/*
- * Copyright (c) 2000 Apple Computer, Inc. All rights reserved.
+ * Copyright (c) 2000-2009 Apple Inc. All rights reserved.
*
- * @APPLE_LICENSE_HEADER_START@
+ * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
*
- * The contents of this file constitute Original Code as defined in and
- * are subject to the Apple Public Source License Version 1.1 (the
- * "License"). You may not use this file except in compliance with the
- * License. Please obtain a copy of the License at
- * http://www.apple.com/publicsource and read it before using this file.
+ * This file contains Original Code and/or Modifications of Original Code
+ * as defined in and that are subject to the Apple Public Source License
+ * Version 2.0 (the 'License'). You may not use this file except in
+ * compliance with the License. The rights granted to you under the License
+ * may not be used to create, or enable the creation or redistribution of,
+ * unlawful or unlicensed copies of an Apple operating system, or to
+ * circumvent, violate, or enable the circumvention or violation of, any
+ * terms of an Apple operating system software license agreement.
*
- * This Original Code and all software distributed under the License are
- * distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY KIND, EITHER
+ * Please obtain a copy of the License at
+ * http://www.opensource.apple.com/apsl/ and read it before using this file.
+ *
+ * The Original Code and all software distributed under the License are
+ * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
* EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
* INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE OR NON-INFRINGEMENT. Please see the
- * License for the specific language governing rights and limitations
- * under the License.
+ * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
+ * Please see the License for the specific language governing rights and
+ * limitations under the License.
*
- * @APPLE_LICENSE_HEADER_END@
+ * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
*/
/*
* Copyright (c) 1988, 1991, 1993
#include <net/if.h>
#include <net/route.h>
+#include <net/dlil.h>
#include <net/raw_cb.h>
#include <netinet/in.h>
+#include <netinet/in_var.h>
+#include <netinet/in_arp.h>
+#include <netinet6/nd6.h>
#include <machine/spl.h>
-extern void m_copydata(struct mbuf *, int, int, caddr_t);
-extern void m_copyback(struct mbuf *, int, int, caddr_t);
-
extern struct rtstat rtstat;
-extern int rttrash;
+extern int check_routeselfref;
+extern struct domain routedomain;
MALLOC_DEFINE(M_RTABLE, "routetbl", "routing tables");
-extern lck_mtx_t *rt_mtx;
static struct sockaddr route_dst = { 2, PF_ROUTE, { 0, } };
static struct sockaddr route_src = { 2, PF_ROUTE, { 0, } };
static struct sockaddr sa_zero = { sizeof(sa_zero), AF_INET, { 0, } };
struct sysctl_req *w_req;
};
-static struct mbuf *
- rt_msg1(int, struct rt_addrinfo *);
+static struct mbuf *rt_msg1(int, struct rt_addrinfo *);
static int rt_msg2(int, struct rt_addrinfo *, caddr_t, struct walkarg *);
static int rt_xaddrs(caddr_t, caddr_t, struct rt_addrinfo *);
static int sysctl_dumpentry(struct radix_node *rn, void *vw);
static int sysctl_iflist(int af, struct walkarg *w);
static int sysctl_iflist2(int af, struct walkarg *w);
static int route_output(struct mbuf *, struct socket *);
-static void rt_setmetrics(u_long, struct rt_metrics *, struct rt_metrics *);
+static void rt_setmetrics(u_int32_t, struct rt_metrics *, struct rt_metrics *);
static void rt_setif(struct rtentry *, struct sockaddr *, struct sockaddr *,
- struct sockaddr *);
+ struct sockaddr *, unsigned int);
+#if IFNET_ROUTE_REFCNT
+static void rt_drainall(void);
+#endif /* IFNET_ROUTE_REFCNT */
+
+#define SIN(sa) ((struct sockaddr_in *)(size_t)(sa))
/* Sleazy use of local variables throughout file, warning!!!! */
#define dst info.rti_info[RTAX_DST]
#define ifaaddr info.rti_info[RTAX_IFA]
#define brdaddr info.rti_info[RTAX_BRD]
