* Copyright (c) 1982, 1986 Regents of the University of California.
* All rights reserved. The Berkeley software License Agreement
* specifies the terms and conditions for redistribution.
* @(#)tcp_output.c 7.3 (Berkeley) %G%
#include "../net/route.h"
u_char tcp_initopt
[4] = { TCPOPT_MAXSEG
, 4, 0x0, 0x0, };
* Tcp output routine: figure out what should be sent and send it.
register struct tcpcb
*tp
;
register struct socket
*so
= tp
->t_inpcb
->inp_socket
;
register struct tcpiphdr
*ti
;
* Determine length of data that should be transmitted,
* and flags that will be used.
* If there is some data or critical controls (SYN, RST)
* to send, then transmit; otherwise, investigate further.
idle
= (tp
->snd_max
== tp
->snd_una
);
off
= tp
->snd_nxt
- tp
->snd_una
;
win
= MIN(tp
->snd_wnd
, tp
->snd_cwnd
);
* If in persist timeout with window of 0, send 1 byte.
* Otherwise, if window is small but nonzero
* and timer expired, we will send what we can
* and go to transmit state.
tp
->t_timer
[TCPT_PERSIST
] = 0;
len
= MIN(so
->so_snd
.sb_cc
, win
) - off
;
flags
= tcp_outflags
[tp
->t_state
];
* If FIN has been sent but not acked,
* but we haven't been called to retransmit,
* len will be -1; transmit if acking, otherwise no need.
* Otherwise, window shrank after we sent into it.
* If window shrank to 0, cancel pending retransmit
* and pull snd_nxt back to (closed) window.
* We will enter persist state below.
* If the window didn't close completely,
if (tp
->t_flags
& TF_ACKNOW
)
tp
->t_timer
[TCPT_REXMT
] = 0;
tp
->snd_nxt
= tp
->snd_una
;
if (len
> tp
->t_maxseg
) {
* Don't send more than one segment if retransmitting
* (or persisting, but then we shouldn't be here).
if (SEQ_LT(tp
->snd_nxt
+ len
, tp
->snd_una
+ so
->so_snd
.sb_cc
))
win
= sbspace(&so
->so_rcv
);
* If our state indicates that FIN should be sent
* and we have not yet done so, or we're retransmitting the FIN,
((tp
->t_flags
& TF_SENTFIN
) == 0 || tp
->snd_nxt
== tp
->snd_una
))
* Send if we owe peer an ACK.
if (tp
->t_flags
& TF_ACKNOW
)
if (flags
& (TH_SYN
|TH_RST
))
if (SEQ_GT(tp
->snd_up
, tp
->snd_una
))
* Sender silly window avoidance. If connection is idle
* and can send all data, a maximum segment,
* at least a maximum default-size segment do it,
* or are forced, do it; otherwise don't bother.
* If peer's buffer is tiny, then send
* when window is at least half open.
* If retransmitting (possibly after persist timer forced us
* to send into a small window), then must resend.
if (len
== tp
->t_maxseg
|| len
>= TCP_MSS
) /* a lot */
if ((idle
|| tp
->t_flags
& TF_NODELAY
) &&
len
+ off
>= so
->so_snd
.sb_cc
)
if (len
>= tp
->max_sndwnd
/ 2)
if (SEQ_LT(tp
->snd_nxt
, tp
->snd_max
))
* Compare available window to amount of window
* known to peer (as advertised window less
* next expected input.) If the difference is 35% or more of the
* maximum possible window, then want to send a window update to peer.
((100*(win
-(tp
->rcv_adv
-tp
->rcv_nxt
))/so
->so_rcv
.sb_hiwat
) >= 35))
* TCP window updates are not reliable, rather a polling protocol
* using ``persist'' packets is used to insure receipt of window
* updates. The three ``states'' for the output side are:
* idle not doing retransmits or persists
* persisting to move a small or zero window
* (re)transmitting and thereby not persisting
* tp->t_timer[TCPT_PERSIST]
* is set when we are in persist state.
* is set when we are called to send a persist packet.
* tp->t_timer[TCPT_REXMT]
* is set when we are retransmitting
* The output side is idle when both timers are zero.
* If send window is too small, there is data to transmit, and no
* retransmit or persist is pending, then go to persist state.
* If nothing happens soon, send when timer expires:
* if window is nonzero, transmit what we can,
* otherwise force out a byte.
if (so
->so_snd
.sb_cc
&& tp
->t_timer
[TCPT_REXMT
] == 0 &&
tp
->t_timer
[TCPT_PERSIST
] == 0) {
* No reason to send a segment, just return.
* Grab a header mbuf, attaching a copy of data to
* be transmitted, and initialize the header from
* the template for sends on this connection.
MGET(m
, M_DONTWAIT
, MT_HEADER
);
m
->m_off
= MMAXOFF
- sizeof (struct tcpiphdr
);
m
->m_len
= sizeof (struct tcpiphdr
);
if (tp
->t_force
&& len
== 1)
else if (SEQ_LT(tp
->snd_nxt
, tp
->snd_max
)) {
tcpstat
.tcps_sndrexmitpack
++;
tcpstat
.tcps_sndrexmitbyte
+= len
;
tcpstat
.tcps_sndbyte
+= len
;
m
->m_next
= m_copy(so
->so_snd
.sb_mb
, off
, len
);
} else if (tp
->t_flags
& TF_ACKNOW
)
else if (flags
& (TH_SYN
|TH_FIN
|TH_RST
))
else if (SEQ_GT(tp
->snd_up
, tp
->snd_una
))
ti
= mtod(m
, struct tcpiphdr
*);
bcopy((caddr_t
)tp
->t_template
, (caddr_t
)ti
, sizeof (struct tcpiphdr
));
* Fill in fields, remembering maximum advertised
* window for use in delaying messages about window sizes.
