netpoll deferred transmit path

[linux-2.6-omap-h63xx.git] / net / ipv4 / tcp_output.c

diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index 061edfae0c29df44ada4e1aec6d055a21eaf7f38..6a8581ab9a23787d784240ca1da65c4d4cbe0b04 100644 (file)
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -269,14 +269,14 @@ static u16 tcp_select_window(struct sock *sk)
        return new_win;
 }
 
-static void tcp_build_and_update_options(__u32 *ptr, struct tcp_sock *tp,
+static void tcp_build_and_update_options(__be32 *ptr, struct tcp_sock *tp,
                                         __u32 tstamp)
 {
        if (tp->rx_opt.tstamp_ok) {
-               *ptr++ = __constant_htonl((TCPOPT_NOP << 24) |
-                                         (TCPOPT_NOP << 16) |
-                                         (TCPOPT_TIMESTAMP << 8) |
-                                         TCPOLEN_TIMESTAMP);
+               *ptr++ = htonl((TCPOPT_NOP << 24) |
+                              (TCPOPT_NOP << 16) |
+                              (TCPOPT_TIMESTAMP << 8) |
+                              TCPOLEN_TIMESTAMP);
                *ptr++ = htonl(tstamp);
                *ptr++ = htonl(tp->rx_opt.ts_recent);
        }
@@ -305,7 +305,7 @@ static void tcp_build_and_update_options(__u32 *ptr, struct tcp_sock *tp,
  * MAX_SYN_SIZE to match the new maximum number of options that you
  * can generate.
  */
-static void tcp_syn_build_options(__u32 *ptr, int mss, int ts, int sack,
+static void tcp_syn_build_options(__be32 *ptr, int mss, int ts, int sack,
                                  int offer_wscale, int wscale, __u32 tstamp,
                                  __u32 ts_recent)
 {
@@ -325,18 +325,27 @@ static void tcp_syn_build_options(__u32 *ptr, int mss, int ts, int sack,
        *ptr++ = htonl((TCPOPT_MSS << 24) | (TCPOLEN_MSS << 16) | mss);
        if (ts) {
                if(sack)
-                       *ptr++ = __constant_htonl((TCPOPT_SACK_PERM << 24) | (TCPOLEN_SACK_PERM << 16) |
-                                                 (TCPOPT_TIMESTAMP << 8) | TCPOLEN_TIMESTAMP);
+                       *ptr++ = htonl((TCPOPT_SACK_PERM << 24) |
+                                      (TCPOLEN_SACK_PERM << 16) |
+                                      (TCPOPT_TIMESTAMP << 8) |
+                                      TCPOLEN_TIMESTAMP);
                else
-                       *ptr++ = __constant_htonl((TCPOPT_NOP << 24) | (TCPOPT_NOP << 16) |
-                                                 (TCPOPT_TIMESTAMP << 8) | TCPOLEN_TIMESTAMP);
+                       *ptr++ = htonl((TCPOPT_NOP << 24) |
+                                      (TCPOPT_NOP << 16) |
+                                      (TCPOPT_TIMESTAMP << 8) |
+                                      TCPOLEN_TIMESTAMP);
                *ptr++ = htonl(tstamp);         /* TSVAL */
                *ptr++ = htonl(ts_recent);      /* TSECR */
        } else if(sack)
-               *ptr++ = __constant_htonl((TCPOPT_NOP << 24) | (TCPOPT_NOP << 16) |
-                                         (TCPOPT_SACK_PERM << 8) | TCPOLEN_SACK_PERM);
+               *ptr++ = htonl((TCPOPT_NOP << 24) |
+                              (TCPOPT_NOP << 16) |
+                              (TCPOPT_SACK_PERM << 8) |
+                              TCPOLEN_SACK_PERM);
        if (offer_wscale)
-               *ptr++ = htonl((TCPOPT_NOP << 24) | (TCPOPT_WINDOW << 16) | (TCPOLEN_WINDOW << 8) | (wscale));
+               *ptr++ = htonl((TCPOPT_NOP << 24) |
+                              (TCPOPT_WINDOW << 16) |
+                              (TCPOLEN_WINDOW << 8) |
+                              (wscale));
 }
 
 /* This routine actually transmits TCP packets queued in by
@@ -417,14 +426,13 @@ static int tcp_transmit_skb(struct sock *sk, struct sk_buff *skb, int clone_it,
 
