tcp: should use number of sack blocks instead of -1
authorIlpo Järvinen <ilpo.jarvinen@helsinki.fi>
Tue, 21 Oct 2008 23:28:36 +0000 (16:28 -0700)
committerDavid S. Miller <davem@davemloft.net>
Tue, 21 Oct 2008 23:28:36 +0000 (16:28 -0700)
commit75e3d8db531b462b875c1adb13eeb6b0be7374c0
treede2d38e5a4c056968eed4b35be4a76948743b583
parent2515ddc6db8eb49a79f0fe5e67ff09ac7c81eab4
tcp: should use number of sack blocks instead of -1

While looking for the recent "sack issue" I also read all eff_sacks
usage that was played around by some relevant commit. I found
out that there's another thing that is asking for a fix (unrelated
to the "sack issue" though).

This feature has probably very little significance in practice.
Opposite direction timeout with bidirectional tcp comes to me as
the most likely scenario though there might be other cases as
well related to non-data segments we send (e.g., response to the
opposite direction segment). Also some ACK losses or option space
wasted for other purposes is necessary to prevent the earlier
SACK feedback getting to the sender.

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
net/ipv4/tcp_output.c