From mboxrd@z Thu Jan 1 00:00:00 1970 From: Matt Mackall Subject: Re: [PATCH] Stop scaring users with "treason uncloaked!" Date: Thu, 18 Dec 2008 17:50:34 -0600 Message-ID: <1229644234.3726.577.camel@calx> References: <1229638581.3726.546.camel@calx> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: David Miller , Linux Kernel Mailing List , netdev To: Ilpo =?ISO-8859-1?Q?J=E4rvinen?= Return-path: Received: from waste.org ([66.93.16.53]:41015 "EHLO waste.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750894AbYLRXu4 (ORCPT ); Thu, 18 Dec 2008 18:50:56 -0500 In-Reply-To: Sender: netdev-owner@vger.kernel.org List-ID: On Fri, 2008-12-19 at 00:33 +0200, Ilpo J=C3=A4rvinen wrote: > On Thu, 18 Dec 2008, Matt Mackall wrote: >=20 > > These debug messages are uninformative and extremely alarming. Rath= er > > than reporting a peer as a treacherous attacker, report it more > > correctly as simply broken.=20 >=20 > Last time (well, ignoring some report on ancient vendor kernel) I've = seen=20 > this to trigger it wasn't the fault of the peer but ours (we sent pas= t the=20 > advertized window and window ended up being shrunk to zero). I agree = that=20 > the message as-is is not very good but your change is not very good=20 > either. Here this is the bulk of messages on my server (which is otherwise fairly quiet). It really doesn't matter whether the message is technically accurate - it's mostly there for the benefit of kernel hackers. The average end user is not going to have the slightest clue what is going on nor shoul= d they even care. An advanced network admin might be able to figure thing= s out by poking at packet traces, but again: not much reason to care. For most admins, the vast majority of these treasonous peers won't be under their administrative control.=20 If you want, we can instead say "Peer %d unexpectedly shrunk window..."= =2E So long as the scary bit is gone, I don't care. --=20 Mathematics is the supreme nostalgia of our time.