[src/trunk]: src/sys/netinet Fix fast path for uni directional transfers

[kardel <kardel%NetBSD.org@localhost>]

details:   https://anonhg.NetBSD.org/src/rev/fe6f4a8e540e
branches:  trunk
changeset: 938116:fe6f4a8e540e
user:      kardel <kardel%NetBSD.org@localhost>
date:      Wed Sep 02 15:08:46 2020 +0000

description:
Fix fast path for uni directional transfers
pure ACK case:

drag snd_wl2 along so only newer
ACKs can update the window size.
also avoids the state where snd_wl2
is eventually larger than th_ack and thus
blocking the window update mechanism and
the connection gets stuck for a loooong
time in the zero sized send window state.

see PR/kern 55567

ok thorpej@, also found in FreeBSD

diffstat:

 sys/netinet/tcp_input.c |  17 +++++++++++++++--
 1 files changed, 15 insertions(+), 2 deletions(-)

diffs (38 lines):

diff -r d5693147782b -r fe6f4a8e540e sys/netinet/tcp_input.c
--- a/sys/netinet/tcp_input.c   Wed Sep 02 14:13:20 2020 +0000
+++ b/sys/netinet/tcp_input.c   Wed Sep 02 15:08:46 2020 +0000
@@ -1,4 +1,4 @@
-/*     $NetBSD: tcp_input.c,v 1.418 2020/07/06 18:49:12 christos Exp $ */
+/*     $NetBSD: tcp_input.c,v 1.419 2020/09/02 15:08:46 kardel Exp $   */
 
 /*
  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
@@ -148,7 +148,7 @@
  */
 
 #include <sys/cdefs.h>
-__KERNEL_RCSID(0, "$NetBSD: tcp_input.c,v 1.418 2020/07/06 18:49:12 christos Exp $");
+__KERNEL_RCSID(0, "$NetBSD: tcp_input.c,v 1.419 2020/09/02 15:08:46 kardel Exp $");
 
 #ifdef _KERNEL_OPT
 #include "opt_inet.h"
@@ -1897,6 +1897,19 @@
                                tp->snd_fack = tp->snd_una;
                                if (SEQ_LT(tp->snd_high, tp->snd_una))
                                        tp->snd_high = tp->snd_una;
+                               /*
+                                * drag snd_wl2 along so only newer
+                                * ACKs can update the window size.
+                                * also avoids the state where snd_wl2
+                                * is eventually larger than th_ack and thus
+                                * blocking the window update mechanism and
+                                * the connection gets stuck for a loooong
+                                * time in the zero sized send window state.
+                                *
+                                * see PR/kern 55567
+                                */
+                               tp->snd_wl2 = tp->snd_una;
+
                                m_freem(m);
 
                                /*