= AF_MAX))
|
1994-05-24 10:09:53 +00:00
|
|
|
senderr(EINVAL);
|
2001-08-31 12:31:09 +00:00
|
|
|
/*
|
|
|
|
* Verify that the caller has the appropriate privilege; RTM_GET
|
|
|
|
* is the only operation the non-superuser is allowed.
|
|
|
|
*/
|
2006-11-06 13:42:10 +00:00
|
|
|
if (rtm->rtm_type != RTM_GET) {
|
|
|
|
error = priv_check(curthread, PRIV_NET_ROUTE);
|
|
|
|
if (error)
|
|
|
|
senderr(error);
|
|
|
|
}
|
2001-08-31 12:31:09 +00:00
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
switch (rtm->rtm_type) {
|
2003-10-03 18:15:54 +00:00
|
|
|
struct rtentry *saved_nrt;
|
1994-05-24 10:09:53 +00:00
|
|
|
|
|
|
|
case RTM_ADD:
|
2004-04-18 00:56:44 +00:00
|
|
|
if (info.rti_info[RTAX_GATEWAY] == NULL)
|
1994-05-24 10:09:53 +00:00
|
|
|
senderr(EINVAL);
|
2004-04-18 00:56:44 +00:00
|
|
|
saved_nrt = NULL;
|
2008-12-26 19:45:24 +00:00
|
|
|
|
This main goals of this project are:
1. separating L2 tables (ARP, NDP) from the L3 routing tables
2. removing as much locking dependencies among these layers as
possible to allow for some parallelism in the search operations
3. simplify the logic in the routing code,
The most notable end result is the obsolescent of the route
cloning (RTF_CLONING) concept, which translated into code reduction
in both IPv4 ARP and IPv6 NDP related modules, and size reduction in
struct rtentry{}. The change in design obsoletes the semantics of
RTF_CLONING, RTF_WASCLONE and RTF_LLINFO routing flags. The userland
applications such as "arp" and "ndp" have been modified to reflect
those changes. The output from "netstat -r" shows only the routing
entries.
Quite a few developers have contributed to this project in the
past: Glebius Smirnoff, Luigi Rizzo, Alessandro Cerri, and
Andre Oppermann. And most recently:
- Kip Macy revised the locking code completely, thus completing
the last piece of the puzzle, Kip has also been conducting
active functional testing
- Sam Leffler has helped me improving/refactoring the code, and
provided valuable reviews
- Julian Elischer setup the perforce tree for me and has helped
me maintaining that branch before the svn conversion
2008-12-15 06:10:57 +00:00
|
|
|
/* support for new ARP code */
|
2008-12-26 19:45:24 +00:00
|
|
|
if (info.rti_info[RTAX_GATEWAY]->sa_family == AF_LINK &&
|
|
|
|
(rtm->rtm_flags & RTF_LLDATA) != 0) {
|
This main goals of this project are:
1. separating L2 tables (ARP, NDP) from the L3 routing tables
2. removing as much locking dependencies among these layers as
possible to allow for some parallelism in the search operations
3. simplify the logic in the routing code,
The most notable end result is the obsolescent of the route
cloning (RTF_CLONING) concept, which translated into code reduction
in both IPv4 ARP and IPv6 NDP related modules, and size reduction in
struct rtentry{}. The change in design obsoletes the semantics of
RTF_CLONING, RTF_WASCLONE and RTF_LLINFO routing flags. The userland
applications such as "arp" and "ndp" have been modified to reflect
those changes. The output from "netstat -r" shows only the routing
entries.
Quite a few developers have contributed to this project in the
past: Glebius Smirnoff, Luigi Rizzo, Alessandro Cerri, and
Andre Oppermann. And most recently:
- Kip Macy revised the locking code completely, thus completing
the last piece of the puzzle, Kip has also been conducting
active functional testing
- Sam Leffler has helped me improving/refactoring the code, and
provided valuable reviews
- Julian Elischer setup the perforce tree for me and has helped
me maintaining that branch before the svn conversion
2008-12-15 06:10:57 +00:00
|
|
|
error = lla_rt_output(rtm, &info);
|
|
|
|
break;
|
|
|
|
}
|
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
|
|
|
error = rtrequest1_fib(RTM_ADD, &info, &saved_nrt,
|
|
|
|
so->so_fibnum);
|
1994-05-24 10:09:53 +00:00
|
|
|
if (error == 0 && saved_nrt) {
|
2003-10-04 03:44:50 +00:00
|
|
|
RT_LOCK(saved_nrt);
|
1994-05-24 10:09:53 +00:00
|
|
|
rt_setmetrics(rtm->rtm_inits,
|
|
|
|
&rtm->rtm_rmx, &saved_nrt->rt_rmx);
|
2006-03-15 19:39:09 +00:00
|
|
|
rtm->rtm_index = saved_nrt->rt_ifp->if_index;
|
2003-11-08 23:36:32 +00:00
|
|
|
RT_REMREF(saved_nrt);
|
2003-10-04 03:44:50 +00:00
|
|
|
RT_UNLOCK(saved_nrt);
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
|
|
|
break;
|
|
|
|
|
|
|
|
case RTM_DELETE:
|
2004-04-18 00:56:44 +00:00
|
|
|
saved_nrt = NULL;
|
This main goals of this project are:
1. separating L2 tables (ARP, NDP) from the L3 routing tables
2. removing as much locking dependencies among these layers as
possible to allow for some parallelism in the search operations
3. simplify the logic in the routing code,
The most notable end result is the obsolescent of the route
cloning (RTF_CLONING) concept, which translated into code reduction
in both IPv4 ARP and IPv6 NDP related modules, and size reduction in
struct rtentry{}. The change in design obsoletes the semantics of
RTF_CLONING, RTF_WASCLONE and RTF_LLINFO routing flags. The userland
applications such as "arp" and "ndp" have been modified to reflect
those changes. The output from "netstat -r" shows only the routing
entries.
Quite a few developers have contributed to this project in the
past: Glebius Smirnoff, Luigi Rizzo, Alessandro Cerri, and
Andre Oppermann. And most recently:
- Kip Macy revised the locking code completely, thus completing
the last piece of the puzzle, Kip has also been conducting
active functional testing
- Sam Leffler has helped me improving/refactoring the code, and
provided valuable reviews
- Julian Elischer setup the perforce tree for me and has helped
me maintaining that branch before the svn conversion
2008-12-15 06:10:57 +00:00
|
|
|
/* support for new ARP code */
|
|
|
|
if (info.rti_info[RTAX_GATEWAY] &&
|
2008-12-26 19:45:24 +00:00
|
|
|
(info.rti_info[RTAX_GATEWAY]->sa_family == AF_LINK) &&
|
|
|
|
(rtm->rtm_flags & RTF_LLDATA) != 0) {
|
This main goals of this project are:
1. separating L2 tables (ARP, NDP) from the L3 routing tables
2. removing as much locking dependencies among these layers as
possible to allow for some parallelism in the search operations
3. simplify the logic in the routing code,
The most notable end result is the obsolescent of the route
cloning (RTF_CLONING) concept, which translated into code reduction
in both IPv4 ARP and IPv6 NDP related modules, and size reduction in
struct rtentry{}. The change in design obsoletes the semantics of
RTF_CLONING, RTF_WASCLONE and RTF_LLINFO routing flags. The userland
applications such as "arp" and "ndp" have been modified to reflect
those changes. The output from "netstat -r" shows only the routing
entries.
Quite a few developers have contributed to this project in the
past: Glebius Smirnoff, Luigi Rizzo, Alessandro Cerri, and
Andre Oppermann. And most recently:
- Kip Macy revised the locking code completely, thus completing
the last piece of the puzzle, Kip has also been conducting
active functional testing
- Sam Leffler has helped me improving/refactoring the code, and
provided valuable reviews
- Julian Elischer setup the perforce tree for me and has helped
me maintaining that branch before the svn conversion
2008-12-15 06:10:57 +00:00
|
|
|
error = lla_rt_output(rtm, &info);
|
|
|
|
break;
|
|
|
|
}
|
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
|
|
|
error = rtrequest1_fib(RTM_DELETE, &info, &saved_nrt,
|
|
|
|
so->so_fibnum);
|
1995-05-10 16:48:52 +00:00
|
|
|
if (error == 0) {
|
2003-10-04 03:44:50 +00:00
|
|
|
RT_LOCK(saved_nrt);
|
2002-12-25 10:21:02 +00:00
|
|
|
rt = saved_nrt;
|
1995-05-10 16:48:52 +00:00
|
|
|
goto report;
|
|
|
|
}
|
1994-05-24 10:09:53 +00:00
|
|
|
break;
|
|
|
|
|
|
|
|
case RTM_GET:
|
|
|
|
case RTM_CHANGE:
|
|
|
|
case RTM_LOCK:
|
Commit step 1 of the vimage project, (network stack)
virtualization work done by Marko Zec (zec@).
This is the first in a series of commits over the course
of the next few weeks.
Mark all uses of global variables to be virtualized
with a V_ prefix.
Use macros to map them back to their global names for
now, so this is a NOP change only.
We hope to have caught at least 85-90% of what is needed
so we do not invalidate a lot of outstanding patches again.
Obtained from: //depot/projects/vimage-commit2/...
Reviewed by: brooks, des, ed, mav, julian,
jamie, kris, rwatson, zec, ...
(various people I forgot, different versions)
md5 (with a bit of help)
Sponsored by: NLnet Foundation, The FreeBSD Foundation
X-MFC after: never
V_Commit_Message_Reviewed_By: more people than the patch
2008-08-17 23:27:27 +00:00
|
|
|
rnh = V_rt_tables[so->so_fibnum][info.rti_info[RTAX_DST]->sa_family];
|
2004-04-18 00:56:44 +00:00
|
|
|
if (rnh == NULL)
|
1995-05-10 16:48:52 +00:00
|
|
|
senderr(EAFNOSUPPORT);
|
2008-12-07 21:15:43 +00:00
|
|
|
RADIX_NODE_HEAD_RLOCK(rnh);
|
2003-10-03 18:15:54 +00:00
|
|
|
rt = (struct rtentry *) rnh->rnh_lookup(info.rti_info[RTAX_DST],
|
|
|
|
info.rti_info[RTAX_NETMASK], rnh);
|
2005-08-11 08:26:31 +00:00
|
|
|
if (rt == NULL) { /* XXX looks bogus */
|
2008-12-07 21:15:43 +00:00
|
|
|
RADIX_NODE_HEAD_RUNLOCK(rnh);
|
1994-05-24 10:09:53 +00:00
|
|
|
senderr(ESRCH);
|
2005-08-11 08:26:31 +00:00
|
|
|
}
|
This patch provides the back end support for equal-cost multi-path
(ECMP) for both IPv4 and IPv6. Previously, multipath route insertion
is disallowed. For example,
route add -net 192.103.54.0/24 10.9.44.1
route add -net 192.103.54.0/24 10.9.44.2
The second route insertion will trigger an error message of
"add net 192.103.54.0/24: gateway 10.2.5.2: route already in table"
Multiple default routes can also be inserted. Here is the netstat
output:
default 10.2.5.1 UGS 0 3074 bge0 =>
default 10.2.5.2 UGS 0 0 bge0
When multipath routes exist, the "route delete" command requires
a specific gateway to be specified or else an error message would
be displayed. For example,
route delete default
would fail and trigger the following error message:
"route: writing to routing socket: No such process"
"delete net default: not in table"
On the other hand,
route delete default 10.2.5.2
would be successful: "delete net default: gateway 10.2.5.2"
One does not have to specify a gateway if there is only a single
route for a particular destination.
I need to perform more testings on address aliases and multiple
interfaces that have the same IP prefixes. This patch as it
stands today is not yet ready for prime time. Therefore, the ECMP
code fragments are fully guarded by the RADIX_MPATH macro.
Include the "options RADIX_MPATH" in the kernel configuration
to enable this feature.
