int sysctl_tcp_westwood;
int sysctl_tcp_vegas_cong_avoid;
-int sysctl_tcp_moderate_rcvbuf;
+int sysctl_tcp_moderate_rcvbuf = 1;
/* Default values of the Vegas variables, in fixed-point representation
* with V_PARAM_SHIFT bits to the right of the binary point.
int sysctl_tcp_vegas_alpha = 1<<V_PARAM_SHIFT;
int sysctl_tcp_vegas_beta = 3<<V_PARAM_SHIFT;
int sysctl_tcp_vegas_gamma = 1<<V_PARAM_SHIFT;
-int sysctl_tcp_bic;
+int sysctl_tcp_bic = 1;
int sysctl_tcp_bic_fast_convergence = 1;
int sysctl_tcp_bic_low_window = 14;
tp->snd_cwnd_stamp = tcp_time_stamp;
}
+static void init_bictcp(struct tcp_opt *tp)
+{
+ tp->bictcp.cnt = 0;
+
+ tp->bictcp.last_max_cwnd = 0;
+ tp->bictcp.last_cwnd = 0;
+ tp->bictcp.last_stamp = 0;
+}
+
/* 5. Recalculate window clamp after socket hit its memory bounds. */
static void tcp_clamp_window(struct sock *sk, struct tcp_opt *tp)
{
* restart window, so that we send ACKs quickly.
*/
tcp_incr_quickack(tp);
- tcp_mem_reclaim(sk);
+ sk_stream_mem_reclaim(sk);
}
}
tp->ack.lrcvtime = now;
tcp_grow_window(sk, tp, skb);
}
-/* Set up a new TCP connection, depending on whether it should be
- * using Vegas or not.
- */
-void tcp_vegas_init(struct tcp_opt *tp)
+/* When starting a new connection, pin down the current choice of
+ * congestion algorithm.
+ */
+void tcp_ca_init(struct tcp_opt *tp)
{
- if (sysctl_tcp_vegas_cong_avoid) {
- tp->vegas.do_vegas = 1;
+ if (sysctl_tcp_westwood)
+ tp->adv_cong = TCP_WESTWOOD;
+ else if (sysctl_tcp_bic)
+ tp->adv_cong = TCP_BIC;
+ else if (sysctl_tcp_vegas_cong_avoid) {
+ tp->adv_cong = TCP_VEGAS;
tp->vegas.baseRTT = 0x7fffffff;
tcp_vegas_enable(tp);
- } else
- tcp_vegas_disable(tp);
+ }
}
/* Do RTT sampling needed for Vegas.
__u32 cwnd = (dst ? dst_metric(dst, RTAX_INITCWND) : 0);
if (!cwnd) {
- if (tp->mss_cache > 1460)
+ if (tp->mss_cache_std > 1460)
cwnd = 2;
else
- cwnd = (tp->mss_cache > 1095) ? 3 : 4;
+ cwnd = (tp->mss_cache_std > 1095) ? 3 : 4;
}
return min_t(__u32, cwnd, tp->snd_cwnd_clamp);
}
* to low value, and then abruptly stops to do it and starts to delay
* ACKs, wait for troubles.
*/
- if (dst_metric(dst, RTAX_RTT) > tp->srtt)
+ if (dst_metric(dst, RTAX_RTT) > tp->srtt) {
tp->srtt = dst_metric(dst, RTAX_RTT);
+ tp->rtt_seq = tp->snd_nxt;
+ }
if (dst_metric(dst, RTAX_RTTVAR) > tp->mdev) {
tp->mdev = dst_metric(dst, RTAX_RTTVAR);
tp->mdev_max = tp->rttvar = max(tp->mdev, TCP_RTO_MIN);
/* This exciting event is worth to be remembered. 8) */
if (ts)
- NET_INC_STATS_BH(TCPTSReorder);
+ NET_INC_STATS_BH(LINUX_MIB_TCPTSREORDER);
else if (IsReno(tp))
- NET_INC_STATS_BH(TCPRenoReorder);
+ NET_INC_STATS_BH(LINUX_MIB_TCPRENOREORDER);
else if (IsFack(tp))
- NET_INC_STATS_BH(TCPFACKReorder);
+ NET_INC_STATS_BH(LINUX_MIB_TCPFACKREORDER);
else
- NET_INC_STATS_BH(TCPSACKReorder);
+ NET_INC_STATS_BH(LINUX_MIB_TCPSACKREORDER);
#if FASTRETRANS_DEBUG > 1
printk(KERN_DEBUG "Disorder%d %d %u f%u s%u rr%d\n",
tp->sack_ok, tp->ca_state,
- tp->reordering, tp->fackets_out, tp->sacked_out,
+ tp->reordering,
+ tcp_get_pcount(&tp->fackets_out),
+ tcp_get_pcount(&tp->sacked_out),
tp->undo_marker ? tp->undo_retrans : 0);
#endif
/* Disable FACK yet. */
unsigned char *ptr = ack_skb->h.raw + TCP_SKB_CB(ack_skb)->sacked;
struct tcp_sack_block *sp = (struct tcp_sack_block *)(ptr+2);
int num_sacks = (ptr[1] - TCPOLEN_SACK_BASE)>>3;
- int reord = tp->packets_out;
+ int reord = tcp_get_pcount(&tp->packets_out);
int prior_fackets;
u32 lost_retrans = 0;
int flag = 0;
tp->mss_cache = tp->mss_cache_std;
}
- if (!tp->sacked_out)
- tp->fackets_out = 0;
- prior_fackets = tp->fackets_out;
+ if (!tcp_get_pcount(&tp->sacked_out))
+ tcp_set_pcount(&tp->fackets_out, 0);
+ prior_fackets = tcp_get_pcount(&tp->fackets_out);
for (i=0; i<num_sacks; i++, sp++) {
struct sk_buff *skb;
if (before(start_seq, ack)) {
dup_sack = 1;
tp->sack_ok |= 4;
- NET_INC_STATS_BH(TCPDSACKRecv);
+ NET_INC_STATS_BH(LINUX_MIB_TCPDSACKRECV);
} else if (num_sacks > 1 &&
!after(end_seq, ntohl(sp[1].end_seq)) &&
!before(start_seq, ntohl(sp[1].start_seq))) {
dup_sack = 1;
tp->sack_ok |= 4;
- NET_INC_STATS_BH(TCPDSACKOfoRecv);
+ NET_INC_STATS_BH(LINUX_MIB_TCPDSACKOFORECV);
}
/* D-SACK for already forgotten data...
