From: David Daney Subject: Re: [PATCH 1/9] ext4: Add -DDEBUG to Makefile Date: Mon, 19 Mar 2012 11:48:12 -0700 Message-ID: <4F677F6C.8030902@gmail.com> References: <7ee60f3b750f4500f9bdcb64f358acbf17987036.1331856300.git.joe@perches.com> <4F63646D.1040700@cavium.com> <20120319043937.GN31682@thunk.org> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit To: Ted Ts'o , David Daney , Joe Perches , Andreas Dilger , linux-ext4@vger.kernel.org, linux-kernel@vger.kernel.org Return-path: Received: from mail-gx0-f174.google.com ([209.85.161.174]:55431 "EHLO mail-gx0-f174.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1757896Ab2CSSsP (ORCPT ); Mon, 19 Mar 2012 14:48:15 -0400 In-Reply-To: <20120319043937.GN31682@thunk.org> Sender: linux-ext4-owner@vger.kernel.org List-ID: On 03/18/2012 09:39 PM, Ted Ts'o wrote: > On Fri, Mar 16, 2012 at 09:03:57AM -0700, David Daney wrote: >> >> In many other susbsystems/drivers, the definition of DEBUG is gated >> by a separate Kconfig symbol used to select debugging just for that >> susbsystem/driver (see CONFIG_MMC_DEBUG for example). > > In ext4 we have many different debugging patches, and in general you'd > never want to enable them all at once (you'd get way too much noise). > The debugging statements are there when a developer is debugging very > specific section of code (i.e., the directory index, or the extent > tree, or the block allocator, etc.). It's really only ext4 developers > who need to use those debugging statements, and even for them it's > quite rare. > > So I've never considered it worthwhile to enable them via a CONFIG_* > menu item; developers who are debugging a specific problem will > generally just drop in the specific #define on an ad hoc basis, and > that works fine. > If you want to unconditionally define DEBUG for all of ext4, that is fine with me. I was just noting that it seemed a little odd to me, and different from what is done elsewhere. Thanks, David Daney