void
tcp_timer_delack(void *xtp)
{
struct tcpcb *tp = xtp;
struct inpcb *inp;
CURVNET_SET(tp->t_vnet);
inp = tp->t_inpcb;
/*
* XXXRW: While this assert is in fact correct, bugs in the tcpcb
* tear-down mean we need it as a work-around for races between
* timers and tcp_discardcb().
*
* KASSERT(inp != NULL, ("tcp_timer_delack: inp == NULL"));
*/
if (inp == NULL) {
tcp_timer_race++;
CURVNET_RESTORE();
return;
}
INP_WLOCK(inp);
if ((inp->inp_flags & INP_DROPPED) || callout_pending(&tp->t_timers->tt_delack)
|| !callout_active(&tp->t_timers->tt_delack)) {
INP_WUNLOCK(inp);
CURVNET_RESTORE();
return;
}
callout_deactivate(&tp->t_timers->tt_delack);
tp->t_flags |= TF_ACKNOW;
TCPSTAT_INC(tcps_delack);
(void) tcp_output(tp);
INP_WUNLOCK(inp);
CURVNET_RESTORE();
}
void
tcp_timer_delack(void *xtp)
{
struct tcpcb *tp = xtp;
struct inpcb *inp;
CURVNET_SET(tp->t_vnet);
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_delack) ||
!callout_active(&tp->t_timers->tt_delack)) {
INP_WUNLOCK(inp);
CURVNET_RESTORE();
return;
}
callout_deactivate(&tp->t_timers->tt_delack);
if ((inp->inp_flags & INP_DROPPED) != 0) {
INP_WUNLOCK(inp);
CURVNET_RESTORE();
return;
}
tp->t_flags |= TF_ACKNOW;
TCPSTAT_INC(tcps_delack);
(void) tp->t_fb->tfb_tcp_output(tp);
INP_WUNLOCK(inp);
CURVNET_RESTORE();
}
void
tcp_twclose(struct tcptw *tw, int reuse)
{
struct socket *so;
struct inpcb *inp;
/*
* At this point, we are in one of two situations:
*
* (1) We have no socket, just an inpcb<->twtcp pair. We can free
* all state.
*
* (2) We have a socket -- if we own a reference, release it and
* notify the socket layer.
*/
inp = tw->tw_inpcb;
//ScenSim-Port// KASSERT((inp->inp_flags & INP_TIMEWAIT), ("tcp_twclose: !timewait"));
//ScenSim-Port// KASSERT(intotw(inp) == tw, ("tcp_twclose: inp_ppcb != tw"));
//ScenSim-Port// INP_INFO_WLOCK_ASSERT(&V_tcbinfo); /* tcp_tw_2msl_stop(). */
//ScenSim-Port// INP_WLOCK_ASSERT(inp);
tw->tw_inpcb = NULL;
tcp_tw_2msl_stop(tw);
inp->inp_ppcb = NULL;
in_pcbdrop(inp);
so = inp->inp_socket;
if (so != NULL) {
/*
* If there's a socket, handle two cases: first, we own a
* strong reference, which we will now release, or we don't
* in which case another reference exists (XXXRW: think
* about this more), and we don't need to take action.
*/
if (inp->inp_flags & INP_SOCKREF) {
inp->inp_flags &= ~INP_SOCKREF;
//ScenSim-Port// INP_WUNLOCK(inp);
//ScenSim-Port// ACCEPT_LOCK();
//ScenSim-Port// SOCK_LOCK(so);
//ScenSim-Port// KASSERT(so->so_state & SS_PROTOREF,
//ScenSim-Port// ("tcp_twclose: INP_SOCKREF && !SS_PROTOREF"));
so->so_state &= ~SS_PROTOREF;
sofree(so);
} else {
/*
* If we don't own the only reference, the socket and
* inpcb need to be left around to be handled by
* tcp_usr_detach() later.