Reviewed by: robert, sam, gnn, julian, kmacy
2008-04-13 05:45:14 +00:00
|
|
|
#ifdef RADIX_MPATH
|
|
|
|
/*
|
|
|
|
* for RTM_CHANGE/LOCK, if we got multipath routes,
|
|
|
|
* we require users to specify a matching RTAX_GATEWAY.
|
|
|
|
*
|
|
|
|
* for RTM_GET, gate is optional even with multipath.
|
|
|
|
* if gate == NULL the first match is returned.
|
|
|
|
* (no need to call rt_mpath_matchgate if gate == NULL)
|
|
|
|
*/
|
|
|
|
if (rn_mpath_capable(rnh) &&
|
|
|
|
(rtm->rtm_type != RTM_GET || info.rti_info[RTAX_GATEWAY])) {
|
|
|
|
rt = rt_mpath_matchgate(rt, info.rti_info[RTAX_GATEWAY]);
|
|
|
|
if (!rt) {
|
2008-12-07 21:15:43 +00:00
|
|
|
RADIX_NODE_HEAD_RUNLOCK(rnh);
|
This patch provides the back end support for equal-cost multi-path
(ECMP) for both IPv4 and IPv6. Previously, multipath route insertion
is disallowed. For example,
route add -net 192.103.54.0/24 10.9.44.1
route add -net 192.103.54.0/24 10.9.44.2
The second route insertion will trigger an error message of
"add net 192.103.54.0/24: gateway 10.2.5.2: route already in table"
Multiple default routes can also be inserted. Here is the netstat
output:
default 10.2.5.1 UGS 0 3074 bge0 =>
default 10.2.5.2 UGS 0 0 bge0
When multipath routes exist, the "route delete" command requires
a specific gateway to be specified or else an error message would
be displayed. For example,
route delete default
would fail and trigger the following error message:
"route: writing to routing socket: No such process"
"delete net default: not in table"
On the other hand,
route delete default 10.2.5.2
would be successful: "delete net default: gateway 10.2.5.2"
One does not have to specify a gateway if there is only a single
route for a particular destination.
I need to perform more testings on address aliases and multiple
interfaces that have the same IP prefixes. This patch as it
stands today is not yet ready for prime time. Therefore, the ECMP
code fragments are fully guarded by the RADIX_MPATH macro.
Include the "options RADIX_MPATH" in the kernel configuration
to enable this feature.
Reviewed by: robert, sam, gnn, julian, kmacy
2008-04-13 05:45:14 +00:00
|
|
|
senderr(ESRCH);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
#endif
|
2003-10-04 03:44:50 +00:00
|
|
|
RT_LOCK(rt);
|
2003-11-08 23:36:32 +00:00
|
|
|
RT_ADDREF(rt);
|
2008-12-07 21:15:43 +00:00
|
|
|
RADIX_NODE_HEAD_RUNLOCK(rnh);
|
2002-12-24 03:03:39 +00:00
|
|
|
|
2005-07-15 09:18:34 +00:00
|
|
|
/*
|
|
|
|
* Fix for PR: 82974
|
|
|
|
*
|
|
|
|
* RTM_CHANGE/LOCK need a perfect match, rn_lookup()
|
|
|
|
* returns a perfect match in case a netmask is
|
|
|
|
* specified. For host routes only a longest prefix
|
|
|
|
* match is returned so it is necessary to compare the
|
|
|
|
* existence of the netmask. If both have a netmask
|
|
|
|
* rnh_lookup() did a perfect match and if none of them
|
|
|
|
* have a netmask both are host routes which is also a
|
|
|
|
* perfect match.
|
|
|
|
*/
|
|
|
|
|
|
|
|
if (rtm->rtm_type != RTM_GET &&
|
|
|
|
(!rt_mask(rt) != !info.rti_info[RTAX_NETMASK])) {
|
|
|
|
RT_UNLOCK(rt);
|
|
|
|
senderr(ESRCH);
|
|
|
|
}
|
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
switch(rtm->rtm_type) {
|
|
|
|
|
|
|
|
case RTM_GET:
|
1995-05-10 16:48:52 +00:00
|
|
|
report:
|
2003-10-04 03:44:50 +00:00
|
|
|
RT_LOCK_ASSERT(rt);
|
2009-01-09 21:57:49 +00:00
|
|
|
if (jailed(curthread->td_ucred) &&
|
|
|
|
((rt->rt_flags & RTF_HOST) == 0 ||
|
|
|
|
!prison_if(curthread->td_ucred, rt_key(rt)))) {
|
|
|
|
RT_UNLOCK(rt);
|
|
|
|
senderr(ESRCH);
|
|
|
|
}
|
2003-10-03 18:15:54 +00:00
|
|
|
info.rti_info[RTAX_DST] = rt_key(rt);
|
|
|
|
info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
|
|
|
|
info.rti_info[RTAX_NETMASK] = rt_mask(rt);
|
This main goals of this project are:
1. separating L2 tables (ARP, NDP) from the L3 routing tables
2. removing as much locking dependencies among these layers as
possible to allow for some parallelism in the search operations
3. simplify the logic in the routing code,
The most notable end result is the obsolescent of the route
cloning (RTF_CLONING) concept, which translated into code reduction
in both IPv4 ARP and IPv6 NDP related modules, and size reduction in
struct rtentry{}. The change in design obsoletes the semantics of
RTF_CLONING, RTF_WASCLONE and RTF_LLINFO routing flags. The userland
applications such as "arp" and "ndp" have been modified to reflect
those changes. The output from "netstat -r" shows only the routing
entries.
Quite a few developers have contributed to this project in the
past: Glebius Smirnoff, Luigi Rizzo, Alessandro Cerri, and
Andre Oppermann. And most recently:
- Kip Macy revised the locking code completely, thus completing
the last piece of the puzzle, Kip has also been conducting
active functional testing
- Sam Leffler has helped me improving/refactoring the code, and
provided valuable reviews
- Julian Elischer setup the perforce tree for me and has helped
me maintaining that branch before the svn conversion
2008-12-15 06:10:57 +00:00
|
|
|
info.rti_info[RTAX_GENMASK] = 0;
|
1994-05-24 10:09:53 +00:00
|
|
|
if (rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) {
|
1994-10-08 22:38:27 +00:00
|
|
|
ifp = rt->rt_ifp;
|
|
|
|
if (ifp) {
|
2004-04-16 08:14:34 +00:00
|
|
|
info.rti_info[RTAX_IFP] =
|
2005-11-11 16:04:59 +00:00
|
|
|
ifp->if_addr->ifa_addr;
|
2009-01-09 21:57:49 +00:00
|
|
|
if (jailed(curthread->td_ucred)) {
|
MFp4:
Bring in updated jail support from bz_jail branch.
This enhances the current jail implementation to permit multiple
addresses per jail. In addtion to IPv4, IPv6 is supported as well.
Due to updated checks it is even possible to have jails without
an IP address at all, which basically gives one a chroot with
restricted process view, no networking,..
SCTP support was updated and supports IPv6 in jails as well.
Cpuset support permits jails to be bound to specific processor
sets after creation.
Jails can have an unrestricted (no duplicate protection, etc.) name
in addition to the hostname. The jail name cannot be changed from
within a jail and is considered to be used for management purposes
or as audit-token in the future.
DDB 'show jails' command was added to aid debugging.
Proper compat support permits 32bit jail binaries to be used on 64bit
systems to manage jails. Also backward compatibility was preserved where
possible: for jail v1 syscalls, as well as with user space management
utilities.
Both jail as well as prison version were updated for the new features.
A gap was intentionally left as the intermediate versions had been
used by various patches floating around the last years.
Bump __FreeBSD_version for the afore mentioned and in kernel changes.
Special thanks to:
- Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches
and Olivier Houchard (cognet) for initial single-IPv6 patches.
- Jeff Roberson (jeff) and Randall Stewart (rrs) for their
help, ideas and review on cpuset and SCTP support.
- Robert Watson (rwatson) for lots and lots of help, discussions,
suggestions and review of most of the patch at various stages.
- John Baldwin (jhb) for his help.
- Simon L. Nielsen (simon) as early adopter testing changes
on cluster machines as well as all the testers and people
who provided feedback the last months on freebsd-jail and
other channels.
- My employer, CK Software GmbH, for the support so I could work on this.
Reviewed by: (see above)
MFC after: 3 months (this is just so that I get the mail)
X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
|
|
|
error = rtm_get_jailed(
|
|
|
|
&info, ifp, rt, &saun,
|
2009-01-09 21:57:49 +00:00
|
|
|
curthread->td_ucred);
|
MFp4:
Bring in updated jail support from bz_jail branch.
This enhances the current jail implementation to permit multiple
addresses per jail. In addtion to IPv4, IPv6 is supported as well.
Due to updated checks it is even possible to have jails without
an IP address at all, which basically gives one a chroot with
restricted process view, no networking,..
SCTP support was updated and supports IPv6 in jails as well.
Cpuset support permits jails to be bound to specific processor
sets after creation.
Jails can have an unrestricted (no duplicate protection, etc.) name
in addition to the hostname. The jail name cannot be changed from
within a jail and is considered to be used for management purposes
or as audit-token in the future.
DDB 'show jails' command was added to aid debugging.
Proper compat support permits 32bit jail binaries to be used on 64bit
systems to manage jails. Also backward compatibility was preserved where
possible: for jail v1 syscalls, as well as with user space management
utilities.
Both jail as well as prison version were updated for the new features.
A gap was intentionally left as the intermediate versions had been
used by various patches floating around the last years.
Bump __FreeBSD_version for the afore mentioned and in kernel changes.
Special thanks to:
- Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches
and Olivier Houchard (cognet) for initial single-IPv6 patches.
- Jeff Roberson (jeff) and Randall Stewart (rrs) for their
help, ideas and review on cpuset and SCTP support.
- Robert Watson (rwatson) for lots and lots of help, discussions,
suggestions and review of most of the patch at various stages.
- John Baldwin (jhb) for his help.
- Simon L. Nielsen (simon) as early adopter testing changes
on cluster machines as well as all the testers and people
who provided feedback the last months on freebsd-jail and
other channels.
- My employer, CK Software GmbH, for the support so I could work on this.
Reviewed by: (see above)
MFC after: 3 months (this is just so that I get the mail)
X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
|
|
|
if (error != 0) {
|
|
|
|
RT_UNLOCK(rt);
|
|
|
|
senderr(ESRCH);
|
|
|
|
}
|
|
|
|
} else {
|
2004-04-26 19:46:52 +00:00
|
|
|
info.rti_info[RTAX_IFA] =
|
2004-05-03 17:15:34 +00:00
|
|
|
rt->rt_ifa->ifa_addr;
|
MFp4:
Bring in updated jail support from bz_jail branch.
This enhances the current jail implementation to permit multiple
addresses per jail. In addtion to IPv4, IPv6 is supported as well.
Due to updated checks it is even possible to have jails without
an IP address at all, which basically gives one a chroot with
restricted process view, no networking,..
SCTP support was updated and supports IPv6 in jails as well.
Cpuset support permits jails to be bound to specific processor
sets after creation.
Jails can have an unrestricted (no duplicate protection, etc.) name
in addition to the hostname. The jail name cannot be changed from
within a jail and is considered to be used for management purposes
or as audit-token in the future.
DDB 'show jails' command was added to aid debugging.
Proper compat support permits 32bit jail binaries to be used on 64bit
systems to manage jails. Also backward compatibility was preserved where
possible: for jail v1 syscalls, as well as with user space management
utilities.
Both jail as well as prison version were updated for the new features.
A gap was intentionally left as the intermediate versions had been
used by various patches floating around the last years.
Bump __FreeBSD_version for the afore mentioned and in kernel changes.
Special thanks to:
- Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches
and Olivier Houchard (cognet) for initial single-IPv6 patches.
- Jeff Roberson (jeff) and Randall Stewart (rrs) for their
help, ideas and review on cpuset and SCTP support.
- Robert Watson (rwatson) for lots and lots of help, discussions,
suggestions and review of most of the patch at various stages.
- John Baldwin (jhb) for his help.
- Simon L. Nielsen (simon) as early adopter testing changes
on cluster machines as well as all the testers and people
who provided feedback the last months on freebsd-jail and
other channels.
- My employer, CK Software GmbH, for the support so I could work on this.