+SYSCTL_NODE(_net, OID_AUTO, idle, CTLFLAG_RW, 0, "idle network monitoring");
+
+#if IFNET_ROUTE_REFCNT
+static struct timeval last_ts;
+
+SYSCTL_NODE(_net_idle, OID_AUTO, route, CTLFLAG_RW, 0, "idle route monitoring");
+
+static int rt_if_idle_drain_interval = RT_IF_IDLE_DRAIN_INTERVAL;
+SYSCTL_INT(_net_idle_route, OID_AUTO, drain_interval, CTLFLAG_RW,
+ &rt_if_idle_drain_interval, 0, "Default interval for draining "
+ "routes when doing interface idle reference counting.");
+#endif /* IFNET_ROUTE_REFCNT */
+
/*
* It really doesn't make any sense at all for this code to share much
* with raw_usrreq.c, since its functionality is so restricted. XXX
rp = sotorawcb(so);
if (error) {
FREE(rp, M_PCB);
- so->so_pcb = 0;
+ so->so_pcb = NULL;
so->so_flags |= SOF_PCBCLEARING;
return error;
}
static int
rts_bind(struct socket *so, struct sockaddr *nam, struct proc *p)
{
- int s, error;
- s = splnet();
+ int error;
error = raw_usrreqs.pru_bind(so, nam, p); /* xxx just EINVAL */
- splx(s);
return error;
}
static int
rts_connect(struct socket *so, struct sockaddr *nam, struct proc *p)
{
- int s, error;
- s = splnet();
+ int error;
error = raw_usrreqs.pru_connect(so, nam, p); /* XXX just EINVAL */
- splx(s);
return error;
}
rts_detach(struct socket *so)
{
struct rawcb *rp = sotorawcb(so);
- int s, error;
+ int error;
- s = splnet();
if (rp != 0) {
switch(rp->rcb_proto.sp_protocol) {
case AF_INET:
route_cb.any_count--;
}
error = raw_usrreqs.pru_detach(so);
- splx(s);
return error;
}
static int
rts_disconnect(struct socket *so)
{
- int s, error;
- s = splnet();
+ int error;
error = raw_usrreqs.pru_disconnect(so);
- splx(s);
return error;
}
static int
rts_peeraddr(struct socket *so, struct sockaddr **nam)
{
- int s, error;
- s = splnet();
+ int error;
error = raw_usrreqs.pru_peeraddr(so, nam);
- splx(s);
return error;
}
rts_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
struct mbuf *control, struct proc *p)
{
- int s, error;
- s = splnet();
+ int error;
error = raw_usrreqs.pru_send(so, flags, m, nam, control, p);
- splx(s);
return error;
}
static int
rts_shutdown(struct socket *so)
{
- int s, error;
- s = splnet();
+ int error;
error = raw_usrreqs.pru_shutdown(so);
- splx(s);
return error;
}
static int
rts_sockaddr(struct socket *so, struct sockaddr **nam)
{
- int s, error;
- s = splnet();
+ int error;
error = raw_usrreqs.pru_sockaddr(so, nam);
- splx(s);
return error;
}
/*ARGSUSED*/
static int
-route_output(m, so)
- struct mbuf *m;
- struct socket *so;
+route_output(struct mbuf *m, struct socket *so)
{
- struct rt_msghdr *rtm = 0;
- struct rtentry *rt = 0;
- struct rtentry *saved_nrt = 0;
+ struct rt_msghdr *rtm = NULL;
+ struct rtentry *rt = NULL;
+ struct rtentry *saved_nrt = NULL;
struct radix_node_head *rnh;
struct rt_addrinfo info;
int len, error = 0;
- struct ifnet *ifp = 0;
+ struct ifnet *ifp = NULL;
#ifndef __APPLE__
struct proc *curproc = current_proc();
#endif
+ struct sockaddr_in dst_in, gate_in;
int sendonlytoself = 0;
+ unsigned int ifscope = IFSCOPE_NONE;
#define senderr(e) { error = e; goto flush;}
- if (m == 0 || ((m->m_len < sizeof(long)) && (m = m_pullup(m, sizeof(long))) == 0))
+ if (m == NULL ||
+ ((m->m_len < sizeof(intptr_t)) && (m = m_pullup(m, sizeof(intptr_t))) == 0))
return (ENOBUFS);
if ((m->m_flags & M_PKTHDR) == 0)
panic("route_output");
/* unlock the socket (but keep a reference) it won't be accessed until raw_input appends to it. */
socket_unlock(so, 0);
- lck_mtx_lock(rt_mtx);
+ lck_mtx_lock(rnh_lock);
len = m->m_pkthdr.len;
if (len < sizeof(*rtm) ||
len != mtod(m, struct rt_msghdr *)->rtm_msglen) {
- dst = 0;
+ dst = NULL;
senderr(EINVAL);
}
R_Malloc(rtm, struct rt_msghdr *, len);
- if (rtm == 0) {
- dst = 0;
+ if (rtm == NULL) {
+ dst = NULL;
senderr(ENOBUFS);
}
m_copydata(m, 0, len, (caddr_t)rtm);
if (rtm->rtm_version != RTM_VERSION) {
- dst = 0;
+ dst = NULL;
senderr(EPROTONOSUPPORT);
}
-
+
/*
* Silent version of RTM_GET for Reachabiltiy APIs. We may change
* all RTM_GETs to be silent in the future, so this is private for now.
