/*
* Wakeup processes waiting on a socket buffer. Do asynchronous notification
* via SIGIO if the socket has the SS_ASYNC flag set.
*
* Called with the socket buffer lock held; will release the lock by the end
* of the function. This allows the caller to acquire the socket buffer lock
* while testing for the need for various sorts of wakeup and hold it through
* to the point where it's no longer required. We currently hold the lock
* through calls out to other subsystems (with the exception of kqueue), and
* then release it to avoid lock order issues. It's not clear that's
* correct.
*/
void
sowakeup(struct socket *so, struct sockbuf *sb)
{
int ret = 0;
SOCKBUF_LOCK_ASSERT(sb);
so_wake_poll(so, sb);
if (sb->sb_flags & SB_WAIT) {
sb->sb_flags &= ~SB_WAIT;
wakeup(&sb->sb_cc);
}
if (sb->sb_upcall != NULL) {
ret = sb->sb_upcall(so, sb->sb_upcallarg, M_DONTWAIT);
if (ret == SU_ISCONNECTED) {
KASSERT(sb == &so->so_rcv,
("SO_SND upcall returned SU_ISCONNECTED"));
soupcall_clear(so, SO_RCV);
}
} else
ret = SU_OK;
SOCKBUF_UNLOCK(sb);
if (ret == SU_ISCONNECTED)
soisconnected(so);
mtx_assert(SOCKBUF_MTX(sb), MA_NOTOWNED);
}
static int
udp_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
{
struct inpcb *inp;
struct inpcbinfo *pcbinfo;
struct sockaddr_in *sin;
int error;
pcbinfo = get_inpcbinfo(so->so_proto->pr_protocol);
inp = sotoinpcb(so);
KASSERT(inp != NULL, ("udp_connect: inp == NULL"));
INP_WLOCK(inp);
if (inp->inp_faddr.s_addr != INADDR_ANY) {
INP_WUNLOCK(inp);
return (EISCONN);
}
sin = (struct sockaddr_in *)nam;
error = prison_remote_ip4(td->td_ucred, &sin->sin_addr);
if (error != 0) {
INP_WUNLOCK(inp);
return (error);
}
INP_HASH_WLOCK(pcbinfo);
error = in_pcbconnect(inp, nam, td->td_ucred);
INP_HASH_WUNLOCK(pcbinfo);
if (error == 0)
soisconnected(so);
INP_WUNLOCK(inp);
return (error);
}
/*
* Wakeup processes waiting on a socket buffer. Do asynchronous notification
* via SIGIO if the socket has the SS_ASYNC flag set.
*
* Called with the socket buffer lock held; will release the lock by the end
* of the function. This allows the caller to acquire the socket buffer lock
* while testing for the need for various sorts of wakeup and hold it through
* to the point where it's no longer required. We currently hold the lock
* through calls out to other subsystems (with the exception of kqueue), and
* then release it to avoid lock order issues. It's not clear that's
* correct.
*/
void
sowakeup(struct socket *so, struct sockbuf *sb)
{
int ret;
SOCKBUF_LOCK_ASSERT(sb);
selwakeuppri(sb->sb_sel, PSOCK);
if (!SEL_WAITING(sb->sb_sel))
sb->sb_flags &= ~SB_SEL;
if (sb->sb_flags & SB_WAIT) {
sb->sb_flags &= ~SB_WAIT;
wakeup(&sb->sb_acc);
}
KNOTE_LOCKED(&sb->sb_sel->si_note, 0);
if (sb->sb_upcall != NULL) {
ret = sb->sb_upcall(so, sb->sb_upcallarg, M_NOWAIT);
if (ret == SU_ISCONNECTED) {
KASSERT(sb == &so->so_rcv,
("SO_SND upcall returned SU_ISCONNECTED"));
soupcall_clear(so, SO_RCV);
}
} else
ret = SU_OK;
if (sb->sb_flags & SB_AIO)
sowakeup_aio(so, sb);
SOCKBUF_UNLOCK(sb);
if (ret == SU_ISCONNECTED)
soisconnected(so);
if ((so->so_state & SS_ASYNC) && so->so_sigio != NULL)
pgsigio(&so->so_sigio, SIGIO, 0);
mtx_assert(SOCKBUF_MTX(sb), MA_NOTOWNED);
}
static void
l2cap_connected(void *arg)
{
struct socket *so = arg;
DPRINTF("Connected\n");
soisconnected(so);
}
static void
rfcomm_connected(void *arg)
{
struct socket *so = arg;
KASSERT(so != NULL);
DPRINTF("Connected\n");
soisconnected(so);
}
static void
sohasdata(struct socket *so, void *arg, int events, int waitflag)
{
if (!soreadable(so))
return;
so->so_upcall = NULL;
so->so_rcv.sb_flags &= ~SB_UPCALL;
soisconnected(so);
return;
}
static void
udp6_connect(netmsg_t msg)
{
struct socket *so = msg->connect.base.nm_so;
struct sockaddr *nam = msg->connect.nm_nam;
struct thread *td = msg->connect.nm_td;
struct sockaddr_in6 *sin6_p;
struct inpcb *inp;
int error;
inp = so->so_pcb;
if (inp == NULL) {
error = EINVAL;
goto out;
}
sin6_p = (struct sockaddr_in6 *)nam;
if (IN6_IS_ADDR_V4MAPPED(&sin6_p->sin6_addr)) {
error = EADDRNOTAVAIL;
goto out;
}
if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr)) {
error = EISCONN;
goto out;
}
if (inp->inp_flags & INP_WILDCARD)
in_pcbremwildcardhash(inp);
if (!prison_remote_ip(td, nam)) {
error = EAFNOSUPPORT; /* IPv4 only jail */
goto out;
}
error = in6_pcbconnect(inp, nam, td);
if (error == 0) {
soisconnected(so);
} else if (error == EAFNOSUPPORT) { /* connection dissolved */
/*
* Follow traditional BSD behavior and retain
* the local port binding. But, fix the old misbehavior
* of overwriting any previously bound local address.
*/
if (!(inp->inp_flags & INP_WASBOUND_NOTANY))
inp->in6p_laddr = kin6addr_any;
in_pcbinswildcardhash(inp);
}
out:
lwkt_replymsg(&msg->connect.base.lmsg, error);
}
/*
* key_attach()
* derived from net/rtsock.c:rts_attach()
*/
static int
key_attach(struct socket *so, int proto, struct thread *p)
{
struct keycb *kp;
int s, error;
if (sotorawcb(so) != 0)
return EISCONN; /* XXX panic? */
kp = (struct keycb *)malloc(sizeof *kp, M_PCB, M_WAITOK); /* XXX */
if (kp == 0)
return ENOBUFS;
bzero(kp, sizeof *kp);
/*
* The splnet() is necessary to block protocols from sending
* error notifications (like RTM_REDIRECT or RTM_LOSING) while
* this PCB is extant but incompletely initialized.
* Probably we should try to do more of this work beforehand and
* eliminate the spl.
*/
s = splnet();
so->so_pcb = (caddr_t)kp;
error = raw_usrreqs.pru_attach(so, proto, p);
kp = (struct keycb *)sotorawcb(so);
if (error) {
free(kp, M_PCB);
so->so_pcb = (caddr_t) 0;
splx(s);
return error;
}
kp->kp_promisc = kp->kp_registered = 0;
if (kp->kp_raw.rcb_proto.sp_protocol == PF_KEY) /* XXX: AF_KEY */
key_cb.key_count++;
key_cb.any_count++;
kp->kp_raw.rcb_laddr = &key_src;
kp->kp_raw.rcb_faddr = &key_dst;
soisconnected(so);
so->so_options |= SO_USELOOPBACK;
splx(s);
return 0;
}
static int lpx_USER_connect( struct socket *so,
struct sockaddr *nam,
struct proc *td )
{
int error;
int s;
struct lpxpcb *lpxp = sotolpxpcb(so);
register struct sockaddr_lpx *slpx = (struct sockaddr_lpx *)nam;
if(slpx)
bzero(&slpx->sipx_addr.x_net, sizeof(slpx->sipx_addr.x_net));
if (!lpx_nullhost(lpxp->lpxp_faddr))
return (EISCONN);
s = splnet();
error = Lpx_PCB_connect(lpxp, nam, td);
splx(s);
if (error == 0)
soisconnected(so);
return (error);
}
static int
ddp_connect(struct socket *so, struct sockaddr *nam, struct proc *p)
{
struct ddpcb *ddp;
int error = 0;
int s;
ddp = sotoddpcb( so );
if ( ddp == NULL ) {
return( EINVAL);
}
if ( ddp->ddp_fsat.sat_port != ATADDR_ANYPORT ) {
return(EISCONN);
}
s = splnet();
error = at_pcbconnect( ddp, nam, p );
splx(s);
if ( error == 0 )
soisconnected( so );
return(error);
}
static int
ddp_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
{
struct ddpcb *ddp;
int error = 0;
ddp = sotoddpcb(so);
KASSERT(ddp != NULL, ("ddp_connect: ddp == NULL"));
DDP_LIST_XLOCK();
DDP_LOCK(ddp);
if (ddp->ddp_fsat.sat_port != ATADDR_ANYPORT) {
DDP_UNLOCK(ddp);
DDP_LIST_XUNLOCK();
return (EISCONN);
}
error = at_pcbconnect( ddp, nam, td );
DDP_UNLOCK(ddp);
DDP_LIST_XUNLOCK();
if (error == 0)
soisconnected(so);
return (error);
}
key_usrreq(struct socket *so, int req, struct mbuf *m, struct mbuf *nam,
struct mbuf *control, struct proc *p)
#endif /*__NetBSD__*/
{
int error = 0;
struct keycb *kp = (struct keycb *)sotorawcb(so);
int s;
#ifdef __NetBSD__
s = splsoftnet();
#else
s = splnet();
#endif
if (req == PRU_ATTACH) {
kp = (struct keycb *)malloc(sizeof(*kp), M_PCB, M_WAITOK);
so->so_pcb = (caddr_t)kp;
if (so->so_pcb)
bzero(so->so_pcb, sizeof(*kp));
}
if (req == PRU_DETACH && kp) {
int af = kp->kp_raw.rcb_proto.sp_protocol;
struct mbuf *n;
if (af == PF_KEY)
key_cb.key_count--;
key_cb.any_count--;
key_freereg(so);
while (kp->kp_queue) {
n = kp->kp_queue->m_nextpkt;
kp->kp_queue->m_nextpkt = NULL;
m_freem(kp->kp_queue);
kp->kp_queue = n;
}
}
#ifndef __NetBSD__
error = raw_usrreq(so, req, m, nam, control);
#else
error = raw_usrreq(so, req, m, nam, control, p);
#endif
m = control = NULL; /* reclaimed in raw_usrreq */
kp = (struct keycb *)sotorawcb(so);
if (req == PRU_ATTACH && kp) {
int af = kp->kp_raw.rcb_proto.sp_protocol;
if (error) {
pfkeystat.sockerr++;
free((caddr_t)kp, M_PCB);
so->so_pcb = (caddr_t) 0;
splx(s);
return (error);
}
kp->kp_promisc = kp->kp_registered = 0;
kp->kp_receive = so->so_receive;
so->so_receive = key_receive;
if (af == PF_KEY) /* XXX: AF_KEY */
key_cb.key_count++;
key_cb.any_count++;
kp->kp_raw.rcb_laddr = &key_src;
kp->kp_raw.rcb_faddr = &key_dst;
soisconnected(so);
so->so_options |= SO_USELOOPBACK;
}
splx(s);
return (error);
}
void
spx_input(struct mbuf *m, struct ipxpcb *ipxp)
{
struct spxpcb *cb;
struct spx *si = mtod(m, struct spx *);
struct socket *so;
struct spx spx_savesi;
int dropsocket = 0;
short ostate = 0;
spxstat.spxs_rcvtotal++;
KASSERT(ipxp != NULL, ("spx_input: ipxpcb == NULL"));
/*
* spx_input() assumes that the caller will hold both the pcb list
* lock and also the ipxp lock. spx_input() will release both before
* returning, and may in fact trade in the ipxp lock for another pcb
* lock following sonewconn().
