static int
uipc_rcvd(struct socket *so, int flags)
{
struct unpcb *unp = sotounpcb(so);
struct socket *so2;
if (unp == 0)
return EINVAL;
switch (so->so_type) {
case SOCK_DGRAM:
panic("uipc_rcvd DGRAM?");
/*NOTREACHED*/
case SOCK_STREAM:
if (unp->unp_conn == 0)
break;
so2 = unp->unp_conn->unp_socket;
/*
* Adjust backpressure on sender
* and wakeup any waiting to write.
*/
so2->so_snd.sb_mbmax += unp->unp_mbcnt - so->so_rcv.sb_mbcnt;
unp->unp_mbcnt = so->so_rcv.sb_mbcnt;
so2->so_snd.sb_hiwat += unp->unp_cc - so->so_rcv.sb_cc;
(void)chgsbsize(so2->so_cred->cr_uid,
(rlim_t)unp->unp_cc - so->so_rcv.sb_cc);
unp->unp_cc = so->so_rcv.sb_cc;
sowwakeup(so2);
break;
default:
panic("uipc_rcvd unknown socktype");
}
return 0;
}
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);
}
}
static void
sdp_nagle_timeout(void *data)
{
struct sdp_sock *ssk = (struct sdp_sock *)data;
struct socket *sk = ssk->socket;
sdp_dbg_data(sk, "last_unacked = %ld\n", ssk->nagle_last_unacked);
if (!callout_active(&ssk->nagle_timer))
return;
callout_deactivate(&ssk->nagle_timer);
if (!ssk->nagle_last_unacked)
goto out;
if (ssk->state == TCPS_CLOSED)
return;
ssk->nagle_last_unacked = 0;
sdp_post_sends(ssk, M_NOWAIT);
sowwakeup(ssk->socket);
out:
if (sk->so_snd.sb_sndptr)
callout_reset(&ssk->nagle_timer, SDP_NAGLE_TIMEOUT,
sdp_nagle_timeout, ssk);
}
/*
* 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);
}
void
socantsendmore(struct socket *so)
{
so->so_state |= SS_CANTSENDMORE;
sowwakeup(so);
}
static int
sdp_process_tx_cq(struct sdp_sock *ssk)
{
struct ib_wc ibwc[SDP_NUM_WC];
int n, i;
int wc_processed = 0;
SDP_WLOCK_ASSERT(ssk);
if (!ssk->tx_ring.cq) {
sdp_dbg(ssk->socket, "tx irq on destroyed tx_cq\n");
return 0;
}
do {
n = ib_poll_cq(ssk->tx_ring.cq, SDP_NUM_WC, ibwc);
for (i = 0; i < n; ++i) {
sdp_process_tx_wc(ssk, ibwc + i);
wc_processed++;
}
} while (n == SDP_NUM_WC);
if (wc_processed) {
sdp_post_sends(ssk, M_DONTWAIT);
sdp_prf1(sk, NULL, "Waking sendmsg. inflight=%d",
(u32) tx_ring_posted(ssk));
sowwakeup(ssk->socket);
}
return wc_processed;
}
/*
* socantsendmore: indicates that no more data will be sent on the
* socket; it would normally be applied to a socket when the user
* informs the system that no more data is to be sent, by the protocol
* code (in case pr_shutdown()).
