void
soisconnected(struct socket *so)
{
struct socket *head;
head = so->so_head;
KASSERT(solocked(so));
KASSERT(head == NULL || solocked2(so, head));
so->so_state &= ~(SS_ISCONNECTING | SS_ISDISCONNECTING);
so->so_state |= SS_ISCONNECTED;
if (head && so->so_onq == &head->so_q0) {
if ((so->so_options & SO_ACCEPTFILTER) == 0) {
soqremque(so, 0);
soqinsque(head, so, 1);
sorwakeup(head);
cv_broadcast(&head->so_cv);
} else {
so->so_upcall =
head->so_accf->so_accept_filter->accf_callback;
so->so_upcallarg = head->so_accf->so_accept_filter_arg;
so->so_rcv.sb_flags |= SB_UPCALL;
so->so_options &= ~SO_ACCEPTFILTER;
(*so->so_upcall)(so, so->so_upcallarg,
POLLIN|POLLRDNORM, M_DONTWAIT);
}
} else {
cv_broadcast(&so->so_cv);
sorwakeup(so);
sowwakeup(so);
}
}
int STKARGFUN
raw_input(struct mbuf *m0,
struct sockproto *proto,
struct sockaddr *src,
struct sockaddr *dst)
{
register struct rawcb *rp;
register struct mbuf *m = m0;
register int sockets = 0;
struct socket *last;
last = 0;
for (rp = rawcb.rcb_next; rp != &rawcb; rp = rp->rcb_next) {
if (rp->rcb_proto.sp_family != proto->sp_family)
continue;
if (rp->rcb_proto.sp_protocol &&
rp->rcb_proto.sp_protocol != proto->sp_protocol)
continue;
/*
* We assume the lower level routines have
* placed the address in a canonical format
* suitable for a structure comparison.
*
* Note that if the lengths are not the same
* the comparison will fail at the first byte.
*/
#define equal(a1, a2) \
(bcmp((caddr_t)(a1), (caddr_t)(a2), a1->sa_len) == 0)
if (rp->rcb_laddr && !equal(rp->rcb_laddr, dst))
continue;
if (rp->rcb_faddr && !equal(rp->rcb_faddr, src))
continue;
if (last) {
struct mbuf *n;
if (n = m_copy(m, 0, (int)M_COPYALL)) {
if (sbappendaddr(&last->so_rcv, src,
n, (struct mbuf *)0) == 0)
/* should notify about lost packet */
m_freem(n);
else {
sorwakeup(last);
sockets++;
}
}
}
last = rp->rcb_socket;
}
if (last) {
if (sbappendaddr(&last->so_rcv, src,
m, (struct mbuf *)0) == 0)
m_freem(m);
else {
sorwakeup(last);
sockets++;
}
} else
m_freem(m);
return (sockets);
}
/*
* Setup generic address and protocol structures
* for raw_input routine, then pass them along with
* mbuf chain.
*/
void
rip_input(struct mbuf *m, int iphlen)
{
struct ip *ip = mtod(m, struct ip *);
register struct inpcb *inp;
struct inpcb *last = 0;
struct mbuf *opts = 0;
ripsrc.sin_addr = ip->ip_src;
for (inp = ripcb.lh_first; inp != NULL; inp = inp->inp_list.le_next) {
if (inp->inp_ip_p && inp->inp_ip_p != ip->ip_p)
continue;
if (inp->inp_laddr.s_addr &&
inp->inp_laddr.s_addr != ip->ip_dst.s_addr)
continue;
if (inp->inp_faddr.s_addr &&
inp->inp_faddr.s_addr != ip->ip_src.s_addr)
continue;
if (last) {
struct mbuf *n = m_copy(m, 0, (int)M_COPYALL);
if (n) {
if (last->inp_flags & INP_CONTROLOPTS ||
last->inp_socket->so_options & SO_TIMESTAMP)
ip_savecontrol(last, &opts, ip, n);
if (sbappendaddr(&last->inp_socket->so_rcv,
(struct sockaddr *)&ripsrc, n,
opts) == 0) {
/* should notify about lost packet */
m_freem(n);
if (opts)
m_freem(opts);
} else
sorwakeup(last->inp_socket);
opts = 0;
}
}
last = inp;
}
if (last) {
if (last->inp_flags & INP_CONTROLOPTS ||
last->inp_socket->so_options & SO_TIMESTAMP)
ip_savecontrol(last, &opts, ip, m);
if (sbappendaddr(&last->inp_socket->so_rcv,
(struct sockaddr *)&ripsrc, m, opts) == 0) {
m_freem(m);
if (opts)
m_freem(opts);
} else
sorwakeup(last->inp_socket);
} else {
m_freem(m);
ipstat.ips_noproto++;
ipstat.ips_delivered--;
}
}
void
socantrcvmore(struct socket *so)
{
so->so_state |= SS_CANTRCVMORE;
sorwakeup(so);
}
/* -----------------------------------------------------------------------------
called from l2tp_rfc when data are present
----------------------------------------------------------------------------- */
int l2tp_input(void *data, mbuf_t m, struct sockaddr *from, int more)
{
struct socket *so = (struct socket *)data;
int err;
lck_mtx_assert(ppp_domain_mutex, LCK_MTX_ASSERT_OWNED);
if (so->so_tpcb) {
// we are hooked to ppp
return l2tp_wan_input(ALIGNED_CAST(struct ppp_link *)so->so_tpcb, m); // Wcast-align fix - we malloc so->so_tpcb
}
if (m) {
if (from == 0) {
// no from address, just free the buffer
mbuf_freem(m);
return 1;
}
if (sbappendaddr(&so->so_rcv, from, (struct mbuf *)m, 0, &err) == 0) {
//IOLog("l2tp_input no space, so = %p\n", so);
return 1;
}
}
if (!more)
sorwakeup(so);
return 0;
}
/*
* Notify a udp user of an asynchronous error;
* just wake up so that he can collect error status.
