#include <sys/param.h>
#include <sys/systm.h>
#include <sys/malloc.h>
#include <sys/mbuf.h>
#include <sys/domain.h>
#include <sys/protosw.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/proc.h>
#ifndef __APPLE__
#include <sys/jail.h>
#endif
#include <sys/kernel.h>
#include <sys/sysctl.h>
#include <libkern/OSAtomic.h>
#include <machine/limits.h>
#ifdef __APPLE__
#include <kern/zalloc.h>
#endif
#include <net/if.h>
#include <net/if_types.h>
#include <net/route.h>
#include <netinet/in.h>
#include <netinet/in_pcb.h>
#include <netinet/in_var.h>
#include <netinet/ip_var.h>
#if INET6
#include <netinet/ip6.h>
#include <netinet6/ip6_var.h>
#endif
#include "faith.h"
#if IPSEC
#include <netinet6/ipsec.h>
#include <netkey/key.h>
#endif
#include <sys/kdebug.h>
#include <sys/random.h>
#if IPSEC
extern int ipsec_bypass;
#endif
#define DBG_FNC_PCB_LOOKUP NETDBG_CODE(DBG_NETTCP, (6 << 8))
#define DBG_FNC_PCB_HLOOKUP NETDBG_CODE(DBG_NETTCP, ((6 << 8) | 1))
struct in_addr zeroin_addr;
int ipport_lowfirstauto = IPPORT_RESERVED - 1;
int ipport_lowlastauto = IPPORT_RESERVEDSTART;
#ifndef __APPLE__
int ipport_firstauto = IPPORT_RESERVED;
int ipport_lastauto = IPPORT_USERRESERVED;
#else
int ipport_firstauto = IPPORT_HIFIRSTAUTO;
int ipport_lastauto = IPPORT_HILASTAUTO;
#endif
int ipport_hifirstauto = IPPORT_HIFIRSTAUTO;
int ipport_hilastauto = IPPORT_HILASTAUTO;
#define RANGECHK(var, min, max) \
if ((var) < (min)) { (var) = (min); } \
else if ((var) > (max)) { (var) = (max); }
static int
sysctl_net_ipport_check SYSCTL_HANDLER_ARGS
{
#pragma unused(arg1, arg2)
int error = sysctl_handle_int(oidp,
oidp->oid_arg1, oidp->oid_arg2, req);
if (!error) {
RANGECHK(ipport_lowfirstauto, 1, IPPORT_RESERVED - 1);
RANGECHK(ipport_lowlastauto, 1, IPPORT_RESERVED - 1);
RANGECHK(ipport_firstauto, IPPORT_RESERVED, USHRT_MAX);
RANGECHK(ipport_lastauto, IPPORT_RESERVED, USHRT_MAX);
RANGECHK(ipport_hifirstauto, IPPORT_RESERVED, USHRT_MAX);
RANGECHK(ipport_hilastauto, IPPORT_RESERVED, USHRT_MAX);
}
return error;
}
#undef RANGECHK
SYSCTL_NODE(_net_inet_ip, IPPROTO_IP, portrange, CTLFLAG_RW|CTLFLAG_LOCKED, 0, "IP Ports");
SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, lowfirst, CTLTYPE_INT|CTLFLAG_RW,
&ipport_lowfirstauto, 0, &sysctl_net_ipport_check, "I", "");
SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, lowlast, CTLTYPE_INT|CTLFLAG_RW,
&ipport_lowlastauto, 0, &sysctl_net_ipport_check, "I", "");
SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, first, CTLTYPE_INT|CTLFLAG_RW,
&ipport_firstauto, 0, &sysctl_net_ipport_check, "I", "");
SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, last, CTLTYPE_INT|CTLFLAG_RW,
&ipport_lastauto, 0, &sysctl_net_ipport_check, "I", "");
SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, hifirst, CTLTYPE_INT|CTLFLAG_RW,
&ipport_hifirstauto, 0, &sysctl_net_ipport_check, "I", "");
SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, hilast, CTLTYPE_INT|CTLFLAG_RW,
&ipport_hilastauto, 0, &sysctl_net_ipport_check, "I", "");
extern int udp_use_randomport;
extern int tcp_use_randomport;
int
in_pcballoc(struct socket *so, struct inpcbinfo *pcbinfo, __unused struct proc *p)
{
struct inpcb *inp;
caddr_t temp;
#if IPSEC
#ifndef __APPLE__
int error;
#endif
#endif
#if CONFIG_MACF_NET
int mac_error;
#endif
if (so->cached_in_sock_layer == 0) {
#if TEMPDEBUG
printf("PCBALLOC calling zalloc for socket %x\n", so);
#endif
inp = (struct inpcb *) zalloc(pcbinfo->ipi_zone);
if (inp == NULL)
return (ENOBUFS);
bzero((caddr_t)inp, sizeof(*inp));
}
else {
#if TEMPDEBUG
printf("PCBALLOC reusing PCB for socket %x\n", so);
#endif
inp = (struct inpcb *) so->so_saved_pcb;
temp = inp->inp_saved_ppcb;
bzero((caddr_t) inp, sizeof(*inp));
inp->inp_saved_ppcb = temp;
}
inp->inp_gencnt = ++pcbinfo->ipi_gencnt;
inp->inp_pcbinfo = pcbinfo;
inp->inp_socket = so;
#if CONFIG_MACF_NET
mac_error = mac_inpcb_label_init(inp, M_WAITOK);
if (mac_error != 0) {