sendonlytoself = 1;
rtm->rtm_type = RTM_GET;
}
-
+
/*
* Perform permission checking, only privileged sockets
* may perform operations other than RTM_GET
*/
if (rtm->rtm_type != RTM_GET && (so->so_state & SS_PRIV) == 0) {
- dst = 0;
+ dst = NULL;
senderr(EPERM);
}
rtm->rtm_pid = proc_selfpid();
info.rti_addrs = rtm->rtm_addrs;
if (rt_xaddrs((caddr_t)(rtm + 1), len + (caddr_t)rtm, &info)) {
- dst = 0;
+ dst = NULL;
senderr(EINVAL);
}
- if (dst == 0 || (dst->sa_family >= AF_MAX)
- || (gate != 0 && (gate->sa_family >= AF_MAX))) {
+ if (dst == NULL || (dst->sa_family >= AF_MAX) ||
+ (gate != NULL && (gate->sa_family >= AF_MAX))) {
senderr(EINVAL);
}
+
+ if (dst->sa_family == AF_INET && dst->sa_len != sizeof (dst_in)) {
+ /* At minimum, we need up to sin_addr */
+ if (dst->sa_len < offsetof(struct sockaddr_in, sin_zero))
+ senderr(EINVAL);
+ bzero(&dst_in, sizeof (dst_in));
+ dst_in.sin_len = sizeof (dst_in);
+ dst_in.sin_family = AF_INET;
+ dst_in.sin_port = SIN(dst)->sin_port;
+ dst_in.sin_addr = SIN(dst)->sin_addr;
+ dst = (struct sockaddr *)&dst_in;
+ }
+
+ if (gate != NULL &&
+ gate->sa_family == AF_INET && gate->sa_len != sizeof (gate_in)) {
+ /* At minimum, we need up to sin_addr */
+ if (gate->sa_len < offsetof(struct sockaddr_in, sin_zero))
+ senderr(EINVAL);
+ bzero(&gate_in, sizeof (gate_in));
+ gate_in.sin_len = sizeof (gate_in);
+ gate_in.sin_family = AF_INET;
+ gate_in.sin_port = SIN(gate)->sin_port;
+ gate_in.sin_addr = SIN(gate)->sin_addr;
+ gate = (struct sockaddr *)&gate_in;
+ }
+
if (genmask) {
struct radix_node *t;
t = rn_addmask((caddr_t)genmask, 0, 1);
else
senderr(ENOBUFS);
}
+
+ /*
+ * If RTF_IFSCOPE flag is set, then rtm_index specifies the scope.
+ */
+ if (rtm->rtm_flags & RTF_IFSCOPE) {
+ /* Scoped routing is for AF_INET only */
+ if (dst->sa_family != AF_INET)
+ senderr(EINVAL);
+ ifscope = rtm->rtm_index;
+ }
+
switch (rtm->rtm_type) {
-
+
case RTM_ADD:
- if (gate == 0)
+ if (gate == NULL)
senderr(EINVAL);
#ifdef __APPLE__
* confusing the routing table with a wrong route to the previous default gateway
*/
{
- extern int check_routeselfref;
#define satosinaddr(sa) (((struct sockaddr_in *)sa)->sin_addr.s_addr)
if (check_routeselfref && (dst && dst->sa_family == AF_INET) &&
}
}
#endif
- error = rtrequest_locked(RTM_ADD, dst, gate, netmask,
- rtm->rtm_flags, &saved_nrt);
+ error = rtrequest_scoped_locked(RTM_ADD, dst, gate,
+ netmask, rtm->rtm_flags, &saved_nrt, ifscope);
if (error == 0 && saved_nrt) {
+ RT_LOCK(saved_nrt);
#ifdef __APPLE__
/*
* If the route request specified an interface with
* dwiggins@bbn.com
*/
- rt_setif(saved_nrt, ifpaddr, ifaaddr, gate);
+ rt_setif(saved_nrt, ifpaddr, ifaaddr, gate,
+ ifscope);
#endif
rt_setmetrics(rtm->rtm_inits,
&rtm->rtm_rmx, &saved_nrt->rt_rmx);
saved_nrt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
saved_nrt->rt_rmx.rmx_locks |=
(rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
- rtunref(saved_nrt);
saved_nrt->rt_genmask = genmask;
+ RT_REMREF_LOCKED(saved_nrt);
+ RT_UNLOCK(saved_nrt);
}
break;
case RTM_DELETE:
- error = rtrequest_locked(RTM_DELETE, dst, gate, netmask,
- rtm->rtm_flags, &saved_nrt);
+ error = rtrequest_scoped_locked(RTM_DELETE, dst,
+ gate, netmask, rtm->rtm_flags, &saved_nrt, ifscope);
if (error == 0) {
- if ((rt = saved_nrt))
- rtref(rt);
+ rt = saved_nrt;
+ RT_LOCK(rt);
goto report;
}
break;
case RTM_GET:
case RTM_CHANGE:
case RTM_LOCK:
- if ((rnh = rt_tables[dst->sa_family]) == 0) {
+ if ((rnh = rt_tables[dst->sa_family]) == NULL)
senderr(EAFNOSUPPORT);
- } else if ((rt = (struct rtentry *)
- rnh->rnh_lookup(dst, netmask, rnh)) != NULL)
- rtref(rt);
- else
+
+ /*
+ * Lookup the best match based on the key-mask pair;
+ * callee adds a reference and checks for root node.