*/
void
udp_notify(struct inpcb *inp, int error)
{
inp->inp_socket->so_error = error;
sorwakeup(inp->inp_socket);
sowwakeup(inp->inp_socket);
}
/*
* subroutine of udp_input(), mainly for source code readability.
* caller must properly init udp_ip6 and udp_in6 beforehand.
*/
static void
udp_append(struct inpcb *last, struct ip *ip, struct mbuf *n, int off,
struct sockaddr_in *udp_in)
{
struct mbuf *opts = NULL;
int ret;
KASSERT(INP_ISIPV4(last), ("not inet inpcb"));
if (last->inp_flags & INP_CONTROLOPTS ||
last->inp_socket->so_options & SO_TIMESTAMP)
ip_savecontrol(last, &opts, ip, n);
m_adj(n, off);
lwkt_gettoken(&last->inp_socket->so_rcv.ssb_token);
ret = ssb_appendaddr(&last->inp_socket->so_rcv,
(struct sockaddr *)udp_in, n, opts);
lwkt_reltoken(&last->inp_socket->so_rcv.ssb_token);
if (ret == 0) {
m_freem(n);
if (opts)
m_freem(opts);
udp_stat.udps_fullsock++;
} else {
sorwakeup(last->inp_socket);
}
}
/*
* socantrcvmore(): indicates that no more data will be received and
* will normally be applied to the socket by a protocol when it detects
* that the peer will send no more data. Data queued for reading in
* the socket may yet be read.
*/
void
socantrcvmore(struct socket *so)
{
KASSERT(solocked(so));
so->so_state |= SS_CANTRCVMORE;
sorwakeup(so);
}
void
soisconnected(struct socket *so)
{
struct socket *head = so->so_head;
so->so_state &= ~(SS_ISCONNECTING|SS_ISDISCONNECTING);
so->so_state |= SS_ISCONNECTED;
if (head && soqremque(so, 0)) {
soqinsque(head, so, 1);
sorwakeup(head);
wakeup_one(&head->so_timeo);
} else {
wakeup(&so->so_timeo);
sorwakeup(so);
sowwakeup(so);
}
}
/*
* When an attempt at a new connection is noted on a socket
* which accepts connections, sonewconn is called. If the
* connection is possible (subject to space constraints, etc.)
* then we allocate a new structure, properly linked into the
* data structure of the original socket, and return this.
* Connstatus may be 0 or SS_ISCONNECTED.
*
* Must be called at splsoftnet()
*/
struct socket *
sonewconn(struct socket *head, int connstatus)
{
struct socket *so;
int soqueue = connstatus ? 1 : 0;
splsoftassert(IPL_SOFTNET);
if (mclpools[0].pr_nout > mclpools[0].pr_hardlimit * 95 / 100)
return (NULL);
if (head->so_qlen + head->so_q0len > head->so_qlimit * 3)
return (NULL);
so = pool_get(&socket_pool, PR_NOWAIT|PR_ZERO);
if (so == NULL)
return (NULL);
so->so_type = head->so_type;
so->so_options = head->so_options &~ SO_ACCEPTCONN;
so->so_linger = head->so_linger;
so->so_state = head->so_state | SS_NOFDREF;
so->so_proto = head->so_proto;
so->so_timeo = head->so_timeo;
so->so_pgid = head->so_pgid;
so->so_euid = head->so_euid;
so->so_ruid = head->so_ruid;
so->so_egid = head->so_egid;
so->so_rgid = head->so_rgid;
so->so_cpid = head->so_cpid;
so->so_siguid = head->so_siguid;
so->so_sigeuid = head->so_sigeuid;
/*
* Inherit watermarks but those may get clamped in low mem situations.
