netdev
[Top] [All Lists]

Re: CONFIG_INET_ECN creates problems

To: rusty@xxxxxxxxxxxxxxxx
Subject: Re: CONFIG_INET_ECN creates problems
From: "David S. Miller" <davem@xxxxxxxxxx>
Date: Tue, 7 Nov 2000 18:08:19 -0800
Cc: hadi@xxxxxxxxxx, netdev@xxxxxxxxxxx
In-reply-to: <20001107020317.EA043820C@xxxxxxxxxxxxxxxxxxxxxxxx> (message from Rusty Russell on Tue, 07 Nov 2000 13:03:16 +1100)
References: <20001107020317.EA043820C@xxxxxxxxxxxxxxxxxxxxxxxx>
Sender: owner-netdev@xxxxxxxxxxx
   From: Rusty Russell <rusty@xxxxxxxxxxxxxxxx>
   Date:   Tue, 07 Nov 2000 13:03:16 +1100

   Penalizes first SYN failures (quite common from .au) and doesn't solve
   the Cisco-sends-RST problem.

Yes it does, and this is one of the biggest problems I have with
the floyd draft.  Any proposed RFC language which advises a TCP
implementation outright to ignore perfectly valid RST packets
is not to be taken seriously.

   We could generalize the Floyd solution
   to N transmits (I suggest N >= 2, rather than one, but it's just a
   inverse of the ECN sysctl), and use two bits in the route cache: one
   to indicate that we've spoken to the host with ECN flags set, and one
   to indicate that we've received a RST for an ECN packet.

Doesn't work through masquerading sites.  DST != same machine nor same
path.  Your scheme also ignores valid RST packets, unacceptable.

   Horrible?  Yes.  But it would allow us to ship ECN by default (good),
   and still communicate with the world (which is the highest priority).

How about we ship it on by default as is? :-)

You may laugh, but I consider it very seriously every day.  Because as
a solution it at least makes sense and will get people to fix their
kit.

Later,
David S. Miller
davem@xxxxxxxxxx

<Prev in Thread] Current Thread [Next in Thread>