*/
IPX_LIST_LOCK_ASSERT();
IPX_LOCK_ASSERT(ipxp);
cb = ipxtospxpcb(ipxp);
KASSERT(cb != NULL, ("spx_input: cb == NULL"));
if (ipxp->ipxp_flags & IPXP_DROPPED)
goto drop;
if (m->m_len < sizeof(*si)) {
if ((m = m_pullup(m, sizeof(*si))) == NULL) {
IPX_UNLOCK(ipxp);
IPX_LIST_UNLOCK();
spxstat.spxs_rcvshort++;
return;
}
si = mtod(m, struct spx *);
}
si->si_seq = ntohs(si->si_seq);
si->si_ack = ntohs(si->si_ack);
si->si_alo = ntohs(si->si_alo);
so = ipxp->ipxp_socket;
KASSERT(so != NULL, ("spx_input: so == NULL"));
#ifdef MAC
if (mac_socket_check_deliver(so, m) != 0)
goto drop;
#endif
if (so->so_options & SO_DEBUG || traceallspxs) {
ostate = cb->s_state;
spx_savesi = *si;
}
if (so->so_options & SO_ACCEPTCONN) {
struct spxpcb *ocb = cb;
so = sonewconn(so, 0);
if (so == NULL)
goto drop;
/*
* This is ugly, but ....
*
* Mark socket as temporary until we're committed to keeping
* it. The code at ``drop'' and ``dropwithreset'' check the
* flag dropsocket to see if the temporary socket created
* here should be discarded. We mark the socket as
* discardable until we're committed to it below in
* TCPS_LISTEN.
*
* XXXRW: In the new world order of real kernel parallelism,
* temporarily allocating the socket when we're "not sure"
* seems like a bad idea, as we might race to remove it if
* the listen socket is closed...?
*
* We drop the lock of the listen socket ipxp, and acquire
* the lock of the new socket ippx.
*/
dropsocket++;
IPX_UNLOCK(ipxp);
ipxp = (struct ipxpcb *)so->so_pcb;
IPX_LOCK(ipxp);
ipxp->ipxp_laddr = si->si_dna;
cb = ipxtospxpcb(ipxp);
cb->s_mtu = ocb->s_mtu; /* preserve sockopts */
cb->s_flags = ocb->s_flags; /* preserve sockopts */
cb->s_flags2 = ocb->s_flags2; /* preserve sockopts */
cb->s_state = TCPS_LISTEN;
}
IPX_LOCK_ASSERT(ipxp);
/*
* Packet received on connection. Reset idle time and keep-alive
* timer.
*/
cb->s_idle = 0;
cb->s_timer[SPXT_KEEP] = SPXTV_KEEP;
switch (cb->s_state) {
case TCPS_LISTEN:{
struct sockaddr_ipx *sipx, ssipx;
struct ipx_addr laddr;
/*
* If somebody here was carying on a conversation and went
* away, and his pen pal thinks he can still talk, we get the
* misdirected packet.
*/
if (spx_hardnosed && (si->si_did != 0 || si->si_seq != 0)) {
spx_istat.gonawy++;
goto dropwithreset;
}
sipx = &ssipx;
bzero(sipx, sizeof *sipx);
sipx->sipx_len = sizeof(*sipx);
sipx->sipx_family = AF_IPX;
sipx->sipx_addr = si->si_sna;
laddr = ipxp->ipxp_laddr;
if (ipx_nullhost(laddr))
ipxp->ipxp_laddr = si->si_dna;
if (ipx_pcbconnect(ipxp, (struct sockaddr *)sipx, &thread0)) {
ipxp->ipxp_laddr = laddr;
spx_istat.noconn++;
goto drop;
}
spx_template(cb);
dropsocket = 0; /* committed to socket */
cb->s_did = si->si_sid;
cb->s_rack = si->si_ack;
cb->s_ralo = si->si_alo;
#define THREEWAYSHAKE
#ifdef THREEWAYSHAKE
cb->s_state = TCPS_SYN_RECEIVED;
cb->s_force = 1 + SPXT_KEEP;
spxstat.spxs_accepts++;
cb->s_timer[SPXT_KEEP] = SPXTV_KEEP;
}
break;
case TCPS_SYN_RECEIVED: {
/*
* This state means that we have heard a response to our
* acceptance of their connection. It is probably logically
* unnecessary in this implementation.
*/
if (si->si_did != cb->s_sid) {
spx_istat.wrncon++;
goto drop;
}
#endif
ipxp->ipxp_fport = si->si_sport;
cb->s_timer[SPXT_REXMT] = 0;
cb->s_timer[SPXT_KEEP] = SPXTV_KEEP;
soisconnected(so);
cb->s_state = TCPS_ESTABLISHED;
spxstat.spxs_accepts++;
}
break;
case TCPS_SYN_SENT:
/*
* This state means that we have gotten a response to our
* attempt to establish a connection. We fill in the data
* from the other side, telling us which port to respond to,
* instead of the well-known one we might have sent to in the
* first place. We also require that this is a response to
* our connection id.
*/
if (si->si_did != cb->s_sid) {
spx_istat.notme++;
goto drop;
}
spxstat.spxs_connects++;
cb->s_did = si->si_sid;
cb->s_rack = si->si_ack;
cb->s_ralo = si->si_alo;
cb->s_dport = ipxp->ipxp_fport = si->si_sport;
cb->s_timer[SPXT_REXMT] = 0;
cb->s_flags |= SF_ACKNOW;
soisconnected(so);
cb->s_state = TCPS_ESTABLISHED;
/*
* Use roundtrip time of connection request for initial rtt.
*/
if (cb->s_rtt) {
cb->s_srtt = cb->s_rtt << 3;
cb->s_rttvar = cb->s_rtt << 1;
SPXT_RANGESET(cb->s_rxtcur,
((cb->s_srtt >> 2) + cb->s_rttvar) >> 1,
SPXTV_MIN, SPXTV_REXMTMAX);
cb->s_rtt = 0;
}
}
static int
udp6_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
{
struct inpcb *inp;
struct inpcbinfo *pcbinfo;
struct sockaddr_in6 *sin6;
int error;
pcbinfo = udp_get_inpcbinfo(so->so_proto->pr_protocol);
inp = sotoinpcb(so);
sin6 = (struct sockaddr_in6 *)nam;
KASSERT(inp != NULL, ("udp6_connect: inp == NULL"));
/*
* XXXRW: Need to clarify locking of v4/v6 flags.
*/
INP_WLOCK(inp);
#ifdef INET
if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) {
struct sockaddr_in sin;
if ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0) {
error = EINVAL;
goto out;
}
if (inp->inp_faddr.s_addr != INADDR_ANY) {
error = EISCONN;
goto out;
}
in6_sin6_2_sin(&sin, sin6);
inp->inp_vflag |= INP_IPV4;
inp->inp_vflag &= ~INP_IPV6;
error = prison_remote_ip4(td->td_ucred, &sin.sin_addr);
if (error != 0)
goto out;
INP_HASH_WLOCK(pcbinfo);
error = in_pcbconnect(inp, (struct sockaddr *)&sin,
td->td_ucred);
INP_HASH_WUNLOCK(pcbinfo);
if (error == 0)
soisconnected(so);
goto out;
}
#endif
if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr)) {
error = EISCONN;
goto out;
}
inp->inp_vflag &= ~INP_IPV4;
inp->inp_vflag |= INP_IPV6;
error = prison_remote_ip6(td->td_ucred, &sin6->sin6_addr);
if (error != 0)
goto out;
INP_HASH_WLOCK(pcbinfo);
error = in6_pcbconnect(inp, nam, td->td_ucred);
INP_HASH_WUNLOCK(pcbinfo);
if (error == 0)
soisconnected(so);
out:
INP_WUNLOCK(inp);
return (error);
}
int
raw_usrreq(struct socket *so,
int req,
struct mbuf *m,
struct mbuf *nam,
struct mbuf *control)
{
register struct rawcb *rp = sotorawcb(so);
register int error = 0;
int len;
if (req == PRU_CONTROL)
return (EOPNOTSUPP);
if (control && control->m_len) {
error = EOPNOTSUPP;
goto release;
}
if (rp == 0) {
error = EINVAL;
goto release;
}
switch (req) {
/*
* Allocate a raw control block and fill in the
* necessary info to allow packets to be routed to
* the appropriate raw interface routine.
*/
case PRU_ATTACH:
if ((so->so_state & SS_PRIV) == 0) {
error = EACCES;
break;
}
error = raw_attach(so, (int)nam);
break;
/*
* Destroy state just before socket deallocation.
* Flush data or not depending on the options.
*/
case PRU_DETACH:
if (rp == 0) {
error = ENOTCONN;
break;
}
raw_detach(rp);
break;
#ifdef notdef
/*
* If a socket isn't bound to a single address,
* the raw input routine will hand it anything
* within that protocol family (assuming there's
* nothing else around it should go to).
*/
case PRU_CONNECT:
if (rp->rcb_faddr) {
error = EISCONN;
break;
}
nam = m_copym(nam, 0, M_COPYALL, M_WAIT);
rp->rcb_faddr = mtod(nam, struct sockaddr *);
soisconnected(so);
break;
case PRU_BIND:
if (rp->rcb_laddr) {
error = EINVAL; /* XXX */
break;
}
error = raw_bind(so, nam);
break;
#endif
case PRU_CONNECT2:
error = EOPNOTSUPP;
goto release;
case PRU_DISCONNECT:
if (rp->rcb_faddr == 0) {
error = ENOTCONN;
break;
}
raw_disconnect(rp);
soisdisconnected(so);
break;
/*
* Mark the connection as being incapable of further input.
*/
case PRU_SHUTDOWN:
socantsendmore(so);
break;
/*
* Ship a packet out. The appropriate raw output
* routine handles any massaging necessary.
*/
case PRU_SEND:
if (nam) {
if (rp->rcb_faddr) {
error = EISCONN;
break;
}
rp->rcb_faddr = mtod(nam, struct sockaddr *);
} else if (rp->rcb_faddr == 0) {
error = ENOTCONN;
break;
}
error = (*so->so_proto->pr_output)(m, so);
m = NULL;
if (nam)
rp->rcb_faddr = 0;
break;
case PRU_ABORT:
raw_disconnect(rp);
sofree(so);
soisdisconnected(so);
break;
case PRU_SENSE:
/*
* stat: don't bother with a blocksize.
*/
return (0);
/*
* Not supported.
*/
case PRU_RCVOOB:
case PRU_RCVD:
return(EOPNOTSUPP);
case PRU_LISTEN:
case PRU_ACCEPT:
case PRU_SENDOOB:
error = EOPNOTSUPP;
break;
case PRU_SOCKADDR:
if (rp->rcb_laddr == 0) {
error = EINVAL;
break;
}
len = rp->rcb_laddr->sa_len;
aligned_bcopy((caddr_t)rp->rcb_laddr, mtod(nam, caddr_t), (unsigned)len);
nam->m_len = len;
break;
case PRU_PEERADDR:
if (rp->rcb_faddr == 0) {
error = ENOTCONN;
break;
}
len = rp->rcb_faddr->sa_len;
aligned_bcopy((caddr_t)rp->rcb_faddr, mtod(nam, caddr_t), (unsigned)len);
nam->m_len = len;
break;
default:
panic("raw_usrreq");
}
int natm_usrreq(struct socket *so, int req, struct mbuf *m, struct mbuf *nam,
struct mbuf *control, struct proc *p)
{
int error = 0, s, s2;
struct natmpcb *npcb;
struct sockaddr_natm *snatm;
struct atm_pseudoioctl api;
struct atm_pseudohdr *aph;
struct atm_rawioctl ario;
struct ifnet *ifp;
int proto = so->so_proto->pr_protocol;
s = splsoftnet();
npcb = (struct natmpcb *) so->so_pcb;
if (npcb == NULL && req != PRU_ATTACH) {
error = EINVAL;
goto done;
}
switch (req) {
case PRU_ATTACH: /* attach protocol to up */
if (npcb) {
error = EISCONN;
break;
}
if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) {
if (proto == PROTO_NATMAAL5)
error = soreserve(so, natm5_sendspace, natm5_recvspace);
else
error = soreserve(so, natm0_sendspace, natm0_recvspace);
if (error)
break;
}
so->so_pcb = (caddr_t) (npcb = npcb_alloc(M_WAITOK));
npcb->npcb_socket = so;
break;
case PRU_DETACH: /* detach protocol from up */
/*
* we turn on 'drain' *before* we sofree.
*/
npcb_free(npcb, NPCB_DESTROY); /* drain */
so->so_pcb = NULL;
sofree(so);
break;
case PRU_CONNECT: /* establish connection to peer */
/*
* validate nam and npcb
*/
if (nam->m_len != sizeof(*snatm)) {
error = EINVAL;
break;
}
snatm = mtod(nam, struct sockaddr_natm *);
if (snatm->snatm_len != sizeof(*snatm) ||
(npcb->npcb_flags & NPCB_FREE) == 0) {
error = EINVAL;
break;
}
if (snatm->snatm_family != AF_NATM) {
error = EAFNOSUPPORT;
break;
}
snatm->snatm_if[IFNAMSIZ-1] = '\0'; /* XXX ensure null termination
since ifunit() uses strcmp */
/*
* convert interface string to ifp, validate.
*/
ifp = ifunit(snatm->snatm_if);
if (ifp == NULL || (ifp->if_flags & IFF_RUNNING) == 0) {
error = ENXIO;
break;
}
if (ifp->if_output != atm_output) {
error = EAFNOSUPPORT;
break;
}
/*
* register us with the NATM PCB layer
*/
if (npcb_add(npcb, ifp, snatm->snatm_vci, snatm->snatm_vpi) != npcb) {
error = EADDRINUSE;
break;
}
/*
* enable rx
*/
ATM_PH_FLAGS(&api.aph) = (proto == PROTO_NATMAAL5) ? ATM_PH_AAL5 : 0;
ATM_PH_VPI(&api.aph) = npcb->npcb_vpi;
ATM_PH_SETVCI(&api.aph, npcb->npcb_vci);
api.rxhand = npcb;
s2 = splnet();
if (ifp->if_ioctl == NULL ||
ifp->if_ioctl(ifp, SIOCATMENA, (caddr_t) &api) != 0) {
splx(s2);
npcb_free(npcb, NPCB_REMOVE);
error = EIO;
break;
}
splx(s2);
soisconnected(so);
break;
case PRU_DISCONNECT: /* disconnect from peer */
if ((npcb->npcb_flags & NPCB_CONNECTED) == 0) {
printf("natm: disconnected check\n");
error = EIO;
break;
}
ifp = npcb->npcb_ifp;
/*
* disable rx
*/
ATM_PH_FLAGS(&api.aph) = ATM_PH_AAL5;
ATM_PH_VPI(&api.aph) = npcb->npcb_vpi;
ATM_PH_SETVCI(&api.aph, npcb->npcb_vci);
api.rxhand = npcb;
s2 = splnet();
if (ifp->if_ioctl != NULL)
ifp->if_ioctl(ifp, SIOCATMDIS, (caddr_t) &api);
splx(s2);
npcb_free(npcb, NPCB_REMOVE);
soisdisconnected(so);
break;
case PRU_SHUTDOWN: /* won't send any more data */
socantsendmore(so);
break;
case PRU_SEND: /* send this data */
if (control && control->m_len) {
m_freem(control);
m_freem(m);
error = EINVAL;
break;
}
/*
* send the data. we must put an atm_pseudohdr on first
*/
M_PREPEND(m, sizeof(*aph), M_WAITOK);
aph = mtod(m, struct atm_pseudohdr *);
ATM_PH_VPI(aph) = npcb->npcb_vpi;
ATM_PH_SETVCI(aph, npcb->npcb_vci);
ATM_PH_FLAGS(aph) = (proto == PROTO_NATMAAL5) ? ATM_PH_AAL5 : 0;
error = atm_output(npcb->npcb_ifp, m, NULL, NULL);
break;
case PRU_SENSE: /* return status into m */
/* return zero? */
break;
case PRU_PEERADDR: /* fetch peer's address */
snatm = mtod(nam, struct sockaddr_natm *);
bzero(snatm, sizeof(*snatm));
nam->m_len = snatm->snatm_len = sizeof(*snatm);
snatm->snatm_family = AF_NATM;
#if defined(__NetBSD__) || defined(__OpenBSD__)
bcopy(npcb->npcb_ifp->if_xname, snatm->snatm_if, sizeof(snatm->snatm_if));
#elif defined(__FreeBSD__)
sprintf(snatm->snatm_if, "%s%d", npcb->npcb_ifp->if_name,
npcb->npcb_ifp->if_unit);
#endif
snatm->snatm_vci = npcb->npcb_vci;
snatm->snatm_vpi = npcb->npcb_vpi;
break;
case PRU_CONTROL: /* control operations on protocol */
/*
* raw atm ioctl. comes in as a SIOCRAWATM. we convert it to
* SIOCXRAWATM and pass it to the driver.
*/
if ((u_long)m == SIOCRAWATM) {
if (npcb->npcb_ifp == NULL) {
error = ENOTCONN;
break;
}
ario.npcb = npcb;
ario.rawvalue = *((int *)nam);
error = npcb->npcb_ifp->if_ioctl(npcb->npcb_ifp,
SIOCXRAWATM, (caddr_t) &ario);
if (!error) {
if (ario.rawvalue)
npcb->npcb_flags |= NPCB_RAW;
else
npcb->npcb_flags &= ~(NPCB_RAW);
}
break;
}
error = EOPNOTSUPP;
break;
case PRU_BIND: /* bind socket to address */
case PRU_LISTEN: /* listen for connection */
case PRU_ACCEPT: /* accept connection from peer */
case PRU_CONNECT2: /* connect two sockets */
case PRU_ABORT: /* abort (fast DISCONNECT, DETACH) */
/* (only happens if LISTEN socket) */
case PRU_RCVD: /* have taken data; more room now */
case PRU_FASTTIMO: /* 200ms timeout */
case PRU_SLOWTIMO: /* 500ms timeout */
case PRU_RCVOOB: /* retrieve out of band data */
case PRU_SENDOOB: /* send out of band data */
case PRU_PROTORCV: /* receive from below */
case PRU_PROTOSEND: /* send to below */
case PRU_SOCKADDR: /* fetch socket's address */
#ifdef DIAGNOSTIC
printf("natm: PRU #%d unsupported\n", req);
#endif
error = EOPNOTSUPP;
break;
default: panic("natm usrreq");
}
done:
splx(s);
return(error);
}
/*
* Completes some final bits of initialization for just established connections
* and changes their state to TCPS_ESTABLISHED.
*
* The ISNs are from after the exchange of SYNs. i.e., the true ISN + 1.
*/
void
make_established(struct toepcb *toep, uint32_t snd_isn, uint32_t rcv_isn,
uint16_t opt)
{
struct inpcb *inp = toep->inp;
struct socket *so = inp->inp_socket;
struct tcpcb *tp = intotcpcb(inp);
long bufsize;
uint32_t iss = be32toh(snd_isn) - 1; /* true ISS */
uint32_t irs = be32toh(rcv_isn) - 1; /* true IRS */
uint16_t tcpopt = be16toh(opt);
struct flowc_tx_params ftxp;
INP_WLOCK_ASSERT(inp);
KASSERT(tp->t_state == TCPS_SYN_SENT ||
tp->t_state == TCPS_SYN_RECEIVED,
("%s: TCP state %s", __func__, tcpstates[tp->t_state]));
CTR4(KTR_CXGBE, "%s: tid %d, toep %p, inp %p",
__func__, toep->tid, toep, inp);
tp->t_state = TCPS_ESTABLISHED;
tp->t_starttime = ticks;
TCPSTAT_INC(tcps_connects);
tp->irs = irs;
tcp_rcvseqinit(tp);
tp->rcv_wnd = toep->rx_credits << 10;
tp->rcv_adv += tp->rcv_wnd;
tp->last_ack_sent = tp->rcv_nxt;
/*
* If we were unable to send all rx credits via opt0, save the remainder
* in rx_credits so that they can be handed over with the next credit
* update.
