dcavalca / rpms / grub2

Forked from rpms/grub2 3 years ago
Clone

Blame SOURCES/0181-tcp-ack-when-we-get-an-OOO-lost-packet.patch

28f7f8
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
ecb9bb
From: Josef Bacik <jbacik@fb.com>
ecb9bb
Date: Wed, 12 Aug 2015 08:16:22 -0700
28f7f8
Subject: [PATCH] tcp: ack when we get an OOO/lost packet
ecb9bb
ecb9bb
While adding tcp window scaling support I was finding that I'd get some packet
ecb9bb
loss or reordering when transferring from large distances and grub would just
ecb9bb
timeout.  This is because we weren't ack'ing when we got our OOO packet, so the
ecb9bb
sender didn't know it needed to retransmit anything, so eventually it would fill
ecb9bb
the window and stop transmitting, and we'd time out.  Fix this by ACK'ing when
ecb9bb
we don't find our next sequence numbered packet.  With this fix I no longer time
ecb9bb
out.  Thanks,
ecb9bb
ecb9bb
Signed-off-by: Josef Bacik <jbacik@fb.com>
ecb9bb
---
ecb9bb
 grub-core/net/tcp.c | 5 ++++-
ecb9bb
 1 file changed, 4 insertions(+), 1 deletion(-)
ecb9bb
ecb9bb
diff --git a/grub-core/net/tcp.c b/grub-core/net/tcp.c
28f7f8
index 2077f5519b2..fa29a2afc6c 100644
ecb9bb
--- a/grub-core/net/tcp.c
ecb9bb
+++ b/grub-core/net/tcp.c
ecb9bb
@@ -882,7 +882,10 @@ grub_net_recv_tcp_packet (struct grub_net_buff *nb,
ecb9bb
 	  grub_priority_queue_pop (sock->pq);
ecb9bb
 	}
ecb9bb
       if (grub_be_to_cpu32 (tcph->seqnr) != sock->their_cur_seq)
ecb9bb
-	return GRUB_ERR_NONE;
ecb9bb
+	{
ecb9bb
+	  ack (sock);
ecb9bb
+	  return GRUB_ERR_NONE;
ecb9bb
+	}
ecb9bb
       while (1)
ecb9bb
 	{
ecb9bb
 	  nb_top_p = grub_priority_queue_top (sock->pq);