/*
* INET An implementation of the TCP/IP protocol suite for the LINUX
* operating system. INET is implemented using the BSD Socket
* interface as the means of communication with the user level.
*
* Implementation of the Transmission Control Protocol(TCP).
*
* Version: $Id: tcp_output.c,v 1.129 2000/11/28 17:04:10 davem Exp $
*
* Authors: Ross Biro, <bir7@leland.Stanford.Edu>
* Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
* Mark Evans, <evansmp@uhura.aston.ac.uk>
* Corey Minyard <wf-rch!minyard@relay.EU.net>
* Florian La Roche, <flla@stud.uni-sb.de>
* Charles Hedrick, <hedrick@klinzhai.rutgers.edu>
* Linus Torvalds, <torvalds@cs.helsinki.fi>
* Alan Cox, <gw4pts@gw4pts.ampr.org>
* Matthew Dillon, <dillon@apollo.west.oic.com>
* Arnt Gulbrandsen, <agulbra@nvg.unit.no>
* Jorge Cwik, <jorge@laser.satlink.net>
*/
/*
* Changes: Pedro Roque : Retransmit queue handled by TCP.
* : Fragmentation on mtu decrease
* : Segment collapse on retransmit
* : AF independence
*
* Linus Torvalds : send_delayed_ack
* David S. Miller : Charge memory using the right skb
* during syn/ack processing.
* David S. Miller : Output engine completely rewritten.
* Andrea Arcangeli: SYNACK carry ts_recent in tsecr.
* Cacophonix Gaul : draft-minshall-nagle-01
* J Hadi Salim : ECN support
*
*/
#include <net/tcp.h>
#include <linux/smp_lock.h>
/* People can turn this off for buggy TCP's found in printers etc. */
int sysctl_tcp_retrans_collapse = 1;
static __inline__
47 void update_send_head(struct sock *sk, struct tcp_opt *tp, struct sk_buff *skb)
{
tp->send_head = skb->next;
50 if (tp->send_head == (struct sk_buff *) &sk->write_queue)
tp->send_head = NULL;
tp->snd_nxt = TCP_SKB_CB(skb)->end_seq;
53 if (tp->packets_out++ == 0)
tcp_reset_xmit_timer(sk, TCP_TIME_RETRANS, tp->rto);
}
/* SND.NXT, if window was not shrunk.
* If window has been shrunk, what should we make? It is not clear at all.
* Using SND.UNA we will fail to open window, SND.NXT is out of window. :-(
* Anything in between SND.UNA...SND.UNA+SND.WND also can be already
* invalid. OK, let's make this for now:
*/
63 static __inline__ __u32 tcp_acceptable_seq(struct sock *sk, struct tcp_opt *tp)
{
65 if (!before(tp->snd_una+tp->snd_wnd, tp->snd_nxt))
66 return tp->snd_nxt;
67 else
68 return tp->snd_una+tp->snd_wnd;
}
/* Calculate mss to advertise in SYN segment.
* RFC1122, RFC1063, draft-ietf-tcpimpl-pmtud-01 state that:
*
* 1. It is independent of path mtu.
* 2. Ideally, it is maximal possible segment size i.e. 65535-40.
* 3. For IPv4 it is reasonable to calculate it from maximal MTU of
* attached devices, because some buggy hosts are confused by
* large MSS.
* 4. We do not make 3, we advertise MSS, calculated from first
* hop device mtu, but allow to raise it to ip_rt_min_advmss.
* This may be overriden via information stored in routing table.
* 5. Value 65535 for MSS is valid in IPv6 and means "as large as possible,
* probably even Jumbo".
*/
85 static __u16 tcp_advertise_mss(struct sock *sk)
{
struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
struct dst_entry *dst = __sk_dst_get(sk);
int mss = tp->advmss;
91 if (dst && dst->advmss < mss) {
mss = dst->advmss;
tp->advmss = mss;
}
96 return (__u16)mss;
}
/* RFC2861. Reset CWND after idle period longer RTO to "restart window".
* This is the first part of cwnd validation mechanism. */
101 static void tcp_cwnd_restart(struct tcp_opt *tp)
{
s32 delta = tcp_time_stamp - tp->lsndtime;
u32 restart_cwnd = tcp_init_cwnd(tp);
u32 cwnd = tp->snd_cwnd;
tp->snd_ssthresh = tcp_current_ssthresh(tp);
restart_cwnd = min(restart_cwnd, cwnd);
110 while ((delta -= tp->rto) > 0 && cwnd > restart_cwnd)
cwnd >>= 1;
tp->snd_cwnd = max(cwnd, restart_cwnd);
tp->snd_cwnd_stamp = tcp_time_stamp;
tp->snd_cwnd_used = 0;
}
117 static __inline__ void tcp_event_data_sent(struct tcp_opt *tp, struct sk_buff *skb)
{
u32 now = tcp_time_stamp;
121 if (!tp->packets_out && (s32)(now - tp->lsndtime) > tp->rto)
tcp_cwnd_restart(tp);
tp->lsndtime = now;
/* If it is a reply for ato after last received
* packet, enter pingpong mode.
*/
129 if ((u32)(now - tp->ack.lrcvtime) < tp->ack.ato)
tp->ack.pingpong = 1;
}
133 static __inline__ void tcp_event_ack_sent(struct sock *sk)
{
struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
tcp_dec_quickack_mode(tp);
tcp_clear_xmit_timer(sk, TCP_TIME_DACK);
}
/* Chose a new window to advertise, update state in tcp_opt for the
* socket, and return result with RFC1323 scaling applied. The return
* value can be stuffed directly into th->window for an outgoing
* frame.
*/
146 static __inline__ u16 tcp_select_window(struct sock *sk)
{
struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
u32 cur_win = tcp_receive_window(tp);
u32 new_win = __tcp_select_window(sk);
/* Never shrink the offered window */
153 if(new_win < cur_win) {
/* Danger Will Robinson!
* Don't update rcv_wup/rcv_wnd here or else
* we will not be able to advertise a zero
* window in time. --DaveM
*
* Relax Will Robinson.
*/
new_win = cur_win;
}
tp->rcv_wnd = new_win;
tp->rcv_wup = tp->rcv_nxt;
/* RFC1323 scaling applied */
new_win >>= tp->rcv_wscale;
#ifdef TCP_FORMAL_WINDOW
if (new_win == 0) {
/* If we advertise zero window, disable fast path. */
tp->pred_flags = 0;
} else if (cur_win == 0 && tp->pred_flags == 0 &&
skb_queue_len(&tp->out_of_order_queue) == 0 &&
!tp->urg_data) {
/* If we open zero window, enable fast path.
