NetBSD/sys/netinet/ip_carp.c

2485 lines
59 KiB
C
Raw Normal View History

/* $NetBSD: ip_carp.c,v 1.99 2018/06/26 06:48:03 msaitoh Exp $ */
/* $OpenBSD: ip_carp.c,v 1.113 2005/11/04 08:11:54 mcbride Exp $ */
/*
* Copyright (c) 2002 Michael Shalayeff. All rights reserved.
* Copyright (c) 2003 Ryan McBride. All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``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 AUTHOR OR HIS RELATIVES 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 MIND, 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.
*/
2015-08-25 01:21:26 +03:00
#ifdef _KERNEL_OPT
#include "opt_inet.h"
#include "opt_mbuftrace.h"
2015-08-25 01:21:26 +03:00
#endif
2007-12-11 15:29:11 +03:00
#include <sys/cdefs.h>
__KERNEL_RCSID(0, "$NetBSD: ip_carp.c,v 1.99 2018/06/26 06:48:03 msaitoh Exp $");
2007-12-11 15:29:11 +03:00
/*
* TODO:
* - iface reconfigure
* - support for hardware checksum calculations;
*
*/
#include <sys/param.h>
#include <sys/proc.h>
#include <sys/mbuf.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/callout.h>
#include <sys/ioctl.h>
#include <sys/errno.h>
#include <sys/device.h>
#include <sys/time.h>
#include <sys/kernel.h>
#include <sys/kauth.h>
#include <sys/sysctl.h>
#include <sys/ucred.h>
#include <sys/syslog.h>
#include <sys/acct.h>
First step of random number subsystem rework described in <20111022023242.BA26F14A158@mail.netbsd.org>. This change includes the following: An initial cleanup and minor reorganization of the entropy pool code in sys/dev/rnd.c and sys/dev/rndpool.c. Several bugs are fixed. Some effort is made to accumulate entropy more quickly at boot time. A generic interface, "rndsink", is added, for stream generators to request that they be re-keyed with good quality entropy from the pool as soon as it is available. The arc4random()/arc4randbytes() implementation in libkern is adjusted to use the rndsink interface for rekeying, which helps address the problem of low-quality keys at boot time. An implementation of the FIPS 140-2 statistical tests for random number generator quality is provided (libkern/rngtest.c). This is based on Greg Rose's implementation from Qualcomm. A new random stream generator, nist_ctr_drbg, is provided. It is based on an implementation of the NIST SP800-90 CTR_DRBG by Henric Jungheim. This generator users AES in a modified counter mode to generate a backtracking-resistant random stream. An abstraction layer, "cprng", is provided for in-kernel consumers of randomness. The arc4random/arc4randbytes API is deprecated for in-kernel use. It is replaced by "cprng_strong". The current cprng_fast implementation wraps the existing arc4random implementation. The current cprng_strong implementation wraps the new CTR_DRBG implementation. Both interfaces are rekeyed from the entropy pool automatically at intervals justifiable from best current cryptographic practice. In some quick tests, cprng_fast() is about the same speed as the old arc4randbytes(), and cprng_strong() is about 20% faster than rnd_extract_data(). Performance is expected to improve. The AES code in src/crypto/rijndael is no longer an optional kernel component, as it is required by cprng_strong, which is not an optional kernel component. The entropy pool output is subjected to the rngtest tests at startup time; if it fails, the system will reboot. There is approximately a 3/10000 chance of a false positive from these tests. Entropy pool _input_ from hardware random numbers is subjected to the rngtest tests at attach time, as well as the FIPS continuous-output test, to detect bad or stuck hardware RNGs; if any are detected, they are detached, but the system continues to run. A problem with rndctl(8) is fixed -- datastructures with pointers in arrays are no longer passed to userspace (this was not a security problem, but rather a major issue for compat32). A new kernel will require a new rndctl. The sysctl kern.arandom() and kern.urandom() nodes are hooked up to the new generators, but the /dev/*random pseudodevices are not, yet. Manual pages for the new kernel interfaces are forthcoming.
2011-11-20 02:51:18 +04:00
#include <sys/cprng.h>
#include <sys/cpu.h>
#include <sys/pserialize.h>
#include <sys/psref.h>
#include <net/if.h>
#include <net/pfil.h>
#include <net/if_types.h>
#include <net/if_ether.h>
#include <net/route.h>
#include <net/netisr.h>
#include <net/net_stats.h>
#include <netinet/if_inarp.h>
#include <netinet/wqinput.h>
#if NFDDI > 0
#include <net/if_fddi.h>
#endif
#if NTOKEN > 0
#include <net/if_token.h>
#endif
#ifdef INET
#include <netinet/in.h>
#include <netinet/in_systm.h>
#include <netinet/in_var.h>
#include <netinet/ip.h>
#include <netinet/ip_var.h>
#include <net/if_dl.h>
#endif
#ifdef INET6
#include <netinet/icmp6.h>
#include <netinet/ip6.h>
#include <netinet6/ip6_var.h>
#include <netinet6/nd6.h>
#include <netinet6/scope6_var.h>
#include <netinet6/in6_var.h>
#endif
#include <net/bpf.h>
#include <sys/sha1.h>
#include <netinet/ip_carp.h>
#include "ioconf.h"
struct carp_mc_entry {
LIST_ENTRY(carp_mc_entry) mc_entries;
union {
struct ether_multi *mcu_enm;
} mc_u;
struct sockaddr_storage mc_addr;
};
#define mc_enm mc_u.mcu_enm
struct carp_softc {
struct ethercom sc_ac;
#define sc_if sc_ac.ec_if
#define sc_carpdev sc_ac.ec_if.if_carpdev
int ah_cookie;
int lh_cookie;
struct ip_moptions sc_imo;
#ifdef INET6
struct ip6_moptions sc_im6o;
#endif /* INET6 */
TAILQ_ENTRY(carp_softc) sc_list;
enum { INIT = 0, BACKUP, MASTER } sc_state;
int sc_suppress;
int sc_bow_out;
int sc_sendad_errors;
#define CARP_SENDAD_MAX_ERRORS 3
int sc_sendad_success;
#define CARP_SENDAD_MIN_SUCCESS 3
int sc_vhid;
int sc_advskew;
int sc_naddrs;
int sc_naddrs6;
int sc_advbase; /* seconds */
int sc_init_counter;
u_int64_t sc_counter;
/* authentication */
#define CARP_HMAC_PAD 64
unsigned char sc_key[CARP_KEY_LEN];
unsigned char sc_pad[CARP_HMAC_PAD];
SHA1_CTX sc_sha1;
u_int32_t sc_hashkey[2];
struct callout sc_ad_tmo; /* advertisement timeout */
struct callout sc_md_tmo; /* master down timeout */
struct callout sc_md6_tmo; /* master down timeout */
LIST_HEAD(__carp_mchead, carp_mc_entry) carp_mc_listhead;
};
int carp_suppress_preempt = 0;
static int carp_opts[CARPCTL_MAXID] = { 0, 1, 0, 0, 0 }; /* XXX for now */
2008-04-15 10:03:28 +04:00
static percpu_t *carpstat_percpu;
#define CARP_STATINC(x) _NET_STATINC(carpstat_percpu, x)
#ifdef MBUFTRACE
static struct mowner carp_proto_mowner_rx = MOWNER_INIT("carp", "rx");
static struct mowner carp_proto_mowner_tx = MOWNER_INIT("carp", "tx");
static struct mowner carp_proto6_mowner_rx = MOWNER_INIT("carp6", "rx");
static struct mowner carp_proto6_mowner_tx = MOWNER_INIT("carp6", "tx");
#endif
struct carp_if {
TAILQ_HEAD(, carp_softc) vhif_vrs;
int vhif_nvrs;
struct ifnet *vhif_ifp;
};
#define CARP_LOG(sc, s) \
if (carp_opts[CARPCTL_LOG]) { \
if (sc) \
log(LOG_INFO, "%s: ", \
(sc)->sc_if.if_xname); \
else \
log(LOG_INFO, "carp: "); \
addlog s; \
addlog("\n"); \
}
static void carp_hmac_prepare(struct carp_softc *);
static void carp_hmac_generate(struct carp_softc *, u_int32_t *,
unsigned char *);
static int carp_hmac_verify(struct carp_softc *, u_int32_t *,
unsigned char *);
static void carp_setroute(struct carp_softc *, int);
static void carp_proto_input_c(struct mbuf *, struct carp_header *,
sa_family_t);
static void carpdetach(struct carp_softc *);
static void carp_prepare_ad(struct mbuf *, struct carp_softc *,
struct carp_header *);
static void carp_send_ad_all(void);
static void carp_send_ad(void *);
static void carp_send_arp(struct carp_softc *);
static void carp_master_down(void *);
static int carp_ioctl(struct ifnet *, u_long, void *);
static void carp_start(struct ifnet *);
static void carp_setrun(struct carp_softc *, sa_family_t);
static void carp_set_state(struct carp_softc *, int);
static int carp_addrcount(struct carp_if *, struct in_ifaddr *, int);
enum { CARP_COUNT_MASTER, CARP_COUNT_RUNNING };
static void carp_multicast_cleanup(struct carp_softc *);
static int carp_set_ifp(struct carp_softc *, struct ifnet *);
static void carp_set_enaddr(struct carp_softc *);
#if 0
static void carp_addr_updated(void *);
#endif
static u_int32_t carp_hash(struct carp_softc *, u_char *);
static int carp_set_addr(struct carp_softc *, struct sockaddr_in *);
static int carp_join_multicast(struct carp_softc *);
#ifdef INET6
static void carp_send_na(struct carp_softc *);
static int carp_set_addr6(struct carp_softc *, struct sockaddr_in6 *);
static int carp_join_multicast6(struct carp_softc *);
#endif
static int carp_clone_create(struct if_clone *, int);
static int carp_clone_destroy(struct ifnet *);
static int carp_ether_addmulti(struct carp_softc *, struct ifreq *);
static int carp_ether_delmulti(struct carp_softc *, struct ifreq *);
static void carp_ether_purgemulti(struct carp_softc *);
static void sysctl_net_inet_carp_setup(struct sysctllog **);
/* workqueue-based pr_input */
static struct wqinput *carp_wqinput;
static void _carp_proto_input(struct mbuf *, int, int);
#ifdef INET6
static struct wqinput *carp6_wqinput;
static void _carp6_proto_input(struct mbuf *, int, int);
#endif
struct if_clone carp_cloner =
IF_CLONE_INITIALIZER("carp", carp_clone_create, carp_clone_destroy);
static __inline u_int16_t
carp_cksum(struct mbuf *m, int len)
{
return (in_cksum(m, len));
}
static __inline u_int16_t
carp6_cksum(struct mbuf *m, uint32_t off, uint32_t len)
{
return (in6_cksum(m, IPPROTO_CARP, off, len));
}
static void
carp_hmac_prepare(struct carp_softc *sc)
{
u_int8_t carp_version = CARP_VERSION, type = CARP_ADVERTISEMENT;
u_int8_t vhid = sc->sc_vhid & 0xff;
SHA1_CTX sha1ctx;
u_int32_t kmd[5];
struct ifaddr *ifa;
int i, found;
struct in_addr last, cur, in;
#ifdef INET6
struct in6_addr last6, cur6, in6;
#endif /* INET6 */
/* compute ipad from key */
2009-03-18 19:00:08 +03:00
memset(sc->sc_pad, 0, sizeof(sc->sc_pad));
memcpy(sc->sc_pad, sc->sc_key, sizeof(sc->sc_key));
for (i = 0; i < sizeof(sc->sc_pad); i++)
sc->sc_pad[i] ^= 0x36;
/* precompute first part of inner hash */
SHA1Init(&sc->sc_sha1);
SHA1Update(&sc->sc_sha1, sc->sc_pad, sizeof(sc->sc_pad));
SHA1Update(&sc->sc_sha1, (void *)&carp_version, sizeof(carp_version));
SHA1Update(&sc->sc_sha1, (void *)&type, sizeof(type));
/* generate a key for the arpbalance hash, before the vhid is hashed */
memcpy(&sha1ctx, &sc->sc_sha1, sizeof(sha1ctx));
SHA1Final((unsigned char *)kmd, &sha1ctx);
sc->sc_hashkey[0] = kmd[0] ^ kmd[1];
sc->sc_hashkey[1] = kmd[2] ^ kmd[3];
/* the rest of the precomputation */
SHA1Update(&sc->sc_sha1, (void *)&vhid, sizeof(vhid));
/* Hash the addresses from smallest to largest, not interface order */
#ifdef INET
cur.s_addr = 0;
do {
int s;
found = 0;
last = cur;
cur.s_addr = 0xffffffff;
s = pserialize_read_enter();
IFADDR_READER_FOREACH(ifa, &sc->sc_if) {
in.s_addr = ifatoia(ifa)->ia_addr.sin_addr.s_addr;
if (ifa->ifa_addr->sa_family == AF_INET &&
ntohl(in.s_addr) > ntohl(last.s_addr) &&
ntohl(in.s_addr) < ntohl(cur.s_addr)) {
cur.s_addr = in.s_addr;
found++;
}
}
pserialize_read_exit(s);
if (found)
SHA1Update(&sc->sc_sha1, (void *)&cur, sizeof(cur));
} while (found);
#endif /* INET */
#ifdef INET6
memset(&cur6, 0x00, sizeof(cur6));
do {
int s;
found = 0;
last6 = cur6;
memset(&cur6, 0xff, sizeof(cur6));
s = pserialize_read_enter();
IFADDR_READER_FOREACH(ifa, &sc->sc_if) {
in6 = ifatoia6(ifa)->ia_addr.sin6_addr;
if (IN6_IS_ADDR_LINKLOCAL(&in6))
in6.s6_addr16[1] = 0;
if (ifa->ifa_addr->sa_family == AF_INET6 &&
memcmp(&in6, &last6, sizeof(in6)) > 0 &&
memcmp(&in6, &cur6, sizeof(in6)) < 0) {
cur6 = in6;
found++;
}
}
pserialize_read_exit(s);
if (found)
SHA1Update(&sc->sc_sha1, (void *)&cur6, sizeof(cur6));
} while (found);
#endif /* INET6 */
/* convert ipad to opad */
for (i = 0; i < sizeof(sc->sc_pad); i++)
sc->sc_pad[i] ^= 0x36 ^ 0x5c;
}
static void
carp_hmac_generate(struct carp_softc *sc, u_int32_t counter[2],
unsigned char md[20])
{
SHA1_CTX sha1ctx;
/* fetch first half of inner hash */
memcpy(&sha1ctx, &sc->sc_sha1, sizeof(sha1ctx));
SHA1Update(&sha1ctx, (void *)counter, sizeof(sc->sc_counter));
SHA1Final(md, &sha1ctx);
/* outer hash */
SHA1Init(&sha1ctx);
SHA1Update(&sha1ctx, sc->sc_pad, sizeof(sc->sc_pad));
SHA1Update(&sha1ctx, md, 20);
SHA1Final(md, &sha1ctx);
}
static int
carp_hmac_verify(struct carp_softc *sc, u_int32_t counter[2],
unsigned char md[20])
{
unsigned char md2[20];
carp_hmac_generate(sc, counter, md2);
2009-03-18 18:14:29 +03:00
return (memcmp(md, md2, sizeof(md2)));
}
static void
carp_setroute(struct carp_softc *sc, int cmd)
{
struct ifaddr *ifa;
int s, bound;
KERNEL_LOCK(1, NULL);
bound = curlwp_bind();
s = pserialize_read_enter();
IFADDR_READER_FOREACH(ifa, &sc->sc_if) {
struct psref psref;
ifa_acquire(ifa, &psref);
pserialize_read_exit(s);
switch (ifa->ifa_addr->sa_family) {
case AF_INET: {
int count = 0;
struct rtentry *rt;
int hr_otherif, nr_ourif;
/*
* Avoid screwing with the routes if there are other
* carp interfaces which are master and have the same
* address.
