tcp: don't clear retransmit_skb_hint when not necessary

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

diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 9da9f19ece8adfc303918709d0da4e59c1b9b910..44a4fffc2cc392c98b9ca5f9e676209865e89f9b 100644 (file)
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -811,12 +811,25 @@ void tcp_update_metrics(struct sock *sk)
        }
 }
 
+/* Numbers are taken from RFC3390.
+ *
+ * John Heffner states:
+ *
+ *     The RFC specifies a window of no more than 4380 bytes
+ *     unless 2*MSS > 4380.  Reading the pseudocode in the RFC
+ *     is a bit misleading because they use a clamp at 4380 bytes
+ *     rather than use a multiplier in the relevant range.
+ */
 __u32 tcp_init_cwnd(struct tcp_sock *tp, struct dst_entry *dst)
 {
        __u32 cwnd = (dst ? dst_metric(dst, RTAX_INITCWND) : 0);
 
-       if (!cwnd)
-               cwnd = rfc3390_bytes_to_packets(tp->mss_cache);
+       if (!cwnd) {
+               if (tp->mss_cache > 1460)
+                       cwnd = 2;
+               else
+                       cwnd = (tp->mss_cache > 1095) ? 3 : 4;
+       }
        return min_t(__u32, cwnd, tp->snd_cwnd_clamp);
 }
 
@@ -966,6 +979,39 @@ static void tcp_update_reordering(struct sock *sk, const int metric,
        }
 }
 
+/* This must be called before lost_out is incremented */
+static void tcp_verify_retransmit_hint(struct tcp_sock *tp, struct sk_buff *skb)
+{
+       if ((tp->retransmit_skb_hint == NULL) ||
+           before(TCP_SKB_CB(skb)->seq,
+                  TCP_SKB_CB(tp->retransmit_skb_hint)->seq))
+               tp->retransmit_skb_hint = skb;
+
+       if (!tp->lost_out ||
+           after(TCP_SKB_CB(skb)->end_seq, tp->retransmit_high))
+               tp->retransmit_high = TCP_SKB_CB(skb)->end_seq;
+}
+
+static void tcp_skb_mark_lost(struct tcp_sock *tp, struct sk_buff *skb)
+{
+       if (!(TCP_SKB_CB(skb)->sacked & (TCPCB_LOST|TCPCB_SACKED_ACKED))) {
+               tcp_verify_retransmit_hint(tp, skb);
+
+               tp->lost_out += tcp_skb_pcount(skb);
+               TCP_SKB_CB(skb)->sacked |= TCPCB_LOST;
+       }
+}
+
+void tcp_skb_mark_lost_uncond_verify(struct tcp_sock *tp, struct sk_buff *skb)
+{
+       tcp_verify_retransmit_hint(tp, skb);
+
+       if (!(TCP_SKB_CB(skb)->sacked & (TCPCB_LOST|TCPCB_SACKED_ACKED))) {
+               tp->lost_out += tcp_skb_pcount(skb);
+               TCP_SKB_CB(skb)->sacked |= TCPCB_LOST;
+       }
+}
+
 /* This procedure tags the retransmission queue when SACKs arrive.
  *
  * We have three tag bits: SACKED(S), RETRANS(R) and LOST(L).
@@ -1142,13 +1188,7 @@ static void tcp_mark_lost_retrans(struct sock *sk)
                        TCP_SKB_CB(skb)->sacked &= ~TCPCB_SACKED_RETRANS;
                        tp->retrans_out -= tcp_skb_pcount(skb);
 
-                       /* clear lost hint */
-                       tp->retransmit_skb_hint = NULL;
-
-                       if (!(TCP_SKB_CB(skb)->sacked & (TCPCB_LOST|TCPCB_SACKED_ACKED))) {
-                               tp->lost_out += tcp_skb_pcount(skb);
-                               TCP_SKB_CB(skb)->sacked |= TCPCB_LOST;
-                       }
+                       tcp_skb_mark_lost_uncond_verify(tp, skb);
                        NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPLOSTRETRANSMIT);
                } else {
                        if (before(ack_seq, new_low_seq))
@@ -1258,9 +1298,6 @@ static int tcp_sacktag_one(struct sk_buff *skb, struct sock *sk,
                                        ~(TCPCB_LOST|TCPCB_SACKED_RETRANS);
                                tp->lost_out -= tcp_skb_pcount(skb);
                                tp->retrans_out -= tcp_skb_pcount(skb);
-
-                               /* clear lost hint */
-                               tp->retransmit_skb_hint = NULL;
                        }
                } else {
                        if (!(sacked & TCPCB_RETRANS)) {
@@ -1279,9 +1316,6 @@ static int tcp_sacktag_one(struct sk_buff *skb, struct sock *sk,
                        if (sacked & TCPCB_LOST) {
                                TCP_SKB_CB(skb)->sacked &= ~TCPCB_LOST;
                                tp->lost_out -= tcp_skb_pcount(skb);
-
-                               /* clear lost hint */
-                               tp->retransmit_skb_hint = NULL;
                        }
                }
 
