Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1161158AbVIPKdI (ORCPT ); Fri, 16 Sep 2005 06:33:08 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1161159AbVIPKdI (ORCPT ); Fri, 16 Sep 2005 06:33:08 -0400 Received: from gockel.physik3.uni-rostock.de ([139.30.44.16]:54918 "EHLO gockel.physik3.uni-rostock.de") by vger.kernel.org with ESMTP id S1161158AbVIPKdH (ORCPT ); Fri, 16 Sep 2005 06:33:07 -0400 Date: Fri, 16 Sep 2005 12:33:03 +0200 (CEST) From: Tim Schmielau To: Jesper Juhl cc: "Randy.Dunlap" , linux-kernel@vger.kernel.org Subject: Re: early printk timings way off In-Reply-To: Message-ID: References: <200509152342.24922.jesper.juhl@gmail.com> <9a87484905091515072c7dd4a8@mail.gmail.com> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 1007 Lines: 26 On Fri, 16 Sep 2005, Tim Schmielau wrote: > On Fri, 16 Sep 2005, Jesper Juhl wrote: > > It also doesn't > > explain why two lines, the first with timing value 0.000, and the next > > with 27.121 don't seem to match reality - the *actual* delta between > > printing those two lines is far lower than 27 seconds. > > Yes, this seems to be different, possibly unrelated problem. > It's interesting that the value jumps _exactly_to_zero_, though. > Will need to dig into the code... Did that. The problem is that printk uses sched_clock() to determine the time, which just isn't supposed to be a reliable long-time clock. We need to base the output on a different clock. Btw, the rate-limiting logic in printk.c looks 'interesting'. Will look into that, too. Tim - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/