*/
if (sc->sc_carpdev != NULL &&
sc->sc_carpdev->if_carp != NULL) {
count = carp_addrcount(
(struct carp_if *)sc->sc_carpdev->if_carp,
ifatoia(ifa), CARP_COUNT_MASTER);
if ((cmd == RTM_ADD && count != 1) ||
(cmd == RTM_DELETE && count != 0))
continue;
}
/* Remove the existing host route, if any */
rtrequest(RTM_DELETE, ifa->ifa_addr,
ifa->ifa_addr, ifa->ifa_netmask,
RTF_HOST, 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 = NULL;
(void)rtrequest(RTM_GET, ifa->ifa_addr, ifa->ifa_addr,
ifa->ifa_netmask, RTF_HOST, &rt);
hr_otherif = (rt && rt->rt_ifp != &sc->sc_if &&
2016-04-04 10:37:07 +03:00
(rt->rt_flags & RTF_CONNECTED));
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 != NULL) {
Make the routing table and rtcaches MP-safe See the following descriptions for details. Proposed on tech-kern and tech-net Overview -------- We protect the routing table with a rwock and protect rtcaches with another rwlock. Each rtentry is protected from being freed or updated via reference counting and psref. Global rwlocks -------------- There are two rwlocks; one for the routing table (rt_lock) and the other for rtcaches (rtcache_lock). rtcache_lock covers all existing rtcaches; there may have room for optimizations (future work). The locking order is rtcache_lock first and rt_lock is next. rtentry references ------------------ References to an rtentry is managed with reference counting and psref. Either of the two mechanisms is used depending on where a rtentry is obtained. Reference counting is used when we obtain a rtentry from the routing table directly via rtalloc1 and rtrequest{,1} while psref is used when we obtain a rtentry from a rtcache via rtcache_* APIs. In both cases, a caller can sleep/block with holding an obtained rtentry. The reasons why we use two different mechanisms are (i) only using reference counting hurts the performance due to atomic instructions (rtcache case) (ii) ease of implementation; applying psref to APIs such rtaloc1 and rtrequest{,1} requires additional works (adding a local variable and an argument). We will finally migrate to use only psref but we can do it when we have a lockless routing table alternative. Reference counting for rtentry ------------------------------ rt_refcnt now doesn't count permanent references such as for rt_timers and rtcaches, instead it is used only for temporal references when obtaining a rtentry via rtalloc1 and rtrequest{,1}. We can do so because destroying a rtentry always involves removing references of rt_timers and rtcaches to the rtentry and we don't need to track such references. This also makes it easy to wait for readers to release references on deleting or updating a rtentry, i.e., we can simply wait until the reference counter is 0 or 1. (If there are permanent references the counter can be arbitrary.) rt_ref increments a reference counter of a rtentry and rt_unref decrements it. rt_ref is called inside APIs (rtalloc1 and rtrequest{,1} so users don't need to care about it while users must call rt_unref to an obtained rtentry after using it. rtfree is removed and we use rt_unref and rt_free instead. rt_unref now just decrements the counter of a given rtentry and rt_free just tries to destroy a given rtentry. See the next section for destructions of rtentries by rt_free. Destructions of rtentries ------------------------- We destroy a rtentry only when we call rtrequst{,1}(RTM_DELETE); the original implementation can destroy in any rtfree where it's the last reference. If we use reference counting or psref, it's easy to understand if the place that a rtentry is destroyed is fixed. rt_free waits for references to a given rtentry to be released before actually destroying the rtentry. rt_free uses a condition variable (cv_wait) (and psref_target_destroy for psref) to wait. Unfortunately rtrequst{,1}(RTM_DELETE) can be called in softint that we cannot use cv_wait. In that case, we have to defer the destruction to a workqueue. rtentry#rt_cv, rtentry#rt_psref and global variables (see rt_free_global) are added to conduct the procedure. Updates of rtentries -------------------- One difficulty to use refcnt/psref instead of rwlock for rtentry is updates of rtentries. We need an additional mechanism to prevent readers from seeing inconsistency of a rtentry being updated. We introduce RTF_UPDATING flag to rtentries that are updating. While the flag is set to a rtentry, users cannot acquire the rtentry. By doing so, we avoid users to see inconsistent rtentries. There are two options when a user tries to acquire a rtentry with the RTF_UPDATING flag; if a user runs in softint context the user fails to acquire a rtentry (NULL is returned). Otherwise a user waits until the update completes by waiting on cv. The procedure of a updater is simpler to destruction of a rtentry. Wait on cv (and psref) and after all readers left, proceed with the update. Global variables (see rt_update_global) are added to conduct the procedure. Currently we apply the mechanism to only RTM_CHANGE in rtsock.c. We would have to apply other codes. See "Known issues" section. psref for rtentry ----------------- When we obtain a rtentry from a rtcache via rtcache_* APIs, psref is used to reference to the rtentry. rtcache_ref acquires a reference to a rtentry with psref and rtcache_unref releases the reference after using it. rtcache_ref is called inside rtcache_* APIs and users don't need to take care of it while users must call rtcache_unref to release the reference. struct psref and int bound that is needed for psref is embedded into struct route. By doing so we don't need to add local variables and additional argument to APIs. However this adds another constraint to psref other than reference counting one's; holding a reference of an rtentry via a rtcache is allowed by just one caller at the same time. So we must not acquire a rtentry via a rtcache twice and avoid a recursive use of a rtcache. And also a rtcache must be arranged to be used by a LWP/softint at the same time somehow. For IP forwarding case, we have per-CPU rtcaches used in softint so the constraint is guaranteed. For a h rtcache of a PCB case, the constraint is guaranteed by the solock of each PCB. Any other cases (pf, ipf, stf and ipsec) are currently guaranteed by only the existence of the global locks (softnet_lock and/or KERNEL_LOCK). If we've found the cases that we cannot guarantee the constraint, we would need to introduce other rtcache APIs that use simple reference counting. psref of rtcache is created with IPL_SOFTNET and so rtcache shouldn't used at an IPL higher than IPL_SOFTNET. Note that rtcache_free is used to invalidate a given rtcache. We don't need another care by my change; just keep them as they are. Performance impact ------------------ When NET_MPSAFE is disabled the performance drop is 3% while when it's enabled the drop is increased to 11%. The difference comes from that currently we don't take any global locks and don't use psref if NET_MPSAFE is disabled. We can optimize the performance of the case of NET_MPSAFE on by reducing lookups of rtcache that uses psref; currently we do two lookups but we should be able to trim one of two. This is a future work. Known issues ------------ There are two known issues to be solved; one is that a caller of rtrequest(RTM_ADD) may change rtentry (see rtinit). We need to prevent new references during the update. Or we may be able to remove the code (perhaps, need more investigations). The other is rtredirect that updates a rtentry. We need to apply our update mechanism, however it's not easy because rtredirect is called in softint and we cannot apply our mechanism simply. One solution is to defer rtredirect to a workqueue but it requires some code restructuring.