if (so->cached_in_sock_layer == 0)
zfree(pcbinfo->ipi_zone, inp);
return (mac_error);
}
mac_inpcb_label_associate(so, inp);
#endif
so->so_pcb = (caddr_t)inp;
if (so->so_proto->pr_flags & PR_PCBLOCK) {
inp->inpcb_mtx = lck_mtx_alloc_init(pcbinfo->mtx_grp, pcbinfo->mtx_attr);
if (inp->inpcb_mtx == NULL) {
printf("in_pcballoc: can't alloc mutex! so=%p\n", so);
return(ENOMEM);
}
}
#if IPSEC
#ifndef __APPLE__
if (ipsec_bypass == 0) {
error = ipsec_init_policy(so, &inp->inp_sp);
if (error != 0) {
zfree(pcbinfo->ipi_zone, inp);
return error;
}
}
#endif
#endif
#if INET6
if (INP_SOCKAF(so) == AF_INET6 && !ip6_mapped_addr_on)
inp->inp_flags |= IN6P_IPV6_V6ONLY;
#endif
#if INET6
if (ip6_auto_flowlabel)
inp->inp_flags |= IN6P_AUTOFLOWLABEL;
#endif
lck_rw_lock_exclusive(pcbinfo->mtx);
inp->inp_gencnt = ++pcbinfo->ipi_gencnt;
LIST_INSERT_HEAD(pcbinfo->listhead, inp, inp_list);
pcbinfo->ipi_count++;
lck_rw_done(pcbinfo->mtx);
return (0);
}
struct inpcb*
in_pcblookup_local_and_cleanup(
struct inpcbinfo *pcbinfo,
struct in_addr laddr,
u_int lport_arg,
int wild_okay)
{
struct inpcb *inp;
inp = in_pcblookup_local(pcbinfo, laddr, lport_arg, wild_okay);
if (inp && inp->inp_wantcnt == WNT_STOPUSING) {
struct socket *so = inp->inp_socket;
lck_mtx_lock(inp->inpcb_mtx);
if (so->so_usecount == 0) {
if (inp->inp_state != INPCB_STATE_DEAD)
in_pcbdetach(inp);
in_pcbdispose(inp);
inp = NULL;
}
else {
lck_mtx_unlock(inp->inpcb_mtx);
}
}
return inp;
}
#ifdef __APPLE_API_PRIVATE
static void
in_pcb_conflict_post_msg(u_int16_t port)
{
struct kev_msg ev_msg;
struct kev_in_portinuse in_portinuse;
in_portinuse.port = ntohs(port);
in_portinuse.req_pid = proc_selfpid();
ev_msg.vendor_code = KEV_VENDOR_APPLE;
ev_msg.kev_class = KEV_NETWORK_CLASS;
ev_msg.kev_subclass = KEV_INET_SUBCLASS;
ev_msg.event_code = KEV_INET_PORTINUSE;
ev_msg.dv[0].data_ptr = &in_portinuse;
ev_msg.dv[0].data_length = sizeof(struct kev_in_portinuse);
ev_msg.dv[1].data_length = 0;
kev_post_msg(&ev_msg);
}
#endif
int
in_pcbbind(struct inpcb *inp, struct sockaddr *nam, struct proc *p)
{
struct socket *so = inp->inp_socket;
unsigned short *lastport;
struct sockaddr_in *sin;
struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
u_short lport = 0, rand_port = 0;
int wild = 0, reuseport = (so->so_options & SO_REUSEPORT);
int error, randomport, conflict = 0;
if (TAILQ_EMPTY(&in_ifaddrhead))
return (EADDRNOTAVAIL);
if (inp->inp_lport || inp->inp_laddr.s_addr != INADDR_ANY)
return (EINVAL);
if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0)
wild = 1;
socket_unlock(so, 0);
lck_rw_lock_exclusive(pcbinfo->mtx);
if (nam) {
sin = (struct sockaddr_in *)nam;
if (nam->sa_len != sizeof (*sin)) {
lck_rw_done(pcbinfo->mtx);
socket_lock(so, 0);
return (EINVAL);
}
#ifdef notdef
if (sin->sin_family != AF_INET) {
lck_rw_done(pcbinfo->mtx);
socket_lock(so, 0);
return (EAFNOSUPPORT);
}
#endif
lport = sin->sin_port;
if (IN_MULTICAST(ntohl(sin->sin_addr.s_addr))) {
if (so->so_options & SO_REUSEADDR)
reuseport = SO_REUSEADDR|SO_REUSEPORT;
} else if (sin->sin_addr.s_addr != INADDR_ANY) {
struct ifaddr *ifa;
sin->sin_port = 0;
if ((ifa = ifa_ifwithaddr((struct sockaddr *)sin)) == 0) {
lck_rw_done(pcbinfo->mtx);
socket_lock(so, 0);
return (EADDRNOTAVAIL);
}
else {
ifafree(ifa);
}
}
if (lport) {
struct inpcb *t;
#if !CONFIG_EMBEDDED
if (ntohs(lport) < IPPORT_RESERVED && proc_suser(p)) {
lck_rw_done(pcbinfo->mtx);
socket_lock(so, 0);
return (EACCES);
}
#endif
if (so->so_uid &&
!IN_MULTICAST(ntohl(sin->sin_addr.s_addr))) {
t = in_pcblookup_local_and_cleanup(inp->inp_pcbinfo,
sin->sin_addr, lport, INPLOOKUP_WILDCARD);
if (t &&
(ntohl(sin->sin_addr.s_addr) != INADDR_ANY ||
ntohl(t->inp_laddr.s_addr) != INADDR_ANY ||
(t->inp_socket->so_options &
SO_REUSEPORT) == 0) &&
(so->so_uid != t->inp_socket->so_uid) &&
((t->inp_socket->so_flags & SOF_REUSESHAREUID) == 0)) {
#if INET6
if (ntohl(sin->sin_addr.s_addr) !=
INADDR_ANY ||
ntohl(t->inp_laddr.s_addr) !=