if (after(end_seq, tp->high_seq))
flag |= FLAG_DATA_LOST;
- for_retrans_queue(skb, sk, tp) {
+ sk_stream_for_retrans_queue(skb, sk) {
u8 sacked = TCP_SKB_CB(skb)->sacked;
int in_sack;
if(!before(TCP_SKB_CB(skb)->seq, end_seq))
break;
- fack_count++;
+ fack_count += tcp_skb_pcount(skb);
in_sack = !after(start_seq, TCP_SKB_CB(skb)->seq) &&
!before(end_seq, TCP_SKB_CB(skb)->end_seq);
*/
if (sacked & TCPCB_LOST) {
TCP_SKB_CB(skb)->sacked &= ~(TCPCB_LOST|TCPCB_SACKED_RETRANS);
- tp->lost_out--;
- tp->retrans_out--;
+ tcp_dec_pcount(&tp->lost_out, skb);
+ tcp_dec_pcount(&tp->retrans_out, skb);
}
} else {
/* New sack for not retransmitted frame,
if (sacked & TCPCB_LOST) {
TCP_SKB_CB(skb)->sacked &= ~TCPCB_LOST;
- tp->lost_out--;
+ tcp_dec_pcount(&tp->lost_out, skb);
}
}
TCP_SKB_CB(skb)->sacked |= TCPCB_SACKED_ACKED;
flag |= FLAG_DATA_SACKED;
- tp->sacked_out++;
+ tcp_inc_pcount(&tp->sacked_out, skb);
- if (fack_count > tp->fackets_out)
- tp->fackets_out = fack_count;
+ if (fack_count > tcp_get_pcount(&tp->fackets_out))
+ tcp_set_pcount(&tp->fackets_out, fack_count);
} else {
if (dup_sack && (sacked&TCPCB_RETRANS))
reord = min(fack_count, reord);
if (dup_sack &&
(TCP_SKB_CB(skb)->sacked&TCPCB_SACKED_RETRANS)) {
TCP_SKB_CB(skb)->sacked &= ~TCPCB_SACKED_RETRANS;
- tp->retrans_out--;
+ tcp_dec_pcount(&tp->retrans_out, skb);
}
}
}
if (lost_retrans && tp->ca_state == TCP_CA_Recovery) {
struct sk_buff *skb;
- for_retrans_queue(skb, sk, tp) {
+ sk_stream_for_retrans_queue(skb, sk) {
if (after(TCP_SKB_CB(skb)->seq, lost_retrans))
break;
if (!after(TCP_SKB_CB(skb)->end_seq, tp->snd_una))
(IsFack(tp) ||
!before(lost_retrans,
TCP_SKB_CB(skb)->ack_seq + tp->reordering *
- tp->mss_cache))) {
+ tp->mss_cache_std))) {
TCP_SKB_CB(skb)->sacked &= ~TCPCB_SACKED_RETRANS;
- tp->retrans_out--;
+ tcp_dec_pcount(&tp->retrans_out, skb);
if (!(TCP_SKB_CB(skb)->sacked&(TCPCB_LOST|TCPCB_SACKED_ACKED))) {
- tp->lost_out++;
+ tcp_inc_pcount(&tp->lost_out, skb);
TCP_SKB_CB(skb)->sacked |= TCPCB_LOST;
flag |= FLAG_DATA_SACKED;
- NET_INC_STATS_BH(TCPLostRetransmit);
+ NET_INC_STATS_BH(LINUX_MIB_TCPLOSTRETRANSMIT);
}
}
}
}
- tp->left_out = tp->sacked_out + tp->lost_out;
+ tcp_set_pcount(&tp->left_out,
+ (tcp_get_pcount(&tp->sacked_out) +
+ tcp_get_pcount(&tp->lost_out)));
- if (reord < tp->fackets_out && tp->ca_state != TCP_CA_Loss)
- tcp_update_reordering(tp, (tp->fackets_out + 1) - reord, 0);
+ if ((reord < tcp_get_pcount(&tp->fackets_out)) &&
+ tp->ca_state != TCP_CA_Loss)
+ tcp_update_reordering(tp,
+ ((tcp_get_pcount(&tp->fackets_out) + 1) -
+ reord), 0);
#if FASTRETRANS_DEBUG > 0
- BUG_TRAP((int)tp->sacked_out >= 0);
- BUG_TRAP((int)tp->lost_out >= 0);
- BUG_TRAP((int)tp->retrans_out >= 0);
+ BUG_TRAP((int)tcp_get_pcount(&tp->sacked_out) >= 0);
+ BUG_TRAP((int)tcp_get_pcount(&tp->lost_out) >= 0);
+ BUG_TRAP((int)tcp_get_pcount(&tp->retrans_out) >= 0);
BUG_TRAP((int)tcp_packets_in_flight(tp) >= 0);
#endif
return flag;
* If something was really lost, it is eventually caught up
* in tcp_enter_frto_loss.
*/
- tp->retrans_out = 0;
+ tcp_set_pcount(&tp->retrans_out, 0);
tp->undo_marker = tp->snd_una;
tp->undo_retrans = 0;
- for_retrans_queue(skb, sk, tp) {
+ sk_stream_for_retrans_queue(skb, sk) {
TCP_SKB_CB(skb)->sacked &= ~TCPCB_RETRANS;
}
tcp_sync_left_out(tp);
struct sk_buff *skb;
int cnt = 0;
- tp->sacked_out = 0;
- tp->lost_out = 0;
- tp->fackets_out = 0;
+ tcp_set_pcount(&tp->sacked_out, 0);
+ tcp_set_pcount(&tp->lost_out, 0);
+ tcp_set_pcount(&tp->fackets_out, 0);
- for_retrans_queue(skb, sk, tp) {
- cnt++;
+ sk_stream_for_retrans_queue(skb, sk) {
+ cnt += tcp_skb_pcount(skb);
TCP_SKB_CB(skb)->sacked &= ~TCPCB_LOST;
if (!(TCP_SKB_CB(skb)->sacked&TCPCB_SACKED_ACKED)) {
/* Do not mark those segments lost that were
* forward transmitted after RTO
*/
- if(!after(TCP_SKB_CB(skb)->end_seq,
+ if (!after(TCP_SKB_CB(skb)->end_seq,
tp->frto_highmark)) {
TCP_SKB_CB(skb)->sacked |= TCPCB_LOST;
- tp->lost_out++;
+ tcp_inc_pcount(&tp->lost_out, skb);
}
} else {
- tp->sacked_out++;
- tp->fackets_out = cnt;
+ tcp_inc_pcount(&tp->sacked_out, skb);
+ tcp_set_pcount(&tp->fackets_out, cnt);
}
}
tcp_sync_left_out(tp);
tcp_set_ca_state(tp, TCP_CA_Loss);
tp->high_seq = tp->frto_highmark;
TCP_ECN_queue_cwr(tp);
+
+ init_bictcp(tp);
}
void tcp_clear_retrans(struct tcp_opt *tp)
{
- tp->left_out = 0;
- tp->retrans_out = 0;
+ tcp_set_pcount(&tp->left_out, 0);
+ tcp_set_pcount(&tp->retrans_out, 0);
- tp->fackets_out = 0;
- tp->sacked_out = 0;
- tp->lost_out = 0;
+ tcp_set_pcount(&tp->fackets_out, 0);
+ tcp_set_pcount(&tp->sacked_out, 0);
+ tcp_set_pcount(&tp->lost_out, 0);
tp->undo_marker = 0;
tp->undo_retrans = 0;
if (!how)
tp->undo_marker = tp->snd_una;
- for_retrans_queue(skb, sk, tp) {
- cnt++;
+ sk_stream_for_retrans_queue(skb, sk) {
+ cnt += tcp_skb_pcount(skb);
if (TCP_SKB_CB(skb)->sacked&TCPCB_RETRANS)
tp->undo_marker = 0;
TCP_SKB_CB(skb)->sacked &= (~TCPCB_TAGBITS)|TCPCB_SACKED_ACKED;
if (!(TCP_SKB_CB(skb)->sacked&TCPCB_SACKED_ACKED) || how) {
TCP_SKB_CB(skb)->sacked &= ~TCPCB_SACKED_ACKED;
TCP_SKB_CB(skb)->sacked |= TCPCB_LOST;
- tp->lost_out++;
+ tcp_inc_pcount(&tp->lost_out, skb);
} else {
- tp->sacked_out++;
- tp->fackets_out = cnt;
+ tcp_inc_pcount(&tp->sacked_out, skb);
+ tcp_set_pcount(&tp->fackets_out, cnt);
}
}