Without this it will be open by the first data packet,
it is too late to merge checksumming to copy.
*/
tcp_fast_path_on(tp);
}
#endif
184 return new_win;
}
/* This routine actually transmits TCP packets queued in by
* tcp_do_sendmsg(). This is used by both the initial
* transmission and possible later retransmissions.
* All SKB's seen here are completely headerless. It is our
* job to build the TCP header, and pass the packet down to
* IP so it can do the same plus pass the packet off to the
* device.
*
* We are working here with either a clone of the original
* SKB, or a fresh unique copy made by the retransmit engine.
*/
199 int tcp_transmit_skb(struct sock *sk, struct sk_buff *skb)
{
201 if(skb != NULL) {
struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
struct tcp_skb_cb *tcb = TCP_SKB_CB(skb);
int tcp_header_size = tp->tcp_header_len;
struct tcphdr *th;
int sysctl_flags;
int err;
#define SYSCTL_FLAG_TSTAMPS 0x1
#define SYSCTL_FLAG_WSCALE 0x2
#define SYSCTL_FLAG_SACK 0x4
sysctl_flags = 0;
214 if (tcb->flags & TCPCB_FLAG_SYN) {
tcp_header_size = sizeof(struct tcphdr) + TCPOLEN_MSS;
216 if(sysctl_tcp_timestamps) {
tcp_header_size += TCPOLEN_TSTAMP_ALIGNED;
sysctl_flags |= SYSCTL_FLAG_TSTAMPS;
}
220 if(sysctl_tcp_window_scaling) {
tcp_header_size += TCPOLEN_WSCALE_ALIGNED;
sysctl_flags |= SYSCTL_FLAG_WSCALE;
}
224 if(sysctl_tcp_sack) {
sysctl_flags |= SYSCTL_FLAG_SACK;
226 if(!(sysctl_flags & SYSCTL_FLAG_TSTAMPS))
tcp_header_size += TCPOLEN_SACKPERM_ALIGNED;
}
229 } else if (tp->eff_sacks) {
/* A SACK is 2 pad bytes, a 2 byte header, plus
* 2 32-bit sequence numbers for each SACK block.
*/
tcp_header_size += (TCPOLEN_SACK_BASE_ALIGNED +
(tp->eff_sacks * TCPOLEN_SACK_PERBLOCK));
}
th = (struct tcphdr *) skb_push(skb, tcp_header_size);
skb->h.th = th;
skb_set_owner_w(skb, sk);
/* Build TCP header and checksum it. */
th->source = sk->sport;
th->dest = sk->dport;
th->seq = htonl(tcb->seq);
th->ack_seq = htonl(tp->rcv_nxt);
*(((__u16 *)th) + 6) = htons(((tcp_header_size >> 2) << 12) | tcb->flags);
246 if (tcb->flags & TCPCB_FLAG_SYN) {
/* RFC1323: The window in SYN & SYN/ACK segments
* is never scaled.
*/
th->window = htons(tp->rcv_wnd);
251 } else {
th->window = htons(tcp_select_window(sk));
}
th->check = 0;
th->urg_ptr = 0;
if (tp->urg_mode &&
258 between(tp->snd_up, tcb->seq+1, tcb->seq+0xFFFF)) {
th->urg_ptr = htons(tp->snd_up-tcb->seq);
th->urg = 1;
}
263 if (tcb->flags & TCPCB_FLAG_SYN) {
tcp_syn_build_options((__u32 *)(th + 1),
tcp_advertise_mss(sk),
(sysctl_flags & SYSCTL_FLAG_TSTAMPS),
(sysctl_flags & SYSCTL_FLAG_SACK),
(sysctl_flags & SYSCTL_FLAG_WSCALE),
tp->rcv_wscale,
tcb->when,
tp->ts_recent);
272 } else {
tcp_build_and_update_options((__u32 *)(th + 1),
tp, tcb->when);
276 TCP_ECN_send(sk, tp, skb, tcp_header_size);
}
tp->af_specific->send_check(sk, th, skb->len, skb);
280 if (tcb->flags & TCPCB_FLAG_ACK)
tcp_event_ack_sent(sk);
283 if (skb->len != tcp_header_size)
tcp_event_data_sent(tp, skb);
TCP_INC_STATS(TcpOutSegs);
err = tp->af_specific->queue_xmit(skb);
289 if (err <= 0)
290 return err;
tcp_enter_cwr(tp);
/* NET_XMIT_CN is special. It does not guarantee,
* that this packet is lost. It tells that device
* is about to start to drop packets or already
* drops some packets of the same priority and
* invokes us to send less aggressively.
*/
300 return err == NET_XMIT_CN ? 0 : err;
}
302 return -ENOBUFS;
#undef SYSCTL_FLAG_TSTAMPS
#undef SYSCTL_FLAG_WSCALE
#undef SYSCTL_FLAG_SACK
}
/* This is the main buffer sending routine. We queue the buffer
* and decide whether to queue or transmit now.
*
* NOTE: probe0 timer is not checked, do not forget tcp_push_pending_frames,
* otherwise socket can stall.
*/
315 void tcp_send_skb(struct sock *sk, struct sk_buff *skb, int force_queue, unsigned cur_mss)
{
struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
/* Advance write_seq and place onto the write_queue. */
tp->write_seq = TCP_SKB_CB(skb)->end_seq;
__skb_queue_tail(&sk->write_queue, skb);
tcp_charge_skb(sk, skb);
324 if (!force_queue && tp->send_head == NULL && tcp_snd_test(tp, skb, cur_mss, tp->nonagle)) {
/* Send it out now. */
TCP_SKB_CB(skb)->when = tcp_time_stamp;
327 if (tcp_transmit_skb(sk, skb_clone(skb, sk->allocation)) == 0) {
tp->snd_nxt = TCP_SKB_CB(skb)->end_seq;
tcp_minshall_update(tp, cur_mss, skb);
330 if (tp->packets_out++ == 0)
tcp_reset_xmit_timer(sk, TCP_TIME_RETRANS, tp->rto);
332 return;
}
}
/* Queue it, remembering where we must start sending. */
336 if (tp->send_head == NULL)
tp->send_head = skb;
}
/* Function to create two new TCP segments. Shrinks the given segment
* to the specified size and appends a new segment with the rest of the
* packet to the list. This won't be called frequently, I hope.
* Remember, these are still headerless SKBs at this point.
