static int
tcp_reass(register struct tcpcb *tp, register struct tcpiphdr *ti,
struct mbuf *m)
{
register struct tcpiphdr *q;
struct socket *so = tp->t_socket;
int flags;
/*
* Call with ti==NULL after become established to
* force pre-ESTABLISHED data up to user socket.
*/
if (ti == NULL)
goto present;
/*
* Find a segment which begins after this one does.
*/
for (q = tcpfrag_list_first(tp); !tcpfrag_list_end(q, tp);
q = tcpiphdr_next(q))
if (SEQ_GT(q->ti_seq, ti->ti_seq))
break;
/*
* If there is a preceding segment, it may provide some of
* our data already. If so, drop the data from the incoming
* segment. If it provides all of our data, drop us.
*/
if (!tcpfrag_list_end(tcpiphdr_prev(q), tp)) {
register int i;
q = tcpiphdr_prev(q);
/* conversion to int (in i) handles seq wraparound */
i = q->ti_seq + q->ti_len - ti->ti_seq;
if (i > 0) {
if (i >= ti->ti_len) {
m_free(m);
/*
* Try to present any queued data
* at the left window edge to the user.
* This is needed after the 3-WHS
* completes.
*/
goto present; /* ??? */
}
m_adj(m, i);
ti->ti_len -= i;
ti->ti_seq += i;
}
q = tcpiphdr_next(q);
}
ti->ti_mbuf = m;
/*
* While we overlap succeeding segments trim them or,
* if they are completely covered, dequeue them.
*/
while (!tcpfrag_list_end(q, tp)) {
register int i = (ti->ti_seq + ti->ti_len) - q->ti_seq;
if (i <= 0)
break;
if (i < q->ti_len) {
q->ti_seq += i;
q->ti_len -= i;
m_adj(q->ti_mbuf, i);
break;
}
q = tcpiphdr_next(q);
m = tcpiphdr_prev(q)->ti_mbuf;
remque(tcpiphdr2qlink(tcpiphdr_prev(q)));
m_free(m);
}
/*
* Stick new segment in its place.
*/
insque(tcpiphdr2qlink(ti), tcpiphdr2qlink(tcpiphdr_prev(q)));
present:
/*
* Present data to user, advancing rcv_nxt through
* completed sequence space.
*/
if (!TCPS_HAVEESTABLISHED(tp->t_state))
return (0);
ti = tcpfrag_list_first(tp);
if (tcpfrag_list_end(ti, tp) || ti->ti_seq != tp->rcv_nxt)
return (0);
if (tp->t_state == TCPS_SYN_RECEIVED && ti->ti_len)
return (0);
do {
tp->rcv_nxt += ti->ti_len;
flags = ti->ti_flags & TH_FIN;
remque(tcpiphdr2qlink(ti));
m = ti->ti_mbuf;
ti = tcpiphdr_next(ti);
if (so->so_state & SS_FCANTSENDMORE)
m_free(m);
else {
if (so->so_emu) {
if (tcp_emu(so,m)) sbappend(so, m);
} else
sbappend(so, m);
}
} while (ti != (struct tcpiphdr *)tp && ti->ti_seq == tp->rcv_nxt);
return (flags);
}
void
tcp_timer_keep(void *xtp)
{
struct tcpcb *tp = xtp;
struct tcptemp *t_template;
struct inpcb *inp;
CURVNET_SET(tp->t_vnet);
#ifdef TCPDEBUG
int ostate;
ostate = tp->t_state;
#endif
inp = tp->t_inpcb;
KASSERT(inp != NULL, ("%s: tp %p tp->t_inpcb == NULL", __func__, tp));
INP_WLOCK(inp);
if (callout_pending(&tp->t_timers->tt_keep) ||
!callout_active(&tp->t_timers->tt_keep)) {
INP_WUNLOCK(inp);
CURVNET_RESTORE();
return;
}
callout_deactivate(&tp->t_timers->tt_keep);
if ((inp->inp_flags & INP_DROPPED) != 0) {
INP_WUNLOCK(inp);
CURVNET_RESTORE();
return;
}
KASSERT((tp->t_timers->tt_flags & TT_STOPPED) == 0,
("%s: tp %p tcpcb can't be stopped here", __func__, tp));
/*
* Because we don't regularly reset the keepalive callout in
* the ESTABLISHED state, it may be that we don't actually need
* to send a keepalive yet. If that occurs, schedule another
* call for the next time the keepalive timer might expire.