tcp_sync_left_out(tp);
*/
if ((skb = skb_peek(&sk->sk_write_queue)) != NULL &&
(TCP_SKB_CB(skb)->sacked & TCPCB_SACKED_ACKED)) {
- NET_INC_STATS_BH(TCPSACKReneging);
+ NET_INC_STATS_BH(LINUX_MIB_TCPSACKRENEGING);
tcp_enter_loss(sk, 1);
tp->retransmits++;
static inline int tcp_fackets_out(struct tcp_opt *tp)
{
- return IsReno(tp) ? tp->sacked_out+1 : tp->fackets_out;
+ return IsReno(tp) ? tcp_get_pcount(&tp->sacked_out)+1 :
+ tcp_get_pcount(&tp->fackets_out);
}
static inline int tcp_skb_timedout(struct tcp_opt *tp, struct sk_buff *skb)
static inline int tcp_head_timedout(struct sock *sk, struct tcp_opt *tp)
{
- return tp->packets_out &&
+ return tcp_get_pcount(&tp->packets_out) &&
tcp_skb_timedout(tp, skb_peek(&sk->sk_write_queue));
}
static int
tcp_time_to_recover(struct sock *sk, struct tcp_opt *tp)
{
+ __u32 packets_out;
+
/* Trick#1: The loss is proven. */
- if (tp->lost_out)
+ if (tcp_get_pcount(&tp->lost_out))
return 1;
/* Not-A-Trick#2 : Classic rule... */
/* Trick#4: It is still not OK... But will it be useful to delay
* recovery more?
*/
- if (tp->packets_out <= tp->reordering &&
- tp->sacked_out >= max_t(__u32, tp->packets_out/2, sysctl_tcp_reordering) &&
+ packets_out = tcp_get_pcount(&tp->packets_out);
+ if (packets_out <= tp->reordering &&
+ tcp_get_pcount(&tp->sacked_out) >= max_t(__u32, packets_out/2, sysctl_tcp_reordering) &&
!tcp_may_send_now(sk, tp)) {
/* We have nothing to send. This connection is limited
* either by receiver window or by application.
{
u32 holes;
- holes = max(tp->lost_out, 1U);
- holes = min(holes, tp->packets_out);
+ holes = max(tcp_get_pcount(&tp->lost_out), 1U);
+ holes = min(holes, tcp_get_pcount(&tp->packets_out));
- if (tp->sacked_out + holes > tp->packets_out) {
- tp->sacked_out = tp->packets_out - holes;
- tcp_update_reordering(tp, tp->packets_out+addend, 0);
+ if ((tcp_get_pcount(&tp->sacked_out) + holes) >
+ tcp_get_pcount(&tp->packets_out)) {
+ tcp_set_pcount(&tp->sacked_out,
+ (tcp_get_pcount(&tp->packets_out) - holes));
+ tcp_update_reordering(tp,
+ tcp_get_pcount(&tp->packets_out)+addend,
+ 0);
}
}
static void tcp_add_reno_sack(struct tcp_opt *tp)
{
- ++tp->sacked_out;
+ tcp_inc_pcount_explicit(&tp->sacked_out, 1);
tcp_check_reno_reordering(tp, 0);
tcp_sync_left_out(tp);
}
{
if (acked > 0) {
/* One ACK acked hole. The rest eat duplicate ACKs. */
- if (acked-1 >= tp->sacked_out)
- tp->sacked_out = 0;
+ if (acked-1 >= tcp_get_pcount(&tp->sacked_out))
+ tcp_set_pcount(&tp->sacked_out, 0);
else
- tp->sacked_out -= acked-1;
+ tcp_dec_pcount_explicit(&tp->sacked_out, acked-1);
}
tcp_check_reno_reordering(tp, acked);
tcp_sync_left_out(tp);
static inline void tcp_reset_reno_sack(struct tcp_opt *tp)
{
- tp->sacked_out = 0;
- tp->left_out = tp->lost_out;
+ tcp_set_pcount(&tp->sacked_out, 0);
+ tcp_set_pcount(&tp->left_out, tcp_get_pcount(&tp->lost_out));
}
/* Mark head of queue up as lost. */
struct sk_buff *skb;
int cnt = packets;
- BUG_TRAP(cnt <= tp->packets_out);
+ BUG_TRAP(cnt <= tcp_get_pcount(&tp->packets_out));
- for_retrans_queue(skb, sk, tp) {
- if (--cnt < 0 || after(TCP_SKB_CB(skb)->end_seq, high_seq))
+ sk_stream_for_retrans_queue(skb, sk) {
+ cnt -= tcp_skb_pcount(skb);
+ if (cnt < 0 || after(TCP_SKB_CB(skb)->end_seq, high_seq))
break;
if (!(TCP_SKB_CB(skb)->sacked&TCPCB_TAGBITS)) {
TCP_SKB_CB(skb)->sacked |= TCPCB_LOST;
- tp->lost_out++;
+ tcp_inc_pcount(&tp->lost_out, skb);
}
}
tcp_sync_left_out(tp);
static void tcp_update_scoreboard(struct sock *sk, struct tcp_opt *tp)
{
if (IsFack(tp)) {
- int lost = tp->fackets_out - tp->reordering;
+ int lost = tcp_get_pcount(&tp->fackets_out) - tp->reordering;
if (lost <= 0)
lost = 1;
tcp_mark_head_lost(sk, tp, lost, tp->high_seq);
if (tcp_head_timedout(sk, tp)) {
struct sk_buff *skb;
- for_retrans_queue(skb, sk, tp) {
+ sk_stream_for_retrans_queue(skb, sk) {
if (tcp_skb_timedout(tp, skb) &&
!(TCP_SKB_CB(skb)->sacked&TCPCB_TAGBITS)) {
TCP_SKB_CB(skb)->sacked |= TCPCB_LOST;
- tp->lost_out++;
+ tcp_inc_pcount(&tp->lost_out, skb);
}
}
tcp_sync_left_out(tp);
printk(KERN_DEBUG "Undo %s %u.%u.%u.%u/%u c%u l%u ss%u/%u p%u\n",
msg,
NIPQUAD(inet->daddr), ntohs(inet->dport),
- tp->snd_cwnd, tp->left_out,
- tp->snd_ssthresh, tp->prior_ssthresh, tp->packets_out);
+ tp->snd_cwnd, tcp_get_pcount(&tp->left_out),
+ tp->snd_ssthresh, tp->prior_ssthresh,
+ tcp_get_pcount(&tp->packets_out));
}
#else
#define DBGUNDO(x...) do { } while (0)
DBGUNDO(sk, tp, tp->ca_state == TCP_CA_Loss ? "loss" : "retrans");
tcp_undo_cwr(tp, 1);
if (tp->ca_state == TCP_CA_Loss)
- NET_INC_STATS_BH(TCPLossUndo);
+ NET_INC_STATS_BH(LINUX_MIB_TCPLOSSUNDO);
else
- NET_INC_STATS_BH(TCPFullUndo);
+ NET_INC_STATS_BH(LINUX_MIB_TCPFULLUNDO);
tp->undo_marker = 0;
}
if (tp->snd_una == tp->high_seq && IsReno(tp)) {
DBGUNDO(sk, tp, "D-SACK");
tcp_undo_cwr(tp, 1);
tp->undo_marker = 0;
- NET_INC_STATS_BH(TCPDSACKUndo);
+ NET_INC_STATS_BH(LINUX_MIB_TCPDSACKUNDO);
}
}
static int tcp_try_undo_partial(struct sock *sk, struct tcp_opt *tp, int acked)
{
/* Partial ACK arrived. Force Hoe's retransmit. */
- int failed = IsReno(tp) || tp->fackets_out>tp->reordering;
+ int failed = IsReno(tp) || tcp_get_pcount(&tp->fackets_out)>tp->reordering;
if (tcp_may_undo(tp)) {
/* Plain luck! Hole if filled with delayed
* packet, rather than with a retransmit.
