NetBSD/sys/net/route.c

1388 lines
36 KiB
C
Raw Normal View History

/* $NetBSD: route.c,v 1.126 2012/01/30 20:01:08 christos Exp $ */
/*-
* Copyright (c) 1998, 2008 The NetBSD Foundation, Inc.
* All rights reserved.
*
* This code is derived from software contributed to The NetBSD Foundation
* by Kevin M. Lahey of the Numerical Aerospace Simulation Facility,
* NASA Ames Research Center.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
* ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
* TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
* PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
* BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*/
/*
* Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
* All rights reserved.
2005-02-27 01:45:09 +03:00
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. Neither the name of the project nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
2005-02-27 01:45:09 +03:00
*
* THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*/
1993-03-21 12:45:37 +03:00
/*
* Copyright (c) 1980, 1986, 1991, 1993
* The Regents of the University of California. All rights reserved.
1993-03-21 12:45:37 +03:00
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. Neither the name of the University nor the names of its contributors
1993-03-21 12:45:37 +03:00
* 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.
*
1997-04-03 01:17:28 +04:00
* @(#)route.c 8.3 (Berkeley) 1/9/95
1993-03-21 12:45:37 +03:00
*/
1993-03-21 21:04:42 +03:00
Eliminate address family-specific route caches (struct route, struct route_in6, struct route_iso), replacing all caches with a struct route. The principle benefit of this change is that all of the protocol families can benefit from route cache-invalidation, which is necessary for correct routing. Route-cache invalidation fixes an ancient PR, kern/3508, at long last; it fixes various other PRs, also. Discussions with and ideas from Joerg Sonnenberger influenced this work tremendously. Of course, all design oversights and bugs are mine. DETAILS 1 I added to each address family a pool of sockaddrs. I have introduced routines for allocating, copying, and duplicating, and freeing sockaddrs: struct sockaddr *sockaddr_alloc(sa_family_t af, int flags); struct sockaddr *sockaddr_copy(struct sockaddr *dst, const struct sockaddr *src); struct sockaddr *sockaddr_dup(const struct sockaddr *src, int flags); void sockaddr_free(struct sockaddr *sa); sockaddr_alloc() returns either a sockaddr from the pool belonging to the specified family, or NULL if the pool is exhausted. The returned sockaddr has the right size for that family; sa_family and sa_len fields are initialized to the family and sockaddr length---e.g., sa_family = AF_INET and sa_len = sizeof(struct sockaddr_in). sockaddr_free() puts the given sockaddr back into its family's pool. sockaddr_dup() and sockaddr_copy() work analogously to strdup() and strcpy(), respectively. sockaddr_copy() KASSERTs that the family of the destination and source sockaddrs are alike. The 'flags' argumet for sockaddr_alloc() and sockaddr_dup() is passed directly to pool_get(9). 2 I added routines for initializing sockaddrs in each address family, sockaddr_in_init(), sockaddr_in6_init(), sockaddr_iso_init(), etc. They are fairly self-explanatory. 3 structs route_in6 and route_iso are no more. All protocol families use struct route. I have changed the route cache, 'struct route', so that it does not contain storage space for a sockaddr. Instead, struct route points to a sockaddr coming from the pool the sockaddr belongs to. I added a new method to struct route, rtcache_setdst(), for setting the cache destination: int rtcache_setdst(struct route *, const struct sockaddr *); rtcache_setdst() returns 0 on success, or ENOMEM if no memory is available to create the sockaddr storage. It is now possible for rtcache_getdst() to return NULL if, say, rtcache_setdst() failed. I check the return value for NULL everywhere in the kernel. 4 Each routing domain (struct domain) has a list of live route caches, dom_rtcache. rtflushall(sa_family_t af) looks up the domain indicated by 'af', walks the domain's list of route caches and invalidates each one.
2007-05-03 00:40:22 +04:00
#include "opt_route.h"
2001-11-13 02:49:33 +03:00
#include <sys/cdefs.h>
__KERNEL_RCSID(0, "$NetBSD: route.c,v 1.126 2012/01/30 20:01:08 christos Exp $");
1998-07-05 10:49:00 +04:00
1993-12-18 03:40:47 +03:00
#include <sys/param.h>
#include <sys/kmem.h>
Eliminate address family-specific route caches (struct route, struct route_in6, struct route_iso), replacing all caches with a struct route. The principle benefit of this change is that all of the protocol families can benefit from route cache-invalidation, which is necessary for correct routing. Route-cache invalidation fixes an ancient PR, kern/3508, at long last; it fixes various other PRs, also. Discussions with and ideas from Joerg Sonnenberger influenced this work tremendously. Of course, all design oversights and bugs are mine. DETAILS 1 I added to each address family a pool of sockaddrs. I have introduced routines for allocating, copying, and duplicating, and freeing sockaddrs: struct sockaddr *sockaddr_alloc(sa_family_t af, int flags); struct sockaddr *sockaddr_copy(struct sockaddr *dst, const struct sockaddr *src); struct sockaddr *sockaddr_dup(const struct sockaddr *src, int flags); void sockaddr_free(struct sockaddr *sa); sockaddr_alloc() returns either a sockaddr from the pool belonging to the specified family, or NULL if the pool is exhausted. The returned sockaddr has the right size for that family; sa_family and sa_len fields are initialized to the family and sockaddr length---e.g., sa_family = AF_INET and sa_len = sizeof(struct sockaddr_in). sockaddr_free() puts the given sockaddr back into its family's pool. sockaddr_dup() and sockaddr_copy() work analogously to strdup() and strcpy(), respectively. sockaddr_copy() KASSERTs that the family of the destination and source sockaddrs are alike. The 'flags' argumet for sockaddr_alloc() and sockaddr_dup() is passed directly to pool_get(9). 2 I added routines for initializing sockaddrs in each address family, sockaddr_in_init(), sockaddr_in6_init(), sockaddr_iso_init(), etc. They are fairly self-explanatory. 3 structs route_in6 and route_iso are no more. All protocol families use struct route. I have changed the route cache, 'struct route', so that it does not contain storage space for a sockaddr. Instead, struct route points to a sockaddr coming from the pool the sockaddr belongs to. I added a new method to struct route, rtcache_setdst(), for setting the cache destination: int rtcache_setdst(struct route *, const struct sockaddr *); rtcache_setdst() returns 0 on success, or ENOMEM if no memory is available to create the sockaddr storage. It is now possible for rtcache_getdst() to return NULL if, say, rtcache_setdst() failed. I check the return value for NULL everywhere in the kernel. 4 Each routing domain (struct domain) has a list of live route caches, dom_rtcache. rtflushall(sa_family_t af) looks up the domain indicated by 'af', walks the domain's list of route caches and invalidates each one.
2007-05-03 00:40:22 +04:00
#include <sys/sysctl.h>
1993-12-18 03:40:47 +03:00
#include <sys/systm.h>
#include <sys/callout.h>
1993-12-18 03:40:47 +03:00
#include <sys/proc.h>
#include <sys/mbuf.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/domain.h>
#include <sys/protosw.h>
#include <sys/kernel.h>
1993-12-18 03:40:47 +03:00
#include <sys/ioctl.h>
#include <sys/pool.h>
#include <sys/kauth.h>
1993-03-21 12:45:37 +03:00
1993-12-18 03:40:47 +03:00
#include <net/if.h>
*** Summary *** When a link-layer address changes (e.g., ifconfig ex0 link 02:de:ad:be:ef:02 active), send a gratuitous ARP and/or a Neighbor Advertisement to update the network-/link-layer address bindings on our LAN peers. Refuse a change of ethernet address to the address 00:00:00:00:00:00 or to any multicast/broadcast address. (Thanks matt@.) Reorder ifnet ioctl operations so that driver ioctls may inherit the functions of their "class"---ether_ioctl(), fddi_ioctl(), et cetera---and the class ioctls may inherit from the generic ioctl, ifioctl_common(), but both driver- and class-ioctls may override the generic behavior. Make network drivers share more code. Distinguish a "factory" link-layer address from others for the purposes of both protecting that address from deletion and computing EUI64. Return consistent, appropriate error codes from network drivers. Improve readability. KNF. *** Details *** In if_attach(), always initialize the interface ioctl routine, ifnet->if_ioctl, if the driver has not already initialized it. Delete if_ioctl == NULL tests everywhere else, because it cannot happen. In the ioctl routines of network interfaces, inherit common ioctl behaviors by calling either ifioctl_common() or whichever ioctl routine is appropriate for the class of interface---e.g., ether_ioctl() for ethernets. Stop (ab)using SIOCSIFADDR and start to use SIOCINITIFADDR. In the user->kernel interface, SIOCSIFADDR's argument was an ifreq, but on the protocol->ifnet interface, SIOCSIFADDR's argument was an ifaddr. That was confusing, and it would work against me as I make it possible for a network interface to overload most ioctls. On the protocol->ifnet interface, replace SIOCSIFADDR with SIOCINITIFADDR. In ifioctl(), return EPERM if userland tries to invoke SIOCINITIFADDR. In ifioctl(), give the interface the first shot at handling most interface ioctls, and give the protocol the second shot, instead of the other way around. Finally, let compatibility code (COMPAT_OSOCK) take a shot. Pull device initialization out of switch statements under SIOCINITIFADDR. For example, pull ..._init() out of any switch statement that looks like this: switch (...->sa_family) { case ...: ..._init(); ... break; ... default: ..._init(); ... break; } Rewrite many if-else clauses that handle all permutations of IFF_UP and IFF_RUNNING to use a switch statement, switch (x & (IFF_UP|IFF_RUNNING)) { case 0: ... break; case IFF_RUNNING: ... break; case IFF_UP: ... break; case IFF_UP|IFF_RUNNING: ... break; } unifdef lots of code containing #ifdef FreeBSD, #ifdef NetBSD, and #ifdef SIOCSIFMTU, especially in fwip(4) and in ndis(4). In ipw(4), remove an if_set_sadl() call that is out of place. In nfe(4), reuse the jumbo MTU logic in ether_ioctl(). Let ethernets register a callback for setting h/w state such as promiscuous mode and the multicast filter in accord with a change in the if_flags: ether_set_ifflags_cb() registers a callback that returns ENETRESET if the caller should reset the ethernet by calling if_init(), 0 on success, != 0 on failure. Pull common code from ex(4), gem(4), nfe(4), sip(4), tlp(4), vge(4) into ether_ioctl(), and register if_flags callbacks for those drivers. Return ENOTTY instead of EINVAL for inappropriate ioctls. In zyd(4), use ENXIO instead of ENOTTY to indicate that the device is not any longer attached. Add to if_set_sadl() a boolean 'factory' argument that indicates whether a link-layer address was assigned by the factory or some other source. In a comment, recommend using the factory address for generating an EUI64, and update in6_get_hw_ifid() to prefer a factory address to any other link-layer address. Add a routing message, RTM_LLINFO_UPD, that tells protocols to update the binding of network-layer addresses to link-layer addresses. Implement this message in IPv4 and IPv6 by sending a gratuitous ARP or a neighbor advertisement, respectively. Generate RTM_LLINFO_UPD messages on a change of an interface's link-layer address. In ether_ioctl(), do not let SIOCALIFADDR set a link-layer address that is broadcast/multicast or equal to 00:00:00:00:00:00. Make ether_ioctl() call ifioctl_common() to handle ioctls that it does not understand. In gif(4), initialize if_softc and use it, instead of assuming that the gif_softc and ifp overlap. Let ifioctl_common() handle SIOCGIFADDR. Sprinkle rtcache_invariants(), which checks on DIAGNOSTIC kernels that certain invariants on a struct route are satisfied. In agr(4), rewrite agr_ioctl_filter() to be a bit more explicit about the ioctls that we do not allow on an agr(4) member interface. bzero -> memset. Delete unnecessary casts to void *. Use sockaddr_in_init() and sockaddr_in6_init(). Compare pointers with NULL instead of "testing truth". Replace some instances of (type *)0 with NULL. Change some K&R prototypes to ANSI C, and join lines.
2008-11-07 03:20:01 +03:00
#include <net/if_dl.h>
1993-12-18 03:40:47 +03:00
#include <net/route.h>
#include <net/raw_cb.h>
1993-03-21 12:45:37 +03:00
1993-12-18 03:40:47 +03:00
#include <netinet/in.h>
#include <netinet/in_var.h>
1993-03-21 12:45:37 +03:00
Eliminate address family-specific route caches (struct route, struct route_in6, struct route_iso), replacing all caches with a struct route. The principle benefit of this change is that all of the protocol families can benefit from route cache-invalidation, which is necessary for correct routing. Route-cache invalidation fixes an ancient PR, kern/3508, at long last; it fixes various other PRs, also. Discussions with and ideas from Joerg Sonnenberger influenced this work tremendously. Of course, all design oversights and bugs are mine. DETAILS 1 I added to each address family a pool of sockaddrs. I have introduced routines for allocating, copying, and duplicating, and freeing sockaddrs: struct sockaddr *sockaddr_alloc(sa_family_t af, int flags); struct sockaddr *sockaddr_copy(struct sockaddr *dst, const struct sockaddr *src); struct sockaddr *sockaddr_dup(const struct sockaddr *src, int flags); void sockaddr_free(struct sockaddr *sa); sockaddr_alloc() returns either a sockaddr from the pool belonging to the specified family, or NULL if the pool is exhausted. The returned sockaddr has the right size for that family; sa_family and sa_len fields are initialized to the family and sockaddr length---e.g., sa_family = AF_INET and sa_len = sizeof(struct sockaddr_in). sockaddr_free() puts the given sockaddr back into its family's pool. sockaddr_dup() and sockaddr_copy() work analogously to strdup() and strcpy(), respectively. sockaddr_copy() KASSERTs that the family of the destination and source sockaddrs are alike. The 'flags' argumet for sockaddr_alloc() and sockaddr_dup() is passed directly to pool_get(9). 2 I added routines for initializing sockaddrs in each address family, sockaddr_in_init(), sockaddr_in6_init(), sockaddr_iso_init(), etc. They are fairly self-explanatory. 3 structs route_in6 and route_iso are no more. All protocol families use struct route. I have changed the route cache, 'struct route', so that it does not contain storage space for a sockaddr. Instead, struct route points to a sockaddr coming from the pool the sockaddr belongs to. I added a new method to struct route, rtcache_setdst(), for setting the cache destination: int rtcache_setdst(struct route *, const struct sockaddr *); rtcache_setdst() returns 0 on success, or ENOMEM if no memory is available to create the sockaddr storage. It is now possible for rtcache_getdst() to return NULL if, say, rtcache_setdst() failed. I check the return value for NULL everywhere in the kernel. 4 Each routing domain (struct domain) has a list of live route caches, dom_rtcache. rtflushall(sa_family_t af) looks up the domain indicated by 'af', walks the domain's list of route caches and invalidates each one.
2007-05-03 00:40:22 +04:00
#ifdef RTFLUSH_DEBUG
#define rtcache_debug() __predict_false(_rtcache_debug)
#else /* RTFLUSH_DEBUG */
#define rtcache_debug() 0
#endif /* RTFLUSH_DEBUG */
1993-12-18 03:40:47 +03:00
2002-05-13 00:40:11 +04:00
struct rtstat rtstat;
1993-03-21 12:45:37 +03:00
int rttrash; /* routes not in table but not freed */
2008-10-04 04:09:34 +04:00
struct pool rtentry_pool;
struct pool rttimer_pool;
struct callout rt_timer_ch; /* callout for rt_timer_timer() */
Eliminate address family-specific route caches (struct route, struct route_in6, struct route_iso), replacing all caches with a struct route. The principle benefit of this change is that all of the protocol families can benefit from route cache-invalidation, which is necessary for correct routing. Route-cache invalidation fixes an ancient PR, kern/3508, at long last; it fixes various other PRs, also. Discussions with and ideas from Joerg Sonnenberger influenced this work tremendously. Of course, all design oversights and bugs are mine. DETAILS 1 I added to each address family a pool of sockaddrs. I have introduced routines for allocating, copying, and duplicating, and freeing sockaddrs: struct sockaddr *sockaddr_alloc(sa_family_t af, int flags); struct sockaddr *sockaddr_copy(struct sockaddr *dst, const struct sockaddr *src); struct sockaddr *sockaddr_dup(const struct sockaddr *src, int flags); void sockaddr_free(struct sockaddr *sa); sockaddr_alloc() returns either a sockaddr from the pool belonging to the specified family, or NULL if the pool is exhausted. The returned sockaddr has the right size for that family; sa_family and sa_len fields are initialized to the family and sockaddr length---e.g., sa_family = AF_INET and sa_len = sizeof(struct sockaddr_in). sockaddr_free() puts the given sockaddr back into its family's pool. sockaddr_dup() and sockaddr_copy() work analogously to strdup() and strcpy(), respectively. sockaddr_copy() KASSERTs that the family of the destination and source sockaddrs are alike. The 'flags' argumet for sockaddr_alloc() and sockaddr_dup() is passed directly to pool_get(9). 2 I added routines for initializing sockaddrs in each address family, sockaddr_in_init(), sockaddr_in6_init(), sockaddr_iso_init(), etc. They are fairly self-explanatory. 3 structs route_in6 and route_iso are no more. All protocol families use struct route. I have changed the route cache, 'struct route', so that it does not contain storage space for a sockaddr. Instead, struct route points to a sockaddr coming from the pool the sockaddr belongs to. I added a new method to struct route, rtcache_setdst(), for setting the cache destination: int rtcache_setdst(struct route *, const struct sockaddr *); rtcache_setdst() returns 0 on success, or ENOMEM if no memory is available to create the sockaddr storage. It is now possible for rtcache_getdst() to return NULL if, say, rtcache_setdst() failed. I check the return value for NULL everywhere in the kernel. 4 Each routing domain (struct domain) has a list of live route caches, dom_rtcache. rtflushall(sa_family_t af) looks up the domain indicated by 'af', walks the domain's list of route caches and invalidates each one.
2007-05-03 00:40:22 +04:00
#ifdef RTFLUSH_DEBUG
static int _rtcache_debug = 0;
#endif /* RTFLUSH_DEBUG */
static kauth_listener_t route_listener;
static int rtdeletemsg(struct rtentry *);
static int rtflushclone1(struct rtentry *, void *);
static void rtflushclone(sa_family_t family, struct rtentry *);
Eliminate address family-specific route caches (struct route, struct route_in6, struct route_iso), replacing all caches with a struct route. The principle benefit of this change is that all of the protocol families can benefit from route cache-invalidation, which is necessary for correct routing. Route-cache invalidation fixes an ancient PR, kern/3508, at long last; it fixes various other PRs, also. Discussions with and ideas from Joerg Sonnenberger influenced this work tremendously. Of course, all design oversights and bugs are mine. DETAILS 1 I added to each address family a pool of sockaddrs. I have introduced routines for allocating, copying, and duplicating, and freeing sockaddrs: struct sockaddr *sockaddr_alloc(sa_family_t af, int flags); struct sockaddr *sockaddr_copy(struct sockaddr *dst, const struct sockaddr *src); struct sockaddr *sockaddr_dup(const struct sockaddr *src, int flags); void sockaddr_free(struct sockaddr *sa); sockaddr_alloc() returns either a sockaddr from the pool belonging to the specified family, or NULL if the pool is exhausted. The returned sockaddr has the right size for that family; sa_family and sa_len fields are initialized to the family and sockaddr length---e.g., sa_family = AF_INET and sa_len = sizeof(struct sockaddr_in). sockaddr_free() puts the given sockaddr back into its family's pool. sockaddr_dup() and sockaddr_copy() work analogously to strdup() and strcpy(), respectively. sockaddr_copy() KASSERTs that the family of the destination and source sockaddrs are alike. The 'flags' argumet for sockaddr_alloc() and sockaddr_dup() is passed directly to pool_get(9). 2 I added routines for initializing sockaddrs in each address family, sockaddr_in_init(), sockaddr_in6_init(), sockaddr_iso_init(), etc. They are fairly self-explanatory. 3 structs route_in6 and route_iso are no more. All protocol families use struct route. I have changed the route cache, 'struct route', so that it does not contain storage space for a sockaddr. Instead, struct route points to a sockaddr coming from the pool the sockaddr belongs to. I added a new method to struct route, rtcache_setdst(), for setting the cache destination: int rtcache_setdst(struct route *, const struct sockaddr *); rtcache_setdst() returns 0 on success, or ENOMEM if no memory is available to create the sockaddr storage. It is now possible for rtcache_getdst() to return NULL if, say, rtcache_setdst() failed. I check the return value for NULL everywhere in the kernel. 4 Each routing domain (struct domain) has a list of live route caches, dom_rtcache. rtflushall(sa_family_t af) looks up the domain indicated by 'af', walks the domain's list of route caches and invalidates each one.
2007-05-03 00:40:22 +04:00
#ifdef RTFLUSH_DEBUG
static void sysctl_net_rtcache_setup(struct sysctllog **);
static void
sysctl_net_rtcache_setup(struct sysctllog **clog)
Eliminate address family-specific route caches (struct route, struct route_in6, struct route_iso), replacing all caches with a struct route. The principle benefit of this change is that all of the protocol families can benefit from route cache-invalidation, which is necessary for correct routing. Route-cache invalidation fixes an ancient PR, kern/3508, at long last; it fixes various other PRs, also. Discussions with and ideas from Joerg Sonnenberger influenced this work tremendously. Of course, all design oversights and bugs are mine. DETAILS 1 I added to each address family a pool of sockaddrs. I have introduced routines for allocating, copying, and duplicating, and freeing sockaddrs: struct sockaddr *sockaddr_alloc(sa_family_t af, int flags); struct sockaddr *sockaddr_copy(struct sockaddr *dst, const struct sockaddr *src); struct sockaddr *sockaddr_dup(const struct sockaddr *src, int flags); void sockaddr_free(struct sockaddr *sa); sockaddr_alloc() returns either a sockaddr from the pool belonging to the specified family, or NULL if the pool is exhausted. The returned sockaddr has the right size for that family; sa_family and sa_len fields are initialized to the family and sockaddr length---e.g., sa_family = AF_INET and sa_len = sizeof(struct sockaddr_in). sockaddr_free() puts the given sockaddr back into its family's pool. sockaddr_dup() and sockaddr_copy() work analogously to strdup() and strcpy(), respectively. sockaddr_copy() KASSERTs that the family of the destination and source sockaddrs are alike. The 'flags' argumet for sockaddr_alloc() and sockaddr_dup() is passed directly to pool_get(9). 2 I added routines for initializing sockaddrs in each address family, sockaddr_in_init(), sockaddr_in6_init(), sockaddr_iso_init(), etc. They are fairly self-explanatory. 3 structs route_in6 and route_iso are no more. All protocol families use struct route. I have changed the route cache, 'struct route', so that it does not contain storage space for a sockaddr. Instead, struct route points to a sockaddr coming from the pool the sockaddr belongs to. I added a new method to struct route, rtcache_setdst(), for setting the cache destination: int rtcache_setdst(struct route *, const struct sockaddr *); rtcache_setdst() returns 0 on success, or ENOMEM if no memory is available to create the sockaddr storage. It is now possible for rtcache_getdst() to return NULL if, say, rtcache_setdst() failed. I check the return value for NULL everywhere in the kernel. 4 Each routing domain (struct domain) has a list of live route caches, dom_rtcache. rtflushall(sa_family_t af) looks up the domain indicated by 'af', walks the domain's list of route caches and invalidates each one.
2007-05-03 00:40:22 +04:00
{
const struct sysctlnode *rnode;
/* XXX do not duplicate */
if (sysctl_createv(clog, 0, NULL, &rnode, CTLFLAG_PERMANENT,
CTLTYPE_NODE, "net", NULL, NULL, 0, NULL, 0, CTL_NET, CTL_EOL) != 0)
return;
if (sysctl_createv(clog, 0, &rnode, &rnode, CTLFLAG_PERMANENT,
CTLTYPE_NODE,
"rtcache", SYSCTL_DESCR("Route cache related settings"),
NULL, 0, NULL, 0, CTL_CREATE, CTL_EOL) != 0)
return;
if (sysctl_createv(clog, 0, &rnode, &rnode,
CTLFLAG_PERMANENT|CTLFLAG_READWRITE, CTLTYPE_INT,
"debug", SYSCTL_DESCR("Debug route caches"),
NULL, 0, &_rtcache_debug, 0, CTL_CREATE, CTL_EOL) != 0)
return;
}
#endif /* RTFLUSH_DEBUG */
struct ifaddr *
rt_get_ifa(struct rtentry *rt)
{
struct ifaddr *ifa;
if ((ifa = rt->rt_ifa) == NULL)
return ifa;
else if (ifa->ifa_getifa == NULL)
return ifa;
#if 0
else if (ifa->ifa_seqno != NULL && *ifa->ifa_seqno == rt->rt_ifa_seqno)
return ifa;
#endif
else {
Take steps to hide the radix_node implementation of the forwarding table from the forwarding table's users: Introduce rt_walktree() for walking the routing table and applying a function to each rtentry. Replace most rn_walktree() calls with it. Use rt_getkey()/rt_setkey() to get/set a route's destination. Keep a pointer to the sockaddr key in the rtentry, so that rtentry users do not have to grovel in the radix_node for the key. Add a RTM_GET method to rtrequest. Use that instead of radix_node lookups in, e.g., carp(4). Add sys/net/link_proto.c, which supplies sockaddr routines for link-layer socket addresses (sockaddr_dl). Cosmetic: Constify. KNF. Stop open-coding LIST_FOREACH, TAILQ_FOREACH, et cetera. Use NULL instead of 0 for null pointers. Use __arraycount(). Reduce gratuitous parenthesization. Stop using variadic arguments for rip6_output(), it is unnecessary. Remove the unnecessary rtentry member rt_genmask and the code to maintain it, since nothing actually used it. Make rt_maskedcopy() easier to read by using meaningful variable names. Extract a subroutine intern_netmask() for looking up a netmask in the masks table. Start converting backslash-ridden IPv6 macros in sys/netinet6/in6_var.h into inline subroutines that one can read without special eyeglasses. One functional change: when the kernel serves an RTM_GET, RTM_LOCK, or RTM_CHANGE request, it applies the netmask (if supplied) to a destination before searching for it in the forwarding table. I have changed sys/netinet/ip_carp.c, carp_setroute(), to remove the unlawful radix_node knowledge. Apart from the changes to carp(4), netiso, ATM, and strip(4), I have run the changes on three nodes in my wireless routing testbed, which involves IPv4 + IPv6 dynamic routing acrobatics, and it's working beautifully so far.
2007-07-20 00:48:52 +04:00
ifa = (*ifa->ifa_getifa)(ifa, rt_getkey(rt));
rt_replace_ifa(rt, ifa);
return ifa;
}
}
static void
rt_set_ifa1(struct rtentry *rt, struct ifaddr *ifa)
{
rt->rt_ifa = ifa;
if (ifa->ifa_seqno != NULL)
rt->rt_ifa_seqno = *ifa->ifa_seqno;
}
/*
* Is this route the connected route for the ifa?
*/
static int
rt_ifa_connected(const struct rtentry *rt, const struct ifaddr *ifa)
{
const struct sockaddr *key, *dst, *odst;
struct sockaddr_storage maskeddst;
key = rt_getkey(rt);
dst = rt->rt_flags & RTF_HOST ? ifa->ifa_dstaddr : ifa->ifa_addr;
if (dst == NULL ||
dst->sa_family != key->sa_family ||
dst->sa_len != key->sa_len)
return 0;
if ((rt->rt_flags & RTF_HOST) == 0 && ifa->ifa_netmask) {
odst = dst;
dst = (struct sockaddr *)&maskeddst;
rt_maskedcopy(odst, (struct sockaddr *)&maskeddst,
ifa->ifa_netmask);
}
return (memcmp(dst, key, dst->sa_len) == 0);
}
void
rt_replace_ifa(struct rtentry *rt, struct ifaddr *ifa)
{
if (rt->rt_ifa &&
rt->rt_ifa != ifa &&
rt->rt_ifa->ifa_flags & IFA_ROUTE &&
rt_ifa_connected(rt, rt->rt_ifa))
{
RT_DPRINTF("rt->_rt_key = %p, ifa = %p, "
"replace deleted IFA_ROUTE\n",
(void *)rt->_rt_key, (void *)rt->rt_ifa);
rt->rt_ifa->ifa_flags &= ~IFA_ROUTE;
if (rt_ifa_connected(rt, ifa)) {
RT_DPRINTF("rt->_rt_key = %p, ifa = %p, "
"replace added IFA_ROUTE\n",
(void *)rt->_rt_key, (void *)ifa);
ifa->ifa_flags |= IFA_ROUTE;
}
}
IFAREF(ifa);
IFAFREE(rt->rt_ifa);
rt_set_ifa1(rt, ifa);
}
static void
rt_set_ifa(struct rtentry *rt, struct ifaddr *ifa)
{
IFAREF(ifa);
rt_set_ifa1(rt, ifa);
}
static int
route_listener_cb(kauth_cred_t cred, kauth_action_t action, void *cookie,
void *arg0, void *arg1, void *arg2, void *arg3)
{
struct rt_msghdr *rtm;
int result;
result = KAUTH_RESULT_DEFER;
rtm = arg1;
if (action != KAUTH_NETWORK_ROUTE)
return result;
if (rtm->rtm_type == RTM_GET)
result = KAUTH_RESULT_ALLOW;
return result;
}
void
rt_init(void)
1993-03-21 12:45:37 +03:00
{
#ifdef RTFLUSH_DEBUG
sysctl_net_rtcache_setup(NULL);
#endif
2008-10-04 04:09:34 +04:00
pool_init(&rtentry_pool, sizeof(struct rtentry), 0, 0, 0, "rtentpl",
NULL, IPL_SOFTNET);
pool_init(&rttimer_pool, sizeof(struct rttimer), 0, 0, 0, "rttmrpl",
NULL, IPL_SOFTNET);
rn_init(); /* initialize all zeroes, all ones, mask table */
rtbl_init();
route_listener = kauth_listen_scope(KAUTH_SCOPE_NETWORK,
route_listener_cb, NULL);
1993-03-21 12:45:37 +03:00
}
Here are various changes designed to protect against bad IPv4 routing caused by stale route caches (struct route). Route caches are sprinkled throughout PCBs, the IP fast-forwarding table, and IP tunnel interfaces (gre, gif, stf). Stale IPv6 and ISO route caches will be treated by separate patches. Thank you to Christoph Badura for suggesting the general approach to invalidating route caches that I take here. Here are the details: Add hooks to struct domain for tracking and for invalidating each domain's route caches: dom_rtcache, dom_rtflush, and dom_rtflushall. Introduce helper subroutines, rtflush(ro) for invalidating a route cache, rtflushall(family) for invalidating all route caches in a routing domain, and rtcache(ro) for notifying the domain of a new cached route. Chain together all IPv4 route caches where ro_rt != NULL. Provide in_rtcache() for adding a route to the chain. Provide in_rtflush() and in_rtflushall() for invalidating IPv4 route caches. In in_rtflush(), set ro_rt to NULL, and remove the route from the chain. In in_rtflushall(), walk the chain and remove every route cache. In rtrequest1(), call rtflushall() to invalidate route caches when a route is added. In gif(4), discard the workaround for stale caches that involves expiring them every so often. Replace the pattern 'RTFREE(ro->ro_rt); ro->ro_rt = NULL;' with a call to rtflush(ro). Update ipflow_fastforward() and all other users of route caches so that they expect a cached route, ro->ro_rt, to turn to NULL. Take care when moving a 'struct route' to rtflush() the source and to rtcache() the destination. In domain initializers, use .dom_xxx tags. KNF here and there.
2006-12-09 08:33:04 +03:00
void
rtflushall(int family)
{
Eliminate address family-specific route caches (struct route, struct route_in6, struct route_iso), replacing all caches with a struct route. The principle benefit of this change is that all of the protocol families can benefit from route cache-invalidation, which is necessary for correct routing. Route-cache invalidation fixes an ancient PR, kern/3508, at long last; it fixes various other PRs, also. Discussions with and ideas from Joerg Sonnenberger influenced this work tremendously. Of course, all design oversights and bugs are mine. DETAILS 1 I added to each address family a pool of sockaddrs. I have introduced routines for allocating, copying, and duplicating, and freeing sockaddrs: struct sockaddr *sockaddr_alloc(sa_family_t af, int flags); struct sockaddr *sockaddr_copy(struct sockaddr *dst, const struct sockaddr *src); struct sockaddr *sockaddr_dup(const struct sockaddr *src, int flags); void sockaddr_free(struct sockaddr *sa); sockaddr_alloc() returns either a sockaddr from the pool belonging to the specified family, or NULL if the pool is exhausted. The returned sockaddr has the right size for that family; sa_family and sa_len fields are initialized to the family and sockaddr length---e.g., sa_family = AF_INET and sa_len = sizeof(struct sockaddr_in). sockaddr_free() puts the given sockaddr back into its family's pool. sockaddr_dup() and sockaddr_copy() work analogously to strdup() and strcpy(), respectively. sockaddr_copy() KASSERTs that the family of the destination and source sockaddrs are alike. The 'flags' argumet for sockaddr_alloc() and sockaddr_dup() is passed directly to pool_get(9). 2 I added routines for initializing sockaddrs in each address family, sockaddr_in_init(), sockaddr_in6_init(), sockaddr_iso_init(), etc. They are fairly self-explanatory. 3 structs route_in6 and route_iso are no more. All protocol families use struct route. I have changed the route cache, 'struct route', so that it does not contain storage space for a sockaddr. Instead, struct route points to a sockaddr coming from the pool the sockaddr belongs to. I added a new method to struct route, rtcache_setdst(), for setting the cache destination: int rtcache_setdst(struct route *, const struct sockaddr *); rtcache_setdst() returns 0 on success, or ENOMEM if no memory is available to create the sockaddr storage. It is now possible for rtcache_getdst() to return NULL if, say, rtcache_setdst() failed. I check the return value for NULL everywhere in the kernel. 4 Each routing domain (struct domain) has a list of live route caches, dom_rtcache. rtflushall(sa_family_t af) looks up the domain indicated by 'af', walks the domain's list of route caches and invalidates each one.
2007-05-03 00:40:22 +04:00
struct domain *dom;
if (rtcache_debug())
printf("%s: enter\n", __func__);
if ((dom = pffinddomain(family)) == NULL)
return;
Here are various changes designed to protect against bad IPv4 routing caused by stale route caches (struct route). Route caches are sprinkled throughout PCBs, the IP fast-forwarding table, and IP tunnel interfaces (gre, gif, stf). Stale IPv6 and ISO route caches will be treated by separate patches. Thank you to Christoph Badura for suggesting the general approach to invalidating route caches that I take here. Here are the details: Add hooks to struct domain for tracking and for invalidating each domain's route caches: dom_rtcache, dom_rtflush, and dom_rtflushall. Introduce helper subroutines, rtflush(ro) for invalidating a route cache, rtflushall(family) for invalidating all route caches in a routing domain, and rtcache(ro) for notifying the domain of a new cached route. Chain together all IPv4 route caches where ro_rt != NULL. Provide in_rtcache() for adding a route to the chain. Provide in_rtflush() and in_rtflushall() for invalidating IPv4 route caches. In in_rtflush(), set ro_rt to NULL, and remove the route from the chain. In in_rtflushall(), walk the chain and remove every route cache. In rtrequest1(), call rtflushall() to invalidate route caches when a route is added. In gif(4), discard the workaround for stale caches that involves expiring them every so often. Replace the pattern 'RTFREE(ro->ro_rt); ro->ro_rt = NULL;' with a call to rtflush(ro). Update ipflow_fastforward() and all other users of route caches so that they expect a cached route, ro->ro_rt, to turn to NULL. Take care when moving a 'struct route' to rtflush() the source and to rtcache() the destination. In domain initializers, use .dom_xxx tags. KNF here and there.
2006-12-09 08:33:04 +03:00
rtcache_invalidate(&dom->dom_rtcache);
Here are various changes designed to protect against bad IPv4 routing caused by stale route caches (struct route). Route caches are sprinkled throughout PCBs, the IP fast-forwarding table, and IP tunnel interfaces (gre, gif, stf). Stale IPv6 and ISO route caches will be treated by separate patches. Thank you to Christoph Badura for suggesting the general approach to invalidating route caches that I take here. Here are the details: Add hooks to struct domain for tracking and for invalidating each domain's route caches: dom_rtcache, dom_rtflush, and dom_rtflushall. Introduce helper subroutines, rtflush(ro) for invalidating a route cache, rtflushall(family) for invalidating all route caches in a routing domain, and rtcache(ro) for notifying the domain of a new cached route. Chain together all IPv4 route caches where ro_rt != NULL. Provide in_rtcache() for adding a route to the chain. Provide in_rtflush() and in_rtflushall() for invalidating IPv4 route caches. In in_rtflush(), set ro_rt to NULL, and remove the route from the chain. In in_rtflushall(), walk the chain and remove every route cache. In rtrequest1(), call rtflushall() to invalidate route caches when a route is added. In gif(4), discard the workaround for stale caches that involves expiring them every so often. Replace the pattern 'RTFREE(ro->ro_rt); ro->ro_rt = NULL;' with a call to rtflush(ro). Update ipflow_fastforward() and all other users of route caches so that they expect a cached route, ro->ro_rt, to turn to NULL. Take care when moving a 'struct route' to rtflush() the source and to rtcache() the destination. In domain initializers, use .dom_xxx tags. KNF here and there.
