From: "Angelo Dell'Aera" <buffer@antifork.org>
To: Stephen Hemminger <shemminger@osdl.org>
Cc: davem@redhat.com, netdev@oss.sgi.com
Subject: Re: [PATCH] TCP westwood bw_sample set never used
Date: Sat, 13 Mar 2004 17:27:12 +0100 [thread overview]
Message-ID: <20040313172712.0d10080a.buffer@antifork.org> (raw)
In-Reply-To: <20040312091505.593443d1@dell_ss3.pdx.osdl.net>
On Fri, 12 Mar 2004 09:15:05 -0800
Stephen Hemminger <shemminger@osdl.org> wrote:
>Found this while looking in detail at TCP. The westwood.bw_sample element
>it tcp_opt is set but not used by the existing code in 2.6.4
Stephen,
I used that field in the first version of the Westwood code just for having
the possibility of doing a bandwidth estimation tracking analysis.
After dropping CONFIG_TCP_WESTWOOD_STATS it's simply useless and so I agree
with your patch. But it's necessary to just point out an important thing.
When you call westwood_filter() you're actually evaluating the bandwidth
estimate. This is done by two constant coefficients low-pass filters one
after the other.
This part of the code was taken clearly separated from the rest of the code
since you could even find a better way to do it (f.e. TCP Westwood which came
before Westwood+ did it in improper manner). Maybe this was the reason
bw_sample was left in the struct westwood. But being it absolutely useless
at this time it could be removed even from there. I simply dropped another
line wrt to your patch.
Regards.
--
Angelo Dell'Aera 'buffer'
Antifork Research, Inc. http://buffer.antifork.org
diff -Naur linux-2.6.4-old/include/linux/tcp.h linux-2.6.4/include/linux/tcp.h
--- linux-2.6.4-old/include/linux/tcp.h 2004-02-18 16:34:15.000000000 +0100
+++ linux-2.6.4/include/linux/tcp.h 2004-03-13 16:56:39.000000000 +0100
@@ -377,7 +377,6 @@
/* TCP Westwood structure */
struct {
- __u32 bw_sample; /* bandwidth sample */
__u32 bw_ns_est; /* first bandwidth estimation..not too smoothed 8) */
__u32 bw_est; /* bandwidth estimate */
__u32 rtt_win_sx; /* here starts a new evaluation... */
diff -Naur linux-2.6.4-old/net/ipv4/tcp_input.c linux-2.6.4/net/ipv4/tcp_input.c
--- linux-2.6.4-old/net/ipv4/tcp_input.c 2004-03-10 17:31:40.000000000 +0100
+++ linux-2.6.4/net/ipv4/tcp_input.c 2004-03-13 16:50:35.000000000 +0100
@@ -2065,7 +2065,6 @@
{
struct tcp_opt *tp = tcp_sk(sk);
- tp->westwood.bw_sample = 0;
tp->westwood.bw_ns_est = 0;
tp->westwood.bw_est = 0;
tp->westwood.accounted = 0;
@@ -2088,14 +2087,13 @@
static void westwood_filter(struct sock *sk, __u32 delta)
{
struct tcp_opt *tp = tcp_sk(sk);
- __u32 sample = tp->westwood.bk / delta;
tp->westwood.bw_ns_est =
- westwood_do_filter(tp->westwood.bw_ns_est, sample);
+ westwood_do_filter(tp->westwood.bw_ns_est,
+ tp->westwood.bk / delta);
tp->westwood.bw_est =
westwood_do_filter(tp->westwood.bw_est,
tp->westwood.bw_ns_est);
- tp->westwood.bw_sample = sample;
}
/* @westwood_update_rttmin
prev parent reply other threads:[~2004-03-13 16:27 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2004-03-12 17:15 [PATCH] TCP westwood bw_sample set never used Stephen Hemminger
2004-03-13 16:27 ` Angelo Dell'Aera [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20040313172712.0d10080a.buffer@antifork.org \
--to=buffer@antifork.org \
--cc=davem@redhat.com \
--cc=netdev@oss.sgi.com \
--cc=shemminger@osdl.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).