*/
if (soreserve(so, head->so_snd.sb_hiwat, head->so_rcv.sb_hiwat)) {
pool_put(&socket_pool, so);
return (NULL);
}
so->so_snd.sb_wat = head->so_snd.sb_wat;
so->so_snd.sb_lowat = head->so_snd.sb_lowat;
so->so_snd.sb_timeo = head->so_snd.sb_timeo;
so->so_rcv.sb_wat = head->so_rcv.sb_wat;
so->so_rcv.sb_lowat = head->so_rcv.sb_lowat;
so->so_rcv.sb_timeo = head->so_rcv.sb_timeo;
soqinsque(head, so, soqueue);
if ((*so->so_proto->pr_usrreq)(so, PRU_ATTACH, NULL, NULL, NULL,
curproc)) {
(void) soqremque(so, soqueue);
pool_put(&socket_pool, so);
return (NULL);
}
if (connstatus) {
sorwakeup(head);
wakeup(&head->so_timeo);
so->so_state |= connstatus;
}
return (so);
}
void
soisdisconnected(struct socket *so)
{
so->so_state &= ~(SS_ISCONNECTING|SS_ISCONNECTED|SS_ISDISCONNECTING);
so->so_state |= (SS_CANTRCVMORE|SS_CANTSENDMORE);
soconnwakeup(so);
sowwakeup(so);
sorwakeup(so);
}
void
soisdisconnecting(struct socket *so)
{
so->so_state &= ~SS_ISCONNECTING;
so->so_state |= (SS_ISDISCONNECTING|SS_CANTRCVMORE|SS_CANTSENDMORE);
wakeup(&so->so_timeo);
sowwakeup(so);
sorwakeup(so);
}
/*
* natmintr: interrupt
*
* Note: we expect a socket pointer in rcvif rather than an interface
* pointer. We can get the interface pointer from the so's PCB if we really
* need it.
*/
void
natmintr(struct mbuf *m)
{
struct socket *so;
struct natmpcb *npcb;
#ifdef DIAGNOSTIC
M_ASSERTPKTHDR(m);
#endif
NATM_LOCK();
npcb = (struct natmpcb *)m->m_pkthdr.rcvif; /* XXX: overloaded */
so = npcb->npcb_socket;
npcb->npcb_inq--;
if (npcb->npcb_flags & NPCB_DRAIN) {
if (npcb->npcb_inq == 0)
free(npcb, M_PCB); /* done! */
NATM_UNLOCK();
m_freem(m);
return;
}
if (npcb->npcb_flags & NPCB_FREE) {
NATM_UNLOCK();
m_freem(m); /* drop */
return;
}
#ifdef NEED_TO_RESTORE_IFP
m->m_pkthdr.rcvif = npcb->npcb_ifp;
#else
#ifdef DIAGNOSTIC
m->m_pkthdr.rcvif = NULL; /* null it out to be safe */
#endif
#endif
if (sbspace(&so->so_rcv) > m->m_pkthdr.len) {
#ifdef NATM_STAT
natm_sookcnt++;
natm_sookbytes += m->m_pkthdr.len;
#endif
sbappendrecord(&so->so_rcv, m);
sorwakeup(so);
NATM_UNLOCK();
} else {
#ifdef NATM_STAT
natm_sodropcnt++;
natm_sodropbytes += m->m_pkthdr.len;
#endif
NATM_UNLOCK();
m_freem(m);
}
}
void
soisconnected(struct socket *so)
{
register struct socket *head = so->so_head;
so->so_state &= ~(SS_ISCONNECTING|SS_ISDISCONNECTING|SS_ISCONFIRMING);
so->so_state |= SS_ISCONNECTED;
if (head && (so->so_state & SS_INCOMP)) {
TAILQ_REMOVE(&head->so_incomp, so, so_list);
head->so_incqlen--;
so->so_state &= ~SS_INCOMP;
TAILQ_INSERT_TAIL(&head->so_comp, so, so_list);
so->so_state |= SS_COMP;
sorwakeup(head);
soconnwakeup(head);
} else {
soconnwakeup(so);
sorwakeup(so);
sowwakeup(so);
}
}
void
soisdisconnected(struct socket *so)
{
KASSERT(solocked(so));
so->so_state &= ~(SS_ISCONNECTING|SS_ISCONNECTED|SS_ISDISCONNECTING);
so->so_state |= (SS_CANTRCVMORE|SS_CANTSENDMORE|SS_ISDISCONNECTED);
cv_broadcast(&so->so_cv);
sowwakeup(so);
sorwakeup(so);
}
int
socket6_send(struct socket *s, struct mbuf *mm, struct sockaddr_in6 *src)
{
if (s) {
if (sbappendaddr(&s->so_rcv, sin6tosa(src), mm, NULL) != 0) {
sorwakeup(s);
return 0;
}
}
m_freem(mm);
return -1;
}
void
soisdisconnecting(struct socket *so)
{
KASSERT(solocked(so));
so->so_state &= ~SS_ISCONNECTING;
so->so_state |= (SS_ISDISCONNECTING|SS_CANTRCVMORE|SS_CANTSENDMORE);
#if 0
cv_broadcast(&so->so_cv);
#endif
sowwakeup(so);
sorwakeup(so);
}
/*
* This may also be called for raw listeners.
*/
void
idp_input(struct mbuf *m, ...)
{
struct nspcb *nsp;
struct idp *idp = mtod(m, struct idp *);
struct ifnet *ifp = m->m_pkthdr.rcvif;
struct sockaddr_ns idp_ns;
va_list ap;
va_start(ap, m);
nsp = va_arg(ap, struct nspcb *);
va_end(ap);
if (nsp == NULL)
panic("No nspcb");
/*
* Construct sockaddr format source address.