*/
if (TCPS_HAVEESTABLISHED(tp->t_state)) {
u_int idletime;
idletime = ticks - tp->t_rcvtime;
if (idletime < TP_KEEPIDLE(tp)) {
callout_reset(&tp->t_timers->tt_keep,
TP_KEEPIDLE(tp) - idletime, tcp_timer_keep, tp);
INP_WUNLOCK(inp);
CURVNET_RESTORE();
return;
}
}
/*
* Keep-alive timer went off; send something
* or drop connection if idle for too long.
*/
TCPSTAT_INC(tcps_keeptimeo);
if (tp->t_state < TCPS_ESTABLISHED)
goto dropit;
if ((always_keepalive || inp->inp_socket->so_options & SO_KEEPALIVE) &&
tp->t_state <= TCPS_CLOSING) {
if (ticks - tp->t_rcvtime >= TP_KEEPIDLE(tp) + TP_MAXIDLE(tp))
goto dropit;
/*
* Send a packet designed to force a response
* if the peer is up and reachable:
* either an ACK if the connection is still alive,
* or an RST if the peer has closed the connection
* due to timeout or reboot.
* Using sequence number tp->snd_una-1
* causes the transmitted zero-length segment
* to lie outside the receive window;
* by the protocol spec, this requires the
* correspondent TCP to respond.
*/
TCPSTAT_INC(tcps_keepprobe);
t_template = tcpip_maketemplate(inp);
if (t_template) {
tcp_respond(tp, t_template->tt_ipgen,
&t_template->tt_t, (struct mbuf *)NULL,
tp->rcv_nxt, tp->snd_una - 1, 0);
free(t_template, M_TEMP);
}
callout_reset(&tp->t_timers->tt_keep, TP_KEEPINTVL(tp),
tcp_timer_keep, tp);
} else
callout_reset(&tp->t_timers->tt_keep, TP_KEEPIDLE(tp),
tcp_timer_keep, tp);
#ifdef TCPDEBUG
if (inp->inp_socket->so_options & SO_DEBUG)
tcp_trace(TA_USER, ostate, tp, (void *)0, (struct tcphdr *)0,
PRU_SLOWTIMO);
#endif
TCP_PROBE2(debug__user, tp, PRU_SLOWTIMO);
INP_WUNLOCK(inp);
CURVNET_RESTORE();
return;
dropit:
TCPSTAT_INC(tcps_keepdrops);
if (tcp_inpinfo_lock_add(inp)) {
tcp_inpinfo_lock_del(inp, tp);
goto out;
}
tp = tcp_drop(tp, ETIMEDOUT);
#ifdef TCPDEBUG
if (tp != NULL && (tp->t_inpcb->inp_socket->so_options & SO_DEBUG))
tcp_trace(TA_USER, ostate, tp, (void *)0, (struct tcphdr *)0,
PRU_SLOWTIMO);
#endif
TCP_PROBE2(debug__user, tp, PRU_SLOWTIMO);
tcp_inpinfo_lock_del(inp, tp);
out:
CURVNET_RESTORE();
}
int
tcp_reass(struct tcpcb *tp, struct tcphdr *th, int *tlenp, struct mbuf *m)
{
struct socket *so = tp->t_inpcb->inp_socket;
struct mbuf *mq, *mp;
int flags, wakeup;
INP_WLOCK_ASSERT(tp->t_inpcb);
/*
* XXX: tcp_reass() is rather inefficient with its data structures
* and should be rewritten (see NetBSD for optimizations).