*/
- if (tp->retrans_out == 0)
+ if (tcp_get_pcount(&tp->retrans_out) == 0)
tp->retrans_stamp = 0;
tcp_update_reordering(tp, tcp_fackets_out(tp)+acked, 1);
DBGUNDO(sk, tp, "Hoe");
tcp_undo_cwr(tp, 0);
- NET_INC_STATS_BH(TCPPartialUndo);
+ NET_INC_STATS_BH(LINUX_MIB_TCPPARTIALUNDO);
/* So... Do not make Hoe's retransmit yet.
* If the first packet was delayed, the rest
{
if (tcp_may_undo(tp)) {
struct sk_buff *skb;
- for_retrans_queue(skb, sk, tp) {
+ sk_stream_for_retrans_queue(skb, sk) {
TCP_SKB_CB(skb)->sacked &= ~TCPCB_LOST;
}
DBGUNDO(sk, tp, "partial loss");
- tp->lost_out = 0;
- tp->left_out = tp->sacked_out;
+ tcp_set_pcount(&tp->lost_out, 0);
+ tcp_set_pcount(&tp->left_out, tcp_get_pcount(&tp->sacked_out));
tcp_undo_cwr(tp, 1);
- NET_INC_STATS_BH(TCPLossUndo);
+ NET_INC_STATS_BH(LINUX_MIB_TCPLOSSUNDO);
tp->retransmits = 0;
tp->undo_marker = 0;
if (!IsReno(tp))
static void tcp_try_to_open(struct sock *sk, struct tcp_opt *tp, int flag)
{
- tp->left_out = tp->sacked_out;
+ tcp_set_pcount(&tp->left_out, tcp_get_pcount(&tp->sacked_out));
- if (tp->retrans_out == 0)
+ if (tcp_get_pcount(&tp->retrans_out) == 0)
tp->retrans_stamp = 0;
if (flag&FLAG_ECE)
if (tp->ca_state != TCP_CA_CWR) {
int state = TCP_CA_Open;
- if (tp->left_out ||
- tp->retrans_out ||
+ if (tcp_get_pcount(&tp->left_out) ||
+ tcp_get_pcount(&tp->retrans_out) ||
tp->undo_marker)
state = TCP_CA_Disorder;
/* Some technical things:
* 1. Reno does not count dupacks (sacked_out) automatically. */
- if (!tp->packets_out)
- tp->sacked_out = 0;
+ if (!tcp_get_pcount(&tp->packets_out))
+ tcp_set_pcount(&tp->sacked_out, 0);
/* 2. SACK counts snd_fack in packets inaccurately. */
- if (tp->sacked_out == 0)
- tp->fackets_out = 0;
+ if (tcp_get_pcount(&tp->sacked_out) == 0)
+ tcp_set_pcount(&tp->fackets_out, 0);
/* Now state machine starts.
* A. ECE, hence prohibit cwnd undoing, the reduction is required. */
tp->prior_ssthresh = 0;
/* B. In all the states check for reneging SACKs. */
- if (tp->sacked_out && tcp_check_sack_reneging(sk, tp))
+ if (tcp_get_pcount(&tp->sacked_out) && tcp_check_sack_reneging(sk, tp))
return;
/* C. Process data loss notification, provided it is valid. */
if ((flag&FLAG_DATA_LOST) &&
before(tp->snd_una, tp->high_seq) &&
tp->ca_state != TCP_CA_Open &&
- tp->fackets_out > tp->reordering) {
- tcp_mark_head_lost(sk, tp, tp->fackets_out-tp->reordering, tp->high_seq);
- NET_INC_STATS_BH(TCPLoss);
+ tcp_get_pcount(&tp->fackets_out) > tp->reordering) {
+ tcp_mark_head_lost(sk, tp, tcp_get_pcount(&tp->fackets_out)-tp->reordering, tp->high_seq);
+ NET_INC_STATS_BH(LINUX_MIB_TCPLOSS);
}
/* D. Synchronize left_out to current state. */
* when high_seq is ACKed. */
if (tp->ca_state == TCP_CA_Open) {
if (!sysctl_tcp_frto)
- BUG_TRAP(tp->retrans_out == 0);
+ BUG_TRAP(tcp_get_pcount(&tp->retrans_out) == 0);
tp->retrans_stamp = 0;
} else if (!before(tp->snd_una, tp->high_seq)) {
switch (tp->ca_state) {
if (IsReno(tp) && is_dupack)
tcp_add_reno_sack(tp);
} else {
- int acked = prior_packets - tp->packets_out;
+ int acked = prior_packets -
+ tcp_get_pcount(&tp->packets_out);
if (IsReno(tp))
tcp_remove_reno_sacks(sk, tp, acked);
is_dupack = tcp_try_undo_partial(sk, tp, acked);
/* Otherwise enter Recovery state */
if (IsReno(tp))
- NET_INC_STATS_BH(TCPRenoRecovery);
+ NET_INC_STATS_BH(LINUX_MIB_TCPRENORECOVERY);
else
- NET_INC_STATS_BH(TCPSackRecovery);
+ NET_INC_STATS_BH(LINUX_MIB_TCPSACKRECOVERY);
tp->high_seq = tp->snd_nxt;
tp->prior_ssthresh = 0;
tp->undo_marker = tp->snd_una;
- tp->undo_retrans = tp->retrans_out;
+ tp->undo_retrans = tcp_get_pcount(&tp->retrans_out);
if (tp->ca_state < TCP_CA_CWR) {
if (!(flag&FLAG_ECE))
static inline __u32 bictcp_cwnd(struct tcp_opt *tp)
{
/* orignal Reno behaviour */
- if (!sysctl_tcp_bic)
+ if (!tcp_is_bic(tp))
return tp->snd_cwnd;
- if (tp->bictcp.last_cwnd == tp->snd_cwnd)
- return tp->bictcp.cnt; /* same cwnd, no update */
-
+ if (tp->bictcp.last_cwnd == tp->snd_cwnd &&
+ (s32)(tcp_time_stamp - tp->bictcp.last_stamp) <= (HZ>>5))
+ return tp->bictcp.cnt;
+
tp->bictcp.last_cwnd = tp->snd_cwnd;
+ tp->bictcp.last_stamp = tcp_time_stamp;
/* start off normal */
if (tp->snd_cwnd <= sysctl_tcp_bic_low_window)
* is the cwnd during the previous RTT.