* Stuff source address and datagram in user buffer.
*/
bzero(&idp_ns, sizeof(idp_ns));
idp_ns.sns_len = sizeof(idp_ns);
idp_ns.sns_family = AF_NS;
idp_ns.sns_addr = idp->idp_sna;
if (ns_neteqnn(idp->idp_sna.x_net, ns_zeronet) && ifp) {
struct ifaddr *ifa;
for (ifa = ifp->if_addrlist.tqh_first; ifa != 0;
ifa = ifa->ifa_list.tqe_next) {
if (ifa->ifa_addr->sa_family == AF_NS) {
idp_ns.sns_addr.x_net =
IA_SNS(ifa)->sns_addr.x_net;
break;
}
}
}
nsp->nsp_rpt = idp->idp_pt;
if ( ! (nsp->nsp_flags & NSP_RAWIN) ) {
m->m_len -= sizeof (struct idp);
m->m_pkthdr.len -= sizeof (struct idp);
m->m_data += sizeof (struct idp);
}
if (sbappendaddr(&nsp->nsp_socket->so_rcv, snstosa(&idp_ns), m,
(struct mbuf *)0) == 0)
goto bad;
sorwakeup(nsp->nsp_socket);
return;
bad:
m_freem(m);
}
/*
* This may also be called for raw listeners.
*/
void
ipx_input(struct mbuf *m, struct ipxpcb *ipxp)
{
struct ipx *ipx = mtod(m, struct ipx *);
struct ifnet *ifp = m->m_pkthdr.rcvif;
struct sockaddr_ipx ipx_ipx;
KASSERT(ipxp != NULL, ("ipx_input: NULL ipxpcb"));
IPX_LOCK_ASSERT(ipxp);
/*
* Construct sockaddr format source address.
* Stuff source address and datagram in user buffer.
*/
ipx_ipx.sipx_len = sizeof(ipx_ipx);
ipx_ipx.sipx_family = AF_IPX;
ipx_ipx.sipx_addr = ipx->ipx_sna;
ipx_ipx.sipx_zero[0] = '\0';
ipx_ipx.sipx_zero[1] = '\0';
if (ipx_neteqnn(ipx->ipx_sna.x_net, ipx_zeronet) && ifp != NULL) {
struct ifaddr *ifa;
for (ifa = TAILQ_FIRST(&ifp->if_addrhead); ifa != NULL;
ifa = TAILQ_NEXT(ifa, ifa_link)) {
if (ifa->ifa_addr->sa_family == AF_IPX) {
ipx_ipx.sipx_addr.x_net =
IA_SIPX(ifa)->sipx_addr.x_net;
break;
}
}
}
ipxp->ipxp_rpt = ipx->ipx_pt;
if ((ipxp->ipxp_flags & IPXP_RAWIN) == 0) {
m->m_len -= sizeof(struct ipx);
m->m_pkthdr.len -= sizeof(struct ipx);
m->m_data += sizeof(struct ipx);
}
#ifdef MAC
if (mac_socket_check_deliver(ipxp->ipxp_socket, m) != 0) {
m_freem(m);
return;
}
#endif
if (sbappendaddr(&ipxp->ipxp_socket->so_rcv,
(struct sockaddr *)&ipx_ipx, m, NULL) == 0)
m_freem(m);
else
sorwakeup(ipxp->ipxp_socket);
}
/* FUNCTION: sorflush()
*
* PARAM1: struct socket * socket structure
*
* RETURNS: none
*
* Closes the "read" half of the socket connection. No more data
* can be received on the socket, and any data currently in the
* socket receive buffer is discarded. Wakeup any processes waiting
* on the socket.
*/
void
sorflush(struct socket * so)
{
struct sockbuf *sb = &so->so_rcv;
int s;
sblock(sb);
socantrcvmore(so);
sbunlock(sb);
sbrelease(sb);
MEMSET((char *)sb, 0, sizeof (*sb));
s = so->so_error;
so->so_error = ESHUTDOWN;
sorwakeup(so);
so->so_error = s;
}
/*
* Notify a udp user of an asynchronous error; just wake up so that they can
* collect error status.
*/
struct inpcb *
udp_notify(struct inpcb *inp, int errno)
{
/*
* While udp_ctlinput() always calls udp_notify() with a read lock
* when invoking it directly, in_pcbnotifyall() currently uses write
* locks due to sharing code with TCP. For now, accept either a read
* or a write lock, but a read lock is sufficient.
*/
INP_LOCK_ASSERT(inp);
inp->inp_socket->so_error = errno;
sorwakeup(inp->inp_socket);
sowwakeup(inp->inp_socket);
return (inp);
}
static void
sco_input(void *arg, struct mbuf *m)
{
struct socket *so = arg;
/*
* since this data is time sensitive, if the buffer
* is full we just dump data until the latest one
* will fit.