+ */
+ rt = rt_lookup(TRUE, dst, netmask, rnh, ifscope);
+ if (rt == NULL)
senderr(ESRCH);
+ RT_LOCK(rt);
+
switch(rtm->rtm_type) {
case RTM_GET: {
struct ifaddr *ifa2;
report:
+ ifa2 = NULL;
+ RT_LOCK_ASSERT_HELD(rt);
dst = rt_key(rt);
gate = rt->rt_gateway;
netmask = rt_mask(rt);
ifnet_lock_shared(ifp);
ifa2 = ifp->if_addrhead.tqh_first;
ifpaddr = ifa2->ifa_addr;
+ ifaref(ifa2);
ifnet_lock_done(ifp);
ifaaddr = rt->rt_ifa->ifa_addr;
rtm->rtm_index = ifp->if_index;
struct rt_msghdr *new_rtm;
R_Malloc(new_rtm, struct rt_msghdr *, len);
if (new_rtm == 0) {
+ RT_UNLOCK(rt);
+ if (ifa2 != NULL)
+ ifafree(ifa2);
senderr(ENOBUFS);
}
Bcopy(rtm, new_rtm, rtm->rtm_msglen);
rtm->rtm_flags = rt->rt_flags;
rtm->rtm_rmx = rt->rt_rmx;
rtm->rtm_addrs = info.rti_addrs;
+ if (ifa2 != NULL)
+ ifafree(ifa2);
}
break;
case RTM_CHANGE:
- if (gate && (error = rt_setgate(rt, rt_key(rt), gate)))
+ if (gate && (error = rt_setgate(rt,
+ rt_key(rt), gate))) {
+ RT_UNLOCK(rt);
senderr(error);
-
+ }
/*
* If they tried to change things but didn't specify
* the required gateway, then just use the old one.
* equivalent to the code found at this very spot
* in BSD.
*/
- rt_setif(rt, ifpaddr, ifaaddr, gate);
+ rt_setif(rt, ifpaddr, ifaaddr, gate,
+ ifscope);
#endif
rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx,
(rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
break;
}
+ RT_UNLOCK(rt);
break;
default:
else
rtm->rtm_flags |= RTF_DONE;
}
- if (rt)
+ if (rt != NULL) {
+ RT_LOCK_ASSERT_NOTHELD(rt);
rtfree_locked(rt);
- lck_mtx_unlock(rt_mtx);
+ }
+ lck_mtx_unlock(rnh_lock);
socket_lock(so, 0); /* relock the socket now */
{
struct rawcb *rp = 0;
}
static void
-rt_setmetrics(which, in, out)
- u_long which;
- struct rt_metrics *in, *out;
+rt_setmetrics(u_int32_t which, struct rt_metrics *in, struct rt_metrics *out)
{
#define metric(f, e) if (which & (f)) out->e = in->e;
metric(RTV_RPIPE, rmx_recvpipe);
* Set route's interface given ifpaddr, ifaaddr, and gateway.
*/
static void
-rt_setif(
- struct rtentry *rt,
- struct sockaddr *Ifpaddr,
- struct sockaddr *Ifaaddr,
- struct sockaddr *Gate)
+rt_setif(struct rtentry *rt, struct sockaddr *Ifpaddr, struct sockaddr *Ifaaddr,
+ struct sockaddr *Gate, unsigned int ifscope)
{
struct ifaddr *ifa = 0;
struct ifnet *ifp = 0;
- lck_mtx_assert(rt_mtx, LCK_MTX_ASSERT_OWNED);
+ lck_mtx_assert(rnh_lock, LCK_MTX_ASSERT_OWNED);
+
+ RT_LOCK_ASSERT_HELD(rt);
+
+ /* trigger route cache reevaluation */
+ if (use_routegenid)
+ routegenid_update();
+
+ /* Don't update a defunct route */
+ if (rt->rt_flags & RTF_CONDEMNED)
+ return;
- /* new gateway could require new ifaddr, ifp;
- flags may also be different; ifp may be specified
- by ll sockaddr when protocol address is ambiguous */
- if (Ifpaddr && (ifa = ifa_ifwithnet(Ifpaddr)) &&
+ /* Add an extra ref for ourselves */
+ RT_ADDREF_LOCKED(rt);
+
+ /*
+ * New gateway could require new ifaddr, ifp; flags may also
+ * be different; ifp may be specified by ll sockaddr when
+ * protocol address is ambiguous.