Reviewed by: (see above)
MFC after: 3 months (this is just so that I get the mail)
X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
|
|
|
}
|
2001-10-17 11:23:59 +00:00
|
|
|
if (ifp->if_flags & IFF_POINTOPOINT)
|
2004-05-03 17:15:34 +00:00
|
|
|
info.rti_info[RTAX_BRD] =
|
|
|
|
rt->rt_ifa->ifa_dstaddr;
|
1994-05-24 10:09:53 +00:00
|
|
|
rtm->rtm_index = ifp->if_index;
|
|
|
|
} else {
|
2004-04-18 00:56:44 +00:00
|
|
|
info.rti_info[RTAX_IFP] = NULL;
|
|
|
|
info.rti_info[RTAX_IFA] = NULL;
|
2002-10-01 15:48:31 +00:00
|
|
|
}
|
2005-06-09 12:20:50 +00:00
|
|
|
} else if ((ifp = rt->rt_ifp) != NULL) {
|
|
|
|
rtm->rtm_index = ifp->if_index;
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
2004-04-17 15:09:36 +00:00
|
|
|
len = rt_msg2(rtm->rtm_type, &info, NULL, NULL);
|
1994-05-24 10:09:53 +00:00
|
|
|
if (len > rtm->rtm_msglen) {
|
|
|
|
struct rt_msghdr *new_rtm;
|
|
|
|
R_Malloc(new_rtm, struct rt_msghdr *, len);
|
2004-04-18 00:56:44 +00:00
|
|
|
if (new_rtm == NULL) {
|
2003-10-04 03:44:50 +00:00
|
|
|
RT_UNLOCK(rt);
|
1994-05-24 10:09:53 +00:00
|
|
|
senderr(ENOBUFS);
|
2003-10-03 18:15:54 +00:00
|
|
|
}
|
2004-04-18 11:47:04 +00:00
|
|
|
bcopy(rtm, new_rtm, rtm->rtm_msglen);
|
1994-05-24 10:09:53 +00:00
|
|
|
Free(rtm); rtm = new_rtm;
|
|
|
|
}
|
2004-04-17 15:09:36 +00:00
|
|
|
(void)rt_msg2(rtm->rtm_type, &info, (caddr_t)rtm, NULL);
|
1994-05-24 10:09:53 +00:00
|
|
|
rtm->rtm_flags = rt->rt_flags;
|
2006-03-15 19:39:09 +00:00
|
|
|
rtm->rtm_use = 0;
|
2003-11-20 20:07:39 +00:00
|
|
|
rt_getmetrics(&rt->rt_rmx, &rtm->rtm_rmx);
|
1994-05-24 10:09:53 +00:00
|
|
|
rtm->rtm_addrs = info.rti_addrs;
|
|
|
|
break;
|
|
|
|
|
|
|
|
case RTM_CHANGE:
|
2003-10-03 18:15:54 +00:00
|
|
|
/*
|
|
|
|
* New gateway could require new ifaddr, ifp;
|
|
|
|
* flags may also be different; ifp may be specified
|
|
|
|
* by ll sockaddr when protocol address is ambiguous
|
|
|
|
*/
|
|
|
|
if (((rt->rt_flags & RTF_GATEWAY) &&
|
|
|
|
info.rti_info[RTAX_GATEWAY] != NULL) ||
|
|
|
|
info.rti_info[RTAX_IFP] != NULL ||
|
|
|
|
(info.rti_info[RTAX_IFA] != NULL &&
|
|
|
|
!sa_equal(info.rti_info[RTAX_IFA],
|
|
|
|
rt->rt_ifa->ifa_addr))) {
|
2005-09-19 16:27:22 +00:00
|
|
|
RT_UNLOCK(rt);
|
2008-12-10 09:21:52 +00:00
|
|
|
RADIX_NODE_HEAD_LOCK(rnh);
|
2008-12-13 22:41:36 +00:00
|
|
|
error = rt_getifa_fib(&info, rt->rt_fibnum);
|
2008-12-10 09:21:52 +00:00
|
|
|
RADIX_NODE_HEAD_UNLOCK(rnh);
|
2008-12-13 22:41:36 +00:00
|
|
|
if (error != 0)
|
|
|
|
senderr(error);
|
2005-09-19 16:27:22 +00:00
|
|
|
RT_LOCK(rt);
|
2001-12-19 16:03:27 +00:00
|
|
|
}
|
2007-03-27 19:36:12 +00:00
|
|
|
if (info.rti_ifa != NULL &&
|
|
|
|
info.rti_ifa != rt->rt_ifa &&
|
|
|
|
rt->rt_ifa != NULL &&
|
2007-03-22 10:51:03 +00:00
|
|
|
rt->rt_ifa->ifa_rtrequest != NULL) {
|
|
|
|
rt->rt_ifa->ifa_rtrequest(RTM_DELETE, rt,
|
|
|
|
&info);
|
|
|
|
IFAFREE(rt->rt_ifa);
|
|
|
|
}
|
|
|
|
if (info.rti_info[RTAX_GATEWAY] != NULL) {
|
2008-12-10 09:21:52 +00:00
|
|
|
RT_UNLOCK(rt);
|
|
|
|
RADIX_NODE_HEAD_LOCK(rnh);
|
|
|
|
RT_LOCK(rt);
|
|
|
|
|
|
|
|
error = rt_setgate(rt, rt_key(rt),
|
|
|
|
info.rti_info[RTAX_GATEWAY]);
|
|
|
|
RADIX_NODE_HEAD_UNLOCK(rnh);
|
|
|
|
if (error != 0) {
|
2007-03-22 10:51:03 +00:00
|
|
|
RT_UNLOCK(rt);
|
|
|
|
senderr(error);
|
|
|
|
}
|
This main goals of this project are:
1. separating L2 tables (ARP, NDP) from the L3 routing tables
2. removing as much locking dependencies among these layers as
possible to allow for some parallelism in the search operations
3. simplify the logic in the routing code,
The most notable end result is the obsolescent of the route
cloning (RTF_CLONING) concept, which translated into code reduction
in both IPv4 ARP and IPv6 NDP related modules, and size reduction in
struct rtentry{}. The change in design obsoletes the semantics of
RTF_CLONING, RTF_WASCLONE and RTF_LLINFO routing flags. The userland
applications such as "arp" and "ndp" have been modified to reflect
those changes. The output from "netstat -r" shows only the routing
entries.
Quite a few developers have contributed to this project in the
past: Glebius Smirnoff, Luigi Rizzo, Alessandro Cerri, and
Andre Oppermann. And most recently:
- Kip Macy revised the locking code completely, thus completing
the last piece of the puzzle, Kip has also been conducting
active functional testing
- Sam Leffler has helped me improving/refactoring the code, and
provided valuable reviews
- Julian Elischer setup the perforce tree for me and has helped
me maintaining that branch before the svn conversion
2008-12-15 06:10:57 +00:00
|
|
|
rt->rt_flags |= RTF_GATEWAY;
|
2003-10-03 18:15:54 +00:00
|
|
|
}
|
2007-03-27 19:36:12 +00:00
|
|
|
if (info.rti_ifa != NULL &&
|
|
|
|
info.rti_ifa != rt->rt_ifa) {
|
|
|
|
IFAREF(info.rti_ifa);
|
2007-03-22 10:51:03 +00:00
|
|
|
rt->rt_ifa = info.rti_ifa;
|
2007-03-27 19:36:12 +00:00
|
|
|
rt->rt_ifp = info.rti_ifp;
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
2006-03-15 19:39:09 +00:00
|
|
|
/* Allow some flags to be toggled on change. */
|
|
|
|
if (rtm->rtm_fmask & RTF_FMASK)
|
|
|
|
rt->rt_flags = (rt->rt_flags &
|
|
|
|
~rtm->rtm_fmask) |
|
|
|
|
(rtm->rtm_flags & rtm->rtm_fmask);
|
1994-05-24 10:09:53 +00:00
|
|
|
rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx,
|
|
|
|
&rt->rt_rmx);
|
2006-03-15 19:39:09 +00:00
|
|
|
rtm->rtm_index = rt->rt_ifp->if_index;
|
1994-05-24 10:09:53 +00:00
|
|
|
if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest)
|
2001-10-17 18:07:05 +00:00
|
|
|
rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, &info);
|
2002-08-25 13:23:09 +00:00
|
|
|
/* FALLTHROUGH */
|
1994-05-24 10:09:53 +00:00
|
|
|
case RTM_LOCK:
|
2003-11-20 20:07:39 +00:00
|
|
|
/* We don't support locks anymore */
|
1994-05-24 10:09:53 +00:00
|
|
|
break;
|
|
|
|
}
|
2003-10-04 03:44:50 +00:00
|
|
|
RT_UNLOCK(rt);
|
1994-05-24 10:09:53 +00:00
|
|
|
break;
|
|
|
|
|
|
|
|
default:
|
|
|
|
senderr(EOPNOTSUPP);
|
|
|
|
}
|
|
|
|
|
|
|
|
flush:
|
|
|
|
if (rtm) {
|
|
|
|
if (error)
|
|
|
|
rtm->rtm_errno = error;
|
1995-05-30 08:16:23 +00:00
|
|
|
else
|
1994-05-24 10:09:53 +00:00
|
|
|
rtm->rtm_flags |= RTF_DONE;
|
|
|
|
}
|
2003-10-03 18:15:54 +00:00
|
|
|
if (rt) /* XXX can this be true? */
|
|
|
|
RTFREE(rt);
|
1994-05-24 10:09:53 +00:00
|
|
|
{
|
2004-04-18 00:56:44 +00:00
|
|
|
struct rawcb *rp = NULL;
|
1994-05-24 10:09:53 +00:00
|
|
|
/*
|
|
|
|
* Check to see if we don't want our own messages.
|
|
|
|
*/
|
|
|
|
if ((so->so_options & SO_USELOOPBACK) == 0) {
|
|
|
|
if (route_cb.any_count <= 1) {
|
|
|
|
if (rtm)
|
|
|
|
Free(rtm);
|
|
|
|
m_freem(m);
|
|
|
|
return (error);
|
|
|
|
}
|
|
|
|
/* There is another listener, so construct message */
|
|
|
|
rp = sotorawcb(so);
|
2002-05-31 11:52:35 +00:00
|
|
|
}
|
1994-05-24 10:09:53 +00:00
|
|
|
if (rtm) {
|
|
|
|
m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm);
|
2001-07-04 11:29:55 +00:00
|
|
|
if (m->m_pkthdr.len < rtm->rtm_msglen) {
|
|
|
|
m_freem(m);
|
|
|
|
m = NULL;
|
|
|
|
} else if (m->m_pkthdr.len > rtm->rtm_msglen)
|
|
|
|
m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len);
|
1994-05-24 10:09:53 +00:00
|
|
|
Free(rtm);
|
|
|
|
}
|
2003-10-03 18:15:54 +00:00
|
|
|
if (m) {
|
|
|
|
if (rp) {
|
|
|
|
/*
|
|
|
|
* XXX insure we don't get a copy by
|
|
|
|
* invalidating our protocol
|
|
|
|
*/
|
|
|
|
unsigned short family = rp->rcb_proto.sp_family;
|
|
|
|
rp->rcb_proto.sp_family = 0;
|
|
|
|
rt_dispatch(m, info.rti_info[RTAX_DST]);
|
|
|
|
rp->rcb_proto.sp_family = family;
|
|
|
|
} else
|
|
|
|
rt_dispatch(m, info.rti_info[RTAX_DST]);
|
|
|
|
}
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
|
|
|
return (error);
|
2003-10-03 18:15:54 +00:00
|
|
|
#undef sa_equal
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
|
|
|
|
1995-11-16 19:00:27 +00:00
|
|
|
static void
|
2004-04-18 00:56:44 +00:00
|
|
|
rt_setmetrics(u_long which, const struct rt_metrics *in,
|
|
|
|
struct rt_metrics_lite *out)
|
1994-05-24 10:09:53 +00:00
|
|
|
{
|
|
|
|
#define metric(f, e) if (which & (f)) out->e = in->e;
|
2003-11-20 20:07:39 +00:00
|
|
|
/*
|
|
|
|
* Only these are stored in the routing entry since introduction
|
|
|
|
* of tcp hostcache. The rest is ignored.
|
|
|
|
*/
|
1994-05-24 10:09:53 +00:00
|
|
|
metric(RTV_MTU, rmx_mtu);
|
2006-07-06 00:24:36 +00:00
|
|
|
/* Userland -> kernel timebase conversion. */
|
|
|
|
if (which & RTV_EXPIRE)
|
|
|
|
out->rmx_expire = in->rmx_expire ?