*/
345 static int tcp_fragment(struct sock *sk, struct sk_buff *skb, u32 len)
{
struct tcp_opt *tp = &sk->tp_pinfo.af_tcp;
struct sk_buff *buff;
int nsize = skb->len - len;
u16 flags;
/* Get a new skb... force flag on. */
buff = tcp_alloc_skb(sk, nsize + MAX_TCP_HEADER, GFP_ATOMIC);
354 if (buff == NULL)
355 return -ENOMEM; /* We'll just try again later. */
tcp_charge_skb(sk, buff);
/* Reserve space for headers. */
skb_reserve(buff, MAX_TCP_HEADER);
/* Correct the sequence numbers. */
TCP_SKB_CB(buff)->seq = TCP_SKB_CB(skb)->seq + len;
TCP_SKB_CB(buff)->end_seq = TCP_SKB_CB(skb)->end_seq;
/* PSH and FIN should only be set in the second packet. */
flags = TCP_SKB_CB(skb)->flags;
TCP_SKB_CB(skb)->flags = flags & ~(TCPCB_FLAG_FIN|TCPCB_FLAG_PSH);
TCP_SKB_CB(buff)->flags = flags;
TCP_SKB_CB(buff)->sacked = TCP_SKB_CB(skb)->sacked&(TCPCB_LOST|TCPCB_EVER_RETRANS|TCPCB_AT_TAIL);
370 if (TCP_SKB_CB(buff)->sacked&TCPCB_LOST) {
tp->lost_out++;
tp->left_out++;
}
TCP_SKB_CB(buff)->sacked &= ~TCPCB_AT_TAIL;
/* Copy and checksum data tail into the new buffer. */
buff->csum = csum_partial_copy_nocheck(skb->data + len, skb_put(buff, nsize),
nsize, 0);
/* This takes care of the FIN sequence number too. */
TCP_SKB_CB(skb)->end_seq = TCP_SKB_CB(buff)->seq;
skb_trim(skb, len);
/* Rechecksum original buffer. */
skb->csum = csum_partial(skb->data, skb->len, 0);
/* Looks stupid, but our code really uses when of
* skbs, which it never sent before. --ANK
*/
TCP_SKB_CB(buff)->when = TCP_SKB_CB(skb)->when;
/* Link BUFF into the send queue. */
__skb_append(skb, buff);
395 return 0;
}
/* This function synchronize snd mss to current pmtu/exthdr set.
tp->user_mss is mss set by user by TCP_MAXSEG. It does NOT counts
for TCP options, but includes only bare TCP header.
tp->mss_clamp is mss negotiated at connection setup.
It is minumum of user_mss and mss received with SYN.
It also does not include TCP options.
tp->pmtu_cookie is last pmtu, seen by this function.
tp->mss_cache is current effective sending mss, including
all tcp options except for SACKs. It is evaluated,
taking into account current pmtu, but never exceeds
tp->mss_clamp.
NOTE1. rfc1122 clearly states that advertised MSS
DOES NOT include either tcp or ip options.
NOTE2. tp->pmtu_cookie and tp->mss_cache are READ ONLY outside
this function. --ANK (980731)
*/
421 int tcp_sync_mss(struct sock *sk, u32 pmtu)
{
struct tcp_opt *tp = &sk->tp_pinfo.af_tcp;
int mss_now;
/* Calculate base mss without TCP options:
It is MMS_S - sizeof(tcphdr) of rfc1122
*/
mss_now = pmtu - tp->af_specific->net_header_len - sizeof(struct tcphdr);
/* Clamp it (mss_clamp does not include tcp options) */
433 if (mss_now > tp->mss_clamp)
mss_now = tp->mss_clamp;
/* Now subtract optional transport overhead */
mss_now -= tp->ext_header_len;
/* Then reserve room for full set of TCP options and 8 bytes of data */
440 if (mss_now < 48)
mss_now = 48;
/* Now subtract TCP options size, not including SACKs */
mss_now -= tp->tcp_header_len - sizeof(struct tcphdr);
/* Bound mss with half of window */
447 if (tp->max_window && mss_now > (tp->max_window>>1))
mss_now = max((tp->max_window>>1), 68 - tp->tcp_header_len);
/* And store cached results */
tp->pmtu_cookie = pmtu;
tp->mss_cache = mss_now;
453 return mss_now;
}
/* This routine writes packets to the network. It advances the
* send_head. This happens as incoming acks open up the remote
* window for us.
*
* Returns 1, if no segments are in flight and we have queued segments, but
* cannot send anything now because of SWS or another problem.
*/
464 int tcp_write_xmit(struct sock *sk)
{
struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
unsigned int mss_now;
/* If we are closed, the bytes will have to remain here.
* In time closedown will finish, we empty the write queue and all
* will be happy.
*/
473 if(sk->state != TCP_CLOSE) {
struct sk_buff *skb;
int sent_pkts = 0;
/* Account for SACKS, we may need to fragment due to this.
* It is just like the real MSS changing on us midstream.
* We also handle things correctly when the user adds some
* IP options mid-stream. Silly to do, but cover it.
*/
mss_now = tcp_current_mss(sk);
while((skb = tp->send_head) &&
485 tcp_snd_test(tp, skb, mss_now, tcp_skb_is_last(sk, skb) ? tp->nonagle : 1)) {
486 if (skb->len > mss_now) {
487 if (tcp_fragment(sk, skb, mss_now))
488 break;
}
TCP_SKB_CB(skb)->when = tcp_time_stamp;
492 if (tcp_transmit_skb(sk, skb_clone(skb, GFP_ATOMIC)))
493 break;
/* Advance the send_head. This one is sent out. */
update_send_head(sk, tp, skb);
tcp_minshall_update(tp, mss_now, skb);
sent_pkts = 1;
}
500 if (sent_pkts) {
tcp_cwnd_validate(sk, tp);
502 return 0;
}
505 return !tp->packets_out && tp->send_head;
}
507 return 0;
}
/* This function returns the amount that we can raise the
* usable window based on the following constraints
*
* 1. The window can never be shrunk once it is offered (RFC 793)
* 2. We limit memory per socket
*
* RFC 1122:
* "the suggested [SWS] avoidance algorithm for the receiver is to keep
* RECV.NEXT + RCV.WIN fixed until:
* RCV.BUFF - RCV.USER - RCV.WINDOW >= min(1/2 RCV.BUFF, MSS)"
*
* i.e. don't raise the right edge of the window until you can raise
* it at least MSS bytes.
*
* Unfortunately, the recommended algorithm breaks header prediction,
* since header prediction assumes th->window stays fixed.