*/
SOCKBUF_LOCK(&so->so_rcv);
bufsize = select_rcv_wnd(so);
SOCKBUF_UNLOCK(&so->so_rcv);
toep->rx_credits = bufsize - tp->rcv_wnd;
tp->iss = iss;
tcp_sendseqinit(tp);
tp->snd_una = iss + 1;
tp->snd_nxt = iss + 1;
tp->snd_max = iss + 1;
assign_rxopt(tp, tcpopt);
SOCKBUF_LOCK(&so->so_snd);
if (so->so_snd.sb_flags & SB_AUTOSIZE && V_tcp_do_autosndbuf)
bufsize = V_tcp_autosndbuf_max;
else
bufsize = sbspace(&so->so_snd);
SOCKBUF_UNLOCK(&so->so_snd);
ftxp.snd_nxt = tp->snd_nxt;
ftxp.rcv_nxt = tp->rcv_nxt;
ftxp.snd_space = bufsize;
ftxp.mss = tp->t_maxseg;
send_flowc_wr(toep, &ftxp);
soisconnected(so);
}
/*ARGSUSED*/
void
spp_input(struct mbuf *m, ...)
{
struct nspcb *nsp;
struct sppcb *cb;
struct spidp *si = mtod(m, struct spidp *);
struct socket *so;
short ostate = 0;
int dropsocket = 0;
va_list ap;
va_start(ap, m);
nsp = va_arg(ap, struct nspcb *);
va_end(ap);
sppstat.spps_rcvtotal++;
if (nsp == 0) {
panic("No nspcb in spp_input");
return;
}
cb = nstosppcb(nsp);
if (cb == 0) goto bad;
if (m->m_len < sizeof(*si)) {
if ((m = m_pullup(m, sizeof(*si))) == 0) {
sppstat.spps_rcvshort++;
return;
}
si = mtod(m, struct spidp *);
}
si->si_seq = ntohs(si->si_seq);
si->si_ack = ntohs(si->si_ack);
si->si_alo = ntohs(si->si_alo);
so = nsp->nsp_socket;
if (so->so_options & SO_DEBUG || traceallspps) {
ostate = cb->s_state;
spp_savesi = *si;
}
if (so->so_options & SO_ACCEPTCONN) {
struct sppcb *ocb = cb;
so = sonewconn(so, 0);
if (so == 0) {
goto drop;
}
/*
* This is ugly, but ....
*
* Mark socket as temporary until we're
* committed to keeping it. The code at
* ``drop'' and ``dropwithreset'' check the
* flag dropsocket to see if the temporary
* socket created here should be discarded.
* We mark the socket as discardable until
* we're committed to it below in TCPS_LISTEN.
*/
dropsocket++;
nsp = (struct nspcb *)so->so_pcb;
nsp->nsp_laddr = si->si_dna;
cb = nstosppcb(nsp);
cb->s_mtu = ocb->s_mtu; /* preserve sockopts */
cb->s_flags = ocb->s_flags; /* preserve sockopts */
cb->s_flags2 = ocb->s_flags2; /* preserve sockopts */
cb->s_state = TCPS_LISTEN;
}
/*
* Packet received on connection.
* reset idle time and keep-alive timer;
*/
cb->s_idle = 0;
cb->s_timer[SPPT_KEEP] = SPPTV_KEEP;
switch (cb->s_state) {
case TCPS_LISTEN:{
struct mbuf *am;
struct sockaddr_ns *sns;
struct ns_addr laddr;
/*
* If somebody here was carying on a conversation
* and went away, and his pen pal thinks he can
* still talk, we get the misdirected packet.
*/
if (spp_hardnosed && (si->si_did != 0 || si->si_seq != 0)) {
spp_istat.gonawy++;
goto dropwithreset;
}
am = m_get(M_DONTWAIT, MT_SONAME);
if (am == NULL)
goto drop;
am->m_len = sizeof (struct sockaddr_ns);
sns = mtod(am, struct sockaddr_ns *);
sns->sns_len = sizeof(*sns);
sns->sns_family = AF_NS;
sns->sns_addr = si->si_sna;
laddr = nsp->nsp_laddr;
if (ns_nullhost(laddr))
nsp->nsp_laddr = si->si_dna;
if (ns_pcbconnect(nsp, am)) {
nsp->nsp_laddr = laddr;
(void) m_free(am);
spp_istat.noconn++;
goto drop;
}
(void) m_free(am);
spp_template(cb);
dropsocket = 0; /* committed to socket */
cb->s_did = si->si_sid;
cb->s_rack = si->si_ack;
cb->s_ralo = si->si_alo;
#define THREEWAYSHAKE
#ifdef THREEWAYSHAKE
cb->s_state = TCPS_SYN_RECEIVED;
cb->s_force = 1 + SPPT_KEEP;
sppstat.spps_accepts++;
cb->s_timer[SPPT_KEEP] = SPPTV_KEEP;
}
break;
/*
* This state means that we have heard a response
* to our acceptance of their connection
* It is probably logically unnecessary in this
* implementation.
*/
case TCPS_SYN_RECEIVED: {
if (si->si_did!=cb->s_sid) {
spp_istat.wrncon++;
goto drop;
}
#endif
nsp->nsp_fport = si->si_sport;
cb->s_timer[SPPT_REXMT] = 0;
cb->s_timer[SPPT_KEEP] = SPPTV_KEEP;
soisconnected(so);
cb->s_state = TCPS_ESTABLISHED;
sppstat.spps_accepts++;
}
break;
/*
* This state means that we have gotten a response
* to our attempt to establish a connection.
* We fill in the data from the other side,
* telling us which port to respond to, instead of the well-
* known one we might have sent to in the first place.
* We also require that this is a response to our
* connection id.
*/
case TCPS_SYN_SENT:
if (si->si_did!=cb->s_sid) {
spp_istat.notme++;
goto drop;
}
sppstat.spps_connects++;
cb->s_did = si->si_sid;
cb->s_rack = si->si_ack;
cb->s_ralo = si->si_alo;
cb->s_dport = nsp->nsp_fport = si->si_sport;
cb->s_timer[SPPT_REXMT] = 0;
cb->s_flags |= SF_ACKNOW;
soisconnected(so);
cb->s_state = TCPS_ESTABLISHED;
/* Use roundtrip time of connection request for initial rtt */
if (cb->s_rtt) {
cb->s_srtt = cb->s_rtt << 3;
cb->s_rttvar = cb->s_rtt << 1;
SPPT_RANGESET(cb->s_rxtcur,
((cb->s_srtt >> 2) + cb->s_rttvar) >> 1,
SPPTV_MIN, SPPTV_REXMTMAX);
cb->s_rtt = 0;
}
}
/*
* TCP input routine, follows pages 65-76 of the
* protocol specification dated September, 1981 very closely.
*/
void
tcp_input(usn_mbuf_t *m, int iphlen)
{
struct tcpiphdr *ti;
struct inpcb *inp;
u_char *optp = NULL;
int optlen;
int len, tlen, off;
struct tcpcb *tp = 0;
int tiflags;
struct usn_socket *so = 0;
int todrop, acked, ourfinisacked;
int needoutput = 0;
short ostate;
struct usn_in_addr laddr;
int dropsocket = 0;
int iss = 0;
u_long tiwin, ts_val, ts_ecr;
int ts_present = 0;
(void)needoutput;
g_tcpstat.tcps_rcvtotal++;
// Get IP and TCP header together in first mbuf.
// Note: IP leaves IP header in first mbuf.
ti = mtod(m, struct tcpiphdr *);
if (iphlen > sizeof (usn_ip_t))
ip_stripoptions(m, (usn_mbuf_t *)0);
if (m->mlen < sizeof (struct tcpiphdr)) {
if ((m = m_pullup(m, sizeof (struct tcpiphdr))) == 0) {
g_tcpstat.tcps_rcvshort++;
return;
}
ti = mtod(m, struct tcpiphdr *);
}
#ifdef DUMP_PAYLOAD
dump_chain(m,"tcp");
#endif
/*
* Checksum extended TCP header and data.
*/
tlen = ntohs(((usn_ip_t *)ti)->ip_len);
len = sizeof (usn_ip_t) + tlen;
ti->ti_next = ti->ti_prev = 0;
ti->ti_x1 = 0;
ti->ti_len = (u_short)tlen;
HTONS(ti->ti_len);
ti->ti_sum = in_cksum(m, len);
if (ti->ti_sum) {
g_tcpstat.tcps_rcvbadsum++;
goto drop;
}
/*
* Check that TCP offset makes sense,
* pull out TCP options and adjust length. XXX
*/
off = ti->ti_off << 2;
if (off < sizeof (struct tcphdr) || off > tlen) {
g_tcpstat.tcps_rcvbadoff++;
goto drop;
}
tlen -= off;
ti->ti_len = tlen;
if (off > sizeof (struct tcphdr)) {
if (m->mlen < sizeof(usn_ip_t) + off) {
if ((m = m_pullup(m, sizeof (usn_ip_t) + off)) == 0) {
g_tcpstat.tcps_rcvshort++;
return;
}
ti = mtod(m, struct tcpiphdr *);
}
optlen = off - sizeof (struct tcphdr);
optp = mtod(m, u_char *) + sizeof (struct tcpiphdr);
// Do quick retrieval of timestamp options ("options
// prediction?"). If timestamp is the only option and it's
// formatted as recommended in RFC 1323 appendix A, we
// quickly get the values now and not bother calling
// tcp_dooptions(), etc.
if ((optlen == TCPOLEN_TSTAMP_APPA ||
(optlen > TCPOLEN_TSTAMP_APPA &&
optp[TCPOLEN_TSTAMP_APPA] == TCPOPT_EOL)) &&
*(u_int *)optp == htonl(TCPOPT_TSTAMP_HDR) &&
(ti->ti_flags & TH_SYN) == 0) {
ts_present = 1;
ts_val = ntohl(*(u_long *)(optp + 4));
ts_ecr = ntohl(*(u_long *)(optp + 8));
optp = NULL; // we've parsed the options
}
}
tiflags = ti->ti_flags;
// Convert TCP protocol specific fields to host format.
NTOHL(ti->ti_seq);
NTOHL(ti->ti_ack);
NTOHS(ti->ti_win);
NTOHS(ti->ti_urp);
// Locate pcb for segment.
findpcb:
inp = g_tcp_last_inpcb;
if (inp->inp_lport != ti->ti_dport ||
inp->inp_fport != ti->ti_sport ||
inp->inp_faddr.s_addr != ti->ti_src.s_addr ||
inp->inp_laddr.s_addr != ti->ti_dst.s_addr) {
inp = in_pcblookup(&g_tcb, ti->ti_src, ti->ti_sport,
ti->ti_dst, ti->ti_dport, INPLOOKUP_WILDCARD);
if (inp)
g_tcp_last_inpcb = inp;
++g_tcpstat.tcps_pcbcachemiss;
}
// If the state is CLOSED (i.e., TCB does not exist) then
// all data in the incoming segment is discarded.