*/
void
socantsendmore(struct socket *so)
{
KASSERT(solocked(so));
so->so_state |= SS_CANTSENDMORE;
sowwakeup(so);
}
/*
* rfcomm_complete(rfcomm_dlc, length)
*
* length bytes are sent and may be removed from socket buffer
*/
static void
rfcomm_complete(void *arg, int length)
{
struct socket *so = arg;
sbdrop(&so->so_snd, length);
sowwakeup(so);
}
/* Wakeup writers if we now have credits. */
static void
sdp_bzcopy_write_space(struct sdp_sock *ssk)
{
struct socket *sk = ssk->socket;
if (tx_credits(ssk) >= ssk->min_bufs && sk)
sowwakeup(sk);
}
static void
l2cap_complete(void *arg, int count)
{
struct socket *so = arg;
while (count-- > 0)
sbdroprecord(&so->so_snd);
sowwakeup(so);
}
static void
sco_complete(void *arg, int num)
{
struct socket *so = arg;
while (num-- > 0)
sbdroprecord(&so->so_snd);
sowwakeup(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);
}
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
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);
}
static inline void
sdp_process_tx_wc(struct sdp_sock *ssk, struct ib_wc *wc)
{
if (likely(wc->wr_id & SDP_OP_SEND)) {
sdp_handle_send_comp(ssk, wc);
return;
}
#ifdef SDP_ZCOPY
if (wc->wr_id & SDP_OP_RDMA) {
/* TODO: handle failed RDMA read cqe */
sdp_dbg_data(ssk->socket,
"TX comp: RDMA read. status: %d\n", wc->status);
sdp_prf1(sk, NULL, "TX comp: RDMA read");
if (!ssk->tx_ring.rdma_inflight) {
sdp_warn(ssk->socket, "ERROR: unexpected RDMA read\n");
return;
}
if (!ssk->tx_ring.rdma_inflight->busy) {
sdp_warn(ssk->socket,
"ERROR: too many RDMA read completions\n");
return;
}
/* Only last RDMA read WR is signalled. Order is guaranteed -
* therefore if Last RDMA read WR is completed - all other
* have, too */
ssk->tx_ring.rdma_inflight->busy = 0;
sowwakeup(ssk->socket);
sdp_dbg_data(ssk->socket, "woke up sleepers\n");
return;
}
#endif
/* Keepalive probe sent cleanup */
sdp_cnt(sdp_keepalive_probes_sent);
if (likely(!wc->status))
return;
sdp_dbg(ssk->socket, " %s consumes KEEPALIVE status %d\n",
__func__, wc->status);
if (wc->status == IB_WC_WR_FLUSH_ERR)
return;
sdp_notify(ssk, ECONNRESET);
}
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);
}
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);
}
}
/*
* 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 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);
}
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);
}
}
/*
* 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;
}
void
so_icmpdu(PACKET p, struct destun * pdp)
{
ip_addr lhost; /* IP address of originator (our iface) */
ip_addr fhost; /* IP address we sent to */
unshort fport; /* TCP/UDP port we sent to */
unshort lport; /* TCP/UDP port we sent from */
struct inpcb * inp;
struct socket * so;
struct tcpcb * tp;
/* extract information about packet which generated DU */
fhost = htonl(pdp->dip.ip_dest);
lhost = htonl(pdp->dip.ip_src);
lport = htons(*(unshort*)(&pdp->ddata[0]));
fport = htons(*(unshort*)(&pdp->ddata[2]));
#ifndef IP_PMTU
/* if it's a datagram-too-big message, ignore it -- As the
* build isn't using PMTU Discovery this packet is most
* probably a Denial of Service Attack.
*/
if(pdp->dcode == DSTFRAG)
{
goto done;
}
#endif /* IP_PMTU */
/* if it's a TCP connection, clean it up */
if (pdp->dip.ip_prot == TCPTP)
{
/* find associated data structs and socket */
inp = in_pcblookup(&tcb, fhost, fport, lhost, lport, INPLOOKUP_WILDCARD);
if (inp == 0)
goto done;
so = inp->inp_socket;
if (so == 0)
goto done;
tp = intotcpcb(inp);
if (tp)
{
if (tp->t_state <= TCPS_LISTEN)
{
goto done;
}
#ifdef ICMP_TCP_DOS
{
struct ip * pip;
struct tcpiphdr * ti;
pip = ip_head(p); /* find IP header */
ti = (struct tcpiphdr *)p->nb_prot;
if(!((tp->snd_una <= ti->ti_seq) && (ti->ti_seq <= tp->snd_nxt)))
goto done;
/* If we get an ICMP Type 3 (Destination Unreachable) - Code 2
* (Protocol Unreachable) message and during the life of a TCP
* connection, then its most probably a Denial of Service Attack.
* As the only other interpretation would be that the support for
* the transport protocol has been removed from the host sending
* the error message during the life of the corresponding
* connection. As in common practice this is higly unlikely in most
* cases, we will treat this message as a DOS attack.
*/
if(pdp->dcode == DSTPROT)
{
if((tp->t_state >= TCPS_ESTABLISHED) && (tp->t_state <= TCPS_TIME_WAIT))
goto done;
}
/* Note some ICMP error messages generated by intermediate routers,
* include more than the recommended 64 bits of the IP Data. If the
* TCP ACK number happens to be present then use it in detecting a
* Denial of Service attack.
*
* This way we can ensure that the TCP Acknowledgement number should
* correspond to data that have already been acknowledged. This way
* we can further reduce the possiblity of considering a spoofed ICMP
* packet by a factor of 2.