|
|
|
|
in->rmx_expire - time_second + time_uptime : 0;
|
1994-05-24 10:09:53 +00:00
|
|
|
#undef metric
|
|
|
|
}
|
|
|
|
|
2003-11-20 20:07:39 +00:00
|
|
|
static void
|
2004-04-18 00:56:44 +00:00
|
|
|
rt_getmetrics(const struct rt_metrics_lite *in, struct rt_metrics *out)
|
2003-11-20 20:07:39 +00:00
|
|
|
{
|
|
|
|
#define metric(e) out->e = in->e;
|
|
|
|
bzero(out, sizeof(*out));
|
|
|
|
metric(rmx_mtu);
|
2006-07-06 00:24:36 +00:00
|
|
|
/* Kernel -> userland timebase conversion. */
|
|
|
|
out->rmx_expire = in->rmx_expire ?
|
|
|
|
in->rmx_expire - time_uptime + time_second : 0;
|
2003-11-20 20:07:39 +00:00
|
|
|
#undef metric
|
|
|
|
}
|
|
|
|
|
1997-07-15 23:25:32 +00:00
|
|
|
/*
|
|
|
|
* Extract the addresses of the passed sockaddrs.
|
|
|
|
* Do a little sanity checking so as to avoid bad memory references.
|
1997-07-18 11:44:24 +00:00
|
|
|
* This data is derived straight from userland.
|
1997-07-15 23:25:32 +00:00
|
|
|
*/
|
1997-07-18 11:44:24 +00:00
|
|
|
static int
|
2003-10-03 18:15:54 +00:00
|
|
|
rt_xaddrs(caddr_t cp, caddr_t cplim, struct rt_addrinfo *rtinfo)
|
1994-05-24 10:09:53 +00:00
|
|
|
{
|
2004-04-13 11:22:22 +00:00
|
|
|
struct sockaddr *sa;
|
|
|
|
int i;
|
1994-05-24 10:09:53 +00:00
|
|
|
|
2003-10-03 18:15:54 +00:00
|
|
|
for (i = 0; i < RTAX_MAX && cp < cplim; i++) {
|
1994-05-24 10:09:53 +00:00
|
|
|
if ((rtinfo->rti_addrs & (1 << i)) == 0)
|
|
|
|
continue;
|
1997-07-16 14:55:14 +00:00
|
|
|
sa = (struct sockaddr *)cp;
|
1997-07-15 23:25:32 +00:00
|
|
|
/*
|
1997-07-18 11:44:24 +00:00
|
|
|
* It won't fit.
|
1997-07-15 23:25:32 +00:00
|
|
|
*/
|
2003-10-03 18:15:54 +00:00
|
|
|
if (cp + sa->sa_len > cplim)
|
1997-07-18 11:44:24 +00:00
|
|
|
return (EINVAL);
|
1997-07-15 23:25:32 +00:00
|
|
|
/*
|
|
|
|
* there are no more.. quit now
|
|
|
|
* If there are more bits, they are in error.
|
|
|
|
* I've seen this. route(1) can evidently generate these.
|
|
|
|
* This causes kernel to core dump.
|
1997-07-18 11:44:24 +00:00
|
|
|
* for compatibility, If we see this, point to a safe address.
|
1997-07-15 23:25:32 +00:00
|
|
|
*/
|
1997-07-18 11:44:24 +00:00
|
|
|
if (sa->sa_len == 0) {
|
|
|
|
rtinfo->rti_info[i] = &sa_zero;
|
|
|
|
return (0); /* should be EINVAL but for compat */
|
|
|
|
}
|
|
|
|
/* accept it */
|
|
|
|
rtinfo->rti_info[i] = sa;
|
2004-04-13 11:22:22 +00:00
|
|
|
cp += SA_SIZE(sa);
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
1997-07-18 11:44:24 +00:00
|
|
|
return (0);
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static struct mbuf *
|
2003-10-03 18:15:54 +00:00
|
|
|
rt_msg1(int type, struct rt_addrinfo *rtinfo)
|
1994-05-24 10:09:53 +00:00
|
|
|
{
|
2004-04-18 00:56:44 +00:00
|
|
|
struct rt_msghdr *rtm;
|
|
|
|
struct mbuf *m;
|
|
|
|
int i;
|
|
|
|
struct sockaddr *sa;
|
1994-05-24 10:09:53 +00:00
|
|
|
int len, dlen;
|
|
|
|
|
|
|
|
switch (type) {
|
|
|
|
|
|
|
|
case RTM_DELADDR:
|
|
|
|
case RTM_NEWADDR:
|
|
|
|
len = sizeof(struct ifa_msghdr);
|
|
|
|
break;
|
|
|
|
|
1997-01-13 21:26:53 +00:00
|
|
|
case RTM_DELMADDR:
|
|
|
|
case RTM_NEWMADDR:
|
|
|
|
len = sizeof(struct ifma_msghdr);
|
|
|
|
break;
|
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
case RTM_IFINFO:
|
|
|
|
len = sizeof(struct if_msghdr);
|
|
|
|
break;
|
|
|
|
|
2002-01-18 14:33:04 +00:00
|
|
|
case RTM_IFANNOUNCE:
|
2004-10-05 19:48:33 +00:00
|
|
|
case RTM_IEEE80211:
|
2002-01-18 14:33:04 +00:00
|
|
|
len = sizeof(struct if_announcemsghdr);
|
|
|
|
break;
|
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
default:
|
|
|
|
len = sizeof(struct rt_msghdr);
|
|
|
|
}
|
2001-06-11 12:39:29 +00:00
|
|
|
if (len > MCLBYTES)
|
1994-05-24 10:09:53 +00:00
|
|
|
panic("rt_msg1");
|
2003-02-19 05:47:46 +00:00
|
|
|
m = m_gethdr(M_DONTWAIT, MT_DATA);
|
2001-06-11 12:39:29 +00:00
|
|
|
if (m && len > MHLEN) {
|
2003-02-19 05:47:46 +00:00
|
|
|
MCLGET(m, M_DONTWAIT);
|
2001-06-11 12:39:29 +00:00
|
|
|
if ((m->m_flags & M_EXT) == 0) {
|
|
|
|
m_free(m);
|
|
|
|
m = NULL;
|
|
|
|
}
|
|
|
|
}
|
2004-04-18 00:56:44 +00:00
|
|
|
if (m == NULL)
|
2001-06-11 12:39:29 +00:00
|
|
|
return (m);
|
1994-05-24 10:09:53 +00:00
|
|
|
m->m_pkthdr.len = m->m_len = len;
|
2004-04-18 00:56:44 +00:00
|
|
|
m->m_pkthdr.rcvif = NULL;
|
1994-05-24 10:09:53 +00:00
|
|
|
rtm = mtod(m, struct rt_msghdr *);
|
|
|
|
bzero((caddr_t)rtm, len);
|
|
|
|
for (i = 0; i < RTAX_MAX; i++) {
|
|
|
|
if ((sa = rtinfo->rti_info[i]) == NULL)
|
|
|
|
continue;
|
|
|
|
rtinfo->rti_addrs |= (1 << i);
|
2004-04-13 11:22:22 +00:00
|
|
|
dlen = SA_SIZE(sa);
|
1994-05-24 10:09:53 +00:00
|
|
|
m_copyback(m, len, dlen, (caddr_t)sa);
|
|
|
|
len += dlen;
|
|
|
|
}
|
|
|
|
if (m->m_pkthdr.len != len) {
|
|
|
|
m_freem(m);
|
|
|
|
return (NULL);
|
|
|
|
}
|
|
|
|
rtm->rtm_msglen = len;
|
|
|
|
rtm->rtm_version = RTM_VERSION;
|
|
|
|
rtm->rtm_type = type;
|
|
|
|
return (m);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
2003-10-03 18:15:54 +00:00
|
|
|
rt_msg2(int type, struct rt_addrinfo *rtinfo, caddr_t cp, struct walkarg *w)
|
1994-05-24 10:09:53 +00:00
|
|
|
{
|
2004-04-18 00:56:44 +00:00
|
|
|
int i;
|
1994-05-24 10:09:53 +00:00
|
|
|
int len, dlen, second_time = 0;
|
|
|
|
caddr_t cp0;
|
|
|
|
|
|
|
|
rtinfo->rti_addrs = 0;
|
|
|
|
again:
|
|
|
|
switch (type) {
|
|
|
|
|
|
|
|
case RTM_DELADDR:
|
|
|
|
case RTM_NEWADDR:
|
|
|
|
len = sizeof(struct ifa_msghdr);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case RTM_IFINFO:
|
|
|
|
len = sizeof(struct if_msghdr);
|
|
|
|
break;
|
|
|
|
|
2003-11-14 18:48:15 +00:00
|
|
|
case RTM_NEWMADDR:
|
|
|
|
len = sizeof(struct ifma_msghdr);
|
|
|
|
break;
|
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
default:
|
|
|
|
len = sizeof(struct rt_msghdr);
|
|
|
|
}
|
1994-10-08 22:38:27 +00:00
|
|
|
cp0 = cp;
|
|
|
|
if (cp0)
|
1994-05-24 10:09:53 +00:00
|
|
|
cp += len;
|
|
|
|
for (i = 0; i < RTAX_MAX; i++) {
|
2004-04-18 00:56:44 +00:00
|
|
|
struct sockaddr *sa;
|
1994-05-24 10:09:53 +00:00
|
|
|
|
2004-04-18 00:56:44 +00:00
|
|
|
if ((sa = rtinfo->rti_info[i]) == NULL)
|
1994-05-24 10:09:53 +00:00
|
|
|
continue;
|
|
|
|
rtinfo->rti_addrs |= (1 << i);
|
2004-04-13 11:22:22 +00:00
|
|
|
dlen = SA_SIZE(sa);
|
1994-05-24 10:09:53 +00:00
|
|
|
if (cp) {
|
|
|
|
bcopy((caddr_t)sa, cp, (unsigned)dlen);
|
|
|
|
cp += dlen;
|
|
|
|
}
|
|
|
|
len += dlen;
|
|
|
|
}
|
2002-01-27 20:39:01 +00:00
|
|
|
len = ALIGN(len);
|
2004-04-18 00:56:44 +00:00
|
|
|
if (cp == NULL && w != NULL && !second_time) {
|
|
|
|
struct walkarg *rw = w;
|
1994-05-24 10:09:53 +00:00
|
|
|
|
1995-11-16 19:00:27 +00:00
|
|
|
if (rw->w_req) {
|
1994-05-24 10:09:53 +00:00
|
|
|
if (rw->w_tmemsize < len) {
|
|
|
|
if (rw->w_tmem)
|
|
|
|
free(rw->w_tmem, M_RTABLE);
|
1994-10-08 22:38:27 +00:00
|
|
|
rw->w_tmem = (caddr_t)
|
|
|
|
malloc(len, M_RTABLE, M_NOWAIT);
|
|
|
|
if (rw->w_tmem)
|
1994-05-24 10:09:53 +00:00
|
|
|
rw->w_tmemsize = len;
|
|
|
|
}
|
|
|
|
if (rw->w_tmem) {
|
|
|
|
cp = rw->w_tmem;
|
|
|
|
second_time = 1;
|
|
|
|
goto again;
|
1995-11-16 19:00:27 +00:00
|
|
|
}
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
if (cp) {
|
2004-04-18 00:56:44 +00:00
|
|
|
struct rt_msghdr *rtm = (struct rt_msghdr *)cp0;
|
1994-05-24 10:09:53 +00:00
|
|
|
|
|
|
|
rtm->rtm_version = RTM_VERSION;
|
|
|
|
rtm->rtm_type = type;
|
|
|
|
rtm->rtm_msglen = len;
|
|
|
|
}
|
|
|
|
return (len);
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* This routine is called to generate a message from the routing
|
|
|
|
* socket indicating that a redirect has occured, a routing lookup
|
|
|
|
* has failed, or that a protocol has detected timeouts to a particular
|
|
|
|
* destination.