// If the TCB exists but is in CLOSED state, it is embryonic,
// but should either do a listen or a connect soon.
if (inp == 0)
goto dropwithreset;
tp = intotcpcb(inp);
DEBUG("found inp cb, laddr=%x, lport=%d, faddr=%x,"
" fport=%d, tp_state=%d, tp_flags=%d",
inp->inp_laddr.s_addr,
inp->inp_lport,
inp->inp_faddr.s_addr,
inp->inp_fport, tp->t_state, tp->t_flags);
if (tp == 0)
goto dropwithreset;
if (tp->t_state == TCPS_CLOSED)
goto drop;
// Unscale the window into a 32-bit value.
if ((tiflags & TH_SYN) == 0)
tiwin = ti->ti_win << tp->snd_scale;
else
tiwin = ti->ti_win;
so = inp->inp_socket;
DEBUG("socket info, options=%x", so->so_options);
if (so->so_options & (SO_DEBUG|SO_ACCEPTCONN)) {
if (so->so_options & SO_DEBUG) {
ostate = tp->t_state;
g_tcp_saveti = *ti;
}
if (so->so_options & SO_ACCEPTCONN) {
if ((tiflags & (TH_RST|TH_ACK|TH_SYN)) != TH_SYN) {
// Note: dropwithreset makes sure we don't
// send a reset in response to a RST.
if (tiflags & TH_ACK) {
g_tcpstat.tcps_badsyn++;
goto dropwithreset;
}
DEBUG("SYN is expected, tiflags=%d", tiflags);
goto drop;
}
so = sonewconn(so, 0);
if (so == 0) {
DEBUG("failed to create new connection, tiflags=%d", tiflags);
goto drop;
}
// Mark socket as temporary until we're
// committed to keeping it. The code at
// ``drop'' and ``dropwithreset'' check the
// flag dropsocket to see if the temporary
// socket created here should be discarded.
// We mark the socket as discardable until
// we're committed to it below in TCPS_LISTEN.
dropsocket++;
inp = (struct inpcb *)so->so_pcb;
inp->inp_laddr = ti->ti_dst;
inp->inp_lport = ti->ti_dport;
// BSD >= 4.3
inp->inp_options = ip_srcroute();
tp = intotcpcb(inp);
tp->t_state = TCPS_LISTEN;
// Compute proper scaling value from buffer space
while (tp->request_r_scale < TCP_MAX_WINSHIFT &&
TCP_MAXWIN << tp->request_r_scale < so->so_rcv->sb_hiwat)
tp->request_r_scale++;
}
}
// Segment received on connection.
// Reset idle time and keep-alive timer.
tp->t_idle = 0;
tp->t_timer[TCPT_KEEP] = g_tcp_keepidle;
// Process options if not in LISTEN state,
// else do it below (after getting remote address).
if (optp && tp->t_state != TCPS_LISTEN)
tcp_dooptions(tp, optp, optlen, ti,
&ts_present, &ts_val, &ts_ecr);
// Header prediction: check for the two common cases
// of a uni-directional data xfer. If the packet has
// no control flags, is in-sequence, the window didn't
// change and we're not retransmitting, it's a
// candidate. If the length is zero and the ack moved
// forward, we're the sender side of the xfer. Just
// free the data acked & wake any higher level process
// that was blocked waiting for space. If the length
// is non-zero and the ack didn't move, we're the
// receiver side. If we're getting packets in-order
// (the reassembly queue is empty), add the data to
// the socket buffer and note that we need a delayed ack.
if (tp->t_state == TCPS_ESTABLISHED &&
(tiflags & (TH_SYN|TH_FIN|TH_RST|TH_URG|TH_ACK)) == TH_ACK &&
(!ts_present || TSTMP_GEQ(ts_val, tp->ts_recent)) &&
ti->ti_seq == tp->rcv_nxt &&
tiwin && tiwin == tp->snd_wnd &&
tp->snd_nxt == tp->snd_max) {
// If last ACK falls within this segment's sequence numbers,
// record the timestamp.
if ( ts_present && TSTMP_GEQ(ts_val, tp->ts_recent) &&
SEQ_LEQ(ti->ti_seq, tp->last_ack_sent) ){
tp->ts_recent_age = g_tcp_now;
tp->ts_recent = ts_val;
}
if (ti->ti_len == 0) {
if (SEQ_GT(ti->ti_ack, tp->snd_una) &&
SEQ_LEQ(ti->ti_ack, tp->snd_max) &&
tp->snd_cwnd >= tp->snd_wnd) {
// this is a pure ack for outstanding data.
++g_tcpstat.tcps_predack;
if (ts_present)
tcp_xmit_timer(tp, g_tcp_now-ts_ecr+1);
else if (tp->t_rtt &&
SEQ_GT(ti->ti_ack, tp->t_rtseq))
tcp_xmit_timer(tp, tp->t_rtt);
acked = ti->ti_ack - tp->snd_una;
g_tcpstat.tcps_rcvackpack++;
g_tcpstat.tcps_rcvackbyte += acked;
TRACE("drop so_snd buffer, drop_bytes=%d, len=%d",
acked, so->so_snd.sb_cc);
sbdrop(so->so_snd, acked);
tp->snd_una = ti->ti_ack;
usn_free_cmbuf(m);
// If all outstanding data are acked, stop
// retransmit timer, otherwise restart timer
// using current (possibly backed-off) value.
// If process is waiting for space,
// wakeup/selwakeup/signal. If data
// are ready to send, let tcp_output
// decide between more output or persist.
if (tp->snd_una == tp->snd_max)
tp->t_timer[TCPT_REXMT] = 0;
else if (tp->t_timer[TCPT_PERSIST] == 0)
tp->t_timer[TCPT_REXMT] = tp->t_rxtcur;
if (so->so_options & SO_DEBUG)
tcp_trace(TA_INPUT, ostate, tp, &g_tcp_saveti, 0);
//if (so->so_snd->sb_flags & SB_NOTIFY) {
// usnet_tcpin_wwakeup(so, USN_TCP_IN, usn_tcpev_sbnotify, 0);
// sowwakeup(so);
//}
// send buffer is available for app thread.
usnet_tcpin_wwakeup(so, USN_TCP_IN, USN_TCPEV_WRITE, 0);
if (so->so_snd->sb_cc)
tcp_output(tp);
return;
}
} else if (ti->ti_ack == tp->snd_una &&
tp->seg_next == (struct tcpiphdr *)tp &&
ti->ti_len <= sbspace(so->so_rcv)) {
// this is a pure, in-sequence data packet
// with nothing on the reassembly queue and
// we have enough buffer space to take it.
++g_tcpstat.tcps_preddat;
tp->rcv_nxt += ti->ti_len;
g_tcpstat.tcps_rcvpack++;
g_tcpstat.tcps_rcvbyte += ti->ti_len;
// Drop TCP, IP headers and TCP options then add data
// to socket buffer.
m->head += sizeof(struct tcpiphdr)+off-sizeof(struct tcphdr);
m->mlen -= sizeof(struct tcpiphdr)+off-sizeof(struct tcphdr);
TRACE("add data to rcv buf");
sbappend(so->so_rcv, m);
sorwakeup(so);
// new data is available for app threads.
usnet_tcpin_rwakeup(so, USN_TCP_IN, USN_TCPEV_READ, m);
if (so->so_options & SO_DEBUG) {
TRACE("tcp trace, so_options=%d", so->so_options);
tcp_trace(TA_INPUT, ostate, tp, &g_tcp_saveti, 0);
}
tp->t_flags |= TF_DELACK;
return;
}
}
// Drop TCP, IP headers and TCP options.
m->head += sizeof(struct tcpiphdr)+off-sizeof(struct tcphdr);
m->mlen -= sizeof(struct tcpiphdr)+off-sizeof(struct tcphdr);
// Calculate amount of space in receive window,
// and then do TCP input processing.
// Receive window is amount of space in rcv queue,
// but not less than advertised window.
{
int win;
win = sbspace(so->so_rcv);
if (win < 0)
win = 0;
tp->rcv_wnd = max(win, (int)(tp->rcv_adv - tp->rcv_nxt));
}
switch (tp->t_state) {
// If the state is LISTEN then ignore segment if it contains an RST.
// If the segment contains an ACK then it is bad and send a RST.
// If it does not contain a SYN then it is not interesting; drop it.
// Don't bother responding if the destination was a broadcast.
// Otherwise initialize tp->rcv_nxt, and tp->irs, select an initial
// tp->iss, and send a segment:
// <SEQ=ISS><ACK=RCV_NXT><CTL=SYN,ACK>
// Also initialize tp->snd_nxt to tp->iss+1 and tp->snd_una to tp->iss.
// Fill in remote peer address fields if not previously specified.
// Enter SYN_RECEIVED state, and process any other fields of this
// segment in this state.
case TCPS_LISTEN: {
usn_mbuf_t *am;
struct usn_sockaddr_in *sin;
if (tiflags & TH_RST)
goto drop;
if (tiflags & TH_ACK)
goto dropwithreset;
if ((tiflags & TH_SYN) == 0)
goto drop;
// RFC1122 4.2.3.10, p. 104: discard bcast/mcast SYN
// in_broadcast() should never return true on a received
// packet with M_BCAST not set.
//if (m->m_flags & (M_BCAST|M_MCAST) ||
// IN_MULTICAST(ntohl(ti->ti_dst.s_addr)))
// goto drop;
am = usn_get_mbuf(0, BUF_MSIZE, 0); // XXX: the size!
if (am == NULL)
goto drop;
am->mlen = sizeof (struct usn_sockaddr_in);
sin = mtod(am, struct usn_sockaddr_in *);
sin->sin_family = AF_INET;
sin->sin_len = sizeof(*sin);
sin->sin_addr = ti->ti_src;
sin->sin_port = ti->ti_sport;
bzero((caddr_t)sin->sin_zero, sizeof(sin->sin_zero));
laddr = inp->inp_laddr;
if (inp->inp_laddr.s_addr == USN_INADDR_ANY)
inp->inp_laddr = ti->ti_dst;
if (in_pcbconnect(inp, am)) {
inp->inp_laddr = laddr;
usn_free_mbuf(am);
goto drop;
}
usn_free_mbuf(am);
tp->t_template = tcp_template(tp);
if (tp->t_template == 0) {
tp = tcp_drop(tp, ENOBUFS);
dropsocket = 0; // socket is already gone
goto drop;
}
if (optp)
tcp_dooptions(tp, optp, optlen, ti,
&ts_present, &ts_val, &ts_ecr);
if (iss)
tp->iss = iss;
else
tp->iss = g_tcp_iss;
g_tcp_iss += TCP_ISSINCR/4;
tp->irs = ti->ti_seq;
tcp_sendseqinit(tp);
tcp_rcvseqinit(tp);
tp->t_flags |= TF_ACKNOW;
TRACE("change tcp state to TCPS_SYN_RECEIVED, state=%d, tp_flags=%d",
tp->t_state, tp->t_flags);
tp->t_state = TCPS_SYN_RECEIVED;
// tcp event
usnet_tcpin_ewakeup(so, USN_TCP_IN, USN_TCPST_SYN_RECEIVED, 0);
tp->t_timer[TCPT_KEEP] = TCPTV_KEEP_INIT;
dropsocket = 0; // committed to socket
g_tcpstat.tcps_accepts++;
goto trimthenstep6;
}
// If the state is SYN_SENT:
// if seg contains an ACK, but not for our SYN, drop the input.