*/
//ScenSim-Port// INP_WUNLOCK(inp);
}
} else
in_pcbfree(inp);
TCPSTAT_INC(tcps_closed);
//ScenSim-Port// crfree(tw->tw_cred);
//ScenSim-Port// tw->tw_cred = NULL;
if (reuse)
return;
uma_zfree(V_tcptw_zone, tw);
}
/*
* Common subroutine to open a TCP connection to remote host specified
* by struct sockaddr_in in mbuf *nam. Call in_pcbbind to assign a local
* port number if needed. Call in_pcbconnect_setup to do the routing and
* to choose a local host address (interface). If there is an existing
* incarnation of the same connection in TIME-WAIT state and if the remote
* host was sending CC options and if the connection duration was < MSL, then
* truncate the previous TIME-WAIT state and proceed.
* Initialize connection parameters and enter SYN-SENT state.
*/
static int
tcp_connect(struct tcpcb *tp, struct sockaddr *nam, struct thread *td)
{
struct inpcb *inp = tp->t_inpcb, *oinp;
struct socket *so = inp->inp_socket;
struct in_addr laddr;
u_short lport;
int error;
INP_INFO_WLOCK_ASSERT(&V_tcbinfo);
INP_WLOCK_ASSERT(inp);
if (inp->inp_lport == 0) {
error = in_pcbbind(inp, (struct sockaddr *)0, td->td_ucred);
if (error)
return error;
}
/*
* Cannot simply call in_pcbconnect, because there might be an
* earlier incarnation of this same connection still in
* TIME_WAIT state, creating an ADDRINUSE error.
*/
laddr = inp->inp_laddr;
lport = inp->inp_lport;
error = in_pcbconnect_setup(inp, nam, &laddr.s_addr, &lport,
&inp->inp_faddr.s_addr, &inp->inp_fport, &oinp, td->td_ucred);
if (error && oinp == NULL)
return error;
if (oinp)
return EADDRINUSE;
inp->inp_laddr = laddr;
in_pcbrehash(inp);
/*
* Compute window scaling to request:
* Scale to fit into sweet spot. See tcp_syncache.c.
* XXX: This should move to tcp_output().
*/
while (tp->request_r_scale < TCP_MAX_WINSHIFT &&
(TCP_MAXWIN << tp->request_r_scale) < sb_max)
tp->request_r_scale++;
soisconnecting(so);
TCPSTAT_INC(tcps_connattempt);
tp->t_state = TCPS_SYN_SENT;
tcp_timer_activate(tp, TT_KEEP, tcp_keepinit);
tp->iss = tcp_new_isn(tp);
tp->t_bw_rtseq = tp->iss;
tcp_sendseqinit(tp);
return 0;
}
/*
* Allocate struct sackhole.
*/
static struct sackhole *
tcp_sackhole_alloc(struct tcpcb *tp, tcp_seq start, tcp_seq end)
{
struct sackhole *hole;
if (tp->snd_numholes >= V_tcp_sack_maxholes ||
V_tcp_sack_globalholes >= V_tcp_sack_globalmaxholes) {
TCPSTAT_INC(tcps_sack_sboverflow);
return NULL;
}
hole = (struct sackhole *)uma_zalloc(V_sack_hole_zone, M_NOWAIT);
if (hole == NULL)
return NULL;
hole->start = start;
hole->end = end;
hole->rxmit = start;
tp->snd_numholes++;
atomic_add_int(&V_tcp_sack_globalholes, 1);
return hole;
}
void
tcp_timer_rexmt(void * xtp)
{
struct tcpcb *tp = xtp;
CURVNET_SET(tp->t_vnet);
int rexmt;
struct inpcb *inp;
#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_rexmt) ||
!callout_active(&tp->t_timers->tt_rexmt)) {
INP_WUNLOCK(inp);
CURVNET_RESTORE();
return;
}
callout_deactivate(&tp->t_timers->tt_rexmt);
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));
tcp_free_sackholes(tp);
if (tp->t_fb->tfb_tcp_rexmit_tmr) {
/* The stack has a timer action too. */
(*tp->t_fb->tfb_tcp_rexmit_tmr)(tp);
}
/*
* Retransmission timer went off. Message has not
* been acked within retransmit interval. Back off
* to a longer retransmit interval and retransmit one segment.