*
* Strictly speaking, keeping th->window fixed violates the receiver
* side SWS prevention criteria. The problem is that under this rule
* a stream of single byte packets will cause the right side of the
* window to always advance by a single byte.
*
* Of course, if the sender implements sender side SWS prevention
* then this will not be a problem.
*
* BSD seems to make the following compromise:
*
* If the free space is less than the 1/4 of the maximum
* space available and the free space is less than 1/2 mss,
* then set the window to 0.
* [ Actually, bsd uses MSS and 1/4 of maximal _window_ ]
* Otherwise, just prevent the window from shrinking
* and from being larger than the largest representable value.
*
* This prevents incremental opening of the window in the regime
* where TCP is limited by the speed of the reader side taking
* data out of the TCP receive queue. It does nothing about
* those cases where the window is constrained on the sender side
* because the pipeline is full.
*
* BSD also seems to "accidentally" limit itself to windows that are a
* multiple of MSS, at least until the free space gets quite small.
* This would appear to be a side effect of the mbuf implementation.
* Combining these two algorithms results in the observed behavior
* of having a fixed window size at almost all times.
*
* Below we obtain similar behavior by forcing the offered window to
* a multiple of the mss when it is feasible to do so.
*
* Note, we don't "adjust" for TIMESTAMP or SACK option bytes.
* Regular options like TIMESTAMP are taken into account.
*/
562 u32 __tcp_select_window(struct sock *sk)
{
struct tcp_opt *tp = &sk->tp_pinfo.af_tcp;
/* MSS for the peer's data. Previous verions used mss_clamp
* here. I don't know if the value based on our guesses
* of peer's MSS is better for the performance. It's more correct
* but may be worse for the performance because of rcv_mss
* fluctuations. --SAW 1998/11/1
*/
unsigned int mss = tp->ack.rcv_mss;
int free_space;
u32 window;
/* Sometimes free_space can be < 0. */
free_space = tcp_space(sk);
577 if (tp->window_clamp < mss)
mss = tp->window_clamp;
580 if (free_space < (int)min(tp->window_clamp, tcp_full_space(sk)) / 2) {
tp->ack.quick = 0;
583 if (tcp_memory_pressure)
tp->rcv_ssthresh = min(tp->rcv_ssthresh, 4*tp->advmss);
586 if (free_space < ((int)mss))
587 return 0;
}
590 if (free_space > tp->rcv_ssthresh)
free_space = tp->rcv_ssthresh;
/* Get the largest window that is a nice multiple of mss.
* Window clamp already applied above.
* If our current window offering is within 1 mss of the
* free space we just keep it. This prevents the divide
* and multiply from happening most of the time.
* We also don't do any window rounding when the free space
* is too small.
*/
window = tp->rcv_wnd;
if ((((int) window) <= (free_space - ((int) mss))) ||
603 (((int) window) > free_space))
window = (((unsigned int) free_space)/mss)*mss;
606 return window;
}
/* Attempt to collapse two adjacent SKB's during retransmission. */
610 static void tcp_retrans_try_collapse(struct sock *sk, struct sk_buff *skb, int mss_now)
{
struct tcp_opt *tp = &sk->tp_pinfo.af_tcp;
struct sk_buff *next_skb = skb->next;
/* The first test we must make is that neither of these two
* SKB's are still referenced by someone else.
*/
618 if(!skb_cloned(skb) && !skb_cloned(next_skb)) {
int skb_size = skb->len, next_skb_size = next_skb->len;
u16 flags = TCP_SKB_CB(skb)->flags;
/* Also punt if next skb has been SACK'd. */
623 if(TCP_SKB_CB(next_skb)->sacked & TCPCB_SACKED_ACKED)
624 return;
/* Next skb is out of window. */
627 if (after(TCP_SKB_CB(next_skb)->end_seq, tp->snd_una+tp->snd_wnd))
628 return;
/* Punt if not enough space exists in the first SKB for
* the data in the second, or the total combined payload
* would exceed the MSS.
*/
if ((next_skb_size > skb_tailroom(skb)) ||
635 ((skb_size + next_skb_size) > mss_now))
636 return;
/* Ok. We will be able to collapse the packet. */
__skb_unlink(next_skb, next_skb->list);
641 if(skb->len % 4) {
/* Must copy and rechecksum all data. */
memcpy(skb_put(skb, next_skb_size), next_skb->data, next_skb_size);
skb->csum = csum_partial(skb->data, skb->len, 0);
645 } else {
/* Optimize, actually we could also combine next_skb->csum
* to skb->csum using a single add w/carry operation too.
*/
skb->csum = csum_partial_copy_nocheck(next_skb->data,
skb_put(skb, next_skb_size),
next_skb_size, skb->csum);
}
/* Update sequence range on original skb. */
TCP_SKB_CB(skb)->end_seq = TCP_SKB_CB(next_skb)->end_seq;
/* Merge over control information. */
flags |= TCP_SKB_CB(next_skb)->flags; /* This moves PSH/FIN etc. over */
TCP_SKB_CB(skb)->flags = flags;
/* All done, get rid of second SKB and account for it so
* packet counting does not break.
*/
TCP_SKB_CB(skb)->sacked |= TCP_SKB_CB(next_skb)->sacked&(TCPCB_EVER_RETRANS|TCPCB_AT_TAIL);
665 if (TCP_SKB_CB(next_skb)->sacked&TCPCB_SACKED_RETRANS)
tp->retrans_out--;
667 if (TCP_SKB_CB(next_skb)->sacked&TCPCB_LOST) {
tp->lost_out--;
tp->left_out--;
}
671 if (!tp->sack_ok && tp->sacked_out) {
/* Reno case is special. Sigh... */
tp->sacked_out--;
tp->left_out--;
}
/* Not quite right: it can be > snd.fack, but
* it is better to underestimate fackets.
*/
679 if (tp->fackets_out)
tp->fackets_out--;
tcp_free_skb(sk, next_skb);
tp->packets_out--;
}
}
/* Do a simple retransmit without using the backoff mechanisms in
* tcp_timer. This is used for path mtu discovery.
* The socket is already locked here.
*/
690 void tcp_simple_retransmit(struct sock *sk)
{
struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
struct sk_buff *skb;
unsigned int mss = tcp_current_mss(sk);
int lost = 0;
697 for_retrans_queue(skb, sk, tp) {
if (skb->len > mss &&
699 !(TCP_SKB_CB(skb)->sacked&TCPCB_SACKED_ACKED)) {
700 if (TCP_SKB_CB(skb)->sacked&TCPCB_SACKED_RETRANS) {
TCP_SKB_CB(skb)->sacked &= ~TCPCB_SACKED_RETRANS;
tp->retrans_out--;
}
704 if (!(TCP_SKB_CB(skb)->sacked&TCPCB_LOST)) {
TCP_SKB_CB(skb)->sacked |= TCPCB_LOST;
tp->lost_out++;
lost = 1;
}
}
}
712 if (!lost)
713 return;
tp->left_out = tp->sacked_out + tp->lost_out;
/* Don't muck with the congestion window here.