2016-12-12 06:55:57 +03:00
rt_unref(rt);
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 = NULL;
}
/* Check for a network route on our interface */
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 = NULL;
(void)rtrequest(RTM_GET, ifa->ifa_addr, ifa->ifa_addr,
ifa->ifa_netmask, 0, &rt);
nr_ourif = (rt && rt->rt_ifp == &sc->sc_if);
switch (cmd) {
case RTM_ADD:
if (hr_otherif) {
ifa->ifa_rtrequest = NULL;
2016-04-04 10:37:07 +03:00
ifa->ifa_flags &= ~RTF_CONNECTED;
rtrequest(RTM_ADD, ifa->ifa_addr,
ifa->ifa_addr, ifa->ifa_netmask,
RTF_UP | RTF_HOST, NULL);
}
if (!hr_otherif || nr_ourif || !rt) {
2016-04-04 10:37:07 +03:00
if (nr_ourif &&
(rt->rt_flags & RTF_CONNECTED) == 0)
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
rtrequest(RTM_DELETE,
ifa->ifa_addr,
ifa->ifa_addr,
ifa->ifa_netmask, 0, NULL);
ifa->ifa_rtrequest = arp_rtrequest;
2016-04-04 10:37:07 +03:00
ifa->ifa_flags |= RTF_CONNECTED;
if (rtrequest(RTM_ADD, ifa->ifa_addr,
ifa->ifa_addr, ifa->ifa_netmask, 0,
NULL) == 0)
ifa->ifa_flags |= IFA_ROUTE;
}
break;
case RTM_DELETE:
break;
default:
break;
}
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 != NULL) {
Make the routing table and rtcaches MP-safe See the following descriptions for details. Proposed on tech-kern and tech-net Overview -------- We protect the routing table with a rwock and protect rtcaches with another rwlock. Each rtentry is protected from being freed or updated via reference counting and psref. Global rwlocks -------------- There are two rwlocks; one for the routing table (rt_lock) and the other for rtcaches (rtcache_lock). rtcache_lock covers all existing rtcaches; there may have room for optimizations (future work). The locking order is rtcache_lock first and rt_lock is next. rtentry references ------------------ References to an rtentry is managed with reference counting and psref. Either of the two mechanisms is used depending on where a rtentry is obtained. Reference counting is used when we obtain a rtentry from the routing table directly via rtalloc1 and rtrequest{,1} while psref is used when we obtain a rtentry from a rtcache via rtcache_* APIs. In both cases, a caller can sleep/block with holding an obtained rtentry. The reasons why we use two different mechanisms are (i) only using reference counting hurts the performance due to atomic instructions (rtcache case) (ii) ease of implementation; applying psref to APIs such rtaloc1 and rtrequest{,1} requires additional works (adding a local variable and an argument). We will finally migrate to use only psref but we can do it when we have a lockless routing table alternative. Reference counting for rtentry ------------------------------ rt_refcnt now doesn't count permanent references such as for rt_timers and rtcaches, instead it is used only for temporal references when obtaining a rtentry via rtalloc1 and rtrequest{,1}. We can do so because destroying a rtentry always involves removing references of rt_timers and rtcaches to the rtentry and we don't need to track such references. This also makes it easy to wait for readers to release references on deleting or updating a rtentry, i.e., we can simply wait until the reference counter is 0 or 1. (If there are permanent references the counter can be arbitrary.) rt_ref increments a reference counter of a rtentry and rt_unref decrements it. rt_ref is called inside APIs (rtalloc1 and rtrequest{,1} so users don't need to care about it while users must call rt_unref to an obtained rtentry after using it. rtfree is removed and we use rt_unref and rt_free instead. rt_unref now just decrements the counter of a given rtentry and rt_free just tries to destroy a given rtentry. See the next section for destructions of rtentries by rt_free. Destructions of rtentries ------------------------- We destroy a rtentry only when we call rtrequst{,1}(RTM_DELETE); the original implementation can destroy in any rtfree where it's the last reference. If we use reference counting or psref, it's easy to understand if the place that a rtentry is destroyed is fixed. rt_free waits for references to a given rtentry to be released before actually destroying the rtentry. rt_free uses a condition variable (cv_wait) (and psref_target_destroy for psref) to wait. Unfortunately rtrequst{,1}(RTM_DELETE) can be called in softint that we cannot use cv_wait. In that case, we have to defer the destruction to a workqueue. rtentry#rt_cv, rtentry#rt_psref and global variables (see rt_free_global) are added to conduct the procedure. Updates of rtentries -------------------- One difficulty to use refcnt/psref instead of rwlock for rtentry is updates of rtentries. We need an additional mechanism to prevent readers from seeing inconsistency of a rtentry being updated. We introduce RTF_UPDATING flag to rtentries that are updating. While the flag is set to a rtentry, users cannot acquire the rtentry. By doing so, we avoid users to see inconsistent rtentries. There are two options when a user tries to acquire a rtentry with the RTF_UPDATING flag; if a user runs in softint context the user fails to acquire a rtentry (NULL is returned). Otherwise a user waits until the update completes by waiting on cv. The procedure of a updater is simpler to destruction of a rtentry. Wait on cv (and psref) and after all readers left, proceed with the update. Global variables (see rt_update_global) are added to conduct the procedure. Currently we apply the mechanism to only RTM_CHANGE in rtsock.c. We would have to apply other codes. See "Known issues" section. psref for rtentry ----------------- When we obtain a rtentry from a rtcache via rtcache_* APIs, psref is used to reference to the rtentry. rtcache_ref acquires a reference to a rtentry with psref and rtcache_unref releases the reference after using it. rtcache_ref is called inside rtcache_* APIs and users don't need to take care of it while users must call rtcache_unref to release the reference. struct psref and int bound that is needed for psref is embedded into struct route. By doing so we don't need to add local variables and additional argument to APIs. However this adds another constraint to psref other than reference counting one's; holding a reference of an rtentry via a rtcache is allowed by just one caller at the same time. So we must not acquire a rtentry via a rtcache twice and avoid a recursive use of a rtcache. And also a rtcache must be arranged to be used by a LWP/softint at the same time somehow. For IP forwarding case, we have per-CPU rtcaches used in softint so the constraint is guaranteed. For a h rtcache of a PCB case, the constraint is guaranteed by the solock of each PCB. Any other cases (pf, ipf, stf and ipsec) are currently guaranteed by only the existence of the global locks (softnet_lock and/or KERNEL_LOCK). If we've found the cases that we cannot guarantee the constraint, we would need to introduce other rtcache APIs that use simple reference counting. psref of rtcache is created with IPL_SOFTNET and so rtcache shouldn't used at an IPL higher than IPL_SOFTNET. Note that rtcache_free is used to invalidate a given rtcache. We don't need another care by my change; just keep them as they are. Performance impact ------------------ When NET_MPSAFE is disabled the performance drop is 3% while when it's enabled the drop is increased to 11%. The difference comes from that currently we don't take any global locks and don't use psref if NET_MPSAFE is disabled. We can optimize the performance of the case of NET_MPSAFE on by reducing lookups of rtcache that uses psref; currently we do two lookups but we should be able to trim one of two. This is a future work. Known issues ------------ There are two known issues to be solved; one is that a caller of rtrequest(RTM_ADD) may change rtentry (see rtinit). We need to prevent new references during the update. Or we may be able to remove the code (perhaps, need more investigations). The other is rtredirect that updates a rtentry. We need to apply our update mechanism, however it's not easy because rtredirect is called in softint and we cannot apply our mechanism simply. One solution is to defer rtredirect to a workqueue but it requires some code restructuring.
2016-12-12 06:55:57 +03:00
rt_unref(rt);
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 = NULL;
}
break;
}
#ifdef INET6
case AF_INET6:
if (cmd == RTM_ADD)
in6_ifaddlocal(ifa);
else
in6_ifremlocal(ifa);
break;
#endif /* INET6 */
default:
break;
}
s = pserialize_read_enter();
ifa_release(ifa, &psref);
}
pserialize_read_exit(s);
curlwp_bindx(bound);
KERNEL_UNLOCK_ONE(NULL);
}
/*
* process input packet.
* we have rearranged checks order compared to the rfc,
* but it seems more efficient this way or not possible otherwise.
*/
static void
_carp_proto_input(struct mbuf *m, int hlen, int proto)
{
struct ip *ip = mtod(m, struct ip *);
struct carp_softc *sc = NULL;
struct carp_header *ch;
2013-10-18 23:48:36 +04:00
int iplen, len;
struct ifnet *rcvif;
2008-04-15 10:03:28 +04:00
CARP_STATINC(CARP_STAT_IPACKETS);
MCLAIM(m, &carp_proto_mowner_rx);
if (!carp_opts[CARPCTL_ALLOW]) {
m_freem(m);
return;
}
rcvif = m_get_rcvif_NOMPSAFE(m);
/* check if received on a valid carp interface */
if (rcvif->if_type != IFT_CARP) {
2008-04-15 10:03:28 +04:00
CARP_STATINC(CARP_STAT_BADIF);
CARP_LOG(sc, ("packet received on non-carp interface: %s",
rcvif->if_xname));
m_freem(m);
return;
}
/* verify that the IP TTL is 255. */
if (ip->ip_ttl != CARP_DFLTTL) {
2008-04-15 10:03:28 +04:00
CARP_STATINC(CARP_STAT_BADTTL);
CARP_LOG(sc, ("received ttl %d != %d on %s", ip->ip_ttl,
CARP_DFLTTL, rcvif->if_xname));
m_freem(m);
return;
}
/*
* verify that the received packet length is
* equal to the CARP header
*/
iplen = ip->ip_hl << 2;
len = iplen + sizeof(*ch);
if (len > m->m_pkthdr.len) {
2008-04-15 10:03:28 +04:00
CARP_STATINC(CARP_STAT_BADLEN);
CARP_LOG(sc, ("packet too short %d on %s", m->m_pkthdr.len,
rcvif->if_xname));
m_freem(m);
return;
}
if ((m = m_pullup(m, len)) == NULL) {
2008-04-15 10:03:28 +04:00
CARP_STATINC(CARP_STAT_HDROPS);
return;
}
ip = mtod(m, struct ip *);
ch = (struct carp_header *)((char *)ip + iplen);
/* verify the CARP checksum */
m->m_data += iplen;
if (carp_cksum(m, len - iplen)) {
2008-04-15 10:03:28 +04:00
CARP_STATINC(CARP_STAT_BADSUM);
CARP_LOG(sc, ("checksum failed on %s",
rcvif->if_xname));
m_freem(m);
return;
}
m->m_data -= iplen;
carp_proto_input_c(m, ch, AF_INET);
}
void
carp_proto_input(struct mbuf *m, ...)
{
wqinput_input(carp_wqinput, m, 0, 0);
}
#ifdef INET6
static void
_carp6_proto_input(struct mbuf *m, int off, int proto)
{
struct carp_softc *sc = NULL;
struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
struct carp_header *ch;
u_int len;
struct ifnet *rcvif;
2008-04-15 10:03:28 +04:00
CARP_STATINC(CARP_STAT_IPACKETS6);
MCLAIM(m, &carp_proto6_mowner_rx);
if (!carp_opts[CARPCTL_ALLOW]) {
m_freem(m);
return;
}
rcvif = m_get_rcvif_NOMPSAFE(m);
/* check if received on a valid carp interface */
if (rcvif->if_type != IFT_CARP) {
2008-04-15 10:03:28 +04:00
CARP_STATINC(CARP_STAT_BADIF);
CARP_LOG(sc, ("packet received on non-carp interface: %s",
rcvif->if_xname));
m_freem(m);
return;
}
/* verify that the IP TTL is 255 */
if (ip6->ip6_hlim != CARP_DFLTTL) {
2008-04-15 10:03:28 +04:00
CARP_STATINC(CARP_STAT_BADTTL);
CARP_LOG(sc, ("received ttl %d != %d on %s", ip6->ip6_hlim,
CARP_DFLTTL, rcvif->if_xname));
m_freem(m);
return;
}
/* verify that we have a complete carp packet */
len = m->m_len;
M_REGION_GET(ch, struct carp_header *, m, off, sizeof(*ch));
if (ch == NULL) {
2008-04-15 10:03:28 +04:00
CARP_STATINC(CARP_STAT_BADLEN);
CARP_LOG(sc, ("packet size %u too small", len));
return;
}
/* verify the CARP checksum */
if (carp6_cksum(m, off, sizeof(*ch))) {
2008-04-15 10:03:28 +04:00
CARP_STATINC(CARP_STAT_BADSUM);
CARP_LOG(sc, ("checksum failed, on %s", rcvif->if_xname));
m_freem(m);
return;
}
carp_proto_input_c(m, ch, AF_INET6);
return;
}
int
carp6_proto_input(struct mbuf **mp, int *offp, int proto)
{
wqinput_input(carp6_wqinput, *mp, *offp, proto);
return IPPROTO_DONE;
}
#endif /* INET6 */
static void
carp_proto_input_c(struct mbuf *m, struct carp_header *ch, sa_family_t af)
{
struct carp_softc *sc;
u_int64_t tmp_counter;
struct timeval sc_tv, ch_tv;
TAILQ_FOREACH(sc, &((struct carp_if *)
m_get_rcvif_NOMPSAFE(m)->if_carpdev->if_carp)->vhif_vrs, sc_list)
if (sc->sc_vhid == ch->carp_vhid)
break;
if (!sc || (sc->sc_if.if_flags & (IFF_UP|IFF_RUNNING)) !=
(IFF_UP|IFF_RUNNING)) {
2008-04-15 10:03:28 +04:00
CARP_STATINC(CARP_STAT_BADVHID);
m_freem(m);
return;
}
/*
* Check if our own advertisement was duplicated
* from a non simplex interface.
* XXX If there is no address on our physical interface
* there is no way to distinguish our ads from the ones
* another carp host might have sent us.