*/
old_wnd = (tp->vegas.beg_snd_nxt - tp->vegas.beg_snd_una) /
- tp->mss_cache;
+ tp->mss_cache_std;
old_snd_cwnd = tp->vegas.beg_snd_cwnd;
/* Save the extent of the current window so we can use this
static __inline__ void tcp_ack_packets_out(struct sock *sk, struct tcp_opt *tp)
{
- if (tp->packets_out==0) {
+ if (!tcp_get_pcount(&tp->packets_out)) {
tcp_clear_xmit_timer(sk, TCP_TIME_RETRANS);
} else {
tcp_reset_xmit_timer(sk, TCP_TIME_RETRANS, tp->rto);
}
}
+/* There is one downside to this scheme. Although we keep the
+ * ACK clock ticking, adjusting packet counters and advancing
+ * congestion window, we do not liberate socket send buffer
+ * space.
+ *
+ * Mucking with skb->truesize and sk->sk_wmem_alloc et al.
+ * then making a write space wakeup callback is a possible
+ * future enhancement. WARNING: it is not trivial to make.
+ */
+static int tcp_tso_acked(struct sock *sk, struct sk_buff *skb,
+ __u32 now, __s32 *seq_rtt)
+{
+ struct tcp_opt *tp = tcp_sk(sk);
+ struct tcp_skb_cb *scb = TCP_SKB_CB(skb);
+ __u32 mss = tcp_skb_mss(skb);
+ __u32 snd_una = tp->snd_una;
+ __u32 orig_seq, seq;
+ __u32 packets_acked = 0;
+ int acked = 0;
+
+ /* If we get here, the whole TSO packet has not been
+ * acked.
+ */
+ BUG_ON(!after(scb->end_seq, snd_una));
+
+ seq = orig_seq = scb->seq;
+ while (!after(seq + mss, snd_una)) {
+ packets_acked++;
+ seq += mss;
+ }
+
+ if (tcp_trim_head(sk, skb, (seq - orig_seq)))
+ return 0;
+
+ if (packets_acked) {
+ __u8 sacked = scb->sacked;
+
+ acked |= FLAG_DATA_ACKED;
+ if (sacked) {
+ if (sacked & TCPCB_RETRANS) {
+ if (sacked & TCPCB_SACKED_RETRANS)
+ tcp_dec_pcount_explicit(&tp->retrans_out,
+ packets_acked);
+ acked |= FLAG_RETRANS_DATA_ACKED;
+ *seq_rtt = -1;
+ } else if (*seq_rtt < 0)
+ *seq_rtt = now - scb->when;
+ if (sacked & TCPCB_SACKED_ACKED)
+ tcp_dec_pcount_explicit(&tp->sacked_out,
+ packets_acked);
+ if (sacked & TCPCB_LOST)
+ tcp_dec_pcount_explicit(&tp->lost_out,
+ packets_acked);
+ if (sacked & TCPCB_URG) {
+ if (tp->urg_mode &&
+ !before(seq, tp->snd_up))
+ tp->urg_mode = 0;
+ }
+ } else if (*seq_rtt < 0)
+ *seq_rtt = now - scb->when;
+
+ if (tcp_get_pcount(&tp->fackets_out)) {
+ __u32 dval = min(tcp_get_pcount(&tp->fackets_out),
+ packets_acked);
+ tcp_dec_pcount_explicit(&tp->fackets_out, dval);
+ }
+ tcp_dec_pcount_explicit(&tp->packets_out, packets_acked);
+
+ BUG_ON(tcp_skb_pcount(skb) == 0);
+ BUG_ON(!before(scb->seq, scb->end_seq));
+ }
+
+ return acked;
+}
+
+
/* Remove acknowledged frames from the retransmission queue. */
static int tcp_clean_rtx_queue(struct sock *sk, __s32 *seq_rtt_p)
{
int acked = 0;
__s32 seq_rtt = -1;
- while ((skb = skb_peek(&sk->sk_write_queue)) && skb != tp->send_head) {
+ while ((skb = skb_peek(&sk->sk_write_queue)) &&
+ skb != sk->sk_send_head) {
struct tcp_skb_cb *scb = TCP_SKB_CB(skb);
__u8 sacked = scb->sacked;
* discard it as it's confirmed to have arrived at
* the other end.
*/
- if (after(scb->end_seq, tp->snd_una))
+ if (after(scb->end_seq, tp->snd_una)) {
+ if (tcp_skb_pcount(skb) > 1)
+ acked |= tcp_tso_acked(sk, skb,
+ now, &seq_rtt);
break;
+ }
/* Initial outgoing SYN's get put onto the write_queue
* just like anything else we transmit. It is not
* connection startup slow start one packet too
* quickly. This is severely frowned upon behavior.
