/*
* Perform any necessary tasks before we enter congestion recovery.
*/
static void
newreno_cong_signal(struct cc_var *ccv, uint32_t type)
{
u_int win;
/* Catch algos which mistakenly leak private signal types. */
KASSERT((type & CC_SIGPRIVMASK) == 0,
("%s: congestion signal type 0x%08x is private\n", __func__, type));
win = max(CCV(ccv, snd_cwnd) / 2 / CCV(ccv, t_maxseg), 2) *
CCV(ccv, t_maxseg);
switch (type) {
case CC_NDUPACK:
if (!IN_FASTRECOVERY(CCV(ccv, t_flags))) {
if (!IN_CONGRECOVERY(CCV(ccv, t_flags)))
CCV(ccv, snd_ssthresh) = win;
ENTER_RECOVERY(CCV(ccv, t_flags));
}
break;
case CC_ECN:
if (!IN_CONGRECOVERY(CCV(ccv, t_flags))) {
CCV(ccv, snd_ssthresh) = win;
CCV(ccv, snd_cwnd) = win;
ENTER_CONGRECOVERY(CCV(ccv, t_flags));
}
break;
}
}
static void
dctcp_ack_received(struct cc_var *ccv, uint16_t type)
{
struct dctcp *dctcp_data;
int bytes_acked = 0;
dctcp_data = ccv->cc_data;
if (CCV(ccv, t_flags) & TF_ECN_PERMIT) {
/*
* DCTCP doesn't treat receipt of ECN marked packet as a
* congestion event. Thus, DCTCP always executes the ACK
* processing out of congestion recovery.
*/
if (IN_CONGRECOVERY(CCV(ccv, t_flags))) {
EXIT_CONGRECOVERY(CCV(ccv, t_flags));
newreno_cc_algo.ack_received(ccv, type);
ENTER_CONGRECOVERY(CCV(ccv, t_flags));
} else
newreno_cc_algo.ack_received(ccv, type);
if (type == CC_DUPACK)
bytes_acked = CCV(ccv, t_maxseg);
if (type == CC_ACK)
bytes_acked = ccv->bytes_this_ack;
/* Update total bytes. */
dctcp_data->bytes_total += bytes_acked;
/* Update total marked bytes. */
if (dctcp_data->ece_curr) {
if (!dctcp_data->ece_prev
&& bytes_acked > CCV(ccv, t_maxseg)) {
dctcp_data->bytes_ecn +=
(bytes_acked - CCV(ccv, t_maxseg));
} else
dctcp_data->bytes_ecn += bytes_acked;
dctcp_data->ece_prev = 1;
} else {
if (dctcp_data->ece_prev
&& bytes_acked > CCV(ccv, t_maxseg))
dctcp_data->bytes_ecn += CCV(ccv, t_maxseg);
dctcp_data->ece_prev = 0;
}
dctcp_data->ece_curr = 0;
/*
* Update the fraction of marked bytes at the end of
* current window size.
*/
if ((IN_FASTRECOVERY(CCV(ccv, t_flags)) &&
SEQ_GEQ(ccv->curack, CCV(ccv, snd_recover))) ||
(!IN_FASTRECOVERY(CCV(ccv, t_flags)) &&
SEQ_GT(ccv->curack, dctcp_data->save_sndnxt)))
dctcp_update_alpha(ccv);
} else
newreno_cc_algo.ack_received(ccv, type);
}
/*
* Perform any necessary tasks before we enter congestion recovery.
*/
static void
newreno_cong_signal(struct cc_var *ccv, uint32_t type)
{
uint32_t cwin, ssthresh_on_loss;
u_int mss;
cwin = CCV(ccv, snd_cwnd);
mss = CCV(ccv, t_maxseg);
ssthresh_on_loss =
max((CCV(ccv, snd_max) - CCV(ccv, snd_una)) / 2 / mss, 2)
* mss;
/* Catch algos which mistakenly leak private signal types. */
KASSERT((type & CC_SIGPRIVMASK) == 0,
("%s: congestion signal type 0x%08x is private\n", __func__, type));
cwin = max(cwin / 2 / mss, 2) * mss;
switch (type) {
case CC_NDUPACK:
if (!IN_FASTRECOVERY(CCV(ccv, t_flags))) {
if (!IN_CONGRECOVERY(CCV(ccv, t_flags))) {
CCV(ccv, snd_ssthresh) = ssthresh_on_loss;
CCV(ccv, snd_cwnd) = cwin;
}
ENTER_RECOVERY(CCV(ccv, t_flags));
}
break;
case CC_ECN:
if (!IN_CONGRECOVERY(CCV(ccv, t_flags))) {
CCV(ccv, snd_ssthresh) = ssthresh_on_loss;
CCV(ccv, snd_cwnd) = cwin;
ENTER_CONGRECOVERY(CCV(ccv, t_flags));
}
break;
case CC_RTO:
CCV(ccv, snd_ssthresh) = ssthresh_on_loss;
CCV(ccv, snd_cwnd) = mss;
break;
}
}
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_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
/*
* Perform any necessary tasks before we enter congestion recovery.
*/
static void
dctcp_cong_signal(struct cc_var *ccv, uint32_t type)
{
struct dctcp *dctcp_data;
u_int win, mss;
dctcp_data = ccv->cc_data;
win = CCV(ccv, snd_cwnd);
mss = CCV(ccv, t_maxseg);
switch (type) {
case CC_NDUPACK:
if (!IN_FASTRECOVERY(CCV(ccv, t_flags))) {
if (!IN_CONGRECOVERY(CCV(ccv, t_flags))) {
CCV(ccv, snd_ssthresh) = mss *
max(win / 2 / mss, 2);
dctcp_data->num_cong_events++;
} else {
/* cwnd has already updated as congestion
* recovery. Reverse cwnd value using
* snd_cwnd_prev and recalculate snd_ssthresh
*/
win = CCV(ccv, snd_cwnd_prev);
CCV(ccv, snd_ssthresh) =
max(win / 2 / mss, 2) * mss;
}
ENTER_RECOVERY(CCV(ccv, t_flags));
}
break;
case CC_ECN:
/*
* Save current snd_cwnd when the host encounters both
* congestion recovery and fast recovery.
*/
CCV(ccv, snd_cwnd_prev) = win;
if (!IN_CONGRECOVERY(CCV(ccv, t_flags))) {
if (V_dctcp_slowstart &&
dctcp_data->num_cong_events++ == 0) {
CCV(ccv, snd_ssthresh) =
mss * max(win / 2 / mss, 2);
dctcp_data->alpha = MAX_ALPHA_VALUE;
dctcp_data->bytes_ecn = 0;
dctcp_data->bytes_total = 0;
dctcp_data->save_sndnxt = CCV(ccv, snd_nxt);
} else
CCV(ccv, snd_ssthresh) = max((win - ((win *
dctcp_data->alpha) >> 11)) / mss, 2) * mss;
CCV(ccv, snd_cwnd) = CCV(ccv, snd_ssthresh);
ENTER_CONGRECOVERY(CCV(ccv, t_flags));
}
dctcp_data->ece_curr = 1;
break;
case CC_RTO:
if (CCV(ccv, t_flags) & TF_ECN_PERMIT) {
CCV(ccv, t_flags) |= TF_ECN_SND_CWR;
dctcp_update_alpha(ccv);
dctcp_data->save_sndnxt += CCV(ccv, t_maxseg);
dctcp_data->num_cong_events++;
}
break;
}
}