*/
if(pip->ip_len >= 32)
{
if(!(ti->ti_seq <= tp->rcv_nxt))
goto done;
}
}
#endif
tcp_close(tp);
}
so->so_error = ECONNREFUSED; /* set error for socket owner */
}
#ifdef UDP_SOCKETS /* this sockets layer supports UDP too */
else if(pdp->dip.ip_prot == UDP_PROT)
{
UDPCONN tmp;
/* search udp table (which keeps hosts in net endian) */
for (tmp = firstudp; tmp; tmp = tmp->u_next)
if ((tmp->u_fport == fport || tmp->u_fport == 0) &&
(tmp->u_fhost == htonl(fhost)) &&
(tmp->u_lport == lport))
{
break; /* found our UDP table entry */
}
if (!tmp)
goto done;
so = (struct socket *)tmp->u_data;
/* May be non-socket (lightweight) UDP connection. */
if (so->so_type != SOCK_DGRAM)
goto done;
so->so_error = ECONNREFUSED; /* set error for socket owner */
/* do a select() notify on socket here */
sorwakeup(so);
sowwakeup(so);
}
#endif /* UDP_SOCKETS */
else
goto done;
#ifdef IP_PMTU
/* if this is a datagram-too-big message, update the Path MTU cache */
if (pdp->dcode == DSTFRAG)
pmtucache_set(pdp->dip.ip_dest, htons(pdp->dno2));
#endif /* IP_PMTU */
done:
LOCK_NET_RESOURCE(FREEQ_RESID);
pk_free(p); /* done with original packet */
UNLOCK_NET_RESOURCE(FREEQ_RESID);
return;
}
/*
* Do a send by putting data in output queue and updating urgent
* marker if URG set. Possibly send more data. Unlike the other
* pru_*() routines, the mbuf chains are our responsibility. We
* must either enqueue them or free them. The other pru_* routines
* generally are caller-frees.
*/
static void
tcp_usr_send(netmsg_t msg)
{
struct socket *so = msg->send.base.nm_so;
int flags = msg->send.nm_flags;
struct mbuf *m = msg->send.nm_m;
int error = 0;
struct inpcb *inp;
struct tcpcb *tp;
TCPDEBUG0;
KKASSERT(msg->send.nm_control == NULL);
KKASSERT(msg->send.nm_addr == NULL);
KKASSERT((flags & PRUS_FREEADDR) == 0);
inp = so->so_pcb;
if (inp == NULL) {
/*
* OOPS! we lost a race, the TCP session got reset after
* we checked SS_CANTSENDMORE, eg: while doing uiomove or a
* network interrupt in the non-critical section of sosend().
*/
m_freem(m);
error = ECONNRESET; /* XXX EPIPE? */
tp = NULL;
TCPDEBUG1();
goto out;
}
tp = intotcpcb(inp);
TCPDEBUG1();
#ifdef foo
/*
* This is no longer necessary, since:
* - sosendtcp() has already checked it for us
* - It does not work with asynchronized send
*/
/*
* Don't let too much OOB data build up
*/
if (flags & PRUS_OOB) {
if (ssb_space(&so->so_snd) < -512) {
m_freem(m);
error = ENOBUFS;
goto out;
}
}
#endif
/*
* Pump the data into the socket.
*/
if (m) {
ssb_appendstream(&so->so_snd, m);
sowwakeup(so);
}
if (flags & PRUS_OOB) {
/*
* 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.
* Otherwise, snd_up should be one lower.
*/
tp->snd_up = tp->snd_una + so->so_snd.ssb_cc;
tp->t_flags |= TF_FORCE;
error = tcp_output(tp);
tp->t_flags &= ~TF_FORCE;
} else {
if (flags & PRUS_EOF) {
/*
* Close the send side of the connection after
* the data is sent.
*/
socantsendmore(so);
tp = tcp_usrclosed(tp);
}
if (tp != NULL && !tcp_output_pending(tp)) {
if (flags & PRUS_MORETOCOME)
tp->t_flags |= TF_MORETOCOME;
error = tcp_output_fair(tp);
if (flags & PRUS_MORETOCOME)
tp->t_flags &= ~TF_MORETOCOME;
}
}
COMMON_END1((flags & PRUS_OOB) ? PRU_SENDOOB :
((flags & PRUS_EOF) ? PRU_SEND_EOF : PRU_SEND),
(flags & PRUS_NOREPLY));
}