INADDR_ANY ||
INP_SOCKAF(so) ==
INP_SOCKAF(t->inp_socket))
#endif
{
#ifdef __APPLE_API_PRIVATE
if ((t->inp_socket->so_flags & SOF_NOTIFYCONFLICT) && ((so->so_flags & SOF_NOTIFYCONFLICT) == 0))
conflict = 1;
lck_rw_done(pcbinfo->mtx);
if (conflict)
in_pcb_conflict_post_msg(lport);
#else
lck_rw_done(pcbinfo->mtx);
#endif
socket_lock(so, 0);
return (EADDRINUSE);
}
}
}
t = in_pcblookup_local_and_cleanup(pcbinfo, sin->sin_addr,
lport, wild);
if (t &&
(reuseport & t->inp_socket->so_options) == 0) {
#if INET6
if (ip6_mapped_addr_on == 0 ||
ntohl(sin->sin_addr.s_addr) !=
INADDR_ANY ||
ntohl(t->inp_laddr.s_addr) !=
INADDR_ANY ||
INP_SOCKAF(so) ==
INP_SOCKAF(t->inp_socket))
#endif
{
#ifdef __APPLE_API_PRIVATE
if ((t->inp_socket->so_flags & SOF_NOTIFYCONFLICT) && ((so->so_flags & SOF_NOTIFYCONFLICT) == 0))
conflict = 1;
lck_rw_done(pcbinfo->mtx);
if (conflict)
in_pcb_conflict_post_msg(lport);
#else
lck_rw_done(pcbinfo->mtx);
#endif
socket_lock(so, 0);
return (EADDRINUSE);
}
}
}
inp->inp_laddr = sin->sin_addr;
}
if (lport == 0) {
u_short first, last;
int count;
randomport = (so->so_flags & SOF_BINDRANDOMPORT) ||
(so->so_type == SOCK_STREAM ? tcp_use_randomport : udp_use_randomport);
inp->inp_flags |= INP_ANONPORT;
if (inp->inp_flags & INP_HIGHPORT) {
first = ipport_hifirstauto;
last = ipport_hilastauto;
lastport = &pcbinfo->lasthi;
} else if (inp->inp_flags & INP_LOWPORT) {
if ((error = proc_suser(p)) != 0) {
lck_rw_done(pcbinfo->mtx);
socket_lock(so, 0);
return error;
}
first = ipport_lowfirstauto;
last = ipport_lowlastauto;
lastport = &pcbinfo->lastlow;
} else {
first = ipport_firstauto;
last = ipport_lastauto;
lastport = &pcbinfo->lastport;
}
if (first == last)
randomport = 0;
if (first > last) {
if (randomport) {
read_random(&rand_port, sizeof(rand_port));
*lastport = first - (rand_port % (first - last));
}
count = first - last;
do {
if (count-- < 0) {
lck_rw_done(pcbinfo->mtx);
socket_lock(so, 0);
inp->inp_laddr.s_addr = INADDR_ANY;
return (EADDRNOTAVAIL);
}
--*lastport;
if (*lastport > first || *lastport < last)
*lastport = first;
lport = htons(*lastport);
} while (in_pcblookup_local_and_cleanup(pcbinfo,
inp->inp_laddr, lport, wild));
} else {
if (randomport) {
read_random(&rand_port, sizeof(rand_port));
*lastport = first + (rand_port % (first - last));
}
count = last - first;
do {
if (count-- < 0) {
lck_rw_done(pcbinfo->mtx);
socket_lock(so, 0);
inp->inp_laddr.s_addr = INADDR_ANY;
return (EADDRNOTAVAIL);
}
++*lastport;
if (*lastport < first || *lastport > last)
*lastport = first;
lport = htons(*lastport);
} while (in_pcblookup_local_and_cleanup(pcbinfo,
inp->inp_laddr, lport, wild));
}
}
socket_lock(so, 0);
inp->inp_lport = lport;
if (in_pcbinshash(inp, 1) != 0) {
inp->inp_laddr.s_addr = INADDR_ANY;
inp->inp_lport = 0;
lck_rw_done(pcbinfo->mtx);
return (EAGAIN);
}
lck_rw_done(pcbinfo->mtx);
sflt_notify(so, sock_evt_bound, NULL);
return (0);
}
int
in_pcbladdr(struct inpcb *inp, struct sockaddr *nam,
struct sockaddr_in **plocal_sin)
{
struct in_ifaddr *ia;
struct sockaddr_in *sin = (struct sockaddr_in *)nam;
if (nam->sa_len != sizeof (*sin))
return (EINVAL);
if (sin->sin_family != AF_INET)
return (EAFNOSUPPORT);
if (sin->sin_port == 0)
return (EADDRNOTAVAIL);
lck_rw_lock_shared(in_ifaddr_rwlock);
if (!TAILQ_EMPTY(&in_ifaddrhead)) {
#define satosin(sa) ((struct sockaddr_in *)(sa))
#define sintosa(sin) ((struct sockaddr *)(sin))
#define ifatoia(ifa) ((struct in_ifaddr *)(ifa))
if (sin->sin_addr.s_addr == INADDR_ANY)
sin->sin_addr = IA_SIN(TAILQ_FIRST(&in_ifaddrhead))->sin_addr;
else if (sin->sin_addr.s_addr == (u_int32_t)INADDR_BROADCAST &&
(TAILQ_FIRST(&in_ifaddrhead)->ia_ifp->if_flags & IFF_BROADCAST))
sin->sin_addr = satosin(&TAILQ_FIRST(&in_ifaddrhead)->ia_broadaddr)->sin_addr;
}
lck_rw_done(in_ifaddr_rwlock);
if (inp->inp_laddr.s_addr == INADDR_ANY) {
struct route *ro;
unsigned int ifscope;
ia = (struct in_ifaddr *)0;
ifscope = (inp->inp_flags & INP_BOUND_IF) ?