        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              = inet->sport;
        th->dest                = inet->dport;
        th->seq                 = htonl(tcb->seq);
        th->ack_seq             = htonl(tp->rcv_nxt);
-       *(((__u16 *)th) + 6)    = htons(((tcp_header_size >> 2) << 12) |
+       *(((__be16 *)th) + 6)   = htons(((tcp_header_size >> 2) << 12) |
                                        tcb->flags);
 
        if (unlikely(tcb->flags & TCPCB_FLAG_SYN)) {
@@ -445,7 +453,7 @@ static int tcp_transmit_skb(struct sock *sk, struct sk_buff *skb, int clone_it,
        }
 
        if (unlikely(tcb->flags & TCPCB_FLAG_SYN)) {
-               tcp_syn_build_options((__u32 *)(th + 1),
+               tcp_syn_build_options((__be32 *)(th + 1),
                                      tcp_advertise_mss(sk),
                                      (sysctl_flags & SYSCTL_FLAG_TSTAMPS),
                                      (sysctl_flags & SYSCTL_FLAG_SACK),
@@ -454,7 +462,7 @@ static int tcp_transmit_skb(struct sock *sk, struct sk_buff *skb, int clone_it,
                                      tcb->when,
                                      tp->rx_opt.ts_recent);
        } else {
-               tcp_build_and_update_options((__u32 *)(th + 1),
+               tcp_build_and_update_options((__be32 *)(th + 1),
                                             tp, tcb->when);
                TCP_ECN_send(sk, tp, skb, tcp_header_size);
        }
@@ -470,19 +478,13 @@ static int tcp_transmit_skb(struct sock *sk, struct sk_buff *skb, int clone_it,
        if (after(tcb->end_seq, tp->snd_nxt) || tcb->seq == tcb->end_seq)
                TCP_INC_STATS(TCP_MIB_OUTSEGS);
 
-       err = icsk->icsk_af_ops->queue_xmit(skb, 0);
+       err = icsk->icsk_af_ops->queue_xmit(skb, sk, 0);
        if (likely(err <= 0))
                return err;
 
        tcp_enter_cwr(sk);
 
-       /* 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.
-        */
-       return err == NET_XMIT_CN ? 0 : err;
+       return net_xmit_eval(err);
 
 #undef SYSCTL_FLAG_TSTAMPS
 #undef SYSCTL_FLAG_WSCALE
@@ -1087,10 +1089,14 @@ static int tcp_tso_should_defer(struct sock *sk, struct tcp_sock *tp, struct sk_
        u32 send_win, cong_win, limit, in_flight;
 
        if (TCP_SKB_CB(skb)->flags & TCPCB_FLAG_FIN)
-               return 0;
+               goto send_now;
 
        if (icsk->icsk_ca_state != TCP_CA_Open)
-               return 0;
+               goto send_now;
+
+       /* Defer for less than two clock ticks. */
+       if (!tp->tso_deferred && ((jiffies<<1)>>1) - (tp->tso_deferred>>1) > 1)
+               goto send_now;
 
        in_flight = tcp_packets_in_flight(tp);
 
@@ -1106,7 +1112,7 @@ static int tcp_tso_should_defer(struct sock *sk, struct tcp_sock *tp, struct sk_
 
        /* If a full-sized TSO skb can be sent, do it. */
        if (limit >= 65536)
-               return 0;
+               goto send_now;
 
        if (sysctl_tcp_tso_win_divisor) {
                u32 chunk = min(tp->snd_wnd, tp->snd_cwnd * tp->mss_cache);
@@ -1116,7 +1122,7 @@ static int tcp_tso_should_defer(struct sock *sk, struct tcp_sock *tp, struct sk_
                 */
                chunk /= sysctl_tcp_tso_win_divisor;
                if (limit >= chunk)
-                       return 0;
+                       goto send_now;
        } else {
                /* Different approach, try not to defer past a single
                 * ACK.  Receiver should ACK every other full sized
@@ -1124,11 +1130,17 @@ static int tcp_tso_should_defer(struct sock *sk, struct tcp_sock *tp, struct sk_
                 * then send now.
                 */
                if (limit > tcp_max_burst(tp) * tp->mss_cache)
-                       return 0;
+                       goto send_now;
        }
 
        /* Ok, it looks like it is advisable to defer.  */
+       tp->tso_deferred = 1 | (jiffies<<1);
+
        return 1;
+
+send_now:
+       tp->tso_deferred = 0;
+       return 0;
 }
 
 /* Create a new MTU probe if we are ready.
@@ -2070,7 +2082,7 @@ struct sk_buff * tcp_make_synack(struct sock *sk, struct dst_entry *dst,
        th->window = htons(req->rcv_wnd);
 
        TCP_SKB_CB(skb)->when = tcp_time_stamp;
-       tcp_syn_build_options((__u32 *)(th + 1), dst_metric(dst, RTAX_ADVMSS), ireq->tstamp_ok,
+       tcp_syn_build_options((__be32 *)(th + 1), dst_metric(dst, RTAX_ADVMSS), ireq->tstamp_ok,
                              ireq->sack_ok, ireq->wscale_ok, ireq->rcv_wscale,
                              TCP_SKB_CB(skb)->when,
                              req->ts_recent);