2006-12-09 08:33:04 +03:00
}
void
rtcache(struct route *ro)
{
Eliminate address family-specific route caches (struct route, struct route_in6, struct route_iso), replacing all caches with a struct route. The principle benefit of this change is that all of the protocol families can benefit from route cache-invalidation, which is necessary for correct routing. Route-cache invalidation fixes an ancient PR, kern/3508, at long last; it fixes various other PRs, also. Discussions with and ideas from Joerg Sonnenberger influenced this work tremendously. Of course, all design oversights and bugs are mine. DETAILS 1 I added to each address family a pool of sockaddrs. I have introduced routines for allocating, copying, and duplicating, and freeing sockaddrs: struct sockaddr *sockaddr_alloc(sa_family_t af, int flags); struct sockaddr *sockaddr_copy(struct sockaddr *dst, const struct sockaddr *src); struct sockaddr *sockaddr_dup(const struct sockaddr *src, int flags); void sockaddr_free(struct sockaddr *sa); sockaddr_alloc() returns either a sockaddr from the pool belonging to the specified family, or NULL if the pool is exhausted. The returned sockaddr has the right size for that family; sa_family and sa_len fields are initialized to the family and sockaddr length---e.g., sa_family = AF_INET and sa_len = sizeof(struct sockaddr_in). sockaddr_free() puts the given sockaddr back into its family's pool. sockaddr_dup() and sockaddr_copy() work analogously to strdup() and strcpy(), respectively. sockaddr_copy() KASSERTs that the family of the destination and source sockaddrs are alike. The 'flags' argumet for sockaddr_alloc() and sockaddr_dup() is passed directly to pool_get(9). 2 I added routines for initializing sockaddrs in each address family, sockaddr_in_init(), sockaddr_in6_init(), sockaddr_iso_init(), etc. They are fairly self-explanatory. 3 structs route_in6 and route_iso are no more. All protocol families use struct route. I have changed the route cache, 'struct route', so that it does not contain storage space for a sockaddr. Instead, struct route points to a sockaddr coming from the pool the sockaddr belongs to. I added a new method to struct route, rtcache_setdst(), for setting the cache destination: int rtcache_setdst(struct route *, const struct sockaddr *); rtcache_setdst() returns 0 on success, or ENOMEM if no memory is available to create the sockaddr storage. It is now possible for rtcache_getdst() to return NULL if, say, rtcache_setdst() failed. I check the return value for NULL everywhere in the kernel. 4 Each routing domain (struct domain) has a list of live route caches, dom_rtcache. rtflushall(sa_family_t af) looks up the domain indicated by 'af', walks the domain's list of route caches and invalidates each one.
2007-05-03 00:40:22 +04:00
struct domain *dom;
Here are various changes designed to protect against bad IPv4 routing caused by stale route caches (struct route). Route caches are sprinkled throughout PCBs, the IP fast-forwarding table, and IP tunnel interfaces (gre, gif, stf). Stale IPv6 and ISO route caches will be treated by separate patches. Thank you to Christoph Badura for suggesting the general approach to invalidating route caches that I take here. Here are the details: Add hooks to struct domain for tracking and for invalidating each domain's route caches: dom_rtcache, dom_rtflush, and dom_rtflushall. Introduce helper subroutines, rtflush(ro) for invalidating a route cache, rtflushall(family) for invalidating all route caches in a routing domain, and rtcache(ro) for notifying the domain of a new cached route. Chain together all IPv4 route caches where ro_rt != NULL. Provide in_rtcache() for adding a route to the chain. Provide in_rtflush() and in_rtflushall() for invalidating IPv4 route caches. In in_rtflush(), set ro_rt to NULL, and remove the route from the chain. In in_rtflushall(), walk the chain and remove every route cache. In rtrequest1(), call rtflushall() to invalidate route caches when a route is added. In gif(4), discard the workaround for stale caches that involves expiring them every so often. Replace the pattern 'RTFREE(ro->ro_rt); ro->ro_rt = NULL;' with a call to rtflush(ro). Update ipflow_fastforward() and all other users of route caches so that they expect a cached route, ro->ro_rt, to turn to NULL. Take care when moving a 'struct route' to rtflush() the source and to rtcache() the destination. In domain initializers, use .dom_xxx tags. KNF here and there.
2006-12-09 08:33:04 +03:00
*** Summary *** When a link-layer address changes (e.g., ifconfig ex0 link 02:de:ad:be:ef:02 active), send a gratuitous ARP and/or a Neighbor Advertisement to update the network-/link-layer address bindings on our LAN peers. Refuse a change of ethernet address to the address 00:00:00:00:00:00 or to any multicast/broadcast address. (Thanks matt@.) Reorder ifnet ioctl operations so that driver ioctls may inherit the functions of their "class"---ether_ioctl(), fddi_ioctl(), et cetera---and the class ioctls may inherit from the generic ioctl, ifioctl_common(), but both driver- and class-ioctls may override the generic behavior. Make network drivers share more code. Distinguish a "factory" link-layer address from others for the purposes of both protecting that address from deletion and computing EUI64. Return consistent, appropriate error codes from network drivers. Improve readability. KNF. *** Details *** In if_attach(), always initialize the interface ioctl routine, ifnet->if_ioctl, if the driver has not already initialized it. Delete if_ioctl == NULL tests everywhere else, because it cannot happen. In the ioctl routines of network interfaces, inherit common ioctl behaviors by calling either ifioctl_common() or whichever ioctl routine is appropriate for the class of interface---e.g., ether_ioctl() for ethernets. Stop (ab)using SIOCSIFADDR and start to use SIOCINITIFADDR. In the user->kernel interface, SIOCSIFADDR's argument was an ifreq, but on the protocol->ifnet interface, SIOCSIFADDR's argument was an ifaddr. That was confusing, and it would work against me as I make it possible for a network interface to overload most ioctls. On the protocol->ifnet interface, replace SIOCSIFADDR with SIOCINITIFADDR. In ifioctl(), return EPERM if userland tries to invoke SIOCINITIFADDR. In ifioctl(), give the interface the first shot at handling most interface ioctls, and give the protocol the second shot, instead of the other way around. Finally, let compatibility code (COMPAT_OSOCK) take a shot. Pull device initialization out of switch statements under SIOCINITIFADDR. For example, pull ..._init() out of any switch statement that looks like this: switch (...->sa_family) { case ...: ..._init(); ... break; ... default: ..._init(); ... break; } Rewrite many if-else clauses that handle all permutations of IFF_UP and IFF_RUNNING to use a switch statement, switch (x & (IFF_UP|IFF_RUNNING)) { case 0: ... break; case IFF_RUNNING: ... break; case IFF_UP: ... break; case IFF_UP|IFF_RUNNING: ... break; } unifdef lots of code containing #ifdef FreeBSD, #ifdef NetBSD, and #ifdef SIOCSIFMTU, especially in fwip(4) and in ndis(4). In ipw(4), remove an if_set_sadl() call that is out of place. In nfe(4), reuse the jumbo MTU logic in ether_ioctl(). Let ethernets register a callback for setting h/w state such as promiscuous mode and the multicast filter in accord with a change in the if_flags: ether_set_ifflags_cb() registers a callback that returns ENETRESET if the caller should reset the ethernet by calling if_init(), 0 on success, != 0 on failure. Pull common code from ex(4), gem(4), nfe(4), sip(4), tlp(4), vge(4) into ether_ioctl(), and register if_flags callbacks for those drivers. Return ENOTTY instead of EINVAL for inappropriate ioctls. In zyd(4), use ENXIO instead of ENOTTY to indicate that the device is not any longer attached. Add to if_set_sadl() a boolean 'factory' argument that indicates whether a link-layer address was assigned by the factory or some other source. In a comment, recommend using the factory address for generating an EUI64, and update in6_get_hw_ifid() to prefer a factory address to any other link-layer address. Add a routing message, RTM_LLINFO_UPD, that tells protocols to update the binding of network-layer addresses to link-layer addresses. Implement this message in IPv4 and IPv6 by sending a gratuitous ARP or a neighbor advertisement, respectively. Generate RTM_LLINFO_UPD messages on a change of an interface's link-layer address. In ether_ioctl(), do not let SIOCALIFADDR set a link-layer address that is broadcast/multicast or equal to 00:00:00:00:00:00. Make ether_ioctl() call ifioctl_common() to handle ioctls that it does not understand. In gif(4), initialize if_softc and use it, instead of assuming that the gif_softc and ifp overlap. Let ifioctl_common() handle SIOCGIFADDR. Sprinkle rtcache_invariants(), which checks on DIAGNOSTIC kernels that certain invariants on a struct route are satisfied. In agr(4), rewrite agr_ioctl_filter() to be a bit more explicit about the ioctls that we do not allow on an agr(4) member interface. bzero -> memset. Delete unnecessary casts to void *. Use sockaddr_in_init() and sockaddr_in6_init(). Compare pointers with NULL instead of "testing truth". Replace some instances of (type *)0 with NULL. Change some K&R prototypes to ANSI C, and join lines.
2008-11-07 03:20:01 +03:00
rtcache_invariants(ro);
KASSERT(ro->_ro_rt != NULL);
KASSERT(ro->ro_invalid == false);
Eliminate address family-specific route caches (struct route, struct route_in6, struct route_iso), replacing all caches with a struct route. The principle benefit of this change is that all of the protocol families can benefit from route cache-invalidation, which is necessary for correct routing. Route-cache invalidation fixes an ancient PR, kern/3508, at long last; it fixes various other PRs, also. Discussions with and ideas from Joerg Sonnenberger influenced this work tremendously. Of course, all design oversights and bugs are mine. DETAILS 1 I added to each address family a pool of sockaddrs. I have introduced routines for allocating, copying, and duplicating, and freeing sockaddrs: struct sockaddr *sockaddr_alloc(sa_family_t af, int flags); struct sockaddr *sockaddr_copy(struct sockaddr *dst, const struct sockaddr *src); struct sockaddr *sockaddr_dup(const struct sockaddr *src, int flags); void sockaddr_free(struct sockaddr *sa); sockaddr_alloc() returns either a sockaddr from the pool belonging to the specified family, or NULL if the pool is exhausted. The returned sockaddr has the right size for that family; sa_family and sa_len fields are initialized to the family and sockaddr length---e.g., sa_family = AF_INET and sa_len = sizeof(struct sockaddr_in). sockaddr_free() puts the given sockaddr back into its family's pool. sockaddr_dup() and sockaddr_copy() work analogously to strdup() and strcpy(), respectively. sockaddr_copy() KASSERTs that the family of the destination and source sockaddrs are alike. The 'flags' argumet for sockaddr_alloc() and sockaddr_dup() is passed directly to pool_get(9). 2 I added routines for initializing sockaddrs in each address family, sockaddr_in_init(), sockaddr_in6_init(), sockaddr_iso_init(), etc. They are fairly self-explanatory. 3 structs route_in6 and route_iso are no more. All protocol families use struct route. I have changed the route cache, 'struct route', so that it does not contain storage space for a sockaddr. Instead, struct route points to a sockaddr coming from the pool the sockaddr belongs to. I added a new method to struct route, rtcache_setdst(), for setting the cache destination: int rtcache_setdst(struct route *, const struct sockaddr *); rtcache_setdst() returns 0 on success, or ENOMEM if no memory is available to create the sockaddr storage. It is now possible for rtcache_getdst() to return NULL if, say, rtcache_setdst() failed. I check the return value for NULL everywhere in the kernel. 4 Each routing domain (struct domain) has a list of live route caches, dom_rtcache. rtflushall(sa_family_t af) looks up the domain indicated by 'af', walks the domain's list of route caches and invalidates each one.
2007-05-03 00:40:22 +04:00
KASSERT(rtcache_getdst(ro) != NULL);
Here are various changes designed to protect against bad IPv4 routing caused by stale route caches (struct route). Route caches are sprinkled throughout PCBs, the IP fast-forwarding table, and IP tunnel interfaces (gre, gif, stf). Stale IPv6 and ISO route caches will be treated by separate patches. Thank you to Christoph Badura for suggesting the general approach to invalidating route caches that I take here. Here are the details: Add hooks to struct domain for tracking and for invalidating each domain's route caches: dom_rtcache, dom_rtflush, and dom_rtflushall. Introduce helper subroutines, rtflush(ro) for invalidating a route cache, rtflushall(family) for invalidating all route caches in a routing domain, and rtcache(ro) for notifying the domain of a new cached route. Chain together all IPv4 route caches where ro_rt != NULL. Provide in_rtcache() for adding a route to the chain. Provide in_rtflush() and in_rtflushall() for invalidating IPv4 route caches. In in_rtflush(), set ro_rt to NULL, and remove the route from the chain. In in_rtflushall(), walk the chain and remove every route cache. In rtrequest1(), call rtflushall() to invalidate route caches when a route is added. In gif(4), discard the workaround for stale caches that involves expiring them every so often. Replace the pattern 'RTFREE(ro->ro_rt); ro->ro_rt = NULL;' with a call to rtflush(ro). Update ipflow_fastforward() and all other users of route caches so that they expect a cached route, ro->ro_rt, to turn to NULL. Take care when moving a 'struct route' to rtflush() the source and to rtcache() the destination. In domain initializers, use .dom_xxx tags. KNF here and there.
2006-12-09 08:33:04 +03:00
Eliminate address family-specific route caches (struct route, struct route_in6, struct route_iso), replacing all caches with a struct route. The principle benefit of this change is that all of the protocol families can benefit from route cache-invalidation, which is necessary for correct routing. Route-cache invalidation fixes an ancient PR, kern/3508, at long last; it fixes various other PRs, also. Discussions with and ideas from Joerg Sonnenberger influenced this work tremendously. Of course, all design oversights and bugs are mine. DETAILS 1 I added to each address family a pool of sockaddrs. I have introduced routines for allocating, copying, and duplicating, and freeing sockaddrs: struct sockaddr *sockaddr_alloc(sa_family_t af, int flags); struct sockaddr *sockaddr_copy(struct sockaddr *dst, const struct sockaddr *src); struct sockaddr *sockaddr_dup(const struct sockaddr *src, int flags); void sockaddr_free(struct sockaddr *sa); sockaddr_alloc() returns either a sockaddr from the pool belonging to the specified family, or NULL if the pool is exhausted. The returned sockaddr has the right size for that family; sa_family and sa_len fields are initialized to the family and sockaddr length---e.g., sa_family = AF_INET and sa_len = sizeof(struct sockaddr_in). sockaddr_free() puts the given sockaddr back into its family's pool. sockaddr_dup() and sockaddr_copy() work analogously to strdup() and strcpy(), respectively. sockaddr_copy() KASSERTs that the family of the destination and source sockaddrs are alike. The 'flags' argumet for sockaddr_alloc() and sockaddr_dup() is passed directly to pool_get(9). 2 I added routines for initializing sockaddrs in each address family, sockaddr_in_init(), sockaddr_in6_init(), sockaddr_iso_init(), etc. They are fairly self-explanatory. 3 structs route_in6 and route_iso are no more. All protocol families use struct route. I have changed the route cache, 'struct route', so that it does not contain storage space for a sockaddr. Instead, struct route points to a sockaddr coming from the pool the sockaddr belongs to. I added a new method to struct route, rtcache_setdst(), for setting the cache destination: int rtcache_setdst(struct route *, const struct sockaddr *); rtcache_setdst() returns 0 on success, or ENOMEM if no memory is available to create the sockaddr storage. It is now possible for rtcache_getdst() to return NULL if, say, rtcache_setdst() failed. I check the return value for NULL everywhere in the kernel. 4 Each routing domain (struct domain) has a list of live route caches, dom_rtcache. rtflushall(sa_family_t af) looks up the domain indicated by 'af', walks the domain's list of route caches and invalidates each one.
2007-05-03 00:40:22 +04:00
if ((dom = pffinddomain(rtcache_getdst(ro)->sa_family)) == NULL)
return;
LIST_INSERT_HEAD(&dom->dom_rtcache, ro, ro_rtcache_next);
*** Summary *** When a link-layer address changes (e.g., ifconfig ex0 link 02:de:ad:be:ef:02 active), send a gratuitous ARP and/or a Neighbor Advertisement to update the network-/link-layer address bindings on our LAN peers. Refuse a change of ethernet address to the address 00:00:00:00:00:00 or to any multicast/broadcast address. (Thanks matt@.) Reorder ifnet ioctl operations so that driver ioctls may inherit the functions of their "class"---ether_ioctl(), fddi_ioctl(), et cetera---and the class ioctls may inherit from the generic ioctl, ifioctl_common(), but both driver- and class-ioctls may override the generic behavior. Make network drivers share more code. Distinguish a "factory" link-layer address from others for the purposes of both protecting that address from deletion and computing EUI64. Return consistent, appropriate error codes from network drivers. Improve readability. KNF. *** Details *** In if_attach(), always initialize the interface ioctl routine, ifnet->if_ioctl, if the driver has not already initialized it. Delete if_ioctl == NULL tests everywhere else, because it cannot happen. In the ioctl routines of network interfaces, inherit common ioctl behaviors by calling either ifioctl_common() or whichever ioctl routine is appropriate for the class of interface---e.g., ether_ioctl() for ethernets. Stop (ab)using SIOCSIFADDR and start to use SIOCINITIFADDR. In the user->kernel interface, SIOCSIFADDR's argument was an ifreq, but on the protocol->ifnet interface, SIOCSIFADDR's argument was an ifaddr. That was confusing, and it would work against me as I make it possible for a network interface to overload most ioctls. On the protocol->ifnet interface, replace SIOCSIFADDR with SIOCINITIFADDR. In ifioctl(), return EPERM if userland tries to invoke SIOCINITIFADDR. In ifioctl(), give the interface the first shot at handling most interface ioctls, and give the protocol the second shot, instead of the other way around. Finally, let compatibility code (COMPAT_OSOCK) take a shot. Pull device initialization out of switch statements under SIOCINITIFADDR. For example, pull ..._init() out of any switch statement that looks like this: switch (...->sa_family) { case ...: ..._init(); ... break; ... default: ..._init(); ... break; } Rewrite many if-else clauses that handle all permutations of IFF_UP and IFF_RUNNING to use a switch statement, switch (x & (IFF_UP|IFF_RUNNING)) { case 0: ... break; case IFF_RUNNING: ... break; case IFF_UP: ... break; case IFF_UP|IFF_RUNNING: ... break; } unifdef lots of code containing #ifdef FreeBSD, #ifdef NetBSD, and #ifdef SIOCSIFMTU, especially in fwip(4) and in ndis(4). In ipw(4), remove an if_set_sadl() call that is out of place. In nfe(4), reuse the jumbo MTU logic in ether_ioctl(). Let ethernets register a callback for setting h/w state such as promiscuous mode and the multicast filter in accord with a change in the if_flags: ether_set_ifflags_cb() registers a callback that returns ENETRESET if the caller should reset the ethernet by calling if_init(), 0 on success, != 0 on failure. Pull common code from ex(4), gem(4), nfe(4), sip(4), tlp(4), vge(4) into ether_ioctl(), and register if_flags callbacks for those drivers. Return ENOTTY instead of EINVAL for inappropriate ioctls. In zyd(4), use ENXIO instead of ENOTTY to indicate that the device is not any longer attached. Add to if_set_sadl() a boolean 'factory' argument that indicates whether a link-layer address was assigned by the factory or some other source. In a comment, recommend using the factory address for generating an EUI64, and update in6_get_hw_ifid() to prefer a factory address to any other link-layer address. Add a routing message, RTM_LLINFO_UPD, that tells protocols to update the binding of network-layer addresses to link-layer addresses. Implement this message in IPv4 and IPv6 by sending a gratuitous ARP or a neighbor advertisement, respectively. Generate RTM_LLINFO_UPD messages on a change of an interface's link-layer address. In ether_ioctl(), do not let SIOCALIFADDR set a link-layer address that is broadcast/multicast or equal to 00:00:00:00:00:00. Make ether_ioctl() call ifioctl_common() to handle ioctls that it does not understand. In gif(4), initialize if_softc and use it, instead of assuming that the gif_softc and ifp overlap. Let ifioctl_common() handle SIOCGIFADDR. Sprinkle rtcache_invariants(), which checks on DIAGNOSTIC kernels that certain invariants on a struct route are satisfied. In agr(4), rewrite agr_ioctl_filter() to be a bit more explicit about the ioctls that we do not allow on an agr(4) member interface. bzero -> memset. Delete unnecessary casts to void *. Use sockaddr_in_init() and sockaddr_in6_init(). Compare pointers with NULL instead of "testing truth". Replace some instances of (type *)0 with NULL. Change some K&R prototypes to ANSI C, and join lines.
2008-11-07 03:20:01 +03:00
rtcache_invariants(ro);
Here are various changes designed to protect against bad IPv4 routing caused by stale route caches (struct route). Route caches are sprinkled throughout PCBs, the IP fast-forwarding table, and IP tunnel interfaces (gre, gif, stf). Stale IPv6 and ISO route caches will be treated by separate patches. Thank you to Christoph Badura for suggesting the general approach to invalidating route caches that I take here. Here are the details: Add hooks to struct domain for tracking and for invalidating each domain's route caches: dom_rtcache, dom_rtflush, and dom_rtflushall. Introduce helper subroutines, rtflush(ro) for invalidating a route cache, rtflushall(family) for invalidating all route caches in a routing domain, and rtcache(ro) for notifying the domain of a new cached route. Chain together all IPv4 route caches where ro_rt != NULL. Provide in_rtcache() for adding a route to the chain. Provide in_rtflush() and in_rtflushall() for invalidating IPv4 route caches. In in_rtflush(), set ro_rt to NULL, and remove the route from the chain. In in_rtflushall(), walk the chain and remove every route cache. In rtrequest1(), call rtflushall() to invalidate route caches when a route is added. In gif(4), discard the workaround for stale caches that involves expiring them every so often. Replace the pattern 'RTFREE(ro->ro_rt); ro->ro_rt = NULL;' with a call to rtflush(ro). Update ipflow_fastforward() and all other users of route caches so that they expect a cached route, ro->ro_rt, to turn to NULL. Take care when moving a 'struct route' to rtflush() the source and to rtcache() the destination. In domain initializers, use .dom_xxx tags. KNF here and there.
2006-12-09 08:33:04 +03:00
}
1993-03-21 12:45:37 +03:00
/*
* Packet routing routines.
*/
struct rtentry *
rtalloc1(const struct sockaddr *dst, int report)
1993-03-21 12:45:37 +03:00
{
rtbl_t *rtbl = rt_gettable(dst->sa_family);
2000-03-30 13:45:33 +04:00
struct rtentry *rt;
struct rtentry *newrt = NULL;
struct rt_addrinfo info;
1995-08-13 03:59:09 +04:00
int s = splsoftnet(), err = 0, msgtype = RTM_MISS;
1993-03-21 12:45:37 +03:00
if (rtbl != NULL && (rt = rt_matchaddr(rtbl, dst)) != NULL) {
newrt = rt;
1993-03-21 12:45:37 +03:00
if (report && (rt->rt_flags & RTF_CLONING)) {
err = rtrequest(RTM_RESOLVE, dst, NULL, NULL, 0,
&newrt);
if (err) {
newrt = rt;
rt->rt_refcnt++;
goto miss;
}
KASSERT(newrt != NULL);
if ((rt = newrt) && (rt->rt_flags & RTF_XRESOLVE)) {
msgtype = RTM_RESOLVE;
goto miss;
}
/* Inform listeners of the new route */
2001-07-18 20:43:09 +04:00
memset(&info, 0, sizeof(info));
Take steps to hide the radix_node implementation of the forwarding table from the forwarding table's users: Introduce rt_walktree() for walking the routing table and applying a function to each rtentry. Replace most rn_walktree() calls with it. Use rt_getkey()/rt_setkey() to get/set a route's destination. Keep a pointer to the sockaddr key in the rtentry, so that rtentry users do not have to grovel in the radix_node for the key. Add a RTM_GET method to rtrequest. Use that instead of radix_node lookups in, e.g., carp(4). Add sys/net/link_proto.c, which supplies sockaddr routines for link-layer socket addresses (sockaddr_dl). Cosmetic: Constify. KNF. Stop open-coding LIST_FOREACH, TAILQ_FOREACH, et cetera. Use NULL instead of 0 for null pointers. Use __arraycount(). Reduce gratuitous parenthesization. Stop using variadic arguments for rip6_output(), it is unnecessary. Remove the unnecessary rtentry member rt_genmask and the code to maintain it, since nothing actually used it. Make rt_maskedcopy() easier to read by using meaningful variable names. Extract a subroutine intern_netmask() for looking up a netmask in the masks table. Start converting backslash-ridden IPv6 macros in sys/netinet6/in6_var.h into inline subroutines that one can read without special eyeglasses. One functional change: when the kernel serves an RTM_GET, RTM_LOCK, or RTM_CHANGE request, it applies the netmask (if supplied) to a destination before searching for it in the forwarding table. I have changed sys/netinet/ip_carp.c, carp_setroute(), to remove the unlawful radix_node knowledge. Apart from the changes to carp(4), netiso, ATM, and strip(4), I have run the changes on three nodes in my wireless routing testbed, which involves IPv4 + IPv6 dynamic routing acrobatics, and it's working beautifully so far.
2007-07-20 00:48:52 +04:00
info.rti_info[RTAX_DST] = rt_getkey(rt);
info.rti_info[RTAX_NETMASK] = rt_mask(rt);
info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
if (rt->rt_ifp != NULL) {
2005-02-27 01:45:09 +03:00
info.rti_info[RTAX_IFP] =
rt->rt_ifp->if_dl->ifa_addr;
info.rti_info[RTAX_IFA] = rt->rt_ifa->ifa_addr;
}
rt_missmsg(RTM_ADD, &info, rt->rt_flags, 0);
1993-03-21 12:45:37 +03:00
} else
rt->rt_refcnt++;
} else {
rtstat.rts_unreach++;
miss: if (report) {
memset((void *)&info, 0, sizeof(info));
info.rti_info[RTAX_DST] = dst;
rt_missmsg(msgtype, &info, 0, err);
}
1993-03-21 12:45:37 +03:00
}
splx(s);
return newrt;
1993-03-21 12:45:37 +03:00
}
void
rtfree(struct rtentry *rt)
1993-03-21 12:45:37 +03:00
{
2000-03-30 13:45:33 +04:00
struct ifaddr *ifa;
if (rt == NULL)
1993-03-21 12:45:37 +03:00
panic("rtfree");
rt->rt_refcnt--;
if (rt->rt_refcnt <= 0 && (rt->rt_flags & RTF_UP) == 0) {
rt_assert_inactive(rt);
rttrash--;
if (rt->rt_refcnt < 0) {
1996-10-13 06:10:01 +04:00
printf("rtfree: %p not freed (neg refs)\n", rt);
return;
}
rt_timer_remove_all(rt, 0);
ifa = rt->rt_ifa;
rt->rt_ifa = NULL;
IFAFREE(ifa);
rt->rt_ifp = NULL;
Take steps to hide the radix_node implementation of the forwarding table from the forwarding table's users: Introduce rt_walktree() for walking the routing table and applying a function to each rtentry. Replace most rn_walktree() calls with it. Use rt_getkey()/rt_setkey() to get/set a route's destination. Keep a pointer to the sockaddr key in the rtentry, so that rtentry users do not have to grovel in the radix_node for the key. Add a RTM_GET method to rtrequest. Use that instead of radix_node lookups in, e.g., carp(4). Add sys/net/link_proto.c, which supplies sockaddr routines for link-layer socket addresses (sockaddr_dl). Cosmetic: Constify. KNF. Stop open-coding LIST_FOREACH, TAILQ_FOREACH, et cetera. Use NULL instead of 0 for null pointers. Use __arraycount(). Reduce gratuitous parenthesization. Stop using variadic arguments for rip6_output(), it is unnecessary. Remove the unnecessary rtentry member rt_genmask and the code to maintain it, since nothing actually used it. Make rt_maskedcopy() easier to read by using meaningful variable names. Extract a subroutine intern_netmask() for looking up a netmask in the masks table. Start converting backslash-ridden IPv6 macros in sys/netinet6/in6_var.h into inline subroutines that one can read without special eyeglasses. One functional change: when the kernel serves an RTM_GET, RTM_LOCK, or RTM_CHANGE request, it applies the netmask (if supplied) to a destination before searching for it in the forwarding table. I have changed sys/netinet/ip_carp.c, carp_setroute(), to remove the unlawful radix_node knowledge. Apart from the changes to carp(4), netiso, ATM, and strip(4), I have run the changes on three nodes in my wireless routing testbed, which involves IPv4 + IPv6 dynamic routing acrobatics, and it's working beautifully so far.
2007-07-20 00:48:52 +04:00
rt_destroy(rt);
pool_put(&rtentry_pool, rt);
1993-03-21 12:45:37 +03:00
}
}
void
ifafree(struct ifaddr *ifa)
{
#ifdef DIAGNOSTIC
if (ifa == NULL)
panic("ifafree: null ifa");
if (ifa->ifa_refcnt != 0)
panic("ifafree: ifa_refcnt != 0 (%d)", ifa->ifa_refcnt);
#endif
#ifdef IFAREF_DEBUG
printf("ifafree: freeing ifaddr %p\n", ifa);
#endif
free(ifa, M_IFADDR);
}
1993-03-21 12:45:37 +03:00
/*
* Force a routing table entry to the specified
* destination to go through the given gateway.
* Normally called as a result of a routing redirect
* message from the network layer.
*
1995-08-13 03:59:09 +04:00
* N.B.: must be called at splsoftnet
1993-03-21 12:45:37 +03:00
*/
1996-02-14 00:59:53 +03:00
void
rtredirect(const struct sockaddr *dst, const struct sockaddr *gateway,
const struct sockaddr *netmask, int flags, const struct sockaddr *src,
struct rtentry **rtp)
1993-03-21 12:45:37 +03:00
{
2000-03-30 13:45:33 +04:00
struct rtentry *rt;
1993-03-21 12:45:37 +03:00
int error = 0;
2009-11-03 03:30:11 +03:00
uint64_t *stat = NULL;
struct rt_addrinfo info;
struct ifaddr *ifa;
1993-03-21 12:45:37 +03:00
/* verify the gateway is directly reachable */
if ((ifa = ifa_ifwithnet(gateway)) == NULL) {
1993-03-21 12:45:37 +03:00
error = ENETUNREACH;
goto out;
1993-03-21 12:45:37 +03:00
}
rt = rtalloc1(dst, 0);
/*
* If the redirect isn't from our current router for this dst,
* it's either old or wrong. If it redirects us to ourselves,
* we have a routing loop, perhaps as a result of an interface
* going down recently.
*/
if (!(flags & RTF_DONE) && rt &&
(sockaddr_cmp(src, rt->rt_gateway) != 0 || rt->rt_ifa != ifa))
1993-03-21 12:45:37 +03:00
error = EINVAL;
else if (ifa_ifwithaddr(gateway))
error = EHOSTUNREACH;
if (error)
goto done;
/*
* Create a new entry if we just got back a wildcard entry
2000-03-14 02:52:25 +03:00
* or the lookup failed. This is necessary for hosts
1993-03-21 12:45:37 +03:00
* which use routing redirects generated by smart gateways
* to dynamically build the routing tables.
*/
if (rt == NULL || (rt_mask(rt) && rt_mask(rt)->sa_len < 2))
1993-03-21 12:45:37 +03:00
goto create;
/*
* Don't listen to the redirect if it's
2005-02-27 01:45:09 +03:00
* for a route to an interface.
1993-03-21 12:45:37 +03:00
*/
if (rt->rt_flags & RTF_GATEWAY) {
if (((rt->rt_flags & RTF_HOST) == 0) && (flags & RTF_HOST)) {
/*
* Changing from route to net => route to host.
* Create new route, rather than smashing route to net.
*/
create:
if (rt != NULL)
rtfree(rt);
1993-03-21 12:45:37 +03:00
flags |= RTF_GATEWAY | RTF_DYNAMIC;
memset(&info, 0, sizeof(info));
info.rti_info[RTAX_DST] = dst;
info.rti_info[RTAX_GATEWAY] = gateway;
info.rti_info[RTAX_NETMASK] = netmask;
info.rti_ifa = ifa;
info.rti_flags = flags;
rt = NULL;
error = rtrequest1(RTM_ADD, &info, &rt);
if (rt != NULL)
flags = rt->rt_flags;
1993-03-21 12:45:37 +03:00
stat = &rtstat.rts_dynamic;
} else {
/*
* Smash the current notion of the gateway to
* this destination. Should check about netmask!!!
*/
rt->rt_flags |= RTF_MODIFIED;
flags |= RTF_MODIFIED;
stat = &rtstat.rts_newgateway;
Take steps to hide the radix_node implementation of the forwarding table from the forwarding table's users: Introduce rt_walktree() for walking the routing table and applying a function to each rtentry. Replace most rn_walktree() calls with it. Use rt_getkey()/rt_setkey() to get/set a route's destination. Keep a pointer to the sockaddr key in the rtentry, so that rtentry users do not have to grovel in the radix_node for the key. Add a RTM_GET method to rtrequest. Use that instead of radix_node lookups in, e.g., carp(4). Add sys/net/link_proto.c, which supplies sockaddr routines for link-layer socket addresses (sockaddr_dl). Cosmetic: Constify. KNF. Stop open-coding LIST_FOREACH, TAILQ_FOREACH, et cetera. Use NULL instead of 0 for null pointers. Use __arraycount(). Reduce gratuitous parenthesization. Stop using variadic arguments for rip6_output(), it is unnecessary. Remove the unnecessary rtentry member rt_genmask and the code to maintain it, since nothing actually used it. Make rt_maskedcopy() easier to read by using meaningful variable names. Extract a subroutine intern_netmask() for looking up a netmask in the masks table. Start converting backslash-ridden IPv6 macros in sys/netinet6/in6_var.h into inline subroutines that one can read without special eyeglasses. One functional change: when the kernel serves an RTM_GET, RTM_LOCK, or RTM_CHANGE request, it applies the netmask (if supplied) to a destination before searching for it in the forwarding table. I have changed sys/netinet/ip_carp.c, carp_setroute(), to remove the unlawful radix_node knowledge. Apart from the changes to carp(4), netiso, ATM, and strip(4), I have run the changes on three nodes in my wireless routing testbed, which involves IPv4 + IPv6 dynamic routing acrobatics, and it's working beautifully so far.
2007-07-20 00:48:52 +04:00
rt_setgate(rt, gateway);
1993-03-21 12:45:37 +03:00
}
} else
error = EHOSTUNREACH;
done:
if (rt) {
if (rtp != NULL && !error)
1993-03-21 12:45:37 +03:00
*rtp = rt;
else
rtfree(rt);
}
out:
1993-03-21 12:45:37 +03:00
if (error)
rtstat.rts_badredirect++;
else if (stat != NULL)
(*stat)++;
memset(&info, 0, sizeof(info));
info.rti_info[RTAX_DST] = dst;
info.rti_info[RTAX_GATEWAY] = gateway;
info.rti_info[RTAX_NETMASK] = netmask;
info.rti_info[RTAX_AUTHOR] = src;
rt_missmsg(RTM_REDIRECT, &info, flags, error);
1993-03-21 12:45:37 +03:00
}
/*
* Delete a route and generate a message
*/
static int
rtdeletemsg(struct rtentry *rt)
{
int error;
struct rt_addrinfo info;
/*
* Request the new route so that the entry is not actually
* deleted. That will allow the information being reported to
* be accurate (and consistent with route_output()).
*/
memset(&info, 0, sizeof(info));
Take steps to hide the radix_node implementation of the forwarding table from the forwarding table's users: Introduce rt_walktree() for walking the routing table and applying a function to each rtentry. Replace most rn_walktree() calls with it. Use rt_getkey()/rt_setkey() to get/set a route's destination. Keep a pointer to the sockaddr key in the rtentry, so that rtentry users do not have to grovel in the radix_node for the key. Add a RTM_GET method to rtrequest. Use that instead of radix_node lookups in, e.g., carp(4). Add sys/net/link_proto.c, which supplies sockaddr routines for link-layer socket addresses (sockaddr_dl). Cosmetic: Constify. KNF. Stop open-coding LIST_FOREACH, TAILQ_FOREACH, et cetera. Use NULL instead of 0 for null pointers. Use __arraycount(). Reduce gratuitous parenthesization. Stop using variadic arguments for rip6_output(), it is unnecessary. Remove the unnecessary rtentry member rt_genmask and the code to maintain it, since nothing actually used it. Make rt_maskedcopy() easier to read by using meaningful variable names. Extract a subroutine intern_netmask() for looking up a netmask in the masks table. Start converting backslash-ridden IPv6 macros in sys/netinet6/in6_var.h into inline subroutines that one can read without special eyeglasses. One functional change: when the kernel serves an RTM_GET, RTM_LOCK, or RTM_CHANGE request, it applies the netmask (if supplied) to a destination before searching for it in the forwarding table. I have changed sys/netinet/ip_carp.c, carp_setroute(), to remove the unlawful radix_node knowledge. Apart from the changes to carp(4), netiso, ATM, and strip(4), I have run the changes on three nodes in my wireless routing testbed, which involves IPv4 + IPv6 dynamic routing acrobatics, and it's working beautifully so far.
2007-07-20 00:48:52 +04:00
info.rti_info[RTAX_DST] = rt_getkey(rt);
info.rti_info[RTAX_NETMASK] = rt_mask(rt);
info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
info.rti_flags = rt->rt_flags;
error = rtrequest1(RTM_DELETE, &info, &rt);
rt_missmsg(RTM_DELETE, &info, info.rti_flags, error);
/* Adjust the refcount */
if (error == 0 && rt->rt_refcnt <= 0) {
rt->rt_refcnt++;
rtfree(rt);
}
return error;
}
static int
rtflushclone1(struct rtentry *rt, void *arg)
{
struct rtentry *parent;
parent = (struct rtentry *)arg;
if ((rt->rt_flags & RTF_CLONED) != 0 && rt->rt_parent == parent)
rtdeletemsg(rt);
return 0;
}
static void
rtflushclone(sa_family_t family, struct rtentry *parent)
{
#ifdef DIAGNOSTIC
if (!parent || (parent->rt_flags & RTF_CLONING) == 0)
panic("rtflushclone: called with a non-cloning route");
#endif
rt_walktree(family, rtflushclone1, (void *)parent);
}
1993-03-21 12:45:37 +03:00
/*
* Routing table ioctl interface.