*/
while (m->m_pkthdr.len > sbspace(&so->so_rcv))
sbdroprecord(&so->so_rcv);
DPRINTFN(10, "received %d bytes\n", m->m_pkthdr.len);
sbappendrecord(&so->so_rcv, m);
sorwakeup(so);
}
key_receive(struct socket *so, struct mbuf **paddr, struct uio *uio,
struct mbuf **mp0, struct mbuf **controlp, int *flagsp)
#endif
{
struct rawcb *rp = sotorawcb(so);
struct keycb *kp = (struct keycb *)rp;
int error;
#ifndef __FreeBSD__
error = (*kp->kp_receive)(so, paddr, uio, mp0, controlp, flagsp);
#else
error = soreceive(so, paddr, uio, mp0, controlp, flagsp);
#endif
if (kp->kp_queue &&
sbspace(&rp->rcb_socket->so_rcv) > kp->kp_queue->m_pkthdr.len)
sorwakeup(so);
return error;
}
static int
soo_drain(struct fileproc *fp, __unused vfs_context_t ctx)
{
int error = 0;
struct socket *so = (struct socket *)fp->f_fglob->fg_data;
if (so) {
socket_lock(so, 1);
so->so_state |= SS_DRAINING;
wakeup((caddr_t)&so->so_timeo);
sorwakeup(so);
sowwakeup(so);
socket_unlock(so, 1);
}
return (error);
}
int
userfw_domain_send_to_socket(struct socket *so, unsigned char *buf, size_t len)
{
struct mbchain m;
int err;
mb_init(&m);
err = mb_put_mem(&m, buf, len, MB_MSYSTEM);
if (err != 0)
{
mb_done(&m);
return err;
}
sbappendstream(&(so->so_rcv), mb_detach(&m));
sorwakeup(so);
return 0;
}
/*
* Receive data on a hook
*/
static int
ngs_rcvdata(hook_p hook, item_p item)
{
struct ngsock *const priv = NG_NODE_PRIVATE(NG_HOOK_NODE(hook));
struct ngpcb *const pcbp = priv->datasock;
struct socket *so;
struct sockaddr_ng *addr;
char *addrbuf[NG_HOOKSIZ + 4];
int addrlen;
struct mbuf *m;
NGI_GET_M(item, m);
NG_FREE_ITEM(item);
/* If there is no data socket, black-hole it. */
if (pcbp == NULL) {
NG_FREE_M(m);
return (0);
}
so = pcbp->ng_socket;
/* Get the return address into a sockaddr. */
addrlen = strlen(NG_HOOK_NAME(hook)); /* <= NG_HOOKSIZ - 1 */
addr = (struct sockaddr_ng *) addrbuf;
addr->sg_len = addrlen + 3;
addr->sg_family = AF_NETGRAPH;
bcopy(NG_HOOK_NAME(hook), addr->sg_data, addrlen);
addr->sg_data[addrlen] = '\0';
/* Try to tell the socket which hook it came in on. */
if (sbappendaddr((struct sockbuf *)&so->so_rcv, (struct sockaddr *)addr, m, NULL) == 0) {
m_freem(m);
TRAP_ERROR;
return (ENOBUFS);
}
sorwakeup(so);
return (0);
}
/*
* send message to the socket.
*/
static int
key_sendup0(struct rawcb *rp, struct mbuf *m, int promisc, int canwait)
{
struct keycb *kp = (struct keycb *)rp;
struct mbuf *n;
int error = 0;
if (canwait) {
if (kp->kp_queue) {
for (n = kp->kp_queue; n && n->m_nextpkt;
n = n->m_nextpkt)
;
n->m_nextpkt = m;
m = kp->kp_queue;
kp->kp_queue = NULL;
} else
m->m_nextpkt = NULL; /* just for safety */
} else
m->m_nextpkt = NULL;
for (; m && error == 0; m = n) {
n = m->m_nextpkt;
if (promisc) {
struct sadb_msg *pmsg;
M_PREPEND(m, sizeof(struct sadb_msg), M_NOWAIT);
if (m && m->m_len < sizeof(struct sadb_msg))
m = m_pullup(m, sizeof(struct sadb_msg));
if (!m) {
pfkeystat.in_nomem++;
error = ENOBUFS;
goto recovery;
}
m->m_pkthdr.len += sizeof(*pmsg);
pmsg = mtod(m, struct sadb_msg *);
bzero(pmsg, sizeof(*pmsg));
pmsg->sadb_msg_version = PF_KEY_V2;
pmsg->sadb_msg_type = SADB_X_PROMISC;
pmsg->sadb_msg_len = PFKEY_UNIT64(m->m_pkthdr.len);
/* pid and seq? */
pfkeystat.in_msgtype[pmsg->sadb_msg_type]++;
}
if (canwait &&
sbspace(&rp->rcb_socket->so_rcv) < m->m_pkthdr.len) {
error = EAGAIN;
goto recovery;
}
m->m_nextpkt = NULL;
if (!sbappendaddr(&rp->rcb_socket->so_rcv,
(struct sockaddr *)&key_src, m, NULL)) {
pfkeystat.in_nomem++;
error = ENOBUFS;
goto recovery;
} else {
sorwakeup(rp->rcb_socket);
error = 0;
}
}
return (error);
recovery:
if (kp->kp_queue) {
/*
* insert m to the head of queue, as normally mbuf on the queue
* is less important than others.