*/
if ((sc->sc_carpdev->if_flags & IFF_SIMPLEX) == 0) {
struct sockaddr sa;
struct ifaddr *ifa;
int s;
2009-03-18 19:00:08 +03:00
memset(&sa, 0, sizeof(sa));
sa.sa_family = af;
s = pserialize_read_enter();
ifa = ifaof_ifpforaddr(&sa, sc->sc_carpdev);
if (ifa && af == AF_INET) {
struct ip *ip = mtod(m, struct ip *);
if (ip->ip_src.s_addr ==
ifatoia(ifa)->ia_addr.sin_addr.s_addr) {
pserialize_read_exit(s);
m_freem(m);
return;
}
}
#ifdef INET6
if (ifa && af == AF_INET6) {
struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
struct in6_addr in6_src, in6_found;
in6_src = ip6->ip6_src;
in6_found = ifatoia6(ifa)->ia_addr.sin6_addr;
if (IN6_IS_ADDR_LINKLOCAL(&in6_src))
in6_src.s6_addr16[1] = 0;
if (IN6_IS_ADDR_LINKLOCAL(&in6_found))
in6_found.s6_addr16[1] = 0;
if (IN6_ARE_ADDR_EQUAL(&in6_src, &in6_found)) {
pserialize_read_exit(s);
m_freem(m);
return;
}
}
#endif /* INET6 */
pserialize_read_exit(s);
}
2009-01-11 05:45:45 +03:00
nanotime(&sc->sc_if.if_lastchange);
sc->sc_if.if_ipackets++;
sc->sc_if.if_ibytes += m->m_pkthdr.len;
/* verify the CARP version. */
if (ch->carp_version != CARP_VERSION) {
2008-04-15 10:03:28 +04:00
CARP_STATINC(CARP_STAT_BADVER);
sc->sc_if.if_ierrors++;
CARP_LOG(sc, ("invalid version %d != %d",
ch->carp_version, CARP_VERSION));
m_freem(m);
return;
}
/* verify the hash */
if (carp_hmac_verify(sc, ch->carp_counter, ch->carp_md)) {
struct ip *ip;
2017-04-19 06:23:06 +03:00
char ipbuf[INET_ADDRSTRLEN];
#ifdef INET6
struct ip6_hdr *ip6;
char ip6buf[INET6_ADDRSTRLEN];
2017-04-19 06:23:06 +03:00
#endif
2008-04-15 10:03:28 +04:00
CARP_STATINC(CARP_STAT_BADAUTH);
sc->sc_if.if_ierrors++;
switch(af) {
case AF_INET:
ip = mtod(m, struct ip *);
CARP_LOG(sc, ("incorrect hash from %s",
IN_PRINT(ipbuf, &ip->ip_src)));
break;
2017-04-19 06:23:06 +03:00
#ifdef INET6
case AF_INET6:
ip6 = mtod(m, struct ip6_hdr *);
CARP_LOG(sc, ("incorrect hash from %s",
IN6_PRINT(ip6buf, &ip6->ip6_src)));
break;
2017-04-19 06:23:06 +03:00
#endif
default: CARP_LOG(sc, ("incorrect hash"));
break;
}
m_freem(m);
return;
}
tmp_counter = ntohl(ch->carp_counter[0]);
tmp_counter = tmp_counter<<32;
tmp_counter += ntohl(ch->carp_counter[1]);
/* XXX Replay protection goes here */
sc->sc_init_counter = 0;
sc->sc_counter = tmp_counter;
sc_tv.tv_sec = sc->sc_advbase;
if (carp_suppress_preempt && sc->sc_advskew < 240)
sc_tv.tv_usec = 240 * 1000000 / 256;
else
sc_tv.tv_usec = sc->sc_advskew * 1000000 / 256;
ch_tv.tv_sec = ch->carp_advbase;
ch_tv.tv_usec = ch->carp_advskew * 1000000 / 256;
switch (sc->sc_state) {
case INIT:
break;
case MASTER:
/*
* If we receive an advertisement from a backup who's going to
* be more frequent than us, go into BACKUP state.
*/
if (timercmp(&sc_tv, &ch_tv, >) ||
timercmp(&sc_tv, &ch_tv, ==)) {
callout_stop(&sc->sc_ad_tmo);
CARP_LOG(sc, ("MASTER -> BACKUP (more frequent advertisement received)"));
carp_set_state(sc, BACKUP);
carp_setrun(sc, 0);
carp_setroute(sc, RTM_DELETE);
}
break;
case BACKUP:
/*
* If we're pre-empting masters who advertise slower than us,
* and this one claims to be slower, treat him as down.
*/
if (carp_opts[CARPCTL_PREEMPT] && timercmp(&sc_tv, &ch_tv, <)) {
CARP_LOG(sc, ("BACKUP -> MASTER (preempting a slower master)"));
carp_master_down(sc);
break;
}
/*
* If the master is going to advertise at such a low frequency
* that he's guaranteed to time out, we'd might as well just
* treat him as timed out now.
*/
sc_tv.tv_sec = sc->sc_advbase * 3;
if (timercmp(&sc_tv, &ch_tv, <)) {
CARP_LOG(sc, ("BACKUP -> MASTER (master timed out)"));
carp_master_down(sc);
break;
}
/*
* Otherwise, we reset the counter and wait for the next
* advertisement.
*/
carp_setrun(sc, af);
break;
}
m_freem(m);
return;
}
/*
* Interface side of the CARP implementation.
*/
/* ARGSUSED */
void
carpattach(int n)
{
if_clone_attach(&carp_cloner);
2008-04-15 10:03:28 +04:00
carpstat_percpu = percpu_alloc(sizeof(uint64_t) * CARP_NSTATS);
}
static int
carp_clone_create(struct if_clone *ifc, int unit)
{
extern int ifqmaxlen;
struct carp_softc *sc;
struct ifnet *ifp;
int rv;
sc = malloc(sizeof(*sc), M_DEVBUF, M_NOWAIT|M_ZERO);
if (!sc)
return (ENOMEM);
sc->sc_suppress = 0;
sc->sc_advbase = CARP_DFLTINTV;
sc->sc_vhid = -1; /* required setting */
sc->sc_advskew = 0;
sc->sc_init_counter = 1;
sc->sc_naddrs = sc->sc_naddrs6 = 0;
#ifdef INET6
sc->sc_im6o.im6o_multicast_hlim = CARP_DFLTTL;
#endif /* INET6 */
callout_init(&sc->sc_ad_tmo, 0);
callout_init(&sc->sc_md_tmo, 0);
callout_init(&sc->sc_md6_tmo, 0);
callout_setfunc(&sc->sc_ad_tmo, carp_send_ad, sc);
callout_setfunc(&sc->sc_md_tmo, carp_master_down, sc);
callout_setfunc(&sc->sc_md6_tmo, carp_master_down, sc);
LIST_INIT(&sc->carp_mc_listhead);
ifp = &sc->sc_if;
ifp->if_softc = sc;
snprintf(ifp->if_xname, sizeof ifp->if_xname, "%s%d", ifc->ifc_name,
unit);
ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
ifp->if_ioctl = carp_ioctl;
ifp->if_start = carp_start;
IFQ_SET_MAXLEN(&ifp->if_snd, ifqmaxlen);
IFQ_SET_READY(&ifp->if_snd);
rv = if_initialize(ifp);
if (rv != 0) {
callout_destroy(&sc->sc_ad_tmo);
callout_destroy(&sc->sc_md_tmo);
callout_destroy(&sc->sc_md6_tmo);
free(ifp->if_softc, M_DEVBUF);
return rv;
}
ether_ifattach(ifp, NULL);
carp_set_enaddr(sc);
/* Overwrite ethernet defaults */
ifp->if_type = IFT_CARP;
ifp->if_output = carp_output;
if_register(ifp);
return (0);
}
static int
carp_clone_destroy(struct ifnet *ifp)
{
struct carp_softc *sc = ifp->if_softc;
carpdetach(ifp->if_softc);
ether_ifdetach(ifp);
if_detach(ifp);
callout_destroy(&sc->sc_ad_tmo);
callout_destroy(&sc->sc_md_tmo);
callout_destroy(&sc->sc_md6_tmo);
free(ifp->if_softc, M_DEVBUF);
return (0);
}
static void
carpdetach(struct carp_softc *sc)
{
struct carp_if *cif;
int s;
callout_stop(&sc->sc_ad_tmo);
callout_stop(&sc->sc_md_tmo);
callout_stop(&sc->sc_md6_tmo);
if (sc->sc_suppress)
carp_suppress_preempt--;
sc->sc_suppress = 0;
if (sc->sc_sendad_errors >= CARP_SENDAD_MAX_ERRORS)
carp_suppress_preempt--;
sc->sc_sendad_errors = 0;
carp_set_state(sc, INIT);
sc->sc_if.if_flags &= ~IFF_UP;
carp_setrun(sc, 0);
carp_multicast_cleanup(sc);
KERNEL_LOCK(1, NULL);
s = splnet();
if (sc->sc_carpdev != NULL) {
/* XXX linkstatehook removal */
cif = (struct carp_if *)sc->sc_carpdev->if_carp;
TAILQ_REMOVE(&cif->vhif_vrs, sc, sc_list);
if (!--cif->vhif_nvrs) {
ifpromisc(sc->sc_carpdev, 0);
sc->sc_carpdev->if_carp = NULL;
2008-12-17 23:51:31 +03:00
free(cif, M_IFADDR);
}
}
sc->sc_carpdev = NULL;
splx(s);
KERNEL_UNLOCK_ONE(NULL);
}
/* Detach an interface from the carp. */
void
carp_ifdetach(struct ifnet *ifp)
{
struct carp_softc *sc, *nextsc;
struct carp_if *cif = (struct carp_if *)ifp->if_carp;
for (sc = TAILQ_FIRST(&cif->vhif_vrs); sc; sc = nextsc) {
nextsc = TAILQ_NEXT(sc, sc_list);
carpdetach(sc);
}
}
static void
carp_prepare_ad(struct mbuf *m, struct carp_softc *sc,
struct carp_header *ch)
{
if (sc->sc_init_counter) {
/* this could also be seconds since unix epoch */
First step of random number subsystem rework described in <20111022023242.BA26F14A158@mail.netbsd.org>. This change includes the following: An initial cleanup and minor reorganization of the entropy pool code in sys/dev/rnd.c and sys/dev/rndpool.c. Several bugs are fixed. Some effort is made to accumulate entropy more quickly at boot time. A generic interface, "rndsink", is added, for stream generators to request that they be re-keyed with good quality entropy from the pool as soon as it is available. The arc4random()/arc4randbytes() implementation in libkern is adjusted to use the rndsink interface for rekeying, which helps address the problem of low-quality keys at boot time. An implementation of the FIPS 140-2 statistical tests for random number generator quality is provided (libkern/rngtest.c). This is based on Greg Rose's implementation from Qualcomm. A new random stream generator, nist_ctr_drbg, is provided. It is based on an implementation of the NIST SP800-90 CTR_DRBG by Henric Jungheim. This generator users AES in a modified counter mode to generate a backtracking-resistant random stream. An abstraction layer, "cprng", is provided for in-kernel consumers of randomness. The arc4random/arc4randbytes API is deprecated for in-kernel use. It is replaced by "cprng_strong". The current cprng_fast implementation wraps the existing arc4random implementation. The current cprng_strong implementation wraps the new CTR_DRBG implementation. Both interfaces are rekeyed from the entropy pool automatically at intervals justifiable from best current cryptographic practice. In some quick tests, cprng_fast() is about the same speed as the old arc4randbytes(), and cprng_strong() is about 20% faster than rnd_extract_data(). Performance is expected to improve. The AES code in src/crypto/rijndael is no longer an optional kernel component, as it is required by cprng_strong, which is not an optional kernel component. The entropy pool output is subjected to the rngtest tests at startup time; if it fails, the system will reboot. There is approximately a 3/10000 chance of a false positive from these tests. Entropy pool _input_ from hardware random numbers is subjected to the rngtest tests at attach time, as well as the FIPS continuous-output test, to detect bad or stuck hardware RNGs; if any are detected, they are detached, but the system continues to run. A problem with rndctl(8) is fixed -- datastructures with pointers in arrays are no longer passed to userspace (this was not a security problem, but rather a major issue for compat32). A new kernel will require a new rndctl. The sysctl kern.arandom() and kern.urandom() nodes are hooked up to the new generators, but the /dev/*random pseudodevices are not, yet. Manual pages for the new kernel interfaces are forthcoming.
