[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: 2.4.x Kernel, ECN And Problem Websites

On Wed, Apr 25, 2001 at 11:48:22AM -0700, Aaron Lehmann wrote:
> Quoting Daniel Stone <daniel@kabuki.openfridge.net>:
> > Why enable ECN at all, if all it effectively does is break stuff? AFAIK,
> > there's no systems out "in the wild" that actually use ECN to make a
> > difference. All that's happening is that peoples' systems are being
> > broken.
> > Which is sub-optimal.
> I would have expected something more intelligent from a "Linux
> kernel developer". ECN is COMPLETELY backward-compatible, and the bits
> it uses are reserved for it. The RFC's instruct these reserved bits to
> be ignored if the device does not support ECN. When firewalls silently
> drop packets just because they have the ECN bits set, those firewalls

Yes, I know this. The bits are officially "reserved" in the RFC. Some people
took this to mean, "must be zero".

> are broken, not Linux or ECN. In short: it's not our problem. I wish

It's not our brokenness, but when you have chunks of the web blacked out, it
really becomes your problem.

> people would stop being so sensationalist about ECN. linux-kernel has
> been tracking delinquent sites for a few months now, and DaveM resolved
> to turn ECN on on vger, which would effictively cut off hotmail users
> from it since hotmail is (was?) one such broken site. All of a sudden

Still is, and AFAIK ECN isn't on vger.

> Slashdot posts a FUD-filled article claiming ECN is enabled by default,
> isn't backward-compatible, and breaks things. I bet that's where this
> thread came from.

I don't read that crap. I'm speaking from the numerous threads, and
experience (I tried it for a while).

Daniel Stone

Reply to: