tcp: don't clear lost_skb_hint when not necessary
Most importantly avoid doing it with cumulative ACK. However,
since we have lost_cnt_hint in the picture as well needing
adjustments, it's not as trivial as dealing with
retransmit_skb_hint (and cannot be done in the all place we
could trivially leave retransmit_skb_hint untouched).
With the previous patch, this should mostly remove O(n^2)
behavior while cumulative ACKs start flowing once rexmit
after a lossy round-trip made it through.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 44a4fff..85627f8 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -2844,6 +2844,7 @@
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();
@@ -2925,9 +2926,11 @@
tcp_unlink_write_queue(skb, sk);
sk_wmem_free_skb(sk, skb);
- tcp_clear_retrans_hints_partial(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 (skb && (TCP_SKB_CB(skb)->sacked & TCPCB_SACKED_ACKED))
@@ -2946,6 +2949,15 @@
/* 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);