2011-11-20 02:51:18 +04:00
sc->sc_counter = cprng_fast64();
} else
sc->sc_counter++;
ch->carp_counter[0] = htonl((sc->sc_counter>>32)&0xffffffff);
ch->carp_counter[1] = htonl(sc->sc_counter&0xffffffff);
carp_hmac_generate(sc, ch->carp_counter, ch->carp_md);
}
static void
carp_send_ad_all(void)
{
struct ifnet *ifp;
struct carp_if *cif;
struct carp_softc *vh;
int s;
int bound = curlwp_bind();
s = pserialize_read_enter();
IFNET_READER_FOREACH(ifp) {
struct psref psref;
if (ifp->if_carp == NULL || ifp->if_type == IFT_CARP)
continue;
if_acquire(ifp, &psref);
pserialize_read_exit(s);
cif = (struct carp_if *)ifp->if_carp;
TAILQ_FOREACH(vh, &cif->vhif_vrs, sc_list) {
if ((vh->sc_if.if_flags & (IFF_UP|IFF_RUNNING)) ==
(IFF_UP|IFF_RUNNING) && vh->sc_state == MASTER)
carp_send_ad(vh);
}
s = pserialize_read_enter();
if_release(ifp, &psref);
}
pserialize_read_exit(s);
curlwp_bindx(bound);
}
static void
carp_send_ad(void *v)
{
struct carp_header ch;
struct timeval tv;
struct carp_softc *sc = v;
struct carp_header *ch_ptr;
struct mbuf *m;
int error, len, advbase, advskew, s;
struct sockaddr sa;
KERNEL_LOCK(1, NULL);
s = splsoftnet();
advbase = advskew = 0; /* Sssssh compiler */
if (sc->sc_carpdev == NULL) {
sc->sc_if.if_oerrors++;
goto retry_later;
}
/* bow out if we've gone to backup (the carp interface is going down) */
if (sc->sc_bow_out) {
sc->sc_bow_out = 0;
advbase = 255;
advskew = 255;
} else {
advbase = sc->sc_advbase;
if (!carp_suppress_preempt || sc->sc_advskew > 240)
advskew = sc->sc_advskew;
else
advskew = 240;
tv.tv_sec = advbase;
tv.tv_usec = advskew * 1000000 / 256;
}
ch.carp_version = CARP_VERSION;
ch.carp_type = CARP_ADVERTISEMENT;
ch.carp_vhid = sc->sc_vhid;
ch.carp_advbase = advbase;
ch.carp_advskew = advskew;
ch.carp_authlen = 7; /* XXX DEFINE */
ch.carp_pad1 = 0; /* must be zero */
ch.carp_cksum = 0;
#ifdef INET
if (sc->sc_naddrs) {
struct ip *ip;
struct ifaddr *ifa;
int _s;
MGETHDR(m, M_DONTWAIT, MT_HEADER);
if (m == NULL) {
sc->sc_if.if_oerrors++;
2008-04-15 10:03:28 +04:00
CARP_STATINC(CARP_STAT_ONOMEM);
/* XXX maybe less ? */
goto retry_later;
}
MCLAIM(m, &carp_proto_mowner_tx);
len = sizeof(*ip) + sizeof(ch);
m->m_pkthdr.len = len;
m_reset_rcvif(m);
m->m_len = len;
MH_ALIGN(m, m->m_len);
m->m_flags |= M_MCAST;
ip = mtod(m, struct ip *);
ip->ip_v = IPVERSION;
ip->ip_hl = sizeof(*ip) >> 2;
ip->ip_tos = IPTOS_LOWDELAY;
ip->ip_len = htons(len);
ip->ip_id = 0; /* no need for id, we don't support fragments */
ip->ip_off = htons(IP_DF);
ip->ip_ttl = CARP_DFLTTL;
ip->ip_p = IPPROTO_CARP;
ip->ip_sum = 0;
2009-03-18 19:00:08 +03:00
memset(&sa, 0, sizeof(sa));
sa.sa_family = AF_INET;
_s = pserialize_read_enter();
ifa = ifaof_ifpforaddr(&sa, sc->sc_carpdev);
if (ifa == NULL)
ip->ip_src.s_addr = 0;
else
ip->ip_src.s_addr =
ifatoia(ifa)->ia_addr.sin_addr.s_addr;
pserialize_read_exit(_s);
ip->ip_dst.s_addr = INADDR_CARP_GROUP;
ch_ptr = (struct carp_header *)(&ip[1]);
memcpy(ch_ptr, &ch, sizeof(ch));
carp_prepare_ad(m, sc, ch_ptr);
m->m_data += sizeof(*ip);
ch_ptr->carp_cksum = carp_cksum(m, len - sizeof(*ip));
m->m_data -= sizeof(*ip);
2009-01-11 05:45:45 +03:00
nanotime(&sc->sc_if.if_lastchange);
sc->sc_if.if_opackets++;
sc->sc_if.if_obytes += len;
2008-04-15 10:03:28 +04:00
CARP_STATINC(CARP_STAT_OPACKETS);
error = ip_output(m, NULL, NULL, IP_RAWOUTPUT, &sc->sc_imo,
NULL);
if (error) {
if (error == ENOBUFS)
2008-04-15 10:03:28 +04:00
CARP_STATINC(CARP_STAT_ONOMEM);
else
CARP_LOG(sc, ("ip_output failed: %d", error));
sc->sc_if.if_oerrors++;
if (sc->sc_sendad_errors < INT_MAX)
sc->sc_sendad_errors++;
if (sc->sc_sendad_errors == CARP_SENDAD_MAX_ERRORS) {
carp_suppress_preempt++;
if (carp_suppress_preempt == 1)
carp_send_ad_all();
}
sc->sc_sendad_success = 0;
} else {
if (sc->sc_sendad_errors >= CARP_SENDAD_MAX_ERRORS) {
if (++sc->sc_sendad_success >=
CARP_SENDAD_MIN_SUCCESS) {
carp_suppress_preempt--;
sc->sc_sendad_errors = 0;
}
} else
sc->sc_sendad_errors = 0;
}
}
#endif /* INET */
#ifdef INET6
if (sc->sc_naddrs6) {
struct ip6_hdr *ip6;
struct ifaddr *ifa;
int _s;
MGETHDR(m, M_DONTWAIT, MT_HEADER);
if (m == NULL) {
sc->sc_if.if_oerrors++;
2008-04-15 10:03:28 +04:00
CARP_STATINC(CARP_STAT_ONOMEM);
/* XXX maybe less ? */
goto retry_later;
}
MCLAIM(m, &carp_proto6_mowner_tx);
len = sizeof(*ip6) + sizeof(ch);
m->m_pkthdr.len = len;
m_reset_rcvif(m);
m->m_len = len;
MH_ALIGN(m, m->m_len);
m->m_flags |= M_MCAST;
ip6 = mtod(m, struct ip6_hdr *);
2009-03-18 19:00:08 +03:00
memset(ip6, 0, sizeof(*ip6));
ip6->ip6_vfc |= IPV6_VERSION;
ip6->ip6_hlim = CARP_DFLTTL;
ip6->ip6_nxt = IPPROTO_CARP;
/* set the source address */
2009-03-18 19:00:08 +03:00
memset(&sa, 0, sizeof(sa));
sa.sa_family = AF_INET6;
_s = pserialize_read_enter();
ifa = ifaof_ifpforaddr(&sa, sc->sc_carpdev);
if (ifa == NULL) /* This should never happen with IPv6 */
2009-03-18 19:00:08 +03:00
memset(&ip6->ip6_src, 0, sizeof(struct in6_addr));
else
bcopy(ifatoia6(ifa)->ia_addr.sin6_addr.s6_addr,
&ip6->ip6_src, sizeof(struct in6_addr));
pserialize_read_exit(_s);
/* set the multicast destination */
ip6->ip6_dst.s6_addr16[0] = htons(0xff02);
ip6->ip6_dst.s6_addr8[15] = 0x12;
if (in6_setscope(&ip6->ip6_dst, &sc->sc_if, NULL) != 0) {
sc->sc_if.if_oerrors++;
m_freem(m);
CARP_LOG(sc, ("in6_setscope failed"));
goto retry_later;
}
ch_ptr = (struct carp_header *)(&ip6[1]);
memcpy(ch_ptr, &ch, sizeof(ch));
carp_prepare_ad(m, sc, ch_ptr);
ch_ptr->carp_cksum = carp6_cksum(m, sizeof(*ip6),
len - sizeof(*ip6));
2009-01-11 05:45:45 +03:00
nanotime(&sc->sc_if.if_lastchange);
sc->sc_if.if_opackets++;
sc->sc_if.if_obytes += len;
2008-04-15 10:03:28 +04:00
CARP_STATINC(CARP_STAT_OPACKETS6);
error = ip6_output(m, NULL, NULL, 0, &sc->sc_im6o, NULL, NULL);
if (error) {
if (error == ENOBUFS)
2008-04-15 10:03:28 +04:00
CARP_STATINC(CARP_STAT_ONOMEM);
else
CARP_LOG(sc, ("ip6_output failed: %d", error));
sc->sc_if.if_oerrors++;
if (sc->sc_sendad_errors < INT_MAX)
sc->sc_sendad_errors++;
if (sc->sc_sendad_errors == CARP_SENDAD_MAX_ERRORS) {
carp_suppress_preempt++;
if (carp_suppress_preempt == 1)
carp_send_ad_all();
}
sc->sc_sendad_success = 0;
} else {
if (sc->sc_sendad_errors >= CARP_SENDAD_MAX_ERRORS) {
if (++sc->sc_sendad_success >=
CARP_SENDAD_MIN_SUCCESS) {
carp_suppress_preempt--;
sc->sc_sendad_errors = 0;
}
} else
sc->sc_sendad_errors = 0;
}
}
#endif /* INET6 */
retry_later:
splx(s);
KERNEL_UNLOCK_ONE(NULL);
if (advbase != 255 || advskew != 255)
callout_schedule(&sc->sc_ad_tmo, tvtohz(&tv));
}
/*
* Broadcast a gratuitous ARP request containing
* the virtual router MAC address for each IP address
* associated with the virtual router.
*/
static void
carp_send_arp(struct carp_softc *sc)
{
struct ifaddr *ifa;
int s, bound;
KERNEL_LOCK(1, NULL);
bound = curlwp_bind();
s = pserialize_read_enter();
IFADDR_READER_FOREACH(ifa, &sc->sc_if) {
struct psref psref;
if (ifa->ifa_addr->sa_family != AF_INET)
continue;
ifa_acquire(ifa, &psref);
pserialize_read_exit(s);
arpannounce(sc->sc_carpdev, ifa, CLLADDR(sc->sc_if.if_sadl));
s = pserialize_read_enter();
ifa_release(ifa, &psref);
}
pserialize_read_exit(s);
curlwp_bindx(bound);
KERNEL_UNLOCK_ONE(NULL);
}
#ifdef INET6
static void
carp_send_na(struct carp_softc *sc)
{
struct ifaddr *ifa;
struct in6_addr *in6;
static struct in6_addr mcast = IN6ADDR_LINKLOCAL_ALLNODES_INIT;
int s, bound;
KERNEL_LOCK(1, NULL);
bound = curlwp_bind();
s = pserialize_read_enter();
IFADDR_READER_FOREACH(ifa, &sc->sc_if) {
struct psref psref;
if (ifa->ifa_addr->sa_family != AF_INET6)
continue;
ifa_acquire(ifa, &psref);
pserialize_read_exit(s);
in6 = &ifatoia6(ifa)->ia_addr.sin6_addr;
nd6_na_output(sc->sc_carpdev, &mcast, in6,
ND_NA_FLAG_OVERRIDE, 1, NULL);
s = pserialize_read_enter();
ifa_release(ifa, &psref);
}
pserialize_read_exit(s);
curlwp_bindx(bound);
KERNEL_UNLOCK_ONE(NULL);
}
#endif /* INET6 */
/*
* Based on bridge_hash() in if_bridge.c
*/
#define mix(a,b,c) \
do { \
a -= b; a -= c; a ^= (c >> 13); \
b -= c; b -= a; b ^= (a << 8); \
c -= a; c -= b; c ^= (b >> 13); \
a -= b; a -= c; a ^= (c >> 12); \
b -= c; b -= a; b ^= (a << 16); \
c -= a; c -= b; c ^= (b >> 5); \
a -= b; a -= c; a ^= (c >> 3); \
b -= c; b -= a; b ^= (a << 10); \
c -= a; c -= b; c ^= (b >> 15); \
} while (0)
static u_int32_t
carp_hash(struct carp_softc *sc, u_char *src)
{
u_int32_t a = 0x9e3779b9, b = sc->sc_hashkey[0], c = sc->sc_hashkey[1];
c += sc->sc_key[3] << 24;
c += sc->sc_key[2] << 16;
c += sc->sc_key[1] << 8;
c += sc->sc_key[0];
b += src[5] << 8;
b += src[4];
a += src[3] << 24;
a += src[2] << 16;
a += src[1] << 8;
a += src[0];
mix(a, b, c);
return (c);
}
static int
carp_addrcount(struct carp_if *cif, struct in_ifaddr *ia, int type)
{
struct carp_softc *vh;
struct ifaddr *ifa;
int count = 0;
TAILQ_FOREACH(vh, &cif->vhif_vrs, sc_list) {
if ((type == CARP_COUNT_RUNNING &&
(vh->sc_if.if_flags & (IFF_UP|IFF_RUNNING)) ==
(IFF_UP|IFF_RUNNING)) ||
(type == CARP_COUNT_MASTER && vh->sc_state == MASTER)) {
int s = pserialize_read_enter();
IFADDR_READER_FOREACH(ifa, &vh->sc_if) {
if (ifa->ifa_addr->sa_family == AF_INET &&
ia->ia_addr.sin_addr.s_addr ==
ifatoia(ifa)->ia_addr.sin_addr.s_addr)
count++;
}
pserialize_read_exit(s);
}
}
return (count);
}
int
carp_iamatch(struct in_ifaddr *ia, u_char *src,
u_int32_t *count, u_int32_t index)
{
struct carp_softc *sc = ia->ia_ifp->if_softc;
if (carp_opts[CARPCTL_ARPBALANCE]) {
/*
* We use the source ip to decide which virtual host should
* handle the request. If we're master of that virtual host,
* then we respond, otherwise, just drop the arp packet on
* the floor.