|
|
|
|
*/
|
|
|
|
void
|
2003-10-03 18:15:54 +00:00
|
|
|
rt_missmsg(int type, struct rt_addrinfo *rtinfo, int flags, int error)
|
1994-05-24 10:09:53 +00:00
|
|
|
{
|
2003-10-03 18:15:54 +00:00
|
|
|
struct rt_msghdr *rtm;
|
|
|
|
struct mbuf *m;
|
1994-05-24 10:09:53 +00:00
|
|
|
struct sockaddr *sa = rtinfo->rti_info[RTAX_DST];
|
|
|
|
|
|
|
|
if (route_cb.any_count == 0)
|
|
|
|
return;
|
|
|
|
m = rt_msg1(type, rtinfo);
|
2004-04-18 00:56:44 +00:00
|
|
|
if (m == NULL)
|
1994-05-24 10:09:53 +00:00
|
|
|
return;
|
|
|
|
rtm = mtod(m, struct rt_msghdr *);
|
|
|
|
rtm->rtm_flags = RTF_DONE | flags;
|
|
|
|
rtm->rtm_errno = error;
|
|
|
|
rtm->rtm_addrs = rtinfo->rti_addrs;
|
2003-10-03 18:15:54 +00:00
|
|
|
rt_dispatch(m, sa);
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* This routine is called to generate a message from the routing
|
|
|
|
* socket indicating that the status of a network interface has changed.
|
|
|
|
*/
|
|
|
|
void
|
2003-10-03 18:15:54 +00:00
|
|
|
rt_ifmsg(struct ifnet *ifp)
|
1994-05-24 10:09:53 +00:00
|
|
|
{
|
2003-10-03 18:15:54 +00:00
|
|
|
struct if_msghdr *ifm;
|
1994-05-24 10:09:53 +00:00
|
|
|
struct mbuf *m;
|
|
|
|
struct rt_addrinfo info;
|
|
|
|
|
|
|
|
if (route_cb.any_count == 0)
|
|
|
|
return;
|
|
|
|
bzero((caddr_t)&info, sizeof(info));
|
|
|
|
m = rt_msg1(RTM_IFINFO, &info);
|
2004-04-18 00:56:44 +00:00
|
|
|
if (m == NULL)
|
1994-05-24 10:09:53 +00:00
|
|
|
return;
|
|
|
|
ifm = mtod(m, struct if_msghdr *);
|
|
|
|
ifm->ifm_index = ifp->if_index;
|
Rename IFF_RUNNING to IFF_DRV_RUNNING, IFF_OACTIVE to IFF_DRV_OACTIVE,
and move both flags from ifnet.if_flags to ifnet.if_drv_flags, making
and documenting the locking of these flags the responsibility of the
device driver, not the network stack. The flags for these two fields
will be mutually exclusive so that they can be exposed to user space as
though they were stored in the same variable.
Provide #defines to provide the old names #ifndef _KERNEL, so that user
applications (such as ifconfig) can use the old flag names. Using the
old names in a device driver will result in a compile error in order to
help device driver writers adopt the new model.
When exposing the interface flags to user space, via interface ioctls
or routing sockets, or the two fields together. Since the driver flags
cannot currently be set for user space, no new logic is currently
required to handle this case.
Add some assertions that general purpose network stack routines, such
as if_setflags(), are not improperly used on driver-owned flags.
With this change, a large number of very minor network stack races are
closed, subject to correct device driver locking. Most were likely
never triggered.
Driver sweep to follow; many thanks to pjd and bz for the line-by-line
review they gave this patch.
Reviewed by: pjd, bz
MFC after: 7 days
2005-08-09 10:16:17 +00:00
|
|
|
ifm->ifm_flags = ifp->if_flags | ifp->if_drv_flags;
|
1994-05-24 10:09:53 +00:00
|
|
|
ifm->ifm_data = ifp->if_data;
|
|
|
|
ifm->ifm_addrs = 0;
|
2003-10-03 18:15:54 +00:00
|
|
|
rt_dispatch(m, NULL);
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* This is called to generate messages from the routing socket
|
|
|
|
* indicating a network interface has had addresses associated with it.
|
|
|
|
* if we ever reverse the logic and replace messages TO the routing
|
|
|
|
* socket indicate a request to configure interfaces, then it will
|
|
|
|
* be unnecessary as the routing socket will automatically generate
|
|
|
|
* copies of it.
|
|
|
|
*/
|
|
|
|
void
|
2003-10-03 18:15:54 +00:00
|
|
|
rt_newaddrmsg(int cmd, struct ifaddr *ifa, int error, struct rtentry *rt)
|
1994-05-24 10:09:53 +00:00
|
|
|
{
|
|
|
|
struct rt_addrinfo info;
|
2004-04-18 00:56:44 +00:00
|
|
|
struct sockaddr *sa = NULL;
|
1994-05-24 10:09:53 +00:00
|
|
|
int pass;
|
2004-04-18 00:56:44 +00:00
|
|
|
struct mbuf *m = NULL;
|
1994-05-24 10:09:53 +00:00
|
|
|
struct ifnet *ifp = ifa->ifa_ifp;
|
|
|
|
|
2005-03-26 21:49:43 +00:00
|
|
|
KASSERT(cmd == RTM_ADD || cmd == RTM_DELETE,
|
|
|
|
("unexpected cmd %u", cmd));
|
2006-11-03 15:23:16 +00:00
|
|
|
#ifdef SCTP
|
|
|
|
/*
|
|
|
|
* notify the SCTP stack
|
|
|
|
* this will only get called when an address is added/deleted
|
|
|
|
* XXX pass the ifaddr struct instead if ifa->ifa_addr...
|
|
|
|
*/
|
|
|
|
sctp_addr_change(ifa, cmd);
|
|
|
|
#endif /* SCTP */
|
1994-05-24 10:09:53 +00:00
|
|
|
if (route_cb.any_count == 0)
|
|
|
|
return;
|
|
|
|
for (pass = 1; pass < 3; pass++) {
|
|
|
|
bzero((caddr_t)&info, sizeof(info));
|
|
|
|
if ((cmd == RTM_ADD && pass == 1) ||
|
|
|
|
(cmd == RTM_DELETE && pass == 2)) {
|
2004-04-18 00:56:44 +00:00
|
|
|
struct ifa_msghdr *ifam;
|
1994-05-24 10:09:53 +00:00
|
|
|
int ncmd = cmd == RTM_ADD ? RTM_NEWADDR : RTM_DELADDR;
|
|
|
|
|
2003-10-03 18:15:54 +00:00
|
|
|
info.rti_info[RTAX_IFA] = sa = ifa->ifa_addr;
|
2005-11-11 16:04:59 +00:00
|
|
|
info.rti_info[RTAX_IFP] = ifp->if_addr->ifa_addr;
|
2003-10-03 18:15:54 +00:00
|
|
|
info.rti_info[RTAX_NETMASK] = ifa->ifa_netmask;
|
|
|
|
info.rti_info[RTAX_BRD] = ifa->ifa_dstaddr;
|
1994-05-24 10:09:53 +00:00
|
|
|
if ((m = rt_msg1(ncmd, &info)) == NULL)
|
|
|
|
continue;
|
|
|
|
ifam = mtod(m, struct ifa_msghdr *);
|
|
|
|
ifam->ifam_index = ifp->if_index;
|
|
|
|
ifam->ifam_metric = ifa->ifa_metric;
|
|
|
|
ifam->ifam_flags = ifa->ifa_flags;
|
|
|
|
ifam->ifam_addrs = info.rti_addrs;
|
|
|
|
}
|
|
|
|
if ((cmd == RTM_ADD && pass == 2) ||
|
|
|
|
(cmd == RTM_DELETE && pass == 1)) {
|
2004-04-18 00:56:44 +00:00
|
|
|
struct rt_msghdr *rtm;
|
1995-05-30 08:16:23 +00:00
|
|
|
|
2004-04-18 00:56:44 +00:00
|
|
|
if (rt == NULL)
|
1994-05-24 10:09:53 +00:00
|
|
|
continue;
|
2003-10-03 18:15:54 +00:00
|
|
|
info.rti_info[RTAX_NETMASK] = rt_mask(rt);
|
|
|
|
info.rti_info[RTAX_DST] = sa = rt_key(rt);
|
|
|
|
info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
|
1994-05-24 10:09:53 +00:00
|
|
|
if ((m = rt_msg1(cmd, &info)) == NULL)
|
|
|
|
continue;
|
|
|
|
rtm = mtod(m, struct rt_msghdr *);
|
|
|
|
rtm->rtm_index = ifp->if_index;
|
|
|
|
rtm->rtm_flags |= rt->rt_flags;
|
|
|
|
rtm->rtm_errno = error;
|
|
|
|
rtm->rtm_addrs = info.rti_addrs;
|
|
|
|
}
|
2003-10-03 18:15:54 +00:00
|
|
|
rt_dispatch(m, sa);
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
1997-01-13 21:26:53 +00:00
|
|
|
/*
|
|
|
|
* This is the analogue to the rt_newaddrmsg which performs the same
|
|
|
|
* function but for multicast group memberhips. This is easier since
|
|
|
|
* there is no route state to worry about.
|
|
|
|
*/
|
|
|
|
void
|
2003-10-03 18:15:54 +00:00
|
|
|
rt_newmaddrmsg(int cmd, struct ifmultiaddr *ifma)
|
1997-01-13 21:26:53 +00:00
|
|
|
{
|
|
|
|
struct rt_addrinfo info;
|
2004-04-18 00:56:44 +00:00
|
|
|
struct mbuf *m = NULL;
|
1997-01-13 21:26:53 +00:00
|
|
|
struct ifnet *ifp = ifma->ifma_ifp;
|
|
|
|
struct ifma_msghdr *ifmam;
|
|
|
|
|
|
|
|
if (route_cb.any_count == 0)
|
|
|
|
return;
|
|
|
|
|
|
|
|
bzero((caddr_t)&info, sizeof(info));
|
2003-10-03 18:15:54 +00:00
|
|
|
info.rti_info[RTAX_IFA] = ifma->ifma_addr;
|
2005-11-11 16:04:59 +00:00
|
|
|
info.rti_info[RTAX_IFP] = ifp ? ifp->if_addr->ifa_addr : NULL;
|
1997-01-13 21:26:53 +00:00
|
|
|
/*
|
|
|
|
* If a link-layer address is present, present it as a ``gateway''
|
|
|
|
* (similarly to how ARP entries, e.g., are presented).
|
|
|
|
*/
|
2003-10-03 18:15:54 +00:00
|
|
|
info.rti_info[RTAX_GATEWAY] = ifma->ifma_lladdr;
|
|
|
|
m = rt_msg1(cmd, &info);
|
|
|
|
if (m == NULL)
|
1997-01-13 21:26:53 +00:00
|
|
|
return;
|
|
|
|
ifmam = mtod(m, struct ifma_msghdr *);
|
2007-03-27 16:11:28 +00:00
|
|
|
KASSERT(ifp != NULL, ("%s: link-layer multicast address w/o ifp\n",
|
|
|
|
__func__));
|
1997-01-13 21:26:53 +00:00
|
|
|
ifmam->ifmam_index = ifp->if_index;
|
|
|
|
ifmam->ifmam_addrs = info.rti_addrs;
|
2003-10-03 18:15:54 +00:00
|
|
|
rt_dispatch(m, ifma->ifma_addr);
|
1997-01-13 21:26:53 +00:00
|
|
|
}
|
1995-11-16 19:00:27 +00:00
|
|
|
|
2004-10-05 19:48:33 +00:00
|
|
|
static struct mbuf *
|
|
|
|
rt_makeifannouncemsg(struct ifnet *ifp, int type, int what,
|
|
|
|
struct rt_addrinfo *info)
|
|
|
|
{
|
|
|
|
struct if_announcemsghdr *ifan;
|
|
|
|
struct mbuf *m;
|
|
|
|
|
|
|
|
if (route_cb.any_count == 0)
|
|
|
|
return NULL;
|
|
|
|
bzero((caddr_t)info, sizeof(*info));
|
|
|
|
m = rt_msg1(type, info);
|
|
|
|
if (m != NULL) {
|
|
|
|
ifan = mtod(m, struct if_announcemsghdr *);
|
|
|
|
ifan->ifan_index = ifp->if_index;
|
|
|
|
strlcpy(ifan->ifan_name, ifp->if_xname,
|
|
|
|
sizeof(ifan->ifan_name));
|
|
|
|
ifan->ifan_what = what;
|
|
|
|
}
|
|
|
|
return m;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* This is called to generate routing socket messages indicating
|
|
|
|
* IEEE80211 wireless events.
|
|
|
|
* XXX we piggyback on the RTM_IFANNOUNCE msg format in a clumsy way.