*/
int
rtioctl(u_long req, void *data, struct lwp *l)
1993-03-21 12:45:37 +03:00
{
return EOPNOTSUPP;
1993-03-21 12:45:37 +03:00
}
struct ifaddr *
ifa_ifwithroute(int flags, const struct sockaddr *dst,
const struct sockaddr *gateway)
1993-03-21 12:45:37 +03:00
{
2000-03-30 13:45:33 +04:00
struct ifaddr *ifa;
1993-03-21 12:45:37 +03:00
if ((flags & RTF_GATEWAY) == 0) {
/*
* If we are adding a route to an interface,
* and the interface is a pt to pt link
* we should search for the destination
* as our clue to the interface. Otherwise
* we can use the local address.
*/
ifa = NULL;
2005-02-27 01:45:09 +03:00
if (flags & RTF_HOST)
1993-03-21 12:45:37 +03:00
ifa = ifa_ifwithdstaddr(dst);
if (ifa == NULL)
1993-03-21 12:45:37 +03:00
ifa = ifa_ifwithaddr(gateway);
} else {
/*
* If we are adding a route to a remote net
* or host, the gateway may still be on the
* other end of a pt to pt link.
*/
ifa = ifa_ifwithdstaddr(gateway);
}
if (ifa == NULL)
1993-03-21 12:45:37 +03:00
ifa = ifa_ifwithnet(gateway);
if (ifa == NULL) {
1993-03-21 12:45:37 +03:00
struct rtentry *rt = rtalloc1(dst, 0);
if (rt == NULL)
return NULL;
1993-03-21 12:45:37 +03:00
rt->rt_refcnt--;
if ((ifa = rt->rt_ifa) == NULL)
return NULL;
1993-03-21 12:45:37 +03:00
}
if (ifa->ifa_addr->sa_family != dst->sa_family) {
struct ifaddr *oifa = ifa;
1993-03-21 12:45:37 +03:00
ifa = ifaof_ifpforaddr(dst, ifa->ifa_ifp);
if (ifa == 0)
ifa = oifa;
}
return ifa;
1993-03-21 12:45:37 +03:00
}
int
rtrequest(int req, const struct sockaddr *dst, const struct sockaddr *gateway,
const struct sockaddr *netmask, int flags, struct rtentry **ret_nrt)
{
struct rt_addrinfo info;
2001-07-18 20:43:09 +04:00
memset(&info, 0, sizeof(info));
info.rti_flags = flags;
info.rti_info[RTAX_DST] = dst;
info.rti_info[RTAX_GATEWAY] = gateway;
info.rti_info[RTAX_NETMASK] = netmask;
return rtrequest1(req, &info, ret_nrt);
}
int
rt_getifa(struct rt_addrinfo *info)
{
struct ifaddr *ifa;
const struct sockaddr *dst = info->rti_info[RTAX_DST];
const struct sockaddr *gateway = info->rti_info[RTAX_GATEWAY];
const struct sockaddr *ifaaddr = info->rti_info[RTAX_IFA];
const struct sockaddr *ifpaddr = info->rti_info[RTAX_IFP];
int flags = info->rti_flags;
/*
* ifp may be specified by sockaddr_dl when protocol address
* is ambiguous
*/
if (info->rti_ifp == NULL && ifpaddr != NULL
&& ifpaddr->sa_family == AF_LINK &&
2008-01-08 06:37:45 +03:00
(ifa = ifa_ifwithnet(ifpaddr)) != NULL)
info->rti_ifp = ifa->ifa_ifp;
if (info->rti_ifa == NULL && ifaaddr != NULL)
info->rti_ifa = ifa_ifwithaddr(ifaaddr);
if (info->rti_ifa == NULL) {
const struct sockaddr *sa;
sa = ifaaddr != NULL ? ifaaddr :
(gateway != NULL ? gateway : dst);
if (sa != NULL && info->rti_ifp != NULL)
info->rti_ifa = ifaof_ifpforaddr(sa, info->rti_ifp);
else if (dst != NULL && gateway != NULL)
info->rti_ifa = ifa_ifwithroute(flags, dst, gateway);
else if (sa != NULL)
info->rti_ifa = ifa_ifwithroute(flags, sa, sa);
}
if ((ifa = info->rti_ifa) == NULL)
return ENETUNREACH;
if (ifa->ifa_getifa != NULL)
info->rti_ifa = ifa = (*ifa->ifa_getifa)(ifa, dst);
if (info->rti_ifp == NULL)
info->rti_ifp = ifa->ifa_ifp;
return 0;
}
int
rtrequest1(int req, struct rt_addrinfo *info, struct rtentry **ret_nrt)
1993-03-21 12:45:37 +03:00
{
int s = splsoftnet();
int error = 0, rc;
struct rtentry *rt, *crt;
rtbl_t *rtbl;
struct ifaddr *ifa, *ifa2;
Take steps to hide the radix_node implementation of the forwarding table from the forwarding table's users: Introduce rt_walktree() for walking the routing table and applying a function to each rtentry. Replace most rn_walktree() calls with it. Use rt_getkey()/rt_setkey() to get/set a route's destination. Keep a pointer to the sockaddr key in the rtentry, so that rtentry users do not have to grovel in the radix_node for the key. Add a RTM_GET method to rtrequest. Use that instead of radix_node lookups in, e.g., carp(4). Add sys/net/link_proto.c, which supplies sockaddr routines for link-layer socket addresses (sockaddr_dl). Cosmetic: Constify. KNF. Stop open-coding LIST_FOREACH, TAILQ_FOREACH, et cetera. Use NULL instead of 0 for null pointers. Use __arraycount(). Reduce gratuitous parenthesization. Stop using variadic arguments for rip6_output(), it is unnecessary. Remove the unnecessary rtentry member rt_genmask and the code to maintain it, since nothing actually used it. Make rt_maskedcopy() easier to read by using meaningful variable names. Extract a subroutine intern_netmask() for looking up a netmask in the masks table. Start converting backslash-ridden IPv6 macros in sys/netinet6/in6_var.h into inline subroutines that one can read without special eyeglasses. One functional change: when the kernel serves an RTM_GET, RTM_LOCK, or RTM_CHANGE request, it applies the netmask (if supplied) to a destination before searching for it in the forwarding table. I have changed sys/netinet/ip_carp.c, carp_setroute(), to remove the unlawful radix_node knowledge. Apart from the changes to carp(4), netiso, ATM, and strip(4), I have run the changes on three nodes in my wireless routing testbed, which involves IPv4 + IPv6 dynamic routing acrobatics, and it's working beautifully so far.
2007-07-20 00:48:52 +04:00
struct sockaddr_storage maskeddst;
const struct sockaddr *dst = info->rti_info[RTAX_DST];
const struct sockaddr *gateway = info->rti_info[RTAX_GATEWAY];
const struct sockaddr *netmask = info->rti_info[RTAX_NETMASK];
int flags = info->rti_flags;
1993-03-21 12:45:37 +03:00
#define senderr(x) { error = x ; goto bad; }
if ((rtbl = rt_gettable(dst->sa_family)) == NULL)
1993-03-21 12:45:37 +03:00
senderr(ESRCH);
if (flags & RTF_HOST)
netmask = NULL;
1993-03-21 12:45:37 +03:00
switch (req) {
case RTM_DELETE:
if (netmask) {
Take steps to hide the radix_node implementation of the forwarding table from the forwarding table's users: Introduce rt_walktree() for walking the routing table and applying a function to each rtentry. Replace most rn_walktree() calls with it. Use rt_getkey()/rt_setkey() to get/set a route's destination. Keep a pointer to the sockaddr key in the rtentry, so that rtentry users do not have to grovel in the radix_node for the key. Add a RTM_GET method to rtrequest. Use that instead of radix_node lookups in, e.g., carp(4). Add sys/net/link_proto.c, which supplies sockaddr routines for link-layer socket addresses (sockaddr_dl). Cosmetic: Constify. KNF. Stop open-coding LIST_FOREACH, TAILQ_FOREACH, et cetera. Use NULL instead of 0 for null pointers. Use __arraycount(). Reduce gratuitous parenthesization. Stop using variadic arguments for rip6_output(), it is unnecessary. Remove the unnecessary rtentry member rt_genmask and the code to maintain it, since nothing actually used it. Make rt_maskedcopy() easier to read by using meaningful variable names. Extract a subroutine intern_netmask() for looking up a netmask in the masks table. Start converting backslash-ridden IPv6 macros in sys/netinet6/in6_var.h into inline subroutines that one can read without special eyeglasses. One functional change: when the kernel serves an RTM_GET, RTM_LOCK, or RTM_CHANGE request, it applies the netmask (if supplied) to a destination before searching for it in the forwarding table. I have changed sys/netinet/ip_carp.c, carp_setroute(), to remove the unlawful radix_node knowledge. Apart from the changes to carp(4), netiso, ATM, and strip(4), I have run the changes on three nodes in my wireless routing testbed, which involves IPv4 + IPv6 dynamic routing acrobatics, and it's working beautifully so far.
2007-07-20 00:48:52 +04:00
rt_maskedcopy(dst, (struct sockaddr *)&maskeddst,
netmask);
dst = (struct sockaddr *)&maskeddst;
}
if ((rt = rt_lookup(rtbl, dst, netmask)) == NULL)
senderr(ESRCH);
if ((rt->rt_flags & RTF_CLONING) != 0) {
/* clean up any cloned children */
rtflushclone(dst->sa_family, rt);
}
if ((rt = rt_deladdr(rtbl, dst, netmask)) == NULL)
1993-03-21 12:45:37 +03:00
senderr(ESRCH);
if (rt->rt_gwroute) {
RTFREE(rt->rt_gwroute);
rt->rt_gwroute = NULL;
}
if (rt->rt_parent) {
rt->rt_parent->rt_refcnt--;
rt->rt_parent = NULL;
}
rt->rt_flags &= ~RTF_UP;
if ((ifa = rt->rt_ifa)) {
if (ifa->ifa_flags & IFA_ROUTE &&
rt_ifa_connected(rt, ifa)) {
RT_DPRINTF("rt->_rt_key = %p, ifa = %p, "
"deleted IFA_ROUTE\n",
(void *)rt->_rt_key, (void *)ifa);
ifa->ifa_flags &= ~IFA_ROUTE;
}
if (ifa->ifa_rtrequest)
ifa->ifa_rtrequest(RTM_DELETE, rt, info);
}
1993-03-21 12:45:37 +03:00
rttrash++;
if (ret_nrt)
*ret_nrt = rt;
else if (rt->rt_refcnt <= 0) {
rt->rt_refcnt++;
1993-03-21 12:45:37 +03:00
rtfree(rt);
}
1993-03-21 12:45:37 +03:00
break;
case RTM_RESOLVE:
if (ret_nrt == NULL || (rt = *ret_nrt) == NULL)
1993-03-21 12:45:37 +03:00
senderr(EINVAL);
if ((rt->rt_flags & RTF_CLONING) == 0)
senderr(EINVAL);
1993-03-21 12:45:37 +03:00
ifa = rt->rt_ifa;
flags = rt->rt_flags & ~(RTF_CLONING | RTF_STATIC);
flags |= RTF_CLONED;
1993-03-21 12:45:37 +03:00
gateway = rt->rt_gateway;
Take steps to hide the radix_node implementation of the forwarding table from the forwarding table's users: Introduce rt_walktree() for walking the routing table and applying a function to each rtentry. Replace most rn_walktree() calls with it. Use rt_getkey()/rt_setkey() to get/set a route's destination. Keep a pointer to the sockaddr key in the rtentry, so that rtentry users do not have to grovel in the radix_node for the key. Add a RTM_GET method to rtrequest. Use that instead of radix_node lookups in, e.g., carp(4). Add sys/net/link_proto.c, which supplies sockaddr routines for link-layer socket addresses (sockaddr_dl). Cosmetic: Constify. KNF. Stop open-coding LIST_FOREACH, TAILQ_FOREACH, et cetera. Use NULL instead of 0 for null pointers. Use __arraycount(). Reduce gratuitous parenthesization. Stop using variadic arguments for rip6_output(), it is unnecessary. Remove the unnecessary rtentry member rt_genmask and the code to maintain it, since nothing actually used it. Make rt_maskedcopy() easier to read by using meaningful variable names. Extract a subroutine intern_netmask() for looking up a netmask in the masks table. Start converting backslash-ridden IPv6 macros in sys/netinet6/in6_var.h into inline subroutines that one can read without special eyeglasses. One functional change: when the kernel serves an RTM_GET, RTM_LOCK, or RTM_CHANGE request, it applies the netmask (if supplied) to a destination before searching for it in the forwarding table. I have changed sys/netinet/ip_carp.c, carp_setroute(), to remove the unlawful radix_node knowledge. Apart from the changes to carp(4), netiso, ATM, and strip(4), I have run the changes on three nodes in my wireless routing testbed, which involves IPv4 + IPv6 dynamic routing acrobatics, and it's working beautifully so far.
2007-07-20 00:48:52 +04:00
flags |= RTF_HOST;
1993-03-21 12:45:37 +03:00
goto makeroute;
case RTM_ADD:
if (info->rti_ifa == NULL && (error = rt_getifa(info)))
senderr(error);
ifa = info->rti_ifa;
1993-03-21 12:45:37 +03:00
makeroute:
/* Already at splsoftnet() so pool_get/pool_put are safe */
rt = pool_get(&rtentry_pool, PR_NOWAIT);
if (rt == NULL)
1993-03-21 12:45:37 +03:00
senderr(ENOBUFS);
memset(rt, 0, sizeof(*rt));
rt->rt_flags = RTF_UP | flags;
LIST_INIT(&rt->rt_timer);
2008-05-14 00:18:11 +04:00
RT_DPRINTF("rt->_rt_key = %p\n", (void *)rt->_rt_key);
if (rt_setkey(rt, dst, M_NOWAIT) == NULL ||
Take steps to hide the radix_node implementation of the forwarding table from the forwarding table's users: Introduce rt_walktree() for walking the routing table and applying a function to each rtentry. Replace most rn_walktree() calls with it. Use rt_getkey()/rt_setkey() to get/set a route's destination. Keep a pointer to the sockaddr key in the rtentry, so that rtentry users do not have to grovel in the radix_node for the key. Add a RTM_GET method to rtrequest. Use that instead of radix_node lookups in, e.g., carp(4). Add sys/net/link_proto.c, which supplies sockaddr routines for link-layer socket addresses (sockaddr_dl). Cosmetic: Constify. KNF. Stop open-coding LIST_FOREACH, TAILQ_FOREACH, et cetera. Use NULL instead of 0 for null pointers. Use __arraycount(). Reduce gratuitous parenthesization. Stop using variadic arguments for rip6_output(), it is unnecessary. Remove the unnecessary rtentry member rt_genmask and the code to maintain it, since nothing actually used it. Make rt_maskedcopy() easier to read by using meaningful variable names. Extract a subroutine intern_netmask() for looking up a netmask in the masks table. Start converting backslash-ridden IPv6 macros in sys/netinet6/in6_var.h into inline subroutines that one can read without special eyeglasses. One functional change: when the kernel serves an RTM_GET, RTM_LOCK, or RTM_CHANGE request, it applies the netmask (if supplied) to a destination before searching for it in the forwarding table. I have changed sys/netinet/ip_carp.c, carp_setroute(), to remove the unlawful radix_node knowledge. Apart from the changes to carp(4), netiso, ATM, and strip(4), I have run the changes on three nodes in my wireless routing testbed, which involves IPv4 + IPv6 dynamic routing acrobatics, and it's working beautifully so far.
2007-07-20 00:48:52 +04:00
rt_setgate(rt, gateway) != 0) {
pool_put(&rtentry_pool, rt);
senderr(ENOBUFS);
}
2008-05-14 00:18:11 +04:00
RT_DPRINTF("rt->_rt_key = %p\n", (void *)rt->_rt_key);
1993-03-21 12:45:37 +03:00
if (netmask) {
Take steps to hide the radix_node implementation of the forwarding table from the forwarding table's users: Introduce rt_walktree() for walking the routing table and applying a function to each rtentry. Replace most rn_walktree() calls with it. Use rt_getkey()/rt_setkey() to get/set a route's destination. Keep a pointer to the sockaddr key in the rtentry, so that rtentry users do not have to grovel in the radix_node for the key. Add a RTM_GET method to rtrequest. Use that instead of radix_node lookups in, e.g., carp(4). Add sys/net/link_proto.c, which supplies sockaddr routines for link-layer socket addresses (sockaddr_dl). Cosmetic: Constify. KNF. Stop open-coding LIST_FOREACH, TAILQ_FOREACH, et cetera. Use NULL instead of 0 for null pointers. Use __arraycount(). Reduce gratuitous parenthesization. Stop using variadic arguments for rip6_output(), it is unnecessary. Remove the unnecessary rtentry member rt_genmask and the code to maintain it, since nothing actually used it. Make rt_maskedcopy() easier to read by using meaningful variable names. Extract a subroutine intern_netmask() for looking up a netmask in the masks table. Start converting backslash-ridden IPv6 macros in sys/netinet6/in6_var.h into inline subroutines that one can read without special eyeglasses. One functional change: when the kernel serves an RTM_GET, RTM_LOCK, or RTM_CHANGE request, it applies the netmask (if supplied) to a destination before searching for it in the forwarding table. I have changed sys/netinet/ip_carp.c, carp_setroute(), to remove the unlawful radix_node knowledge. Apart from the changes to carp(4), netiso, ATM, and strip(4), I have run the changes on three nodes in my wireless routing testbed, which involves IPv4 + IPv6 dynamic routing acrobatics, and it's working beautifully so far.
2007-07-20 00:48:52 +04:00
rt_maskedcopy(dst, (struct sockaddr *)&maskeddst,
netmask);
rt_setkey(rt, (struct sockaddr *)&maskeddst, M_NOWAIT);
2008-05-14 00:18:11 +04:00
RT_DPRINTF("rt->_rt_key = %p\n", (void *)rt->_rt_key);
Take steps to hide the radix_node implementation of the forwarding table from the forwarding table's users: Introduce rt_walktree() for walking the routing table and applying a function to each rtentry. Replace most rn_walktree() calls with it. Use rt_getkey()/rt_setkey() to get/set a route's destination. Keep a pointer to the sockaddr key in the rtentry, so that rtentry users do not have to grovel in the radix_node for the key. Add a RTM_GET method to rtrequest. Use that instead of radix_node lookups in, e.g., carp(4). Add sys/net/link_proto.c, which supplies sockaddr routines for link-layer socket addresses (sockaddr_dl). Cosmetic: Constify. KNF. Stop open-coding LIST_FOREACH, TAILQ_FOREACH, et cetera. Use NULL instead of 0 for null pointers. Use __arraycount(). Reduce gratuitous parenthesization. Stop using variadic arguments for rip6_output(), it is unnecessary. Remove the unnecessary rtentry member rt_genmask and the code to maintain it, since nothing actually used it. Make rt_maskedcopy() easier to read by using meaningful variable names. Extract a subroutine intern_netmask() for looking up a netmask in the masks table. Start converting backslash-ridden IPv6 macros in sys/netinet6/in6_var.h into inline subroutines that one can read without special eyeglasses. One functional change: when the kernel serves an RTM_GET, RTM_LOCK, or RTM_CHANGE request, it applies the netmask (if supplied) to a destination before searching for it in the forwarding table. I have changed sys/netinet/ip_carp.c, carp_setroute(), to remove the unlawful radix_node knowledge. Apart from the changes to carp(4), netiso, ATM, and strip(4), I have run the changes on three nodes in my wireless routing testbed, which involves IPv4 + IPv6 dynamic routing acrobatics, and it's working beautifully so far.
2007-07-20 00:48:52 +04:00
} else {
rt_setkey(rt, dst, M_NOWAIT);
2008-05-14 00:18:11 +04:00
RT_DPRINTF("rt->_rt_key = %p\n", (void *)rt->_rt_key);
Take steps to hide the radix_node implementation of the forwarding table from the forwarding table's users: Introduce rt_walktree() for walking the routing table and applying a function to each rtentry. Replace most rn_walktree() calls with it. Use rt_getkey()/rt_setkey() to get/set a route's destination. Keep a pointer to the sockaddr key in the rtentry, so that rtentry users do not have to grovel in the radix_node for the key. Add a RTM_GET method to rtrequest. Use that instead of radix_node lookups in, e.g., carp(4). Add sys/net/link_proto.c, which supplies sockaddr routines for link-layer socket addresses (sockaddr_dl). Cosmetic: Constify. KNF. Stop open-coding LIST_FOREACH, TAILQ_FOREACH, et cetera. Use NULL instead of 0 for null pointers. Use __arraycount(). Reduce gratuitous parenthesization. Stop using variadic arguments for rip6_output(), it is unnecessary. Remove the unnecessary rtentry member rt_genmask and the code to maintain it, since nothing actually used it. Make rt_maskedcopy() easier to read by using meaningful variable names. Extract a subroutine intern_netmask() for looking up a netmask in the masks table. Start converting backslash-ridden IPv6 macros in sys/netinet6/in6_var.h into inline subroutines that one can read without special eyeglasses. One functional change: when the kernel serves an RTM_GET, RTM_LOCK, or RTM_CHANGE request, it applies the netmask (if supplied) to a destination before searching for it in the forwarding table. I have changed sys/netinet/ip_carp.c, carp_setroute(), to remove the unlawful radix_node knowledge. Apart from the changes to carp(4), netiso, ATM, and strip(4), I have run the changes on three nodes in my wireless routing testbed, which involves IPv4 + IPv6 dynamic routing acrobatics, and it's working beautifully so far.
2007-07-20 00:48:52 +04:00
}
rt_set_ifa(rt, ifa);
if (info->rti_info[RTAX_TAG] != NULL)
rt_settag(rt, info->rti_info[RTAX_TAG]);
2008-05-14 00:18:11 +04:00
RT_DPRINTF("rt->_rt_key = %p\n", (void *)rt->_rt_key);
if (info->rti_info[RTAX_IFP] != NULL &&
(ifa2 = ifa_ifwithnet(info->rti_info[RTAX_IFP])) != NULL &&
ifa2->ifa_ifp != NULL)
rt->rt_ifp = ifa2->ifa_ifp;
else
rt->rt_ifp = ifa->ifa_ifp;
if (req == RTM_RESOLVE) {
1993-03-21 12:45:37 +03:00
rt->rt_rmx = (*ret_nrt)->rt_rmx; /* copy metrics */
rt->rt_parent = *ret_nrt;
rt->rt_parent->rt_refcnt++;
}
2008-05-14 00:18:11 +04:00
RT_DPRINTF("rt->_rt_key = %p\n", (void *)rt->_rt_key);
rc = rt_addaddr(rtbl, rt, netmask);
2008-05-14 00:18:11 +04:00
RT_DPRINTF("rt->_rt_key = %p\n", (void *)rt->_rt_key);
if (rc != 0 && (crt = rtalloc1(rt_getkey(rt), 0)) != NULL) {
/* overwrite cloned route */
if ((crt->rt_flags & RTF_CLONED) != 0) {
rtdeletemsg(crt);
rc = rt_addaddr(rtbl, rt, netmask);
}
RTFREE(crt);
2008-05-14 00:18:11 +04:00
RT_DPRINTF("rt->_rt_key = %p\n", (void *)rt->_rt_key);
}
2008-05-14 00:18:11 +04:00
RT_DPRINTF("rt->_rt_key = %p\n", (void *)rt->_rt_key);
if (rc != 0) {
IFAFREE(ifa);
if ((rt->rt_flags & RTF_CLONED) != 0 && rt->rt_parent)
rtfree(rt->rt_parent);
if (rt->rt_gwroute)
rtfree(rt->rt_gwroute);
Take steps to hide the radix_node implementation of the forwarding table from the forwarding table's users: Introduce rt_walktree() for walking the routing table and applying a function to each rtentry. Replace most rn_walktree() calls with it. Use rt_getkey()/rt_setkey() to get/set a route's destination. Keep a pointer to the sockaddr key in the rtentry, so that rtentry users do not have to grovel in the radix_node for the key. Add a RTM_GET method to rtrequest. Use that instead of radix_node lookups in, e.g., carp(4). Add sys/net/link_proto.c, which supplies sockaddr routines for link-layer socket addresses (sockaddr_dl). Cosmetic: Constify. KNF. Stop open-coding LIST_FOREACH, TAILQ_FOREACH, et cetera. Use NULL instead of 0 for null pointers. Use __arraycount(). Reduce gratuitous parenthesization. Stop using variadic arguments for rip6_output(), it is unnecessary. Remove the unnecessary rtentry member rt_genmask and the code to maintain it, since nothing actually used it. Make rt_maskedcopy() easier to read by using meaningful variable names. Extract a subroutine intern_netmask() for looking up a netmask in the masks table. Start converting backslash-ridden IPv6 macros in sys/netinet6/in6_var.h into inline subroutines that one can read without special eyeglasses. One functional change: when the kernel serves an RTM_GET, RTM_LOCK, or RTM_CHANGE request, it applies the netmask (if supplied) to a destination before searching for it in the forwarding table. I have changed sys/netinet/ip_carp.c, carp_setroute(), to remove the unlawful radix_node knowledge. Apart from the changes to carp(4), netiso, ATM, and strip(4), I have run the changes on three nodes in my wireless routing testbed, which involves IPv4 + IPv6 dynamic routing acrobatics, and it's working beautifully so far.
2007-07-20 00:48:52 +04:00
rt_destroy(rt);
pool_put(&rtentry_pool, rt);
senderr(rc);
}
2008-05-14 00:18:11 +04:00
RT_DPRINTF("rt->_rt_key = %p\n", (void *)rt->_rt_key);
1993-03-21 12:45:37 +03:00
if (ifa->ifa_rtrequest)
ifa->ifa_rtrequest(req, rt, info);
2008-05-14 00:18:11 +04:00
RT_DPRINTF("rt->_rt_key = %p\n", (void *)rt->_rt_key);
1993-03-21 12:45:37 +03:00
if (ret_nrt) {
*ret_nrt = rt;
rt->rt_refcnt++;
}
if ((rt->rt_flags & RTF_CLONING) != 0) {
/* clean up any cloned children */
rtflushclone(dst->sa_family, rt);
}
Here are various changes designed to protect against bad IPv4 routing caused by stale route caches (struct route). Route caches are sprinkled throughout PCBs, the IP fast-forwarding table, and IP tunnel interfaces (gre, gif, stf). Stale IPv6 and ISO route caches will be treated by separate patches. Thank you to Christoph Badura for suggesting the general approach to invalidating route caches that I take here. Here are the details: Add hooks to struct domain for tracking and for invalidating each domain's route caches: dom_rtcache, dom_rtflush, and dom_rtflushall. Introduce helper subroutines, rtflush(ro) for invalidating a route cache, rtflushall(family) for invalidating all route caches in a routing domain, and rtcache(ro) for notifying the domain of a new cached route. Chain together all IPv4 route caches where ro_rt != NULL. Provide in_rtcache() for adding a route to the chain. Provide in_rtflush() and in_rtflushall() for invalidating IPv4 route caches. In in_rtflush(), set ro_rt to NULL, and remove the route from the chain. In in_rtflushall(), walk the chain and remove every route cache. In rtrequest1(), call rtflushall() to invalidate route caches when a route is added. In gif(4), discard the workaround for stale caches that involves expiring them every so often. Replace the pattern 'RTFREE(ro->ro_rt); ro->ro_rt = NULL;' with a call to rtflush(ro). Update ipflow_fastforward() and all other users of route caches so that they expect a cached route, ro->ro_rt, to turn to NULL. Take care when moving a 'struct route' to rtflush() the source and to rtcache() the destination. In domain initializers, use .dom_xxx tags. KNF here and there.
2006-12-09 08:33:04 +03:00
rtflushall(dst->sa_family);
1993-03-21 12:45:37 +03:00
break;
case RTM_GET:
Take steps to hide the radix_node implementation of the forwarding table from the forwarding table's users: Introduce rt_walktree() for walking the routing table and applying a function to each rtentry. Replace most rn_walktree() calls with it. Use rt_getkey()/rt_setkey() to get/set a route's destination. Keep a pointer to the sockaddr key in the rtentry, so that rtentry users do not have to grovel in the radix_node for the key. Add a RTM_GET method to rtrequest. Use that instead of radix_node lookups in, e.g., carp(4). Add sys/net/link_proto.c, which supplies sockaddr routines for link-layer socket addresses (sockaddr_dl). Cosmetic: Constify. KNF. Stop open-coding LIST_FOREACH, TAILQ_FOREACH, et cetera. Use NULL instead of 0 for null pointers. Use __arraycount(). Reduce gratuitous parenthesization. Stop using variadic arguments for rip6_output(), it is unnecessary. Remove the unnecessary rtentry member rt_genmask and the code to maintain it, since nothing actually used it. Make rt_maskedcopy() easier to read by using meaningful variable names. Extract a subroutine intern_netmask() for looking up a netmask in the masks table. Start converting backslash-ridden IPv6 macros in sys/netinet6/in6_var.h into inline subroutines that one can read without special eyeglasses. One functional change: when the kernel serves an RTM_GET, RTM_LOCK, or RTM_CHANGE request, it applies the netmask (if supplied) to a destination before searching for it in the forwarding table. I have changed sys/netinet/ip_carp.c, carp_setroute(), to remove the unlawful radix_node knowledge. Apart from the changes to carp(4), netiso, ATM, and strip(4), I have run the changes on three nodes in my wireless routing testbed, which involves IPv4 + IPv6 dynamic routing acrobatics, and it's working beautifully so far.
2007-07-20 00:48:52 +04:00
if (netmask != NULL) {
rt_maskedcopy(dst, (struct sockaddr *)&maskeddst,
netmask);
dst = (struct sockaddr *)&maskeddst;
}
if ((rt = rt_lookup(rtbl, dst, netmask)) == NULL)
senderr(ESRCH);
if (ret_nrt != NULL) {
*ret_nrt = rt;
rt->rt_refcnt++;
}
break;
1993-03-21 12:45:37 +03:00
}
bad:
splx(s);
return error;
1993-03-21 12:45:37 +03:00
}
int
Take steps to hide the radix_node implementation of the forwarding table from the forwarding table's users: Introduce rt_walktree() for walking the routing table and applying a function to each rtentry. Replace most rn_walktree() calls with it. Use rt_getkey()/rt_setkey() to get/set a route's destination. Keep a pointer to the sockaddr key in the rtentry, so that rtentry users do not have to grovel in the radix_node for the key. Add a RTM_GET method to rtrequest. Use that instead of radix_node lookups in, e.g., carp(4). Add sys/net/link_proto.c, which supplies sockaddr routines for link-layer socket addresses (sockaddr_dl). Cosmetic: Constify. KNF. Stop open-coding LIST_FOREACH, TAILQ_FOREACH, et cetera. Use NULL instead of 0 for null pointers. Use __arraycount(). Reduce gratuitous parenthesization. Stop using variadic arguments for rip6_output(), it is unnecessary. Remove the unnecessary rtentry member rt_genmask and the code to maintain it, since nothing actually used it. Make rt_maskedcopy() easier to read by using meaningful variable names. Extract a subroutine intern_netmask() for looking up a netmask in the masks table. Start converting backslash-ridden IPv6 macros in sys/netinet6/in6_var.h into inline subroutines that one can read without special eyeglasses. One functional change: when the kernel serves an RTM_GET, RTM_LOCK, or RTM_CHANGE request, it applies the netmask (if supplied) to a destination before searching for it in the forwarding table. I have changed sys/netinet/ip_carp.c, carp_setroute(), to remove the unlawful radix_node knowledge. Apart from the changes to carp(4), netiso, ATM, and strip(4), I have run the changes on three nodes in my wireless routing testbed, which involves IPv4 + IPv6 dynamic routing acrobatics, and it's working beautifully so far.
2007-07-20 00:48:52 +04:00
rt_setgate(struct rtentry *rt, const struct sockaddr *gate)
{
Take steps to hide the radix_node implementation of the forwarding table from the forwarding table's users: Introduce rt_walktree() for walking the routing table and applying a function to each rtentry. Replace most rn_walktree() calls with it. Use rt_getkey()/rt_setkey() to get/set a route's destination. Keep a pointer to the sockaddr key in the rtentry, so that rtentry users do not have to grovel in the radix_node for the key. Add a RTM_GET method to rtrequest. Use that instead of radix_node lookups in, e.g., carp(4). Add sys/net/link_proto.c, which supplies sockaddr routines for link-layer socket addresses (sockaddr_dl). Cosmetic: Constify. KNF. Stop open-coding LIST_FOREACH, TAILQ_FOREACH, et cetera. Use NULL instead of 0 for null pointers. Use __arraycount(). Reduce gratuitous parenthesization. Stop using variadic arguments for rip6_output(), it is unnecessary. Remove the unnecessary rtentry member rt_genmask and the code to maintain it, since nothing actually used it. Make rt_maskedcopy() easier to read by using meaningful variable names. Extract a subroutine intern_netmask() for looking up a netmask in the masks table. Start converting backslash-ridden IPv6 macros in sys/netinet6/in6_var.h into inline subroutines that one can read without special eyeglasses. One functional change: when the kernel serves an RTM_GET, RTM_LOCK, or RTM_CHANGE request, it applies the netmask (if supplied) to a destination before searching for it in the forwarding table. I have changed sys/netinet/ip_carp.c, carp_setroute(), to remove the unlawful radix_node knowledge. Apart from the changes to carp(4), netiso, ATM, and strip(4), I have run the changes on three nodes in my wireless routing testbed, which involves IPv4 + IPv6 dynamic routing acrobatics, and it's working beautifully so far.
2007-07-20 00:48:52 +04:00
KASSERT(rt != rt->rt_gwroute);
KASSERT(rt->_rt_key != NULL);
2008-05-14 00:18:11 +04:00
RT_DPRINTF("rt->_rt_key = %p\n", (void *)rt->_rt_key);
Take steps to hide the radix_node implementation of the forwarding table from the forwarding table's users: Introduce rt_walktree() for walking the routing table and applying a function to each rtentry. Replace most rn_walktree() calls with it. Use rt_getkey()/rt_setkey() to get/set a route's destination. Keep a pointer to the sockaddr key in the rtentry, so that rtentry users do not have to grovel in the radix_node for the key. Add a RTM_GET method to rtrequest. Use that instead of radix_node lookups in, e.g., carp(4). Add sys/net/link_proto.c, which supplies sockaddr routines for link-layer socket addresses (sockaddr_dl). Cosmetic: Constify. KNF. Stop open-coding LIST_FOREACH, TAILQ_FOREACH, et cetera. Use NULL instead of 0 for null pointers. Use __arraycount(). Reduce gratuitous parenthesization. Stop using variadic arguments for rip6_output(), it is unnecessary. Remove the unnecessary rtentry member rt_genmask and the code to maintain it, since nothing actually used it. Make rt_maskedcopy() easier to read by using meaningful variable names. Extract a subroutine intern_netmask() for looking up a netmask in the masks table. Start converting backslash-ridden IPv6 macros in sys/netinet6/in6_var.h into inline subroutines that one can read without special eyeglasses. One functional change: when the kernel serves an RTM_GET, RTM_LOCK, or RTM_CHANGE request, it applies the netmask (if supplied) to a destination before searching for it in the forwarding table. I have changed sys/netinet/ip_carp.c, carp_setroute(), to remove the unlawful radix_node knowledge. Apart from the changes to carp(4), netiso, ATM, and strip(4), I have run the changes on three nodes in my wireless routing testbed, which involves IPv4 + IPv6 dynamic routing acrobatics, and it's working beautifully so far.
2007-07-20 00:48:52 +04:00
if (rt->rt_gwroute) {
RTFREE(rt->rt_gwroute);
rt->rt_gwroute = NULL;
}
Take steps to hide the radix_node implementation of the forwarding table from the forwarding table's users: Introduce rt_walktree() for walking the routing table and applying a function to each rtentry. Replace most rn_walktree() calls with it. Use rt_getkey()/rt_setkey() to get/set a route's destination. Keep a pointer to the sockaddr key in the rtentry, so that rtentry users do not have to grovel in the radix_node for the key. Add a RTM_GET method to rtrequest. Use that instead of radix_node lookups in, e.g., carp(4). Add sys/net/link_proto.c, which supplies sockaddr routines for link-layer socket addresses (sockaddr_dl). Cosmetic: Constify. KNF. Stop open-coding LIST_FOREACH, TAILQ_FOREACH, et cetera. Use NULL instead of 0 for null pointers. Use __arraycount(). Reduce gratuitous parenthesization. Stop using variadic arguments for rip6_output(), it is unnecessary. Remove the unnecessary rtentry member rt_genmask and the code to maintain it, since nothing actually used it. Make rt_maskedcopy() easier to read by using meaningful variable names. Extract a subroutine intern_netmask() for looking up a netmask in the masks table. Start converting backslash-ridden IPv6 macros in sys/netinet6/in6_var.h into inline subroutines that one can read without special eyeglasses. One functional change: when the kernel serves an RTM_GET, RTM_LOCK, or RTM_CHANGE request, it applies the netmask (if supplied) to a destination before searching for it in the forwarding table. I have changed sys/netinet/ip_carp.c, carp_setroute(), to remove the unlawful radix_node knowledge. Apart from the changes to carp(4), netiso, ATM, and strip(4), I have run the changes on three nodes in my wireless routing testbed, which involves IPv4 + IPv6 dynamic routing acrobatics, and it's working beautifully so far.