*/
if (++tp->t_rxtshift > TCP_MAXRXTSHIFT) {
tp->t_rxtshift = TCP_MAXRXTSHIFT;
TCPSTAT_INC(tcps_timeoutdrop);
if (tcp_inpinfo_lock_add(inp)) {
tcp_inpinfo_lock_del(inp, tp);
goto out;
}
tp = tcp_drop(tp, tp->t_softerror ?
tp->t_softerror : ETIMEDOUT);
tcp_inpinfo_lock_del(inp, tp);
goto out;
}
if (tp->t_state == TCPS_SYN_SENT) {
/*
* If the SYN was retransmitted, indicate CWND to be
* limited to 1 segment in cc_conn_init().
*/
tp->snd_cwnd = 1;
} else if (tp->t_rxtshift == 1) {
/*
* first retransmit; record ssthresh and cwnd so they can
* be recovered if this turns out to be a "bad" retransmit.
* A retransmit is considered "bad" if an ACK for this
* segment is received within RTT/2 interval; the assumption
* here is that the ACK was already in flight. See
* "On Estimating End-to-End Network Path Properties" by
* Allman and Paxson for more details.
*/
tp->snd_cwnd_prev = tp->snd_cwnd;
tp->snd_ssthresh_prev = tp->snd_ssthresh;
tp->snd_recover_prev = tp->snd_recover;
if (IN_FASTRECOVERY(tp->t_flags))
tp->t_flags |= TF_WASFRECOVERY;
else
tp->t_flags &= ~TF_WASFRECOVERY;
if (IN_CONGRECOVERY(tp->t_flags))
tp->t_flags |= TF_WASCRECOVERY;
else
tp->t_flags &= ~TF_WASCRECOVERY;
tp->t_badrxtwin = ticks + (tp->t_srtt >> (TCP_RTT_SHIFT + 1));
tp->t_flags |= TF_PREVVALID;
} else
void
tcp_timer_persist(void *xtp)
{
struct tcpcb *tp = xtp;
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_persist) ||
!callout_active(&tp->t_timers->tt_persist)) {
INP_WUNLOCK(inp);
CURVNET_RESTORE();
return;
}
callout_deactivate(&tp->t_timers->tt_persist);
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));
/*
* Persistence timer into zero window.
* Force a byte to be output, if possible.
*/
TCPSTAT_INC(tcps_persisttimeo);
/*
* Hack: if the peer is dead/unreachable, we do not
* time out if the window is closed. After a full
* backoff, drop the connection if the idle time
* (no responses to probes) reaches the maximum
* backoff that we would use if retransmitting.
*/
if (tp->t_rxtshift == TCP_MAXRXTSHIFT &&
(ticks - tp->t_rcvtime >= tcp_maxpersistidle ||
ticks - tp->t_rcvtime >= TCP_REXMTVAL(tp) * tcp_totbackoff)) {
TCPSTAT_INC(tcps_persistdrop);
if (tcp_inpinfo_lock_add(inp)) {
tcp_inpinfo_lock_del(inp, tp);
goto out;
}
tp = tcp_drop(tp, ETIMEDOUT);
tcp_inpinfo_lock_del(inp, tp);
goto out;
}
/*
* If the user has closed the socket then drop a persisting
* connection after a much reduced timeout.
*/
if (tp->t_state > TCPS_CLOSE_WAIT &&
(ticks - tp->t_rcvtime) >= TCPTV_PERSMAX) {
TCPSTAT_INC(tcps_persistdrop);
if (tcp_inpinfo_lock_add(inp)) {
tcp_inpinfo_lock_del(inp, tp);
goto out;
}
tp = tcp_drop(tp, ETIMEDOUT);
tcp_inpinfo_lock_del(inp, tp);
goto out;
}
tcp_setpersist(tp);
tp->t_flags |= TF_FORCEDATA;
(void) tp->t_fb->tfb_tcp_output(tp);
tp->t_flags &= ~TF_FORCEDATA;
#ifdef TCPDEBUG
if (tp != NULL && tp->t_inpcb->inp_socket->so_options & SO_DEBUG)
tcp_trace(TA_USER, ostate, tp, NULL, NULL, PRU_SLOWTIMO);
#endif
TCP_PROBE2(debug__user, tp, PRU_SLOWTIMO);
INP_WUNLOCK(inp);
out:
CURVNET_RESTORE();
}
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();
}
void
tcp_timer_2msl(void *xtp)
{
struct tcpcb *tp = xtp;
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);
tcp_free_sackholes(tp);
if (callout_pending(&tp->t_timers->tt_2msl) ||
!callout_active(&tp->t_timers->tt_2msl)) {
INP_WUNLOCK(tp->t_inpcb);
CURVNET_RESTORE();
return;
}
callout_deactivate(&tp->t_timers->tt_2msl);
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));
/*
* 2 MSL timeout in shutdown went off. If we're closed but
* still waiting for peer to close and connection has been idle
* too long delete connection control block. Otherwise, check
* again in a bit.