*/
if (m) {
m->m_nextpkt = kp->kp_queue;
kp->kp_queue = m;
}
} else {
/* recover the queue */
if (!m) {
/* first ENOBUFS case */
kp->kp_queue = n;
} else {
kp->kp_queue = m;
m->m_nextpkt = n;
}
}
return (error);
}
/*
* When an attempt at a new connection is noted on a socket
* which accepts connections, sonewconn is called. If the
* connection is possible (subject to space constraints, etc.)
* then we allocate a new structure, propoerly linked into the
* data structure of the original socket, and return this.
* Connstatus may be 0, SS_ISCONFIRMING, or SS_ISCONNECTED.
*/
struct socket *
sonewconn(struct socket *head, int connstatus)
{
struct socket *so;
int soqueue, error;
KASSERT(connstatus == 0 || connstatus == SS_ISCONFIRMING ||
connstatus == SS_ISCONNECTED);
KASSERT(solocked(head));
if ((head->so_options & SO_ACCEPTFILTER) != 0)
connstatus = 0;
soqueue = connstatus ? 1 : 0;
if (head->so_qlen + head->so_q0len > 3 * head->so_qlimit / 2)
return NULL;
so = soget(false);
if (so == NULL)
return NULL;
mutex_obj_hold(head->so_lock);
// so->so_lock = head->so_lock;
so->so_type = head->so_type;
so->so_options = head->so_options &~ SO_ACCEPTCONN;
so->so_linger = head->so_linger;
so->so_state = head->so_state | SS_NOFDREF;
so->so_proto = head->so_proto;
so->so_timeo = head->so_timeo;
#if 0 /* VADIM */
so->so_pgid = head->so_pgid;
so->so_send = head->so_send;
so->so_receive = head->so_receive;
so->so_uidinfo = head->so_uidinfo;
so->so_cpid = head->so_cpid;
#else
so->glueing_block = NULL;
#endif
#ifdef MBUFTRACE
so->so_mowner = head->so_mowner;
so->so_rcv.sb_mowner = head->so_rcv.sb_mowner;
so->so_snd.sb_mowner = head->so_snd.sb_mowner;
#endif
if (soreserve(so, head->so_snd.sb_hiwat, head->so_rcv.sb_hiwat) != 0)
goto out;
so->so_snd.sb_lowat = head->so_snd.sb_lowat;
so->so_rcv.sb_lowat = head->so_rcv.sb_lowat;
so->so_rcv.sb_timeo = head->so_rcv.sb_timeo;
so->so_snd.sb_timeo = head->so_snd.sb_timeo;
so->so_rcv.sb_flags |= head->so_rcv.sb_flags & (SB_AUTOSIZE | SB_ASYNC);
so->so_snd.sb_flags |= head->so_snd.sb_flags & (SB_AUTOSIZE | SB_ASYNC);
soqinsque(head, so, soqueue);
error = (*so->so_proto->pr_usrreq)(so, PRU_ATTACH, NULL, NULL,
NULL);
KASSERT(solocked(so));
if (error != 0) {
(void) soqremque(so, soqueue);
out:
/*
* Remove acccept filter if one is present.
* XXX Is this really needed?
*/
#if 0 /*VADIM*/
if (so->so_accf != NULL)
(void)accept_filt_clear(so);
#endif
soput(so);
return NULL;
}
if (connstatus) {
sorwakeup(head);
#if 0
cv_broadcast(&head->so_cv);
#endif
so->so_state |= connstatus;
}
return so;
}
int
udp_input(struct mbuf **mp, int *offp, int proto)
{
struct sockaddr_in udp_in = { sizeof udp_in, AF_INET };
int iphlen;
struct ip *ip;
struct udphdr *uh;
struct inpcb *inp;
struct mbuf *m;
struct mbuf *opts = NULL;
int len, off;
struct ip save_ip;
struct inpcbinfo *pcbinfo = &udbinfo[mycpuid];
off = *offp;
m = *mp;
*mp = NULL;
iphlen = off;
udp_stat.udps_ipackets++;
/*
* Strip IP options, if any; should skip this,
* make available to user, and use on returned packets,
* but we don't yet have a way to check the checksum
* with options still present.
*/
if (iphlen > sizeof(struct ip)) {
ip_stripoptions(m);
iphlen = sizeof(struct ip);
}
/*
* IP and UDP headers are together in first mbuf.
* Already checked and pulled up in ip_demux().
*/
KASSERT(m->m_len >= iphlen + sizeof(struct udphdr),
("UDP header not in one mbuf"));
ip = mtod(m, struct ip *);
uh = (struct udphdr *)((caddr_t)ip + iphlen);
/* destination port of 0 is illegal, based on RFC768. */
if (uh->uh_dport == 0)
goto bad;
/*
* Make mbuf data length reflect UDP length.
* If not enough data to reflect UDP length, drop.
*/
len = ntohs((u_short)uh->uh_ulen);
if (ip->ip_len != len) {
if (len > ip->ip_len || len < sizeof(struct udphdr)) {
udp_stat.udps_badlen++;
goto bad;
}
m_adj(m, len - ip->ip_len);
/* ip->ip_len = len; */
}
/*
* Save a copy of the IP header in case we want restore it
* for sending an ICMP error message in response.