@@ -1311,7 +1345,6 @@ static int tcp_sacktag_one(struct sk_buff *skb, struct sock *sk,
        if (dup_sack && (TCP_SKB_CB(skb)->sacked & TCPCB_SACKED_RETRANS)) {
                TCP_SKB_CB(skb)->sacked &= ~TCPCB_SACKED_RETRANS;
                tp->retrans_out -= tcp_skb_pcount(skb);
-               tp->retransmit_skb_hint = NULL;
        }
 
        return flag;
@@ -1854,6 +1887,7 @@ static void tcp_enter_frto_loss(struct sock *sk, int allowed_segments, int flag)
                if (!(TCP_SKB_CB(skb)->sacked & TCPCB_SACKED_ACKED)) {
                        TCP_SKB_CB(skb)->sacked |= TCPCB_LOST;
                        tp->lost_out += tcp_skb_pcount(skb);
+                       tp->retransmit_high = TCP_SKB_CB(skb)->end_seq;
                }
        }
        tcp_verify_left_out(tp);
@@ -1870,7 +1904,7 @@ static void tcp_enter_frto_loss(struct sock *sk, int allowed_segments, int flag)
        tp->high_seq = tp->snd_nxt;
        TCP_ECN_queue_cwr(tp);
 
-       tcp_clear_retrans_hints_partial(tp);
+       tcp_clear_all_retrans_hints(tp);
 }
 
 static void tcp_clear_retrans_partial(struct tcp_sock *tp)
@@ -1921,12 +1955,11 @@ void tcp_enter_loss(struct sock *sk, int how)
                /* Push undo marker, if it was plain RTO and nothing
                 * was retransmitted. */
                tp->undo_marker = tp->snd_una;
-               tcp_clear_retrans_hints_partial(tp);
        } else {
                tp->sacked_out = 0;
                tp->fackets_out = 0;
-               tcp_clear_all_retrans_hints(tp);
        }
+       tcp_clear_all_retrans_hints(tp);
 
        tcp_for_write_queue(skb, sk) {
                if (skb == tcp_send_head(sk))
@@ -1939,6 +1972,7 @@ void tcp_enter_loss(struct sock *sk, int how)
                        TCP_SKB_CB(skb)->sacked &= ~TCPCB_SACKED_ACKED;
                        TCP_SKB_CB(skb)->sacked |= TCPCB_LOST;
                        tp->lost_out += tcp_skb_pcount(skb);
+                       tp->retransmit_high = TCP_SKB_CB(skb)->end_seq;
                }
        }
        tcp_verify_left_out(tp);
@@ -2144,19 +2178,6 @@ static int tcp_time_to_recover(struct sock *sk)
        return 0;
 }
 
-/* RFC: This is from the original, I doubt that this is necessary at all:
- * clear xmit_retrans hint if seq of this skb is beyond hint. How could we
- * retransmitted past LOST markings in the first place? I'm not fully sure
- * about undo and end of connection cases, which can cause R without L?
- */
-static void tcp_verify_retransmit_hint(struct tcp_sock *tp, struct sk_buff *skb)
-{
-       if ((tp->retransmit_skb_hint != NULL) &&
-           before(TCP_SKB_CB(skb)->seq,
-                  TCP_SKB_CB(tp->retransmit_skb_hint)->seq))
-               tp->retransmit_skb_hint = NULL;
-}
-
 /* Mark head of queue up as lost. With RFC3517 SACK, the packets is
  * is against sacked "cnt", otherwise it's against facked "cnt"
  */
@@ -2204,11 +2225,7 @@ static void tcp_mark_head_lost(struct sock *sk, int packets)
                        cnt = packets;
                }
 
-               if (!(TCP_SKB_CB(skb)->sacked & (TCPCB_SACKED_ACKED|TCPCB_LOST))) {
-                       TCP_SKB_CB(skb)->sacked |= TCPCB_LOST;
-                       tp->lost_out += tcp_skb_pcount(skb);
-                       tcp_verify_retransmit_hint(tp, skb);
-               }
+               tcp_skb_mark_lost(tp, skb);
        }
        tcp_verify_left_out(tp);
 }
@@ -2250,11 +2267,7 @@ static void tcp_update_scoreboard(struct sock *sk, int fast_rexmit)
                        if (!tcp_skb_timedout(sk, skb))
                                break;
 
-                       if (!(TCP_SKB_CB(skb)->sacked & (TCPCB_SACKED_ACKED|TCPCB_LOST))) {
-                               TCP_SKB_CB(skb)->sacked |= TCPCB_LOST;
-                               tp->lost_out += tcp_skb_pcount(skb);
-                               tcp_verify_retransmit_hint(tp, skb);
-                       }
+                       tcp_skb_mark_lost(tp, skb);
                }
 
                tp->scoreboard_skb_hint = skb;
@@ -2365,10 +2378,6 @@ static void tcp_undo_cwr(struct sock *sk, const int undo)
        }
        tcp_moderate_cwnd(tp);
        tp->snd_cwnd_stamp = tcp_time_stamp;
-
-       /* There is something screwy going on with the retrans hints after
-          an undo */
-       tcp_clear_all_retrans_hints(tp);
 }
 
 static inline int tcp_may_undo(struct tcp_sock *tp)
@@ -2916,7 +2925,9 @@ static int tcp_clean_rtx_queue(struct sock *sk, int prior_fackets)
 
                tcp_unlink_write_queue(skb, sk);
                sk_wmem_free_skb(sk, skb);
-               tcp_clear_all_retrans_hints(tp);
+               tcp_clear_retrans_hints_partial(tp);
+               if (skb == tp->retransmit_skb_hint)
+                       tp->retransmit_skb_hint = NULL;
        }
 
        if (skb && (TCP_SKB_CB(skb)->sacked & TCPCB_SACKED_ACKED))