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
rt_ieee80211msg(struct ifnet *ifp, int what, void *data, size_t data_len)
|
|
|
|
{
|
|
|
|
struct mbuf *m;
|
|
|
|
struct rt_addrinfo info;
|
|
|
|
|
|
|
|
m = rt_makeifannouncemsg(ifp, RTM_IEEE80211, what, &info);
|
|
|
|
if (m != NULL) {
|
|
|
|
/*
|
|
|
|
* Append the ieee80211 data. Try to stick it in the
|
|
|
|
* mbuf containing the ifannounce msg; otherwise allocate
|
|
|
|
* a new mbuf and append.
|
|
|
|
*
|
|
|
|
* NB: we assume m is a single mbuf.
|
|
|
|
*/
|
|
|
|
if (data_len > M_TRAILINGSPACE(m)) {
|
|
|
|
struct mbuf *n = m_get(M_NOWAIT, MT_DATA);
|
|
|
|
if (n == NULL) {
|
|
|
|
m_freem(m);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
bcopy(data, mtod(n, void *), data_len);
|
|
|
|
n->m_len = data_len;
|
|
|
|
m->m_next = n;
|
|
|
|
} else if (data_len > 0) {
|
|
|
|
bcopy(data, mtod(m, u_int8_t *) + m->m_len, data_len);
|
|
|
|
m->m_len += data_len;
|
|
|
|
}
|
|
|
|
if (m->m_flags & M_PKTHDR)
|
|
|
|
m->m_pkthdr.len += data_len;
|
|
|
|
mtod(m, struct if_announcemsghdr *)->ifan_msglen += data_len;
|
|
|
|
rt_dispatch(m, NULL);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2002-01-18 14:33:04 +00:00
|
|
|
/*
|
|
|
|
* This is called to generate routing socket messages indicating
|
|
|
|
* network interface arrival and departure.
|
|
|
|
*/
|
|
|
|
void
|
2003-10-03 18:15:54 +00:00
|
|
|
rt_ifannouncemsg(struct ifnet *ifp, int what)
|
2002-01-18 14:33:04 +00:00
|
|
|
{
|
|
|
|
struct mbuf *m;
|
|
|
|
struct rt_addrinfo info;
|
|
|
|
|
2004-10-05 19:48:33 +00:00
|
|
|
m = rt_makeifannouncemsg(ifp, RTM_IFANNOUNCE, what, &info);
|
|
|
|
if (m != NULL)
|
|
|
|
rt_dispatch(m, NULL);
|
|
|
|
}
|
2002-01-18 14:33:04 +00:00
|
|
|
|
2003-10-03 18:15:54 +00:00
|
|
|
static void
|
2004-04-18 00:56:44 +00:00
|
|
|
rt_dispatch(struct mbuf *m, const struct sockaddr *sa)
|
2003-10-03 18:15:54 +00:00
|
|
|
{
|
Step 1.5 of importing the network stack virtualization infrastructure
from the vimage project, as per plan established at devsummit 08/08:
http://wiki.freebsd.org/Image/Notes200808DevSummit
Introduce INIT_VNET_*() initializer macros, VNET_FOREACH() iterator
macros, and CURVNET_SET() context setting macros, all currently
resolving to NOPs.
Prepare for virtualization of selected SYSCTL objects by introducing a
family of SYSCTL_V_*() macros, currently resolving to their global
counterparts, i.e. SYSCTL_V_INT() == SYSCTL_INT().
Move selected #defines from sys/sys/vimage.h to newly introduced header
files specific to virtualized subsystems (sys/net/vnet.h,
sys/netinet/vinet.h etc.).
All the changes are verified to have zero functional impact at this
point in time by doing MD5 comparision between pre- and post-change
object files(*).
(*) netipsec/keysock.c did not validate depending on compile time options.
Implemented by: julian, bz, brooks, zec
Reviewed by: julian, bz, brooks, kris, rwatson, ...
Approved by: julian (mentor)
Obtained from: //depot/projects/vimage-commit2/...
X-MFC after: never
Sponsored by: NLnet Foundation, The FreeBSD Foundation
2008-10-02 15:37:58 +00:00
|
|
|
INIT_VNET_NET(curvnet);
|
Introduce a netisr to deliver kernel-generated routing, avoiding
recursive entering of the socket code from the routing code:
- Modify rt_dispatch() to bundle up the sockaddr family, if any,
associated with a pending mbuf to dispatch to routing sockets, in
an m_tag on the mbuf.
- Allocate NETISR_ROUTE for use by routing sockets.
- Introduce rtsintrq, an ifqueue to be used by the netisr, and
introduce rts_input(), a function to unbundle the tagged sockaddr
and inject the mbuf and address into raw_input(), which previously
occurred in rt_dispatch().
- Introduce rts_init() to initialize rtsintrq, its mutex, and
register the netisr. Perform this at the same point in system
initialization as setup of the domains.
This change introduces asynchrony between the generation of a
pending routing socket message and delivery to sockets for use
by userspace. It avoids socket->routing->rtsock->socket use and
helps to avoid lock order reversals between the routing code and
socket code (in particular, raw socket control blocks), as route
locks are held over calls to rt_dispatch().
Reviewed by: "George V.Neville-Neil" <gnn@neville-neil.com>
Conceptual head nod by: sam
2004-06-09 02:48:23 +00:00
|
|
|
struct m_tag *tag;
|
2003-10-03 18:15:54 +00:00
|
|
|
|
Introduce a netisr to deliver kernel-generated routing, avoiding
recursive entering of the socket code from the routing code:
- Modify rt_dispatch() to bundle up the sockaddr family, if any,
associated with a pending mbuf to dispatch to routing sockets, in
an m_tag on the mbuf.
- Allocate NETISR_ROUTE for use by routing sockets.
- Introduce rtsintrq, an ifqueue to be used by the netisr, and
introduce rts_input(), a function to unbundle the tagged sockaddr
and inject the mbuf and address into raw_input(), which previously
occurred in rt_dispatch().
- Introduce rts_init() to initialize rtsintrq, its mutex, and
register the netisr. Perform this at the same point in system
initialization as setup of the domains.
This change introduces asynchrony between the generation of a
pending routing socket message and delivery to sockets for use
by userspace. It avoids socket->routing->rtsock->socket use and
helps to avoid lock order reversals between the routing code and
socket code (in particular, raw socket control blocks), as route
locks are held over calls to rt_dispatch().
Reviewed by: "George V.Neville-Neil" <gnn@neville-neil.com>
Conceptual head nod by: sam
2004-06-09 02:48:23 +00:00
|
|
|
/*
|
|
|
|
* Preserve the family from the sockaddr, if any, in an m_tag for
|
|
|
|
* use when injecting the mbuf into the routing socket buffer from
|
|
|
|
* the netisr.
|
|
|
|
*/
|
|
|
|
if (sa != NULL) {
|
|
|
|
tag = m_tag_get(PACKET_TAG_RTSOCKFAM, sizeof(unsigned short),
|
|
|
|
M_NOWAIT);
|
|
|
|
if (tag == NULL) {
|
|
|
|
m_freem(m);
|
|
|
|
return;
|
|
|
|
}
|
2005-02-23 22:50:19 +00:00
|
|
|
*(unsigned short *)(tag + 1) = sa->sa_family;
|
Introduce a netisr to deliver kernel-generated routing, avoiding
recursive entering of the socket code from the routing code:
- Modify rt_dispatch() to bundle up the sockaddr family, if any,
associated with a pending mbuf to dispatch to routing sockets, in
an m_tag on the mbuf.
- Allocate NETISR_ROUTE for use by routing sockets.
- Introduce rtsintrq, an ifqueue to be used by the netisr, and
introduce rts_input(), a function to unbundle the tagged sockaddr
and inject the mbuf and address into raw_input(), which previously
occurred in rt_dispatch().
- Introduce rts_init() to initialize rtsintrq, its mutex, and
register the netisr. Perform this at the same point in system
initialization as setup of the domains.
This change introduces asynchrony between the generation of a
pending routing socket message and delivery to sockets for use
by userspace. It avoids socket->routing->rtsock->socket use and
helps to avoid lock order reversals between the routing code and
socket code (in particular, raw socket control blocks), as route
locks are held over calls to rt_dispatch().
Reviewed by: "George V.Neville-Neil" <gnn@neville-neil.com>
Conceptual head nod by: sam
2004-06-09 02:48:23 +00:00
|
|
|
m_tag_prepend(m, tag);
|
|
|
|
}
|
2004-08-27 18:33:08 +00:00
|
|
|
netisr_queue(NETISR_ROUTE, m); /* mbuf is free'd on failure. */
|
2003-10-03 18:15:54 +00:00
|
|
|
}
|
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
/*
|
|
|
|
* This is used in dumping the kernel table via sysctl().
|
|
|
|
*/
|
2002-09-28 17:15:38 +00:00
|
|
|
static int
|
2003-10-03 18:15:54 +00:00
|
|
|
sysctl_dumpentry(struct radix_node *rn, void *vw)
|
1994-05-24 10:09:53 +00:00
|
|
|
{
|
2003-10-03 18:15:54 +00:00
|
|
|
struct walkarg *w = vw;
|
|
|
|
struct rtentry *rt = (struct rtentry *)rn;
|
1994-05-24 10:09:53 +00:00
|
|
|
int error = 0, size;
|
|
|
|
struct rt_addrinfo info;
|
|
|
|
|
|
|
|
if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg))
|
|
|
|
return 0;
|
2009-01-09 21:57:49 +00:00
|
|
|
if (jailed(w->w_req->td->td_ucred) &&
|
|
|
|
((rt->rt_flags & RTF_HOST) == 0 ||
|
|
|
|
!prison_if(w->w_req->td->td_ucred, rt_key(rt))))
|
|
|
|
return (0);
|
1994-05-24 10:09:53 +00:00
|
|
|
bzero((caddr_t)&info, sizeof(info));
|
2003-10-03 18:15:54 +00:00
|
|
|
info.rti_info[RTAX_DST] = rt_key(rt);
|
|
|
|
info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
|
|
|
|
info.rti_info[RTAX_NETMASK] = rt_mask(rt);
|
This main goals of this project are:
1. separating L2 tables (ARP, NDP) from the L3 routing tables
2. removing as much locking dependencies among these layers as
possible to allow for some parallelism in the search operations
3. simplify the logic in the routing code,
The most notable end result is the obsolescent of the route
cloning (RTF_CLONING) concept, which translated into code reduction
in both IPv4 ARP and IPv6 NDP related modules, and size reduction in
struct rtentry{}. The change in design obsoletes the semantics of
RTF_CLONING, RTF_WASCLONE and RTF_LLINFO routing flags. The userland
applications such as "arp" and "ndp" have been modified to reflect
those changes. The output from "netstat -r" shows only the routing
entries.