// if seg contains a RST, then drop the connection.
// if seg does not contain SYN, then drop it.
// Otherwise this is an acceptable SYN segment
// initialize tp->rcv_nxt and tp->irs
// if seg contains ack then advance tp->snd_una
// if SYN has been acked change to ESTABLISHED else SYN_RCVD state
// arrange for segment to be acked (eventually)
// continue processing rest of data/controls, beginning with URG
case TCPS_SYN_SENT:
if ((tiflags & TH_ACK) &&
(SEQ_LEQ(ti->ti_ack, tp->iss) ||
SEQ_GT(ti->ti_ack, tp->snd_max)))
goto dropwithreset;
if (tiflags & TH_RST) {
if (tiflags & TH_ACK)
tp = tcp_drop(tp, ECONNREFUSED);
goto drop;
}
if ((tiflags & TH_SYN) == 0)
goto drop;
if (tiflags & TH_ACK) {
tp->snd_una = ti->ti_ack;
if (SEQ_LT(tp->snd_nxt, tp->snd_una))
tp->snd_nxt = tp->snd_una;
tp->t_timer[TCPT_REXMT] = 0;
}
tp->irs = ti->ti_seq;
tcp_rcvseqinit(tp);
tp->t_flags |= TF_ACKNOW;
TRACE("ack now, tp flags=%d", tp->t_flags);
// XXX: remove second test.
if (tiflags & TH_ACK /*&& SEQ_GT(tp->snd_una, tp->iss)*/) {
g_tcpstat.tcps_connects++;
soisconnected(so);
TRACE("change tcp state to TCPS_ESTABLISHED,"
" state=%d, tp_flags=%d", tp->t_state, tp->t_flags);
tp->t_state = TCPS_ESTABLISHED;
// Do window scaling on this connection?
if ((tp->t_flags & (TF_RCVD_SCALE|TF_REQ_SCALE)) ==
(TF_RCVD_SCALE|TF_REQ_SCALE)) {
tp->snd_scale = tp->requested_s_scale;
tp->rcv_scale = tp->request_r_scale;
}
tcp_reass(tp, (struct tcpiphdr *)0, (usn_mbuf_t *)0);
// if we didn't have to retransmit the SYN,
// use its rtt as our initial srtt & rtt var.
if (tp->t_rtt)
tcp_xmit_timer(tp, tp->t_rtt);
} else {
TRACE("change tcp state to TCPS_SYN_RECEIVED, state=%d, tp_flags=%d",
tp->t_state, tp->t_flags);
tp->t_state = TCPS_SYN_RECEIVED;
// tcp event
usnet_tcpin_ewakeup(so, USN_TCP_IN, USN_TCPST_SYN_RECEIVED, 0);
}
trimthenstep6:
// Advance ti->ti_seq to correspond to first data byte.
// If data, trim to stay within window,
// dropping FIN if necessary.
ti->ti_seq++;
if (ti->ti_len > tp->rcv_wnd) {
todrop = ti->ti_len - tp->rcv_wnd;
m_adj(m, -todrop);
ti->ti_len = tp->rcv_wnd;
tiflags &= ~TH_FIN;
g_tcpstat.tcps_rcvpackafterwin++;
g_tcpstat.tcps_rcvbyteafterwin += todrop;
}
tp->snd_wl1 = ti->ti_seq - 1;
tp->rcv_up = ti->ti_seq;
goto step6;
}
// States other than LISTEN or SYN_SENT.
// First check timestamp, if present.
// Then check that at least some bytes of segment are within
// receive window. If segment begins before rcv_nxt,
// drop leading data (and SYN); if nothing left, just ack.
//
// RFC 1323 PAWS: If we have a timestamp reply on this segment
// and it's less than ts_recent, drop it.
if (ts_present && (tiflags & TH_RST) == 0 && tp->ts_recent &&
TSTMP_LT(ts_val, tp->ts_recent)) {
// Check to see if ts_recent is over 24 days old.
if ((int)(g_tcp_now - tp->ts_recent_age) > TCP_PAWS_IDLE) {
// Invalidate ts_recent. If this segment updates
// ts_recent, the age will be reset later and ts_recent
// will get a valid value. If it does not, setting
// ts_recent to zero will at least satisfy the
// requirement that zero be placed in the timestamp
// echo reply when ts_recent isn't valid. The
// age isn't reset until we get a valid ts_recent
// because we don't want out-of-order segments to be
// dropped when ts_recent is old.
tp->ts_recent = 0;
} else {
g_tcpstat.tcps_rcvduppack++;
g_tcpstat.tcps_rcvdupbyte += ti->ti_len;
g_tcpstat.tcps_pawsdrop++;
goto dropafterack;
}
}
todrop = tp->rcv_nxt - ti->ti_seq;
if (todrop > 0) {
if (tiflags & TH_SYN) {
tiflags &= ~TH_SYN;
ti->ti_seq++;
if (ti->ti_urp > 1)
ti->ti_urp--;
else
tiflags &= ~TH_URG;
todrop--;
}
if ( todrop >= ti->ti_len ||
( todrop == ti->ti_len && (tiflags & TH_FIN ) == 0 ) ) {
// Any valid FIN must be to the left of the window.
// At this point the FIN must be a duplicate or
// out of sequence; drop it.
tiflags &= ~TH_FIN;
// Send an ACK to resynchronize and drop any data
// But keep on processing for RST or ACK.
tp->t_flags |= TF_ACKNOW;
TRACE("send ack now to resync, tp_flags=%d", tp->t_flags);
todrop = ti->ti_len;
g_tcpstat.tcps_rcvdupbyte += ti->ti_len;
g_tcpstat.tcps_rcvduppack++;
} else {
g_tcpstat.tcps_rcvpartduppack++;
g_tcpstat.tcps_rcvpartdupbyte += ti->ti_len;
}
m_adj(m, todrop);
ti->ti_seq += todrop;
ti->ti_len -= todrop;
if (ti->ti_urp > todrop)
ti->ti_urp -= todrop;
else {
tiflags &= ~TH_URG;
ti->ti_urp = 0;
}
}
// If new data are received on a connection after the
// user processes are gone, then RST the other end.
if ((so->so_state & USN_NOFDREF) &&
tp->t_state > TCPS_CLOSE_WAIT && ti->ti_len) {
tp = tcp_close(tp);
g_tcpstat.tcps_rcvafterclose++;
goto dropwithreset;
}
// If segment ends after window, drop trailing data
// (and PUSH and FIN); if nothing left, just ACK.
todrop = (ti->ti_seq+ti->ti_len) - (tp->rcv_nxt+tp->rcv_wnd);
if (todrop > 0) {
g_tcpstat.tcps_rcvpackafterwin++;
if (todrop >= ti->ti_len) {
g_tcpstat.tcps_rcvbyteafterwin += ti->ti_len;
// If a new connection request is received
// while in TIME_WAIT, drop the old connection
// and start over if the sequence numbers
// are above the previous ones.
if (tiflags & TH_SYN &&
tp->t_state == TCPS_TIME_WAIT &&
SEQ_GT(ti->ti_seq, tp->rcv_nxt)) {
iss = tp->snd_nxt + TCP_ISSINCR;
tp = tcp_close(tp);
goto findpcb;
}
// If window is closed can only take segments at
// window edge, and have to drop data and PUSH from
// incoming segments. Continue processing, but
// remember to ack. Otherwise, drop segment
// and ack.
if (tp->rcv_wnd == 0 && ti->ti_seq == tp->rcv_nxt) {
tp->t_flags |= TF_ACKNOW;
g_tcpstat.tcps_rcvwinprobe++;
} else
goto dropafterack;
} else
g_tcpstat.tcps_rcvbyteafterwin += todrop;
m_adj(m, -todrop);
ti->ti_len -= todrop;
tiflags &= ~(TH_PUSH|TH_FIN);
}
// check valid timestamp. Replace code above.
if (ts_present && TSTMP_GEQ(ts_val, tp->ts_recent) &&
SEQ_LEQ(ti->ti_seq, tp->last_ack_sent) ) {
tp->ts_recent_age = g_tcp_now;
tp->ts_recent = ts_val;
}
// If the RST bit is set examine the state:
// SYN_RECEIVED STATE:
// If passive open, return to LISTEN state.
// If active open, inform user that connection was refused.
// ESTABLISHED, FIN_WAIT_1, FIN_WAIT2, CLOSE_WAIT STATES:
// Inform user that connection was reset, and close tcb.
// CLOSING, LAST_ACK, TIME_WAIT STATES
// Close the tcb.
if (tiflags&TH_RST) switch (tp->t_state) {
case TCPS_SYN_RECEIVED:
so->so_error = ECONNREFUSED;
goto close;
case TCPS_ESTABLISHED:
case TCPS_FIN_WAIT_1:
case TCPS_FIN_WAIT_2:
case TCPS_CLOSE_WAIT:
so->so_error = ECONNRESET;
close:
DEBUG("change tcp state to TCPS_CLOSED, state=%d", tp->t_state);
tp->t_state = TCPS_CLOSED;
// tcp event
usnet_tcpin_ewakeup(so, USN_TCP_IN, USN_TCPST_CLOSED, 0);
g_tcpstat.tcps_drops++;
tp = tcp_close(tp);
goto drop;
case TCPS_CLOSING:
case TCPS_LAST_ACK:
case TCPS_TIME_WAIT:
tp = tcp_close(tp);
goto drop;
}
// If a SYN is in the window, then this is an
// error and we send an RST and drop the connection.
if (tiflags & TH_SYN) {
tp = tcp_drop(tp, ECONNRESET);
goto dropwithreset;
}
// If the ACK bit is off we drop the segment and return.
if ((tiflags & TH_ACK) == 0)
goto drop;
// Ack processing.
switch (tp->t_state) {
// In SYN_RECEIVED state if the ack ACKs our SYN then enter
// ESTABLISHED state and continue processing, otherwise
// send an RST.
case TCPS_SYN_RECEIVED:
if (SEQ_GT(tp->snd_una, ti->ti_ack) ||
SEQ_GT(ti->ti_ack, tp->snd_max))
goto dropwithreset;
g_tcpstat.tcps_connects++;
DEBUG("change tcp state to TCPS_ESTABLISHED, state=%d", tp->t_state);
tp->t_state = TCPS_ESTABLISHED;
soisconnected(so);
// Do window scaling?
if ((tp->t_flags & (TF_RCVD_SCALE|TF_REQ_SCALE)) ==
(TF_RCVD_SCALE|TF_REQ_SCALE)) {
tp->snd_scale = tp->requested_s_scale;
tp->rcv_scale = tp->request_r_scale;
}
tcp_reass(tp, (struct tcpiphdr *)0, (usn_mbuf_t *)0);
tp->snd_wl1 = ti->ti_seq - 1;
// fall into ...
