tcp: fix FRTO undo on cumulative ACK of SACKed range
authorNeal Cardwell <ncardwell@google.com>
Thu, 2 Apr 2015 00:26:46 +0000 (20:26 -0400)
committerDavid S. Miller <davem@davemloft.net>
Thu, 2 Apr 2015 20:35:58 +0000 (16:35 -0400)
commit666b805150efd62f05810ff0db08f44a2370c937
tree265d4aba6edfc0b63a07ba13f2c05eafe4e2ad4f
parent0c36820e2ab7d943ab1188230fdf2149826d33c0
tcp: fix FRTO undo on cumulative ACK of SACKed range

On processing cumulative ACKs, the FRTO code was not checking the
SACKed bit, meaning that there could be a spurious FRTO undo on a
cumulative ACK of a previously SACKed skb.

The FRTO code should only consider a cumulative ACK to indicate that
an original/unretransmitted skb is newly ACKed if the skb was not yet
SACKed.

The effect of the spurious FRTO undo would typically be to make the
connection think that all previously-sent packets were in flight when
they really weren't, leading to a stall and an RTO.

Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Fixes: e33099f96d99c ("tcp: implement RFC5682 F-RTO")
Signed-off-by: David S. Miller <davem@davemloft.net>
net/ipv4/tcp_input.c