2007-07-20 00:48:52 +04:00
KASSERT(rt->_rt_key != NULL);
2008-05-14 00:18:11 +04:00
RT_DPRINTF("rt->_rt_key = %p\n", (void *)rt->_rt_key);
Take steps to hide the radix_node implementation of the forwarding table from the forwarding table's users: Introduce rt_walktree() for walking the routing table and applying a function to each rtentry. Replace most rn_walktree() calls with it. Use rt_getkey()/rt_setkey() to get/set a route's destination. Keep a pointer to the sockaddr key in the rtentry, so that rtentry users do not have to grovel in the radix_node for the key. Add a RTM_GET method to rtrequest. Use that instead of radix_node lookups in, e.g., carp(4). Add sys/net/link_proto.c, which supplies sockaddr routines for link-layer socket addresses (sockaddr_dl). Cosmetic: Constify. KNF. Stop open-coding LIST_FOREACH, TAILQ_FOREACH, et cetera. Use NULL instead of 0 for null pointers. Use __arraycount(). Reduce gratuitous parenthesization. Stop using variadic arguments for rip6_output(), it is unnecessary. Remove the unnecessary rtentry member rt_genmask and the code to maintain it, since nothing actually used it. Make rt_maskedcopy() easier to read by using meaningful variable names. Extract a subroutine intern_netmask() for looking up a netmask in the masks table. Start converting backslash-ridden IPv6 macros in sys/netinet6/in6_var.h into inline subroutines that one can read without special eyeglasses. One functional change: when the kernel serves an RTM_GET, RTM_LOCK, or RTM_CHANGE request, it applies the netmask (if supplied) to a destination before searching for it in the forwarding table. I have changed sys/netinet/ip_carp.c, carp_setroute(), to remove the unlawful radix_node knowledge. Apart from the changes to carp(4), netiso, ATM, and strip(4), I have run the changes on three nodes in my wireless routing testbed, which involves IPv4 + IPv6 dynamic routing acrobatics, and it's working beautifully so far.
2007-07-20 00:48:52 +04:00
if (rt->rt_gateway != NULL)
sockaddr_free(rt->rt_gateway);
KASSERT(rt->_rt_key != NULL);
2008-05-14 00:18:11 +04:00
RT_DPRINTF("rt->_rt_key = %p\n", (void *)rt->_rt_key);
if ((rt->rt_gateway = sockaddr_dup(gate, M_NOWAIT)) == NULL)
Take steps to hide the radix_node implementation of the forwarding table from the forwarding table's users: Introduce rt_walktree() for walking the routing table and applying a function to each rtentry. Replace most rn_walktree() calls with it. Use rt_getkey()/rt_setkey() to get/set a route's destination. Keep a pointer to the sockaddr key in the rtentry, so that rtentry users do not have to grovel in the radix_node for the key. Add a RTM_GET method to rtrequest. Use that instead of radix_node lookups in, e.g., carp(4). Add sys/net/link_proto.c, which supplies sockaddr routines for link-layer socket addresses (sockaddr_dl). Cosmetic: Constify. KNF. Stop open-coding LIST_FOREACH, TAILQ_FOREACH, et cetera. Use NULL instead of 0 for null pointers. Use __arraycount(). Reduce gratuitous parenthesization. Stop using variadic arguments for rip6_output(), it is unnecessary. Remove the unnecessary rtentry member rt_genmask and the code to maintain it, since nothing actually used it. Make rt_maskedcopy() easier to read by using meaningful variable names. Extract a subroutine intern_netmask() for looking up a netmask in the masks table. Start converting backslash-ridden IPv6 macros in sys/netinet6/in6_var.h into inline subroutines that one can read without special eyeglasses. One functional change: when the kernel serves an RTM_GET, RTM_LOCK, or RTM_CHANGE request, it applies the netmask (if supplied) to a destination before searching for it in the forwarding table. I have changed sys/netinet/ip_carp.c, carp_setroute(), to remove the unlawful radix_node knowledge. Apart from the changes to carp(4), netiso, ATM, and strip(4), I have run the changes on three nodes in my wireless routing testbed, which involves IPv4 + IPv6 dynamic routing acrobatics, and it's working beautifully so far.
2007-07-20 00:48:52 +04:00
return ENOMEM;
KASSERT(rt->_rt_key != NULL);
2008-05-14 00:18:11 +04:00
RT_DPRINTF("rt->_rt_key = %p\n", (void *)rt->_rt_key);
Take steps to hide the radix_node implementation of the forwarding table from the forwarding table's users: Introduce rt_walktree() for walking the routing table and applying a function to each rtentry. Replace most rn_walktree() calls with it. Use rt_getkey()/rt_setkey() to get/set a route's destination. Keep a pointer to the sockaddr key in the rtentry, so that rtentry users do not have to grovel in the radix_node for the key. Add a RTM_GET method to rtrequest. Use that instead of radix_node lookups in, e.g., carp(4). Add sys/net/link_proto.c, which supplies sockaddr routines for link-layer socket addresses (sockaddr_dl). Cosmetic: Constify. KNF. Stop open-coding LIST_FOREACH, TAILQ_FOREACH, et cetera. Use NULL instead of 0 for null pointers. Use __arraycount(). Reduce gratuitous parenthesization. Stop using variadic arguments for rip6_output(), it is unnecessary. Remove the unnecessary rtentry member rt_genmask and the code to maintain it, since nothing actually used it. Make rt_maskedcopy() easier to read by using meaningful variable names. Extract a subroutine intern_netmask() for looking up a netmask in the masks table. Start converting backslash-ridden IPv6 macros in sys/netinet6/in6_var.h into inline subroutines that one can read without special eyeglasses. One functional change: when the kernel serves an RTM_GET, RTM_LOCK, or RTM_CHANGE request, it applies the netmask (if supplied) to a destination before searching for it in the forwarding table. I have changed sys/netinet/ip_carp.c, carp_setroute(), to remove the unlawful radix_node knowledge. Apart from the changes to carp(4), netiso, ATM, and strip(4), I have run the changes on three nodes in my wireless routing testbed, which involves IPv4 + IPv6 dynamic routing acrobatics, and it's working beautifully so far.
2007-07-20 00:48:52 +04:00
if (rt->rt_flags & RTF_GATEWAY) {
Take steps to hide the radix_node implementation of the forwarding table from the forwarding table's users: Introduce rt_walktree() for walking the routing table and applying a function to each rtentry. Replace most rn_walktree() calls with it. Use rt_getkey()/rt_setkey() to get/set a route's destination. Keep a pointer to the sockaddr key in the rtentry, so that rtentry users do not have to grovel in the radix_node for the key. Add a RTM_GET method to rtrequest. Use that instead of radix_node lookups in, e.g., carp(4). Add sys/net/link_proto.c, which supplies sockaddr routines for link-layer socket addresses (sockaddr_dl). Cosmetic: Constify. KNF. Stop open-coding LIST_FOREACH, TAILQ_FOREACH, et cetera. Use NULL instead of 0 for null pointers. Use __arraycount(). Reduce gratuitous parenthesization. Stop using variadic arguments for rip6_output(), it is unnecessary. Remove the unnecessary rtentry member rt_genmask and the code to maintain it, since nothing actually used it. Make rt_maskedcopy() easier to read by using meaningful variable names. Extract a subroutine intern_netmask() for looking up a netmask in the masks table. Start converting backslash-ridden IPv6 macros in sys/netinet6/in6_var.h into inline subroutines that one can read without special eyeglasses. One functional change: when the kernel serves an RTM_GET, RTM_LOCK, or RTM_CHANGE request, it applies the netmask (if supplied) to a destination before searching for it in the forwarding table. I have changed sys/netinet/ip_carp.c, carp_setroute(), to remove the unlawful radix_node knowledge. Apart from the changes to carp(4), netiso, ATM, and strip(4), I have run the changes on three nodes in my wireless routing testbed, which involves IPv4 + IPv6 dynamic routing acrobatics, and it's working beautifully so far.
2007-07-20 00:48:52 +04:00
KASSERT(rt->_rt_key != NULL);
2008-05-14 00:18:11 +04:00
RT_DPRINTF("rt->_rt_key = %p\n", (void *)rt->_rt_key);
rt->rt_gwroute = rtalloc1(gate, 1);
/*
* If we switched gateways, grab the MTU from the new
* gateway route if the current MTU, if the current MTU is
* greater than the MTU of gateway.
* Note that, if the MTU of gateway is 0, we will reset the
* MTU of the route to run PMTUD again from scratch. XXX
*/
Take steps to hide the radix_node implementation of the forwarding table from the forwarding table's users: Introduce rt_walktree() for walking the routing table and applying a function to each rtentry. Replace most rn_walktree() calls with it. Use rt_getkey()/rt_setkey() to get/set a route's destination. Keep a pointer to the sockaddr key in the rtentry, so that rtentry users do not have to grovel in the radix_node for the key. Add a RTM_GET method to rtrequest. Use that instead of radix_node lookups in, e.g., carp(4). Add sys/net/link_proto.c, which supplies sockaddr routines for link-layer socket addresses (sockaddr_dl). Cosmetic: Constify. KNF. Stop open-coding LIST_FOREACH, TAILQ_FOREACH, et cetera. Use NULL instead of 0 for null pointers. Use __arraycount(). Reduce gratuitous parenthesization. Stop using variadic arguments for rip6_output(), it is unnecessary. Remove the unnecessary rtentry member rt_genmask and the code to maintain it, since nothing actually used it. Make rt_maskedcopy() easier to read by using meaningful variable names. Extract a subroutine intern_netmask() for looking up a netmask in the masks table. Start converting backslash-ridden IPv6 macros in sys/netinet6/in6_var.h into inline subroutines that one can read without special eyeglasses. One functional change: when the kernel serves an RTM_GET, RTM_LOCK, or RTM_CHANGE request, it applies the netmask (if supplied) to a destination before searching for it in the forwarding table. I have changed sys/netinet/ip_carp.c, carp_setroute(), to remove the unlawful radix_node knowledge. Apart from the changes to carp(4), netiso, ATM, and strip(4), I have run the changes on three nodes in my wireless routing testbed, which involves IPv4 + IPv6 dynamic routing acrobatics, and it's working beautifully so far.
2007-07-20 00:48:52 +04:00
KASSERT(rt->_rt_key != NULL);
2008-05-14 00:18:11 +04:00
RT_DPRINTF("rt->_rt_key = %p\n", (void *)rt->_rt_key);
if (rt->rt_gwroute
&& !(rt->rt_rmx.rmx_locks & RTV_MTU)
&& rt->rt_rmx.rmx_mtu
&& rt->rt_rmx.rmx_mtu > rt->rt_gwroute->rt_rmx.rmx_mtu) {
rt->rt_rmx.rmx_mtu = rt->rt_gwroute->rt_rmx.rmx_mtu;
}
}
Take steps to hide the radix_node implementation of the forwarding table from the forwarding table's users: Introduce rt_walktree() for walking the routing table and applying a function to each rtentry. Replace most rn_walktree() calls with it. Use rt_getkey()/rt_setkey() to get/set a route's destination. Keep a pointer to the sockaddr key in the rtentry, so that rtentry users do not have to grovel in the radix_node for the key. Add a RTM_GET method to rtrequest. Use that instead of radix_node lookups in, e.g., carp(4). Add sys/net/link_proto.c, which supplies sockaddr routines for link-layer socket addresses (sockaddr_dl). Cosmetic: Constify. KNF. Stop open-coding LIST_FOREACH, TAILQ_FOREACH, et cetera. Use NULL instead of 0 for null pointers. Use __arraycount(). Reduce gratuitous parenthesization. Stop using variadic arguments for rip6_output(), it is unnecessary. Remove the unnecessary rtentry member rt_genmask and the code to maintain it, since nothing actually used it. Make rt_maskedcopy() easier to read by using meaningful variable names. Extract a subroutine intern_netmask() for looking up a netmask in the masks table. Start converting backslash-ridden IPv6 macros in sys/netinet6/in6_var.h into inline subroutines that one can read without special eyeglasses. One functional change: when the kernel serves an RTM_GET, RTM_LOCK, or RTM_CHANGE request, it applies the netmask (if supplied) to a destination before searching for it in the forwarding table. I have changed sys/netinet/ip_carp.c, carp_setroute(), to remove the unlawful radix_node knowledge. Apart from the changes to carp(4), netiso, ATM, and strip(4), I have run the changes on three nodes in my wireless routing testbed, which involves IPv4 + IPv6 dynamic routing acrobatics, and it's working beautifully so far.
2007-07-20 00:48:52 +04:00
KASSERT(rt->_rt_key != NULL);
2008-05-14 00:18:11 +04:00
RT_DPRINTF("rt->_rt_key = %p\n", (void *)rt->_rt_key);
return 0;
}
void
rt_maskedcopy(const struct sockaddr *src, struct sockaddr *dst,
const struct sockaddr *netmask)
1993-03-21 12:45:37 +03:00
{
Take steps to hide the radix_node implementation of the forwarding table from the forwarding table's users: Introduce rt_walktree() for walking the routing table and applying a function to each rtentry. Replace most rn_walktree() calls with it. Use rt_getkey()/rt_setkey() to get/set a route's destination. Keep a pointer to the sockaddr key in the rtentry, so that rtentry users do not have to grovel in the radix_node for the key. Add a RTM_GET method to rtrequest. Use that instead of radix_node lookups in, e.g., carp(4). Add sys/net/link_proto.c, which supplies sockaddr routines for link-layer socket addresses (sockaddr_dl). Cosmetic: Constify. KNF. Stop open-coding LIST_FOREACH, TAILQ_FOREACH, et cetera. Use NULL instead of 0 for null pointers. Use __arraycount(). Reduce gratuitous parenthesization. Stop using variadic arguments for rip6_output(), it is unnecessary. Remove the unnecessary rtentry member rt_genmask and the code to maintain it, since nothing actually used it. Make rt_maskedcopy() easier to read by using meaningful variable names. Extract a subroutine intern_netmask() for looking up a netmask in the masks table. Start converting backslash-ridden IPv6 macros in sys/netinet6/in6_var.h into inline subroutines that one can read without special eyeglasses. One functional change: when the kernel serves an RTM_GET, RTM_LOCK, or RTM_CHANGE request, it applies the netmask (if supplied) to a destination before searching for it in the forwarding table. I have changed sys/netinet/ip_carp.c, carp_setroute(), to remove the unlawful radix_node knowledge. Apart from the changes to carp(4), netiso, ATM, and strip(4), I have run the changes on three nodes in my wireless routing testbed, which involves IPv4 + IPv6 dynamic routing acrobatics, and it's working beautifully so far.
2007-07-20 00:48:52 +04:00
const char *netmaskp = &netmask->sa_data[0],
*srcp = &src->sa_data[0];
char *dstp = &dst->sa_data[0];
const char *maskend = (char *)dst + MIN(netmask->sa_len, src->sa_len);
const char *srcend = (char *)dst + src->sa_len;
Take steps to hide the radix_node implementation of the forwarding table from the forwarding table's users: Introduce rt_walktree() for walking the routing table and applying a function to each rtentry. Replace most rn_walktree() calls with it. Use rt_getkey()/rt_setkey() to get/set a route's destination. Keep a pointer to the sockaddr key in the rtentry, so that rtentry users do not have to grovel in the radix_node for the key. Add a RTM_GET method to rtrequest. Use that instead of radix_node lookups in, e.g., carp(4). Add sys/net/link_proto.c, which supplies sockaddr routines for link-layer socket addresses (sockaddr_dl). Cosmetic: Constify. KNF. Stop open-coding LIST_FOREACH, TAILQ_FOREACH, et cetera. Use NULL instead of 0 for null pointers. Use __arraycount(). Reduce gratuitous parenthesization. Stop using variadic arguments for rip6_output(), it is unnecessary. Remove the unnecessary rtentry member rt_genmask and the code to maintain it, since nothing actually used it. Make rt_maskedcopy() easier to read by using meaningful variable names. Extract a subroutine intern_netmask() for looking up a netmask in the masks table. Start converting backslash-ridden IPv6 macros in sys/netinet6/in6_var.h into inline subroutines that one can read without special eyeglasses. One functional change: when the kernel serves an RTM_GET, RTM_LOCK, or RTM_CHANGE request, it applies the netmask (if supplied) to a destination before searching for it in the forwarding table. I have changed sys/netinet/ip_carp.c, carp_setroute(), to remove the unlawful radix_node knowledge. Apart from the changes to carp(4), netiso, ATM, and strip(4), I have run the changes on three nodes in my wireless routing testbed, which involves IPv4 + IPv6 dynamic routing acrobatics, and it's working beautifully so far.
2007-07-20 00:48:52 +04:00
dst->sa_len = src->sa_len;
dst->sa_family = src->sa_family;
while (dstp < maskend)
*dstp++ = *srcp++ & *netmaskp++;
if (dstp < srcend)
memset(dstp, 0, (size_t)(srcend - dstp));
1993-03-21 12:45:37 +03:00
}
1993-03-21 12:45:37 +03:00
/*
* Set up or tear down a routing table entry, normally
1993-03-21 12:45:37 +03:00
* for an interface.
*/
int
rtinit(struct ifaddr *ifa, int cmd, int flags)
1993-03-21 12:45:37 +03:00
{
2000-03-30 13:45:33 +04:00
struct rtentry *rt;
struct sockaddr *dst, *odst;
Take steps to hide the radix_node implementation of the forwarding table from the forwarding table's users: Introduce rt_walktree() for walking the routing table and applying a function to each rtentry. Replace most rn_walktree() calls with it. Use rt_getkey()/rt_setkey() to get/set a route's destination. Keep a pointer to the sockaddr key in the rtentry, so that rtentry users do not have to grovel in the radix_node for the key. Add a RTM_GET method to rtrequest. Use that instead of radix_node lookups in, e.g., carp(4). Add sys/net/link_proto.c, which supplies sockaddr routines for link-layer socket addresses (sockaddr_dl). Cosmetic: Constify. KNF. Stop open-coding LIST_FOREACH, TAILQ_FOREACH, et cetera. Use NULL instead of 0 for null pointers. Use __arraycount(). Reduce gratuitous parenthesization. Stop using variadic arguments for rip6_output(), it is unnecessary. Remove the unnecessary rtentry member rt_genmask and the code to maintain it, since nothing actually used it. Make rt_maskedcopy() easier to read by using meaningful variable names. Extract a subroutine intern_netmask() for looking up a netmask in the masks table. Start converting backslash-ridden IPv6 macros in sys/netinet6/in6_var.h into inline subroutines that one can read without special eyeglasses. One functional change: when the kernel serves an RTM_GET, RTM_LOCK, or RTM_CHANGE request, it applies the netmask (if supplied) to a destination before searching for it in the forwarding table. I have changed sys/netinet/ip_carp.c, carp_setroute(), to remove the unlawful radix_node knowledge. Apart from the changes to carp(4), netiso, ATM, and strip(4), I have run the changes on three nodes in my wireless routing testbed, which involves IPv4 + IPv6 dynamic routing acrobatics, and it's working beautifully so far.
2007-07-20 00:48:52 +04:00
struct sockaddr_storage maskeddst;
struct rtentry *nrt = NULL;
1993-03-21 12:45:37 +03:00
int error;
struct rt_addrinfo info;
*** Summary *** When a link-layer address changes (e.g., ifconfig ex0 link 02:de:ad:be:ef:02 active), send a gratuitous ARP and/or a Neighbor Advertisement to update the network-/link-layer address bindings on our LAN peers. Refuse a change of ethernet address to the address 00:00:00:00:00:00 or to any multicast/broadcast address. (Thanks matt@.) Reorder ifnet ioctl operations so that driver ioctls may inherit the functions of their "class"---ether_ioctl(), fddi_ioctl(), et cetera---and the class ioctls may inherit from the generic ioctl, ifioctl_common(), but both driver- and class-ioctls may override the generic behavior. Make network drivers share more code. Distinguish a "factory" link-layer address from others for the purposes of both protecting that address from deletion and computing EUI64. Return consistent, appropriate error codes from network drivers. Improve readability. KNF. *** Details *** In if_attach(), always initialize the interface ioctl routine, ifnet->if_ioctl, if the driver has not already initialized it. Delete if_ioctl == NULL tests everywhere else, because it cannot happen. In the ioctl routines of network interfaces, inherit common ioctl behaviors by calling either ifioctl_common() or whichever ioctl routine is appropriate for the class of interface---e.g., ether_ioctl() for ethernets. Stop (ab)using SIOCSIFADDR and start to use SIOCINITIFADDR. In the user->kernel interface, SIOCSIFADDR's argument was an ifreq, but on the protocol->ifnet interface, SIOCSIFADDR's argument was an ifaddr. That was confusing, and it would work against me as I make it possible for a network interface to overload most ioctls. On the protocol->ifnet interface, replace SIOCSIFADDR with SIOCINITIFADDR. In ifioctl(), return EPERM if userland tries to invoke SIOCINITIFADDR. In ifioctl(), give the interface the first shot at handling most interface ioctls, and give the protocol the second shot, instead of the other way around. Finally, let compatibility code (COMPAT_OSOCK) take a shot. Pull device initialization out of switch statements under SIOCINITIFADDR. For example, pull ..._init() out of any switch statement that looks like this: switch (...->sa_family) { case ...: ..._init(); ... break; ... default: ..._init(); ... break; } Rewrite many if-else clauses that handle all permutations of IFF_UP and IFF_RUNNING to use a switch statement, switch (x & (IFF_UP|IFF_RUNNING)) { case 0: ... break; case IFF_RUNNING: ... break; case IFF_UP: ... break; case IFF_UP|IFF_RUNNING: ... break; } unifdef lots of code containing #ifdef FreeBSD, #ifdef NetBSD, and #ifdef SIOCSIFMTU, especially in fwip(4) and in ndis(4). In ipw(4), remove an if_set_sadl() call that is out of place. In nfe(4), reuse the jumbo MTU logic in ether_ioctl(). Let ethernets register a callback for setting h/w state such as promiscuous mode and the multicast filter in accord with a change in the if_flags: ether_set_ifflags_cb() registers a callback that returns ENETRESET if the caller should reset the ethernet by calling if_init(), 0 on success, != 0 on failure. Pull common code from ex(4), gem(4), nfe(4), sip(4), tlp(4), vge(4) into ether_ioctl(), and register if_flags callbacks for those drivers. Return ENOTTY instead of EINVAL for inappropriate ioctls. In zyd(4), use ENXIO instead of ENOTTY to indicate that the device is not any longer attached. Add to if_set_sadl() a boolean 'factory' argument that indicates whether a link-layer address was assigned by the factory or some other source. In a comment, recommend using the factory address for generating an EUI64, and update in6_get_hw_ifid() to prefer a factory address to any other link-layer address. Add a routing message, RTM_LLINFO_UPD, that tells protocols to update the binding of network-layer addresses to link-layer addresses. Implement this message in IPv4 and IPv6 by sending a gratuitous ARP or a neighbor advertisement, respectively. Generate RTM_LLINFO_UPD messages on a change of an interface's link-layer address. In ether_ioctl(), do not let SIOCALIFADDR set a link-layer address that is broadcast/multicast or equal to 00:00:00:00:00:00. Make ether_ioctl() call ifioctl_common() to handle ioctls that it does not understand. In gif(4), initialize if_softc and use it, instead of assuming that the gif_softc and ifp overlap. Let ifioctl_common() handle SIOCGIFADDR. Sprinkle rtcache_invariants(), which checks on DIAGNOSTIC kernels that certain invariants on a struct route are satisfied. In agr(4), rewrite agr_ioctl_filter() to be a bit more explicit about the ioctls that we do not allow on an agr(4) member interface. bzero -> memset. Delete unnecessary casts to void *. Use sockaddr_in_init() and sockaddr_in6_init(). Compare pointers with NULL instead of "testing truth". Replace some instances of (type *)0 with NULL. Change some K&R prototypes to ANSI C, and join lines.
2008-11-07 03:20:01 +03:00
struct sockaddr_dl *sdl;
const struct sockaddr_dl *ifsdl;
1993-03-21 12:45:37 +03:00
dst = flags & RTF_HOST ? ifa->ifa_dstaddr : ifa->ifa_addr;
if (cmd == RTM_DELETE) {
if ((flags & RTF_HOST) == 0 && ifa->ifa_netmask) {
/* Delete subnet route for this interface */
odst = dst;
Take steps to hide the radix_node implementation of the forwarding table from the forwarding table's users: Introduce rt_walktree() for walking the routing table and applying a function to each rtentry. Replace most rn_walktree() calls with it. Use rt_getkey()/rt_setkey() to get/set a route's destination. Keep a pointer to the sockaddr key in the rtentry, so that rtentry users do not have to grovel in the radix_node for the key. Add a RTM_GET method to rtrequest. Use that instead of radix_node lookups in, e.g., carp(4). Add sys/net/link_proto.c, which supplies sockaddr routines for link-layer socket addresses (sockaddr_dl). Cosmetic: Constify. KNF. Stop open-coding LIST_FOREACH, TAILQ_FOREACH, et cetera. Use NULL instead of 0 for null pointers. Use __arraycount(). Reduce gratuitous parenthesization. Stop using variadic arguments for rip6_output(), it is unnecessary. Remove the unnecessary rtentry member rt_genmask and the code to maintain it, since nothing actually used it. Make rt_maskedcopy() easier to read by using meaningful variable names. Extract a subroutine intern_netmask() for looking up a netmask in the masks table. Start converting backslash-ridden IPv6 macros in sys/netinet6/in6_var.h into inline subroutines that one can read without special eyeglasses. One functional change: when the kernel serves an RTM_GET, RTM_LOCK, or RTM_CHANGE request, it applies the netmask (if supplied) to a destination before searching for it in the forwarding table. I have changed sys/netinet/ip_carp.c, carp_setroute(), to remove the unlawful radix_node knowledge. Apart from the changes to carp(4), netiso, ATM, and strip(4), I have run the changes on three nodes in my wireless routing testbed, which involves IPv4 + IPv6 dynamic routing acrobatics, and it's working beautifully so far.
2007-07-20 00:48:52 +04:00
dst = (struct sockaddr *)&maskeddst;
rt_maskedcopy(odst, dst, ifa->ifa_netmask);
1993-03-21 12:45:37 +03:00
}
1996-02-14 00:59:53 +03:00
if ((rt = rtalloc1(dst, 0)) != NULL) {
1993-03-21 12:45:37 +03:00
rt->rt_refcnt--;
if (rt->rt_ifa != ifa)
return (flags & RTF_HOST) ? EHOSTUNREACH
: ENETUNREACH;
1993-03-21 12:45:37 +03:00
}
}
2001-07-18 20:43:09 +04:00
memset(&info, 0, sizeof(info));
info.rti_ifa = ifa;
info.rti_flags = flags | ifa->ifa_flags;
info.rti_info[RTAX_DST] = dst;
info.rti_info[RTAX_GATEWAY] = ifa->ifa_addr;
/*
* XXX here, it seems that we are assuming that ifa_netmask is NULL
* for RTF_HOST. bsdi4 passes NULL explicitly (via intermediate
* variable) when RTF_HOST is 1. still not sure if i can safely
* change it to meet bsdi4 behavior.
*/
*** Summary *** When a link-layer address changes (e.g., ifconfig ex0 link 02:de:ad:be:ef:02 active), send a gratuitous ARP and/or a Neighbor Advertisement to update the network-/link-layer address bindings on our LAN peers. Refuse a change of ethernet address to the address 00:00:00:00:00:00 or to any multicast/broadcast address. (Thanks matt@.) Reorder ifnet ioctl operations so that driver ioctls may inherit the functions of their "class"---ether_ioctl(), fddi_ioctl(), et cetera---and the class ioctls may inherit from the generic ioctl, ifioctl_common(), but both driver- and class-ioctls may override the generic behavior. Make network drivers share more code. Distinguish a "factory" link-layer address from others for the purposes of both protecting that address from deletion and computing EUI64. Return consistent, appropriate error codes from network drivers. Improve readability. KNF. *** Details *** In if_attach(), always initialize the interface ioctl routine, ifnet->if_ioctl, if the driver has not already initialized it. Delete if_ioctl == NULL tests everywhere else, because it cannot happen. In the ioctl routines of network interfaces, inherit common ioctl behaviors by calling either ifioctl_common() or whichever ioctl routine is appropriate for the class of interface---e.g., ether_ioctl() for ethernets. Stop (ab)using SIOCSIFADDR and start to use SIOCINITIFADDR. In the user->kernel interface, SIOCSIFADDR's argument was an ifreq, but on the protocol->ifnet interface, SIOCSIFADDR's argument was an ifaddr. That was confusing, and it would work against me as I make it possible for a network interface to overload most ioctls. On the protocol->ifnet interface, replace SIOCSIFADDR with SIOCINITIFADDR. In ifioctl(), return EPERM if userland tries to invoke SIOCINITIFADDR. In ifioctl(), give the interface the first shot at handling most interface ioctls, and give the protocol the second shot, instead of the other way around. Finally, let compatibility code (COMPAT_OSOCK) take a shot. Pull device initialization out of switch statements under SIOCINITIFADDR. For example, pull ..._init() out of any switch statement that looks like this: switch (...->sa_family) { case ...: ..._init(); ... break; ... default: ..._init(); ... break; } Rewrite many if-else clauses that handle all permutations of IFF_UP and IFF_RUNNING to use a switch statement, switch (x & (IFF_UP|IFF_RUNNING)) { case 0: ... break; case IFF_RUNNING: ... break; case IFF_UP: ... break; case IFF_UP|IFF_RUNNING: ... break; } unifdef lots of code containing #ifdef FreeBSD, #ifdef NetBSD, and #ifdef SIOCSIFMTU, especially in fwip(4) and in ndis(4). In ipw(4), remove an if_set_sadl() call that is out of place. In nfe(4), reuse the jumbo MTU logic in ether_ioctl(). Let ethernets register a callback for setting h/w state such as promiscuous mode and the multicast filter in accord with a change in the if_flags: ether_set_ifflags_cb() registers a callback that returns ENETRESET if the caller should reset the ethernet by calling if_init(), 0 on success, != 0 on failure. Pull common code from ex(4), gem(4), nfe(4), sip(4), tlp(4), vge(4) into ether_ioctl(), and register if_flags callbacks for those drivers. Return ENOTTY instead of EINVAL for inappropriate ioctls. In zyd(4), use ENXIO instead of ENOTTY to indicate that the device is not any longer attached. Add to if_set_sadl() a boolean 'factory' argument that indicates whether a link-layer address was assigned by the factory or some other source. In a comment, recommend using the factory address for generating an EUI64, and update in6_get_hw_ifid() to prefer a factory address to any other link-layer address. Add a routing message, RTM_LLINFO_UPD, that tells protocols to update the binding of network-layer addresses to link-layer addresses. Implement this message in IPv4 and IPv6 by sending a gratuitous ARP or a neighbor advertisement, respectively. Generate RTM_LLINFO_UPD messages on a change of an interface's link-layer address. In ether_ioctl(), do not let SIOCALIFADDR set a link-layer address that is broadcast/multicast or equal to 00:00:00:00:00:00. Make ether_ioctl() call ifioctl_common() to handle ioctls that it does not understand. In gif(4), initialize if_softc and use it, instead of assuming that the gif_softc and ifp overlap. Let ifioctl_common() handle SIOCGIFADDR. Sprinkle rtcache_invariants(), which checks on DIAGNOSTIC kernels that certain invariants on a struct route are satisfied. In agr(4), rewrite agr_ioctl_filter() to be a bit more explicit about the ioctls that we do not allow on an agr(4) member interface. bzero -> memset. Delete unnecessary casts to void *. Use sockaddr_in_init() and sockaddr_in6_init(). Compare pointers with NULL instead of "testing truth". Replace some instances of (type *)0 with NULL. Change some K&R prototypes to ANSI C, and join lines.
2008-11-07 03:20:01 +03:00
if (cmd != RTM_LLINFO_UPD)
info.rti_info[RTAX_NETMASK] = ifa->ifa_netmask;
error = rtrequest1((cmd == RTM_LLINFO_UPD) ? RTM_GET : cmd, &info,
&nrt);
if (error != 0 || (rt = nrt) == NULL)
;
else switch (cmd) {
case RTM_DELETE:
rt_newaddrmsg(cmd, ifa, error, nrt);
if (rt->rt_refcnt <= 0) {
rt->rt_refcnt++;
rtfree(rt);
}
*** Summary *** When a link-layer address changes (e.g., ifconfig ex0 link 02:de:ad:be:ef:02 active), send a gratuitous ARP and/or a Neighbor Advertisement to update the network-/link-layer address bindings on our LAN peers. Refuse a change of ethernet address to the address 00:00:00:00:00:00 or to any multicast/broadcast address. (Thanks matt@.) Reorder ifnet ioctl operations so that driver ioctls may inherit the functions of their "class"---ether_ioctl(), fddi_ioctl(), et cetera---and the class ioctls may inherit from the generic ioctl, ifioctl_common(), but both driver- and class-ioctls may override the generic behavior. Make network drivers share more code. Distinguish a "factory" link-layer address from others for the purposes of both protecting that address from deletion and computing EUI64. Return consistent, appropriate error codes from network drivers. Improve readability. KNF. *** Details *** In if_attach(), always initialize the interface ioctl routine, ifnet->if_ioctl, if the driver has not already initialized it. Delete if_ioctl == NULL tests everywhere else, because it cannot happen. In the ioctl routines of network interfaces, inherit common ioctl behaviors by calling either ifioctl_common() or whichever ioctl routine is appropriate for the class of interface---e.g., ether_ioctl() for ethernets. Stop (ab)using SIOCSIFADDR and start to use SIOCINITIFADDR. In the user->kernel interface, SIOCSIFADDR's argument was an ifreq, but on the protocol->ifnet interface, SIOCSIFADDR's argument was an ifaddr. That was confusing, and it would work against me as I make it possible for a network interface to overload most ioctls. On the protocol->ifnet interface, replace SIOCSIFADDR with SIOCINITIFADDR. In ifioctl(), return EPERM if userland tries to invoke SIOCINITIFADDR. In ifioctl(), give the interface the first shot at handling most interface ioctls, and give the protocol the second shot, instead of the other way around. Finally, let compatibility code (COMPAT_OSOCK) take a shot. Pull device initialization out of switch statements under SIOCINITIFADDR. For example, pull ..._init() out of any switch statement that looks like this: switch (...->sa_family) { case ...: ..._init(); ... break; ... default: ..._init(); ... break; } Rewrite many if-else clauses that handle all permutations of IFF_UP and IFF_RUNNING to use a switch statement, switch (x & (IFF_UP|IFF_RUNNING)) { case 0: ... break; case IFF_RUNNING: ... break; case IFF_UP: ... break; case IFF_UP|IFF_RUNNING: ... break; } unifdef lots of code containing #ifdef FreeBSD, #ifdef NetBSD, and #ifdef SIOCSIFMTU, especially in fwip(4) and in ndis(4). In ipw(4), remove an if_set_sadl() call that is out of place. In nfe(4), reuse the jumbo MTU logic in ether_ioctl(). Let ethernets register a callback for setting h/w state such as promiscuous mode and the multicast filter in accord with a change in the if_flags: ether_set_ifflags_cb() registers a callback that returns ENETRESET if the caller should reset the ethernet by calling if_init(), 0 on success, != 0 on failure. Pull common code from ex(4), gem(4), nfe(4), sip(4), tlp(4), vge(4) into ether_ioctl(), and register if_flags callbacks for those drivers. Return ENOTTY instead of EINVAL for inappropriate ioctls. In zyd(4), use ENXIO instead of ENOTTY to indicate that the device is not any longer attached. Add to if_set_sadl() a boolean 'factory' argument that indicates whether a link-layer address was assigned by the factory or some other source. In a comment, recommend using the factory address for generating an EUI64, and update in6_get_hw_ifid() to prefer a factory address to any other link-layer address. Add a routing message, RTM_LLINFO_UPD, that tells protocols to update the binding of network-layer addresses to link-layer addresses. Implement this message in IPv4 and IPv6 by sending a gratuitous ARP or a neighbor advertisement, respectively. Generate RTM_LLINFO_UPD messages on a change of an interface's link-layer address. In ether_ioctl(), do not let SIOCALIFADDR set a link-layer address that is broadcast/multicast or equal to 00:00:00:00:00:00. Make ether_ioctl() call ifioctl_common() to handle ioctls that it does not understand. In gif(4), initialize if_softc and use it, instead of assuming that the gif_softc and ifp overlap. Let ifioctl_common() handle SIOCGIFADDR. Sprinkle rtcache_invariants(), which checks on DIAGNOSTIC kernels that certain invariants on a struct route are satisfied. In agr(4), rewrite agr_ioctl_filter() to be a bit more explicit about the ioctls that we do not allow on an agr(4) member interface. bzero -> memset. Delete unnecessary casts to void *. Use sockaddr_in_init() and sockaddr_in6_init(). Compare pointers with NULL instead of "testing truth". Replace some instances of (type *)0 with NULL. Change some K&R prototypes to ANSI C, and join lines.