Quite a few developers have contributed to this project in the
past: Glebius Smirnoff, Luigi Rizzo, Alessandro Cerri, and
Andre Oppermann. And most recently:
- Kip Macy revised the locking code completely, thus completing
the last piece of the puzzle, Kip has also been conducting
active functional testing
- Sam Leffler has helped me improving/refactoring the code, and
provided valuable reviews
- Julian Elischer setup the perforce tree for me and has helped
me maintaining that branch before the svn conversion
2008-12-15 06:10:57 +00:00
|
|
|
info.rti_info[RTAX_GENMASK] = 0;
|
2001-10-17 11:23:59 +00:00
|
|
|
if (rt->rt_ifp) {
|
2005-11-11 16:04:59 +00:00
|
|
|
info.rti_info[RTAX_IFP] = rt->rt_ifp->if_addr->ifa_addr;
|
2003-10-03 18:15:54 +00:00
|
|
|
info.rti_info[RTAX_IFA] = rt->rt_ifa->ifa_addr;
|
2001-10-17 11:23:59 +00:00
|
|
|
if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
|
2003-10-03 18:15:54 +00:00
|
|
|
info.rti_info[RTAX_BRD] = rt->rt_ifa->ifa_dstaddr;
|
2001-10-17 11:23:59 +00:00
|
|
|
}
|
2004-04-17 15:09:36 +00:00
|
|
|
size = rt_msg2(RTM_GET, &info, NULL, w);
|
1995-11-16 19:00:27 +00:00
|
|
|
if (w->w_req && w->w_tmem) {
|
2003-10-03 18:15:54 +00:00
|
|
|
struct rt_msghdr *rtm = (struct rt_msghdr *)w->w_tmem;
|
1994-05-24 10:09:53 +00:00
|
|
|
|
|
|
|
rtm->rtm_flags = rt->rt_flags;
|
2003-11-20 20:07:39 +00:00
|
|
|
rtm->rtm_use = rt->rt_rmx.rmx_pksent;
|
|
|
|
rt_getmetrics(&rt->rt_rmx, &rtm->rtm_rmx);
|
1994-05-24 10:09:53 +00:00
|
|
|
rtm->rtm_index = rt->rt_ifp->if_index;
|
|
|
|
rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0;
|
|
|
|
rtm->rtm_addrs = info.rti_addrs;
|
1995-11-16 19:00:27 +00:00
|
|
|
error = SYSCTL_OUT(w->w_req, (caddr_t)rtm, size);
|
|
|
|
return (error);
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
|
|
|
return (error);
|
|
|
|
}
|
|
|
|
|
2002-09-28 17:15:38 +00:00
|
|
|
static int
|
2003-10-03 18:15:54 +00:00
|
|
|
sysctl_iflist(int af, struct walkarg *w)
|
1994-05-24 10:09:53 +00:00
|
|
|
{
|
Step 1.5 of importing the network stack virtualization infrastructure
from the vimage project, as per plan established at devsummit 08/08:
http://wiki.freebsd.org/Image/Notes200808DevSummit
Introduce INIT_VNET_*() initializer macros, VNET_FOREACH() iterator
macros, and CURVNET_SET() context setting macros, all currently
resolving to NOPs.
Prepare for virtualization of selected SYSCTL objects by introducing a
family of SYSCTL_V_*() macros, currently resolving to their global
counterparts, i.e. SYSCTL_V_INT() == SYSCTL_INT().
Move selected #defines from sys/sys/vimage.h to newly introduced header
files specific to virtualized subsystems (sys/net/vnet.h,
sys/netinet/vinet.h etc.).
All the changes are verified to have zero functional impact at this
point in time by doing MD5 comparision between pre- and post-change
object files(*).
(*) netipsec/keysock.c did not validate depending on compile time options.
Implemented by: julian, bz, brooks, zec
Reviewed by: julian, bz, brooks, kris, rwatson, ...
Approved by: julian (mentor)
Obtained from: //depot/projects/vimage-commit2/...
X-MFC after: never
Sponsored by: NLnet Foundation, The FreeBSD Foundation
2008-10-02 15:37:58 +00:00
|
|
|
INIT_VNET_NET(curvnet);
|
2003-10-03 18:15:54 +00:00
|
|
|
struct ifnet *ifp;
|
|
|
|
struct ifaddr *ifa;
|
|
|
|
struct rt_addrinfo info;
|
|
|
|
int len, error = 0;
|
1994-05-24 10:09:53 +00:00
|
|
|
|
|
|
|
bzero((caddr_t)&info, sizeof(info));
|
2005-09-10 15:12:24 +00:00
|
|
|
IFNET_RLOCK();
|
Commit step 1 of the vimage project, (network stack)
virtualization work done by Marko Zec (zec@).
This is the first in a series of commits over the course
of the next few weeks.
Mark all uses of global variables to be virtualized
with a V_ prefix.
Use macros to map them back to their global names for
now, so this is a NOP change only.
We hope to have caught at least 85-90% of what is needed
so we do not invalidate a lot of outstanding patches again.
Obtained from: //depot/projects/vimage-commit2/...
Reviewed by: brooks, des, ed, mav, julian,
jamie, kris, rwatson, zec, ...
(various people I forgot, different versions)
md5 (with a bit of help)
Sponsored by: NLnet Foundation, The FreeBSD Foundation
X-MFC after: never
V_Commit_Message_Reviewed_By: more people than the patch
2008-08-17 23:27:27 +00:00
|
|
|
TAILQ_FOREACH(ifp, &V_ifnet, if_link) {
|
1994-05-24 10:09:53 +00:00
|
|
|
if (w->w_arg && w->w_arg != ifp->if_index)
|
|
|
|
continue;
|
2005-11-11 16:04:59 +00:00
|
|
|
ifa = ifp->if_addr;
|
2003-10-03 18:15:54 +00:00
|
|
|
info.rti_info[RTAX_IFP] = ifa->ifa_addr;
|
2004-04-17 15:09:36 +00:00
|
|
|
len = rt_msg2(RTM_IFINFO, &info, NULL, w);
|
2004-04-18 00:56:44 +00:00
|
|
|
info.rti_info[RTAX_IFP] = NULL;
|
1995-11-16 19:00:27 +00:00
|
|
|
if (w->w_req && w->w_tmem) {
|
2003-10-03 18:15:54 +00:00
|
|
|
struct if_msghdr *ifm;
|
1994-05-24 10:09:53 +00:00
|
|
|
|
|
|
|
ifm = (struct if_msghdr *)w->w_tmem;
|
|
|
|
ifm->ifm_index = ifp->if_index;
|
Rename IFF_RUNNING to IFF_DRV_RUNNING, IFF_OACTIVE to IFF_DRV_OACTIVE,
and move both flags from ifnet.if_flags to ifnet.if_drv_flags, making
and documenting the locking of these flags the responsibility of the
device driver, not the network stack. The flags for these two fields
will be mutually exclusive so that they can be exposed to user space as
though they were stored in the same variable.
Provide #defines to provide the old names #ifndef _KERNEL, so that user
applications (such as ifconfig) can use the old flag names. Using the
old names in a device driver will result in a compile error in order to
help device driver writers adopt the new model.
When exposing the interface flags to user space, via interface ioctls
or routing sockets, or the two fields together. Since the driver flags
cannot currently be set for user space, no new logic is currently
required to handle this case.
Add some assertions that general purpose network stack routines, such
as if_setflags(), are not improperly used on driver-owned flags.
With this change, a large number of very minor network stack races are
closed, subject to correct device driver locking. Most were likely
never triggered.
Driver sweep to follow; many thanks to pjd and bz for the line-by-line
review they gave this patch.
Reviewed by: pjd, bz
MFC after: 7 days
2005-08-09 10:16:17 +00:00
|
|
|
ifm->ifm_flags = ifp->if_flags | ifp->if_drv_flags;
|
1994-05-24 10:09:53 +00:00
|
|
|
ifm->ifm_data = ifp->if_data;
|
|
|
|
ifm->ifm_addrs = info.rti_addrs;
|
1995-11-16 19:00:27 +00:00
|
|
|
error = SYSCTL_OUT(w->w_req,(caddr_t)ifm, len);
|
1994-10-08 22:38:27 +00:00
|
|
|
if (error)
|
2001-09-29 05:08:04 +00:00
|
|
|
goto done;
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
2004-04-18 00:56:44 +00:00
|
|
|
while ((ifa = TAILQ_NEXT(ifa, ifa_link)) != NULL) {
|
1994-05-24 10:09:53 +00:00
|
|
|
if (af && af != ifa->ifa_addr->sa_family)
|
|
|
|
continue;
|
2009-01-09 23:57:59 +00:00
|
|
|
if (jailed(w->w_req->td->td_ucred) &&
|
|
|
|
!prison_if(w->w_req->td->td_ucred, ifa->ifa_addr))
|
This Implements the mumbled about "Jail" feature.
This is a seriously beefed up chroot kind of thing. The process
is jailed along the same lines as a chroot does it, but with
additional tough restrictions imposed on what the superuser can do.
For all I know, it is safe to hand over the root bit inside a
prison to the customer living in that prison, this is what
it was developed for in fact: "real virtual servers".
Each prison has an ip number associated with it, which all IP
communications will be coerced to use and each prison has its own
hostname.
Needless to say, you need more RAM this way, but the advantage is
that each customer can run their own particular version of apache
and not stomp on the toes of their neighbors.
It generally does what one would expect, but setting up a jail
still takes a little knowledge.
A few notes:
I have no scripts for setting up a jail, don't ask me for them.
The IP number should be an alias on one of the interfaces.
mount a /proc in each jail, it will make ps more useable.
/proc/<pid>/status tells the hostname of the prison for
jailed processes.
Quotas are only sensible if you have a mountpoint per prison.
There are no privisions for stopping resource-hogging.
Some "#ifdef INET" and similar may be missing (send patches!)
If somebody wants to take it from here and develop it into
more of a "virtual machine" they should be most welcome!
Tools, comments, patches & documentation most welcome.
Have fun...
Sponsored by: http://www.rndassociates.com/
Run for almost a year by: http://www.servetheweb.com/
1999-04-28 11:38:52 +00:00
|
|
|
continue;
|
2003-10-03 18:15:54 +00:00
|
|
|
info.rti_info[RTAX_IFA] = ifa->ifa_addr;
|
|
|
|
info.rti_info[RTAX_NETMASK] = ifa->ifa_netmask;
|
|
|
|
info.rti_info[RTAX_BRD] = ifa->ifa_dstaddr;
|
2004-04-17 15:09:36 +00:00
|
|
|
len = rt_msg2(RTM_NEWADDR, &info, NULL, w);
|
1995-11-16 19:00:27 +00:00
|
|
|
if (w->w_req && w->w_tmem) {
|
2003-10-03 18:15:54 +00:00
|
|
|
struct ifa_msghdr *ifam;
|
1994-05-24 10:09:53 +00:00
|
|
|
|
|
|
|
ifam = (struct ifa_msghdr *)w->w_tmem;
|
|
|
|
ifam->ifam_index = ifa->ifa_ifp->if_index;
|
|
|
|
ifam->ifam_flags = ifa->ifa_flags;
|
|
|
|
ifam->ifam_metric = ifa->ifa_metric;
|
|
|
|
ifam->ifam_addrs = info.rti_addrs;
|
1995-11-16 19:00:27 +00:00
|
|
|
error = SYSCTL_OUT(w->w_req, w->w_tmem, len);
|
1994-10-08 22:38:27 +00:00
|
|
|
if (error)
|
2001-09-29 05:08:04 +00:00
|
|
|
goto done;
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
|
|
|
}
|
2003-10-03 18:15:54 +00:00
|
|
|
info.rti_info[RTAX_IFA] = info.rti_info[RTAX_NETMASK] =
|
2004-04-18 00:56:44 +00:00
|
|
|
info.rti_info[RTAX_BRD] = NULL;
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
2001-09-29 05:08:04 +00:00
|
|
|
done:
|
2005-09-10 15:12:24 +00:00
|
|
|
IFNET_RUNLOCK();
|
2001-09-29 05:08:04 +00:00
|
|
|
return (error);
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
|
|
|
|
2009-01-09 16:22:32 +00:00
|
|
|
static int
|
2004-04-16 08:14:34 +00:00
|
|
|
sysctl_ifmalist(int af, struct walkarg *w)
|
2003-11-14 18:48:15 +00:00
|
|
|
{
|
Step 1.5 of importing the network stack virtualization infrastructure
from the vimage project, as per plan established at devsummit 08/08:
http://wiki.freebsd.org/Image/Notes200808DevSummit
Introduce INIT_VNET_*() initializer macros, VNET_FOREACH() iterator
macros, and CURVNET_SET() context setting macros, all currently
resolving to NOPs.
Prepare for virtualization of selected SYSCTL objects by introducing a
family of SYSCTL_V_*() macros, currently resolving to their global
counterparts, i.e. SYSCTL_V_INT() == SYSCTL_INT().
Move selected #defines from sys/sys/vimage.h to newly introduced header
files specific to virtualized subsystems (sys/net/vnet.h,
sys/netinet/vinet.h etc.).
All the changes are verified to have zero functional impact at this
point in time by doing MD5 comparision between pre- and post-change
object files(*).
(*) netipsec/keysock.c did not validate depending on compile time options.
Implemented by: julian, bz, brooks, zec
Reviewed by: julian, bz, brooks, kris, rwatson, ...
Approved by: julian (mentor)
Obtained from: //depot/projects/vimage-commit2/...