*/
/*
* Call with th==NULL after become established to
* force pre-ESTABLISHED data up to user socket.
*/
if (th == NULL)
goto present;
M_ASSERTPKTHDR(m);
KASSERT(*tlenp == m->m_pkthdr.len, ("%s: tlenp %u len %u", __func__,
*tlenp, m->m_pkthdr.len));
/*
* Limit the number of segments that can be queued to reduce the
* potential for mbuf exhaustion. For best performance, we want to be
* able to queue a full window's worth of segments. The size of the
* socket receive buffer determines our advertised window and grows
* automatically when socket buffer autotuning is enabled. Use it as the
* basis for our queue limit.
* Always let the missing segment through which caused this queue.
* NB: Access to the socket buffer is left intentionally unlocked as we
* can tolerate stale information here.
*/
if ((th->th_seq != tp->rcv_nxt || !TCPS_HAVEESTABLISHED(tp->t_state)) &&
tp->t_segqlen + m->m_pkthdr.len >= sbspace(&so->so_rcv)) {
char *s;
TCPSTAT_INC(tcps_rcvreassfull);
*tlenp = 0;
if ((s = tcp_log_addrs(&tp->t_inpcb->inp_inc, th, NULL,
NULL))) {
log(LOG_DEBUG, "%s; %s: queue limit reached, "
"segment dropped\n", s, __func__);
free(s, M_TCPLOG);
}
m_freem(m);
return (0);
}
/*
* Find a segment which begins after this one does.
*/
mp = NULL;
for (mq = tp->t_segq; mq != NULL; mq = mq->m_nextpkt) {
if (SEQ_GT(M_TCPHDR(mq)->th_seq, th->th_seq))
break;
mp = mq;
}
/*
* If there is a preceding segment, it may provide some of
* our data already. If so, drop the data from the incoming
* segment. If it provides all of our data, drop us.
*/
if (mp != NULL) {
int i;
/* conversion to int (in i) handles seq wraparound */
i = M_TCPHDR(mp)->th_seq + mp->m_pkthdr.len - th->th_seq;
if (i > 0) {
if (i >= *tlenp) {
TCPSTAT_INC(tcps_rcvduppack);
TCPSTAT_ADD(tcps_rcvdupbyte, *tlenp);
m_freem(m);
/*
* Try to present any queued data
* at the left window edge to the user.
* This is needed after the 3-WHS
* completes.
*/
goto present; /* ??? */
}
m_adj(m, i);
*tlenp -= i;
th->th_seq += i;
}
}
tp->t_rcvoopack++;
TCPSTAT_INC(tcps_rcvoopack);
TCPSTAT_ADD(tcps_rcvoobyte, *tlenp);
/*
* While we overlap succeeding segments trim them or,
* if they are completely covered, dequeue them.
*/
while (mq) {
struct mbuf *nq;
int i;
i = (th->th_seq + *tlenp) - M_TCPHDR(mq)->th_seq;
if (i <= 0)
break;
if (i < mq->m_pkthdr.len) {
M_TCPHDR(mq)->th_seq += i;
m_adj(mq, i);
tp->t_segqlen -= i;
break;
}
nq = mq->m_nextpkt;
tp->t_segqlen -= mq->m_pkthdr.len;
m_freem(mq);
if (mp)
mp->m_nextpkt = nq;
else
tp->t_segq = nq;
mq = nq;
}
/*
* Insert the new segment queue entry into place. Try to collapse
* mbuf chains if segments are adjacent.
*/
if (mp) {
if (M_TCPHDR(mp)->th_seq + mp->m_pkthdr.len == th->th_seq)
m_catpkt(mp, m);
else {
m->m_nextpkt = mp->m_nextpkt;
mp->m_nextpkt = m;
m->m_pkthdr.pkt_tcphdr = th;
}
} else {
mq = tp->t_segq;
tp->t_segq = m;
if (mq && th->th_seq + *tlenp == M_TCPHDR(mq)->th_seq) {
m->m_nextpkt = mq->m_nextpkt;
mq->m_nextpkt = NULL;
m_catpkt(m, mq);
} else
m->m_nextpkt = mq;
m->m_pkthdr.pkt_tcphdr = th;
}
tp->t_segqlen += *tlenp;
present:
/*
* Present data to user, advancing rcv_nxt through
* completed sequence space.