2008-11-07 03:20:01 +03:00
break;
case RTM_LLINFO_UPD:
rt->rt_refcnt--;
RT_DPRINTF("%s: updating%s\n", __func__,
((rt->rt_flags & RTF_LLINFO) == 0) ? " (no llinfo)" : "");
ifsdl = ifa->ifa_ifp->if_sadl;
if ((rt->rt_flags & RTF_LLINFO) != 0 &&
(sdl = satosdl(rt->rt_gateway)) != NULL &&
sdl->sdl_family == AF_LINK &&
sockaddr_dl_setaddr(sdl, sdl->sdl_len, CLLADDR(ifsdl),
ifa->ifa_ifp->if_addrlen) == NULL) {
error = EINVAL;
break;
}
if (cmd == RTM_LLINFO_UPD && ifa->ifa_rtrequest != NULL)
ifa->ifa_rtrequest(RTM_LLINFO_UPD, rt, &info);
rt_newaddrmsg(RTM_CHANGE, ifa, error, nrt);
break;
case RTM_ADD:
rt->rt_refcnt--;
if (rt->rt_ifa != ifa) {
1997-04-03 01:17:28 +04:00
printf("rtinit: wrong ifa (%p) was (%p)\n", ifa,
rt->rt_ifa);
*** Summary *** When a link-layer address changes (e.g., ifconfig ex0 link 02:de:ad:be:ef:02 active), send a gratuitous ARP and/or a Neighbor Advertisement to update the network-/link-layer address bindings on our LAN peers. Refuse a change of ethernet address to the address 00:00:00:00:00:00 or to any multicast/broadcast address. (Thanks matt@.) Reorder ifnet ioctl operations so that driver ioctls may inherit the functions of their "class"---ether_ioctl(), fddi_ioctl(), et cetera---and the class ioctls may inherit from the generic ioctl, ifioctl_common(), but both driver- and class-ioctls may override the generic behavior. Make network drivers share more code. Distinguish a "factory" link-layer address from others for the purposes of both protecting that address from deletion and computing EUI64. Return consistent, appropriate error codes from network drivers. Improve readability. KNF. *** Details *** In if_attach(), always initialize the interface ioctl routine, ifnet->if_ioctl, if the driver has not already initialized it. Delete if_ioctl == NULL tests everywhere else, because it cannot happen. In the ioctl routines of network interfaces, inherit common ioctl behaviors by calling either ifioctl_common() or whichever ioctl routine is appropriate for the class of interface---e.g., ether_ioctl() for ethernets. Stop (ab)using SIOCSIFADDR and start to use SIOCINITIFADDR. In the user->kernel interface, SIOCSIFADDR's argument was an ifreq, but on the protocol->ifnet interface, SIOCSIFADDR's argument was an ifaddr. That was confusing, and it would work against me as I make it possible for a network interface to overload most ioctls. On the protocol->ifnet interface, replace SIOCSIFADDR with SIOCINITIFADDR. In ifioctl(), return EPERM if userland tries to invoke SIOCINITIFADDR. In ifioctl(), give the interface the first shot at handling most interface ioctls, and give the protocol the second shot, instead of the other way around. Finally, let compatibility code (COMPAT_OSOCK) take a shot. Pull device initialization out of switch statements under SIOCINITIFADDR. For example, pull ..._init() out of any switch statement that looks like this: switch (...->sa_family) { case ...: ..._init(); ... break; ... default: ..._init(); ... break; } Rewrite many if-else clauses that handle all permutations of IFF_UP and IFF_RUNNING to use a switch statement, switch (x & (IFF_UP|IFF_RUNNING)) { case 0: ... break; case IFF_RUNNING: ... break; case IFF_UP: ... break; case IFF_UP|IFF_RUNNING: ... break; } unifdef lots of code containing #ifdef FreeBSD, #ifdef NetBSD, and #ifdef SIOCSIFMTU, especially in fwip(4) and in ndis(4). In ipw(4), remove an if_set_sadl() call that is out of place. In nfe(4), reuse the jumbo MTU logic in ether_ioctl(). Let ethernets register a callback for setting h/w state such as promiscuous mode and the multicast filter in accord with a change in the if_flags: ether_set_ifflags_cb() registers a callback that returns ENETRESET if the caller should reset the ethernet by calling if_init(), 0 on success, != 0 on failure. Pull common code from ex(4), gem(4), nfe(4), sip(4), tlp(4), vge(4) into ether_ioctl(), and register if_flags callbacks for those drivers. Return ENOTTY instead of EINVAL for inappropriate ioctls. In zyd(4), use ENXIO instead of ENOTTY to indicate that the device is not any longer attached. Add to if_set_sadl() a boolean 'factory' argument that indicates whether a link-layer address was assigned by the factory or some other source. In a comment, recommend using the factory address for generating an EUI64, and update in6_get_hw_ifid() to prefer a factory address to any other link-layer address. Add a routing message, RTM_LLINFO_UPD, that tells protocols to update the binding of network-layer addresses to link-layer addresses. Implement this message in IPv4 and IPv6 by sending a gratuitous ARP or a neighbor advertisement, respectively. Generate RTM_LLINFO_UPD messages on a change of an interface's link-layer address. In ether_ioctl(), do not let SIOCALIFADDR set a link-layer address that is broadcast/multicast or equal to 00:00:00:00:00:00. Make ether_ioctl() call ifioctl_common() to handle ioctls that it does not understand. In gif(4), initialize if_softc and use it, instead of assuming that the gif_softc and ifp overlap. Let ifioctl_common() handle SIOCGIFADDR. Sprinkle rtcache_invariants(), which checks on DIAGNOSTIC kernels that certain invariants on a struct route are satisfied. In agr(4), rewrite agr_ioctl_filter() to be a bit more explicit about the ioctls that we do not allow on an agr(4) member interface. bzero -> memset. Delete unnecessary casts to void *. Use sockaddr_in_init() and sockaddr_in6_init(). Compare pointers with NULL instead of "testing truth". Replace some instances of (type *)0 with NULL. Change some K&R prototypes to ANSI C, and join lines.
2008-11-07 03:20:01 +03:00
if (rt->rt_ifa->ifa_rtrequest != NULL) {
rt->rt_ifa->ifa_rtrequest(RTM_DELETE, rt,
&info);
}
rt_replace_ifa(rt, ifa);
rt->rt_ifp = ifa->ifa_ifp;
*** Summary *** When a link-layer address changes (e.g., ifconfig ex0 link 02:de:ad:be:ef:02 active), send a gratuitous ARP and/or a Neighbor Advertisement to update the network-/link-layer address bindings on our LAN peers. Refuse a change of ethernet address to the address 00:00:00:00:00:00 or to any multicast/broadcast address. (Thanks matt@.) Reorder ifnet ioctl operations so that driver ioctls may inherit the functions of their "class"---ether_ioctl(), fddi_ioctl(), et cetera---and the class ioctls may inherit from the generic ioctl, ifioctl_common(), but both driver- and class-ioctls may override the generic behavior. Make network drivers share more code. Distinguish a "factory" link-layer address from others for the purposes of both protecting that address from deletion and computing EUI64. Return consistent, appropriate error codes from network drivers. Improve readability. KNF. *** Details *** In if_attach(), always initialize the interface ioctl routine, ifnet->if_ioctl, if the driver has not already initialized it. Delete if_ioctl == NULL tests everywhere else, because it cannot happen. In the ioctl routines of network interfaces, inherit common ioctl behaviors by calling either ifioctl_common() or whichever ioctl routine is appropriate for the class of interface---e.g., ether_ioctl() for ethernets. Stop (ab)using SIOCSIFADDR and start to use SIOCINITIFADDR. In the user->kernel interface, SIOCSIFADDR's argument was an ifreq, but on the protocol->ifnet interface, SIOCSIFADDR's argument was an ifaddr. That was confusing, and it would work against me as I make it possible for a network interface to overload most ioctls. On the protocol->ifnet interface, replace SIOCSIFADDR with SIOCINITIFADDR. In ifioctl(), return EPERM if userland tries to invoke SIOCINITIFADDR. In ifioctl(), give the interface the first shot at handling most interface ioctls, and give the protocol the second shot, instead of the other way around. Finally, let compatibility code (COMPAT_OSOCK) take a shot. Pull device initialization out of switch statements under SIOCINITIFADDR. For example, pull ..._init() out of any switch statement that looks like this: switch (...->sa_family) { case ...: ..._init(); ... break; ... default: ..._init(); ... break; } Rewrite many if-else clauses that handle all permutations of IFF_UP and IFF_RUNNING to use a switch statement, switch (x & (IFF_UP|IFF_RUNNING)) { case 0: ... break; case IFF_RUNNING: ... break; case IFF_UP: ... break; case IFF_UP|IFF_RUNNING: ... break; } unifdef lots of code containing #ifdef FreeBSD, #ifdef NetBSD, and #ifdef SIOCSIFMTU, especially in fwip(4) and in ndis(4). In ipw(4), remove an if_set_sadl() call that is out of place. In nfe(4), reuse the jumbo MTU logic in ether_ioctl(). Let ethernets register a callback for setting h/w state such as promiscuous mode and the multicast filter in accord with a change in the if_flags: ether_set_ifflags_cb() registers a callback that returns ENETRESET if the caller should reset the ethernet by calling if_init(), 0 on success, != 0 on failure. Pull common code from ex(4), gem(4), nfe(4), sip(4), tlp(4), vge(4) into ether_ioctl(), and register if_flags callbacks for those drivers. Return ENOTTY instead of EINVAL for inappropriate ioctls. In zyd(4), use ENXIO instead of ENOTTY to indicate that the device is not any longer attached. Add to if_set_sadl() a boolean 'factory' argument that indicates whether a link-layer address was assigned by the factory or some other source. In a comment, recommend using the factory address for generating an EUI64, and update in6_get_hw_ifid() to prefer a factory address to any other link-layer address. Add a routing message, RTM_LLINFO_UPD, that tells protocols to update the binding of network-layer addresses to link-layer addresses. Implement this message in IPv4 and IPv6 by sending a gratuitous ARP or a neighbor advertisement, respectively. Generate RTM_LLINFO_UPD messages on a change of an interface's link-layer address. In ether_ioctl(), do not let SIOCALIFADDR set a link-layer address that is broadcast/multicast or equal to 00:00:00:00:00:00. Make ether_ioctl() call ifioctl_common() to handle ioctls that it does not understand. In gif(4), initialize if_softc and use it, instead of assuming that the gif_softc and ifp overlap. Let ifioctl_common() handle SIOCGIFADDR. Sprinkle rtcache_invariants(), which checks on DIAGNOSTIC kernels that certain invariants on a struct route are satisfied. In agr(4), rewrite agr_ioctl_filter() to be a bit more explicit about the ioctls that we do not allow on an agr(4) member interface. bzero -> memset. Delete unnecessary casts to void *. Use sockaddr_in_init() and sockaddr_in6_init(). Compare pointers with NULL instead of "testing truth". Replace some instances of (type *)0 with NULL. Change some K&R prototypes to ANSI C, and join lines.
2008-11-07 03:20:01 +03:00
if (ifa->ifa_rtrequest != NULL)
ifa->ifa_rtrequest(RTM_ADD, rt, &info);
}
rt_newaddrmsg(cmd, ifa, error, nrt);
*** Summary *** When a link-layer address changes (e.g., ifconfig ex0 link 02:de:ad:be:ef:02 active), send a gratuitous ARP and/or a Neighbor Advertisement to update the network-/link-layer address bindings on our LAN peers. Refuse a change of ethernet address to the address 00:00:00:00:00:00 or to any multicast/broadcast address. (Thanks matt@.) Reorder ifnet ioctl operations so that driver ioctls may inherit the functions of their "class"---ether_ioctl(), fddi_ioctl(), et cetera---and the class ioctls may inherit from the generic ioctl, ifioctl_common(), but both driver- and class-ioctls may override the generic behavior. Make network drivers share more code. Distinguish a "factory" link-layer address from others for the purposes of both protecting that address from deletion and computing EUI64. Return consistent, appropriate error codes from network drivers. Improve readability. KNF. *** Details *** In if_attach(), always initialize the interface ioctl routine, ifnet->if_ioctl, if the driver has not already initialized it. Delete if_ioctl == NULL tests everywhere else, because it cannot happen. In the ioctl routines of network interfaces, inherit common ioctl behaviors by calling either ifioctl_common() or whichever ioctl routine is appropriate for the class of interface---e.g., ether_ioctl() for ethernets. Stop (ab)using SIOCSIFADDR and start to use SIOCINITIFADDR. In the user->kernel interface, SIOCSIFADDR's argument was an ifreq, but on the protocol->ifnet interface, SIOCSIFADDR's argument was an ifaddr. That was confusing, and it would work against me as I make it possible for a network interface to overload most ioctls. On the protocol->ifnet interface, replace SIOCSIFADDR with SIOCINITIFADDR. In ifioctl(), return EPERM if userland tries to invoke SIOCINITIFADDR. In ifioctl(), give the interface the first shot at handling most interface ioctls, and give the protocol the second shot, instead of the other way around. Finally, let compatibility code (COMPAT_OSOCK) take a shot. Pull device initialization out of switch statements under SIOCINITIFADDR. For example, pull ..._init() out of any switch statement that looks like this: switch (...->sa_family) { case ...: ..._init(); ... break; ... default: ..._init(); ... break; } Rewrite many if-else clauses that handle all permutations of IFF_UP and IFF_RUNNING to use a switch statement, switch (x & (IFF_UP|IFF_RUNNING)) { case 0: ... break; case IFF_RUNNING: ... break; case IFF_UP: ... break; case IFF_UP|IFF_RUNNING: ... break; } unifdef lots of code containing #ifdef FreeBSD, #ifdef NetBSD, and #ifdef SIOCSIFMTU, especially in fwip(4) and in ndis(4). In ipw(4), remove an if_set_sadl() call that is out of place. In nfe(4), reuse the jumbo MTU logic in ether_ioctl(). Let ethernets register a callback for setting h/w state such as promiscuous mode and the multicast filter in accord with a change in the if_flags: ether_set_ifflags_cb() registers a callback that returns ENETRESET if the caller should reset the ethernet by calling if_init(), 0 on success, != 0 on failure. Pull common code from ex(4), gem(4), nfe(4), sip(4), tlp(4), vge(4) into ether_ioctl(), and register if_flags callbacks for those drivers. Return ENOTTY instead of EINVAL for inappropriate ioctls. In zyd(4), use ENXIO instead of ENOTTY to indicate that the device is not any longer attached. Add to if_set_sadl() a boolean 'factory' argument that indicates whether a link-layer address was assigned by the factory or some other source. In a comment, recommend using the factory address for generating an EUI64, and update in6_get_hw_ifid() to prefer a factory address to any other link-layer address. Add a routing message, RTM_LLINFO_UPD, that tells protocols to update the binding of network-layer addresses to link-layer addresses. Implement this message in IPv4 and IPv6 by sending a gratuitous ARP or a neighbor advertisement, respectively. Generate RTM_LLINFO_UPD messages on a change of an interface's link-layer address. In ether_ioctl(), do not let SIOCALIFADDR set a link-layer address that is broadcast/multicast or equal to 00:00:00:00:00:00. Make ether_ioctl() call ifioctl_common() to handle ioctls that it does not understand. In gif(4), initialize if_softc and use it, instead of assuming that the gif_softc and ifp overlap. Let ifioctl_common() handle SIOCGIFADDR. Sprinkle rtcache_invariants(), which checks on DIAGNOSTIC kernels that certain invariants on a struct route are satisfied. In agr(4), rewrite agr_ioctl_filter() to be a bit more explicit about the ioctls that we do not allow on an agr(4) member interface. bzero -> memset. Delete unnecessary casts to void *. Use sockaddr_in_init() and sockaddr_in6_init(). Compare pointers with NULL instead of "testing truth". Replace some instances of (type *)0 with NULL. Change some K&R prototypes to ANSI C, and join lines.
2008-11-07 03:20:01 +03:00
break;
1993-03-21 12:45:37 +03:00
}
return error;
1993-03-21 12:45:37 +03:00
}
/*
* Route timer routines. These routes allow functions to be called
* for various routes at any time. This is useful in supporting
* path MTU discovery and redirect route deletion.
*
* This is similar to some BSDI internal functions, but it provides
* for multiple queues for efficiency's sake...
*/
LIST_HEAD(, rttimer_queue) rttimer_queue_head;
static int rt_init_done = 0;
#define RTTIMER_CALLOUT(r) do { \
if (r->rtt_func != NULL) { \
(*r->rtt_func)(r->rtt_rt, r); \
} else { \
rtrequest((int) RTM_DELETE, \
Take steps to hide the radix_node implementation of the forwarding table from the forwarding table's users: Introduce rt_walktree() for walking the routing table and applying a function to each rtentry. Replace most rn_walktree() calls with it. Use rt_getkey()/rt_setkey() to get/set a route's destination. Keep a pointer to the sockaddr key in the rtentry, so that rtentry users do not have to grovel in the radix_node for the key. Add a RTM_GET method to rtrequest. Use that instead of radix_node lookups in, e.g., carp(4). Add sys/net/link_proto.c, which supplies sockaddr routines for link-layer socket addresses (sockaddr_dl). Cosmetic: Constify. KNF. Stop open-coding LIST_FOREACH, TAILQ_FOREACH, et cetera. Use NULL instead of 0 for null pointers. Use __arraycount(). Reduce gratuitous parenthesization. Stop using variadic arguments for rip6_output(), it is unnecessary. Remove the unnecessary rtentry member rt_genmask and the code to maintain it, since nothing actually used it. Make rt_maskedcopy() easier to read by using meaningful variable names. Extract a subroutine intern_netmask() for looking up a netmask in the masks table. Start converting backslash-ridden IPv6 macros in sys/netinet6/in6_var.h into inline subroutines that one can read without special eyeglasses. One functional change: when the kernel serves an RTM_GET, RTM_LOCK, or RTM_CHANGE request, it applies the netmask (if supplied) to a destination before searching for it in the forwarding table. I have changed sys/netinet/ip_carp.c, carp_setroute(), to remove the unlawful radix_node knowledge. Apart from the changes to carp(4), netiso, ATM, and strip(4), I have run the changes on three nodes in my wireless routing testbed, which involves IPv4 + IPv6 dynamic routing acrobatics, and it's working beautifully so far.
2007-07-20 00:48:52 +04:00
rt_getkey(r->rtt_rt), \
0, 0, 0, 0); \
} \
} while (/*CONSTCOND*/0)
2005-02-27 01:45:09 +03:00
/*
* Some subtle order problems with domain initialization mean that
* we cannot count on this being run from rt_init before various
* protocol initializations are done. Therefore, we make sure
* that this is run when the first queue is added...
*/
2005-02-27 01:45:09 +03:00
void
rt_timer_init(void)
{
assert(rt_init_done == 0);
LIST_INIT(&rttimer_queue_head);
callout_init(&rt_timer_ch, 0);
callout_reset(&rt_timer_ch, hz, rt_timer_timer, NULL);
rt_init_done = 1;
}
struct rttimer_queue *
rt_timer_queue_create(u_int timeout)
{
struct rttimer_queue *rtq;
if (rt_init_done == 0)
rt_timer_init();
R_Malloc(rtq, struct rttimer_queue *, sizeof *rtq);
if (rtq == NULL)
return NULL;
memset(rtq, 0, sizeof(*rtq));
rtq->rtq_timeout = timeout;
TAILQ_INIT(&rtq->rtq_head);
LIST_INSERT_HEAD(&rttimer_queue_head, rtq, rtq_link);
return rtq;
}
void
rt_timer_queue_change(struct rttimer_queue *rtq, long timeout)
{
rtq->rtq_timeout = timeout;
}
void
rt_timer_queue_remove_all(struct rttimer_queue *rtq, int destroy)
{
struct rttimer *r;
while ((r = TAILQ_FIRST(&rtq->rtq_head)) != NULL) {
LIST_REMOVE(r, rtt_link);
TAILQ_REMOVE(&rtq->rtq_head, r, rtt_next);
if (destroy)
RTTIMER_CALLOUT(r);
/* we are already at splsoftnet */
pool_put(&rttimer_pool, r);
if (rtq->rtq_count > 0)
rtq->rtq_count--;
else
printf("rt_timer_queue_remove_all: "
"rtq_count reached 0\n");
}
}
void
rt_timer_queue_destroy(struct rttimer_queue *rtq, int destroy)
{
rt_timer_queue_remove_all(rtq, destroy);
LIST_REMOVE(rtq, rtq_link);
/*
* Caller is responsible for freeing the rttimer_queue structure.
*/
}
unsigned long
rt_timer_count(struct rttimer_queue *rtq)
{
return rtq->rtq_count;
}
2005-02-27 01:45:09 +03:00
void
rt_timer_remove_all(struct rtentry *rt, int destroy)
{
struct rttimer *r;
while ((r = LIST_FIRST(&rt->rt_timer)) != NULL) {
LIST_REMOVE(r, rtt_link);
TAILQ_REMOVE(&r->rtt_queue->rtq_head, r, rtt_next);
if (destroy)
RTTIMER_CALLOUT(r);
if (r->rtt_queue->rtq_count > 0)
r->rtt_queue->rtq_count--;
else
printf("rt_timer_remove_all: rtq_count reached 0\n");
/* we are already at splsoftnet */
2000-12-11 10:52:48 +03:00
pool_put(&rttimer_pool, r);
}
}
2005-02-27 01:45:09 +03:00
int
rt_timer_add(struct rtentry *rt,
void (*func)(struct rtentry *, struct rttimer *),
struct rttimer_queue *queue)
{
struct rttimer *r;
int s;
/*
* If there's already a timer with this action, destroy it before
* we add a new one.
*/
LIST_FOREACH(r, &rt->rt_timer, rtt_link) {
if (r->rtt_func == func)
break;
}
if (r != NULL) {
LIST_REMOVE(r, rtt_link);
TAILQ_REMOVE(&r->rtt_queue->rtq_head, r, rtt_next);
if (r->rtt_queue->rtq_count > 0)
r->rtt_queue->rtq_count--;
else
printf("rt_timer_add: rtq_count reached 0\n");
} else {
s = splsoftnet();
r = pool_get(&rttimer_pool, PR_NOWAIT);
splx(s);
if (r == NULL)
return ENOBUFS;
}
memset(r, 0, sizeof(*r));
r->rtt_rt = rt;
r->rtt_time = time_uptime;
r->rtt_func = func;
r->rtt_queue = queue;
LIST_INSERT_HEAD(&rt->rt_timer, r, rtt_link);
TAILQ_INSERT_TAIL(&queue->rtq_head, r, rtt_next);
r->rtt_queue->rtq_count++;
2005-02-27 01:45:09 +03:00
return 0;
}
/* ARGSUSED */
void
rt_timer_timer(void *arg)
{
struct rttimer_queue *rtq;
struct rttimer *r;
int s;
s = splsoftnet();
LIST_FOREACH(rtq, &rttimer_queue_head, rtq_link) {
while ((r = TAILQ_FIRST(&rtq->rtq_head)) != NULL &&
(r->rtt_time + rtq->rtq_timeout) < time_uptime) {
LIST_REMOVE(r, rtt_link);
TAILQ_REMOVE(&rtq->rtq_head, r, rtt_next);
RTTIMER_CALLOUT(r);
pool_put(&rttimer_pool, r);
if (rtq->rtq_count > 0)
rtq->rtq_count--;
else
printf("rt_timer_timer: rtq_count reached 0\n");
}
}
splx(s);
callout_reset(&rt_timer_ch, hz, rt_timer_timer, NULL);
}
static struct rtentry *
_rtcache_init(struct route *ro, int flag)
{
*** Summary *** When a link-layer address changes (e.g., ifconfig ex0 link 02:de:ad:be:ef:02 active), send a gratuitous ARP and/or a Neighbor Advertisement to update the network-/link-layer address bindings on our LAN peers. Refuse a change of ethernet address to the address 00:00:00:00:00:00 or to any multicast/broadcast address. (Thanks matt@.) Reorder ifnet ioctl operations so that driver ioctls may inherit the functions of their "class"---ether_ioctl(), fddi_ioctl(), et cetera---and the class ioctls may inherit from the generic ioctl, ifioctl_common(), but both driver- and class-ioctls may override the generic behavior. Make network drivers share more code. Distinguish a "factory" link-layer address from others for the purposes of both protecting that address from deletion and computing EUI64. Return consistent, appropriate error codes from network drivers. Improve readability. KNF. *** Details *** In if_attach(), always initialize the interface ioctl routine, ifnet->if_ioctl, if the driver has not already initialized it. Delete if_ioctl == NULL tests everywhere else, because it cannot happen. In the ioctl routines of network interfaces, inherit common ioctl behaviors by calling either ifioctl_common() or whichever ioctl routine is appropriate for the class of interface---e.g., ether_ioctl() for ethernets. Stop (ab)using SIOCSIFADDR and start to use SIOCINITIFADDR. In the user->kernel interface, SIOCSIFADDR's argument was an ifreq, but on the protocol->ifnet interface, SIOCSIFADDR's argument was an ifaddr. That was confusing, and it would work against me as I make it possible for a network interface to overload most ioctls. On the protocol->ifnet interface, replace SIOCSIFADDR with SIOCINITIFADDR. In ifioctl(), return EPERM if userland tries to invoke SIOCINITIFADDR. In ifioctl(), give the interface the first shot at handling most interface ioctls, and give the protocol the second shot, instead of the other way around. Finally, let compatibility code (COMPAT_OSOCK) take a shot. Pull device initialization out of switch statements under SIOCINITIFADDR. For example, pull ..._init() out of any switch statement that looks like this: switch (...->sa_family) { case ...: ..._init(); ... break; ... default: ..._init(); ... break; } Rewrite many if-else clauses that handle all permutations of IFF_UP and IFF_RUNNING to use a switch statement, switch (x & (IFF_UP|IFF_RUNNING)) { case 0: ... break; case IFF_RUNNING: ... break; case IFF_UP: ... break; case IFF_UP|IFF_RUNNING: ... break; } unifdef lots of code containing #ifdef FreeBSD, #ifdef NetBSD, and #ifdef SIOCSIFMTU, especially in fwip(4) and in ndis(4). In ipw(4), remove an if_set_sadl() call that is out of place. In nfe(4), reuse the jumbo MTU logic in ether_ioctl(). Let ethernets register a callback for setting h/w state such as promiscuous mode and the multicast filter in accord with a change in the if_flags: ether_set_ifflags_cb() registers a callback that returns ENETRESET if the caller should reset the ethernet by calling if_init(), 0 on success, != 0 on failure. Pull common code from ex(4), gem(4), nfe(4), sip(4), tlp(4), vge(4) into ether_ioctl(), and register if_flags callbacks for those drivers. Return ENOTTY instead of EINVAL for inappropriate ioctls. In zyd(4), use ENXIO instead of ENOTTY to indicate that the device is not any longer attached. Add to if_set_sadl() a boolean 'factory' argument that indicates whether a link-layer address was assigned by the factory or some other source. In a comment, recommend using the factory address for generating an EUI64, and update in6_get_hw_ifid() to prefer a factory address to any other link-layer address. Add a routing message, RTM_LLINFO_UPD, that tells protocols to update the binding of network-layer addresses to link-layer addresses. Implement this message in IPv4 and IPv6 by sending a gratuitous ARP or a neighbor advertisement, respectively. Generate RTM_LLINFO_UPD messages on a change of an interface's link-layer address. In ether_ioctl(), do not let SIOCALIFADDR set a link-layer address that is broadcast/multicast or equal to 00:00:00:00:00:00. Make ether_ioctl() call ifioctl_common() to handle ioctls that it does not understand. In gif(4), initialize if_softc and use it, instead of assuming that the gif_softc and ifp overlap. Let ifioctl_common() handle SIOCGIFADDR. Sprinkle rtcache_invariants(), which checks on DIAGNOSTIC kernels that certain invariants on a struct route are satisfied. In agr(4), rewrite agr_ioctl_filter() to be a bit more explicit about the ioctls that we do not allow on an agr(4) member interface. bzero -> memset. Delete unnecessary casts to void *. Use sockaddr_in_init() and sockaddr_in6_init(). Compare pointers with NULL instead of "testing truth". Replace some instances of (type *)0 with NULL. Change some K&R prototypes to ANSI C, and join lines.
2008-11-07 03:20:01 +03:00
rtcache_invariants(ro);
KASSERT(ro->_ro_rt == NULL);
Eliminate address family-specific route caches (struct route, struct route_in6, struct route_iso), replacing all caches with a struct route. The principle benefit of this change is that all of the protocol families can benefit from route cache-invalidation, which is necessary for correct routing. Route-cache invalidation fixes an ancient PR, kern/3508, at long last; it fixes various other PRs, also. Discussions with and ideas from Joerg Sonnenberger influenced this work tremendously. Of course, all design oversights and bugs are mine. DETAILS 1 I added to each address family a pool of sockaddrs. I have introduced routines for allocating, copying, and duplicating, and freeing sockaddrs: struct sockaddr *sockaddr_alloc(sa_family_t af, int flags); struct sockaddr *sockaddr_copy(struct sockaddr *dst, const struct sockaddr *src); struct sockaddr *sockaddr_dup(const struct sockaddr *src, int flags); void sockaddr_free(struct sockaddr *sa); sockaddr_alloc() returns either a sockaddr from the pool belonging to the specified family, or NULL if the pool is exhausted. The returned sockaddr has the right size for that family; sa_family and sa_len fields are initialized to the family and sockaddr length---e.g., sa_family = AF_INET and sa_len = sizeof(struct sockaddr_in). sockaddr_free() puts the given sockaddr back into its family's pool. sockaddr_dup() and sockaddr_copy() work analogously to strdup() and strcpy(), respectively. sockaddr_copy() KASSERTs that the family of the destination and source sockaddrs are alike. The 'flags' argumet for sockaddr_alloc() and sockaddr_dup() is passed directly to pool_get(9). 2 I added routines for initializing sockaddrs in each address family, sockaddr_in_init(), sockaddr_in6_init(), sockaddr_iso_init(), etc. They are fairly self-explanatory. 3 structs route_in6 and route_iso are no more. All protocol families use struct route. I have changed the route cache, 'struct route', so that it does not contain storage space for a sockaddr. Instead, struct route points to a sockaddr coming from the pool the sockaddr belongs to. I added a new method to struct route, rtcache_setdst(), for setting the cache destination: int rtcache_setdst(struct route *, const struct sockaddr *); rtcache_setdst() returns 0 on success, or ENOMEM if no memory is available to create the sockaddr storage. It is now possible for rtcache_getdst() to return NULL if, say, rtcache_setdst() failed. I check the return value for NULL everywhere in the kernel. 4 Each routing domain (struct domain) has a list of live route caches, dom_rtcache. rtflushall(sa_family_t af) looks up the domain indicated by 'af', walks the domain's list of route caches and invalidates each one.
2007-05-03 00:40:22 +04:00
if (rtcache_getdst(ro) == NULL)
return NULL;
ro->ro_invalid = false;
if ((ro->_ro_rt = rtalloc1(rtcache_getdst(ro), flag)) != NULL)
rtcache(ro);
*** Summary *** When a link-layer address changes (e.g., ifconfig ex0 link 02:de:ad:be:ef:02 active), send a gratuitous ARP and/or a Neighbor Advertisement to update the network-/link-layer address bindings on our LAN peers. Refuse a change of ethernet address to the address 00:00:00:00:00:00 or to any multicast/broadcast address. (Thanks matt@.) Reorder ifnet ioctl operations so that driver ioctls may inherit the functions of their "class"---ether_ioctl(), fddi_ioctl(), et cetera---and the class ioctls may inherit from the generic ioctl, ifioctl_common(), but both driver- and class-ioctls may override the generic behavior. Make network drivers share more code. Distinguish a "factory" link-layer address from others for the purposes of both protecting that address from deletion and computing EUI64. Return consistent, appropriate error codes from network drivers. Improve readability. KNF. *** Details *** In if_attach(), always initialize the interface ioctl routine, ifnet->if_ioctl, if the driver has not already initialized it. Delete if_ioctl == NULL tests everywhere else, because it cannot happen. In the ioctl routines of network interfaces, inherit common ioctl behaviors by calling either ifioctl_common() or whichever ioctl routine is appropriate for the class of interface---e.g., ether_ioctl() for ethernets. Stop (ab)using SIOCSIFADDR and start to use SIOCINITIFADDR. In the user->kernel interface, SIOCSIFADDR's argument was an ifreq, but on the protocol->ifnet interface, SIOCSIFADDR's argument was an ifaddr. That was confusing, and it would work against me as I make it possible for a network interface to overload most ioctls. On the protocol->ifnet interface, replace SIOCSIFADDR with SIOCINITIFADDR. In ifioctl(), return EPERM if userland tries to invoke SIOCINITIFADDR. In ifioctl(), give the interface the first shot at handling most interface ioctls, and give the protocol the second shot, instead of the other way around. Finally, let compatibility code (COMPAT_OSOCK) take a shot. Pull device initialization out of switch statements under SIOCINITIFADDR. For example, pull ..._init() out of any switch statement that looks like this: switch (...->sa_family) { case ...: ..._init(); ... break; ... default: ..._init(); ... break; } Rewrite many if-else clauses that handle all permutations of IFF_UP and IFF_RUNNING to use a switch statement, switch (x & (IFF_UP|IFF_RUNNING)) { case 0: ... break; case IFF_RUNNING: ... break; case IFF_UP: ... break; case IFF_UP|IFF_RUNNING: ... break; } unifdef lots of code containing #ifdef FreeBSD, #ifdef NetBSD, and #ifdef SIOCSIFMTU, especially in fwip(4) and in ndis(4). In ipw(4), remove an if_set_sadl() call that is out of place. In nfe(4), reuse the jumbo MTU logic in ether_ioctl(). Let ethernets register a callback for setting h/w state such as promiscuous mode and the multicast filter in accord with a change in the if_flags: ether_set_ifflags_cb() registers a callback that returns ENETRESET if the caller should reset the ethernet by calling if_init(), 0 on success, != 0 on failure. Pull common code from ex(4), gem(4), nfe(4), sip(4), tlp(4), vge(4) into ether_ioctl(), and register if_flags callbacks for those drivers. Return ENOTTY instead of EINVAL for inappropriate ioctls. In zyd(4), use ENXIO instead of ENOTTY to indicate that the device is not any longer attached. Add to if_set_sadl() a boolean 'factory' argument that indicates whether a link-layer address was assigned by the factory or some other source. In a comment, recommend using the factory address for generating an EUI64, and update in6_get_hw_ifid() to prefer a factory address to any other link-layer address. Add a routing message, RTM_LLINFO_UPD, that tells protocols to update the binding of network-layer addresses to link-layer addresses. Implement this message in IPv4 and IPv6 by sending a gratuitous ARP or a neighbor advertisement, respectively. Generate RTM_LLINFO_UPD messages on a change of an interface's link-layer address. In ether_ioctl(), do not let SIOCALIFADDR set a link-layer address that is broadcast/multicast or equal to 00:00:00:00:00:00. Make ether_ioctl() call ifioctl_common() to handle ioctls that it does not understand. In gif(4), initialize if_softc and use it, instead of assuming that the gif_softc and ifp overlap. Let ifioctl_common() handle SIOCGIFADDR. Sprinkle rtcache_invariants(), which checks on DIAGNOSTIC kernels that certain invariants on a struct route are satisfied. In agr(4), rewrite agr_ioctl_filter() to be a bit more explicit about the ioctls that we do not allow on an agr(4) member interface. bzero -> memset. Delete unnecessary casts to void *. Use sockaddr_in_init() and sockaddr_in6_init(). Compare pointers with NULL instead of "testing truth". Replace some instances of (type *)0 with NULL. Change some K&R prototypes to ANSI C, and join lines.
2008-11-07 03:20:01 +03:00
rtcache_invariants(ro);
return ro->_ro_rt;
}
struct rtentry *
rtcache_init(struct route *ro)
{
return _rtcache_init(ro, 1);
}
struct rtentry *
rtcache_init_noclone(struct route *ro)
{
return _rtcache_init(ro, 0);
}
Eliminate address family-specific route caches (struct route, struct route_in6, struct route_iso), replacing all caches with a struct route. The principle benefit of this change is that all of the protocol families can benefit from route cache-invalidation, which is necessary for correct routing. Route-cache invalidation fixes an ancient PR, kern/3508, at long last; it fixes various other PRs, also. Discussions with and ideas from Joerg Sonnenberger influenced this work tremendously. Of course, all design oversights and bugs are mine. DETAILS 1 I added to each address family a pool of sockaddrs. I have introduced routines for allocating, copying, and duplicating, and freeing sockaddrs: struct sockaddr *sockaddr_alloc(sa_family_t af, int flags); struct sockaddr *sockaddr_copy(struct sockaddr *dst, const struct sockaddr *src); struct sockaddr *sockaddr_dup(const struct sockaddr *src, int flags); void sockaddr_free(struct sockaddr *sa); sockaddr_alloc() returns either a sockaddr from the pool belonging to the specified family, or NULL if the pool is exhausted. The returned sockaddr has the right size for that family; sa_family and sa_len fields are initialized to the family and sockaddr length---e.g., sa_family = AF_INET and sa_len = sizeof(struct sockaddr_in). sockaddr_free() puts the given sockaddr back into its family's pool. sockaddr_dup() and sockaddr_copy() work analogously to strdup() and strcpy(), respectively. sockaddr_copy() KASSERTs that the family of the destination and source sockaddrs are alike. The 'flags' argumet for sockaddr_alloc() and sockaddr_dup() is passed directly to pool_get(9). 2 I added routines for initializing sockaddrs in each address family, sockaddr_in_init(), sockaddr_in6_init(), sockaddr_iso_init(), etc. They are fairly self-explanatory. 3 structs route_in6 and route_iso are no more. All protocol families use struct route. I have changed the route cache, 'struct route', so that it does not contain storage space for a sockaddr. Instead, struct route points to a sockaddr coming from the pool the sockaddr belongs to. I added a new method to struct route, rtcache_setdst(), for setting the cache destination: int rtcache_setdst(struct route *, const struct sockaddr *); rtcache_setdst() returns 0 on success, or ENOMEM if no memory is available to create the sockaddr storage. It is now possible for rtcache_getdst() to return NULL if, say, rtcache_setdst() failed. I check the return value for NULL everywhere in the kernel. 4 Each routing domain (struct domain) has a list of live route caches, dom_rtcache. rtflushall(sa_family_t af) looks up the domain indicated by 'af', walks the domain's list of route caches and invalidates each one.
