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 (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);
if ((inp->inp_flags & INP_DROPPED) != 0) {
INP_WUNLOCK(inp);
INP_INFO_RUNLOCK(&V_tcbinfo);
CURVNET_RESTORE();
return;
}
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_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_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 (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);
if ((inp->inp_flags & INP_DROPPED) != 0) {
INP_WUNLOCK(inp);
INP_INFO_WUNLOCK(&V_tcbinfo);
CURVNET_RESTORE();
return;
}
/*
* 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;
}
/*
* 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);
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();
}
/*
* Read from so's socket into sb_snd, updating all relevant sbuf fields
* NOTE: This will only be called if it is select()ed for reading, so
* a read() of 0 (or less) means it's disconnected
*/
int
soread(PNATState pData, struct socket *so)
{
int n, nn, lss, total;
struct sbuf *sb = &so->so_snd;
size_t len = sb->sb_datalen - sb->sb_cc;
struct iovec iov[2];
int mss = so->so_tcpcb->t_maxseg;
STAM_PROFILE_START(&pData->StatIOread, a);
STAM_COUNTER_RESET(&pData->StatIORead_in_1);
STAM_COUNTER_RESET(&pData->StatIORead_in_2);
QSOCKET_LOCK(tcb);
SOCKET_LOCK(so);
QSOCKET_UNLOCK(tcb);
LogFlow(("soread: so = %R[natsock]\n", so));
Log2(("%s: so = %R[natsock] so->so_snd = %R[sbuf]\n", RT_GCC_EXTENSION __PRETTY_FUNCTION__, so, sb));
/*
* No need to check if there's enough room to read.
* soread wouldn't have been called if there weren't
*/
len = sb->sb_datalen - sb->sb_cc;
iov[0].iov_base = sb->sb_wptr;
iov[1].iov_base = 0;
iov[1].iov_len = 0;
if (sb->sb_wptr < sb->sb_rptr)
{
iov[0].iov_len = sb->sb_rptr - sb->sb_wptr;
/* Should never succeed, but... */
if (iov[0].iov_len > len)
iov[0].iov_len = len;
if (iov[0].iov_len > mss)
iov[0].iov_len -= iov[0].iov_len%mss;
n = 1;
}
else
{
iov[0].iov_len = (sb->sb_data + sb->sb_datalen) - sb->sb_wptr;
/* Should never succeed, but... */
if (iov[0].iov_len > len)
iov[0].iov_len = len;
len -= iov[0].iov_len;
if (len)
{
iov[1].iov_base = sb->sb_data;
iov[1].iov_len = sb->sb_rptr - sb->sb_data;
if (iov[1].iov_len > len)
iov[1].iov_len = len;
total = iov[0].iov_len + iov[1].iov_len;
if (total > mss)
{
lss = total % mss;
if (iov[1].iov_len > lss)
{
iov[1].iov_len -= lss;
n = 2;
}
else
{
lss -= iov[1].iov_len;
iov[0].iov_len -= lss;
n = 1;
}
}
else
n = 2;
}
else
{
if (iov[0].iov_len > mss)
iov[0].iov_len -= iov[0].iov_len%mss;
n = 1;
}
}
#ifdef HAVE_READV
nn = readv(so->s, (struct iovec *)iov, n);
#else
nn = recv(so->s, iov[0].iov_base, iov[0].iov_len, (so->so_tcpcb->t_force? MSG_OOB:0));
#endif
Log2(("%s: read(1) nn = %d bytes\n", RT_GCC_EXTENSION __PRETTY_FUNCTION__, nn));
Log2(("%s: so = %R[natsock] so->so_snd = %R[sbuf]\n", RT_GCC_EXTENSION __PRETTY_FUNCTION__, so, sb));
if (nn <= 0)
{
/*
* Special case for WSAEnumNetworkEvents: If we receive 0 bytes that
* _could_ mean that the connection is closed. But we will receive an
* FD_CLOSE event later if the connection was _really_ closed. With
* www.youtube.com I see this very often. Closing the socket too early
* would be dangerous.