*
* If in TIME_WAIT state just ignore as this timeout is handled in
* tcp_tw_2msl_scan().
*
* If fastrecycle of FIN_WAIT_2, in FIN_WAIT_2 and receiver has closed,
* there's no point in hanging onto FIN_WAIT_2 socket. Just close it.
* Ignore fact that there were recent incoming segments.
*/
if ((inp->inp_flags & INP_TIMEWAIT) != 0) {
INP_WUNLOCK(inp);
CURVNET_RESTORE();
return;
}
if (tcp_fast_finwait2_recycle && tp->t_state == TCPS_FIN_WAIT_2 &&
tp->t_inpcb && tp->t_inpcb->inp_socket &&
(tp->t_inpcb->inp_socket->so_rcv.sb_state & SBS_CANTRCVMORE)) {
TCPSTAT_INC(tcps_finwait2_drops);
if (tcp_inpinfo_lock_add(inp)) {
tcp_inpinfo_lock_del(inp, tp);
goto out;
}
tp = tcp_close(tp);
tcp_inpinfo_lock_del(inp, tp);
goto out;
} else {
if (ticks - tp->t_rcvtime <= TP_MAXIDLE(tp)) {
callout_reset(&tp->t_timers->tt_2msl,
TP_KEEPINTVL(tp), tcp_timer_2msl, tp);
} else {
if (tcp_inpinfo_lock_add(inp)) {
tcp_inpinfo_lock_del(inp, tp);
goto out;
}
tp = tcp_close(tp);
tcp_inpinfo_lock_del(inp, tp);
goto out;
}
}
#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);
if (tp != NULL)
INP_WUNLOCK(inp);
out:
CURVNET_RESTORE();
}
void
tcp_timer_rexmt(void * xtp)
{
struct tcpcb *tp = xtp;
CURVNET_SET(tp->t_vnet);
int rexmt;
int headlocked;
struct inpcb *inp;
#ifdef TCPDEBUG
int ostate;
ostate = tp->t_state;
#endif
INP_INFO_RLOCK(&V_tcbinfo);
inp = tp->t_inpcb;
/*
* XXXRW: While this assert is in fact correct, bugs in the tcpcb
* tear-down mean we need it as a work-around for races between
* timers and tcp_discardcb().
*
* KASSERT(inp != NULL, ("tcp_timer_rexmt: inp == NULL"));
*/
if (inp == NULL) {
tcp_timer_race++;
INP_INFO_RUNLOCK(&V_tcbinfo);
CURVNET_RESTORE();
return;
}
INP_WLOCK(inp);
if ((inp->inp_flags & INP_DROPPED) || callout_pending(&tp->t_timers->tt_rexmt)
|| !callout_active(&tp->t_timers->tt_rexmt)) {
INP_WUNLOCK(inp);
INP_INFO_RUNLOCK(&V_tcbinfo);
CURVNET_RESTORE();
return;
}
callout_deactivate(&tp->t_timers->tt_rexmt);
tcp_free_sackholes(tp);
/*
* Retransmission timer went off. Message has not
* been acked within retransmit interval. Back off
* to a longer retransmit interval and retransmit one segment.
*/
if (++tp->t_rxtshift > TCP_MAXRXTSHIFT) {
tp->t_rxtshift = TCP_MAXRXTSHIFT;
TCPSTAT_INC(tcps_timeoutdrop);
in_pcbref(inp);
INP_INFO_RUNLOCK(&V_tcbinfo);
INP_WUNLOCK(inp);
INP_INFO_WLOCK(&V_tcbinfo);
INP_WLOCK(inp);
if (in_pcbrele_wlocked(inp)) {
INP_INFO_WUNLOCK(&V_tcbinfo);
CURVNET_RESTORE();
return;
}
if (inp->inp_flags & INP_DROPPED) {
INP_WUNLOCK(inp);
INP_INFO_WUNLOCK(&V_tcbinfo);
CURVNET_RESTORE();
return;
}
tp = tcp_drop(tp, tp->t_softerror ?
tp->t_softerror : ETIMEDOUT);
headlocked = 1;
goto out;
}
INP_INFO_RUNLOCK(&V_tcbinfo);
headlocked = 0;
if (tp->t_rxtshift == 1) {
/*
* first retransmit; record ssthresh and cwnd so they can
* be recovered if this turns out to be a "bad" retransmit.