2007-05-03 00:40:22 +04:00
struct rtentry *
Eliminate address family-specific route caches (struct route, struct route_in6, struct route_iso), replacing all caches with a struct route. The principle benefit of this change is that all of the protocol families can benefit from route cache-invalidation, which is necessary for correct routing. Route-cache invalidation fixes an ancient PR, kern/3508, at long last; it fixes various other PRs, also. Discussions with and ideas from Joerg Sonnenberger influenced this work tremendously. Of course, all design oversights and bugs are mine. DETAILS 1 I added to each address family a pool of sockaddrs. I have introduced routines for allocating, copying, and duplicating, and freeing sockaddrs: struct sockaddr *sockaddr_alloc(sa_family_t af, int flags); struct sockaddr *sockaddr_copy(struct sockaddr *dst, const struct sockaddr *src); struct sockaddr *sockaddr_dup(const struct sockaddr *src, int flags); void sockaddr_free(struct sockaddr *sa); sockaddr_alloc() returns either a sockaddr from the pool belonging to the specified family, or NULL if the pool is exhausted. The returned sockaddr has the right size for that family; sa_family and sa_len fields are initialized to the family and sockaddr length---e.g., sa_family = AF_INET and sa_len = sizeof(struct sockaddr_in). sockaddr_free() puts the given sockaddr back into its family's pool. sockaddr_dup() and sockaddr_copy() work analogously to strdup() and strcpy(), respectively. sockaddr_copy() KASSERTs that the family of the destination and source sockaddrs are alike. The 'flags' argumet for sockaddr_alloc() and sockaddr_dup() is passed directly to pool_get(9). 2 I added routines for initializing sockaddrs in each address family, sockaddr_in_init(), sockaddr_in6_init(), sockaddr_iso_init(), etc. They are fairly self-explanatory. 3 structs route_in6 and route_iso are no more. All protocol families use struct route. I have changed the route cache, 'struct route', so that it does not contain storage space for a sockaddr. Instead, struct route points to a sockaddr coming from the pool the sockaddr belongs to. I added a new method to struct route, rtcache_setdst(), for setting the cache destination: int rtcache_setdst(struct route *, const struct sockaddr *); rtcache_setdst() returns 0 on success, or ENOMEM if no memory is available to create the sockaddr storage. It is now possible for rtcache_getdst() to return NULL if, say, rtcache_setdst() failed. I check the return value for NULL everywhere in the kernel. 4 Each routing domain (struct domain) has a list of live route caches, dom_rtcache. rtflushall(sa_family_t af) looks up the domain indicated by 'af', walks the domain's list of route caches and invalidates each one.
2007-05-03 00:40:22 +04:00
rtcache_update(struct route *ro, int clone)
{
rtcache_clear(ro);
return _rtcache_init(ro, clone);
Eliminate address family-specific route caches (struct route, struct route_in6, struct route_iso), replacing all caches with a struct route. The principle benefit of this change is that all of the protocol families can benefit from route cache-invalidation, which is necessary for correct routing. Route-cache invalidation fixes an ancient PR, kern/3508, at long last; it fixes various other PRs, also. Discussions with and ideas from Joerg Sonnenberger influenced this work tremendously. Of course, all design oversights and bugs are mine. DETAILS 1 I added to each address family a pool of sockaddrs. I have introduced routines for allocating, copying, and duplicating, and freeing sockaddrs: struct sockaddr *sockaddr_alloc(sa_family_t af, int flags); struct sockaddr *sockaddr_copy(struct sockaddr *dst, const struct sockaddr *src); struct sockaddr *sockaddr_dup(const struct sockaddr *src, int flags); void sockaddr_free(struct sockaddr *sa); sockaddr_alloc() returns either a sockaddr from the pool belonging to the specified family, or NULL if the pool is exhausted. The returned sockaddr has the right size for that family; sa_family and sa_len fields are initialized to the family and sockaddr length---e.g., sa_family = AF_INET and sa_len = sizeof(struct sockaddr_in). sockaddr_free() puts the given sockaddr back into its family's pool. sockaddr_dup() and sockaddr_copy() work analogously to strdup() and strcpy(), respectively. sockaddr_copy() KASSERTs that the family of the destination and source sockaddrs are alike. The 'flags' argumet for sockaddr_alloc() and sockaddr_dup() is passed directly to pool_get(9). 2 I added routines for initializing sockaddrs in each address family, sockaddr_in_init(), sockaddr_in6_init(), sockaddr_iso_init(), etc. They are fairly self-explanatory. 3 structs route_in6 and route_iso are no more. All protocol families use struct route. I have changed the route cache, 'struct route', so that it does not contain storage space for a sockaddr. Instead, struct route points to a sockaddr coming from the pool the sockaddr belongs to. I added a new method to struct route, rtcache_setdst(), for setting the cache destination: int rtcache_setdst(struct route *, const struct sockaddr *); rtcache_setdst() returns 0 on success, or ENOMEM if no memory is available to create the sockaddr storage. It is now possible for rtcache_getdst() to return NULL if, say, rtcache_setdst() failed. I check the return value for NULL everywhere in the kernel. 4 Each routing domain (struct domain) has a list of live route caches, dom_rtcache. rtflushall(sa_family_t af) looks up the domain indicated by 'af', walks the domain's list of route caches and invalidates each one.
2007-05-03 00:40:22 +04:00
}
void
Eliminate address family-specific route caches (struct route, struct route_in6, struct route_iso), replacing all caches with a struct route. The principle benefit of this change is that all of the protocol families can benefit from route cache-invalidation, which is necessary for correct routing. Route-cache invalidation fixes an ancient PR, kern/3508, at long last; it fixes various other PRs, also. Discussions with and ideas from Joerg Sonnenberger influenced this work tremendously. Of course, all design oversights and bugs are mine. DETAILS 1 I added to each address family a pool of sockaddrs. I have introduced routines for allocating, copying, and duplicating, and freeing sockaddrs: struct sockaddr *sockaddr_alloc(sa_family_t af, int flags); struct sockaddr *sockaddr_copy(struct sockaddr *dst, const struct sockaddr *src); struct sockaddr *sockaddr_dup(const struct sockaddr *src, int flags); void sockaddr_free(struct sockaddr *sa); sockaddr_alloc() returns either a sockaddr from the pool belonging to the specified family, or NULL if the pool is exhausted. The returned sockaddr has the right size for that family; sa_family and sa_len fields are initialized to the family and sockaddr length---e.g., sa_family = AF_INET and sa_len = sizeof(struct sockaddr_in). sockaddr_free() puts the given sockaddr back into its family's pool. sockaddr_dup() and sockaddr_copy() work analogously to strdup() and strcpy(), respectively. sockaddr_copy() KASSERTs that the family of the destination and source sockaddrs are alike. The 'flags' argumet for sockaddr_alloc() and sockaddr_dup() is passed directly to pool_get(9). 2 I added routines for initializing sockaddrs in each address family, sockaddr_in_init(), sockaddr_in6_init(), sockaddr_iso_init(), etc. They are fairly self-explanatory. 3 structs route_in6 and route_iso are no more. All protocol families use struct route. I have changed the route cache, 'struct route', so that it does not contain storage space for a sockaddr. Instead, struct route points to a sockaddr coming from the pool the sockaddr belongs to. I added a new method to struct route, rtcache_setdst(), for setting the cache destination: int rtcache_setdst(struct route *, const struct sockaddr *); rtcache_setdst() returns 0 on success, or ENOMEM if no memory is available to create the sockaddr storage. It is now possible for rtcache_getdst() to return NULL if, say, rtcache_setdst() failed. I check the return value for NULL everywhere in the kernel. 4 Each routing domain (struct domain) has a list of live route caches, dom_rtcache. rtflushall(sa_family_t af) looks up the domain indicated by 'af', walks the domain's list of route caches and invalidates each one.
2007-05-03 00:40:22 +04:00
rtcache_copy(struct route *new_ro, const struct route *old_ro)
{
struct rtentry *rt;
KASSERT(new_ro != old_ro);
*** Summary *** When a link-layer address changes (e.g., ifconfig ex0 link 02:de:ad:be:ef:02 active), send a gratuitous ARP and/or a Neighbor Advertisement to update the network-/link-layer address bindings on our LAN peers. Refuse a change of ethernet address to the address 00:00:00:00:00:00 or to any multicast/broadcast address. (Thanks matt@.) Reorder ifnet ioctl operations so that driver ioctls may inherit the functions of their "class"---ether_ioctl(), fddi_ioctl(), et cetera---and the class ioctls may inherit from the generic ioctl, ifioctl_common(), but both driver- and class-ioctls may override the generic behavior. Make network drivers share more code. Distinguish a "factory" link-layer address from others for the purposes of both protecting that address from deletion and computing EUI64. Return consistent, appropriate error codes from network drivers. Improve readability. KNF. *** Details *** In if_attach(), always initialize the interface ioctl routine, ifnet->if_ioctl, if the driver has not already initialized it. Delete if_ioctl == NULL tests everywhere else, because it cannot happen. In the ioctl routines of network interfaces, inherit common ioctl behaviors by calling either ifioctl_common() or whichever ioctl routine is appropriate for the class of interface---e.g., ether_ioctl() for ethernets. Stop (ab)using SIOCSIFADDR and start to use SIOCINITIFADDR. In the user->kernel interface, SIOCSIFADDR's argument was an ifreq, but on the protocol->ifnet interface, SIOCSIFADDR's argument was an ifaddr. That was confusing, and it would work against me as I make it possible for a network interface to overload most ioctls. On the protocol->ifnet interface, replace SIOCSIFADDR with SIOCINITIFADDR. In ifioctl(), return EPERM if userland tries to invoke SIOCINITIFADDR. In ifioctl(), give the interface the first shot at handling most interface ioctls, and give the protocol the second shot, instead of the other way around. Finally, let compatibility code (COMPAT_OSOCK) take a shot. Pull device initialization out of switch statements under SIOCINITIFADDR. For example, pull ..._init() out of any switch statement that looks like this: switch (...->sa_family) { case ...: ..._init(); ... break; ... default: ..._init(); ... break; } Rewrite many if-else clauses that handle all permutations of IFF_UP and IFF_RUNNING to use a switch statement, switch (x & (IFF_UP|IFF_RUNNING)) { case 0: ... break; case IFF_RUNNING: ... break; case IFF_UP: ... break; case IFF_UP|IFF_RUNNING: ... break; } unifdef lots of code containing #ifdef FreeBSD, #ifdef NetBSD, and #ifdef SIOCSIFMTU, especially in fwip(4) and in ndis(4). In ipw(4), remove an if_set_sadl() call that is out of place. In nfe(4), reuse the jumbo MTU logic in ether_ioctl(). Let ethernets register a callback for setting h/w state such as promiscuous mode and the multicast filter in accord with a change in the if_flags: ether_set_ifflags_cb() registers a callback that returns ENETRESET if the caller should reset the ethernet by calling if_init(), 0 on success, != 0 on failure. Pull common code from ex(4), gem(4), nfe(4), sip(4), tlp(4), vge(4) into ether_ioctl(), and register if_flags callbacks for those drivers. Return ENOTTY instead of EINVAL for inappropriate ioctls. In zyd(4), use ENXIO instead of ENOTTY to indicate that the device is not any longer attached. Add to if_set_sadl() a boolean 'factory' argument that indicates whether a link-layer address was assigned by the factory or some other source. In a comment, recommend using the factory address for generating an EUI64, and update in6_get_hw_ifid() to prefer a factory address to any other link-layer address. Add a routing message, RTM_LLINFO_UPD, that tells protocols to update the binding of network-layer addresses to link-layer addresses. Implement this message in IPv4 and IPv6 by sending a gratuitous ARP or a neighbor advertisement, respectively. Generate RTM_LLINFO_UPD messages on a change of an interface's link-layer address. In ether_ioctl(), do not let SIOCALIFADDR set a link-layer address that is broadcast/multicast or equal to 00:00:00:00:00:00. Make ether_ioctl() call ifioctl_common() to handle ioctls that it does not understand. In gif(4), initialize if_softc and use it, instead of assuming that the gif_softc and ifp overlap. Let ifioctl_common() handle SIOCGIFADDR. Sprinkle rtcache_invariants(), which checks on DIAGNOSTIC kernels that certain invariants on a struct route are satisfied. In agr(4), rewrite agr_ioctl_filter() to be a bit more explicit about the ioctls that we do not allow on an agr(4) member interface. bzero -> memset. Delete unnecessary casts to void *. Use sockaddr_in_init() and sockaddr_in6_init(). Compare pointers with NULL instead of "testing truth". Replace some instances of (type *)0 with NULL. Change some K&R prototypes to ANSI C, and join lines.
2008-11-07 03:20:01 +03:00
rtcache_invariants(new_ro);
rtcache_invariants(old_ro);
if ((rt = rtcache_validate(old_ro)) != NULL)
rt->rt_refcnt++;
Eliminate address family-specific route caches (struct route, struct route_in6, struct route_iso), replacing all caches with a struct route. The principle benefit of this change is that all of the protocol families can benefit from route cache-invalidation, which is necessary for correct routing. Route-cache invalidation fixes an ancient PR, kern/3508, at long last; it fixes various other PRs, also. Discussions with and ideas from Joerg Sonnenberger influenced this work tremendously. Of course, all design oversights and bugs are mine. DETAILS 1 I added to each address family a pool of sockaddrs. I have introduced routines for allocating, copying, and duplicating, and freeing sockaddrs: struct sockaddr *sockaddr_alloc(sa_family_t af, int flags); struct sockaddr *sockaddr_copy(struct sockaddr *dst, const struct sockaddr *src); struct sockaddr *sockaddr_dup(const struct sockaddr *src, int flags); void sockaddr_free(struct sockaddr *sa); sockaddr_alloc() returns either a sockaddr from the pool belonging to the specified family, or NULL if the pool is exhausted. The returned sockaddr has the right size for that family; sa_family and sa_len fields are initialized to the family and sockaddr length---e.g., sa_family = AF_INET and sa_len = sizeof(struct sockaddr_in). sockaddr_free() puts the given sockaddr back into its family's pool. sockaddr_dup() and sockaddr_copy() work analogously to strdup() and strcpy(), respectively. sockaddr_copy() KASSERTs that the family of the destination and source sockaddrs are alike. The 'flags' argumet for sockaddr_alloc() and sockaddr_dup() is passed directly to pool_get(9). 2 I added routines for initializing sockaddrs in each address family, sockaddr_in_init(), sockaddr_in6_init(), sockaddr_iso_init(), etc. They are fairly self-explanatory. 3 structs route_in6 and route_iso are no more. All protocol families use struct route. I have changed the route cache, 'struct route', so that it does not contain storage space for a sockaddr. Instead, struct route points to a sockaddr coming from the pool the sockaddr belongs to. I added a new method to struct route, rtcache_setdst(), for setting the cache destination: int rtcache_setdst(struct route *, const struct sockaddr *); rtcache_setdst() returns 0 on success, or ENOMEM if no memory is available to create the sockaddr storage. It is now possible for rtcache_getdst() to return NULL if, say, rtcache_setdst() failed. I check the return value for NULL everywhere in the kernel. 4 Each routing domain (struct domain) has a list of live route caches, dom_rtcache. rtflushall(sa_family_t af) looks up the domain indicated by 'af', walks the domain's list of route caches and invalidates each one.
2007-05-03 00:40:22 +04:00
if (rtcache_getdst(old_ro) == NULL ||
rtcache_setdst(new_ro, rtcache_getdst(old_ro)) != 0)
return;
new_ro->ro_invalid = false;
if ((new_ro->_ro_rt = rt) != NULL)
rtcache(new_ro);
*** Summary *** When a link-layer address changes (e.g., ifconfig ex0 link 02:de:ad:be:ef:02 active), send a gratuitous ARP and/or a Neighbor Advertisement to update the network-/link-layer address bindings on our LAN peers. Refuse a change of ethernet address to the address 00:00:00:00:00:00 or to any multicast/broadcast address. (Thanks matt@.) Reorder ifnet ioctl operations so that driver ioctls may inherit the functions of their "class"---ether_ioctl(), fddi_ioctl(), et cetera---and the class ioctls may inherit from the generic ioctl, ifioctl_common(), but both driver- and class-ioctls may override the generic behavior. Make network drivers share more code. Distinguish a "factory" link-layer address from others for the purposes of both protecting that address from deletion and computing EUI64. Return consistent, appropriate error codes from network drivers. Improve readability. KNF. *** Details *** In if_attach(), always initialize the interface ioctl routine, ifnet->if_ioctl, if the driver has not already initialized it. Delete if_ioctl == NULL tests everywhere else, because it cannot happen. In the ioctl routines of network interfaces, inherit common ioctl behaviors by calling either ifioctl_common() or whichever ioctl routine is appropriate for the class of interface---e.g., ether_ioctl() for ethernets. Stop (ab)using SIOCSIFADDR and start to use SIOCINITIFADDR. In the user->kernel interface, SIOCSIFADDR's argument was an ifreq, but on the protocol->ifnet interface, SIOCSIFADDR's argument was an ifaddr. That was confusing, and it would work against me as I make it possible for a network interface to overload most ioctls. On the protocol->ifnet interface, replace SIOCSIFADDR with SIOCINITIFADDR. In ifioctl(), return EPERM if userland tries to invoke SIOCINITIFADDR. In ifioctl(), give the interface the first shot at handling most interface ioctls, and give the protocol the second shot, instead of the other way around. Finally, let compatibility code (COMPAT_OSOCK) take a shot. Pull device initialization out of switch statements under SIOCINITIFADDR. For example, pull ..._init() out of any switch statement that looks like this: switch (...->sa_family) { case ...: ..._init(); ... break; ... default: ..._init(); ... break; } Rewrite many if-else clauses that handle all permutations of IFF_UP and IFF_RUNNING to use a switch statement, switch (x & (IFF_UP|IFF_RUNNING)) { case 0: ... break; case IFF_RUNNING: ... break; case IFF_UP: ... break; case IFF_UP|IFF_RUNNING: ... break; } unifdef lots of code containing #ifdef FreeBSD, #ifdef NetBSD, and #ifdef SIOCSIFMTU, especially in fwip(4) and in ndis(4). In ipw(4), remove an if_set_sadl() call that is out of place. In nfe(4), reuse the jumbo MTU logic in ether_ioctl(). Let ethernets register a callback for setting h/w state such as promiscuous mode and the multicast filter in accord with a change in the if_flags: ether_set_ifflags_cb() registers a callback that returns ENETRESET if the caller should reset the ethernet by calling if_init(), 0 on success, != 0 on failure. Pull common code from ex(4), gem(4), nfe(4), sip(4), tlp(4), vge(4) into ether_ioctl(), and register if_flags callbacks for those drivers. Return ENOTTY instead of EINVAL for inappropriate ioctls. In zyd(4), use ENXIO instead of ENOTTY to indicate that the device is not any longer attached. Add to if_set_sadl() a boolean 'factory' argument that indicates whether a link-layer address was assigned by the factory or some other source. In a comment, recommend using the factory address for generating an EUI64, and update in6_get_hw_ifid() to prefer a factory address to any other link-layer address. Add a routing message, RTM_LLINFO_UPD, that tells protocols to update the binding of network-layer addresses to link-layer addresses. Implement this message in IPv4 and IPv6 by sending a gratuitous ARP or a neighbor advertisement, respectively. Generate RTM_LLINFO_UPD messages on a change of an interface's link-layer address. In ether_ioctl(), do not let SIOCALIFADDR set a link-layer address that is broadcast/multicast or equal to 00:00:00:00:00:00. Make ether_ioctl() call ifioctl_common() to handle ioctls that it does not understand. In gif(4), initialize if_softc and use it, instead of assuming that the gif_softc and ifp overlap. Let ifioctl_common() handle SIOCGIFADDR. Sprinkle rtcache_invariants(), which checks on DIAGNOSTIC kernels that certain invariants on a struct route are satisfied. In agr(4), rewrite agr_ioctl_filter() to be a bit more explicit about the ioctls that we do not allow on an agr(4) member interface. bzero -> memset. Delete unnecessary casts to void *. Use sockaddr_in_init() and sockaddr_in6_init(). Compare pointers with NULL instead of "testing truth". Replace some instances of (type *)0 with NULL. Change some K&R prototypes to ANSI C, and join lines.
2008-11-07 03:20:01 +03:00
rtcache_invariants(new_ro);
}
static struct dom_rtlist invalid_routes = LIST_HEAD_INITIALIZER(dom_rtlist);
void
rtcache_invalidate(struct dom_rtlist *rtlist)
{
struct route *ro;
while ((ro = LIST_FIRST(rtlist)) != NULL) {
*** Summary *** When a link-layer address changes (e.g., ifconfig ex0 link 02:de:ad:be:ef:02 active), send a gratuitous ARP and/or a Neighbor Advertisement to update the network-/link-layer address bindings on our LAN peers. Refuse a change of ethernet address to the address 00:00:00:00:00:00 or to any multicast/broadcast address. (Thanks matt@.) Reorder ifnet ioctl operations so that driver ioctls may inherit the functions of their "class"---ether_ioctl(), fddi_ioctl(), et cetera---and the class ioctls may inherit from the generic ioctl, ifioctl_common(), but both driver- and class-ioctls may override the generic behavior. Make network drivers share more code. Distinguish a "factory" link-layer address from others for the purposes of both protecting that address from deletion and computing EUI64. Return consistent, appropriate error codes from network drivers. Improve readability. KNF. *** Details *** In if_attach(), always initialize the interface ioctl routine, ifnet->if_ioctl, if the driver has not already initialized it. Delete if_ioctl == NULL tests everywhere else, because it cannot happen. In the ioctl routines of network interfaces, inherit common ioctl behaviors by calling either ifioctl_common() or whichever ioctl routine is appropriate for the class of interface---e.g., ether_ioctl() for ethernets. Stop (ab)using SIOCSIFADDR and start to use SIOCINITIFADDR. In the user->kernel interface, SIOCSIFADDR's argument was an ifreq, but on the protocol->ifnet interface, SIOCSIFADDR's argument was an ifaddr. That was confusing, and it would work against me as I make it possible for a network interface to overload most ioctls. On the protocol->ifnet interface, replace SIOCSIFADDR with SIOCINITIFADDR. In ifioctl(), return EPERM if userland tries to invoke SIOCINITIFADDR. In ifioctl(), give the interface the first shot at handling most interface ioctls, and give the protocol the second shot, instead of the other way around. Finally, let compatibility code (COMPAT_OSOCK) take a shot. Pull device initialization out of switch statements under SIOCINITIFADDR. For example, pull ..._init() out of any switch statement that looks like this: switch (...->sa_family) { case ...: ..._init(); ... break; ... default: ..._init(); ... break; } Rewrite many if-else clauses that handle all permutations of IFF_UP and IFF_RUNNING to use a switch statement, switch (x & (IFF_UP|IFF_RUNNING)) { case 0: ... break; case IFF_RUNNING: ... break; case IFF_UP: ... break; case IFF_UP|IFF_RUNNING: ... break; } unifdef lots of code containing #ifdef FreeBSD, #ifdef NetBSD, and #ifdef SIOCSIFMTU, especially in fwip(4) and in ndis(4). In ipw(4), remove an if_set_sadl() call that is out of place. In nfe(4), reuse the jumbo MTU logic in ether_ioctl(). Let ethernets register a callback for setting h/w state such as promiscuous mode and the multicast filter in accord with a change in the if_flags: ether_set_ifflags_cb() registers a callback that returns ENETRESET if the caller should reset the ethernet by calling if_init(), 0 on success, != 0 on failure. Pull common code from ex(4), gem(4), nfe(4), sip(4), tlp(4), vge(4) into ether_ioctl(), and register if_flags callbacks for those drivers. Return ENOTTY instead of EINVAL for inappropriate ioctls. In zyd(4), use ENXIO instead of ENOTTY to indicate that the device is not any longer attached. Add to if_set_sadl() a boolean 'factory' argument that indicates whether a link-layer address was assigned by the factory or some other source. In a comment, recommend using the factory address for generating an EUI64, and update in6_get_hw_ifid() to prefer a factory address to any other link-layer address. Add a routing message, RTM_LLINFO_UPD, that tells protocols to update the binding of network-layer addresses to link-layer addresses. Implement this message in IPv4 and IPv6 by sending a gratuitous ARP or a neighbor advertisement, respectively. Generate RTM_LLINFO_UPD messages on a change of an interface's link-layer address. In ether_ioctl(), do not let SIOCALIFADDR set a link-layer address that is broadcast/multicast or equal to 00:00:00:00:00:00. Make ether_ioctl() call ifioctl_common() to handle ioctls that it does not understand. In gif(4), initialize if_softc and use it, instead of assuming that the gif_softc and ifp overlap. Let ifioctl_common() handle SIOCGIFADDR. Sprinkle rtcache_invariants(), which checks on DIAGNOSTIC kernels that certain invariants on a struct route are satisfied. In agr(4), rewrite agr_ioctl_filter() to be a bit more explicit about the ioctls that we do not allow on an agr(4) member interface. bzero -> memset. Delete unnecessary casts to void *. Use sockaddr_in_init() and sockaddr_in6_init(). Compare pointers with NULL instead of "testing truth". Replace some instances of (type *)0 with NULL. Change some K&R prototypes to ANSI C, and join lines.
2008-11-07 03:20:01 +03:00
rtcache_invariants(ro);
KASSERT(ro->_ro_rt != NULL);
ro->ro_invalid = true;
LIST_REMOVE(ro, ro_rtcache_next);
LIST_INSERT_HEAD(&invalid_routes, ro, ro_rtcache_next);
*** Summary *** When a link-layer address changes (e.g., ifconfig ex0 link 02:de:ad:be:ef:02 active), send a gratuitous ARP and/or a Neighbor Advertisement to update the network-/link-layer address bindings on our LAN peers. Refuse a change of ethernet address to the address 00:00:00:00:00:00 or to any multicast/broadcast address. (Thanks matt@.) Reorder ifnet ioctl operations so that driver ioctls may inherit the functions of their "class"---ether_ioctl(), fddi_ioctl(), et cetera---and the class ioctls may inherit from the generic ioctl, ifioctl_common(), but both driver- and class-ioctls may override the generic behavior. Make network drivers share more code. Distinguish a "factory" link-layer address from others for the purposes of both protecting that address from deletion and computing EUI64. Return consistent, appropriate error codes from network drivers. Improve readability. KNF. *** Details *** In if_attach(), always initialize the interface ioctl routine, ifnet->if_ioctl, if the driver has not already initialized it. Delete if_ioctl == NULL tests everywhere else, because it cannot happen. In the ioctl routines of network interfaces, inherit common ioctl behaviors by calling either ifioctl_common() or whichever ioctl routine is appropriate for the class of interface---e.g., ether_ioctl() for ethernets. Stop (ab)using SIOCSIFADDR and start to use SIOCINITIFADDR. In the user->kernel interface, SIOCSIFADDR's argument was an ifreq, but on the protocol->ifnet interface, SIOCSIFADDR's argument was an ifaddr. That was confusing, and it would work against me as I make it possible for a network interface to overload most ioctls. On the protocol->ifnet interface, replace SIOCSIFADDR with SIOCINITIFADDR. In ifioctl(), return EPERM if userland tries to invoke SIOCINITIFADDR. In ifioctl(), give the interface the first shot at handling most interface ioctls, and give the protocol the second shot, instead of the other way around. Finally, let compatibility code (COMPAT_OSOCK) take a shot. Pull device initialization out of switch statements under SIOCINITIFADDR. For example, pull ..._init() out of any switch statement that looks like this: switch (...->sa_family) { case ...: ..._init(); ... break; ... default: ..._init(); ... break; } Rewrite many if-else clauses that handle all permutations of IFF_UP and IFF_RUNNING to use a switch statement, switch (x & (IFF_UP|IFF_RUNNING)) { case 0: ... break; case IFF_RUNNING: ... break; case IFF_UP: ... break; case IFF_UP|IFF_RUNNING: ... break; } unifdef lots of code containing #ifdef FreeBSD, #ifdef NetBSD, and #ifdef SIOCSIFMTU, especially in fwip(4) and in ndis(4). In ipw(4), remove an if_set_sadl() call that is out of place. In nfe(4), reuse the jumbo MTU logic in ether_ioctl(). Let ethernets register a callback for setting h/w state such as promiscuous mode and the multicast filter in accord with a change in the if_flags: ether_set_ifflags_cb() registers a callback that returns ENETRESET if the caller should reset the ethernet by calling if_init(), 0 on success, != 0 on failure. Pull common code from ex(4), gem(4), nfe(4), sip(4), tlp(4), vge(4) into ether_ioctl(), and register if_flags callbacks for those drivers. Return ENOTTY instead of EINVAL for inappropriate ioctls. In zyd(4), use ENXIO instead of ENOTTY to indicate that the device is not any longer attached. Add to if_set_sadl() a boolean 'factory' argument that indicates whether a link-layer address was assigned by the factory or some other source. In a comment, recommend using the factory address for generating an EUI64, and update in6_get_hw_ifid() to prefer a factory address to any other link-layer address. Add a routing message, RTM_LLINFO_UPD, that tells protocols to update the binding of network-layer addresses to link-layer addresses. Implement this message in IPv4 and IPv6 by sending a gratuitous ARP or a neighbor advertisement, respectively. Generate RTM_LLINFO_UPD messages on a change of an interface's link-layer address. In ether_ioctl(), do not let SIOCALIFADDR set a link-layer address that is broadcast/multicast or equal to 00:00:00:00:00:00. Make ether_ioctl() call ifioctl_common() to handle ioctls that it does not understand. In gif(4), initialize if_softc and use it, instead of assuming that the gif_softc and ifp overlap. Let ifioctl_common() handle SIOCGIFADDR. Sprinkle rtcache_invariants(), which checks on DIAGNOSTIC kernels that certain invariants on a struct route are satisfied. In agr(4), rewrite agr_ioctl_filter() to be a bit more explicit about the ioctls that we do not allow on an agr(4) member interface. bzero -> memset. Delete unnecessary casts to void *. Use sockaddr_in_init() and sockaddr_in6_init(). Compare pointers with NULL instead of "testing truth". Replace some instances of (type *)0 with NULL. Change some K&R prototypes to ANSI C, and join lines.
2008-11-07 03:20:01 +03:00
rtcache_invariants(ro);
}
}
void
rtcache_clear(struct route *ro)
{
*** Summary *** When a link-layer address changes (e.g., ifconfig ex0 link 02:de:ad:be:ef:02 active), send a gratuitous ARP and/or a Neighbor Advertisement to update the network-/link-layer address bindings on our LAN peers. Refuse a change of ethernet address to the address 00:00:00:00:00:00 or to any multicast/broadcast address. (Thanks matt@.) Reorder ifnet ioctl operations so that driver ioctls may inherit the functions of their "class"---ether_ioctl(), fddi_ioctl(), et cetera---and the class ioctls may inherit from the generic ioctl, ifioctl_common(), but both driver- and class-ioctls may override the generic behavior. Make network drivers share more code. Distinguish a "factory" link-layer address from others for the purposes of both protecting that address from deletion and computing EUI64. Return consistent, appropriate error codes from network drivers. Improve readability. KNF. *** Details *** In if_attach(), always initialize the interface ioctl routine, ifnet->if_ioctl, if the driver has not already initialized it. Delete if_ioctl == NULL tests everywhere else, because it cannot happen. In the ioctl routines of network interfaces, inherit common ioctl behaviors by calling either ifioctl_common() or whichever ioctl routine is appropriate for the class of interface---e.g., ether_ioctl() for ethernets. Stop (ab)using SIOCSIFADDR and start to use SIOCINITIFADDR. In the user->kernel interface, SIOCSIFADDR's argument was an ifreq, but on the protocol->ifnet interface, SIOCSIFADDR's argument was an ifaddr. That was confusing, and it would work against me as I make it possible for a network interface to overload most ioctls. On the protocol->ifnet interface, replace SIOCSIFADDR with SIOCINITIFADDR. In ifioctl(), return EPERM if userland tries to invoke SIOCINITIFADDR. In ifioctl(), give the interface the first shot at handling most interface ioctls, and give the protocol the second shot, instead of the other way around. Finally, let compatibility code (COMPAT_OSOCK) take a shot. Pull device initialization out of switch statements under SIOCINITIFADDR. For example, pull ..._init() out of any switch statement that looks like this: switch (...->sa_family) { case ...: ..._init(); ... break; ... default: ..._init(); ... break; } Rewrite many if-else clauses that handle all permutations of IFF_UP and IFF_RUNNING to use a switch statement, switch (x & (IFF_UP|IFF_RUNNING)) { case 0: ... break; case IFF_RUNNING: ... break; case IFF_UP: ... break; case IFF_UP|IFF_RUNNING: ... break; } unifdef lots of code containing #ifdef FreeBSD, #ifdef NetBSD, and #ifdef SIOCSIFMTU, especially in fwip(4) and in ndis(4). In ipw(4), remove an if_set_sadl() call that is out of place. In nfe(4), reuse the jumbo MTU logic in ether_ioctl(). Let ethernets register a callback for setting h/w state such as promiscuous mode and the multicast filter in accord with a change in the if_flags: ether_set_ifflags_cb() registers a callback that returns ENETRESET if the caller should reset the ethernet by calling if_init(), 0 on success, != 0 on failure. Pull common code from ex(4), gem(4), nfe(4), sip(4), tlp(4), vge(4) into ether_ioctl(), and register if_flags callbacks for those drivers. Return ENOTTY instead of EINVAL for inappropriate ioctls. In zyd(4), use ENXIO instead of ENOTTY to indicate that the device is not any longer attached. Add to if_set_sadl() a boolean 'factory' argument that indicates whether a link-layer address was assigned by the factory or some other source. In a comment, recommend using the factory address for generating an EUI64, and update in6_get_hw_ifid() to prefer a factory address to any other link-layer address. Add a routing message, RTM_LLINFO_UPD, that tells protocols to update the binding of network-layer addresses to link-layer addresses. Implement this message in IPv4 and IPv6 by sending a gratuitous ARP or a neighbor advertisement, respectively. Generate RTM_LLINFO_UPD messages on a change of an interface's link-layer address. In ether_ioctl(), do not let SIOCALIFADDR set a link-layer address that is broadcast/multicast or equal to 00:00:00:00:00:00. Make ether_ioctl() call ifioctl_common() to handle ioctls that it does not understand. In gif(4), initialize if_softc and use it, instead of assuming that the gif_softc and ifp overlap. Let ifioctl_common() handle SIOCGIFADDR. Sprinkle rtcache_invariants(), which checks on DIAGNOSTIC kernels that certain invariants on a struct route are satisfied. In agr(4), rewrite agr_ioctl_filter() to be a bit more explicit about the ioctls that we do not allow on an agr(4) member interface. bzero -> memset. Delete unnecessary casts to void *. Use sockaddr_in_init() and sockaddr_in6_init(). Compare pointers with NULL instead of "testing truth". Replace some instances of (type *)0 with NULL. Change some K&R prototypes to ANSI C, and join lines.
