tcp: fix bogus RTT for CC when retransmissions are acked
authorKenneth Klette Jonassen <kennetkl@ifi.uio.no>
Sat, 11 Apr 2015 00:17:49 +0000 (02:17 +0200)
committerDavid S. Miller <davem@davemloft.net>
Mon, 13 Apr 2015 17:54:25 +0000 (13:54 -0400)
commit3d0d26c7976bf190c3f1d2efbc31462db8246bc0
tree09b3b238bb8292ad5a2eacc770ea63b5ef457d74
parent4577139b2dabf58973d59d157aae4ddd3bde863a
tcp: fix bogus RTT for CC when retransmissions are acked

Since retransmitted segments are not used for RTT estimation, previously
SACKed segments present in the rtx queue are used. This estimation can be
several times larger than the actual RTT. When a cumulative ack covers both
previously SACKed and retransmitted segments, CC may thus get a bogus RTT.

Such segments previously had an RTT estimation in tcp_sacktag_one(), so it
seems reasonable to not reuse them in tcp_clean_rtx_queue() at all.

Afaik, this has had no effect on SRTT/RTO because of Karn's check.

Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no>
Acked-by: Neal Cardwell <ncardwell@google.com>
Tested-by: Neal Cardwell <ncardwell@google.com>
Acked-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
net/ipv4/tcp_input.c