* Reason is that we do not increase amount of _data_
* in network, but units changed and effective
* cwnd/ssthresh really reduced now.
*/
722 if (tp->ca_state != TCP_CA_Loss) {
tp->high_seq = tp->snd_nxt;
tp->snd_ssthresh = tcp_current_ssthresh(tp);
tp->prior_ssthresh = 0;
tp->undo_marker = 0;
tp->ca_state = TCP_CA_Loss;
}
tcp_xmit_retransmit_queue(sk);
}
/* This retransmits one SKB. Policy decisions and retransmit queue
* state updates are done by the caller. Returns non-zero if an
* error occurred which prevented the send.
*/
736 int tcp_retransmit_skb(struct sock *sk, struct sk_buff *skb)
{
struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
unsigned int cur_mss = tcp_current_mss(sk);
int err;
/* Do not sent more than we queued. 1/4 is reserved for possible
* copying overhead: frgagmentation, tunneling, mangling etc.
*/
745 if (atomic_read(&sk->wmem_alloc) > min(sk->wmem_queued+(sk->wmem_queued>>2),sk->sndbuf))
746 return -EAGAIN;
748 if(skb->len > cur_mss) {
749 if(tcp_fragment(sk, skb, cur_mss))
750 return -ENOMEM; /* We'll try again later. */
/* New SKB created, account for it. */
tp->packets_out++;
}
/* Collapse two adjacent packets if worthwhile and we can. */
if(!(TCP_SKB_CB(skb)->flags & TCPCB_FLAG_SYN) &&
(skb->len < (cur_mss >> 1)) &&
(skb->next != tp->send_head) &&
(skb->next != (struct sk_buff *)&sk->write_queue) &&
761 (sysctl_tcp_retrans_collapse != 0))
tcp_retrans_try_collapse(sk, skb, cur_mss);
764 if(tp->af_specific->rebuild_header(sk))
765 return -EHOSTUNREACH; /* Routing failure or similar. */
/* Some Solaris stacks overoptimize and ignore the FIN on a
* retransmit when old data is attached. So strip it off
* since it is cheap to do so and saves bytes on the network.
*/
if(skb->len > 0 &&
(TCP_SKB_CB(skb)->flags & TCPCB_FLAG_FIN) &&
773 tp->snd_una == (TCP_SKB_CB(skb)->end_seq - 1)) {
TCP_SKB_CB(skb)->seq = TCP_SKB_CB(skb)->end_seq - 1;
skb_trim(skb, 0);
skb->csum = 0;
}
/* Make a copy, if the first transmission SKB clone we made
* is still in somebody's hands, else make a clone.
*/
TCP_SKB_CB(skb)->when = tcp_time_stamp;
err = tcp_transmit_skb(sk, (skb_cloned(skb) ?
skb_copy(skb, GFP_ATOMIC):
skb_clone(skb, GFP_ATOMIC)));
788 if (err == 0) {
/* Update global TCP statistics. */
TCP_INC_STATS(TcpRetransSegs);
#if FASTRETRANS_DEBUG > 0
793 if (TCP_SKB_CB(skb)->sacked&TCPCB_SACKED_RETRANS) {
794 if (net_ratelimit())
printk(KERN_DEBUG "retrans_out leaked.\n");
}
#endif
TCP_SKB_CB(skb)->sacked |= TCPCB_RETRANS;
tp->retrans_out++;
/* Save stamp of the first retransmit. */
802 if (!tp->retrans_stamp)
tp->retrans_stamp = TCP_SKB_CB(skb)->when;
tp->undo_retrans++;
/* snd_nxt is stored to detect loss of retransmitted segment,
* see tcp_input.c tcp_sacktag_write_queue().
*/
TCP_SKB_CB(skb)->ack_seq = tp->snd_nxt;
}
812 return err;
}
/* This gets called after a retransmit timeout, and the initially
* retransmitted data is acknowledged. It tries to continue
* resending the rest of the retransmit queue, until either
* we've sent it all or the congestion window limit is reached.
* If doing SACK, the first ACK which comes back for a timeout
* based retransmit packet might feed us FACK information again.
* If so, we use it to avoid unnecessarily retransmissions.
*/
823 void tcp_xmit_retransmit_queue(struct sock *sk)
{
struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
struct sk_buff *skb;
int packet_cnt = tp->lost_out;
/* First pass: retransmit lost packets. */
830 if (packet_cnt) {
831 for_retrans_queue(skb, sk, tp) {
__u8 sacked = TCP_SKB_CB(skb)->sacked;
834 if (tcp_packets_in_flight(tp) >= tp->snd_cwnd)
835 return;
837 if (sacked&TCPCB_LOST) {
838 if (!(sacked&(TCPCB_SACKED_ACKED|TCPCB_SACKED_RETRANS))) {
839 if (tcp_retransmit_skb(sk, skb))
840 return;
841 if (tp->ca_state != TCP_CA_Loss)
NET_INC_STATS_BH(TCPFastRetrans);
843 else
NET_INC_STATS_BH(TCPSlowStartRetrans);
846 if (skb == skb_peek(&sk->write_queue))
tcp_reset_xmit_timer(sk, TCP_TIME_RETRANS, tp->rto);
}
850 if (--packet_cnt <= 0)
851 break;
}
}
}
/* OK, demanded retransmission is finished. */
/* Forward retransmissions are possible only during Recovery. */
859 if (tp->ca_state != TCP_CA_Recovery)
860 return;
/* No forward retransmissions in Reno are possible. */
863 if (!tp->sack_ok)
864 return;
/* Yeah, we have to make difficult choice between forward transmission
* and retransmission... Both ways have their merits...
*
* For now we do not retrnamsit anything, while we have some new
* segments to send.