2008-11-07 03:20:01 +03:00
rtcache_invariants(ro);
if (ro->_ro_rt == NULL)
return;
LIST_REMOVE(ro, ro_rtcache_next);
RTFREE(ro->_ro_rt);
ro->_ro_rt = NULL;
*** Summary *** When a link-layer address changes (e.g., ifconfig ex0 link 02:de:ad:be:ef:02 active), send a gratuitous ARP and/or a Neighbor Advertisement to update the network-/link-layer address bindings on our LAN peers. Refuse a change of ethernet address to the address 00:00:00:00:00:00 or to any multicast/broadcast address. (Thanks matt@.) Reorder ifnet ioctl operations so that driver ioctls may inherit the functions of their "class"---ether_ioctl(), fddi_ioctl(), et cetera---and the class ioctls may inherit from the generic ioctl, ifioctl_common(), but both driver- and class-ioctls may override the generic behavior. Make network drivers share more code. Distinguish a "factory" link-layer address from others for the purposes of both protecting that address from deletion and computing EUI64. Return consistent, appropriate error codes from network drivers. Improve readability. KNF. *** Details *** In if_attach(), always initialize the interface ioctl routine, ifnet->if_ioctl, if the driver has not already initialized it. Delete if_ioctl == NULL tests everywhere else, because it cannot happen. In the ioctl routines of network interfaces, inherit common ioctl behaviors by calling either ifioctl_common() or whichever ioctl routine is appropriate for the class of interface---e.g., ether_ioctl() for ethernets. Stop (ab)using SIOCSIFADDR and start to use SIOCINITIFADDR. In the user->kernel interface, SIOCSIFADDR's argument was an ifreq, but on the protocol->ifnet interface, SIOCSIFADDR's argument was an ifaddr. That was confusing, and it would work against me as I make it possible for a network interface to overload most ioctls. On the protocol->ifnet interface, replace SIOCSIFADDR with SIOCINITIFADDR. In ifioctl(), return EPERM if userland tries to invoke SIOCINITIFADDR. In ifioctl(), give the interface the first shot at handling most interface ioctls, and give the protocol the second shot, instead of the other way around. Finally, let compatibility code (COMPAT_OSOCK) take a shot. Pull device initialization out of switch statements under SIOCINITIFADDR. For example, pull ..._init() out of any switch statement that looks like this: switch (...->sa_family) { case ...: ..._init(); ... break; ... default: ..._init(); ... break; } Rewrite many if-else clauses that handle all permutations of IFF_UP and IFF_RUNNING to use a switch statement, switch (x & (IFF_UP|IFF_RUNNING)) { case 0: ... break; case IFF_RUNNING: ... break; case IFF_UP: ... break; case IFF_UP|IFF_RUNNING: ... break; } unifdef lots of code containing #ifdef FreeBSD, #ifdef NetBSD, and #ifdef SIOCSIFMTU, especially in fwip(4) and in ndis(4). In ipw(4), remove an if_set_sadl() call that is out of place. In nfe(4), reuse the jumbo MTU logic in ether_ioctl(). Let ethernets register a callback for setting h/w state such as promiscuous mode and the multicast filter in accord with a change in the if_flags: ether_set_ifflags_cb() registers a callback that returns ENETRESET if the caller should reset the ethernet by calling if_init(), 0 on success, != 0 on failure. Pull common code from ex(4), gem(4), nfe(4), sip(4), tlp(4), vge(4) into ether_ioctl(), and register if_flags callbacks for those drivers. Return ENOTTY instead of EINVAL for inappropriate ioctls. In zyd(4), use ENXIO instead of ENOTTY to indicate that the device is not any longer attached. Add to if_set_sadl() a boolean 'factory' argument that indicates whether a link-layer address was assigned by the factory or some other source. In a comment, recommend using the factory address for generating an EUI64, and update in6_get_hw_ifid() to prefer a factory address to any other link-layer address. Add a routing message, RTM_LLINFO_UPD, that tells protocols to update the binding of network-layer addresses to link-layer addresses. Implement this message in IPv4 and IPv6 by sending a gratuitous ARP or a neighbor advertisement, respectively. Generate RTM_LLINFO_UPD messages on a change of an interface's link-layer address. In ether_ioctl(), do not let SIOCALIFADDR set a link-layer address that is broadcast/multicast or equal to 00:00:00:00:00:00. Make ether_ioctl() call ifioctl_common() to handle ioctls that it does not understand. In gif(4), initialize if_softc and use it, instead of assuming that the gif_softc and ifp overlap. Let ifioctl_common() handle SIOCGIFADDR. Sprinkle rtcache_invariants(), which checks on DIAGNOSTIC kernels that certain invariants on a struct route are satisfied. In agr(4), rewrite agr_ioctl_filter() to be a bit more explicit about the ioctls that we do not allow on an agr(4) member interface. bzero -> memset. Delete unnecessary casts to void *. Use sockaddr_in_init() and sockaddr_in6_init(). Compare pointers with NULL instead of "testing truth". Replace some instances of (type *)0 with NULL. Change some K&R prototypes to ANSI C, and join lines.
2008-11-07 03:20:01 +03:00
ro->ro_invalid = false;
rtcache_invariants(ro);
}
Eliminate address family-specific route caches (struct route, struct route_in6, struct route_iso), replacing all caches with a struct route. The principle benefit of this change is that all of the protocol families can benefit from route cache-invalidation, which is necessary for correct routing. Route-cache invalidation fixes an ancient PR, kern/3508, at long last; it fixes various other PRs, also. Discussions with and ideas from Joerg Sonnenberger influenced this work tremendously. Of course, all design oversights and bugs are mine. DETAILS 1 I added to each address family a pool of sockaddrs. I have introduced routines for allocating, copying, and duplicating, and freeing sockaddrs: struct sockaddr *sockaddr_alloc(sa_family_t af, int flags); struct sockaddr *sockaddr_copy(struct sockaddr *dst, const struct sockaddr *src); struct sockaddr *sockaddr_dup(const struct sockaddr *src, int flags); void sockaddr_free(struct sockaddr *sa); sockaddr_alloc() returns either a sockaddr from the pool belonging to the specified family, or NULL if the pool is exhausted. The returned sockaddr has the right size for that family; sa_family and sa_len fields are initialized to the family and sockaddr length---e.g., sa_family = AF_INET and sa_len = sizeof(struct sockaddr_in). sockaddr_free() puts the given sockaddr back into its family's pool. sockaddr_dup() and sockaddr_copy() work analogously to strdup() and strcpy(), respectively. sockaddr_copy() KASSERTs that the family of the destination and source sockaddrs are alike. The 'flags' argumet for sockaddr_alloc() and sockaddr_dup() is passed directly to pool_get(9). 2 I added routines for initializing sockaddrs in each address family, sockaddr_in_init(), sockaddr_in6_init(), sockaddr_iso_init(), etc. They are fairly self-explanatory. 3 structs route_in6 and route_iso are no more. All protocol families use struct route. I have changed the route cache, 'struct route', so that it does not contain storage space for a sockaddr. Instead, struct route points to a sockaddr coming from the pool the sockaddr belongs to. I added a new method to struct route, rtcache_setdst(), for setting the cache destination: int rtcache_setdst(struct route *, const struct sockaddr *); rtcache_setdst() returns 0 on success, or ENOMEM if no memory is available to create the sockaddr storage. It is now possible for rtcache_getdst() to return NULL if, say, rtcache_setdst() failed. I check the return value for NULL everywhere in the kernel. 4 Each routing domain (struct domain) has a list of live route caches, dom_rtcache. rtflushall(sa_family_t af) looks up the domain indicated by 'af', walks the domain's list of route caches and invalidates each one.
2007-05-03 00:40:22 +04:00
struct rtentry *
rtcache_lookup2(struct route *ro, const struct sockaddr *dst, int clone,
int *hitp)
Eliminate address family-specific route caches (struct route, struct route_in6, struct route_iso), replacing all caches with a struct route. The principle benefit of this change is that all of the protocol families can benefit from route cache-invalidation, which is necessary for correct routing. Route-cache invalidation fixes an ancient PR, kern/3508, at long last; it fixes various other PRs, also. Discussions with and ideas from Joerg Sonnenberger influenced this work tremendously. Of course, all design oversights and bugs are mine. DETAILS 1 I added to each address family a pool of sockaddrs. I have introduced routines for allocating, copying, and duplicating, and freeing sockaddrs: struct sockaddr *sockaddr_alloc(sa_family_t af, int flags); struct sockaddr *sockaddr_copy(struct sockaddr *dst, const struct sockaddr *src); struct sockaddr *sockaddr_dup(const struct sockaddr *src, int flags); void sockaddr_free(struct sockaddr *sa); sockaddr_alloc() returns either a sockaddr from the pool belonging to the specified family, or NULL if the pool is exhausted. The returned sockaddr has the right size for that family; sa_family and sa_len fields are initialized to the family and sockaddr length---e.g., sa_family = AF_INET and sa_len = sizeof(struct sockaddr_in). sockaddr_free() puts the given sockaddr back into its family's pool. sockaddr_dup() and sockaddr_copy() work analogously to strdup() and strcpy(), respectively. sockaddr_copy() KASSERTs that the family of the destination and source sockaddrs are alike. The 'flags' argumet for sockaddr_alloc() and sockaddr_dup() is passed directly to pool_get(9). 2 I added routines for initializing sockaddrs in each address family, sockaddr_in_init(), sockaddr_in6_init(), sockaddr_iso_init(), etc. They are fairly self-explanatory. 3 structs route_in6 and route_iso are no more. All protocol families use struct route. I have changed the route cache, 'struct route', so that it does not contain storage space for a sockaddr. Instead, struct route points to a sockaddr coming from the pool the sockaddr belongs to. I added a new method to struct route, rtcache_setdst(), for setting the cache destination: int rtcache_setdst(struct route *, const struct sockaddr *); rtcache_setdst() returns 0 on success, or ENOMEM if no memory is available to create the sockaddr storage. It is now possible for rtcache_getdst() to return NULL if, say, rtcache_setdst() failed. I check the return value for NULL everywhere in the kernel. 4 Each routing domain (struct domain) has a list of live route caches, dom_rtcache. rtflushall(sa_family_t af) looks up the domain indicated by 'af', walks the domain's list of route caches and invalidates each one.
2007-05-03 00:40:22 +04:00
{
const struct sockaddr *odst;
struct rtentry *rt = NULL;
Eliminate address family-specific route caches (struct route, struct route_in6, struct route_iso), replacing all caches with a struct route. The principle benefit of this change is that all of the protocol families can benefit from route cache-invalidation, which is necessary for correct routing. Route-cache invalidation fixes an ancient PR, kern/3508, at long last; it fixes various other PRs, also. Discussions with and ideas from Joerg Sonnenberger influenced this work tremendously. Of course, all design oversights and bugs are mine. DETAILS 1 I added to each address family a pool of sockaddrs. I have introduced routines for allocating, copying, and duplicating, and freeing sockaddrs: struct sockaddr *sockaddr_alloc(sa_family_t af, int flags); struct sockaddr *sockaddr_copy(struct sockaddr *dst, const struct sockaddr *src); struct sockaddr *sockaddr_dup(const struct sockaddr *src, int flags); void sockaddr_free(struct sockaddr *sa); sockaddr_alloc() returns either a sockaddr from the pool belonging to the specified family, or NULL if the pool is exhausted. The returned sockaddr has the right size for that family; sa_family and sa_len fields are initialized to the family and sockaddr length---e.g., sa_family = AF_INET and sa_len = sizeof(struct sockaddr_in). sockaddr_free() puts the given sockaddr back into its family's pool. sockaddr_dup() and sockaddr_copy() work analogously to strdup() and strcpy(), respectively. sockaddr_copy() KASSERTs that the family of the destination and source sockaddrs are alike. The 'flags' argumet for sockaddr_alloc() and sockaddr_dup() is passed directly to pool_get(9). 2 I added routines for initializing sockaddrs in each address family, sockaddr_in_init(), sockaddr_in6_init(), sockaddr_iso_init(), etc. They are fairly self-explanatory. 3 structs route_in6 and route_iso are no more. All protocol families use struct route. I have changed the route cache, 'struct route', so that it does not contain storage space for a sockaddr. Instead, struct route points to a sockaddr coming from the pool the sockaddr belongs to. I added a new method to struct route, rtcache_setdst(), for setting the cache destination: int rtcache_setdst(struct route *, const struct sockaddr *); rtcache_setdst() returns 0 on success, or ENOMEM if no memory is available to create the sockaddr storage. It is now possible for rtcache_getdst() to return NULL if, say, rtcache_setdst() failed. I check the return value for NULL everywhere in the kernel. 4 Each routing domain (struct domain) has a list of live route caches, dom_rtcache. rtflushall(sa_family_t af) looks up the domain indicated by 'af', walks the domain's list of route caches and invalidates each one.
2007-05-03 00:40:22 +04:00
*** Summary *** When a link-layer address changes (e.g., ifconfig ex0 link 02:de:ad:be:ef:02 active), send a gratuitous ARP and/or a Neighbor Advertisement to update the network-/link-layer address bindings on our LAN peers. Refuse a change of ethernet address to the address 00:00:00:00:00:00 or to any multicast/broadcast address. (Thanks matt@.) Reorder ifnet ioctl operations so that driver ioctls may inherit the functions of their "class"---ether_ioctl(), fddi_ioctl(), et cetera---and the class ioctls may inherit from the generic ioctl, ifioctl_common(), but both driver- and class-ioctls may override the generic behavior. Make network drivers share more code. Distinguish a "factory" link-layer address from others for the purposes of both protecting that address from deletion and computing EUI64. Return consistent, appropriate error codes from network drivers. Improve readability. KNF. *** Details *** In if_attach(), always initialize the interface ioctl routine, ifnet->if_ioctl, if the driver has not already initialized it. Delete if_ioctl == NULL tests everywhere else, because it cannot happen. In the ioctl routines of network interfaces, inherit common ioctl behaviors by calling either ifioctl_common() or whichever ioctl routine is appropriate for the class of interface---e.g., ether_ioctl() for ethernets. Stop (ab)using SIOCSIFADDR and start to use SIOCINITIFADDR. In the user->kernel interface, SIOCSIFADDR's argument was an ifreq, but on the protocol->ifnet interface, SIOCSIFADDR's argument was an ifaddr. That was confusing, and it would work against me as I make it possible for a network interface to overload most ioctls. On the protocol->ifnet interface, replace SIOCSIFADDR with SIOCINITIFADDR. In ifioctl(), return EPERM if userland tries to invoke SIOCINITIFADDR. In ifioctl(), give the interface the first shot at handling most interface ioctls, and give the protocol the second shot, instead of the other way around. Finally, let compatibility code (COMPAT_OSOCK) take a shot. Pull device initialization out of switch statements under SIOCINITIFADDR. For example, pull ..._init() out of any switch statement that looks like this: switch (...->sa_family) { case ...: ..._init(); ... break; ... default: ..._init(); ... break; } Rewrite many if-else clauses that handle all permutations of IFF_UP and IFF_RUNNING to use a switch statement, switch (x & (IFF_UP|IFF_RUNNING)) { case 0: ... break; case IFF_RUNNING: ... break; case IFF_UP: ... break; case IFF_UP|IFF_RUNNING: ... break; } unifdef lots of code containing #ifdef FreeBSD, #ifdef NetBSD, and #ifdef SIOCSIFMTU, especially in fwip(4) and in ndis(4). In ipw(4), remove an if_set_sadl() call that is out of place. In nfe(4), reuse the jumbo MTU logic in ether_ioctl(). Let ethernets register a callback for setting h/w state such as promiscuous mode and the multicast filter in accord with a change in the if_flags: ether_set_ifflags_cb() registers a callback that returns ENETRESET if the caller should reset the ethernet by calling if_init(), 0 on success, != 0 on failure. Pull common code from ex(4), gem(4), nfe(4), sip(4), tlp(4), vge(4) into ether_ioctl(), and register if_flags callbacks for those drivers. Return ENOTTY instead of EINVAL for inappropriate ioctls. In zyd(4), use ENXIO instead of ENOTTY to indicate that the device is not any longer attached. Add to if_set_sadl() a boolean 'factory' argument that indicates whether a link-layer address was assigned by the factory or some other source. In a comment, recommend using the factory address for generating an EUI64, and update in6_get_hw_ifid() to prefer a factory address to any other link-layer address. Add a routing message, RTM_LLINFO_UPD, that tells protocols to update the binding of network-layer addresses to link-layer addresses. Implement this message in IPv4 and IPv6 by sending a gratuitous ARP or a neighbor advertisement, respectively. Generate RTM_LLINFO_UPD messages on a change of an interface's link-layer address. In ether_ioctl(), do not let SIOCALIFADDR set a link-layer address that is broadcast/multicast or equal to 00:00:00:00:00:00. Make ether_ioctl() call ifioctl_common() to handle ioctls that it does not understand. In gif(4), initialize if_softc and use it, instead of assuming that the gif_softc and ifp overlap. Let ifioctl_common() handle SIOCGIFADDR. Sprinkle rtcache_invariants(), which checks on DIAGNOSTIC kernels that certain invariants on a struct route are satisfied. In agr(4), rewrite agr_ioctl_filter() to be a bit more explicit about the ioctls that we do not allow on an agr(4) member interface. bzero -> memset. Delete unnecessary casts to void *. Use sockaddr_in_init() and sockaddr_in6_init(). Compare pointers with NULL instead of "testing truth". Replace some instances of (type *)0 with NULL. Change some K&R prototypes to ANSI C, and join lines.
2008-11-07 03:20:01 +03:00
rtcache_invariants(ro);
Eliminate address family-specific route caches (struct route, struct route_in6, struct route_iso), replacing all caches with a struct route. The principle benefit of this change is that all of the protocol families can benefit from route cache-invalidation, which is necessary for correct routing. Route-cache invalidation fixes an ancient PR, kern/3508, at long last; it fixes various other PRs, also. Discussions with and ideas from Joerg Sonnenberger influenced this work tremendously. Of course, all design oversights and bugs are mine. DETAILS 1 I added to each address family a pool of sockaddrs. I have introduced routines for allocating, copying, and duplicating, and freeing sockaddrs: struct sockaddr *sockaddr_alloc(sa_family_t af, int flags); struct sockaddr *sockaddr_copy(struct sockaddr *dst, const struct sockaddr *src); struct sockaddr *sockaddr_dup(const struct sockaddr *src, int flags); void sockaddr_free(struct sockaddr *sa); sockaddr_alloc() returns either a sockaddr from the pool belonging to the specified family, or NULL if the pool is exhausted. The returned sockaddr has the right size for that family; sa_family and sa_len fields are initialized to the family and sockaddr length---e.g., sa_family = AF_INET and sa_len = sizeof(struct sockaddr_in). sockaddr_free() puts the given sockaddr back into its family's pool. sockaddr_dup() and sockaddr_copy() work analogously to strdup() and strcpy(), respectively. sockaddr_copy() KASSERTs that the family of the destination and source sockaddrs are alike. The 'flags' argumet for sockaddr_alloc() and sockaddr_dup() is passed directly to pool_get(9). 2 I added routines for initializing sockaddrs in each address family, sockaddr_in_init(), sockaddr_in6_init(), sockaddr_iso_init(), etc. They are fairly self-explanatory. 3 structs route_in6 and route_iso are no more. All protocol families use struct route. I have changed the route cache, 'struct route', so that it does not contain storage space for a sockaddr. Instead, struct route points to a sockaddr coming from the pool the sockaddr belongs to. I added a new method to struct route, rtcache_setdst(), for setting the cache destination: int rtcache_setdst(struct route *, const struct sockaddr *); rtcache_setdst() returns 0 on success, or ENOMEM if no memory is available to create the sockaddr storage. It is now possible for rtcache_getdst() to return NULL if, say, rtcache_setdst() failed. I check the return value for NULL everywhere in the kernel. 4 Each routing domain (struct domain) has a list of live route caches, dom_rtcache. rtflushall(sa_family_t af) looks up the domain indicated by 'af', walks the domain's list of route caches and invalidates each one.
2007-05-03 00:40:22 +04:00
odst = rtcache_getdst(ro);
if (odst == NULL)
;
else if (sockaddr_cmp(odst, dst) != 0)
rtcache_free(ro);
else if ((rt = rtcache_validate(ro)) == NULL)
rtcache_clear(ro);
Eliminate address family-specific route caches (struct route, struct route_in6, struct route_iso), replacing all caches with a struct route. The principle benefit of this change is that all of the protocol families can benefit from route cache-invalidation, which is necessary for correct routing. Route-cache invalidation fixes an ancient PR, kern/3508, at long last; it fixes various other PRs, also. Discussions with and ideas from Joerg Sonnenberger influenced this work tremendously. Of course, all design oversights and bugs are mine. DETAILS 1 I added to each address family a pool of sockaddrs. I have introduced routines for allocating, copying, and duplicating, and freeing sockaddrs: struct sockaddr *sockaddr_alloc(sa_family_t af, int flags); struct sockaddr *sockaddr_copy(struct sockaddr *dst, const struct sockaddr *src); struct sockaddr *sockaddr_dup(const struct sockaddr *src, int flags); void sockaddr_free(struct sockaddr *sa); sockaddr_alloc() returns either a sockaddr from the pool belonging to the specified family, or NULL if the pool is exhausted. The returned sockaddr has the right size for that family; sa_family and sa_len fields are initialized to the family and sockaddr length---e.g., sa_family = AF_INET and sa_len = sizeof(struct sockaddr_in). sockaddr_free() puts the given sockaddr back into its family's pool. sockaddr_dup() and sockaddr_copy() work analogously to strdup() and strcpy(), respectively. sockaddr_copy() KASSERTs that the family of the destination and source sockaddrs are alike. The 'flags' argumet for sockaddr_alloc() and sockaddr_dup() is passed directly to pool_get(9). 2 I added routines for initializing sockaddrs in each address family, sockaddr_in_init(), sockaddr_in6_init(), sockaddr_iso_init(), etc. They are fairly self-explanatory. 3 structs route_in6 and route_iso are no more. All protocol families use struct route. I have changed the route cache, 'struct route', so that it does not contain storage space for a sockaddr. Instead, struct route points to a sockaddr coming from the pool the sockaddr belongs to. I added a new method to struct route, rtcache_setdst(), for setting the cache destination: int rtcache_setdst(struct route *, const struct sockaddr *); rtcache_setdst() returns 0 on success, or ENOMEM if no memory is available to create the sockaddr storage. It is now possible for rtcache_getdst() to return NULL if, say, rtcache_setdst() failed. I check the return value for NULL everywhere in the kernel. 4 Each routing domain (struct domain) has a list of live route caches, dom_rtcache. rtflushall(sa_family_t af) looks up the domain indicated by 'af', walks the domain's list of route caches and invalidates each one.
2007-05-03 00:40:22 +04:00
if (rt == NULL) {
*hitp = 0;
if (rtcache_setdst(ro, dst) == 0)
rt = _rtcache_init(ro, clone);
} else
*hitp = 1;
Eliminate address family-specific route caches (struct route, struct route_in6, struct route_iso), replacing all caches with a struct route. The principle benefit of this change is that all of the protocol families can benefit from route cache-invalidation, which is necessary for correct routing. Route-cache invalidation fixes an ancient PR, kern/3508, at long last; it fixes various other PRs, also. Discussions with and ideas from Joerg Sonnenberger influenced this work tremendously. Of course, all design oversights and bugs are mine. DETAILS 1 I added to each address family a pool of sockaddrs. I have introduced routines for allocating, copying, and duplicating, and freeing sockaddrs: struct sockaddr *sockaddr_alloc(sa_family_t af, int flags); struct sockaddr *sockaddr_copy(struct sockaddr *dst, const struct sockaddr *src); struct sockaddr *sockaddr_dup(const struct sockaddr *src, int flags); void sockaddr_free(struct sockaddr *sa); sockaddr_alloc() returns either a sockaddr from the pool belonging to the specified family, or NULL if the pool is exhausted. The returned sockaddr has the right size for that family; sa_family and sa_len fields are initialized to the family and sockaddr length---e.g., sa_family = AF_INET and sa_len = sizeof(struct sockaddr_in). sockaddr_free() puts the given sockaddr back into its family's pool. sockaddr_dup() and sockaddr_copy() work analogously to strdup() and strcpy(), respectively. sockaddr_copy() KASSERTs that the family of the destination and source sockaddrs are alike. The 'flags' argumet for sockaddr_alloc() and sockaddr_dup() is passed directly to pool_get(9). 2 I added routines for initializing sockaddrs in each address family, sockaddr_in_init(), sockaddr_in6_init(), sockaddr_iso_init(), etc. They are fairly self-explanatory. 3 structs route_in6 and route_iso are no more. All protocol families use struct route. I have changed the route cache, 'struct route', so that it does not contain storage space for a sockaddr. Instead, struct route points to a sockaddr coming from the pool the sockaddr belongs to. I added a new method to struct route, rtcache_setdst(), for setting the cache destination: int rtcache_setdst(struct route *, const struct sockaddr *); rtcache_setdst() returns 0 on success, or ENOMEM if no memory is available to create the sockaddr storage. It is now possible for rtcache_getdst() to return NULL if, say, rtcache_setdst() failed. I check the return value for NULL everywhere in the kernel. 4 Each routing domain (struct domain) has a list of live route caches, dom_rtcache. rtflushall(sa_family_t af) looks up the domain indicated by 'af', walks the domain's list of route caches and invalidates each one.
2007-05-03 00:40:22 +04:00
*** Summary *** When a link-layer address changes (e.g., ifconfig ex0 link 02:de:ad:be:ef:02 active), send a gratuitous ARP and/or a Neighbor Advertisement to update the network-/link-layer address bindings on our LAN peers. Refuse a change of ethernet address to the address 00:00:00:00:00:00 or to any multicast/broadcast address. (Thanks matt@.) Reorder ifnet ioctl operations so that driver ioctls may inherit the functions of their "class"---ether_ioctl(), fddi_ioctl(), et cetera---and the class ioctls may inherit from the generic ioctl, ifioctl_common(), but both driver- and class-ioctls may override the generic behavior. Make network drivers share more code. Distinguish a "factory" link-layer address from others for the purposes of both protecting that address from deletion and computing EUI64. Return consistent, appropriate error codes from network drivers. Improve readability. KNF. *** Details *** In if_attach(), always initialize the interface ioctl routine, ifnet->if_ioctl, if the driver has not already initialized it. Delete if_ioctl == NULL tests everywhere else, because it cannot happen. In the ioctl routines of network interfaces, inherit common ioctl behaviors by calling either ifioctl_common() or whichever ioctl routine is appropriate for the class of interface---e.g., ether_ioctl() for ethernets. Stop (ab)using SIOCSIFADDR and start to use SIOCINITIFADDR. In the user->kernel interface, SIOCSIFADDR's argument was an ifreq, but on the protocol->ifnet interface, SIOCSIFADDR's argument was an ifaddr. That was confusing, and it would work against me as I make it possible for a network interface to overload most ioctls. On the protocol->ifnet interface, replace SIOCSIFADDR with SIOCINITIFADDR. In ifioctl(), return EPERM if userland tries to invoke SIOCINITIFADDR. In ifioctl(), give the interface the first shot at handling most interface ioctls, and give the protocol the second shot, instead of the other way around. Finally, let compatibility code (COMPAT_OSOCK) take a shot. Pull device initialization out of switch statements under SIOCINITIFADDR. For example, pull ..._init() out of any switch statement that looks like this: switch (...->sa_family) { case ...: ..._init(); ... break; ... default: ..._init(); ... break; } Rewrite many if-else clauses that handle all permutations of IFF_UP and IFF_RUNNING to use a switch statement, switch (x & (IFF_UP|IFF_RUNNING)) { case 0: ... break; case IFF_RUNNING: ... break; case IFF_UP: ... break; case IFF_UP|IFF_RUNNING: ... break; } unifdef lots of code containing #ifdef FreeBSD, #ifdef NetBSD, and #ifdef SIOCSIFMTU, especially in fwip(4) and in ndis(4). In ipw(4), remove an if_set_sadl() call that is out of place. In nfe(4), reuse the jumbo MTU logic in ether_ioctl(). Let ethernets register a callback for setting h/w state such as promiscuous mode and the multicast filter in accord with a change in the if_flags: ether_set_ifflags_cb() registers a callback that returns ENETRESET if the caller should reset the ethernet by calling if_init(), 0 on success, != 0 on failure. Pull common code from ex(4), gem(4), nfe(4), sip(4), tlp(4), vge(4) into ether_ioctl(), and register if_flags callbacks for those drivers. Return ENOTTY instead of EINVAL for inappropriate ioctls. In zyd(4), use ENXIO instead of ENOTTY to indicate that the device is not any longer attached. Add to if_set_sadl() a boolean 'factory' argument that indicates whether a link-layer address was assigned by the factory or some other source. In a comment, recommend using the factory address for generating an EUI64, and update in6_get_hw_ifid() to prefer a factory address to any other link-layer address. Add a routing message, RTM_LLINFO_UPD, that tells protocols to update the binding of network-layer addresses to link-layer addresses. Implement this message in IPv4 and IPv6 by sending a gratuitous ARP or a neighbor advertisement, respectively. Generate RTM_LLINFO_UPD messages on a change of an interface's link-layer address. In ether_ioctl(), do not let SIOCALIFADDR set a link-layer address that is broadcast/multicast or equal to 00:00:00:00:00:00. Make ether_ioctl() call ifioctl_common() to handle ioctls that it does not understand. In gif(4), initialize if_softc and use it, instead of assuming that the gif_softc and ifp overlap. Let ifioctl_common() handle SIOCGIFADDR. Sprinkle rtcache_invariants(), which checks on DIAGNOSTIC kernels that certain invariants on a struct route are satisfied. In agr(4), rewrite agr_ioctl_filter() to be a bit more explicit about the ioctls that we do not allow on an agr(4) member interface. bzero -> memset. Delete unnecessary casts to void *. Use sockaddr_in_init() and sockaddr_in6_init(). Compare pointers with NULL instead of "testing truth". Replace some instances of (type *)0 with NULL. Change some K&R prototypes to ANSI C, and join lines.
2008-11-07 03:20:01 +03:00
rtcache_invariants(ro);
return rt;
Eliminate address family-specific route caches (struct route, struct route_in6, struct route_iso), replacing all caches with a struct route. The principle benefit of this change is that all of the protocol families can benefit from route cache-invalidation, which is necessary for correct routing. Route-cache invalidation fixes an ancient PR, kern/3508, at long last; it fixes various other PRs, also. Discussions with and ideas from Joerg Sonnenberger influenced this work tremendously. Of course, all design oversights and bugs are mine. DETAILS 1 I added to each address family a pool of sockaddrs. I have introduced routines for allocating, copying, and duplicating, and freeing sockaddrs: struct sockaddr *sockaddr_alloc(sa_family_t af, int flags); struct sockaddr *sockaddr_copy(struct sockaddr *dst, const struct sockaddr *src); struct sockaddr *sockaddr_dup(const struct sockaddr *src, int flags); void sockaddr_free(struct sockaddr *sa); sockaddr_alloc() returns either a sockaddr from the pool belonging to the specified family, or NULL if the pool is exhausted. The returned sockaddr has the right size for that family; sa_family and sa_len fields are initialized to the family and sockaddr length---e.g., sa_family = AF_INET and sa_len = sizeof(struct sockaddr_in). sockaddr_free() puts the given sockaddr back into its family's pool. sockaddr_dup() and sockaddr_copy() work analogously to strdup() and strcpy(), respectively. sockaddr_copy() KASSERTs that the family of the destination and source sockaddrs are alike. The 'flags' argumet for sockaddr_alloc() and sockaddr_dup() is passed directly to pool_get(9). 2 I added routines for initializing sockaddrs in each address family, sockaddr_in_init(), sockaddr_in6_init(), sockaddr_iso_init(), etc. They are fairly self-explanatory. 3 structs route_in6 and route_iso are no more. All protocol families use struct route. I have changed the route cache, 'struct route', so that it does not contain storage space for a sockaddr. Instead, struct route points to a sockaddr coming from the pool the sockaddr belongs to. I added a new method to struct route, rtcache_setdst(), for setting the cache destination: int rtcache_setdst(struct route *, const struct sockaddr *); rtcache_setdst() returns 0 on success, or ENOMEM if no memory is available to create the sockaddr storage. It is now possible for rtcache_getdst() to return NULL if, say, rtcache_setdst() failed. I check the return value for NULL everywhere in the kernel. 4 Each routing domain (struct domain) has a list of live route caches, dom_rtcache. rtflushall(sa_family_t af) looks up the domain indicated by 'af', walks the domain's list of route caches and invalidates each one.
2007-05-03 00:40:22 +04:00
}
void
rtcache_free(struct route *ro)
{
rtcache_clear(ro);
if (ro->ro_sa != NULL) {
sockaddr_free(ro->ro_sa);
ro->ro_sa = NULL;
}
*** Summary *** When a link-layer address changes (e.g., ifconfig ex0 link 02:de:ad:be:ef:02 active), send a gratuitous ARP and/or a Neighbor Advertisement to update the network-/link-layer address bindings on our LAN peers. Refuse a change of ethernet address to the address 00:00:00:00:00:00 or to any multicast/broadcast address. (Thanks matt@.) Reorder ifnet ioctl operations so that driver ioctls may inherit the functions of their "class"---ether_ioctl(), fddi_ioctl(), et cetera---and the class ioctls may inherit from the generic ioctl, ifioctl_common(), but both driver- and class-ioctls may override the generic behavior. Make network drivers share more code. Distinguish a "factory" link-layer address from others for the purposes of both protecting that address from deletion and computing EUI64. Return consistent, appropriate error codes from network drivers. Improve readability. KNF. *** Details *** In if_attach(), always initialize the interface ioctl routine, ifnet->if_ioctl, if the driver has not already initialized it. Delete if_ioctl == NULL tests everywhere else, because it cannot happen. In the ioctl routines of network interfaces, inherit common ioctl behaviors by calling either ifioctl_common() or whichever ioctl routine is appropriate for the class of interface---e.g., ether_ioctl() for ethernets. Stop (ab)using SIOCSIFADDR and start to use SIOCINITIFADDR. In the user->kernel interface, SIOCSIFADDR's argument was an ifreq, but on the protocol->ifnet interface, SIOCSIFADDR's argument was an ifaddr. That was confusing, and it would work against me as I make it possible for a network interface to overload most ioctls. On the protocol->ifnet interface, replace SIOCSIFADDR with SIOCINITIFADDR. In ifioctl(), return EPERM if userland tries to invoke SIOCINITIFADDR. In ifioctl(), give the interface the first shot at handling most interface ioctls, and give the protocol the second shot, instead of the other way around. Finally, let compatibility code (COMPAT_OSOCK) take a shot. Pull device initialization out of switch statements under SIOCINITIFADDR. For example, pull ..._init() out of any switch statement that looks like this: switch (...->sa_family) { case ...: ..._init(); ... break; ... default: ..._init(); ... break; } Rewrite many if-else clauses that handle all permutations of IFF_UP and IFF_RUNNING to use a switch statement, switch (x & (IFF_UP|IFF_RUNNING)) { case 0: ... break; case IFF_RUNNING: ... break; case IFF_UP: ... break; case IFF_UP|IFF_RUNNING: ... break; } unifdef lots of code containing #ifdef FreeBSD, #ifdef NetBSD, and #ifdef SIOCSIFMTU, especially in fwip(4) and in ndis(4). In ipw(4), remove an if_set_sadl() call that is out of place. In nfe(4), reuse the jumbo MTU logic in ether_ioctl(). Let ethernets register a callback for setting h/w state such as promiscuous mode and the multicast filter in accord with a change in the if_flags: ether_set_ifflags_cb() registers a callback that returns ENETRESET if the caller should reset the ethernet by calling if_init(), 0 on success, != 0 on failure. Pull common code from ex(4), gem(4), nfe(4), sip(4), tlp(4), vge(4) into ether_ioctl(), and register if_flags callbacks for those drivers. Return ENOTTY instead of EINVAL for inappropriate ioctls. In zyd(4), use ENXIO instead of ENOTTY to indicate that the device is not any longer attached. Add to if_set_sadl() a boolean 'factory' argument that indicates whether a link-layer address was assigned by the factory or some other source. In a comment, recommend using the factory address for generating an EUI64, and update in6_get_hw_ifid() to prefer a factory address to any other link-layer address. Add a routing message, RTM_LLINFO_UPD, that tells protocols to update the binding of network-layer addresses to link-layer addresses. Implement this message in IPv4 and IPv6 by sending a gratuitous ARP or a neighbor advertisement, respectively. Generate RTM_LLINFO_UPD messages on a change of an interface's link-layer address. In ether_ioctl(), do not let SIOCALIFADDR set a link-layer address that is broadcast/multicast or equal to 00:00:00:00:00:00. Make ether_ioctl() call ifioctl_common() to handle ioctls that it does not understand. In gif(4), initialize if_softc and use it, instead of assuming that the gif_softc and ifp overlap. Let ifioctl_common() handle SIOCGIFADDR. Sprinkle rtcache_invariants(), which checks on DIAGNOSTIC kernels that certain invariants on a struct route are satisfied. In agr(4), rewrite agr_ioctl_filter() to be a bit more explicit about the ioctls that we do not allow on an agr(4) member interface. bzero -> memset. Delete unnecessary casts to void *. Use sockaddr_in_init() and sockaddr_in6_init(). Compare pointers with NULL instead of "testing truth". Replace some instances of (type *)0 with NULL. Change some K&R prototypes to ANSI C, and join lines.
2008-11-07 03:20:01 +03:00
rtcache_invariants(ro);
}
Eliminate address family-specific route caches (struct route, struct route_in6, struct route_iso), replacing all caches with a struct route. The principle benefit of this change is that all of the protocol families can benefit from route cache-invalidation, which is necessary for correct routing. Route-cache invalidation fixes an ancient PR, kern/3508, at long last; it fixes various other PRs, also. Discussions with and ideas from Joerg Sonnenberger influenced this work tremendously. Of course, all design oversights and bugs are mine. DETAILS 1 I added to each address family a pool of sockaddrs. I have introduced routines for allocating, copying, and duplicating, and freeing sockaddrs: struct sockaddr *sockaddr_alloc(sa_family_t af, int flags); struct sockaddr *sockaddr_copy(struct sockaddr *dst, const struct sockaddr *src); struct sockaddr *sockaddr_dup(const struct sockaddr *src, int flags); void sockaddr_free(struct sockaddr *sa); sockaddr_alloc() returns either a sockaddr from the pool belonging to the specified family, or NULL if the pool is exhausted. The returned sockaddr has the right size for that family; sa_family and sa_len fields are initialized to the family and sockaddr length---e.g., sa_family = AF_INET and sa_len = sizeof(struct sockaddr_in). sockaddr_free() puts the given sockaddr back into its family's pool. sockaddr_dup() and sockaddr_copy() work analogously to strdup() and strcpy(), respectively. sockaddr_copy() KASSERTs that the family of the destination and source sockaddrs are alike. The 'flags' argumet for sockaddr_alloc() and sockaddr_dup() is passed directly to pool_get(9). 2 I added routines for initializing sockaddrs in each address family, sockaddr_in_init(), sockaddr_in6_init(), sockaddr_iso_init(), etc. They are fairly self-explanatory. 3 structs route_in6 and route_iso are no more. All protocol families use struct route. I have changed the route cache, 'struct route', so that it does not contain storage space for a sockaddr. Instead, struct route points to a sockaddr coming from the pool the sockaddr belongs to. I added a new method to struct route, rtcache_setdst(), for setting the cache destination: int rtcache_setdst(struct route *, const struct sockaddr *); rtcache_setdst() returns 0 on success, or ENOMEM if no memory is available to create the sockaddr storage. It is now possible for rtcache_getdst() to return NULL if, say, rtcache_setdst() failed. I check the return value for NULL everywhere in the kernel. 4 Each routing domain (struct domain) has a list of live route caches, dom_rtcache. rtflushall(sa_family_t af) looks up the domain indicated by 'af', walks the domain's list of route caches and invalidates each one.