+ */
+ if (Ifpaddr && (ifa = ifa_ifwithnet_scoped(Ifpaddr, ifscope)) &&
(ifp = ifa->ifa_ifp) && (Ifaaddr || Gate)) {
- ifafree(ifa);
- ifa = ifaof_ifpforaddr(Ifaaddr ? Ifaaddr : Gate,
- ifp);
- }
- else
- {
+ ifafree(ifa);
+ ifa = ifaof_ifpforaddr(Ifaaddr ? Ifaaddr : Gate, ifp);
+ } else {
if (ifa) {
ifafree(ifa);
ifa = 0;
if (Ifpaddr && (ifp = if_withname(Ifpaddr)) ) {
if (Gate) {
ifa = ifaof_ifpforaddr(Gate, ifp);
- }
- else {
+ } else {
ifnet_lock_shared(ifp);
ifa = TAILQ_FIRST(&ifp->if_addrhead);
- ifaref(ifa);
+ if (ifa != NULL)
+ ifaref(ifa);
ifnet_lock_done(ifp);
}
- }
- else if (Ifaaddr && (ifa = ifa_ifwithaddr(Ifaaddr))) {
- ifp = ifa->ifa_ifp;
- }
- else if (Gate && (ifa = ifa_ifwithroute(rt->rt_flags,
- rt_key(rt), Gate))) {
+ } else if (Ifaaddr &&
+ (ifa = ifa_ifwithaddr_scoped(Ifaaddr, ifscope))) {
ifp = ifa->ifa_ifp;
+ } else if (Gate != NULL) {
+ /*
+ * Safe to drop rt_lock and use rt_key, since holding
+ * rnh_lock here prevents another thread from calling
+ * rt_setgate() on this route. We cannot hold the
+ * lock across ifa_ifwithroute since the lookup done
+ * by that routine may point to the same route.
+ */
+ RT_UNLOCK(rt);
+ if ((ifa = ifa_ifwithroute_scoped_locked(rt->rt_flags,
+ rt_key(rt), Gate, ifscope)) != NULL)
+ ifp = ifa->ifa_ifp;
+ RT_LOCK(rt);
+ /* Don't update a defunct route */
+ if (rt->rt_flags & RTF_CONDEMNED) {
+ if (ifa != NULL)
+ ifafree(ifa);
+ /* Release extra ref */
+ RT_REMREF_LOCKED(rt);
+ return;
+ }
}
}
if (ifa) {
struct ifaddr *oifa = rt->rt_ifa;
if (oifa != ifa) {
- if (oifa && oifa->ifa_rtrequest)
- oifa->ifa_rtrequest(RTM_DELETE,
- rt, Gate);
+ if (oifa && oifa->ifa_rtrequest)
+ oifa->ifa_rtrequest(RTM_DELETE, rt, Gate);
rtsetifa(rt, ifa);
- rt->rt_ifp = ifp;
- rt->rt_rmx.rmx_mtu = ifp->if_mtu;
- if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest)
+#if IFNET_ROUTE_REFCNT
+ /*
+ * Adjust route ref count for the interfaces.
+ */
+ if (rt->rt_if_ref_fn != NULL && rt->rt_ifp != ifp) {
+ rt->rt_if_ref_fn(ifp, 1);
+ rt->rt_if_ref_fn(rt->rt_ifp, -1);
+ }
+#endif /* IFNET_ROUTE_REFCNT */
+ rt->rt_ifp = ifp;
+ /*
+ * If this is the (non-scoped) default route, record
+ * the interface index used for the primary ifscope.