* A retransmit is considered "bad" if an ACK for this
* segment is received within RTT/2 interval; the assumption
* here is that the ACK was already in flight. See
* "On Estimating End-to-End Network Path Properties" by
* Allman and Paxson for more details.
*/
tp->snd_cwnd_prev = tp->snd_cwnd;
tp->snd_ssthresh_prev = tp->snd_ssthresh;
tp->snd_recover_prev = tp->snd_recover;
if (IN_FASTRECOVERY(tp->t_flags))
tp->t_flags |= TF_WASFRECOVERY;
else
tp->t_flags &= ~TF_WASFRECOVERY;
if (IN_CONGRECOVERY(tp->t_flags))
tp->t_flags |= TF_WASCRECOVERY;
else
tp->t_flags &= ~TF_WASCRECOVERY;
tp->t_badrxtwin = ticks + (tp->t_srtt >> (TCP_RTT_SHIFT + 1));
tp->t_flags |= TF_PREVVALID;
} else
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 *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_INFO_WLOCK(&V_tcbinfo);
inp = tp->t_inpcb;
/*
* XXXRW: While this assert is in fact correct, bugs in the tcpcb
* tear-down mean we need it as a work-around for races between
* timers and tcp_discardcb().
*
* KASSERT(inp != NULL, ("tcp_timer_keep: inp == NULL"));
*/
if (inp == NULL) {
tcp_timer_race++;
INP_INFO_WUNLOCK(&V_tcbinfo);
CURVNET_RESTORE();
return;
}
INP_WLOCK(inp);
if ((inp->inp_flags & INP_DROPPED) || callout_pending(&tp->t_timers->tt_keep)
|| !callout_active(&tp->t_timers->tt_keep)) {
INP_WUNLOCK(inp);
INP_INFO_WUNLOCK(&V_tcbinfo);
CURVNET_RESTORE();
return;
}
callout_deactivate(&tp->t_timers->tt_keep);
/*
* 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_on(&tp->t_timers->tt_keep, TP_KEEPINTVL(tp),
tcp_timer_keep, tp, INP_CPU(inp));
} else
callout_reset_on(&tp->t_timers->tt_keep, TP_KEEPIDLE(tp),
tcp_timer_keep, tp, INP_CPU(inp));
#ifdef TCPDEBUG
if (inp->inp_socket->so_options & SO_DEBUG)
tcp_trace(TA_USER, ostate, tp, (void *)0, (struct tcphdr *)0,
PRU_SLOWTIMO);
#endif
INP_WUNLOCK(inp);
INP_INFO_WUNLOCK(&V_tcbinfo);
CURVNET_RESTORE();
return;
dropit:
TCPSTAT_INC(tcps_keepdrops);
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
if (tp != NULL)
INP_WUNLOCK(tp->t_inpcb);
INP_INFO_WUNLOCK(&V_tcbinfo);
CURVNET_RESTORE();
}
void
tcp_timer_2msl(void *xtp)
{
struct tcpcb *tp = xtp;
struct inpcb *inp;
CURVNET_SET(tp->t_vnet);
#ifdef TCPDEBUG
int ostate;
ostate = tp->t_state;
#endif
/*
* XXXRW: Does this actually happen?
*/
INP_INFO_WLOCK(&V_tcbinfo);
inp = tp->t_inpcb;
/*
* XXXRW: While this assert is in fact correct, bugs in the tcpcb
* tear-down mean we need it as a work-around for races between
* timers and tcp_discardcb().