X-MFC after: never
Sponsored by: NLnet Foundation, The FreeBSD Foundation
2008-10-02 15:37:58 +00:00
|
|
|
INIT_VNET_NET(curvnet);
|
2004-04-16 08:14:34 +00:00
|
|
|
struct ifnet *ifp;
|
2003-11-14 18:48:15 +00:00
|
|
|
struct ifmultiaddr *ifma;
|
|
|
|
struct rt_addrinfo info;
|
|
|
|
int len, error = 0;
|
2004-04-16 08:14:34 +00:00
|
|
|
struct ifaddr *ifa;
|
2003-11-14 18:48:15 +00:00
|
|
|
|
|
|
|
bzero((caddr_t)&info, sizeof(info));
|
2005-09-10 15:12:24 +00:00
|
|
|
IFNET_RLOCK();
|
Commit step 1 of the vimage project, (network stack)
virtualization work done by Marko Zec (zec@).
This is the first in a series of commits over the course
of the next few weeks.
Mark all uses of global variables to be virtualized
with a V_ prefix.
Use macros to map them back to their global names for
now, so this is a NOP change only.
We hope to have caught at least 85-90% of what is needed
so we do not invalidate a lot of outstanding patches again.
Obtained from: //depot/projects/vimage-commit2/...
Reviewed by: brooks, des, ed, mav, julian,
jamie, kris, rwatson, zec, ...
(various people I forgot, different versions)
md5 (with a bit of help)
Sponsored by: NLnet Foundation, The FreeBSD Foundation
X-MFC after: never
V_Commit_Message_Reviewed_By: more people than the patch
2008-08-17 23:27:27 +00:00
|
|
|
TAILQ_FOREACH(ifp, &V_ifnet, if_link) {
|
2003-11-14 18:48:15 +00:00
|
|
|
if (w->w_arg && w->w_arg != ifp->if_index)
|
|
|
|
continue;
|
2005-11-11 16:04:59 +00:00
|
|
|
ifa = ifp->if_addr;
|
2004-04-17 15:09:36 +00:00
|
|
|
info.rti_info[RTAX_IFP] = ifa ? ifa->ifa_addr : NULL;
|
2005-09-10 15:12:24 +00:00
|
|
|
IF_ADDR_LOCK(ifp);
|
2003-11-14 18:48:15 +00:00
|
|
|
TAILQ_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link) {
|
|
|
|
if (af && af != ifma->ifma_addr->sa_family)
|
|
|
|
continue;
|
2009-01-09 23:57:59 +00:00
|
|
|
if (jailed(w->w_req->td->td_ucred) &&
|
|
|
|
!prison_if(w->w_req->td->td_ucred, ifma->ifma_addr))
|
2003-11-14 18:48:15 +00:00
|
|
|
continue;
|
|
|
|
info.rti_info[RTAX_IFA] = ifma->ifma_addr;
|
2004-04-17 15:09:36 +00:00
|
|
|
info.rti_info[RTAX_GATEWAY] =
|
|
|
|
(ifma->ifma_addr->sa_family != AF_LINK) ?
|
|
|
|
ifma->ifma_lladdr : NULL;
|
|
|
|
len = rt_msg2(RTM_NEWMADDR, &info, NULL, w);
|
2003-11-14 18:48:15 +00:00
|
|
|
if (w->w_req && w->w_tmem) {
|
2004-04-16 08:14:34 +00:00
|
|
|
struct ifma_msghdr *ifmam;
|
2003-11-14 18:48:15 +00:00
|
|
|
|
|
|
|
ifmam = (struct ifma_msghdr *)w->w_tmem;
|
|
|
|
ifmam->ifmam_index = ifma->ifma_ifp->if_index;
|
|
|
|
ifmam->ifmam_flags = 0;
|
|
|
|
ifmam->ifmam_addrs = info.rti_addrs;
|
|
|
|
error = SYSCTL_OUT(w->w_req, w->w_tmem, len);
|
2005-09-10 15:12:24 +00:00
|
|
|
if (error) {
|
|
|
|
IF_ADDR_UNLOCK(ifp);
|
2003-11-14 18:48:15 +00:00
|
|
|
goto done;
|
2005-09-10 15:12:24 +00:00
|
|
|
}
|
2003-11-14 18:48:15 +00:00
|
|
|
}
|
|
|
|
}
|
2005-09-10 15:12:24 +00:00
|
|
|
IF_ADDR_UNLOCK(ifp);
|
2003-11-14 18:48:15 +00:00
|
|
|
}
|
|
|
|
done:
|
2005-09-10 15:12:24 +00:00
|
|
|
IFNET_RUNLOCK();
|
2003-11-14 18:48:15 +00:00
|
|
|
return (error);
|
|
|
|
}
|
|
|
|
|
1995-11-16 19:00:27 +00:00
|
|
|
static int
|
2000-07-04 11:25:35 +00:00
|
|
|
sysctl_rtsock(SYSCTL_HANDLER_ARGS)
|
1994-05-24 10:09:53 +00:00
|
|
|
{
|
Step 1.5 of importing the network stack virtualization infrastructure
from the vimage project, as per plan established at devsummit 08/08:
http://wiki.freebsd.org/Image/Notes200808DevSummit
Introduce INIT_VNET_*() initializer macros, VNET_FOREACH() iterator
macros, and CURVNET_SET() context setting macros, all currently
resolving to NOPs.
Prepare for virtualization of selected SYSCTL objects by introducing a
family of SYSCTL_V_*() macros, currently resolving to their global
counterparts, i.e. SYSCTL_V_INT() == SYSCTL_INT().
Move selected #defines from sys/sys/vimage.h to newly introduced header
files specific to virtualized subsystems (sys/net/vnet.h,
sys/netinet/vinet.h etc.).
All the changes are verified to have zero functional impact at this
point in time by doing MD5 comparision between pre- and post-change
object files(*).
(*) netipsec/keysock.c did not validate depending on compile time options.
Implemented by: julian, bz, brooks, zec
Reviewed by: julian, bz, brooks, kris, rwatson, ...
Approved by: julian (mentor)
Obtained from: //depot/projects/vimage-commit2/...
X-MFC after: never
Sponsored by: NLnet Foundation, The FreeBSD Foundation
2008-10-02 15:37:58 +00:00
|
|
|
INIT_VNET_NET(curvnet);
|
1995-11-16 19:00:27 +00:00
|
|
|
int *name = (int *)arg1;
|
|
|
|
u_int namelen = arg2;
|
2003-10-03 18:15:54 +00:00
|
|
|
struct radix_node_head *rnh;
|
2005-09-10 15:12:24 +00:00
|
|
|
int i, lim, error = EINVAL;
|
2004-04-12 20:26:01 +00:00
|
|
|
u_char af;
|
1994-05-24 10:09:53 +00:00
|
|
|
struct walkarg w;
|
|
|
|
|
1995-11-16 19:00:27 +00:00
|
|
|
name ++;
|
|
|
|
namelen--;
|
|
|
|
if (req->newptr)
|
1994-05-24 10:09:53 +00:00
|
|
|
return (EPERM);
|
|
|
|
if (namelen != 3)
|
2002-01-25 00:50:49 +00:00
|
|
|
return ((namelen < 3) ? EISDIR : ENOTDIR);
|
1994-05-24 10:09:53 +00:00
|
|
|
af = name[0];
|
2002-12-25 10:51:20 +00:00
|
|
|
if (af > AF_MAX)
|
|
|
|
return (EINVAL);
|
2004-04-18 11:47:04 +00:00
|
|
|
bzero(&w, sizeof(w));
|
1994-05-24 10:09:53 +00:00
|
|
|
w.w_op = name[1];
|
|
|
|
w.w_arg = name[2];
|
1995-11-16 19:00:27 +00:00
|
|
|
w.w_req = req;
|
1994-05-24 10:09:53 +00:00
|
|
|
|
2005-09-10 15:12:24 +00:00
|
|
|
error = sysctl_wire_old_buffer(req, 0);
|
|
|
|
if (error)
|
|
|
|
return (error);
|
1994-05-24 10:09:53 +00:00
|
|
|
switch (w.w_op) {
|
|
|
|
|
|
|
|
case NET_RT_DUMP:
|
|
|
|
case NET_RT_FLAGS:
|
2004-04-12 20:26:01 +00:00
|
|
|
if (af == 0) { /* dump all tables */
|
|
|
|
i = 1;
|
|
|
|
lim = AF_MAX;
|
|
|
|
} else /* dump only one table */
|
|
|
|
i = lim = af;
|
2008-12-26 19:45:24 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* take care of llinfo entries, the caller must
|
|
|
|
* specify an AF
|
|
|
|
*/
|
|
|
|
if (w.w_op == NET_RT_FLAGS && w.w_arg == 0) {
|
|
|
|
if (af != 0)
|
|
|
|
error = lltable_sysctl_dumparp(af, w.w_req);
|
|
|
|
else
|
|
|
|
error = EINVAL;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
/*
|
|
|
|
* take care of routing entries
|
|
|
|
*/
|
2004-04-12 20:26:01 +00:00
|
|
|
for (error = 0; error == 0 && i <= lim; i++)
|
2009-01-09 23:57:59 +00:00
|
|
|
if ((rnh = V_rt_tables[req->td->td_proc->p_fibnum][i]) != NULL) {
|
2005-09-10 15:12:24 +00:00
|
|
|
RADIX_NODE_HEAD_LOCK(rnh);
|
2002-12-24 03:03:39 +00:00
|
|
|
error = rnh->rnh_walktree(rnh,
|
2005-09-10 15:12:24 +00:00
|
|
|
sysctl_dumpentry, &w);
|
|
|
|
RADIX_NODE_HEAD_UNLOCK(rnh);
|
2004-04-12 20:26:01 +00:00
|
|
|
} else if (af != 0)
|
2002-12-24 03:03:39 +00:00
|
|
|
error = EAFNOSUPPORT;
|
1994-05-24 10:09:53 +00:00
|
|
|
break;
|
|
|
|
|
|
|
|
case NET_RT_IFLIST:
|
|
|
|
error = sysctl_iflist(af, &w);
|
2003-11-14 18:48:15 +00:00
|
|
|
break;
|
|
|
|
|
|
|
|
case NET_RT_IFMALIST:
|
|
|
|
error = sysctl_ifmalist(af, &w);
|
|
|
|
break;
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
|
|
|
if (w.w_tmem)
|
|
|
|
free(w.w_tmem, M_RTABLE);
|
|
|
|
return (error);
|
|
|
|
}
|
|
|
|
|
1997-04-27 20:01:29 +00:00
|
|
|
SYSCTL_NODE(_net, PF_ROUTE, routetable, CTLFLAG_RD, sysctl_rtsock, "");
|
1995-11-16 19:00:27 +00:00
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
/*
|
|
|
|
* Definitions of protocols supported in the ROUTE domain.
|
|
|
|
*/
|
|
|
|
|
2005-09-07 10:06:14 +00:00
|
|
|
static struct domain routedomain; /* or at least forward */
|
1994-05-24 10:09:53 +00:00
|
|
|
|
1995-11-16 19:00:27 +00:00
|
|
|
static struct protosw routesw[] = {
|
2005-11-09 13:29:16 +00:00
|
|
|
{
|
|
|
|
.pr_type = SOCK_RAW,
|
|
|
|
.pr_domain = &routedomain,
|
|
|
|
.pr_flags = PR_ATOMIC|PR_ADDR,
|
|
|
|
.pr_output = route_output,
|
|
|
|
.pr_ctlinput = raw_ctlinput,
|
|
|
|
.pr_init = raw_init,
|
|
|
|
.pr_usrreqs = &route_usrreqs
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
|
|
|
};
|
|
|
|
|
2005-11-09 13:29:16 +00:00
|
|
|
static struct domain routedomain = {
|
|
|
|
.dom_family = PF_ROUTE,
|
|
|
|
.dom_name = "route",
|
|
|
|
.dom_protosw = routesw,
|
|
|
|
.dom_protoswNPROTOSW = &routesw[sizeof(routesw)/sizeof(routesw[0])]
|
|
|
|
};
|
1995-05-10 16:48:52 +00:00
|
|
|
|
1995-05-11 00:13:26 +00:00
|
|
|
DOMAIN_SET(route);
|