inp->inp_boundif : IFSCOPE_NONE;
ro = &inp->inp_route;
if (ro->ro_rt != NULL)
RT_LOCK_SPIN(ro->ro_rt);
if (ro->ro_rt && (ro->ro_dst.sa_family != AF_INET ||
satosin(&ro->ro_dst)->sin_addr.s_addr !=
sin->sin_addr.s_addr ||
inp->inp_socket->so_options & SO_DONTROUTE ||
ro->ro_rt->generation_id != route_generation)) {
RT_UNLOCK(ro->ro_rt);
rtfree(ro->ro_rt);
ro->ro_rt = NULL;
}
if ((inp->inp_socket->so_options & SO_DONTROUTE) == 0 &&
(ro->ro_rt == NULL || ro->ro_rt->rt_ifp == NULL)) {
if (ro->ro_rt != NULL)
RT_UNLOCK(ro->ro_rt);
bzero(&ro->ro_dst, sizeof(struct sockaddr_in));
ro->ro_dst.sa_family = AF_INET;
ro->ro_dst.sa_len = sizeof(struct sockaddr_in);
((struct sockaddr_in *) &ro->ro_dst)->sin_addr =
sin->sin_addr;
rtalloc_scoped_ign(ro, 0, ifscope);
if (ro->ro_rt != NULL)
RT_LOCK_SPIN(ro->ro_rt);
}
if (ro->ro_rt != NULL) {
RT_LOCK_ASSERT_HELD(ro->ro_rt);
if (!(ro->ro_rt->rt_ifp->if_flags & IFF_LOOPBACK)) {
ia = ifatoia(ro->ro_rt->rt_ifa);
if (ia)
ifaref(&ia->ia_ifa);
}
RT_UNLOCK(ro->ro_rt);
}
if (ia == 0) {
u_short fport = sin->sin_port;
sin->sin_port = 0;
ia = ifatoia(ifa_ifwithdstaddr(sintosa(sin)));
if (ia == 0) {
ia = ifatoia(ifa_ifwithnet_scoped(sintosa(sin),
ifscope));
}
sin->sin_port = fport;
if (ia == 0) {
lck_rw_lock_shared(in_ifaddr_rwlock);
ia = TAILQ_FIRST(&in_ifaddrhead);
if (ia)
ifaref(&ia->ia_ifa);
lck_rw_done(in_ifaddr_rwlock);
}
if (ia == 0)
return (EADDRNOTAVAIL);
}
if (IN_MULTICAST(ntohl(sin->sin_addr.s_addr)) &&
inp->inp_moptions != NULL) {
struct ip_moptions *imo;
struct ifnet *ifp;
imo = inp->inp_moptions;
if (imo->imo_multicast_ifp != NULL && (ia == NULL ||
ia->ia_ifp != imo->imo_multicast_ifp)) {
ifp = imo->imo_multicast_ifp;
if (ia)
ifafree(&ia->ia_ifa);
lck_rw_lock_shared(in_ifaddr_rwlock);
TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) {
if (ia->ia_ifp == ifp)
break;
}
if (ia)
ifaref(&ia->ia_ifa);
lck_rw_done(in_ifaddr_rwlock);
if (ia == 0)
return (EADDRNOTAVAIL);
}
}
*plocal_sin = &ia->ia_addr;
ifafree(&ia->ia_ifa);
}
return(0);
}
int
in_pcbconnect(struct inpcb *inp, struct sockaddr *nam, struct proc *p)
{
struct sockaddr_in *ifaddr;
struct sockaddr_in *sin = (struct sockaddr_in *)nam;
struct inpcb *pcb;
int error;
if ((error = in_pcbladdr(inp, nam, &ifaddr)) != 0)
return(error);
socket_unlock(inp->inp_socket, 0);
pcb = in_pcblookup_hash(inp->inp_pcbinfo, sin->sin_addr, sin->sin_port,
inp->inp_laddr.s_addr ? inp->inp_laddr : ifaddr->sin_addr,
inp->inp_lport, 0, NULL);
socket_lock(inp->inp_socket, 0);
if (pcb != NULL) {
in_pcb_checkstate(pcb, WNT_RELEASE, pcb == inp ? 1 : 0);
return (EADDRINUSE);
}
if (inp->inp_laddr.s_addr == INADDR_ANY) {
if (inp->inp_lport == 0) {
error = in_pcbbind(inp, (struct sockaddr *)0, p);
if (error)
return (error);
}
if (!lck_rw_try_lock_exclusive(inp->inp_pcbinfo->mtx)) {
socket_unlock(inp->inp_socket, 0);
lck_rw_lock_exclusive(inp->inp_pcbinfo->mtx);
socket_lock(inp->inp_socket, 0);
}
inp->inp_laddr = ifaddr->sin_addr;
inp->inp_flags |= INP_INADDR_ANY;
}
else {
if (!lck_rw_try_lock_exclusive(inp->inp_pcbinfo->mtx)) {
socket_unlock(inp->inp_socket, 0);
lck_rw_lock_exclusive(inp->inp_pcbinfo->mtx);
socket_lock(inp->inp_socket, 0);
}
}
inp->inp_faddr = sin->sin_addr;
inp->inp_fport = sin->sin_port;
in_pcbrehash(inp);
lck_rw_done(inp->inp_pcbinfo->mtx);
return (0);
}
void
in_pcbdisconnect(struct inpcb *inp)
{
inp->inp_faddr.s_addr = INADDR_ANY;
inp->inp_fport = 0;
if (!lck_rw_try_lock_exclusive(inp->inp_pcbinfo->mtx)) {
socket_unlock(inp->inp_socket, 0);
lck_rw_lock_exclusive(inp->inp_pcbinfo->mtx);
socket_lock(inp->inp_socket, 0);
}
in_pcbrehash(inp);
lck_rw_done(inp->inp_pcbinfo->mtx);
if (inp->inp_socket->so_state & SS_NOFDREF)
in_pcbdetach(inp);
}
void
in_pcbdetach(struct inpcb *inp)
{
struct socket *so = inp->inp_socket;
if (so->so_pcb == 0) {