*/
/* Count the elegible carp interfaces with this address */
if (*count == 0)
*count = carp_addrcount(
(struct carp_if *)ia->ia_ifp->if_carpdev->if_carp,
ia, CARP_COUNT_RUNNING);
/* This should never happen, but... */
if (*count == 0)
return (0);
if (carp_hash(sc, src) % *count == index - 1 &&
sc->sc_state == MASTER) {
return (1);
}
} else {
if (sc->sc_state == MASTER)
return (1);
}
return (0);
}
#ifdef INET6
struct ifaddr *
carp_iamatch6(void *v, struct in6_addr *taddr)
{
struct carp_if *cif = v;
struct carp_softc *vh;
struct ifaddr *ifa;
TAILQ_FOREACH(vh, &cif->vhif_vrs, sc_list) {
int s = pserialize_read_enter();
IFADDR_READER_FOREACH(ifa, &vh->sc_if) {
if (IN6_ARE_ADDR_EQUAL(taddr,
&ifatoia6(ifa)->ia_addr.sin6_addr) &&
((vh->sc_if.if_flags & (IFF_UP|IFF_RUNNING)) ==
(IFF_UP|IFF_RUNNING)) && vh->sc_state == MASTER)
return (ifa);
}
pserialize_read_exit(s);
}
return (NULL);
}
#endif /* INET6 */
struct ifnet *
carp_ourether(void *v, struct ether_header *eh, u_char iftype, int src)
{
struct carp_if *cif = (struct carp_if *)v;
struct carp_softc *vh;
u_int8_t *ena;
if (src)
ena = (u_int8_t *)&eh->ether_shost;
else
ena = (u_int8_t *)&eh->ether_dhost;
switch (iftype) {
case IFT_ETHER:
case IFT_FDDI:
if (ena[0] || ena[1] || ena[2] != 0x5e || ena[3] || ena[4] != 1)
return (NULL);
break;
case IFT_ISO88025:
if (ena[0] != 3 || ena[1] || ena[4] || ena[5])
return (NULL);
break;
default:
return (NULL);
break;
}
TAILQ_FOREACH(vh, &cif->vhif_vrs, sc_list)
if ((vh->sc_if.if_flags & (IFF_UP|IFF_RUNNING)) ==
(IFF_UP|IFF_RUNNING) && vh->sc_state == MASTER &&
2009-03-18 18:14:29 +03:00
!memcmp(ena, CLLADDR(vh->sc_if.if_sadl),
ETHER_ADDR_LEN)) {
return (&vh->sc_if);
}
return (NULL);
}
int
carp_input(struct mbuf *m, u_int8_t *shost, u_int8_t *dhost, u_int16_t etype)
{
struct ether_header eh;
struct carp_if *cif = (struct carp_if *)m_get_rcvif_NOMPSAFE(m)->if_carp;
struct ifnet *ifp;
memcpy(&eh.ether_shost, shost, sizeof(eh.ether_shost));
memcpy(&eh.ether_dhost, dhost, sizeof(eh.ether_dhost));
eh.ether_type = etype;
if (m->m_flags & (M_BCAST|M_MCAST)) {
struct carp_softc *vh;
struct mbuf *m0;
/*
* XXX Should really check the list of multicast addresses
* for each CARP interface _before_ copying.
*/
TAILQ_FOREACH(vh, &cif->vhif_vrs, sc_list) {
m0 = m_copym(m, 0, M_COPYALL, M_DONTWAIT);
if (m0 == NULL)
continue;
m_set_rcvif(m0, &vh->sc_if);
ether_input(&vh->sc_if, m0);
}
return (1);
}
ifp = carp_ourether(cif, &eh, m_get_rcvif_NOMPSAFE(m)->if_type, 0);
if (ifp == NULL) {
return (1);
}
m_set_rcvif(m, ifp);
bpf_mtap(ifp, m, BPF_D_IN);
ifp->if_ipackets++;
ether_input(ifp, m);
return (0);
}
static void
carp_master_down(void *v)
{
struct carp_softc *sc = v;
switch (sc->sc_state) {
case INIT:
printf("%s: master_down event in INIT state\n",
sc->sc_if.if_xname);
break;
case MASTER:
break;
case BACKUP:
CARP_LOG(sc, ("INIT -> MASTER (preempting)"));
carp_set_state(sc, MASTER);
carp_send_ad(sc);
carp_send_arp(sc);
#ifdef INET6
carp_send_na(sc);
#endif /* INET6 */
carp_setrun(sc, 0);
carp_setroute(sc, RTM_ADD);
break;
}
}
/*
* When in backup state, af indicates whether to reset the master down timer
* for v4 or v6. If it's set to zero, reset the ones which are already pending.
*/
static void
carp_setrun(struct carp_softc *sc, sa_family_t af)
{
struct timeval tv;
if (sc->sc_carpdev == NULL) {
sc->sc_if.if_flags &= ~IFF_RUNNING;
carp_set_state(sc, INIT);
return;
}
if (sc->sc_if.if_flags & IFF_UP && sc->sc_vhid > 0 &&
(sc->sc_naddrs || sc->sc_naddrs6) && !sc->sc_suppress) {
sc->sc_if.if_flags |= IFF_RUNNING;
} else {
sc->sc_if.if_flags &= ~IFF_RUNNING;
carp_setroute(sc, RTM_DELETE);
return;
}
switch (sc->sc_state) {
case INIT:
carp_set_state(sc, BACKUP);
carp_setroute(sc, RTM_DELETE);
carp_setrun(sc, 0);
break;
case BACKUP:
callout_stop(&sc->sc_ad_tmo);
tv.tv_sec = 3 * sc->sc_advbase;
tv.tv_usec = sc->sc_advskew * 1000000 / 256;
switch (af) {
#ifdef INET
case AF_INET:
callout_schedule(&sc->sc_md_tmo, tvtohz(&tv));
break;
#endif /* INET */
#ifdef INET6
case AF_INET6:
callout_schedule(&sc->sc_md6_tmo, tvtohz(&tv));
break;
#endif /* INET6 */
default:
if (sc->sc_naddrs)
callout_schedule(&sc->sc_md_tmo, tvtohz(&tv));
#ifdef INET6
if (sc->sc_naddrs6)
callout_schedule(&sc->sc_md6_tmo, tvtohz(&tv));
#endif /* INET6 */
break;
}
break;
case MASTER:
tv.tv_sec = sc->sc_advbase;
tv.tv_usec = sc->sc_advskew * 1000000 / 256;
callout_schedule(&sc->sc_ad_tmo, tvtohz(&tv));
break;
}
}
static void
carp_multicast_cleanup(struct carp_softc *sc)
{
struct ip_moptions *imo = &sc->sc_imo;
#ifdef INET6
struct ip6_moptions *im6o = &sc->sc_im6o;
#endif
u_int16_t n = imo->imo_num_memberships;
/* Clean up our own multicast memberships */
while (n-- > 0) {
if (imo->imo_membership[n] != NULL) {
in_delmulti(imo->imo_membership[n]);
imo->imo_membership[n] = NULL;
}
}
imo->imo_num_memberships = 0;
imo->imo_multicast_if_index = 0;
#ifdef INET6
while (!LIST_EMPTY(&im6o->im6o_memberships)) {
struct in6_multi_mship *imm =
LIST_FIRST(&im6o->im6o_memberships);
LIST_REMOVE(imm, i6mm_chain);
in6_leavegroup(imm);
}
im6o->im6o_multicast_if_index = 0;
#endif
/* And any other multicast memberships */
carp_ether_purgemulti(sc);
}
static int
carp_set_ifp(struct carp_softc *sc, struct ifnet *ifp)
{
struct carp_if *cif, *ncif = NULL;
struct carp_softc *vr, *after = NULL;
int myself = 0, error = 0;
int s;
if (ifp == sc->sc_carpdev)
return (0);
if (ifp != NULL) {
if ((ifp->if_flags & IFF_MULTICAST) == 0)
return (EADDRNOTAVAIL);
if (ifp->if_type == IFT_CARP)
return (EINVAL);
if (ifp->if_carp == NULL) {
2008-12-17 23:51:31 +03:00
ncif = malloc(sizeof(*cif), M_IFADDR, M_NOWAIT);
if (ncif == NULL)
return (ENOBUFS);
if ((error = ifpromisc(ifp, 1))) {
2008-12-17 23:51:31 +03:00
free(ncif, M_IFADDR);
return (error);
}
ncif->vhif_ifp = ifp;
TAILQ_INIT(&ncif->vhif_vrs);
} else {
cif = (struct carp_if *)ifp->if_carp;
TAILQ_FOREACH(vr, &cif->vhif_vrs, sc_list)
if (vr != sc && vr->sc_vhid == sc->sc_vhid)
return (EINVAL);
}
/* detach from old interface */
if (sc->sc_carpdev != NULL)
carpdetach(sc);
/* join multicast groups */
if (sc->sc_naddrs < 0 &&
(error = carp_join_multicast(sc)) != 0) {
if (ncif != NULL)
2008-12-17 23:51:31 +03:00
free(ncif, M_IFADDR);
return (error);
}
#ifdef INET6
if (sc->sc_naddrs6 < 0 &&
(error = carp_join_multicast6(sc)) != 0) {
if (ncif != NULL)
2008-12-17 23:51:31 +03:00
free(ncif, M_IFADDR);
carp_multicast_cleanup(sc);
return (error);
}
#endif
/* attach carp interface to physical interface */
if (ncif != NULL)
ifp->if_carp = (void *)ncif;
sc->sc_carpdev = ifp;
sc->sc_if.if_capabilities = ifp->if_capabilities &
(IFCAP_TSOv4 | IFCAP_TSOv6 |
IFCAP_CSUM_IPv4_Tx|IFCAP_CSUM_IPv4_Rx|
IFCAP_CSUM_TCPv4_Tx|IFCAP_CSUM_TCPv4_Rx|
IFCAP_CSUM_UDPv4_Tx|IFCAP_CSUM_UDPv4_Rx|
IFCAP_CSUM_TCPv6_Tx|IFCAP_CSUM_TCPv6_Rx|
IFCAP_CSUM_UDPv6_Tx|IFCAP_CSUM_UDPv6_Rx);
cif = (struct carp_if *)ifp->if_carp;
TAILQ_FOREACH(vr, &cif->vhif_vrs, sc_list) {
if (vr == sc)
myself = 1;
if (vr->sc_vhid < sc->sc_vhid)
after = vr;
}
if (!myself) {
/* We're trying to keep things in order */
if (after == NULL) {
TAILQ_INSERT_TAIL(&cif->vhif_vrs, sc, sc_list);
} else {
TAILQ_INSERT_AFTER(&cif->vhif_vrs, after,
sc, sc_list);
}
cif->vhif_nvrs++;
}
if (sc->sc_naddrs || sc->sc_naddrs6)
sc->sc_if.if_flags |= IFF_UP;
carp_set_enaddr(sc);
KERNEL_LOCK(1, NULL);
s = splnet();
/* XXX linkstatehooks establish */
carp_carpdev_state(ifp);
splx(s);
KERNEL_UNLOCK_ONE(NULL);
} else {
carpdetach(sc);
sc->sc_if.if_flags &= ~(IFF_UP|IFF_RUNNING);
}
return (0);
}
static void
carp_set_enaddr(struct carp_softc *sc)
{
uint8_t enaddr[ETHER_ADDR_LEN];
if (sc->sc_carpdev && sc->sc_carpdev->if_type == IFT_ISO88025) {
enaddr[0] = 3;
enaddr[1] = 0;
enaddr[2] = 0x40 >> (sc->sc_vhid - 1);
enaddr[3] = 0x40000 >> (sc->sc_vhid - 1);
enaddr[4] = 0;
enaddr[5] = 0;
} else {
enaddr[0] = 0;
enaddr[1] = 0;
enaddr[2] = 0x5e;
enaddr[3] = 0;
enaddr[4] = 1;
enaddr[5] = sc->sc_vhid;
}
*** 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_set_sadl(&sc->sc_if, enaddr, sizeof(enaddr), false);
}
#if 0
static void
carp_addr_updated(void *v)
{
struct carp_softc *sc = (struct carp_softc *) v;
struct ifaddr *ifa;
int new_naddrs = 0, new_naddrs6 = 0;
IFADDR_READER_FOREACH(ifa, &sc->sc_if) {
if (ifa->ifa_addr->sa_family == AF_INET)
new_naddrs++;
else if (ifa->ifa_addr->sa_family == AF_INET6)
new_naddrs6++;
}
/* Handle a callback after SIOCDIFADDR */
if (new_naddrs < sc->sc_naddrs || new_naddrs6 < sc->sc_naddrs6) {
struct in_addr mc_addr;
sc->sc_naddrs = new_naddrs;
sc->sc_naddrs6 = new_naddrs6;
/* Re-establish multicast membership removed by in_control */
mc_addr.s_addr = INADDR_CARP_GROUP;
if (!in_multi_group(mc_addr, &sc->sc_if, 0)) {
2009-03-18 19:00:08 +03:00
memset(&sc->sc_imo, 0, sizeof(sc->sc_imo));
if (sc->sc_carpdev != NULL && sc->sc_naddrs > 0)
carp_join_multicast(sc);
}
if (sc->sc_naddrs == 0 && sc->sc_naddrs6 == 0) {
sc->sc_if.if_flags &= ~IFF_UP;
carp_set_state(sc, INIT);
} else
carp_hmac_prepare(sc);
}
carp_setrun(sc, 0);
}
#endif