* If resending a FIN, be sure not to use a new sequence number.
if (flags
& TH_FIN
&& tp
->t_flags
& TF_SENTFIN
&&
tp
->snd_nxt
!= tp
->snd_una
)
ti
->ti_seq
= htonl(tp
->snd_nxt
);
ti
->ti_ack
= htonl(tp
->rcv_nxt
);
* Before ESTABLISHED, force sending of initial options
* unless TCP set to not do any options.
if (tp
->t_state
< TCPS_ESTABLISHED
&& (tp
->t_flags
& TF_NOOPT
) == 0) {
mss
= MIN(so
->so_rcv
.sb_hiwat
/ 2, tcp_mss(tp
));
if (mss
> IP_MSS
- sizeof(struct tcpiphdr
)) {
optlen
= sizeof (tcp_initopt
);
*(u_short
*)(opt
+ 2) = htons(mss
);
} else if (tp
->t_tcpopt
) {
opt
= mtod(tp
->t_tcpopt
, u_char
*);
optlen
= tp
->t_tcpopt
->m_len
;
m
->m_next
= m_get(M_DONTWAIT
, MT_DATA
);
bcopy((caddr_t
)opt
, mtod(m0
, caddr_t
), optlen
);
opt
= (u_char
*)(mtod(m0
, caddr_t
) + optlen
);
while (m0
->m_len
& 0x3) {
ti
->ti_off
= (sizeof (struct tcphdr
) + optlen
) >> 2;
* Calculate receive window. Don't shrink window,
* but avoid silly window syndrome.
if (win
< so
->so_rcv
.sb_hiwat
/ 4 && win
< tp
->t_maxseg
)
if (win
< (int)(tp
->rcv_adv
- tp
->rcv_nxt
))
win
= (int)(tp
->rcv_adv
- tp
->rcv_nxt
);
ti
->ti_win
= htons((u_short
)win
);
if (SEQ_GT(tp
->snd_up
, tp
->snd_nxt
)) {
ti
->ti_urp
= htons((u_short
)(tp
->snd_up
- tp
->snd_nxt
));
* If no urgent pointer to send, then we pull
* the urgent pointer to the left edge of the send window
* so that it doesn't drift into the send window on sequence
tp
->snd_up
= tp
->snd_una
; /* drag it along */
* If anything to send and we can send it all, set PUSH.
* (This will keep happy those implementations which only
* give data to the user when a buffer fills or a PUSH comes in.)
if (len
&& off
+len
== so
->so_snd
.sb_cc
)
* Put TCP length in extended header, and then
* checksum extended header and data.
ti
->ti_len
= htons((u_short
)(sizeof(struct tcphdr
) +
ti
->ti_sum
= in_cksum(m
, sizeof (struct tcpiphdr
) + (int)optlen
+ len
);
* In transmit state, time the transmission and arrange for
* the retransmit. In persist state, just set snd_max.
if (tp
->t_force
== 0 || tp
->t_timer
[TCPT_PERSIST
] == 0) {
* Advance snd_nxt over sequence space of this segment.
tp
->t_flags
|= TF_SENTFIN
;
if (SEQ_GT(tp
->snd_nxt
, tp
->snd_max
)) {
tp
->snd_max
= tp
->snd_nxt
;
* Time this transmission if not a retransmission and
* not currently timing anything.
tp
->t_rtseq
= tp
->snd_nxt
- len
;
tcpstat
.tcps_segstimed
++;
* Set retransmit timer if not currently set,
* and not doing an ack or a keep-alive probe.
* Initial value for retransmit timer is tcp_beta*tp->t_srtt.
* Initialize shift counter which is used for backoff
if (tp
->t_timer
[TCPT_REXMT
] == 0 &&
tp
->snd_nxt
!= tp
->snd_una
) {
TCPT_RANGESET(tp
->t_timer
[TCPT_REXMT
],
tcp_beta
* (tp
->t_srtt
? tp
->t_srtt
: TCPTV_SRTTDFLT
),
tp
->t_timer
[TCPT_PERSIST
] = 0;
if (SEQ_GT(tp
->snd_nxt
+ len
, tp
->snd_max
))
tp
->snd_max
= tp
->snd_nxt
+ len
;
if (so
->so_options
& SO_DEBUG
)
tcp_trace(TA_OUTPUT
, tp
->t_state
, tp
, ti
, 0);
* Fill in IP length and desired time to live and
((struct ip
*)ti
)->ip_len
= sizeof (struct tcpiphdr
) + optlen
+ len
;
((struct ip
*)ti
)->ip_ttl
= TCP_TTL
;
error
= ip_output(m
, tp
->t_inpcb
->inp_options
, &tp
->t_inpcb
->inp_route
,
so
->so_options
& SO_DONTROUTE
);
* Data sent (as far as we can tell).
* If this advertises a larger window than any other segment,
* then remember the size of the advertised window.
* Any pending ACK has now been sent.
if (win
> 0 && SEQ_GT(tp
->rcv_nxt
+win
, tp
->rcv_adv
))
tp
->rcv_adv
= tp
->rcv_nxt
+ win
;
tp
->t_flags
&= ~(TF_ACKNOW
|TF_DELACK
);
register struct tcpcb
*tp
;
if (tp
->t_timer
[TCPT_REXMT
])
panic("tcp_output REXMT");
* Start/restart persistance timer.
TCPT_RANGESET(tp
->t_timer
[TCPT_PERSIST
],
((int)(tcp_beta
* tp
->t_srtt
)) << tp
->t_rxtshift
,
TCPTV_PERSMIN
, TCPTV_MAX
);
if (tp
->t_rxtshift
>= TCP_MAXRXTSHIFT
)