*/
save_ip = *ip;
/*
* Checksum extended UDP header and data.
*/
if (uh->uh_sum) {
if (m->m_pkthdr.csum_flags & CSUM_DATA_VALID) {
if (m->m_pkthdr.csum_flags & CSUM_PSEUDO_HDR)
uh->uh_sum = m->m_pkthdr.csum_data;
else
uh->uh_sum = in_pseudo(ip->ip_src.s_addr,
ip->ip_dst.s_addr, htonl((u_short)len +
m->m_pkthdr.csum_data + IPPROTO_UDP));
uh->uh_sum ^= 0xffff;
} else {
char b[9];
bcopy(((struct ipovly *)ip)->ih_x1, b, 9);
bzero(((struct ipovly *)ip)->ih_x1, 9);
((struct ipovly *)ip)->ih_len = uh->uh_ulen;
uh->uh_sum = in_cksum(m, len + sizeof(struct ip));
bcopy(b, ((struct ipovly *)ip)->ih_x1, 9);
}
if (uh->uh_sum) {
udp_stat.udps_badsum++;
m_freem(m);
return(IPPROTO_DONE);
}
} else
udp_stat.udps_nosum++;
if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) ||
in_broadcast(ip->ip_dst, m->m_pkthdr.rcvif)) {
struct inpcbhead *connhead;
struct inpcontainer *ic, *ic_marker;
struct inpcontainerhead *ichead;
struct udp_mcast_arg arg;
struct inpcb *last;
int error;
/*
* Deliver a multicast or broadcast datagram to *all* sockets
* for which the local and remote addresses and ports match
* those of the incoming datagram. This allows more than
* one process to receive multi/broadcasts on the same port.
* (This really ought to be done for unicast datagrams as
* well, but that would cause problems with existing
* applications that open both address-specific sockets and
* a wildcard socket listening to the same port -- they would
* end up receiving duplicates of every unicast datagram.
* Those applications open the multiple sockets to overcome an
* inadequacy of the UDP socket interface, but for backwards
* compatibility we avoid the problem here rather than
* fixing the interface. Maybe 4.5BSD will remedy this?)
*/
/*
* Construct sockaddr format source address.
*/
udp_in.sin_port = uh->uh_sport;
udp_in.sin_addr = ip->ip_src;
arg.udp_in = &udp_in;
/*
* Locate pcb(s) for datagram.
* (Algorithm copied from raw_intr().)
*/
last = NULL;
arg.iphlen = iphlen;
connhead = &pcbinfo->hashbase[
INP_PCBCONNHASH(ip->ip_src.s_addr, uh->uh_sport,
ip->ip_dst.s_addr, uh->uh_dport, pcbinfo->hashmask)];
LIST_FOREACH(inp, connhead, inp_hash) {
#ifdef INET6
if (!INP_ISIPV4(inp))
continue;
#endif
if (!in_hosteq(inp->inp_faddr, ip->ip_src) ||
!in_hosteq(inp->inp_laddr, ip->ip_dst) ||
inp->inp_fport != uh->uh_sport ||
inp->inp_lport != uh->uh_dport)
continue;
arg.inp = inp;
arg.last = last;
arg.ip = ip;
arg.m = m;
error = udp_mcast_input(&arg);
if (error == ERESTART)
continue;
last = arg.last;
if (error == EJUSTRETURN)
goto done;
}
ichead = &pcbinfo->wildcardhashbase[
INP_PCBWILDCARDHASH(uh->uh_dport,
pcbinfo->wildcardhashmask)];
ic_marker = in_pcbcontainer_marker(mycpuid);
GET_PCBINFO_TOKEN(pcbinfo);
LIST_INSERT_HEAD(ichead, ic_marker, ic_list);
while ((ic = LIST_NEXT(ic_marker, ic_list)) != NULL) {
LIST_REMOVE(ic_marker, ic_list);
LIST_INSERT_AFTER(ic, ic_marker, ic_list);
inp = ic->ic_inp;
if (inp->inp_flags & INP_PLACEMARKER)
continue;
#ifdef INET6
if (!INP_ISIPV4(inp))
continue;
#endif
if (inp->inp_lport != uh->uh_dport)
continue;
if (inp->inp_laddr.s_addr != INADDR_ANY &&
inp->inp_laddr.s_addr != ip->ip_dst.s_addr)
continue;
arg.inp = inp;
arg.last = last;
arg.ip = ip;
arg.m = m;
error = udp_mcast_input(&arg);
if (error == ERESTART)
continue;
last = arg.last;
if (error == EJUSTRETURN)
break;
}
LIST_REMOVE(ic_marker, ic_list);
REL_PCBINFO_TOKEN(pcbinfo);
done:
if (last == NULL) {
/*
* No matching pcb found; discard datagram.
* (No need to send an ICMP Port Unreachable
* for a broadcast or multicast datgram.)