*/
if (!TCPS_HAVEESTABLISHED(tp->t_state))
return (0);
flags = 0;
wakeup = 0;
SOCKBUF_LOCK(&so->so_rcv);
while ((mq = tp->t_segq) != NULL &&
M_TCPHDR(mq)->th_seq == tp->rcv_nxt) {
tp->t_segq = mq->m_nextpkt;
tp->rcv_nxt += mq->m_pkthdr.len;
tp->t_segqlen -= mq->m_pkthdr.len;
flags = M_TCPHDR(mq)->th_flags & TH_FIN;
if (so->so_rcv.sb_state & SBS_CANTRCVMORE)
m_freem(mq);
else {
mq->m_nextpkt = NULL;
sbappendstream_locked(&so->so_rcv, mq, 0);
wakeup = 1;
}
}
ND6_HINT(tp);
if (wakeup)
sorwakeup_locked(so);
else
SOCKBUF_UNLOCK(&so->so_rcv);
return (flags);
}
void
tcp_timer_keep(void *arg)
{
struct tcpcb *tp = arg;
struct socket *so = NULL; /* Quell compiler warning */
#ifdef TCP_DEBUG
short ostate;
#endif
mutex_enter(softnet_lock);
if ((tp->t_flags & TF_DEAD) != 0) {
mutex_exit(softnet_lock);
return;
}
if (!callout_expired(&tp->t_timer[TCPT_KEEP])) {
mutex_exit(softnet_lock);
return;
}
KERNEL_LOCK(1, NULL);
#ifdef TCP_DEBUG
ostate = tp->t_state;
#endif /* TCP_DEBUG */
/*
* Keep-alive timer went off; send something
* or drop connection if idle for too long.
*/
TCP_STATINC(TCP_STAT_KEEPTIMEO);
if (TCPS_HAVEESTABLISHED(tp->t_state) == 0)
goto dropit;
#ifdef INET
if (tp->t_inpcb)
so = tp->t_inpcb->inp_socket;
#endif
#ifdef INET6
if (tp->t_in6pcb)
so = tp->t_in6pcb->in6p_socket;
#endif
KASSERT(so != NULL);
if (so->so_options & SO_KEEPALIVE &&
tp->t_state <= TCPS_CLOSE_WAIT) {
if ((tp->t_maxidle > 0) &&
((tcp_now - tp->t_rcvtime) >=
tp->t_keepidle + tp->t_maxidle))
goto dropit;
/*
* Send a packet designed to force a response
* if the peer is up and reachable:
* either an ACK if the connection is still alive,
* or an RST if the peer has closed the connection
* due to timeout or reboot.
* Using sequence number tp->snd_una-1
* causes the transmitted zero-length segment
* to lie outside the receive window;
* by the protocol spec, this requires the
* correspondent TCP to respond.
*/
TCP_STATINC(TCP_STAT_KEEPPROBE);
if (tcp_compat_42) {
/*
* The keepalive packet must have nonzero
* length to get a 4.2 host to respond.
*/
(void)tcp_respond(tp, tp->t_template,
NULL, NULL, tp->rcv_nxt - 1,
tp->snd_una - 1, 0);
} else {
(void)tcp_respond(tp, tp->t_template,
NULL, NULL, tp->rcv_nxt,
tp->snd_una - 1, 0);
}
TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepintvl);
} else
TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepidle);
#ifdef TCP_DEBUG
if (tp && so->so_options & SO_DEBUG)
tcp_trace(TA_USER, ostate, tp, NULL,
PRU_SLOWTIMO | (TCPT_KEEP << 8));
#endif
KERNEL_UNLOCK_ONE(NULL);
mutex_exit(softnet_lock);
return;
dropit:
TCP_STATINC(TCP_STAT_KEEPDROPS);
(void) tcp_drop(tp, ETIMEDOUT);
KERNEL_UNLOCK_ONE(NULL);
mutex_exit(softnet_lock);
}
int tcp_reass(struct tcpcb *tp, struct tcpiphdr *ti, struct mbuf *m)
{
register struct tcpiphdr *q;
struct socket *so = tp->t_socket;
int flags;
/*
* Call with ti==0 after become established to
* force pre-ESTABLISHED data up to user socket.
