In message <20041028195016.CCDAC2329@fnord.ir.bbn.com>Greg Troxel writes
>I went to enable ECN for TCP on a 2.0_RC4 box, and didn't find the
>sysctl that I remember using on FreeBSD in 1999.  From a quick grep of
>the code, I conclude that there is no support for ECN in tcp, but that
>the ALTQ code will do ECN marking.

Thats my understand also, from when I looked into ECN this July.
KAME ALTQ (courtesy of Kenjiro Cho <kjc@csl.sony.co.jp>) has patches
against NetBSD-1.5.2, which added TCP end-host ECN support. I have a
copy of that patch if you have difficutly finding a copy.

>Can anyone point me to an TCP/ECN implementation for 2.0, or point me
>in the right direction if it is there?


>Also, I wonder if there is some objection to adding ECN support, or if
>it's just that no one has gotten around to it.

Adding ECN gets positive support from me. However, the Usual Suspects
here are increasingly interested in SACK. It'd be nice if any work you
(or others) did on ECN dovetailed with those working on SACK, and
FACK, or rate-halving, or whatever is best-practice these days.