*/
- if(!(scb->flags & TCPCB_FLAG_SYN)) {
+ if (!(scb->flags & TCPCB_FLAG_SYN)) {
acked |= FLAG_DATA_ACKED;
} else {
acked |= FLAG_SYN_ACKED;
}
if (sacked) {
- if(sacked & TCPCB_RETRANS) {
+ if (sacked & TCPCB_RETRANS) {
if(sacked & TCPCB_SACKED_RETRANS)
- tp->retrans_out--;
+ tcp_dec_pcount(&tp->retrans_out, skb);
acked |= FLAG_RETRANS_DATA_ACKED;
seq_rtt = -1;
} else if (seq_rtt < 0)
seq_rtt = now - scb->when;
- if(sacked & TCPCB_SACKED_ACKED)
- tp->sacked_out--;
- if(sacked & TCPCB_LOST)
- tp->lost_out--;
- if(sacked & TCPCB_URG) {
+ if (sacked & TCPCB_SACKED_ACKED)
+ tcp_dec_pcount(&tp->sacked_out, skb);
+ if (sacked & TCPCB_LOST)
+ tcp_dec_pcount(&tp->lost_out, skb);
+ if (sacked & TCPCB_URG) {
if (tp->urg_mode &&
!before(scb->end_seq, tp->snd_up))
tp->urg_mode = 0;
}
} else if (seq_rtt < 0)
seq_rtt = now - scb->when;
- if (tp->fackets_out)
- tp->fackets_out--;
- tp->packets_out--;
+ tcp_dec_pcount_approx(&tp->fackets_out, skb);
+ tcp_packets_out_dec(tp, skb);
__skb_unlink(skb, skb->list);
- tcp_free_skb(sk, skb);
+ sk_stream_free_skb(sk, skb);
}
if (acked&FLAG_ACKED) {
}
#if FASTRETRANS_DEBUG > 0
- BUG_TRAP((int)tp->sacked_out >= 0);
- BUG_TRAP((int)tp->lost_out >= 0);
- BUG_TRAP((int)tp->retrans_out >= 0);
- if (!tp->packets_out && tp->sack_ok) {
- if (tp->lost_out) {
- printk(KERN_DEBUG "Leak l=%u %d\n", tp->lost_out,
- tp->ca_state);
- tp->lost_out = 0;
+ BUG_TRAP((int)tcp_get_pcount(&tp->sacked_out) >= 0);
+ BUG_TRAP((int)tcp_get_pcount(&tp->lost_out) >= 0);
+ BUG_TRAP((int)tcp_get_pcount(&tp->retrans_out) >= 0);
+ if (!tcp_get_pcount(&tp->packets_out) && tp->sack_ok) {
+ if (tcp_get_pcount(&tp->lost_out)) {
+ printk(KERN_DEBUG "Leak l=%u %d\n",
+ tcp_get_pcount(&tp->lost_out),
+ tp->ca_state);
+ tcp_set_pcount(&tp->lost_out, 0);
}
- if (tp->sacked_out) {
- printk(KERN_DEBUG "Leak s=%u %d\n", tp->sacked_out,
- tp->ca_state);
- tp->sacked_out = 0;
+ if (tcp_get_pcount(&tp->sacked_out)) {
+ printk(KERN_DEBUG "Leak s=%u %d\n",
+ tcp_get_pcount(&tp->sacked_out),
+ tp->ca_state);
+ tcp_set_pcount(&tp->sacked_out, 0);
}
- if (tp->retrans_out) {
- printk(KERN_DEBUG "Leak r=%u %d\n", tp->retrans_out,
- tp->ca_state);
- tp->retrans_out = 0;
+ if (tcp_get_pcount(&tp->retrans_out)) {
+ printk(KERN_DEBUG "Leak r=%u %d\n",
+ tcp_get_pcount(&tp->retrans_out),
+ tp->ca_state);
+ tcp_set_pcount(&tp->retrans_out, 0);
}
}
#endif
/* Was it a usable window open? */
- if (!after(TCP_SKB_CB(tp->send_head)->end_seq,
+ if (!after(TCP_SKB_CB(sk->sk_send_head)->end_seq,
tp->snd_una + tp->snd_wnd)) {
tp->backoff = 0;
tcp_clear_xmit_timer(sk, TCP_TIME_PROBE0);
* WESTWOOD_RTT_MIN minimum bound since we could be on a LAN!
*/
-static inline __u32 westwood_update_rttmin(struct sock *sk)
+static inline __u32 westwood_update_rttmin(const struct sock *sk)
{
- struct tcp_opt *tp = tcp_sk(sk);
+ const struct tcp_opt *tp = tcp_sk(sk);
__u32 rttmin = tp->westwood.rtt_min;
- if (tp->westwood.rtt == 0)
- return(rttmin);
-
- if (tp->westwood.rtt < tp->westwood.rtt_min || !rttmin)
+ if (tp->westwood.rtt != 0 &&
+ (tp->westwood.rtt < tp->westwood.rtt_min || !rttmin))
rttmin = tp->westwood.rtt;
- return(rttmin);
+ return rttmin;
}
/*
* Evaluate increases for dk.
*/
-static inline __u32 westwood_acked(struct sock *sk)
+static inline __u32 westwood_acked(const struct sock *sk)
{
- struct tcp_opt *tp = tcp_sk(sk);
+ const struct tcp_opt *tp = tcp_sk(sk);
- return ((tp->snd_una) - (tp->westwood.snd_una));
+ return tp->snd_una - tp->westwood.snd_una;
}
/*
* window, 1 if the sample has to be considered in the next window.
*/
-static int westwood_new_window(struct sock *sk)
+static int westwood_new_window(const struct sock *sk)
{
- struct tcp_opt *tp = tcp_sk(sk);
+ const struct tcp_opt *tp = tcp_sk(sk);
__u32 left_bound;
__u32 rtt;
int ret = 0;
struct tcp_opt *tp = tcp_sk(sk);
__u32 delta = now - tp->westwood.rtt_win_sx;
- if (!delta)
- return;
+ if (delta) {
+ if (tp->westwood.rtt)
+ westwood_filter(sk, delta);
- if (tp->westwood.rtt)
- westwood_filter(sk, delta);
-
- tp->westwood.bk = 0;
- tp->westwood.rtt_win_sx = tcp_time_stamp;
+ tp->westwood.bk = 0;
+ tp->westwood.rtt_win_sx = tcp_time_stamp;
+ }
}
{
struct tcp_opt *tp = tcp_sk(sk);
- tp->westwood.accounted += tp->mss_cache;
- tp->westwood.cumul_ack = tp->mss_cache;
+ tp->westwood.accounted += tp->mss_cache_std;
+ tp->westwood.cumul_ack = tp->mss_cache_std;
}
static inline int westwood_may_change_cumul(struct tcp_opt *tp)
{
- return ((tp->westwood.cumul_ack) > tp->mss_cache);
+ return (tp->westwood.cumul_ack > tp->mss_cache_std);
}
static inline void westwood_partial_update(struct tcp_opt *tp)
{
tp->westwood.accounted -= tp->westwood.cumul_ack;
- tp->westwood.cumul_ack = tp->mss_cache;
+ tp->westwood.cumul_ack = tp->mss_cache_std;
}
static inline void westwood_complete_update(struct tcp_opt *tp)
* delayed or partial acks.
*/
-static __u32 westwood_acked_count(struct sock *sk)
+static inline __u32 westwood_acked_count(struct sock *sk)
{
struct tcp_opt *tp = tcp_sk(sk);
if (westwood_may_change_cumul(tp)) {
/* Partial or delayed ack */
- if ((tp->westwood.accounted) >= (tp->westwood.cumul_ack))
+ if (tp->westwood.accounted >= tp->westwood.cumul_ack)
westwood_partial_update(tp);
else
westwood_complete_update(tp);
tcp_westwood_fast_bw(sk, skb);
flag |= FLAG_WIN_UPDATE;
- NET_INC_STATS_BH(TCPHPAcks);
+ NET_INC_STATS_BH(LINUX_MIB_TCPHPACKS);
} else {
if (ack_seq != TCP_SKB_CB(skb)->end_seq)
flag |= FLAG_DATA;
else
- NET_INC_STATS_BH(TCPPureAcks);
+ NET_INC_STATS_BH(LINUX_MIB_TCPPUREACKS);
flag |= tcp_ack_update_window(sk, tp, skb, ack, ack_seq);
*/
sk->sk_err_soft = 0;
tp->rcv_tstamp = tcp_time_stamp;
- prior_packets = tp->packets_out;
+ prior_packets = tcp_get_pcount(&tp->packets_out);
if (!prior_packets)
goto no_queue;
* being used to time the probes, and is probably far higher than
* it needs to be for normal retransmission.