panic("in_pcbdetach: inp=%p so=%p proto=%d so_pcb is null!\n",
inp, so, so->so_proto->pr_protocol);
}
#if IPSEC
if (ipsec_bypass == 0) {
ipsec4_delete_pcbpolicy(inp);
}
#endif
if (in_pcb_checkstate(inp, WNT_STOPUSING, 1) != WNT_STOPUSING)
panic("in_pcbdetach so=%p prot=%x couldn't set to STOPUSING\n", so, so->so_proto->pr_protocol);
#if TEMPDEBUG
if (so->cached_in_sock_layer)
printf("in_pcbdetach for cached socket %x flags=%x\n", so, so->so_flags);
else
printf("in_pcbdetach for allocated socket %x flags=%x\n", so, so->so_flags);
#endif
if ((so->so_flags & SOF_PCBCLEARING) == 0) {
struct rtentry *rt;
inp->inp_vflag = 0;
if (inp->inp_options)
(void)m_free(inp->inp_options);
if ((rt = inp->inp_route.ro_rt) != NULL) {
inp->inp_route.ro_rt = NULL;
rtfree(rt);
}
ip_freemoptions(inp->inp_moptions);
inp->inp_moptions = NULL;
sofreelastref(so, 0);
inp->inp_state = INPCB_STATE_DEAD;
so->so_flags |= SOF_PCBCLEARING;
}
}
void
in_pcbdispose(struct inpcb *inp)
{
struct socket *so = inp->inp_socket;
struct inpcbinfo *ipi = inp->inp_pcbinfo;
#if TEMPDEBUG
if (inp->inp_state != INPCB_STATE_DEAD) {
printf("in_pcbdispose: not dead yet? so=%p\n", so);
}
#endif
if (so && so->so_usecount != 0)
panic("in_pcbdispose: use count=%x so=%p\n", so->so_usecount, so);
lck_rw_assert(ipi->mtx, LCK_RW_ASSERT_EXCLUSIVE);
inp->inp_gencnt = ++ipi->ipi_gencnt;
in_pcbremlists(inp);
if (so) {
if (so->so_proto->pr_flags & PR_PCBLOCK) {
sofreelastref(so, 0);
if (so->so_rcv.sb_cc || so->so_snd.sb_cc) {
#if TEMPDEBUG
printf("in_pcbdispose sb not cleaned up so=%p rc_cci=%x snd_cc=%x\n",
so, so->so_rcv.sb_cc, so->so_snd.sb_cc);
#endif
sbrelease(&so->so_rcv);
sbrelease(&so->so_snd);
}
if (so->so_head != NULL)
panic("in_pcbdispose, so=%p head still exist\n", so);
lck_mtx_unlock(inp->inpcb_mtx);
lck_mtx_free(inp->inpcb_mtx, ipi->mtx_grp);
}
so->so_flags |= SOF_PCBCLEARING;
so->so_saved_pcb = (caddr_t) inp;
so->so_pcb = 0;
inp->inp_socket = 0;
#if CONFIG_MACF_NET
mac_inpcb_label_destroy(inp);
#endif
if (inp->inp_route.ro_rt != NULL) {
rtfree(inp->inp_route.ro_rt);
inp->inp_route.ro_rt = NULL;
}
if (so->cached_in_sock_layer == 0) {
zfree(ipi->ipi_zone, inp);
}
sodealloc(so);
}
#if TEMPDEBUG
else
printf("in_pcbdispose: no socket for inp=%p\n", inp);
#endif
}
int
in_setsockaddr(struct socket *so, struct sockaddr **nam)
{
struct inpcb *inp;
struct sockaddr_in *sin;
MALLOC(sin, struct sockaddr_in *, sizeof *sin, M_SONAME, M_WAITOK);
if (sin == NULL)
return ENOBUFS;
bzero(sin, sizeof *sin);
sin->sin_family = AF_INET;
sin->sin_len = sizeof(*sin);
inp = sotoinpcb(so);
if (!inp) {
FREE(sin, M_SONAME);
return ECONNRESET;
}
sin->sin_port = inp->inp_lport;
sin->sin_addr = inp->inp_laddr;
*nam = (struct sockaddr *)sin;
return 0;
}
int
in_setpeeraddr(struct socket *so, struct sockaddr **nam)
{
struct inpcb *inp;
struct sockaddr_in *sin;
MALLOC(sin, struct sockaddr_in *, sizeof *sin, M_SONAME, M_WAITOK);
if (sin == NULL)
return ENOBUFS;
bzero((caddr_t)sin, sizeof (*sin));
sin->sin_family = AF_INET;
sin->sin_len = sizeof(*sin);
inp = sotoinpcb(so);
if (!inp) {
FREE(sin, M_SONAME);
return ECONNRESET;
}
sin->sin_port = inp->inp_fport;
sin->sin_addr = inp->inp_faddr;
*nam = (struct sockaddr *)sin;
return 0;
}
void
in_pcbnotifyall(struct inpcbinfo *pcbinfo, struct in_addr faddr,
int errno, void (*notify)(struct inpcb *, int))
{
struct inpcb *inp;
lck_rw_lock_shared(pcbinfo->mtx);
LIST_FOREACH(inp, pcbinfo->listhead, inp_list) {
#if INET6
if ((inp->inp_vflag & INP_IPV4) == 0)
continue;
#endif
if (inp->inp_faddr.s_addr != faddr.s_addr ||
inp->inp_socket == NULL)
continue;
if (in_pcb_checkstate(inp, WNT_ACQUIRE, 0) == WNT_STOPUSING)
continue;
socket_lock(inp->inp_socket, 1);
(*notify)(inp, errno);
(void)in_pcb_checkstate(inp, WNT_RELEASE, 1);
socket_unlock(inp->inp_socket, 1);
}
lck_rw_done(pcbinfo->mtx);