+ */
+ if (rt_inet_default(rt, rt_key(rt)))
+ set_primary_ifscope(rt->rt_ifp->if_index);
+ rt->rt_rmx.rmx_mtu = ifp->if_mtu;
+ if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest)
rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, Gate);
- } else {
ifafree(ifa);
- goto call_ifareq;
+ /* Release extra ref */
+ RT_REMREF_LOCKED(rt);
+ return;
}
ifafree(ifa);
- return;
}
- call_ifareq:
+
/* XXX: to reset gateway to correct value, at RTM_CHANGE */
if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest)
rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, Gate);
-}
+ /* Release extra ref */
+ RT_REMREF_LOCKED(rt);
+}
-#define ROUNDUP(a) \
- ((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long))
-#define ADVANCE(x, n) (x += ROUNDUP((n)->sa_len))
+#define ROUNDUP32(a) \
+ ((a) > 0 ? (1 + (((a) - 1) | (sizeof(uint32_t) - 1))) : sizeof(uint32_t))
+#define ADVANCE32(x, n) (x += ROUNDUP32((n)->sa_len))
/*
* This data is derived straight from userland.
*/
static int
-rt_xaddrs(cp, cplim, rtinfo)
- caddr_t cp, cplim;
- struct rt_addrinfo *rtinfo;
+rt_xaddrs(caddr_t cp, caddr_t cplim, struct rt_addrinfo *rtinfo)
{
struct sockaddr *sa;
int i;
/* accept it */
rtinfo->rti_info[i] = sa;
- ADVANCE(cp, sa);
+ ADVANCE32(cp, sa);
}
return (0);
}
static struct mbuf *
-rt_msg1(
- int type,
- struct rt_addrinfo *rtinfo)
+rt_msg1(int type, struct rt_addrinfo *rtinfo)
{
struct rt_msghdr *rtm;
struct mbuf *m;
int i;
- struct sockaddr *sa;
int len, dlen;
switch (type) {
rtm = mtod(m, struct rt_msghdr *);
bzero((caddr_t)rtm, len);
for (i = 0; i < RTAX_MAX; i++) {
+ struct sockaddr *sa, *hint;
+ struct sockaddr_storage ss;
+
if ((sa = rtinfo->rti_info[i]) == NULL)
continue;
+
+ switch (i) {
+ case RTAX_DST:
+ case RTAX_NETMASK:
+ if ((hint = rtinfo->rti_info[RTAX_DST]) == NULL)
+ hint = rtinfo->rti_info[RTAX_IFA];
+
+ /* Scrub away any trace of embedded interface scope */
+ sa = rtm_scrub_ifscope(i, hint, sa, &ss);
+ break;
+
+ default:
+ break;
+ }
+
rtinfo->rti_addrs |= (1 << i);
- dlen = ROUNDUP(sa->sa_len);
+ dlen = ROUNDUP32(sa->sa_len);
m_copyback(m, len, dlen, (caddr_t)sa);
len += dlen;
}
}
static int
-rt_msg2(type, rtinfo, cp, w)
- int type;
- struct rt_addrinfo *rtinfo;
- caddr_t cp;
- struct walkarg *w;
+rt_msg2(int type, struct rt_addrinfo *rtinfo, caddr_t cp, struct walkarg *w)
{
int i;
int len, dlen, second_time = 0;
if (cp0)
cp += len;
for (i = 0; i < RTAX_MAX; i++) {
- struct sockaddr *sa;
+ struct sockaddr *sa, *hint;
+ struct sockaddr_storage ss;
if ((sa = rtinfo->rti_info[i]) == 0)
continue;
+
+ switch (i) {
+ case RTAX_DST:
+ case RTAX_NETMASK:
+ if ((hint = rtinfo->rti_info[RTAX_DST]) == NULL)
+ hint = rtinfo->rti_info[RTAX_IFA];
+
+ /* Scrub away any trace of embedded interface scope */
+ sa = rtm_scrub_ifscope(i, hint, sa, &ss);
+ break;
+
+ default:
+ break;
+ }
+
rtinfo->rti_addrs |= (1 << i);
- dlen = ROUNDUP(sa->sa_len);
+ dlen = ROUNDUP32(sa->sa_len);
if (cp) {
bcopy((caddr_t)sa, cp, (unsigned)dlen);
cp += dlen;
* destination.
*/
void
-rt_missmsg(type, rtinfo, flags, error)
- int type, flags, error;
- struct rt_addrinfo *rtinfo;
+rt_missmsg(int type, struct rt_addrinfo *rtinfo, int flags, int error)
{
struct rt_msghdr *rtm;
struct mbuf *m;
struct sockaddr *sa = rtinfo->rti_info[RTAX_DST];
- lck_mtx_assert(rt_mtx, LCK_MTX_ASSERT_OWNED);
-
if (route_cb.any_count == 0)
return;
m = rt_msg1(type, rtinfo);
ifm = mtod(m, struct if_msghdr *);
ifm->ifm_index = ifp->if_index;
ifm->ifm_flags = (u_short)ifp->if_flags;
- if_data_internal_to_if_data(&ifp->if_data, &ifm->ifm_data);
+ if_data_internal_to_if_data(ifp, &ifp->if_data, &ifm->ifm_data);
ifm->ifm_addrs = 0;
route_proto.sp_protocol = 0;
raw_input(m, &route_proto, &route_src, &route_dst);
* copies of it.