static int
carp_set_addr(struct carp_softc *sc, struct sockaddr_in *sin)
{
struct ifnet *ifp = sc->sc_carpdev;
struct in_ifaddr *ia, *ia_if;
int error = 0;
int s;
if (sin->sin_addr.s_addr == 0) {
if (!(sc->sc_if.if_flags & IFF_UP))
carp_set_state(sc, INIT);
if (sc->sc_naddrs)
sc->sc_if.if_flags |= IFF_UP;
carp_setrun(sc, 0);
return (0);
}
/* we have to do this by hand to ensure we don't match on ourselves */
ia_if = NULL;
s = pserialize_read_enter();
IN_ADDRLIST_READER_FOREACH(ia) {
/* and, yeah, we need a multicast-capable iface too */
if (ia->ia_ifp != &sc->sc_if &&
ia->ia_ifp->if_type != IFT_CARP &&
(ia->ia_ifp->if_flags & IFF_MULTICAST) &&
(sin->sin_addr.s_addr & ia->ia_subnetmask) ==
ia->ia_subnet) {
if (!ia_if)
ia_if = ia;
}
}
if (ia_if) {
ia = ia_if;
if (ifp) {
if (ifp != ia->ia_ifp)
return (EADDRNOTAVAIL);
} else {
/* FIXME NOMPSAFE */
ifp = ia->ia_ifp;
}
}
pserialize_read_exit(s);
if ((error = carp_set_ifp(sc, ifp)))
return (error);
if (sc->sc_carpdev == NULL)
return (EADDRNOTAVAIL);
if (sc->sc_naddrs == 0 && (error = carp_join_multicast(sc)) != 0)
return (error);
sc->sc_naddrs++;
if (sc->sc_carpdev != NULL)
sc->sc_if.if_flags |= IFF_UP;
carp_set_state(sc, INIT);
carp_setrun(sc, 0);
/*
* Hook if_addrhooks so that we get a callback after in_ifinit has run,
* to correct any inappropriate routes that it inserted.
*/
if (sc->ah_cookie == 0) {
/* XXX link address hook */
}
return (0);
}
static int
carp_join_multicast(struct carp_softc *sc)
{
struct ip_moptions *imo = &sc->sc_imo, tmpimo;
struct in_addr addr;
2009-03-18 19:00:08 +03:00
memset(&tmpimo, 0, sizeof(tmpimo));
addr.s_addr = INADDR_CARP_GROUP;
if ((tmpimo.imo_membership[0] =
in_addmulti(&addr, &sc->sc_if)) == NULL) {
return (ENOBUFS);
}
imo->imo_membership[0] = tmpimo.imo_membership[0];
imo->imo_num_memberships = 1;
imo->imo_multicast_if_index = sc->sc_if.if_index;
imo->imo_multicast_ttl = CARP_DFLTTL;
imo->imo_multicast_loop = 0;
return (0);
}
#ifdef INET6
static int
carp_set_addr6(struct carp_softc *sc, struct sockaddr_in6 *sin6)
{
struct ifnet *ifp = sc->sc_carpdev;
struct in6_ifaddr *ia, *ia_if;
int error = 0;
int s;
if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
if (!(sc->sc_if.if_flags & IFF_UP))
carp_set_state(sc, INIT);
if (sc->sc_naddrs6)
sc->sc_if.if_flags |= IFF_UP;
carp_setrun(sc, 0);
return (0);
}
/* we have to do this by hand to ensure we don't match on ourselves */
ia_if = NULL;
s = pserialize_read_enter();
IN6_ADDRLIST_READER_FOREACH(ia) {
int i;
for (i = 0; i < 4; i++) {
if ((sin6->sin6_addr.s6_addr32[i] &
ia->ia_prefixmask.sin6_addr.s6_addr32[i]) !=
(ia->ia_addr.sin6_addr.s6_addr32[i] &
ia->ia_prefixmask.sin6_addr.s6_addr32[i]))
break;
}
/* and, yeah, we need a multicast-capable iface too */
if (ia->ia_ifp != &sc->sc_if &&
ia->ia_ifp->if_type != IFT_CARP &&
(ia->ia_ifp->if_flags & IFF_MULTICAST) &&
(i == 4)) {
if (!ia_if)
ia_if = ia;
}
}
pserialize_read_exit(s);
if (ia_if) {
ia = ia_if;
if (sc->sc_carpdev) {
if (sc->sc_carpdev != ia->ia_ifp)
return (EADDRNOTAVAIL);
} else {
ifp = ia->ia_ifp;
}
}
if ((error = carp_set_ifp(sc, ifp)))
return (error);
if (sc->sc_carpdev == NULL)
return (EADDRNOTAVAIL);
if (sc->sc_naddrs6 == 0 && (error = carp_join_multicast6(sc)) != 0)
return (error);
sc->sc_naddrs6++;
if (sc->sc_carpdev != NULL)
sc->sc_if.if_flags |= IFF_UP;
carp_set_state(sc, INIT);
carp_setrun(sc, 0);
return (0);
}
static int
carp_join_multicast6(struct carp_softc *sc)
{
struct in6_multi_mship *imm, *imm2;
struct ip6_moptions *im6o = &sc->sc_im6o;
struct sockaddr_in6 addr6;
int error;
/* Join IPv6 CARP multicast group */
2009-03-18 19:00:08 +03:00
memset(&addr6, 0, sizeof(addr6));
addr6.sin6_family = AF_INET6;
addr6.sin6_len = sizeof(addr6);
addr6.sin6_addr.s6_addr16[0] = htons(0xff02);
addr6.sin6_addr.s6_addr16[1] = htons(sc->sc_if.if_index);
addr6.sin6_addr.s6_addr8[15] = 0x12;
if ((imm = in6_joingroup(&sc->sc_if,
&addr6.sin6_addr, &error, 0)) == NULL) {
return (error);
}
/* join solicited multicast address */
2009-03-18 19:00:08 +03:00
memset(&addr6.sin6_addr, 0, sizeof(addr6.sin6_addr));
addr6.sin6_addr.s6_addr16[0] = htons(0xff02);
addr6.sin6_addr.s6_addr16[1] = htons(sc->sc_if.if_index);
addr6.sin6_addr.s6_addr32[1] = 0;
addr6.sin6_addr.s6_addr32[2] = htonl(1);
addr6.sin6_addr.s6_addr32[3] = 0;
addr6.sin6_addr.s6_addr8[12] = 0xff;
if ((imm2 = in6_joingroup(&sc->sc_if,
&addr6.sin6_addr, &error, 0)) == NULL) {
in6_leavegroup(imm);
return (error);
}
/* apply v6 multicast membership */
im6o->im6o_multicast_if_index = sc->sc_if.if_index;
if (imm)
LIST_INSERT_HEAD(&im6o->im6o_memberships, imm,
i6mm_chain);
if (imm2)
LIST_INSERT_HEAD(&im6o->im6o_memberships, imm2,
i6mm_chain);
return (0);
}
#endif /* INET6 */
static int
*** 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
carp_ioctl(struct ifnet *ifp, u_long cmd, void *data)
{
struct lwp *l = curlwp; /* XXX */
struct carp_softc *sc = ifp->if_softc, *vr;
struct carpreq carpr;
struct ifaddr *ifa;
struct ifreq *ifr;
struct ifnet *cdev = NULL;
int error = 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
ifa = (struct ifaddr *)data;
ifr = (struct ifreq *)data;
switch (cmd) {
*** 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
case SIOCINITIFADDR:
switch (ifa->ifa_addr->sa_family) {
#ifdef INET
case AF_INET:
sc->sc_if.if_flags |= IFF_UP;
memcpy(ifa->ifa_dstaddr, ifa->ifa_addr,
sizeof(struct sockaddr));
error = carp_set_addr(sc, satosin(ifa->ifa_addr));
break;
#endif /* INET */
#ifdef INET6
case AF_INET6:
sc->sc_if.if_flags|= IFF_UP;
error = carp_set_addr6(sc, satosin6(ifa->ifa_addr));
break;
#endif /* INET6 */
default:
error = EAFNOSUPPORT;
break;
}
break;
case SIOCSIFFLAGS:
*** 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 ((error = ifioctl_common(ifp, cmd, data)) != 0)
break;
if (sc->sc_state != INIT && !(ifr->ifr_flags & IFF_UP)) {
callout_stop(&sc->sc_ad_tmo);
callout_stop(&sc->sc_md_tmo);
callout_stop(&sc->sc_md6_tmo);
if (sc->sc_state == MASTER) {
/* we need the interface up to bow out */
sc->sc_if.if_flags |= IFF_UP;
sc->sc_bow_out = 1;
carp_send_ad(sc);
}
sc->sc_if.if_flags &= ~IFF_UP;
carp_set_state(sc, INIT);
carp_setrun(sc, 0);
} else if (sc->sc_state == INIT && (ifr->ifr_flags & IFF_UP)) {
sc->sc_if.if_flags |= IFF_UP;
carp_setrun(sc, 0);
}
break;
case SIOCSVH:
2006-10-25 22:11:22 +04:00
if (l == NULL)
break;
if ((error = kauth_authorize_network(l->l_cred,
KAUTH_NETWORK_INTERFACE,
2006-10-30 03:58:21 +03:00
KAUTH_REQ_NETWORK_INTERFACE_SETPRIV, ifp, (void *)cmd,
2006-10-25 22:11:22 +04:00
NULL)) != 0)
break;
if ((error = copyin(ifr->ifr_data, &carpr, sizeof carpr)))
break;
error = 1;
if (carpr.carpr_carpdev[0] != '\0' &&
(cdev = ifunit(carpr.carpr_carpdev)) == NULL)
return (EINVAL);
if ((error = carp_set_ifp(sc, cdev)))
return (error);
if (sc->sc_state != INIT && carpr.carpr_state != sc->sc_state) {
switch (carpr.carpr_state) {
case BACKUP:
callout_stop(&sc->sc_ad_tmo);
carp_set_state(sc, BACKUP);
carp_setrun(sc, 0);
carp_setroute(sc, RTM_DELETE);
break;
case MASTER:
carp_master_down(sc);
break;
default:
break;
}
}
if (carpr.carpr_vhid > 0) {
if (carpr.carpr_vhid > 255) {
error = EINVAL;
break;
}
if (sc->sc_carpdev) {
struct carp_if *cif;
cif = (struct carp_if *)sc->sc_carpdev->if_carp;
TAILQ_FOREACH(vr, &cif->vhif_vrs, sc_list)
if (vr != sc &&
vr->sc_vhid == carpr.carpr_vhid)
return (EINVAL);
}
sc->sc_vhid = carpr.carpr_vhid;
carp_set_enaddr(sc);
carp_set_state(sc, INIT);
error--;
}
if (carpr.carpr_advbase > 0 || carpr.carpr_advskew > 0) {
if (carpr.carpr_advskew > 254) {
error = EINVAL;
break;
}
if (carpr.carpr_advbase > 255) {
error = EINVAL;
break;
}
sc->sc_advbase = carpr.carpr_advbase;
sc->sc_advskew = carpr.carpr_advskew;
error--;
}
memcpy(sc->sc_key, carpr.carpr_key, sizeof(sc->sc_key));
if (error > 0)
error = EINVAL;
else {
error = 0;
carp_setrun(sc, 0);
}
break;
case SIOCGVH:
2009-03-18 19:00:08 +03:00
memset(&carpr, 0, sizeof(carpr));
if (sc->sc_carpdev != NULL)
strlcpy(carpr.carpr_carpdev, sc->sc_carpdev->if_xname,
IFNAMSIZ);
carpr.carpr_state = sc->sc_state;
carpr.carpr_vhid = sc->sc_vhid;
carpr.carpr_advbase = sc->sc_advbase;
carpr.carpr_advskew = sc->sc_advskew;
2006-05-25 19:22:05 +04:00
if ((l != NULL) && (error = kauth_authorize_network(l->l_cred,
2006-10-25 22:11:22 +04:00
KAUTH_NETWORK_INTERFACE,
2006-10-30 03:58:21 +03:00
KAUTH_REQ_NETWORK_INTERFACE_SETPRIV, ifp, (void *)cmd,
2009-05-13 01:48:42 +04:00
NULL)) == 0)
memcpy(carpr.carpr_key, sc->sc_key,
sizeof(carpr.carpr_key));
error = copyout(&carpr, ifr->ifr_data, sizeof(carpr));
break;
case SIOCADDMULTI:
error = carp_ether_addmulti(sc, ifr);
break;
case SIOCDELMULTI:
error = carp_ether_delmulti(sc, ifr);
break;
case SIOCSIFCAP:
2012-08-20 20:01:37 +04:00
if ((error = ifioctl_common(ifp, cmd, data)) == ENETRESET)
error = 0;
break;
default:
*** 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
error = ether_ioctl(ifp, cmd, data);
}
carp_hmac_prepare(sc);
return (error);
}
/*
* Start output on carp interface. This function should never be called.