}
void
in_losing(struct inpcb *inp)
{
struct rtentry *rt;
struct rt_addrinfo info;
if ((rt = inp->inp_route.ro_rt) != NULL) {
struct in_ifaddr *ia;
bzero((caddr_t)&info, sizeof(info));
RT_LOCK(rt);
info.rti_info[RTAX_DST] =
(struct sockaddr *)&inp->inp_route.ro_dst;
info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
info.rti_info[RTAX_NETMASK] = rt_mask(rt);
rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0);
if (rt->rt_flags & RTF_DYNAMIC) {
rt->rt_flags |= RTF_CONDEMNED;
RT_UNLOCK(rt);
(void) rtrequest(RTM_DELETE, rt_key(rt),
rt->rt_gateway, rt_mask(rt), rt->rt_flags,
(struct rtentry **)0);
} else {
RT_UNLOCK(rt);
}
if ((ia = ifa_foraddr(inp->inp_laddr.s_addr)) != NULL) {
inp->inp_route.ro_rt = NULL;
rtfree(rt);
ifafree(&ia->ia_ifa);
}
}
}
void
in_rtchange(struct inpcb *inp, __unused int errno)
{
struct rtentry *rt;
if ((rt = inp->inp_route.ro_rt) != NULL) {
struct in_ifaddr *ia;
if ((ia = ifa_foraddr(inp->inp_laddr.s_addr)) == NULL) {
return;
}
ifafree(&ia->ia_ifa);
rtfree(rt);
inp->inp_route.ro_rt = NULL;
}
}
struct inpcb *
in_pcblookup_local(struct inpcbinfo *pcbinfo, struct in_addr laddr,
unsigned int lport_arg, int wild_okay)
{
struct inpcb *inp;
int matchwild = 3, wildcard;
u_short lport = lport_arg;
KERNEL_DEBUG(DBG_FNC_PCB_LOOKUP | DBG_FUNC_START, 0,0,0,0,0);
if (!wild_okay) {
struct inpcbhead *head;
head = &pcbinfo->hashbase[INP_PCBHASH(INADDR_ANY, lport, 0, pcbinfo->hashmask)];
LIST_FOREACH(inp, head, inp_hash) {
#if INET6
if ((inp->inp_vflag & INP_IPV4) == 0)
continue;
#endif
if (inp->inp_faddr.s_addr == INADDR_ANY &&
inp->inp_laddr.s_addr == laddr.s_addr &&
inp->inp_lport == lport) {
return (inp);
}
}
KERNEL_DEBUG(DBG_FNC_PCB_LOOKUP | DBG_FUNC_END, 0,0,0,0,0);
return (NULL);
} else {
struct inpcbporthead *porthash;
struct inpcbport *phd;
struct inpcb *match = NULL;
porthash = &pcbinfo->porthashbase[INP_PCBPORTHASH(lport,
pcbinfo->porthashmask)];
LIST_FOREACH(phd, porthash, phd_hash) {
if (phd->phd_port == lport)
break;
}
if (phd != NULL) {
LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) {
wildcard = 0;
#if INET6
if ((inp->inp_vflag & INP_IPV4) == 0)
continue;
#endif
if (inp->inp_faddr.s_addr != INADDR_ANY)
wildcard++;
if (inp->inp_laddr.s_addr != INADDR_ANY) {
if (laddr.s_addr == INADDR_ANY)
wildcard++;
else if (inp->inp_laddr.s_addr != laddr.s_addr)
continue;
} else {
if (laddr.s_addr != INADDR_ANY)
wildcard++;
}
if (wildcard < matchwild) {
match = inp;
matchwild = wildcard;
if (matchwild == 0) {
break;
}
}
}
}
KERNEL_DEBUG(DBG_FNC_PCB_LOOKUP | DBG_FUNC_END, match,0,0,0,0);
return (match);
}
}
struct inpcb *
in_pcblookup_hash(
struct inpcbinfo *pcbinfo,
struct in_addr faddr,
u_int fport_arg,
struct in_addr laddr,
u_int lport_arg,
int wildcard,
__unused struct ifnet *ifp)
{
struct inpcbhead *head;
struct inpcb *inp;
u_short fport = fport_arg, lport = lport_arg;
lck_rw_lock_shared(pcbinfo->mtx);
head = &pcbinfo->hashbase[INP_PCBHASH(faddr.s_addr, lport, fport, pcbinfo->hashmask)];
LIST_FOREACH(inp, head, inp_hash) {
#if INET6
if ((inp->inp_vflag & INP_IPV4) == 0)
continue;
#endif
if (inp->inp_faddr.s_addr == faddr.s_addr &&
inp->inp_laddr.s_addr == laddr.s_addr &&
inp->inp_fport == fport &&
inp->inp_lport == lport) {
if (in_pcb_checkstate(inp, WNT_ACQUIRE, 0) != WNT_STOPUSING) {
lck_rw_done(pcbinfo->mtx);
return (inp);
}
else {
lck_rw_done(pcbinfo->mtx);
return(NULL);
}
}
}
if (wildcard) {
struct inpcb *local_wild = NULL;
#if INET6
struct inpcb *local_wild_mapped = NULL;
#endif
head = &pcbinfo->hashbase[INP_PCBHASH(INADDR_ANY, lport, 0, pcbinfo->hashmask)];
LIST_FOREACH(inp, head, inp_hash) {
#if INET6
if ((inp->inp_vflag & INP_IPV4) == 0)
continue;
#endif
if (inp->inp_faddr.s_addr == INADDR_ANY &&
inp->inp_lport == lport) {
#if defined(NFAITH) && NFAITH > 0
if (ifp && ifp->if_type == IFT_FAITH &&