// In ESTABLISHED state: drop duplicate ACKs; ACK out of range
// ACKs. If the ack is in the range
// tp->snd_una < ti->ti_ack <= tp->snd_max
// then advance tp->snd_una to ti->ti_ack and drop
// data from the retransmission queue. If this ACK reflects
// more up to date window information we update our window information.
case TCPS_ESTABLISHED:
case TCPS_FIN_WAIT_1:
case TCPS_FIN_WAIT_2:
case TCPS_CLOSE_WAIT:
case TCPS_CLOSING:
case TCPS_LAST_ACK:
case TCPS_TIME_WAIT:
if (SEQ_LEQ(ti->ti_ack, tp->snd_una)) {
if (ti->ti_len == 0 && tiwin == tp->snd_wnd) {
g_tcpstat.tcps_rcvdupack++;
// If we have outstanding data (other than
// a window probe), this is a completely
// duplicate ack (ie, window info didn't
// change), the ack is the biggest we've
// seen and we've seen exactly our rexmt
// threshhold of them, assume a packet
// has been dropped and retransmit it.
// Kludge snd_nxt & the congestion
// window so we send only this one
// packet.
//
// We know we're losing at the current
// window size so do congestion avoidance
// (set ssthresh to half the current window
// and pull our congestion window back to
// the new ssthresh).
//
// Dup acks mean that packets have left the
// network (they're now cached at the receiver)
// so bump cwnd by the amount in the receiver
// to keep a constant cwnd packets in the
// network.
if (tp->t_timer[TCPT_REXMT] == 0 ||
ti->ti_ack != tp->snd_una)
tp->t_dupacks = 0;
else if (++tp->t_dupacks == g_tcprexmtthresh) {
// congestion avoidance
tcp_seq onxt = tp->snd_nxt;
u_int win =
min(tp->snd_wnd, tp->snd_cwnd) / 2 / tp->t_maxseg;
if (win < 2)
win = 2;
tp->snd_ssthresh = win * tp->t_maxseg;
tp->t_timer[TCPT_REXMT] = 0;
tp->t_rtt = 0;
tp->snd_nxt = ti->ti_ack;
tp->snd_cwnd = tp->t_maxseg;
tcp_output(tp);
tp->snd_cwnd = tp->snd_ssthresh +
tp->t_maxseg * tp->t_dupacks;
if (SEQ_GT(onxt, tp->snd_nxt))
tp->snd_nxt = onxt;
goto drop;
} else if (tp->t_dupacks > g_tcprexmtthresh) {
tp->snd_cwnd += tp->t_maxseg;
tcp_output(tp);
goto drop;
}
} else
tp->t_dupacks = 0;
break;
}
// If the congestion window was inflated to account
// for the other side's cached packets, retract it.
if (tp->t_dupacks > g_tcprexmtthresh &&
tp->snd_cwnd > tp->snd_ssthresh)
tp->snd_cwnd = tp->snd_ssthresh;
tp->t_dupacks = 0;
if (SEQ_GT(ti->ti_ack, tp->snd_max)) {
g_tcpstat.tcps_rcvacktoomuch++;
goto dropafterack;
}
acked = ti->ti_ack - tp->snd_una;
g_tcpstat.tcps_rcvackpack++;
g_tcpstat.tcps_rcvackbyte += acked;
// If we have a timestamp reply, update smoothed
// round trip time. If no timestamp is present but
// transmit timer is running and timed sequence
// number was acked, update smoothed round trip time.
// Since we now have an rtt measurement, cancel the
// timer backoff (cf., Phil Karn's retransmit alg.).
// Recompute the initial retransmit timer.
if (ts_present)
tcp_xmit_timer(tp, g_tcp_now-ts_ecr+1);
else if (tp->t_rtt && SEQ_GT(ti->ti_ack, tp->t_rtseq))
tcp_xmit_timer(tp,tp->t_rtt);
// If all outstanding data is acked, stop retransmit
// timer and remember to restart (more output or persist).
// If there is more data to be acked, restart retransmit
// timer, using current (possibly backed-off) value.
if (ti->ti_ack == tp->snd_max) {
tp->t_timer[TCPT_REXMT] = 0;
DEBUG("change needoutput to 1");
needoutput = 1;
tp->t_flags |= TF_NEEDOUTPUT;
} else if (tp->t_timer[TCPT_PERSIST] == 0)
tp->t_timer[TCPT_REXMT] = tp->t_rxtcur;
// When new data is acked, open the congestion window.
// If the window gives us less than ssthresh packets
// in flight, open exponentially (maxseg per packet).
// Otherwise open linearly: maxseg per window
// (maxseg * (maxseg / cwnd) per packet).
{
u_int cw = tp->snd_cwnd;
u_int incr = tp->t_maxseg;
if (cw > tp->snd_ssthresh)
incr = incr * incr / cw;
tp->snd_cwnd = min(cw + incr, TCP_MAXWIN<<tp->snd_scale);
}
if (acked > so->so_snd->sb_cc) {
tp->snd_wnd -= so->so_snd->sb_cc;
DEBUG("drop all so_snd buffer, drop_bytes=%d, acked=%d",
so->so_snd->sb_cc, acked);
sbdrop(so->so_snd, (int)so->so_snd->sb_cc);
ourfinisacked = 1;
} else {
DEBUG("drop so_snd buffer, drop_bytes=%d, len=%d", acked, so->so_snd->sb_cc);
sbdrop(so->so_snd, acked);
tp->snd_wnd -= acked;
ourfinisacked = 0;
}
//if (so->so_snd->sb_flags & SB_NOTIFY) {
sowwakeup(so);
usnet_tcpin_wwakeup(so, USN_TCP_IN, USN_TCPEV_WRITE, 0);
//}
tp->snd_una = ti->ti_ack;
if (SEQ_LT(tp->snd_nxt, tp->snd_una))
tp->snd_nxt = tp->snd_una;
switch (tp->t_state) {
// In FIN_WAIT_1 STATE in addition to the processing
// for the ESTABLISHED state if our FIN is now acknowledged
// then enter FIN_WAIT_2.
case TCPS_FIN_WAIT_1:
if (ourfinisacked) {
// If we can't receive any more
// data, then closing user can proceed.
// Starting the timer is contrary to the
// specification, but if we don't get a FIN
// we'll hang forever.
if (so->so_state & USN_CANTRCVMORE) {
soisdisconnected(so);
tp->t_timer[TCPT_2MSL] = g_tcp_maxidle;
}
DEBUG("change tcp state to TCPS_FIN_WAIT_2, state=%d", tp->t_state);
tp->t_state = TCPS_FIN_WAIT_2;
usnet_tcpin_ewakeup(so, USN_TCP_IN, USN_TCPST_FIN_WAIT2, 0);
}
break;
// In CLOSING STATE in addition to the processing for
// the ESTABLISHED state if the ACK acknowledges our FIN
// then enter the TIME-WAIT state, otherwise ignore
// the segment.
case TCPS_CLOSING:
if (ourfinisacked) {
DEBUG("change tcp state to TCPS_TIME_WAIT, state=%d", tp->t_state);
tp->t_state = TCPS_TIME_WAIT;
usnet_tcpin_ewakeup(so, USN_TCP_IN, USN_TCPST_TIME_WAIT, 0);
tcp_canceltimers(tp);
tp->t_timer[TCPT_2MSL] = 2 * TCPTV_MSL;
soisdisconnected(so);
}
break;
// In LAST_ACK, we may still be waiting for data to drain
// and/or to be acked, as well as for the ack of our FIN.
// If our FIN is now acknowledged, delete the TCB,
// enter the closed state and return.
case TCPS_LAST_ACK:
if (ourfinisacked) {
tp = tcp_close(tp);
goto drop;
}
break;
// In TIME_WAIT state the only thing that should arrive
// is a retransmission of the remote FIN. Acknowledge
// it and restart the finack timer.
case TCPS_TIME_WAIT:
tp->t_timer[TCPT_2MSL] = 2 * TCPTV_MSL;
goto dropafterack;
}
}
step6:
// Update window information.
// Don't look at window if no ACK: TAC's send garbage on first SYN.
if ((tiflags & TH_ACK) &&
(SEQ_LT(tp->snd_wl1, ti->ti_seq) ||
(tp->snd_wl1 == ti->ti_seq && (SEQ_LT(tp->snd_wl2, ti->ti_ack) ||
(tp->snd_wl2 == ti->ti_ack && tiwin > tp->snd_wnd) )) )) {
// keep track of pure window updates
if (ti->ti_len == 0 &&
tp->snd_wl2 == ti->ti_ack && tiwin > tp->snd_wnd)
g_tcpstat.tcps_rcvwinupd++;
tp->snd_wnd = tiwin;
tp->snd_wl1 = ti->ti_seq;
tp->snd_wl2 = ti->ti_ack;
if (tp->snd_wnd > tp->max_sndwnd)
tp->max_sndwnd = tp->snd_wnd;
DEBUG("change needoutput to 1");
tp->t_flags |= TF_NEEDOUTPUT;
needoutput = 1;
}
// Process segments with URG.
if ((tiflags & TH_URG) && ti->ti_urp &&
TCPS_HAVERCVDFIN(tp->t_state) == 0) {
// This is a kludge, but if we receive and accept
// random urgent pointers, we'll crash in
// soreceive. It's hard to imagine someone
// actually wanting to send this much urgent data.
if (ti->ti_urp + so->so_rcv->sb_cc > g_sb_max) {
ti->ti_urp = 0; // XXX
tiflags &= ~TH_URG; // XXX
goto dodata; // XXX
}
// If this segment advances the known urgent pointer,
// then mark the data stream. This should not happen
// in CLOSE_WAIT, CLOSING, LAST_ACK or TIME_WAIT STATES since
// a FIN has been received from the remote side.
// In these states we ignore the URG.
//
// According to RFC961 (Assigned Protocols),
// the urgent pointer points to the last octet
// of urgent data. We continue, however,
// to consider it to indicate the first octet
// of data past the urgent section as the original
// spec states (in one of two places).
if (SEQ_GT(ti->ti_seq+ti->ti_urp, tp->rcv_up)) {
tp->rcv_up = ti->ti_seq + ti->ti_urp;
so->so_oobmark = so->so_rcv->sb_cc +
(tp->rcv_up - tp->rcv_nxt) - 1;
if (so->so_oobmark == 0)
so->so_state |= USN_RCVATMARK;
sohasoutofband(so);
// send async event to app threads.
usnet_tcpin_ewakeup(so, USN_TCP_IN, USN_TCPEV_OUTOFBOUND, 0);
tp->t_oobflags &= ~(TCPOOB_HAVEDATA | TCPOOB_HADDATA);
}
// Remove out of band data so doesn't get presented to user.