*/
static void
carp_start(struct ifnet *ifp)
{
#ifdef DEBUG
printf("%s: start called\n", ifp->if_xname);
#endif
}
int
carp_output(struct ifnet *ifp, struct mbuf *m, const struct sockaddr *sa,
const struct rtentry *rt)
{
struct carp_softc *sc = ((struct carp_softc *)ifp->if_softc);
KASSERT(KERNEL_LOCKED_P());
if (sc->sc_carpdev != NULL && sc->sc_state == MASTER) {
2016-06-20 11:08:13 +03:00
return if_output_lock(sc->sc_carpdev, ifp, m, sa, rt);
} else {
m_freem(m);
return (ENETUNREACH);
}
}
static void
carp_set_state(struct carp_softc *sc, int state)
{
static const char *carp_states[] = { CARP_STATES };
int link_state;
if (sc->sc_state == state)
return;
CARP_LOG(sc, ("state transition from: %s -> to: %s", carp_states[sc->sc_state], carp_states[state]));
sc->sc_state = state;
switch (state) {
case BACKUP:
link_state = LINK_STATE_DOWN;
break;
case MASTER:
link_state = LINK_STATE_UP;
break;
default:
link_state = LINK_STATE_UNKNOWN;
break;
}
/*
* The lock is needed to serialize a call of
* if_link_state_change_softint from here and a call from softint.
*/
KERNEL_LOCK(1, NULL);
if_link_state_change_softint(&sc->sc_if, link_state);
KERNEL_UNLOCK_ONE(NULL);
}
void
carp_carpdev_state(void *v)
{
struct carp_if *cif;
struct carp_softc *sc;
struct ifnet *ifp = v;
if (ifp->if_type == IFT_CARP)
return;
cif = (struct carp_if *)ifp->if_carp;
TAILQ_FOREACH(sc, &cif->vhif_vrs, sc_list) {
int suppressed = sc->sc_suppress;
if (sc->sc_carpdev->if_link_state == LINK_STATE_DOWN ||
!(sc->sc_carpdev->if_flags & IFF_UP)) {
sc->sc_if.if_flags &= ~IFF_RUNNING;
callout_stop(&sc->sc_ad_tmo);
callout_stop(&sc->sc_md_tmo);
callout_stop(&sc->sc_md6_tmo);
carp_set_state(sc, INIT);
sc->sc_suppress = 1;
carp_setrun(sc, 0);
if (!suppressed) {
carp_suppress_preempt++;
if (carp_suppress_preempt == 1)
carp_send_ad_all();
}
} else {
carp_set_state(sc, INIT);
sc->sc_suppress = 0;
carp_setrun(sc, 0);
if (suppressed)
carp_suppress_preempt--;
}
}
}
static int
carp_ether_addmulti(struct carp_softc *sc, struct ifreq *ifr)
{
const struct sockaddr *sa = ifreq_getaddr(SIOCADDMULTI, ifr);
struct ifnet *ifp;
struct carp_mc_entry *mc;
u_int8_t addrlo[ETHER_ADDR_LEN], addrhi[ETHER_ADDR_LEN];
int error;
ifp = sc->sc_carpdev;
if (ifp == NULL)
return (EINVAL);
error = ether_addmulti(sa, &sc->sc_ac);
if (error != ENETRESET)
return (error);
/*
* This is new multicast address. We have to tell parent
* about it. Also, remember this multicast address so that
* we can delete them on unconfigure.
*/
2008-12-17 23:51:31 +03:00
mc = malloc(sizeof(struct carp_mc_entry), M_DEVBUF, M_NOWAIT);
if (mc == NULL) {
error = ENOMEM;
goto alloc_failed;
}
/*
* As ether_addmulti() returns ENETRESET, following two
* statement shouldn't fail.
*/
(void)ether_multiaddr(sa, addrlo, addrhi);
ETHER_LOCK(&sc->sc_ac);
mc->mc_enm = ether_lookup_multi(addrlo, addrhi, &sc->sc_ac);
ETHER_UNLOCK(&sc->sc_ac);
memcpy(&mc->mc_addr, sa, sa->sa_len);
LIST_INSERT_HEAD(&sc->carp_mc_listhead, mc, mc_entries);
error = if_mcast_op(ifp, SIOCADDMULTI, sa);
if (error != 0)
goto ioctl_failed;
return (error);
ioctl_failed:
LIST_REMOVE(mc, mc_entries);
2008-12-17 23:51:31 +03:00
free(mc, M_DEVBUF);
alloc_failed:
(void)ether_delmulti(sa, &sc->sc_ac);
return (error);
}
static int
carp_ether_delmulti(struct carp_softc *sc, struct ifreq *ifr)
{
const struct sockaddr *sa = ifreq_getaddr(SIOCDELMULTI, ifr);
struct ifnet *ifp;
struct ether_multi *enm;
struct carp_mc_entry *mc;
u_int8_t addrlo[ETHER_ADDR_LEN], addrhi[ETHER_ADDR_LEN];
int error;
ifp = sc->sc_carpdev;
if (ifp == NULL)
return (EINVAL);
/*
* Find a key to lookup carp_mc_entry. We have to do this
* before calling ether_delmulti for obvious reason.
*/
if ((error = ether_multiaddr(sa, addrlo, addrhi)) != 0)
return (error);
ETHER_LOCK(&sc->sc_ac);
enm = ether_lookup_multi(addrlo, addrhi, &sc->sc_ac);
ETHER_UNLOCK(&sc->sc_ac);
if (enm == NULL)
return (EINVAL);
LIST_FOREACH(mc, &sc->carp_mc_listhead, mc_entries)
if (mc->mc_enm == enm)
break;
/* We won't delete entries we didn't add */
if (mc == NULL)
return (EINVAL);
error = ether_delmulti(sa, &sc->sc_ac);
if (error != ENETRESET)
return (error);
/* We no longer use this multicast address. Tell parent so. */
error = if_mcast_op(ifp, SIOCDELMULTI, sa);
if (error == 0) {
/* And forget about this address. */
LIST_REMOVE(mc, mc_entries);
2008-12-17 23:51:31 +03:00
free(mc, M_DEVBUF);
} else
(void)ether_addmulti(sa, &sc->sc_ac);
return (error);
}
/*
* Delete any multicast address we have asked to add from parent
* interface. Called when the carp is being unconfigured.
*/
static void
carp_ether_purgemulti(struct carp_softc *sc)
{
struct ifnet *ifp = sc->sc_carpdev; /* Parent. */
struct carp_mc_entry *mc;
if (ifp == NULL)
return;
while ((mc = LIST_FIRST(&sc->carp_mc_listhead)) != NULL) {
(void)if_mcast_op(ifp, SIOCDELMULTI, sstosa(&mc->mc_addr));
LIST_REMOVE(mc, mc_entries);
2008-12-17 23:51:31 +03:00
free(mc, M_DEVBUF);
}
}
2008-04-15 10:03:28 +04:00
static int
sysctl_net_inet_carp_stats(SYSCTLFN_ARGS)
{
return (NETSTAT_SYSCTL(carpstat_percpu, CARP_NSTATS));
2008-04-15 10:03:28 +04:00
}
void
carp_init(void)
{
sysctl_net_inet_carp_setup(NULL);
#ifdef MBUFTRACE
MOWNER_ATTACH(&carp_proto_mowner_rx);
MOWNER_ATTACH(&carp_proto_mowner_tx);
MOWNER_ATTACH(&carp_proto6_mowner_rx);
MOWNER_ATTACH(&carp_proto6_mowner_tx);
#endif
carp_wqinput = wqinput_create("carp", _carp_proto_input);
#ifdef INET6
carp6_wqinput = wqinput_create("carp6", _carp6_proto_input);
#endif
}
static void
sysctl_net_inet_carp_setup(struct sysctllog **clog)
{
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
CTLTYPE_NODE, "inet", NULL,
NULL, 0, NULL, 0,
CTL_NET, PF_INET, CTL_EOL);
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
CTLTYPE_NODE, "carp",
SYSCTL_DESCR("CARP related settings"),
NULL, 0, NULL, 0,
CTL_NET, PF_INET, IPPROTO_CARP, CTL_EOL);
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
CTLTYPE_INT, "preempt",
SYSCTL_DESCR("Enable CARP Preempt"),
NULL, 0, &carp_opts[CARPCTL_PREEMPT], 0,
CTL_NET, PF_INET, IPPROTO_CARP,
CTL_CREATE, CTL_EOL);
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
CTLTYPE_INT, "arpbalance",
SYSCTL_DESCR("Enable ARP balancing"),
NULL, 0, &carp_opts[CARPCTL_ARPBALANCE], 0,
CTL_NET, PF_INET, IPPROTO_CARP,
CTL_CREATE, CTL_EOL);
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
CTLTYPE_INT, "allow",
SYSCTL_DESCR("Enable CARP"),
NULL, 0, &carp_opts[CARPCTL_ALLOW], 0,
CTL_NET, PF_INET, IPPROTO_CARP,
CTL_CREATE, CTL_EOL);
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
CTLTYPE_INT, "log",
SYSCTL_DESCR("CARP logging"),
NULL, 0, &carp_opts[CARPCTL_LOG], 0,
CTL_NET, PF_INET, IPPROTO_CARP,
CTL_CREATE, CTL_EOL);
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
CTLTYPE_STRUCT, "stats",
2006-05-25 19:22:05 +04:00
SYSCTL_DESCR("CARP statistics"),
2008-04-15 10:03:28 +04:00
sysctl_net_inet_carp_stats, 0, NULL, 0,
CTL_NET, PF_INET, IPPROTO_CARP, CARPCTL_STATS,
CTL_EOL);
}