(inp->inp_flags & INP_FAITH) == 0)
continue;
#endif
if (inp->inp_laddr.s_addr == laddr.s_addr) {
if (in_pcb_checkstate(inp, WNT_ACQUIRE, 0) != WNT_STOPUSING) {
lck_rw_done(pcbinfo->mtx);
return (inp);
}
else {
lck_rw_done(pcbinfo->mtx);
return(NULL);
}
}
else if (inp->inp_laddr.s_addr == INADDR_ANY) {
#if INET6
if (INP_CHECK_SOCKAF(inp->inp_socket,
AF_INET6))
local_wild_mapped = inp;
else
#endif
local_wild = inp;
}
}
}
if (local_wild == NULL) {
#if INET6
if (local_wild_mapped != NULL) {
if (in_pcb_checkstate(local_wild_mapped, WNT_ACQUIRE, 0) != WNT_STOPUSING) {
lck_rw_done(pcbinfo->mtx);
return (local_wild_mapped);
}
else {
lck_rw_done(pcbinfo->mtx);
return(NULL);
}
}
#endif
lck_rw_done(pcbinfo->mtx);
return (NULL);
}
if (in_pcb_checkstate(local_wild, WNT_ACQUIRE, 0) != WNT_STOPUSING) {
lck_rw_done(pcbinfo->mtx);
return (local_wild);
}
else {
lck_rw_done(pcbinfo->mtx);
return(NULL);
}
}
lck_rw_done(pcbinfo->mtx);
return (NULL);
}
int
in_pcbinshash(struct inpcb *inp, int locked)
{
struct inpcbhead *pcbhash;
struct inpcbporthead *pcbporthash;
struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
struct inpcbport *phd;
u_int32_t hashkey_faddr;
if (!locked) {
if (!lck_rw_try_lock_exclusive(pcbinfo->mtx)) {
socket_unlock(inp->inp_socket, 0);
lck_rw_lock_exclusive(pcbinfo->mtx);
socket_lock(inp->inp_socket, 0);
}
}
#if INET6
if (inp->inp_vflag & INP_IPV6)
hashkey_faddr = inp->in6p_faddr.s6_addr32[3] ;
else
#endif
hashkey_faddr = inp->inp_faddr.s_addr;
inp->hash_element = INP_PCBHASH(hashkey_faddr, inp->inp_lport, inp->inp_fport, pcbinfo->hashmask);
pcbhash = &pcbinfo->hashbase[inp->hash_element];
pcbporthash = &pcbinfo->porthashbase[INP_PCBPORTHASH(inp->inp_lport,
pcbinfo->porthashmask)];
LIST_FOREACH(phd, pcbporthash, phd_hash) {
if (phd->phd_port == inp->inp_lport)
break;
}
if (phd == NULL) {
MALLOC(phd, struct inpcbport *, sizeof(struct inpcbport), M_PCB, M_WAITOK);
if (phd == NULL) {
if (!locked)
lck_rw_done(pcbinfo->mtx);
return (ENOBUFS);
}
phd->phd_port = inp->inp_lport;
LIST_INIT(&phd->phd_pcblist);
LIST_INSERT_HEAD(pcbporthash, phd, phd_hash);
}
inp->inp_phd = phd;
LIST_INSERT_HEAD(&phd->phd_pcblist, inp, inp_portlist);
LIST_INSERT_HEAD(pcbhash, inp, inp_hash);
if (!locked)
lck_rw_done(pcbinfo->mtx);
return (0);
}
void
in_pcbrehash(struct inpcb *inp)
{
struct inpcbhead *head;
u_int32_t hashkey_faddr;
#if INET6
if (inp->inp_vflag & INP_IPV6)
hashkey_faddr = inp->in6p_faddr.s6_addr32[3] ;
else
#endif
hashkey_faddr = inp->inp_faddr.s_addr;
inp->hash_element = INP_PCBHASH(hashkey_faddr, inp->inp_lport,
inp->inp_fport, inp->inp_pcbinfo->hashmask);
head = &inp->inp_pcbinfo->hashbase[inp->hash_element];
LIST_REMOVE(inp, inp_hash);
LIST_INSERT_HEAD(head, inp, inp_hash);
}
void
in_pcbremlists(struct inpcb *inp)
{
inp->inp_gencnt = ++inp->inp_pcbinfo->ipi_gencnt;
if (inp->inp_lport) {
struct inpcbport *phd = inp->inp_phd;
LIST_REMOVE(inp, inp_hash);
LIST_REMOVE(inp, inp_portlist);
if (phd != NULL && (LIST_FIRST(&phd->phd_pcblist) == NULL)) {
LIST_REMOVE(phd, phd_hash);
FREE(phd, M_PCB);
}
}
LIST_REMOVE(inp, inp_list);
inp->inp_pcbinfo->ipi_count--;
}
int
in_pcb_checkstate(struct inpcb *pcb, int mode, int locked)
{
volatile UInt32 *wantcnt = (volatile UInt32 *)&pcb->inp_wantcnt;
UInt32 origwant;
UInt32 newwant;
switch (mode) {
case WNT_STOPUSING:
if (locked == 0)
socket_lock(pcb->inp_socket, 1);
pcb->inp_state = INPCB_STATE_DEAD;
stopusing:
if (pcb->inp_socket->so_usecount < 0)
panic("in_pcb_checkstate STOP pcb=%p so=%p usecount is negative\n", pcb, pcb->inp_socket);
if (locked == 0)
socket_unlock(pcb->inp_socket, 1);
origwant = *wantcnt;
if ((UInt16) origwant == 0xffff )
return (WNT_STOPUSING);
newwant = 0xffff;
if ((UInt16) origwant == 0) {
OSCompareAndSwap(origwant, newwant, wantcnt) ;
}
return (WNT_STOPUSING);