*/
int status;
unsigned long pending = 0;
status = ioctlsocket(so->s, FIONREAD, &pending);
if (status < 0)
Log(("NAT:%s: error in WSAIoctl: %d\n", RT_GCC_EXTENSION __PRETTY_FUNCTION__, errno));
if (nn == 0 && (pending != 0))
{
SOCKET_UNLOCK(so);
STAM_PROFILE_STOP(&pData->StatIOread, a);
return 0;
}
if ( nn < 0
&& soIgnorableErrorCode(errno))
{
SOCKET_UNLOCK(so);
STAM_PROFILE_STOP(&pData->StatIOread, a);
return 0;
}
else
{
int fUninitiolizedTemplate = 0;
fUninitiolizedTemplate = RT_BOOL(( sototcpcb(so)
&& ( sototcpcb(so)->t_template.ti_src.s_addr == INADDR_ANY
|| sototcpcb(so)->t_template.ti_dst.s_addr == INADDR_ANY)));
/* nn == 0 means peer has performed an orderly shutdown */
Log2(("%s: disconnected, nn = %d, errno = %d (%s)\n",
RT_GCC_EXTENSION __PRETTY_FUNCTION__, nn, errno, strerror(errno)));
sofcantrcvmore(so);
if (!fUninitiolizedTemplate)
tcp_sockclosed(pData, sototcpcb(so));
else
tcp_drop(pData, sototcpcb(so), errno);
SOCKET_UNLOCK(so);
STAM_PROFILE_STOP(&pData->StatIOread, a);
return -1;
}
}
STAM_STATS(
if (n == 1)
{
STAM_COUNTER_INC(&pData->StatIORead_in_1);
STAM_COUNTER_ADD(&pData->StatIORead_in_1_bytes, nn);
}
else
{
STAM_COUNTER_INC(&pData->StatIORead_in_2);
STAM_COUNTER_ADD(&pData->StatIORead_in_2_1st_bytes, nn);
}
);
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 (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);
if ((inp->inp_flags & INP_DROPPED) != 0) {
INP_WUNLOCK(inp);
INP_INFO_WUNLOCK(&V_tcbinfo);
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_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();
}
/* Caller should be in critical section */
static struct tcpcb *
tcp_timer_keep_handler(struct tcpcb *tp)
{
struct tcptemp *t_template;
#ifdef TCPDEBUG
int ostate = tp->t_state;
#endif
/*
* Keep-alive timer went off; send something
* or drop connection if idle for too long.
*/
tcpstat.tcps_keeptimeo++;
if (tp->t_state < TCPS_ESTABLISHED)
goto dropit;
if ((always_keepalive || (tp->t_flags & TF_KEEPALIVE) ||
(tp->t_inpcb->inp_socket->so_options & SO_KEEPALIVE)) &&
tp->t_state <= TCPS_CLOSING) {
if ((ticks - 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.
*/
tcpstat.tcps_keepprobe++;
t_template = tcp_maketemplate(tp);
if (t_template) {
tcp_respond(tp, t_template->tt_ipgen,
&t_template->tt_t, NULL,
tp->rcv_nxt, tp->snd_una - 1, 0);
tcp_freetemplate(t_template);
}
tcp_callout_reset(tp, tp->tt_keep, tp->t_keepintvl,
tcp_timer_keep);
} else {
tcp_callout_reset(tp, tp->tt_keep, tp->t_keepidle,
tcp_timer_keep);
}
#ifdef TCPDEBUG
if (tp->t_inpcb->inp_socket->so_options & SO_DEBUG)
tcp_trace(TA_USER, ostate, tp, NULL, NULL, PRU_SLOWTIMO);
#endif
return tp;
dropit:
tcpstat.tcps_keepdrops++;
tp = tcp_drop(tp, ETIMEDOUT);
#ifdef TCPDEBUG
if (tp && (tp->t_inpcb->inp_socket->so_options & SO_DEBUG))
tcp_trace(TA_USER, ostate, tp, NULL, NULL, PRU_SLOWTIMO);
#endif
return tp;
}
