tcp: call tcp_ack_tstamp() when not fully acked

When skb is coalesced tcp_ack_tstamp() still needs to be called when not
fully acked in tcp_clean_rtx_queue(), otherwise SCM_TSTAMP_ACK
timestamps may never be fired. Since the original patch series had
dependent commits, this patch fixes the issue instead of reverting by
restoring calls to tcp_ack_tstamp() when skb is not fully acked.

Fixes: fdb7eb21ddd3 ("tcp: stamp SCM_TSTAMP_ACK later in tcp_clean_rtx_queue()")
Signed-off-by: Yousuk Seung <ysseung@google.com>
Signed-off-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Soheil Hassas Yeganeh <soheil@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 8479b84..12c26c9 100644 (file)
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -3172,8 +3172,11 @@ static int tcp_clean_rtx_queue(struct sock *sk, u32 prior_fack,
        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;
+       if (skb) {
+               tcp_ack_tstamp(sk, skb, prior_snd_una);
+               if (TCP_SKB_CB(skb)->sacked & TCPCB_SACKED_ACKED)
+                       flag |= FLAG_SACK_RENEGING;
+       }
 
        if (likely(first_ackt) && !(flag & FLAG_RETRANS_DATA_ACKED)) {
                seq_rtt_us = tcp_stamp_us_delta(tp->tcp_mstamp, first_ackt);