*/
873 if (tcp_may_send_now(sk, tp))
874 return;
packet_cnt = 0;
878 for_retrans_queue(skb, sk, tp) {
879 if(++packet_cnt > tp->fackets_out)
880 break;
882 if (tcp_packets_in_flight(tp) >= tp->snd_cwnd)
883 break;
885 if(TCP_SKB_CB(skb)->sacked & TCPCB_TAGBITS)
886 continue;
/* Ok, retransmit it. */
889 if(tcp_retransmit_skb(sk, skb))
890 break;
892 if (skb == skb_peek(&sk->write_queue))
tcp_reset_xmit_timer(sk, TCP_TIME_RETRANS, tp->rto);
NET_INC_STATS_BH(TCPForwardRetrans);
}
}
/* Send a fin. The caller locks the socket for us. This cannot be
* allowed to fail queueing a FIN frame under any circumstances.
*/
903 void tcp_send_fin(struct sock *sk)
{
struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
struct sk_buff *skb = skb_peek_tail(&sk->write_queue);
unsigned int mss_now;
/* Optimization, tack on the FIN if we have a queue of
* unsent frames. But be careful about outgoing SACKS
* and IP options.
*/
mss_now = tcp_current_mss(sk);
/* Please, find seven differences of 2.3.33 and loook
* what I broke here. 8) --ANK
*/
919 if(tp->send_head != NULL) {
/* tcp_write_xmit() takes care of the rest. */
TCP_SKB_CB(skb)->flags |= TCPCB_FLAG_FIN;
TCP_SKB_CB(skb)->end_seq++;
tp->write_seq++;
/* Special case to avoid Nagle bogosity. If this
* segment is the last segment, and it was queued
* due to Nagle/SWS-avoidance, send it out now.
*/
if(tp->send_head == skb &&
930 !after(tp->write_seq, tp->snd_una + tp->snd_wnd)) {
TCP_SKB_CB(skb)->when = tcp_time_stamp;
932 if (!tcp_transmit_skb(sk, skb_clone(skb, GFP_KERNEL)))
update_send_head(sk, tp, skb);
934 else
tcp_check_probe_timer(sk, tp);
}
937 } else {
/* Socket is locked, keep trying until memory is available. */
939 for (;;) {
skb = alloc_skb(MAX_TCP_HEADER, GFP_KERNEL);
941 if (skb)
942 break;
current->policy |= SCHED_YIELD;
schedule();
}
/* Reserve space for headers and prepare control bits. */
skb_reserve(skb, MAX_TCP_HEADER);
skb->csum = 0;
TCP_SKB_CB(skb)->flags = (TCPCB_FLAG_ACK | TCPCB_FLAG_FIN);
TCP_SKB_CB(skb)->sacked = 0;
/* FIN eats a sequence byte, write_seq advanced by tcp_send_skb(). */
TCP_SKB_CB(skb)->seq = tp->write_seq;
TCP_SKB_CB(skb)->end_seq = TCP_SKB_CB(skb)->seq + 1;
tcp_send_skb(sk, skb, 0, mss_now);
__tcp_push_pending_frames(sk, tp, mss_now, 1);
}
}
/* We get here when a process closes a file descriptor (either due to
* an explicit close() or as a byproduct of exit()'ing) and there
* was unread data in the receive queue. This behavior is recommended
* by draft-ietf-tcpimpl-prob-03.txt section 3.10. -DaveM
*/
966 void tcp_send_active_reset(struct sock *sk, int priority)
{
struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
struct sk_buff *skb;
/* NOTE: No TCP options attached and we never retransmit this. */
skb = alloc_skb(MAX_TCP_HEADER, priority);
973 if (!skb) {
NET_INC_STATS(TCPAbortFailed);
975 return;
}
/* Reserve space for headers and prepare control bits. */
skb_reserve(skb, MAX_TCP_HEADER);
skb->csum = 0;
TCP_SKB_CB(skb)->flags = (TCPCB_FLAG_ACK | TCPCB_FLAG_RST);
TCP_SKB_CB(skb)->sacked = 0;
/* Send it off. */
TCP_SKB_CB(skb)->seq = tcp_acceptable_seq(sk, tp);
TCP_SKB_CB(skb)->end_seq = TCP_SKB_CB(skb)->seq;
TCP_SKB_CB(skb)->when = tcp_time_stamp;
988 if (tcp_transmit_skb(sk, skb))
NET_INC_STATS(TCPAbortFailed);
}
/* WARNING: This routine must only be called when we have already sent
* a SYN packet that crossed the incoming SYN that caused this routine
* to get called. If this assumption fails then the initial rcv_wnd
* and rcv_wscale values will not be correct.
*/
997 int tcp_send_synack(struct sock *sk)
{
struct sk_buff* skb;
skb = skb_peek(&sk->write_queue);
1002 if (skb == NULL || !(TCP_SKB_CB(skb)->flags&TCPCB_FLAG_SYN)) {
printk(KERN_DEBUG "tcp_send_synack: wrong queue state\n");
1004 return -EFAULT;
}
1006 if (!(TCP_SKB_CB(skb)->flags&TCPCB_FLAG_ACK)) {
1007 if (skb_cloned(skb)) {
struct sk_buff *nskb = skb_copy(skb, GFP_ATOMIC);
1009 if (nskb == NULL)
1010 return -ENOMEM;
__skb_unlink(skb, &sk->write_queue);
__skb_queue_head(&sk->write_queue, nskb);
tcp_free_skb(sk, skb);
tcp_charge_skb(sk, nskb);
skb = nskb;
}
TCP_SKB_CB(skb)->flags |= TCPCB_FLAG_ACK;
1019 TCP_ECN_send_synack(&sk->tp_pinfo.af_tcp, skb);
}
TCP_SKB_CB(skb)->when = tcp_time_stamp;
1022 return tcp_transmit_skb(sk, skb_clone(skb, GFP_ATOMIC));
}
/*
* Prepare a SYN-ACK.