break;
case WNT_ACQUIRE:
do {
origwant = *wantcnt;
if ((UInt16) origwant == 0xffff ) {
return (WNT_STOPUSING);
}
newwant = origwant + 1;
} while (!OSCompareAndSwap(origwant, newwant, wantcnt));
return (WNT_ACQUIRE);
break;
case WNT_RELEASE:
if (locked == 0)
socket_lock(pcb->inp_socket, 1);
do {
origwant = *wantcnt;
if ((UInt16) origwant == 0x0 )
panic("in_pcb_checkstate pcb=%p release with zero count", pcb);
if ((UInt16) origwant == 0xffff ) {
#if TEMPDEBUG
printf("in_pcb_checkstate: REL PCB was STOPUSING while release. odd pcb=%p\n", pcb);
#endif
if (locked == 0)
socket_unlock(pcb->inp_socket, 1);
return (WNT_STOPUSING);
}
newwant = origwant - 1;
} while (!OSCompareAndSwap(origwant, newwant, wantcnt));
if (pcb->inp_state == INPCB_STATE_DEAD)
goto stopusing;
if (pcb->inp_socket->so_usecount < 0)
panic("in_pcb_checkstate RELEASE pcb=%p so=%p usecount is negative\n", pcb, pcb->inp_socket);
if (locked == 0)
socket_unlock(pcb->inp_socket, 1);
return (WNT_RELEASE);
break;
default:
panic("in_pcb_checkstate: so=%p not a valid state =%x\n", pcb->inp_socket, mode);
}
return (mode);
}
void
inpcb_to_compat(
struct inpcb *inp,
struct inpcb_compat *inp_compat)
{
bzero(inp_compat, sizeof(*inp_compat));
inp_compat->inp_fport = inp->inp_fport;
inp_compat->inp_lport = inp->inp_lport;
inp_compat->nat_owner = inp->nat_owner;
inp_compat->nat_cookie = inp->nat_cookie;
inp_compat->inp_gencnt = inp->inp_gencnt;
inp_compat->inp_flags = inp->inp_flags;
inp_compat->inp_flow = inp->inp_flow;
inp_compat->inp_vflag = inp->inp_vflag;
inp_compat->inp_ip_ttl = inp->inp_ip_ttl;
inp_compat->inp_ip_p = inp->inp_ip_p;
inp_compat->inp_dependfaddr.inp6_foreign = inp->inp_dependfaddr.inp6_foreign;
inp_compat->inp_dependladdr.inp6_local = inp->inp_dependladdr.inp6_local;
inp_compat->inp_depend4.inp4_ip_tos = inp->inp_depend4.inp4_ip_tos;
inp_compat->inp_depend6.inp6_hlim = inp->inp_depend6.inp6_hlim;
inp_compat->inp_depend6.inp6_cksum = inp->inp_depend6.inp6_cksum;
inp_compat->inp_depend6.inp6_ifindex = inp->inp_depend6.inp6_ifindex;
inp_compat->inp_depend6.inp6_hops = inp->inp_depend6.inp6_hops;
}
#if !CONFIG_EMBEDDED
void
inpcb_to_xinpcb64(
struct inpcb *inp,
struct xinpcb64 *xinp)
{
xinp->inp_fport = inp->inp_fport;
xinp->inp_lport = inp->inp_lport;
xinp->inp_gencnt = inp->inp_gencnt;
xinp->inp_flags = inp->inp_flags;
xinp->inp_flow = inp->inp_flow;
xinp->inp_vflag = inp->inp_vflag;
xinp->inp_ip_ttl = inp->inp_ip_ttl;
xinp->inp_ip_p = inp->inp_ip_p;
xinp->inp_dependfaddr.inp6_foreign = inp->inp_dependfaddr.inp6_foreign;
xinp->inp_dependladdr.inp6_local = inp->inp_dependladdr.inp6_local;
xinp->inp_depend4.inp4_ip_tos = inp->inp_depend4.inp4_ip_tos;
xinp->inp_depend6.inp6_hlim = inp->inp_depend6.inp6_hlim;
xinp->inp_depend6.inp6_cksum = inp->inp_depend6.inp6_cksum;
xinp->inp_depend6.inp6_ifindex = inp->inp_depend6.inp6_ifindex;
xinp->inp_depend6.inp6_hops = inp->inp_depend6.inp6_hops;
}
#endif
void
inp_route_copyout(struct inpcb *inp, struct route *dst)
{
struct route *src = &inp->inp_route;
lck_mtx_assert(inp->inpcb_mtx, LCK_MTX_ASSERT_OWNED);
if (src->ro_rt != NULL && rt_key(src->ro_rt)->sa_family != AF_INET) {
rtfree(src->ro_rt);
src->ro_rt = NULL;
}
bcopy(src, dst, sizeof (*dst));
if (dst->ro_rt != NULL)
RT_ADDREF(dst->ro_rt);
}
void
inp_route_copyin(struct inpcb *inp, struct route *src)
{
struct route *dst = &inp->inp_route;
lck_mtx_assert(inp->inpcb_mtx, LCK_MTX_ASSERT_OWNED);
if (src->ro_rt != NULL && rt_key(src->ro_rt)->sa_family != AF_INET)
panic("%s: wrong or corrupted route: %p", __func__, src);
if (dst->ro_rt == NULL) {
bcopy(src, dst, sizeof (*dst));
} else if (src->ro_rt != NULL) {
if (dst->ro_rt == src->ro_rt) {
dst->ro_flags = src->ro_flags;
rtfree(src->ro_rt);
src->ro_rt = NULL;
} else {
rtfree(dst->ro_rt);
bcopy(src, dst, sizeof (*dst));
}
}
}