*/
if (ti == 0)
goto present;
/*
* Find a segment which begins after this one does.
*/
for (q = (struct tcpiphdr *)tp->seg_next; q != (struct tcpiphdr *)tp;
q = (struct tcpiphdr *)q->ti_next)
if (SEQ_GT(q->ti_seq, ti->ti_seq))
break;
/*
* If there is a preceding segment, it may provide some of
* our data already. If so, drop the data from the incoming
* segment. If it provides all of our data, drop us.
*/
if ((struct tcpiphdr *)q->ti_prev != (struct tcpiphdr *)tp) {
register int i;
q = (struct tcpiphdr *)q->ti_prev;
/* conversion to int (in i) handles seq wraparound */
i = q->ti_seq + q->ti_len - ti->ti_seq;
if (i > 0) {
if (i >= ti->ti_len) {
tcpstat.tcps_rcvduppack++;
tcpstat.tcps_rcvdupbyte += ti->ti_len;
m_freem(m);
/*
* Try to present any queued data
* at the left window edge to the user.
* This is needed after the 3-WHS
* completes.
*/
goto present; /* ??? */
}
m_adj(m, i);
ti->ti_len -= i;
ti->ti_seq += i;
}
q = (struct tcpiphdr *)(q->ti_next);
}
tcpstat.tcps_rcvoopack++;
tcpstat.tcps_rcvoobyte += ti->ti_len;
REASS_MBUF(ti) = (mbufp_32) m; /* XXX */
/*
* While we overlap succeeding segments trim them or,
* if they are completely covered, dequeue them.
*/
while (q != (struct tcpiphdr *)tp) {
register int i = (ti->ti_seq + ti->ti_len) - q->ti_seq;
if (i <= 0)
break;
if (i < q->ti_len) {
q->ti_seq += i;
q->ti_len -= i;
m_adj((struct mbuf *) REASS_MBUF(q), i);
break;
}
q = (struct tcpiphdr *)q->ti_next;
m = (struct mbuf *) REASS_MBUF((struct tcpiphdr *)q->ti_prev);
slirp_remque((void *)(q->ti_prev));
m_freem(m);
}
/*
* Stick new segment in its place.
*/
slirp_insque(ti, (void *)(q->ti_prev));
present:
/*
* Present data to user, advancing rcv_nxt through
* completed sequence space.
*/
if (!TCPS_HAVEESTABLISHED(tp->t_state))
return (0);
ti = (struct tcpiphdr *) tp->seg_next;
if (ti == (struct tcpiphdr *)tp || ti->ti_seq != tp->rcv_nxt)
return (0);
if (tp->t_state == TCPS_SYN_RECEIVED && ti->ti_len)
return (0);
do {
tp->rcv_nxt += ti->ti_len;
flags = ti->ti_flags & TH_FIN;
slirp_remque(ti);
m = (struct mbuf *) REASS_MBUF(ti); /* XXX */
ti = (struct tcpiphdr *)ti->ti_next;
/* if (so->so_state & SS_FCANTRCVMORE) */
if (so->so_state & SS_FCANTSENDMORE)
m_freem(m);
else {
if (so->so_emu) {
if (tcp_emu(so,m)) sbappend(so, m);
} else
sbappend(so, m);
}
} while (ti != (struct tcpiphdr *)tp && ti->ti_seq == tp->rcv_nxt);
/* sorwakeup(so); */
return (flags);
}