*/
1028 struct sk_buff * tcp_make_synack(struct sock *sk, struct dst_entry *dst,
struct open_request *req)
{
struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
struct tcphdr *th;
int tcp_header_size;
struct sk_buff *skb;
skb = sock_wmalloc(sk, MAX_TCP_HEADER + 15, 1, GFP_ATOMIC);
1037 if (skb == NULL)
1038 return NULL;
/* Reserve space for headers. */
skb_reserve(skb, MAX_TCP_HEADER);
skb->dst = dst_clone(dst);
tcp_header_size = (sizeof(struct tcphdr) + TCPOLEN_MSS +
(req->tstamp_ok ? TCPOLEN_TSTAMP_ALIGNED : 0) +
(req->wscale_ok ? TCPOLEN_WSCALE_ALIGNED : 0) +
/* SACK_PERM is in the place of NOP NOP of TS */
((req->sack_ok && !req->tstamp_ok) ? TCPOLEN_SACKPERM_ALIGNED : 0));
skb->h.th = th = (struct tcphdr *) skb_push(skb, tcp_header_size);
memset(th, 0, sizeof(struct tcphdr));
th->syn = 1;
th->ack = 1;
1055 TCP_ECN_make_synack(req, th);
th->source = sk->sport;
th->dest = req->rmt_port;
TCP_SKB_CB(skb)->seq = req->snt_isn;
TCP_SKB_CB(skb)->end_seq = TCP_SKB_CB(skb)->seq + 1;
th->seq = htonl(TCP_SKB_CB(skb)->seq);
th->ack_seq = htonl(req->rcv_isn + 1);
1062 if (req->rcv_wnd == 0) { /* ignored for retransmitted syns */
__u8 rcv_wscale;
/* Set this up on the first call only */
req->window_clamp = tp->window_clamp ? : dst->window;
/* tcp_full_space because it is guaranteed to be the first packet */
tcp_select_initial_window(tcp_full_space(sk),
dst->advmss - (req->tstamp_ok ? TCPOLEN_TSTAMP_ALIGNED : 0),
&req->rcv_wnd,
&req->window_clamp,
req->wscale_ok,
&rcv_wscale);
req->rcv_wscale = rcv_wscale;
}
/* RFC1323: The window in SYN & SYN/ACK segments is never scaled. */
th->window = htons(req->rcv_wnd);
TCP_SKB_CB(skb)->when = tcp_time_stamp;
tcp_syn_build_options((__u32 *)(th + 1), dst->advmss, req->tstamp_ok,
req->sack_ok, req->wscale_ok, req->rcv_wscale,
TCP_SKB_CB(skb)->when,
req->ts_recent);
skb->csum = 0;
th->doff = (tcp_header_size >> 2);
TCP_INC_STATS(TcpOutSegs);
1088 return skb;
}
1091 int tcp_connect(struct sock *sk, struct sk_buff *buff)
{
struct dst_entry *dst = __sk_dst_get(sk);
struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
/* Reserve space for headers. */
skb_reserve(buff, MAX_TCP_HEADER);
/* We'll fix this up when we get a response from the other end.
* See tcp_input.c:tcp_rcv_state_process case TCP_SYN_SENT.
*/
tp->tcp_header_len = sizeof(struct tcphdr) +
(sysctl_tcp_timestamps ? TCPOLEN_TSTAMP_ALIGNED : 0);
/* If user gave his TCP_MAXSEG, record it to clamp */
1106 if (tp->user_mss)
tp->mss_clamp = tp->user_mss;
tp->max_window = 0;
tcp_sync_mss(sk, dst->pmtu);
1111 if (!tp->window_clamp)
tp->window_clamp = dst->window;
tp->advmss = dst->advmss;
tcp_initialize_rcv_mss(sk);
tcp_select_initial_window(tcp_full_space(sk),
tp->advmss - (tp->ts_recent_stamp ? tp->tcp_header_len - sizeof(struct tcphdr) : 0),
&tp->rcv_wnd,
&tp->window_clamp,
sysctl_tcp_window_scaling,
&tp->rcv_wscale);
tp->rcv_ssthresh = tp->rcv_wnd;
/* Socket identity change complete, no longer
* in TCP_CLOSE, so enter ourselves into the
* hash tables.
*/
tcp_set_state(sk,TCP_SYN_SENT);
1130 if (tp->af_specific->hash_connecting(sk))
1131 goto err_out;
sk->err = 0;
sk->done = 0;
tp->snd_wnd = 0;
tcp_init_wl(tp, tp->write_seq, 0);
tp->snd_una = tp->write_seq;
tp->snd_sml = tp->write_seq;
tp->rcv_nxt = 0;
tp->rcv_wup = 0;
tp->copied_seq = 0;
tp->rto = TCP_TIMEOUT_INIT;
tp->retransmits = 0;
tcp_clear_retrans(tp);
TCP_SKB_CB(buff)->flags = TCPCB_FLAG_SYN;
1148 TCP_ECN_send_syn(tp, buff);
TCP_SKB_CB(buff)->sacked = 0;
buff->csum = 0;
TCP_SKB_CB(buff)->seq = tp->write_seq++;
TCP_SKB_CB(buff)->end_seq = tp->write_seq;
tp->snd_nxt = tp->write_seq;
tp->pushed_seq = tp->write_seq;
/* Send it off. */
TCP_SKB_CB(buff)->when = tcp_time_stamp;
tp->retrans_stamp = TCP_SKB_CB(buff)->when;
__skb_queue_tail(&sk->write_queue, buff);
tcp_charge_skb(sk, buff);
tp->packets_out++;
tcp_transmit_skb(sk, skb_clone(buff, GFP_KERNEL));
TCP_INC_STATS(TcpActiveOpens);
/* Timer for repeating the SYN until an answer. */
tcp_reset_xmit_timer(sk, TCP_TIME_RETRANS, tp->rto);
1167 return 0;
err_out:
tcp_set_state(sk,TCP_CLOSE);
kfree_skb(buff);
1172 return -EADDRNOTAVAIL;
}
/* Send out a delayed ack, the caller does the policy checking
* to see if we should even be here. See tcp_input.c:tcp_ack_snd_check()
* for details.
*/
1179 void tcp_send_delayed_ack(struct sock *sk)
{
struct tcp_opt *tp = &sk->tp_pinfo.af_tcp;
int ato = tp->ack.ato;
unsigned long timeout;
1185 if (ato > TCP_DELACK_MIN) {
int max_ato = HZ/2;
1188 if (tp->ack.pingpong || (tp->ack.pending&TCP_ACK_PUSHED))
max_ato = TCP_DELACK_MAX;
/* Slow path, intersegment interval is "high". */
/* If some rtt estimate is known, use it to bound delayed ack.
* Do not use tp->rto here, use results of rtt measurements
* directly.
*/
1197 if (tp->srtt) {
int rtt = max(tp->srtt>>3, TCP_DELACK_MIN);
1200 if (rtt < max_ato)
max_ato = rtt;
}
ato = min(ato, max_ato);
}
/* Stay within the limit we were given */
timeout = jiffies + ato;
/* Use new timeout only if there wasn't a older one earlier. */
1211 if (tp->ack.pending&TCP_ACK_TIMER) {
/* If delack timer was blocked or is about to expire,
* send ACK now.