*
* KASSERT(inp != NULL, ("tcp_timer_2msl: inp == NULL"));
*/
if (inp == NULL) {
tcp_timer_race++;
INP_INFO_WUNLOCK(&V_tcbinfo);
CURVNET_RESTORE();
return;
}
INP_WLOCK(inp);
tcp_free_sackholes(tp);
if ((inp->inp_flags & INP_DROPPED) || callout_pending(&tp->t_timers->tt_2msl) ||
!callout_active(&tp->t_timers->tt_2msl)) {
INP_WUNLOCK(tp->t_inpcb);
INP_INFO_WUNLOCK(&V_tcbinfo);
CURVNET_RESTORE();
return;
}
callout_deactivate(&tp->t_timers->tt_2msl);
/*
* 2 MSL timeout in shutdown went off. If we're closed but
* still waiting for peer to close and connection has been idle
* too long, or if 2MSL time is up from TIME_WAIT, delete connection
* control block. Otherwise, check again in a bit.
*
* If fastrecycle of FIN_WAIT_2, in FIN_WAIT_2 and receiver has closed,
* there's no point in hanging onto FIN_WAIT_2 socket. Just close it.
* Ignore fact that there were recent incoming segments.
*/
if (tcp_fast_finwait2_recycle && tp->t_state == TCPS_FIN_WAIT_2 &&
tp->t_inpcb && tp->t_inpcb->inp_socket &&
(tp->t_inpcb->inp_socket->so_rcv.sb_state & SBS_CANTRCVMORE)) {
TCPSTAT_INC(tcps_finwait2_drops);
tp = tcp_close(tp);
} else {
if (tp->t_state != TCPS_TIME_WAIT &&
ticks - tp->t_rcvtime <= TP_MAXIDLE(tp))
callout_reset_on(&tp->t_timers->tt_2msl,
TP_KEEPINTVL(tp), tcp_timer_2msl, tp, INP_CPU(inp));
else
tp = tcp_close(tp);
}
#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
if (tp != NULL)
INP_WUNLOCK(inp);
INP_INFO_WUNLOCK(&V_tcbinfo);
CURVNET_RESTORE();
}
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_INFO_RLOCK(&V_tcbinfo);
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);
INP_INFO_RUNLOCK(&V_tcbinfo);
CURVNET_RESTORE();
return;
}
callout_deactivate(&tp->t_timers->tt_keep);
if ((inp->inp_flags & INP_DROPPED) != 0) {
INP_WUNLOCK(inp);
INP_INFO_RUNLOCK(&V_tcbinfo);
CURVNET_RESTORE();
return;
}
KASSERT((tp->t_timers->tt_flags & TT_STOPPED) == 0,
("%s: tp %p tcpcb can't be stopped here", __func__, tp));
KASSERT((tp->t_timers->tt_flags & TT_KEEP) != 0,
("%s: tp %p keep callout should be running", __func__, tp));
/*
* 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);
}
if (!callout_reset(&tp->t_timers->tt_keep, TP_KEEPINTVL(tp),
tcp_timer_keep, tp)) {
tp->t_timers->tt_flags &= ~TT_KEEP_RST;
}
} else if (!callout_reset(&tp->t_timers->tt_keep, TP_KEEPIDLE(tp),
tcp_timer_keep, tp)) {
tp->t_timers->tt_flags &= ~TT_KEEP_RST;
}
#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);
INP_INFO_RUNLOCK(&V_tcbinfo);
CURVNET_RESTORE();
return;
dropit:
TCPSTAT_INC(tcps_keepdrops);
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);
if (tp != NULL)
INP_WUNLOCK(tp->t_inpcb);
INP_INFO_RUNLOCK(&V_tcbinfo);
CURVNET_RESTORE();
}
void
tcp_timer_persist(void *xtp)
{
struct tcpcb *tp = xtp;
struct inpcb *inp;
CURVNET_SET(tp->t_vnet);
#ifdef TCPDEBUG
int ostate;
ostate = tp->t_state;
#endif
INP_INFO_WLOCK(&V_tcbinfo);
inp = tp->t_inpcb;
/*
* XXXRW: While this assert is in fact correct, bugs in the tcpcb
* tear-down mean we need it as a work-around for races between
* timers and tcp_discardcb().