2007-05-03 00:40:22 +04:00
int
rtcache_setdst(struct route *ro, const struct sockaddr *sa)
{
Eliminate address family-specific route caches (struct route, struct route_in6, struct route_iso), replacing all caches with a struct route. The principle benefit of this change is that all of the protocol families can benefit from route cache-invalidation, which is necessary for correct routing. Route-cache invalidation fixes an ancient PR, kern/3508, at long last; it fixes various other PRs, also. Discussions with and ideas from Joerg Sonnenberger influenced this work tremendously. Of course, all design oversights and bugs are mine. DETAILS 1 I added to each address family a pool of sockaddrs. I have introduced routines for allocating, copying, and duplicating, and freeing sockaddrs: struct sockaddr *sockaddr_alloc(sa_family_t af, int flags); struct sockaddr *sockaddr_copy(struct sockaddr *dst, const struct sockaddr *src); struct sockaddr *sockaddr_dup(const struct sockaddr *src, int flags); void sockaddr_free(struct sockaddr *sa); sockaddr_alloc() returns either a sockaddr from the pool belonging to the specified family, or NULL if the pool is exhausted. The returned sockaddr has the right size for that family; sa_family and sa_len fields are initialized to the family and sockaddr length---e.g., sa_family = AF_INET and sa_len = sizeof(struct sockaddr_in). sockaddr_free() puts the given sockaddr back into its family's pool. sockaddr_dup() and sockaddr_copy() work analogously to strdup() and strcpy(), respectively. sockaddr_copy() KASSERTs that the family of the destination and source sockaddrs are alike. The 'flags' argumet for sockaddr_alloc() and sockaddr_dup() is passed directly to pool_get(9). 2 I added routines for initializing sockaddrs in each address family, sockaddr_in_init(), sockaddr_in6_init(), sockaddr_iso_init(), etc. They are fairly self-explanatory. 3 structs route_in6 and route_iso are no more. All protocol families use struct route. I have changed the route cache, 'struct route', so that it does not contain storage space for a sockaddr. Instead, struct route points to a sockaddr coming from the pool the sockaddr belongs to. I added a new method to struct route, rtcache_setdst(), for setting the cache destination: int rtcache_setdst(struct route *, const struct sockaddr *); rtcache_setdst() returns 0 on success, or ENOMEM if no memory is available to create the sockaddr storage. It is now possible for rtcache_getdst() to return NULL if, say, rtcache_setdst() failed. I check the return value for NULL everywhere in the kernel. 4 Each routing domain (struct domain) has a list of live route caches, dom_rtcache. rtflushall(sa_family_t af) looks up the domain indicated by 'af', walks the domain's list of route caches and invalidates each one.
2007-05-03 00:40:22 +04:00
KASSERT(sa != NULL);
*** Summary *** When a link-layer address changes (e.g., ifconfig ex0 link 02:de:ad:be:ef:02 active), send a gratuitous ARP and/or a Neighbor Advertisement to update the network-/link-layer address bindings on our LAN peers. Refuse a change of ethernet address to the address 00:00:00:00:00:00 or to any multicast/broadcast address. (Thanks matt@.) Reorder ifnet ioctl operations so that driver ioctls may inherit the functions of their "class"---ether_ioctl(), fddi_ioctl(), et cetera---and the class ioctls may inherit from the generic ioctl, ifioctl_common(), but both driver- and class-ioctls may override the generic behavior. Make network drivers share more code. Distinguish a "factory" link-layer address from others for the purposes of both protecting that address from deletion and computing EUI64. Return consistent, appropriate error codes from network drivers. Improve readability. KNF. *** Details *** In if_attach(), always initialize the interface ioctl routine, ifnet->if_ioctl, if the driver has not already initialized it. Delete if_ioctl == NULL tests everywhere else, because it cannot happen. In the ioctl routines of network interfaces, inherit common ioctl behaviors by calling either ifioctl_common() or whichever ioctl routine is appropriate for the class of interface---e.g., ether_ioctl() for ethernets. Stop (ab)using SIOCSIFADDR and start to use SIOCINITIFADDR. In the user->kernel interface, SIOCSIFADDR's argument was an ifreq, but on the protocol->ifnet interface, SIOCSIFADDR's argument was an ifaddr. That was confusing, and it would work against me as I make it possible for a network interface to overload most ioctls. On the protocol->ifnet interface, replace SIOCSIFADDR with SIOCINITIFADDR. In ifioctl(), return EPERM if userland tries to invoke SIOCINITIFADDR. In ifioctl(), give the interface the first shot at handling most interface ioctls, and give the protocol the second shot, instead of the other way around. Finally, let compatibility code (COMPAT_OSOCK) take a shot. Pull device initialization out of switch statements under SIOCINITIFADDR. For example, pull ..._init() out of any switch statement that looks like this: switch (...->sa_family) { case ...: ..._init(); ... break; ... default: ..._init(); ... break; } Rewrite many if-else clauses that handle all permutations of IFF_UP and IFF_RUNNING to use a switch statement, switch (x & (IFF_UP|IFF_RUNNING)) { case 0: ... break; case IFF_RUNNING: ... break; case IFF_UP: ... break; case IFF_UP|IFF_RUNNING: ... break; } unifdef lots of code containing #ifdef FreeBSD, #ifdef NetBSD, and #ifdef SIOCSIFMTU, especially in fwip(4) and in ndis(4). In ipw(4), remove an if_set_sadl() call that is out of place. In nfe(4), reuse the jumbo MTU logic in ether_ioctl(). Let ethernets register a callback for setting h/w state such as promiscuous mode and the multicast filter in accord with a change in the if_flags: ether_set_ifflags_cb() registers a callback that returns ENETRESET if the caller should reset the ethernet by calling if_init(), 0 on success, != 0 on failure. Pull common code from ex(4), gem(4), nfe(4), sip(4), tlp(4), vge(4) into ether_ioctl(), and register if_flags callbacks for those drivers. Return ENOTTY instead of EINVAL for inappropriate ioctls. In zyd(4), use ENXIO instead of ENOTTY to indicate that the device is not any longer attached. Add to if_set_sadl() a boolean 'factory' argument that indicates whether a link-layer address was assigned by the factory or some other source. In a comment, recommend using the factory address for generating an EUI64, and update in6_get_hw_ifid() to prefer a factory address to any other link-layer address. Add a routing message, RTM_LLINFO_UPD, that tells protocols to update the binding of network-layer addresses to link-layer addresses. Implement this message in IPv4 and IPv6 by sending a gratuitous ARP or a neighbor advertisement, respectively. Generate RTM_LLINFO_UPD messages on a change of an interface's link-layer address. In ether_ioctl(), do not let SIOCALIFADDR set a link-layer address that is broadcast/multicast or equal to 00:00:00:00:00:00. Make ether_ioctl() call ifioctl_common() to handle ioctls that it does not understand. In gif(4), initialize if_softc and use it, instead of assuming that the gif_softc and ifp overlap. Let ifioctl_common() handle SIOCGIFADDR. Sprinkle rtcache_invariants(), which checks on DIAGNOSTIC kernels that certain invariants on a struct route are satisfied. In agr(4), rewrite agr_ioctl_filter() to be a bit more explicit about the ioctls that we do not allow on an agr(4) member interface. bzero -> memset. Delete unnecessary casts to void *. Use sockaddr_in_init() and sockaddr_in6_init(). Compare pointers with NULL instead of "testing truth". Replace some instances of (type *)0 with NULL. Change some K&R prototypes to ANSI C, and join lines.
2008-11-07 03:20:01 +03:00
rtcache_invariants(ro);
Eliminate address family-specific route caches (struct route, struct route_in6, struct route_iso), replacing all caches with a struct route. The principle benefit of this change is that all of the protocol families can benefit from route cache-invalidation, which is necessary for correct routing. Route-cache invalidation fixes an ancient PR, kern/3508, at long last; it fixes various other PRs, also. Discussions with and ideas from Joerg Sonnenberger influenced this work tremendously. Of course, all design oversights and bugs are mine. DETAILS 1 I added to each address family a pool of sockaddrs. I have introduced routines for allocating, copying, and duplicating, and freeing sockaddrs: struct sockaddr *sockaddr_alloc(sa_family_t af, int flags); struct sockaddr *sockaddr_copy(struct sockaddr *dst, const struct sockaddr *src); struct sockaddr *sockaddr_dup(const struct sockaddr *src, int flags); void sockaddr_free(struct sockaddr *sa); sockaddr_alloc() returns either a sockaddr from the pool belonging to the specified family, or NULL if the pool is exhausted. The returned sockaddr has the right size for that family; sa_family and sa_len fields are initialized to the family and sockaddr length---e.g., sa_family = AF_INET and sa_len = sizeof(struct sockaddr_in). sockaddr_free() puts the given sockaddr back into its family's pool. sockaddr_dup() and sockaddr_copy() work analogously to strdup() and strcpy(), respectively. sockaddr_copy() KASSERTs that the family of the destination and source sockaddrs are alike. The 'flags' argumet for sockaddr_alloc() and sockaddr_dup() is passed directly to pool_get(9). 2 I added routines for initializing sockaddrs in each address family, sockaddr_in_init(), sockaddr_in6_init(), sockaddr_iso_init(), etc. They are fairly self-explanatory. 3 structs route_in6 and route_iso are no more. All protocol families use struct route. I have changed the route cache, 'struct route', so that it does not contain storage space for a sockaddr. Instead, struct route points to a sockaddr coming from the pool the sockaddr belongs to. I added a new method to struct route, rtcache_setdst(), for setting the cache destination: int rtcache_setdst(struct route *, const struct sockaddr *); rtcache_setdst() returns 0 on success, or ENOMEM if no memory is available to create the sockaddr storage. It is now possible for rtcache_getdst() to return NULL if, say, rtcache_setdst() failed. I check the return value for NULL everywhere in the kernel. 4 Each routing domain (struct domain) has a list of live route caches, dom_rtcache. rtflushall(sa_family_t af) looks up the domain indicated by 'af', walks the domain's list of route caches and invalidates each one.
2007-05-03 00:40:22 +04:00
if (ro->ro_sa != NULL && ro->ro_sa->sa_family == sa->sa_family) {
rtcache_clear(ro);
*** Summary *** When a link-layer address changes (e.g., ifconfig ex0 link 02:de:ad:be:ef:02 active), send a gratuitous ARP and/or a Neighbor Advertisement to update the network-/link-layer address bindings on our LAN peers. Refuse a change of ethernet address to the address 00:00:00:00:00:00 or to any multicast/broadcast address. (Thanks matt@.) Reorder ifnet ioctl operations so that driver ioctls may inherit the functions of their "class"---ether_ioctl(), fddi_ioctl(), et cetera---and the class ioctls may inherit from the generic ioctl, ifioctl_common(), but both driver- and class-ioctls may override the generic behavior. Make network drivers share more code. Distinguish a "factory" link-layer address from others for the purposes of both protecting that address from deletion and computing EUI64. Return consistent, appropriate error codes from network drivers. Improve readability. KNF. *** Details *** In if_attach(), always initialize the interface ioctl routine, ifnet->if_ioctl, if the driver has not already initialized it. Delete if_ioctl == NULL tests everywhere else, because it cannot happen. In the ioctl routines of network interfaces, inherit common ioctl behaviors by calling either ifioctl_common() or whichever ioctl routine is appropriate for the class of interface---e.g., ether_ioctl() for ethernets. Stop (ab)using SIOCSIFADDR and start to use SIOCINITIFADDR. In the user->kernel interface, SIOCSIFADDR's argument was an ifreq, but on the protocol->ifnet interface, SIOCSIFADDR's argument was an ifaddr. That was confusing, and it would work against me as I make it possible for a network interface to overload most ioctls. On the protocol->ifnet interface, replace SIOCSIFADDR with SIOCINITIFADDR. In ifioctl(), return EPERM if userland tries to invoke SIOCINITIFADDR. In ifioctl(), give the interface the first shot at handling most interface ioctls, and give the protocol the second shot, instead of the other way around. Finally, let compatibility code (COMPAT_OSOCK) take a shot. Pull device initialization out of switch statements under SIOCINITIFADDR. For example, pull ..._init() out of any switch statement that looks like this: switch (...->sa_family) { case ...: ..._init(); ... break; ... default: ..._init(); ... break; } Rewrite many if-else clauses that handle all permutations of IFF_UP and IFF_RUNNING to use a switch statement, switch (x & (IFF_UP|IFF_RUNNING)) { case 0: ... break; case IFF_RUNNING: ... break; case IFF_UP: ... break; case IFF_UP|IFF_RUNNING: ... break; } unifdef lots of code containing #ifdef FreeBSD, #ifdef NetBSD, and #ifdef SIOCSIFMTU, especially in fwip(4) and in ndis(4). In ipw(4), remove an if_set_sadl() call that is out of place. In nfe(4), reuse the jumbo MTU logic in ether_ioctl(). Let ethernets register a callback for setting h/w state such as promiscuous mode and the multicast filter in accord with a change in the if_flags: ether_set_ifflags_cb() registers a callback that returns ENETRESET if the caller should reset the ethernet by calling if_init(), 0 on success, != 0 on failure. Pull common code from ex(4), gem(4), nfe(4), sip(4), tlp(4), vge(4) into ether_ioctl(), and register if_flags callbacks for those drivers. Return ENOTTY instead of EINVAL for inappropriate ioctls. In zyd(4), use ENXIO instead of ENOTTY to indicate that the device is not any longer attached. Add to if_set_sadl() a boolean 'factory' argument that indicates whether a link-layer address was assigned by the factory or some other source. In a comment, recommend using the factory address for generating an EUI64, and update in6_get_hw_ifid() to prefer a factory address to any other link-layer address. Add a routing message, RTM_LLINFO_UPD, that tells protocols to update the binding of network-layer addresses to link-layer addresses. Implement this message in IPv4 and IPv6 by sending a gratuitous ARP or a neighbor advertisement, respectively. Generate RTM_LLINFO_UPD messages on a change of an interface's link-layer address. In ether_ioctl(), do not let SIOCALIFADDR set a link-layer address that is broadcast/multicast or equal to 00:00:00:00:00:00. Make ether_ioctl() call ifioctl_common() to handle ioctls that it does not understand. In gif(4), initialize if_softc and use it, instead of assuming that the gif_softc and ifp overlap. Let ifioctl_common() handle SIOCGIFADDR. Sprinkle rtcache_invariants(), which checks on DIAGNOSTIC kernels that certain invariants on a struct route are satisfied. In agr(4), rewrite agr_ioctl_filter() to be a bit more explicit about the ioctls that we do not allow on an agr(4) member interface. bzero -> memset. Delete unnecessary casts to void *. Use sockaddr_in_init() and sockaddr_in6_init(). Compare pointers with NULL instead of "testing truth". Replace some instances of (type *)0 with NULL. Change some K&R prototypes to ANSI C, and join lines.
2008-11-07 03:20:01 +03:00
if (sockaddr_copy(ro->ro_sa, ro->ro_sa->sa_len, sa) != NULL) {
rtcache_invariants(ro);
return 0;
*** Summary *** When a link-layer address changes (e.g., ifconfig ex0 link 02:de:ad:be:ef:02 active), send a gratuitous ARP and/or a Neighbor Advertisement to update the network-/link-layer address bindings on our LAN peers. Refuse a change of ethernet address to the address 00:00:00:00:00:00 or to any multicast/broadcast address. (Thanks matt@.) Reorder ifnet ioctl operations so that driver ioctls may inherit the functions of their "class"---ether_ioctl(), fddi_ioctl(), et cetera---and the class ioctls may inherit from the generic ioctl, ifioctl_common(), but both driver- and class-ioctls may override the generic behavior. Make network drivers share more code. Distinguish a "factory" link-layer address from others for the purposes of both protecting that address from deletion and computing EUI64. Return consistent, appropriate error codes from network drivers. Improve readability. KNF. *** Details *** In if_attach(), always initialize the interface ioctl routine, ifnet->if_ioctl, if the driver has not already initialized it. Delete if_ioctl == NULL tests everywhere else, because it cannot happen. In the ioctl routines of network interfaces, inherit common ioctl behaviors by calling either ifioctl_common() or whichever ioctl routine is appropriate for the class of interface---e.g., ether_ioctl() for ethernets. Stop (ab)using SIOCSIFADDR and start to use SIOCINITIFADDR. In the user->kernel interface, SIOCSIFADDR's argument was an ifreq, but on the protocol->ifnet interface, SIOCSIFADDR's argument was an ifaddr. That was confusing, and it would work against me as I make it possible for a network interface to overload most ioctls. On the protocol->ifnet interface, replace SIOCSIFADDR with SIOCINITIFADDR. In ifioctl(), return EPERM if userland tries to invoke SIOCINITIFADDR. In ifioctl(), give the interface the first shot at handling most interface ioctls, and give the protocol the second shot, instead of the other way around. Finally, let compatibility code (COMPAT_OSOCK) take a shot. Pull device initialization out of switch statements under SIOCINITIFADDR. For example, pull ..._init() out of any switch statement that looks like this: switch (...->sa_family) { case ...: ..._init(); ... break; ... default: ..._init(); ... break; } Rewrite many if-else clauses that handle all permutations of IFF_UP and IFF_RUNNING to use a switch statement, switch (x & (IFF_UP|IFF_RUNNING)) { case 0: ... break; case IFF_RUNNING: ... break; case IFF_UP: ... break; case IFF_UP|IFF_RUNNING: ... break; } unifdef lots of code containing #ifdef FreeBSD, #ifdef NetBSD, and #ifdef SIOCSIFMTU, especially in fwip(4) and in ndis(4). In ipw(4), remove an if_set_sadl() call that is out of place. In nfe(4), reuse the jumbo MTU logic in ether_ioctl(). Let ethernets register a callback for setting h/w state such as promiscuous mode and the multicast filter in accord with a change in the if_flags: ether_set_ifflags_cb() registers a callback that returns ENETRESET if the caller should reset the ethernet by calling if_init(), 0 on success, != 0 on failure. Pull common code from ex(4), gem(4), nfe(4), sip(4), tlp(4), vge(4) into ether_ioctl(), and register if_flags callbacks for those drivers. Return ENOTTY instead of EINVAL for inappropriate ioctls. In zyd(4), use ENXIO instead of ENOTTY to indicate that the device is not any longer attached. Add to if_set_sadl() a boolean 'factory' argument that indicates whether a link-layer address was assigned by the factory or some other source. In a comment, recommend using the factory address for generating an EUI64, and update in6_get_hw_ifid() to prefer a factory address to any other link-layer address. Add a routing message, RTM_LLINFO_UPD, that tells protocols to update the binding of network-layer addresses to link-layer addresses. Implement this message in IPv4 and IPv6 by sending a gratuitous ARP or a neighbor advertisement, respectively. Generate RTM_LLINFO_UPD messages on a change of an interface's link-layer address. In ether_ioctl(), do not let SIOCALIFADDR set a link-layer address that is broadcast/multicast or equal to 00:00:00:00:00:00. Make ether_ioctl() call ifioctl_common() to handle ioctls that it does not understand. In gif(4), initialize if_softc and use it, instead of assuming that the gif_softc and ifp overlap. Let ifioctl_common() handle SIOCGIFADDR. Sprinkle rtcache_invariants(), which checks on DIAGNOSTIC kernels that certain invariants on a struct route are satisfied. In agr(4), rewrite agr_ioctl_filter() to be a bit more explicit about the ioctls that we do not allow on an agr(4) member interface. bzero -> memset. Delete unnecessary casts to void *. Use sockaddr_in_init() and sockaddr_in6_init(). Compare pointers with NULL instead of "testing truth". Replace some instances of (type *)0 with NULL. Change some K&R prototypes to ANSI C, and join lines.
2008-11-07 03:20:01 +03:00
}
sockaddr_free(ro->ro_sa);
Eliminate address family-specific route caches (struct route, struct route_in6, struct route_iso), replacing all caches with a struct route. The principle benefit of this change is that all of the protocol families can benefit from route cache-invalidation, which is necessary for correct routing. Route-cache invalidation fixes an ancient PR, kern/3508, at long last; it fixes various other PRs, also. Discussions with and ideas from Joerg Sonnenberger influenced this work tremendously. Of course, all design oversights and bugs are mine. DETAILS 1 I added to each address family a pool of sockaddrs. I have introduced routines for allocating, copying, and duplicating, and freeing sockaddrs: struct sockaddr *sockaddr_alloc(sa_family_t af, int flags); struct sockaddr *sockaddr_copy(struct sockaddr *dst, const struct sockaddr *src); struct sockaddr *sockaddr_dup(const struct sockaddr *src, int flags); void sockaddr_free(struct sockaddr *sa); sockaddr_alloc() returns either a sockaddr from the pool belonging to the specified family, or NULL if the pool is exhausted. The returned sockaddr has the right size for that family; sa_family and sa_len fields are initialized to the family and sockaddr length---e.g., sa_family = AF_INET and sa_len = sizeof(struct sockaddr_in). sockaddr_free() puts the given sockaddr back into its family's pool. sockaddr_dup() and sockaddr_copy() work analogously to strdup() and strcpy(), respectively. sockaddr_copy() KASSERTs that the family of the destination and source sockaddrs are alike. The 'flags' argumet for sockaddr_alloc() and sockaddr_dup() is passed directly to pool_get(9). 2 I added routines for initializing sockaddrs in each address family, sockaddr_in_init(), sockaddr_in6_init(), sockaddr_iso_init(), etc. They are fairly self-explanatory. 3 structs route_in6 and route_iso are no more. All protocol families use struct route. I have changed the route cache, 'struct route', so that it does not contain storage space for a sockaddr. Instead, struct route points to a sockaddr coming from the pool the sockaddr belongs to. I added a new method to struct route, rtcache_setdst(), for setting the cache destination: int rtcache_setdst(struct route *, const struct sockaddr *); rtcache_setdst() returns 0 on success, or ENOMEM if no memory is available to create the sockaddr storage. It is now possible for rtcache_getdst() to return NULL if, say, rtcache_setdst() failed. I check the return value for NULL everywhere in the kernel. 4 Each routing domain (struct domain) has a list of live route caches, dom_rtcache. rtflushall(sa_family_t af) looks up the domain indicated by 'af', walks the domain's list of route caches and invalidates each one.
2007-05-03 00:40:22 +04:00
} else if (ro->ro_sa != NULL)
rtcache_free(ro); /* free ro_sa, wrong family */
KASSERT(ro->_ro_rt == NULL);
if ((ro->ro_sa = sockaddr_dup(sa, M_NOWAIT)) == NULL) {
*** Summary *** When a link-layer address changes (e.g., ifconfig ex0 link 02:de:ad:be:ef:02 active), send a gratuitous ARP and/or a Neighbor Advertisement to update the network-/link-layer address bindings on our LAN peers. Refuse a change of ethernet address to the address 00:00:00:00:00:00 or to any multicast/broadcast address. (Thanks matt@.) Reorder ifnet ioctl operations so that driver ioctls may inherit the functions of their "class"---ether_ioctl(), fddi_ioctl(), et cetera---and the class ioctls may inherit from the generic ioctl, ifioctl_common(), but both driver- and class-ioctls may override the generic behavior. Make network drivers share more code. Distinguish a "factory" link-layer address from others for the purposes of both protecting that address from deletion and computing EUI64. Return consistent, appropriate error codes from network drivers. Improve readability. KNF. *** Details *** In if_attach(), always initialize the interface ioctl routine, ifnet->if_ioctl, if the driver has not already initialized it. Delete if_ioctl == NULL tests everywhere else, because it cannot happen. In the ioctl routines of network interfaces, inherit common ioctl behaviors by calling either ifioctl_common() or whichever ioctl routine is appropriate for the class of interface---e.g., ether_ioctl() for ethernets. Stop (ab)using SIOCSIFADDR and start to use SIOCINITIFADDR. In the user->kernel interface, SIOCSIFADDR's argument was an ifreq, but on the protocol->ifnet interface, SIOCSIFADDR's argument was an ifaddr. That was confusing, and it would work against me as I make it possible for a network interface to overload most ioctls. On the protocol->ifnet interface, replace SIOCSIFADDR with SIOCINITIFADDR. In ifioctl(), return EPERM if userland tries to invoke SIOCINITIFADDR. In ifioctl(), give the interface the first shot at handling most interface ioctls, and give the protocol the second shot, instead of the other way around. Finally, let compatibility code (COMPAT_OSOCK) take a shot. Pull device initialization out of switch statements under SIOCINITIFADDR. For example, pull ..._init() out of any switch statement that looks like this: switch (...->sa_family) { case ...: ..._init(); ... break; ... default: ..._init(); ... break; } Rewrite many if-else clauses that handle all permutations of IFF_UP and IFF_RUNNING to use a switch statement, switch (x & (IFF_UP|IFF_RUNNING)) { case 0: ... break; case IFF_RUNNING: ... break; case IFF_UP: ... break; case IFF_UP|IFF_RUNNING: ... break; } unifdef lots of code containing #ifdef FreeBSD, #ifdef NetBSD, and #ifdef SIOCSIFMTU, especially in fwip(4) and in ndis(4). In ipw(4), remove an if_set_sadl() call that is out of place. In nfe(4), reuse the jumbo MTU logic in ether_ioctl(). Let ethernets register a callback for setting h/w state such as promiscuous mode and the multicast filter in accord with a change in the if_flags: ether_set_ifflags_cb() registers a callback that returns ENETRESET if the caller should reset the ethernet by calling if_init(), 0 on success, != 0 on failure. Pull common code from ex(4), gem(4), nfe(4), sip(4), tlp(4), vge(4) into ether_ioctl(), and register if_flags callbacks for those drivers. Return ENOTTY instead of EINVAL for inappropriate ioctls. In zyd(4), use ENXIO instead of ENOTTY to indicate that the device is not any longer attached. Add to if_set_sadl() a boolean 'factory' argument that indicates whether a link-layer address was assigned by the factory or some other source. In a comment, recommend using the factory address for generating an EUI64, and update in6_get_hw_ifid() to prefer a factory address to any other link-layer address. Add a routing message, RTM_LLINFO_UPD, that tells protocols to update the binding of network-layer addresses to link-layer addresses. Implement this message in IPv4 and IPv6 by sending a gratuitous ARP or a neighbor advertisement, respectively. Generate RTM_LLINFO_UPD messages on a change of an interface's link-layer address. In ether_ioctl(), do not let SIOCALIFADDR set a link-layer address that is broadcast/multicast or equal to 00:00:00:00:00:00. Make ether_ioctl() call ifioctl_common() to handle ioctls that it does not understand. In gif(4), initialize if_softc and use it, instead of assuming that the gif_softc and ifp overlap. Let ifioctl_common() handle SIOCGIFADDR. Sprinkle rtcache_invariants(), which checks on DIAGNOSTIC kernels that certain invariants on a struct route are satisfied. In agr(4), rewrite agr_ioctl_filter() to be a bit more explicit about the ioctls that we do not allow on an agr(4) member interface. bzero -> memset. Delete unnecessary casts to void *. Use sockaddr_in_init() and sockaddr_in6_init(). Compare pointers with NULL instead of "testing truth". Replace some instances of (type *)0 with NULL. Change some K&R prototypes to ANSI C, and join lines.
2008-11-07 03:20:01 +03:00
rtcache_invariants(ro);
Eliminate address family-specific route caches (struct route, struct route_in6, struct route_iso), replacing all caches with a struct route. The principle benefit of this change is that all of the protocol families can benefit from route cache-invalidation, which is necessary for correct routing. Route-cache invalidation fixes an ancient PR, kern/3508, at long last; it fixes various other PRs, also. Discussions with and ideas from Joerg Sonnenberger influenced this work tremendously. Of course, all design oversights and bugs are mine. DETAILS 1 I added to each address family a pool of sockaddrs. I have introduced routines for allocating, copying, and duplicating, and freeing sockaddrs: struct sockaddr *sockaddr_alloc(sa_family_t af, int flags); struct sockaddr *sockaddr_copy(struct sockaddr *dst, const struct sockaddr *src); struct sockaddr *sockaddr_dup(const struct sockaddr *src, int flags); void sockaddr_free(struct sockaddr *sa); sockaddr_alloc() returns either a sockaddr from the pool belonging to the specified family, or NULL if the pool is exhausted. The returned sockaddr has the right size for that family; sa_family and sa_len fields are initialized to the family and sockaddr length---e.g., sa_family = AF_INET and sa_len = sizeof(struct sockaddr_in). sockaddr_free() puts the given sockaddr back into its family's pool. sockaddr_dup() and sockaddr_copy() work analogously to strdup() and strcpy(), respectively. sockaddr_copy() KASSERTs that the family of the destination and source sockaddrs are alike. The 'flags' argumet for sockaddr_alloc() and sockaddr_dup() is passed directly to pool_get(9). 2 I added routines for initializing sockaddrs in each address family, sockaddr_in_init(), sockaddr_in6_init(), sockaddr_iso_init(), etc. They are fairly self-explanatory. 3 structs route_in6 and route_iso are no more. All protocol families use struct route. I have changed the route cache, 'struct route', so that it does not contain storage space for a sockaddr. Instead, struct route points to a sockaddr coming from the pool the sockaddr belongs to. I added a new method to struct route, rtcache_setdst(), for setting the cache destination: int rtcache_setdst(struct route *, const struct sockaddr *); rtcache_setdst() returns 0 on success, or ENOMEM if no memory is available to create the sockaddr storage. It is now possible for rtcache_getdst() to return NULL if, say, rtcache_setdst() failed. I check the return value for NULL everywhere in the kernel. 4 Each routing domain (struct domain) has a list of live route caches, dom_rtcache. rtflushall(sa_family_t af) looks up the domain indicated by 'af', walks the domain's list of route caches and invalidates each one.
2007-05-03 00:40:22 +04:00
return ENOMEM;
}
*** Summary *** When a link-layer address changes (e.g., ifconfig ex0 link 02:de:ad:be:ef:02 active), send a gratuitous ARP and/or a Neighbor Advertisement to update the network-/link-layer address bindings on our LAN peers. Refuse a change of ethernet address to the address 00:00:00:00:00:00 or to any multicast/broadcast address. (Thanks matt@.) Reorder ifnet ioctl operations so that driver ioctls may inherit the functions of their "class"---ether_ioctl(), fddi_ioctl(), et cetera---and the class ioctls may inherit from the generic ioctl, ifioctl_common(), but both driver- and class-ioctls may override the generic behavior. Make network drivers share more code. Distinguish a "factory" link-layer address from others for the purposes of both protecting that address from deletion and computing EUI64. Return consistent, appropriate error codes from network drivers. Improve readability. KNF. *** Details *** In if_attach(), always initialize the interface ioctl routine, ifnet->if_ioctl, if the driver has not already initialized it. Delete if_ioctl == NULL tests everywhere else, because it cannot happen. In the ioctl routines of network interfaces, inherit common ioctl behaviors by calling either ifioctl_common() or whichever ioctl routine is appropriate for the class of interface---e.g., ether_ioctl() for ethernets. Stop (ab)using SIOCSIFADDR and start to use SIOCINITIFADDR. In the user->kernel interface, SIOCSIFADDR's argument was an ifreq, but on the protocol->ifnet interface, SIOCSIFADDR's argument was an ifaddr. That was confusing, and it would work against me as I make it possible for a network interface to overload most ioctls. On the protocol->ifnet interface, replace SIOCSIFADDR with SIOCINITIFADDR. In ifioctl(), return EPERM if userland tries to invoke SIOCINITIFADDR. In ifioctl(), give the interface the first shot at handling most interface ioctls, and give the protocol the second shot, instead of the other way around. Finally, let compatibility code (COMPAT_OSOCK) take a shot. Pull device initialization out of switch statements under SIOCINITIFADDR. For example, pull ..._init() out of any switch statement that looks like this: switch (...->sa_family) { case ...: ..._init(); ... break; ... default: ..._init(); ... break; } Rewrite many if-else clauses that handle all permutations of IFF_UP and IFF_RUNNING to use a switch statement, switch (x & (IFF_UP|IFF_RUNNING)) { case 0: ... break; case IFF_RUNNING: ... break; case IFF_UP: ... break; case IFF_UP|IFF_RUNNING: ... break; } unifdef lots of code containing #ifdef FreeBSD, #ifdef NetBSD, and #ifdef SIOCSIFMTU, especially in fwip(4) and in ndis(4). In ipw(4), remove an if_set_sadl() call that is out of place. In nfe(4), reuse the jumbo MTU logic in ether_ioctl(). Let ethernets register a callback for setting h/w state such as promiscuous mode and the multicast filter in accord with a change in the if_flags: ether_set_ifflags_cb() registers a callback that returns ENETRESET if the caller should reset the ethernet by calling if_init(), 0 on success, != 0 on failure. Pull common code from ex(4), gem(4), nfe(4), sip(4), tlp(4), vge(4) into ether_ioctl(), and register if_flags callbacks for those drivers. Return ENOTTY instead of EINVAL for inappropriate ioctls. In zyd(4), use ENXIO instead of ENOTTY to indicate that the device is not any longer attached. Add to if_set_sadl() a boolean 'factory' argument that indicates whether a link-layer address was assigned by the factory or some other source. In a comment, recommend using the factory address for generating an EUI64, and update in6_get_hw_ifid() to prefer a factory address to any other link-layer address. Add a routing message, RTM_LLINFO_UPD, that tells protocols to update the binding of network-layer addresses to link-layer addresses. Implement this message in IPv4 and IPv6 by sending a gratuitous ARP or a neighbor advertisement, respectively. Generate RTM_LLINFO_UPD messages on a change of an interface's link-layer address. In ether_ioctl(), do not let SIOCALIFADDR set a link-layer address that is broadcast/multicast or equal to 00:00:00:00:00:00. Make ether_ioctl() call ifioctl_common() to handle ioctls that it does not understand. In gif(4), initialize if_softc and use it, instead of assuming that the gif_softc and ifp overlap. Let ifioctl_common() handle SIOCGIFADDR. Sprinkle rtcache_invariants(), which checks on DIAGNOSTIC kernels that certain invariants on a struct route are satisfied. In agr(4), rewrite agr_ioctl_filter() to be a bit more explicit about the ioctls that we do not allow on an agr(4) member interface. bzero -> memset. Delete unnecessary casts to void *. Use sockaddr_in_init() and sockaddr_in6_init(). Compare pointers with NULL instead of "testing truth". Replace some instances of (type *)0 with NULL. Change some K&R prototypes to ANSI C, and join lines.
2008-11-07 03:20:01 +03:00
rtcache_invariants(ro);
Eliminate address family-specific route caches (struct route, struct route_in6, struct route_iso), replacing all caches with a struct route. The principle benefit of this change is that all of the protocol families can benefit from route cache-invalidation, which is necessary for correct routing. Route-cache invalidation fixes an ancient PR, kern/3508, at long last; it fixes various other PRs, also. Discussions with and ideas from Joerg Sonnenberger influenced this work tremendously. Of course, all design oversights and bugs are mine. DETAILS 1 I added to each address family a pool of sockaddrs. I have introduced routines for allocating, copying, and duplicating, and freeing sockaddrs: struct sockaddr *sockaddr_alloc(sa_family_t af, int flags); struct sockaddr *sockaddr_copy(struct sockaddr *dst, const struct sockaddr *src); struct sockaddr *sockaddr_dup(const struct sockaddr *src, int flags); void sockaddr_free(struct sockaddr *sa); sockaddr_alloc() returns either a sockaddr from the pool belonging to the specified family, or NULL if the pool is exhausted. The returned sockaddr has the right size for that family; sa_family and sa_len fields are initialized to the family and sockaddr length---e.g., sa_family = AF_INET and sa_len = sizeof(struct sockaddr_in). sockaddr_free() puts the given sockaddr back into its family's pool. sockaddr_dup() and sockaddr_copy() work analogously to strdup() and strcpy(), respectively. sockaddr_copy() KASSERTs that the family of the destination and source sockaddrs are alike. The 'flags' argumet for sockaddr_alloc() and sockaddr_dup() is passed directly to pool_get(9). 2 I added routines for initializing sockaddrs in each address family, sockaddr_in_init(), sockaddr_in6_init(), sockaddr_iso_init(), etc. They are fairly self-explanatory. 3 structs route_in6 and route_iso are no more. All protocol families use struct route. I have changed the route cache, 'struct route', so that it does not contain storage space for a sockaddr. Instead, struct route points to a sockaddr coming from the pool the sockaddr belongs to. I added a new method to struct route, rtcache_setdst(), for setting the cache destination: int rtcache_setdst(struct route *, const struct sockaddr *); rtcache_setdst() returns 0 on success, or ENOMEM if no memory is available to create the sockaddr storage. It is now possible for rtcache_getdst() to return NULL if, say, rtcache_setdst() failed. I check the return value for NULL everywhere in the kernel. 4 Each routing domain (struct domain) has a list of live route caches, dom_rtcache. rtflushall(sa_family_t af) looks up the domain indicated by 'af', walks the domain's list of route caches and invalidates each one.
2007-05-03 00:40:22 +04:00
return 0;
}
const struct sockaddr *
rt_settag(struct rtentry *rt, const struct sockaddr *tag)
{
if (rt->rt_tag != tag) {
if (rt->rt_tag != NULL)
sockaddr_free(rt->rt_tag);
rt->rt_tag = sockaddr_dup(tag, M_NOWAIT);
}
return rt->rt_tag;
}
struct sockaddr *
rt_gettag(struct rtentry *rt)
{
return rt->rt_tag;
}