*/
1215 if (tp->ack.blocked || time_before_eq(tp->ack.timeout, jiffies+(ato>>2))) {
tcp_send_ack(sk);
1217 return;
}
1220 if (!time_before(timeout, tp->ack.timeout))
timeout = tp->ack.timeout;
}
tp->ack.pending |= TCP_ACK_SCHED|TCP_ACK_TIMER;
tp->ack.timeout = timeout;
1225 if (!mod_timer(&tp->delack_timer, timeout))
sock_hold(sk);
#ifdef TCP_FORMAL_WINDOW
/* Explanation. Header prediction path does not handle
* case of zero window. If we send ACK immediately, pred_flags
* are reset when sending ACK. If rcv_nxt is advanced and
* ack is not sent, than delayed ack is scheduled.
* Hence, it is the best place to check for zero window.
*/
if (tp->pred_flags) {
if (tcp_receive_window(tp) == 0)
tp->pred_flags = 0;
} else {
if (skb_queue_len(&tp->out_of_order_queue) == 0 &&
!tp->urg_data)
tcp_fast_path_on(tp);
}
#endif
}
/* This routine sends an ack and also updates the window. */
1247 void tcp_send_ack(struct sock *sk)
{
/* If we have been reset, we may not send again. */
1250 if(sk->state != TCP_CLOSE) {
struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
struct sk_buff *buff;
/* We are not putting this on the write queue, so
* tcp_transmit_skb() will set the ownership to this
* sock.
*/
buff = alloc_skb(MAX_TCP_HEADER, GFP_ATOMIC);
1259 if (buff == NULL) {
tcp_schedule_ack(tp);
tp->ack.ato = TCP_ATO_MIN;
tcp_reset_xmit_timer(sk, TCP_TIME_DACK, TCP_DELACK_MAX);
1263 return;
}
/* Reserve space for headers and prepare control bits. */
skb_reserve(buff, MAX_TCP_HEADER);
buff->csum = 0;
TCP_SKB_CB(buff)->flags = TCPCB_FLAG_ACK;
TCP_SKB_CB(buff)->sacked = 0;
/* Send it off, this clears delayed acks for us. */
TCP_SKB_CB(buff)->seq = TCP_SKB_CB(buff)->end_seq = tcp_acceptable_seq(sk, tp);
TCP_SKB_CB(buff)->when = tcp_time_stamp;
tcp_transmit_skb(sk, buff);
}
}
/* This routine sends a packet with an out of date sequence
* number. It assumes the other end will try to ack it.
*
* Question: what should we make while urgent mode?
* 4.4BSD forces sending single byte of data. We cannot send
* out of window data, because we have SND.NXT==SND.MAX...
*
* Current solution: to send TWO zero-length segments in urgent mode:
* one is with SEG.SEQ=SND.UNA to deliver urgent pointer, another is
* out-of-date with SND.UNA-1 to probe window.
*/
1290 static int tcp_xmit_probe_skb(struct sock *sk, int urgent)
{
struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
struct sk_buff *skb;
/* We don't queue it, tcp_transmit_skb() sets ownership. */
skb = alloc_skb(MAX_TCP_HEADER, GFP_ATOMIC);
1297 if (skb == NULL)
1298 return -1;
/* Reserve space for headers and set control bits. */
skb_reserve(skb, MAX_TCP_HEADER);
skb->csum = 0;
TCP_SKB_CB(skb)->flags = TCPCB_FLAG_ACK;
TCP_SKB_CB(skb)->sacked = urgent;
/* Use a previous sequence. This should cause the other
* end to send an ack. Don't queue or clone SKB, just
* send it.
*/
TCP_SKB_CB(skb)->seq = urgent ? tp->snd_una : tp->snd_una - 1;
TCP_SKB_CB(skb)->end_seq = TCP_SKB_CB(skb)->seq;
TCP_SKB_CB(skb)->when = tcp_time_stamp;
1313 return tcp_transmit_skb(sk, skb);
}
1316 int tcp_write_wakeup(struct sock *sk)
{
1318 if (sk->state != TCP_CLOSE) {
struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
struct sk_buff *skb;
if ((skb = tp->send_head) != NULL &&
1323 before(TCP_SKB_CB(skb)->seq, tp->snd_una+tp->snd_wnd)) {
int err;
int mss = tcp_current_mss(sk);
int seg_size = tp->snd_una+tp->snd_wnd-TCP_SKB_CB(skb)->seq;
1328 if (before(tp->pushed_seq, TCP_SKB_CB(skb)->end_seq))
tp->pushed_seq = TCP_SKB_CB(skb)->end_seq;
/* We are probing the opening of a window
* but the window size is != 0
* must have been a result SWS avoidance ( sender )
*/
if (seg_size < TCP_SKB_CB(skb)->end_seq - TCP_SKB_CB(skb)->seq ||
1336 skb->len > mss) {
seg_size = min(seg_size, mss);
TCP_SKB_CB(skb)->flags |= TCPCB_FLAG_PSH;
1339 if (tcp_fragment(sk, skb, seg_size))
1340 return -1;
}
TCP_SKB_CB(skb)->flags |= TCPCB_FLAG_PSH;
TCP_SKB_CB(skb)->when = tcp_time_stamp;
err = tcp_transmit_skb(sk, skb_clone(skb, GFP_ATOMIC));
1345 if (!err) {
update_send_head(sk, tp, skb);
}
1348 return err;
1349 } else {
if (tp->urg_mode &&
1351 between(tp->snd_up, tp->snd_una+1, tp->snd_una+0xFFFF))
tcp_xmit_probe_skb(sk, TCPCB_URG);
1353 return tcp_xmit_probe_skb(sk, 0);
}
}
1356 return -1;
}
/* A window probe timeout has occurred. If window is not closed send
* a partial packet else a zero probe.
*/
1362 void tcp_send_probe0(struct sock *sk)
{
struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
int err;
err = tcp_write_wakeup(sk);
1369 if (tp->packets_out || !tp->send_head) {
/* Cancel probe timer, if it is not required. */
tp->probes_out = 0;
tp->backoff = 0;
1373 return;
}
1376 if (err <= 0) {
tp->backoff++;
tp->probes_out++;
tcp_reset_xmit_timer (sk, TCP_TIME_PROBE0,
min(tp->rto << tp->backoff, TCP_RTO_MAX));
1381 } else {
/* If packet was not sent due to local congestion,
* do not backoff and do not remember probes_out.
* Let local senders to fight for local resources.
*
* Use accumulated backoff yet.
*/
1388 if (!tp->probes_out)
tp->probes_out=1;
tcp_reset_xmit_timer (sk, TCP_TIME_PROBE0,
min(tp->rto << tp->backoff, TCP_RESOURCE_PROBE_INTERVAL));
}
}