*/
udp_stat.udps_noportbcast++;
goto bad;
}
#ifdef IPSEC
/* check AH/ESP integrity. */
if (ipsec4_in_reject_so(m, last->inp_socket)) {
ipsecstat.in_polvio++;
goto bad;
}
#endif /*IPSEC*/
#ifdef FAST_IPSEC
/* check AH/ESP integrity. */
if (ipsec4_in_reject(m, last))
goto bad;
#endif /*FAST_IPSEC*/
udp_append(last, ip, m, iphlen + sizeof(struct udphdr),
&udp_in);
return(IPPROTO_DONE);
}
/*
* Locate pcb for datagram.
*/
inp = in_pcblookup_pkthash(pcbinfo, ip->ip_src, uh->uh_sport,
ip->ip_dst, uh->uh_dport, TRUE, m->m_pkthdr.rcvif,
udp_reuseport_ext ? m : NULL);
if (inp == NULL) {
if (log_in_vain) {
char buf[sizeof "aaa.bbb.ccc.ddd"];
strcpy(buf, inet_ntoa(ip->ip_dst));
log(LOG_INFO,
"Connection attempt to UDP %s:%d from %s:%d\n",
buf, ntohs(uh->uh_dport), inet_ntoa(ip->ip_src),
ntohs(uh->uh_sport));
}
udp_stat.udps_noport++;
if (m->m_flags & (M_BCAST | M_MCAST)) {
udp_stat.udps_noportbcast++;
goto bad;
}
if (blackhole)
goto bad;
#ifdef ICMP_BANDLIM
if (badport_bandlim(BANDLIM_ICMP_UNREACH) < 0)
goto bad;
#endif
*ip = save_ip;
ip->ip_len += iphlen;
icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PORT, 0, 0);
return(IPPROTO_DONE);
}
KASSERT(INP_ISIPV4(inp), ("not inet inpcb"));
#ifdef IPSEC
if (ipsec4_in_reject_so(m, inp->inp_socket)) {
ipsecstat.in_polvio++;
goto bad;
}
#endif /*IPSEC*/
#ifdef FAST_IPSEC
if (ipsec4_in_reject(m, inp))
goto bad;
#endif /*FAST_IPSEC*/
/*
* Check the minimum TTL for socket.
*/
if (ip->ip_ttl < inp->inp_ip_minttl)
goto bad;
/*
* Construct sockaddr format source address.
* Stuff source address and datagram in user buffer.
*/
udp_in.sin_port = uh->uh_sport;
udp_in.sin_addr = ip->ip_src;
if ((inp->inp_flags & INP_CONTROLOPTS) ||
(inp->inp_socket->so_options & SO_TIMESTAMP))
ip_savecontrol(inp, &opts, ip, m);
m_adj(m, iphlen + sizeof(struct udphdr));
lwkt_gettoken(&inp->inp_socket->so_rcv.ssb_token);
if (ssb_appendaddr(&inp->inp_socket->so_rcv,
(struct sockaddr *)&udp_in, m, opts) == 0) {
lwkt_reltoken(&inp->inp_socket->so_rcv.ssb_token);
udp_stat.udps_fullsock++;
goto bad;
}
lwkt_reltoken(&inp->inp_socket->so_rcv.ssb_token);
sorwakeup(inp->inp_socket);
return(IPPROTO_DONE);
bad:
m_freem(m);
if (opts)
m_freem(opts);
return(IPPROTO_DONE);
}
void
natmintr()
{
int s;
struct mbuf *m;
struct socket *so;
struct natmpcb *npcb;
next:
s = splnet();
IF_DEQUEUE(&natmintrq, m);
splx(s);
if (m == NULL)
return;
#ifdef DIAGNOSTIC
if ((m->m_flags & M_PKTHDR) == 0)
panic("natmintr no HDR");
#endif
npcb = (struct natmpcb *) m->m_pkthdr.rcvif; /* XXX: overloaded */
so = npcb->npcb_socket;
s = splnet(); /* could have atm devs @ different levels */
npcb->npcb_inq--;
splx(s);
if (npcb->npcb_flags & NPCB_DRAIN) {
m_freem(m);
if (npcb->npcb_inq == 0)
free(npcb, M_PCB); /* done! */
goto next;
}
if (npcb->npcb_flags & NPCB_FREE) {
m_freem(m); /* drop */
goto next;
}
#ifdef NEED_TO_RESTORE_IFP
m->m_pkthdr.rcvif = npcb->npcb_ifp;
#else
#ifdef DIAGNOSTIC
m->m_pkthdr.rcvif = NULL; /* null it out to be safe */
#endif
#endif
if (sbspace(&so->so_rcv) > m->m_pkthdr.len ||
((npcb->npcb_flags & NPCB_RAW) != 0 && so->so_rcv.sb_cc < NPCB_RAWCC) ) {
#ifdef NATM_STAT
natm_sookcnt++;
natm_sookbytes += m->m_pkthdr.len;
#endif
sbappendrecord(&so->so_rcv, m);
sorwakeup(so);
} else {
#ifdef NATM_STAT
natm_sodropcnt++;
natm_sodropbytes += m->m_pkthdr.len;
#endif
m_freem(m);
}
goto next;
}