*/
- if (tp->send_head)
+ if (sk->sk_send_head)
tcp_ack_probe(sk);
return 1;
__skb_queue_purge(&tp->out_of_order_queue);
if (tp->sack_ok)
tcp_sack_reset(tp);
- tcp_mem_reclaim(sk);
+ sk_stream_mem_reclaim(sk);
if (!sock_flag(sk, SOCK_DEAD)) {
sk->sk_state_change(sk);
{
if (tp->sack_ok && sysctl_tcp_dsack) {
if (before(seq, tp->rcv_nxt))
- NET_INC_STATS_BH(TCPDSACKOldSent);
+ NET_INC_STATS_BH(LINUX_MIB_TCPDSACKOLDSENT);
else
- NET_INC_STATS_BH(TCPDSACKOfoSent);
+ NET_INC_STATS_BH(LINUX_MIB_TCPDSACKOFOSENT);
tp->dsack = 1;
tp->duplicate_sack[0].start_seq = seq;
if (TCP_SKB_CB(skb)->end_seq != TCP_SKB_CB(skb)->seq &&
before(TCP_SKB_CB(skb)->seq, tp->rcv_nxt)) {
- NET_INC_STATS_BH(DelayedACKLost);
+ NET_INC_STATS_BH(LINUX_MIB_DELAYEDACKLOST);
tcp_enter_quickack_mode(tp);
if (tp->sack_ok && sysctl_tcp_dsack) {
}
}
-static inline int tcp_rmem_schedule(struct sock *sk, struct sk_buff *skb)
-{
- return (int)skb->truesize <= sk->sk_forward_alloc ||
- tcp_mem_schedule(sk, skb->truesize, 1);
-}
-
static int tcp_prune_queue(struct sock *sk);
static void tcp_data_queue(struct sock *sk, struct sk_buff *skb)
queue_and_out:
if (eaten < 0 &&
(atomic_read(&sk->sk_rmem_alloc) > sk->sk_rcvbuf ||
- !tcp_rmem_schedule(sk, skb))) {
- if (tcp_prune_queue(sk) < 0 || !tcp_rmem_schedule(sk, skb))
+ !sk_stream_rmem_schedule(sk, skb))) {
+ if (tcp_prune_queue(sk) < 0 ||
+ !sk_stream_rmem_schedule(sk, skb))
goto drop;
}
- tcp_set_owner_r(skb, sk);
+ sk_stream_set_owner_r(skb, sk);
__skb_queue_tail(&sk->sk_receive_queue, skb);
}
tp->rcv_nxt = TCP_SKB_CB(skb)->end_seq;
if (!after(TCP_SKB_CB(skb)->end_seq, tp->rcv_nxt)) {
/* A retransmit, 2nd most common case. Force an immediate ack. */
- NET_INC_STATS_BH(DelayedACKLost);
+ NET_INC_STATS_BH(LINUX_MIB_DELAYEDACKLOST);
tcp_dsack_set(tp, TCP_SKB_CB(skb)->seq, TCP_SKB_CB(skb)->end_seq);
out_of_window:
TCP_ECN_check_ce(tp, skb);
if (atomic_read(&sk->sk_rmem_alloc) > sk->sk_rcvbuf ||
- !tcp_rmem_schedule(sk, skb)) {
- if (tcp_prune_queue(sk) < 0 || !tcp_rmem_schedule(sk, skb))
+ !sk_stream_rmem_schedule(sk, skb)) {
+ if (tcp_prune_queue(sk) < 0 ||
+ !sk_stream_rmem_schedule(sk, skb))
goto drop;
}
SOCK_DEBUG(sk, "out of order segment: rcv_next %X seq %X - %X\n",
tp->rcv_nxt, TCP_SKB_CB(skb)->seq, TCP_SKB_CB(skb)->end_seq);
- tcp_set_owner_r(skb, sk);
+ sk_stream_set_owner_r(skb, sk);
if (!skb_peek(&tp->out_of_order_queue)) {
/* Initial out of order segment, build 1 SACK. */
struct sk_buff *next = skb->next;
__skb_unlink(skb, skb->list);
__kfree_skb(skb);
- NET_INC_STATS_BH(TCPRcvCollapsed);
+ NET_INC_STATS_BH(LINUX_MIB_TCPRCVCOLLAPSED);
skb = next;
continue;
}
memcpy(nskb->cb, skb->cb, sizeof(skb->cb));
TCP_SKB_CB(nskb)->seq = TCP_SKB_CB(nskb)->end_seq = start;
__skb_insert(nskb, skb->prev, skb, skb->list);
- tcp_set_owner_r(nskb, sk);
+ sk_stream_set_owner_r(nskb, sk);
/* Copy data, releasing collapsed skbs. */
while (copy > 0) {
struct sk_buff *next = skb->next;
__skb_unlink(skb, skb->list);
__kfree_skb(skb);
- NET_INC_STATS_BH(TCPRcvCollapsed);
+ NET_INC_STATS_BH(LINUX_MIB_TCPRCVCOLLAPSED);
skb = next;
if (skb == tail || skb->h.th->syn || skb->h.th->fin)
return;
SOCK_DEBUG(sk, "prune_queue: c=%x\n", tp->copied_seq);
- NET_INC_STATS_BH(PruneCalled);
+ NET_INC_STATS_BH(LINUX_MIB_PRUNECALLED);
if (atomic_read(&sk->sk_rmem_alloc) >= sk->sk_rcvbuf)
tcp_clamp_window(sk, tp);
tcp_collapse(sk, sk->sk_receive_queue.next,
(struct sk_buff*)&sk->sk_receive_queue,
tp->copied_seq, tp->rcv_nxt);
- tcp_mem_reclaim(sk);
+ sk_stream_mem_reclaim(sk);
if (atomic_read(&sk->sk_rmem_alloc) <= sk->sk_rcvbuf)
return 0;
/* First, purge the out_of_order queue. */
if (skb_queue_len(&tp->out_of_order_queue)) {
- NET_ADD_STATS_BH(OfoPruned,
+ NET_ADD_STATS_BH(LINUX_MIB_OFOPRUNED,
skb_queue_len(&tp->out_of_order_queue));
__skb_queue_purge(&tp->out_of_order_queue);
*/
if (tp->sack_ok)
tcp_sack_reset(tp);
- tcp_mem_reclaim(sk);
+ sk_stream_mem_reclaim(sk);
}
if (atomic_read(&sk->sk_rmem_alloc) <= sk->sk_rcvbuf)
* drop receive data on the floor. It will get retransmitted
* and hopefully then we'll have sufficient space.