*
* Since this is coming from the interface, it is expected that the
- * interface will be locked.
+ * interface will be locked. Caller must hold rt_lock.
*/
void
-rt_newaddrmsg(cmd, ifa, error, rt)
- int cmd, error;
- struct ifaddr *ifa;
- struct rtentry *rt;
+rt_newaddrmsg(int cmd, struct ifaddr *ifa, int error, struct rtentry *rt)
{
struct rt_addrinfo info;
struct sockaddr *sa = 0;
struct mbuf *m = 0;
struct ifnet *ifp = ifa->ifa_ifp;
+ RT_LOCK_ASSERT_HELD(rt);
+
if (route_cb.any_count == 0)
return;
for (pass = 1; pass < 3; pass++) {
struct ifa_msghdr *ifam;
int ncmd = cmd == RTM_ADD ? RTM_NEWADDR : RTM_DELADDR;
+ /* Lock ifp for if_addrhead */
+ ifnet_lock_shared(ifp);
ifaaddr = sa = ifa->ifa_addr;
ifpaddr = ifp->if_addrhead.tqh_first->ifa_addr;
netmask = ifa->ifa_netmask;
brdaddr = ifa->ifa_dstaddr;
- if ((m = rt_msg1(ncmd, &info)) == NULL)
+ if ((m = rt_msg1(ncmd, &info)) == NULL) {
+ ifnet_lock_done(ifp);
continue;
+ }
+ ifnet_lock_done(ifp);
ifam = mtod(m, struct ifa_msghdr *);
ifam->ifam_index = ifp->if_index;
ifam->ifam_metric = ifa->ifa_metric;
* there is no route state to worry about.
*/
void
-rt_newmaddrmsg(cmd, ifma)
- int cmd;
- struct ifmultiaddr *ifma;
+rt_newmaddrmsg(int cmd, struct ifmultiaddr *ifma)
{
struct rt_addrinfo info;
struct mbuf *m = 0;
bzero((caddr_t)&info, sizeof(info));
ifaaddr = ifma->ifma_addr;
+ /* Lock ifp for if_addrhead */
+ if (ifp != NULL)
+ ifnet_lock_shared(ifp);
if (ifp && ifp->if_addrhead.tqh_first)
ifpaddr = ifp->if_addrhead.tqh_first->ifa_addr;
else
* (similarly to how ARP entries, e.g., are presented).
*/
gate = ifma->ifma_ll->ifma_addr;
- if ((m = rt_msg1(cmd, &info)) == NULL)
+ if ((m = rt_msg1(cmd, &info)) == NULL) {
+ if (ifp != NULL)
+ ifnet_lock_done(ifp);
return;
+ }
+ if (ifp != NULL)
+ ifnet_lock_done(ifp);
ifmam = mtod(m, struct ifma_msghdr *);
ifmam->ifmam_index = ifp ? ifp->if_index : 0;
ifmam->ifmam_addrs = info.rti_addrs;
* This is used in dumping the kernel table via sysctl().