// This can happen independent of advancing the URG pointer,
// but if two URG's are pending at once, some out-of-band
// data may creep in... ick.
if (ti->ti_urp <= ti->ti_len
#ifdef SO_OOBINLINE
&& (so->so_options & SO_OOBINLINE) == 0
#endif
)
tcp_pulloutofband(so, ti, m);
} else
// If no out of band data is expected,
// pull receive urgent pointer along
// with the receive window.
if (SEQ_GT(tp->rcv_nxt, tp->rcv_up))
tp->rcv_up = tp->rcv_nxt;
dodata: // XXX
#ifdef DUMP_PAYLOAD
DEBUG("Handle data");
dump_chain(m,"tcp");
#endif
// Process the segment text, merging it into the TCP sequencing queue,
// and arranging for acknowledgment of receipt if necessary.
// This process logically involves adjusting tp->rcv_wnd as data
// is presented to the user (this happens in tcp_usrreq.c,
// case PRU_RCVD). If a FIN has already been received on this
// connection then we just ignore the text.
if ((ti->ti_len || (tiflags&TH_FIN)) &&
TCPS_HAVERCVDFIN(tp->t_state) == 0) {
TCP_REASS(tp, ti, m, so, tiflags);
// Note the amount of data that peer has sent into
// our window, in order to estimate the sender's
// buffer size.
len = so->so_rcv->sb_hiwat - (tp->rcv_adv - tp->rcv_nxt);
} else {
usn_free_cmbuf(m);
tiflags &= ~TH_FIN;
}
// If FIN is received ACK the FIN and let the user know
// that the connection is closing.
if (tiflags & TH_FIN) {
if (TCPS_HAVERCVDFIN(tp->t_state) == 0) {
socantrcvmore(so);
tp->t_flags |= TF_ACKNOW;
TRACE("ack FIN now, tp flags=%d", tp->t_flags);
tp->rcv_nxt++;
}
switch (tp->t_state) {
// In SYN_RECEIVED and ESTABLISHED STATES
// enter the CLOSE_WAIT state.
case TCPS_SYN_RECEIVED:
case TCPS_ESTABLISHED:
TRACE("change tcp state to TCPS_CLOSE_WAIT, state=%d", tp->t_state);
tp->t_state = TCPS_CLOSE_WAIT;
soewakeup(so, 0);
usnet_tcpin_ewakeup(so, USN_TCP_IN, USN_TCPST_CLOSE_WAIT, 0);
break;
// If still in FIN_WAIT_1 STATE FIN has not been acked so
// enter the CLOSING state.
case TCPS_FIN_WAIT_1:
TRACE("change tcp state to TCPS_CLOSING, state=%d", tp->t_state);
tp->t_state = TCPS_CLOSING;
usnet_tcpin_ewakeup(so, USN_TCP_IN, USN_TCPST_CLOSING, 0);
break;
// In FIN_WAIT_2 state enter the TIME_WAIT state,
// starting the time-wait timer, turning off the other
// standard timers.
case TCPS_FIN_WAIT_2:
TRACE("change tcp state to TCPS_TIME_WAIT, state=%d", tp->t_state);
tp->t_state = TCPS_TIME_WAIT;
tcp_canceltimers(tp);
tp->t_timer[TCPT_2MSL] = 2 * TCPTV_MSL;
soisdisconnected(so);
usnet_tcpin_ewakeup(so, USN_TCP_IN, USN_TCPST_TIME_WAIT, 0);
break;
// In TIME_WAIT state restart the 2 MSL time_wait timer.
case TCPS_TIME_WAIT:
tp->t_timer[TCPT_2MSL] = 2 * TCPTV_MSL;
break;
}
}
if (so->so_options & SO_DEBUG) {
TRACE("tcp trace, so_options=%d", so->so_options);
tcp_trace(TA_INPUT, ostate, tp, &g_tcp_saveti, 0);
}
// Return any desired output.
//if (needoutput || (tp->t_flags & TF_ACKNOW)){
if (tp->t_flags & TF_NEEDOUTPUT || (tp->t_flags & TF_ACKNOW)){
TRACE("ack now or need to ouput, tp->t_flags=%d", tp->t_flags);
tcp_output(tp);
}
return;
dropafterack:
TRACE("dropafterack");
// Generate an ACK dropping incoming segment if it occupies
// sequence space, where the ACK reflects our state.
if (tiflags & TH_RST)
goto drop;
usn_free_cmbuf(m);
tp->t_flags |= TF_ACKNOW;
TRACE("ack now, tp flags=%d", tp->t_flags);
tcp_output(tp);
return;
dropwithreset:
TRACE("dropwithreset");
// Generate a RST, dropping incoming segment.
// Make ACK acceptable to originator of segment.
// Don't bother to respond if destination was broadcast/multicast.
#define USN_MULTICAST(i) (((u_int)(i) & 0xf0000000) == 0xe0000000)
if ((tiflags & TH_RST) || m->flags & (BUF_BCAST|BUF_MCAST) ||
USN_MULTICAST(ntohl(ti->ti_dst.s_addr)))
goto drop;
if (tiflags & TH_ACK)
tcp_respond(tp, ti, m, (tcp_seq)0, ti->ti_ack, TH_RST);
else {
if (tiflags & TH_SYN)
ti->ti_len++;
tcp_respond(tp, ti, m, ti->ti_seq+ti->ti_len, (tcp_seq)0,
TH_RST|TH_ACK);
}
// destroy temporarily created socket
if (dropsocket)
soabort(so);
return;
drop:
TRACE("drop");
// Drop space held by incoming segment and return.
if (tp && (tp->t_inpcb->inp_socket->so_options & SO_DEBUG)) {
TRACE("tcp trace: drop a socket");
tcp_trace(TA_DROP, ostate, tp, &g_tcp_saveti, 0);
}
usn_free_cmbuf(m);
// destroy temporarily created socket
if (dropsocket)
soabort(so);
return;
}
static int
natm_usr_connect(struct socket *so, struct sockaddr *nam, struct thread *p)
{
struct natmpcb *npcb;
struct sockaddr_natm *snatm;
struct atmio_openvcc op;
struct ifnet *ifp;
int error = 0;
int proto = so->so_proto->pr_protocol;
npcb = (struct natmpcb *)so->so_pcb;
KASSERT(npcb != NULL, ("natm_usr_connect: npcb == NULL"));
/*
* Validate nam and npcb.
*/
NATM_LOCK();
snatm = (struct sockaddr_natm *)nam;
if (snatm->snatm_len != sizeof(*snatm) ||
(npcb->npcb_flags & NPCB_FREE) == 0) {
NATM_UNLOCK();
return (EINVAL);
}
if (snatm->snatm_family != AF_NATM) {
NATM_UNLOCK();
return (EAFNOSUPPORT);
}
snatm->snatm_if[IFNAMSIZ - 1] = '\0'; /* XXX ensure null termination
since ifunit() uses strcmp */
/*
* Convert interface string to ifp, validate.
*/
ifp = ifunit(snatm->snatm_if);
if (ifp == NULL || (ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
NATM_UNLOCK();
return (ENXIO);
}
if (ifp->if_output != atm_output) {
NATM_UNLOCK();
return (EAFNOSUPPORT);
}
/*
* Register us with the NATM PCB layer.
*/
if (npcb_add(npcb, ifp, snatm->snatm_vci, snatm->snatm_vpi) != npcb) {
NATM_UNLOCK();
return (EADDRINUSE);
}
/*
* Open the channel.
*
* XXXRW: Eventually desirable to hold mutex over ioctl?
*/
bzero(&op, sizeof(op));
op.rxhand = npcb;
op.param.flags = ATMIO_FLAG_PVC;
op.param.vpi = npcb->npcb_vpi;
op.param.vci = npcb->npcb_vci;
op.param.rmtu = op.param.tmtu = ifp->if_mtu;
op.param.aal = (proto == PROTO_NATMAAL5) ? ATMIO_AAL_5 : ATMIO_AAL_0;
op.param.traffic = ATMIO_TRAFFIC_UBR;
NATM_UNLOCK();
if (ifp->if_ioctl == NULL ||
ifp->if_ioctl(ifp, SIOCATMOPENVCC, (caddr_t)&op) != 0)
return (EIO);
soisconnected(so);
return (error);
}
static int
natm_usr_connect(struct socket *so, struct mbuf *nam)
{
struct natmpcb *npcb;
struct sockaddr_natm *snatm;
struct atm_pseudoioctl api;
struct atm_pseudohdr *aph;
struct ifnet *ifp;
int error = 0;
int s2, s = SPLSOFTNET();
int proto = so->so_proto->pr_protocol;
npcb = (struct natmpcb *) so->so_pcb;
if (npcb == NULL) {
error = EINVAL;
goto out;
}
/*
* validate nam and npcb
*/
if (nam->m_len != sizeof(*snatm)) {
error = EINVAL;
goto out;
}
snatm = mtod(nam, struct sockaddr_natm *);
if (snatm->snatm_len != sizeof(*snatm) ||
(npcb->npcb_flags & NPCB_FREE) == 0) {
error = EINVAL;
goto out;
}
if (snatm->snatm_family != AF_NATM) {
error = EAFNOSUPPORT;
goto out;
}
snatm->snatm_if[IFNAMSIZ-1] = '\0'; /* XXX ensure null termination
since ifunit() uses strcmp */
/*
* convert interface string to ifp, validate.
*/
ifp = ifunit(snatm->snatm_if);
if (ifp == NULL || (ifp->if_flags & IFF_RUNNING) == 0) {
error = ENXIO;
goto out;
}
if (ifp->if_output != atm_output) {
error = EAFNOSUPPORT;
goto out;
}
/*
* register us with the NATM PCB layer
*/
if (npcb_add(npcb, ifp, snatm->snatm_vci, snatm->snatm_vpi) != npcb) {
error = EADDRINUSE;
goto out;
}
/*
* enable rx
*/
ATM_PH_FLAGS(&api.aph) = (proto == PROTO_NATMAAL5) ? ATM_PH_AAL5 : 0;
ATM_PH_VPI(&api.aph) = npcb->npcb_vpi;
ATM_PH_SETVCI(&api.aph, npcb->npcb_vci);
api.rxhand = npcb;
s2 = splimp();
if (ifp->if_ioctl == NULL ||
ifp->if_ioctl(ifp, SIOCATMENA, (caddr_t) &api) != 0) {
splx(s2);
npcb_free(npcb, NPCB_REMOVE);
error = EIO;
goto out;
}
splx(s2);
soisconnected(so);
out:
splx(s);
return (error);
}