*
* KASSERT(inp != NULL, ("tcp_timer_persist: inp == NULL"));
*/
if (inp == NULL) {
tcp_timer_race++;
INP_INFO_WUNLOCK(&V_tcbinfo);
CURVNET_RESTORE();
return;
}
INP_WLOCK(inp);
if ((inp->inp_flags & INP_DROPPED) || callout_pending(&tp->t_timers->tt_persist)
|| !callout_active(&tp->t_timers->tt_persist)) {
INP_WUNLOCK(inp);
INP_INFO_WUNLOCK(&V_tcbinfo);
CURVNET_RESTORE();
return;
}
callout_deactivate(&tp->t_timers->tt_persist);
/*
* Persistance timer into zero window.
* Force a byte to be output, if possible.
*/
TCPSTAT_INC(tcps_persisttimeo);
/*
* Hack: if the peer is dead/unreachable, we do not
* time out if the window is closed. After a full
* backoff, drop the connection if the idle time
* (no responses to probes) reaches the maximum
* backoff that we would use if retransmitting.
*/
if (tp->t_rxtshift == TCP_MAXRXTSHIFT &&
(ticks - tp->t_rcvtime >= tcp_maxpersistidle ||
ticks - tp->t_rcvtime >= TCP_REXMTVAL(tp) * tcp_totbackoff)) {
TCPSTAT_INC(tcps_persistdrop);
tp = tcp_drop(tp, ETIMEDOUT);
goto out;
}
tcp_setpersist(tp);
tp->t_flags |= TF_FORCEDATA;
(void) tcp_output(tp);
tp->t_flags &= ~TF_FORCEDATA;
out:
#ifdef TCPDEBUG
if (tp != NULL && tp->t_inpcb->inp_socket->so_options & SO_DEBUG)
tcp_trace(TA_USER, ostate, tp, NULL, NULL, PRU_SLOWTIMO);
#endif
if (tp != NULL)
INP_WUNLOCK(inp);
INP_INFO_WUNLOCK(&V_tcbinfo);
CURVNET_RESTORE();
}
static int
tcp6_connect(struct tcpcb *tp, struct sockaddr *nam, struct thread *td)
{
struct inpcb *inp = tp->t_inpcb, *oinp;
struct socket *so = inp->inp_socket;
struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam;
struct in6_addr addr6;
int error;
INP_INFO_WLOCK_ASSERT(&V_tcbinfo);
INP_WLOCK_ASSERT(inp);
if (inp->inp_lport == 0) {
error = in6_pcbbind(inp, (struct sockaddr *)0, td->td_ucred);
if (error)
return error;
}
/*
* Cannot simply call in_pcbconnect, because there might be an
* earlier incarnation of this same connection still in
* TIME_WAIT state, creating an ADDRINUSE error.
* in6_pcbladdr() also handles scope zone IDs.
*/
error = in6_pcbladdr(inp, nam, &addr6);
if (error)
return error;
oinp = in6_pcblookup_hash(inp->inp_pcbinfo,
&sin6->sin6_addr, sin6->sin6_port,
IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)
? &addr6
: &inp->in6p_laddr,
inp->inp_lport, 0, NULL);
if (oinp)
return EADDRINUSE;
if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr))
inp->in6p_laddr = addr6;
inp->in6p_faddr = sin6->sin6_addr;
inp->inp_fport = sin6->sin6_port;
/* update flowinfo - draft-itojun-ipv6-flowlabel-api-00 */
inp->inp_flow &= ~IPV6_FLOWLABEL_MASK;
if (inp->inp_flags & IN6P_AUTOFLOWLABEL)
inp->inp_flow |=
(htonl(ip6_randomflowlabel()) & IPV6_FLOWLABEL_MASK);
in_pcbrehash(inp);
/* Compute window scaling to request. */
while (tp->request_r_scale < TCP_MAX_WINSHIFT &&
(TCP_MAXWIN << tp->request_r_scale) < sb_max)
tp->request_r_scale++;
soisconnecting(so);
TCPSTAT_INC(tcps_connattempt);
tp->t_state = TCPS_SYN_SENT;
tcp_timer_activate(tp, TT_KEEP, tcp_keepinit);
tp->iss = tcp_new_isn(tp);
tp->t_bw_rtseq = tp->iss;
tcp_sendseqinit(tp);
return 0;
}
/*
* 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);
}