*/
int
-sysctl_dumpentry(rn, vw)
- struct radix_node *rn;
- void *vw;
+sysctl_dumpentry(struct radix_node *rn, void *vw)
{
struct walkarg *w = vw;
struct rtentry *rt = (struct rtentry *)rn;
int error = 0, size;
struct rt_addrinfo info;
- if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg))
+ RT_LOCK(rt);
+ if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg)) {
+ RT_UNLOCK(rt);
return 0;
+ }
bzero((caddr_t)&info, sizeof(info));
dst = rt_key(rt);
gate = rt->rt_gateway;
rtm->rtm_errno = 0;
rtm->rtm_addrs = info.rti_addrs;
error = SYSCTL_OUT(w->w_req, (caddr_t)rtm, size);
+ RT_UNLOCK(rt);
return (error);
}
} else {
rtm->rtm_reserved = 0;
rtm->rtm_addrs = info.rti_addrs;
error = SYSCTL_OUT(w->w_req, (caddr_t)rtm, size);
+ RT_UNLOCK(rt);
return (error);
}
}
+ RT_UNLOCK(rt);
return (error);
}
ifm = (struct if_msghdr *)w->w_tmem;
ifm->ifm_index = ifp->if_index;
ifm->ifm_flags = (u_short)ifp->if_flags;
- if_data_internal_to_if_data(&ifp->if_data, &ifm->ifm_data);
+ if_data_internal_to_if_data(ifp, &ifp->if_data, &ifm->ifm_data);
ifm->ifm_addrs = info.rti_addrs;
error = SYSCTL_OUT(w->w_req,(caddr_t)ifm, len);
if (error) {
while ((ifa = ifa->ifa_link.tqe_next) != 0) {
if (af && af != ifa->ifa_addr->sa_family)
continue;
-#ifndef __APPLE__
- if (curproc->p_prison && prison_if(curproc, ifa->ifa_addr))
- continue;
-#endif
ifaaddr = ifa->ifa_addr;
netmask = ifa->ifa_netmask;
brdaddr = ifa->ifa_dstaddr;
ifm->ifm_snd_maxlen = ifp->if_snd.ifq_maxlen;
ifm->ifm_snd_drops = ifp->if_snd.ifq_drops;
ifm->ifm_timer = ifp->if_timer;
- if_data_internal_to_if_data64(&ifp->if_data, &ifm->ifm_data);
+ if_data_internal_to_if_data64(ifp, &ifp->if_data, &ifm->ifm_data);
error = SYSCTL_OUT(w->w_req, w->w_tmem, len);
if (error) {
ifnet_lock_done(ifp);
return 0;
}
+#if IFNET_ROUTE_REFCNT
+/*
+ * Called from pfslowtimo(), protected by domain_proto_mtx
+ */
+static void
+rt_drainall(void)
+{
+ struct timeval delta_ts, current_ts;
+
+ /*
+ * This test is done without holding rnh_lock; in the even that
+ * we read stale value, it will only cause an extra (or miss)
+ * drain and is therefore harmless.
+ */
+ if (ifnet_aggressive_drainers == 0) {
+ if (timerisset(&last_ts))
+ timerclear(&last_ts);
+ return;
+ }
+
+ microuptime(¤t_ts);
+ timersub(¤t_ts, &last_ts, &delta_ts);
+
+ if (delta_ts.tv_sec >= rt_if_idle_drain_interval) {
+ timerclear(&last_ts);
+
+ in_rtqdrain(); /* protocol cloned routes: INET */
+ in6_rtqdrain(); /* protocol cloned routes: INET6 */
+ in_arpdrain(NULL); /* cloned routes: ARP */
+ nd6_drain(NULL); /* cloned routes: ND6 */
+
+ last_ts.tv_sec = current_ts.tv_sec;
+ last_ts.tv_usec = current_ts.tv_usec;
+ }
+}
+
+void
+rt_aggdrain(int on)
+{
+ lck_mtx_assert(rnh_lock, LCK_MTX_ASSERT_OWNED);
+
+ if (on)
+ routedomain.dom_protosw->pr_flags |= PR_AGGDRAIN;
+ else
+ routedomain.dom_protosw->pr_flags &= ~PR_AGGDRAIN;
+}
+#endif /* IFNET_ROUTE_REFCNT */
static int
sysctl_rtsock SYSCTL_HANDLER_ARGS
{
+#pragma unused(oidp)
int *name = (int *)arg1;
u_int namelen = arg2;
struct radix_node_head *rnh;
case NET_RT_DUMP:
case NET_RT_DUMP2:
case NET_RT_FLAGS:
- lck_mtx_lock(rt_mtx);
+ lck_mtx_lock(rnh_lock);
for (i = 1; i <= AF_MAX; i++)
if ((rnh = rt_tables[i]) && (af == 0 || af == i) &&
(error = rnh->rnh_walktree(rnh,
sysctl_dumpentry, &w)))
break;
- lck_mtx_unlock(rt_mtx);
+ lck_mtx_unlock(rnh_lock);
break;
case NET_RT_IFLIST:
error = sysctl_iflist(af, &w);
/*
* Definitions of protocols supported in the ROUTE domain.
*/
-
-struct domain routedomain; /* or at least forward */
-
static struct protosw routesw[] = {
{ SOCK_RAW, &routedomain, 0, PR_ATOMIC|PR_ADDR,
0, route_output, raw_ctlinput, 0,
0,
- raw_init, 0, 0, 0,
+ raw_init, 0, 0,
+#if IFNET_ROUTE_REFCNT
+ rt_drainall,
+#else
+ 0,
+#endif /* IFNET_ROUTE_REFCNT */
0,
&route_usrreqs,
0, 0, 0,
struct domain routedomain =
{ PF_ROUTE, "route", route_init, 0, 0,
routesw,
- 0, 0, 0, 0, 0, 0, 0, 0,
+ NULL, NULL, 0, 0, 0, 0, NULL, 0,
{ 0, 0 } };
DOMAIN_SET(route);
projects / apple / xnu.git / blobdiff