netdev - About TCP Westwood implementation...

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [day] [month] [year] [list]

Date:	Thu, 24 Dec 2009 11:45:20 +0900
From:	"Hwang Jae-Hyun" <jhhwang@...korea.ac.kr>
To:	<netdev@...r.kernel.org>
Subject: About TCP Westwood implementation...

Hi,

(I re-send this posting that was sent to linux-net.)
I think that the current Westwood algorithm implemented in the Linux kernel
(since 2.6.13) is not same with one described in its original paper.
In the Linux implementation, Westwood uses 'tcp_reno_ssthresh' function as a
'ssthresh' interface as below codes.

---------------------------------------------------
<tcp_westwood.c>
...
static struct tcp_congestion_ops tcp_westwood = {
	.init		= tcp_westwood_init,
	.ssthresh	= tcp_reno_ssthresh,
	.cong_avoid	= tcp_reno_cong_avoid,
	.min_cwnd	= tcp_westwood_bw_rttmin,
	.cwnd_event	= tcp_westwood_event,
	.get_info	= tcp_westwood_info,
	.pkts_acked	= tcp_westwood_pkts_acked,

	.owner		= THIS_MODULE,
	.name		= "westwood"
};
----------------------------------------------------

But, Westwood basically sets its ssthresh value based on bandwidth
estimation, not by cwnd/2 as Reno does.
In this case, Linux-Westwood would call 'tcp_reno_ssthresh' to set its
ssthresh value when retransmission timeout expires.
I don't know why Westwood uses Reno's ssthresh function.

In my opinion,
1) '.ssthresh' interface should be changed from 'tcp_reno_ssthresh' to
'tcp_westwood_bw_rttmin'.
Then, '.min_cwnd' interface can be removed.
2) In 'tcp_westwood_event()' function,
 2.1) The case of 'CA_EVENT_COMPLETE_CWR' can be removed. (corresponding
codes are already executed)
 2.2) In the case of 'CA_EVENT_FRTO', 'tp->snd_ssthresh =
tcp_westwood_bw_rttmin(sk);' can be removed because '.ssthresh' interface is
called before tcp_ca_event(sk, CA_EVENT_FRTO) is called.

Thanks,
Jaehyun Hwang

--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html