*/
- NET_INC_STATS_BH(RcvPruned);
+ NET_INC_STATS_BH(LINUX_MIB_RCVPRUNED);
/* Massive buffer overcommit. */
tp->pred_flags = 0;
/* When incoming ACK allowed to free some skb from write_queue,
- * we remember this event in flag tp->queue_shrunk and wake up socket
+ * we remember this event in flag sk->sk_queue_shrunk and wake up socket
* on the exit from tcp input handler.
*
* PROBLEM: sndbuf expansion does not work well with largesend.
{
struct tcp_opt *tp = tcp_sk(sk);
- if (tp->packets_out < tp->snd_cwnd &&
+ if (tcp_get_pcount(&tp->packets_out) < tp->snd_cwnd &&
!(sk->sk_userlocks & SOCK_SNDBUF_LOCK) &&
!tcp_memory_pressure &&
atomic_read(&tcp_memory_allocated) < sysctl_tcp_mem[0]) {
- int sndmem = max_t(u32, tp->mss_clamp, tp->mss_cache) +
+ int sndmem = max_t(u32, tp->mss_clamp, tp->mss_cache_std) +
MAX_TCP_HEADER + 16 + sizeof(struct sk_buff),
demanded = max_t(unsigned int, tp->snd_cwnd,
tp->reordering + 1);
static inline void tcp_check_space(struct sock *sk)
{
- struct tcp_opt *tp = tcp_sk(sk);
-
- if (tp->queue_shrunk) {
- tp->queue_shrunk = 0;
+ if (sk->sk_queue_shrunk) {
+ sk->sk_queue_shrunk = 0;
if (sk->sk_socket &&
test_bit(SOCK_NOSPACE, &sk->sk_socket->flags))
tcp_new_space(sk);
static __inline__ void tcp_data_snd_check(struct sock *sk)
{
- struct tcp_opt *tp = tcp_sk(sk);
- struct sk_buff *skb = tp->send_head;
+ struct sk_buff *skb = sk->sk_send_head;
if (skb != NULL)
__tcp_data_snd_check(sk, skb);
tcp_data_snd_check(sk);
return 0;
} else { /* Header too small */
- TCP_INC_STATS_BH(TcpInErrs);
+ TCP_INC_STATS_BH(TCP_MIB_INERRS);
goto discard;
}
} else {
__skb_pull(skb, tcp_header_len);
tp->rcv_nxt = TCP_SKB_CB(skb)->end_seq;
- NET_INC_STATS_BH(TCPHPHitsToUser);
+ NET_INC_STATS_BH(LINUX_MIB_TCPHPHITSTOUSER);
eaten = 1;
}
}
if ((int)skb->truesize > sk->sk_forward_alloc)
goto step5;
- NET_INC_STATS_BH(TCPHPHits);
+ NET_INC_STATS_BH(LINUX_MIB_TCPHPHITS);
/* Bulk data transfer: receiver */
__skb_pull(skb,tcp_header_len);
__skb_queue_tail(&sk->sk_receive_queue, skb);
- tcp_set_owner_r(skb, sk);
+ sk_stream_set_owner_r(skb, sk);
tp->rcv_nxt = TCP_SKB_CB(skb)->end_seq;
}
if (tcp_fast_parse_options(skb, th, tp) && tp->saw_tstamp &&
tcp_paws_discard(tp, skb)) {
if (!th->rst) {
- NET_INC_STATS_BH(PAWSEstabRejected);
+ NET_INC_STATS_BH(LINUX_MIB_PAWSESTABREJECTED);
tcp_send_dupack(sk, skb);
goto discard;
}
tcp_replace_ts_recent(tp, TCP_SKB_CB(skb)->seq);
if (th->syn && !before(TCP_SKB_CB(skb)->seq, tp->rcv_nxt)) {
- TCP_INC_STATS_BH(TcpInErrs);
- NET_INC_STATS_BH(TCPAbortOnSyn);
+ TCP_INC_STATS_BH(TCP_MIB_INERRS);
+ NET_INC_STATS_BH(LINUX_MIB_TCPABORTONSYN);
tcp_reset(sk);
return 1;
}
return 0;
csum_error:
- TCP_INC_STATS_BH(TcpInErrs);
+ TCP_INC_STATS_BH(TCP_MIB_INERRS);
discard:
__kfree_skb(skb);
if (tp->saw_tstamp && tp->rcv_tsecr &&
!between(tp->rcv_tsecr, tp->retrans_stamp,
tcp_time_stamp)) {
- NET_INC_STATS_BH(PAWSActiveRejected);
+ NET_INC_STATS_BH(LINUX_MIB_PAWSACTIVEREJECTED);
goto reset_and_undo;
}
sk_wake_async(sk, 0, POLL_OUT);
}
- if (tp->write_pending || tp->defer_accept || tp->ack.pingpong) {
+ if (sk->sk_write_pending || tp->defer_accept || tp->ack.pingpong) {
/* Save one ACK. Data will be ready after
* several ticks, if write_pending is set.
*
return 1;
init_westwood(sk);
+ init_bictcp(tp);
/* Now we have several options: In theory there is
* nothing else in the frame. KA9Q has an option to
case TCP_SYN_SENT:
init_westwood(sk);
+ init_bictcp(tp);
queued = tcp_rcv_synsent_state_process(sk, skb, th, len);
if (queued >= 0)
if (tcp_fast_parse_options(skb, th, tp) && tp->saw_tstamp &&
tcp_paws_discard(tp, skb)) {
if (!th->rst) {
- NET_INC_STATS_BH(PAWSEstabRejected);
+ NET_INC_STATS_BH(LINUX_MIB_PAWSESTABREJECTED);
tcp_send_dupack(sk, skb);
goto discard;
}
* Check for a SYN in window.
*/
if (th->syn && !before(TCP_SKB_CB(skb)->seq, tp->rcv_nxt)) {
- NET_INC_STATS_BH(TCPAbortOnSyn);
+ NET_INC_STATS_BH(LINUX_MIB_TCPABORTONSYN);
tcp_reset(sk);
return 1;
}
(TCP_SKB_CB(skb)->end_seq != TCP_SKB_CB(skb)->seq &&
after(TCP_SKB_CB(skb)->end_seq - th->fin, tp->rcv_nxt))) {
tcp_done(sk);
- NET_INC_STATS_BH(TCPAbortOnData);
+ NET_INC_STATS_BH(LINUX_MIB_TCPABORTONDATA);
return 1;
}
if (sk->sk_shutdown & RCV_SHUTDOWN) {
if (TCP_SKB_CB(skb)->end_seq != TCP_SKB_CB(skb)->seq &&
after(TCP_SKB_CB(skb)->end_seq - th->fin, tp->rcv_nxt)) {
- NET_INC_STATS_BH(TCPAbortOnData);
+ NET_INC_STATS_BH(LINUX_MIB_TCPABORTONDATA);
tcp_reset(sk);
return 1;
}
git://git.onelab.eu / linux-2.6.git / blobdiff