}
}
+/* 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);
}
}
}
+/* 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).
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))
~(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)) {
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;
}
}
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;
return 0;
skb = tcp_write_queue_head(sk);
+ if (tcp_skb_is_last(sk, skb))
+ return 1;
skb = tcp_write_queue_next(sk, skb); /* Skips head */
tcp_for_write_queue_from(skb, sk) {
if (skb == tcp_send_head(sk))
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);
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)
/* 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))
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);
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"
*/
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);
}
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;
}
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)
* is before the ack sequence we can discard it as it's confirmed to have
* arrived at the other end.
*/
-static int tcp_clean_rtx_queue(struct sock *sk, int prior_fackets)
+static int tcp_clean_rtx_queue(struct sock *sk, int prior_fackets,
+ u32 prior_snd_una)
{
struct tcp_sock *tp = tcp_sk(sk);
const struct inet_connection_sock *icsk = inet_csk(sk);
int flag = 0;
u32 pkts_acked = 0;
u32 reord = tp->packets_out;
+ u32 prior_sacked = tp->sacked_out;
s32 seq_rtt = -1;
s32 ca_seq_rtt = -1;
ktime_t last_ackt = net_invalid_timestamp();
if (sacked & TCPCB_LOST)
tp->lost_out -= acked_pcount;
- if (unlikely(tp->urg_mode && !before(end_seq, tp->snd_up)))
- tp->urg_mode = 0;
-
tp->packets_out -= acked_pcount;
pkts_acked += acked_pcount;
tcp_unlink_write_queue(skb, sk);
sk_wmem_free_skb(sk, skb);
- tcp_clear_all_retrans_hints(tp);
+ tp->scoreboard_skb_hint = NULL;
+ if (skb == tp->retransmit_skb_hint)
+ tp->retransmit_skb_hint = NULL;
+ if (skb == tp->lost_skb_hint)
+ tp->lost_skb_hint = NULL;
}
+ if (likely(between(tp->snd_up, prior_snd_una, tp->snd_una)))
+ tp->snd_up = tp->snd_una;
+
if (skb && (TCP_SKB_CB(skb)->sacked & TCPCB_SACKED_ACKED))
flag |= FLAG_SACK_RENEGING;
/* Non-retransmitted hole got filled? That's reordering */
if (reord < prior_fackets)
tcp_update_reordering(sk, tp->fackets_out - reord, 0);
+
+ /* No need to care for underflows here because
+ * the lost_skb_hint gets NULLed if we're past it
+ * (or something non-trivial happened)
+ */
+ if (tcp_is_fack(tp))
+ tp->lost_cnt_hint -= pkts_acked;
+ else
+ tp->lost_cnt_hint -= prior_sacked - tp->sacked_out;
}
tp->fackets_out -= min(pkts_acked, tp->fackets_out);
goto no_queue;
/* See if we can take anything off of the retransmit queue. */
- flag |= tcp_clean_rtx_queue(sk, prior_fackets);
+ flag |= tcp_clean_rtx_queue(sk, prior_fackets, prior_snd_una);
if (tp->frto_counter)
frto_cwnd = tcp_process_frto(sk, flag);
skb1 = skb1->prev;
}
}
- __skb_insert(skb, skb1, skb1->next, &tp->out_of_order_queue);
+ __skb_queue_after(&tp->out_of_order_queue, skb1, skb);
/* And clean segments covered by new one as whole. */
while ((skb1 = skb->next) !=
memcpy(nskb->head, skb->head, header);
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, list);
+ __skb_queue_before(list, skb, nskb);
skb_set_owner_r(nskb, sk);
/* Copy data, releasing collapsed skbs. */