From: FD Cami Subject: Re: [PATCH] extX: convert prink(KERN_WARNING) to extX_warning() Date: Tue, 24 Jun 2008 00:20:23 +0200 Message-ID: <20080624002023.0fadc1d1@olorin> References: <20080622201841.669ff882@olorin> <20080623203107.GD6239@webber.adilger.int> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Cc: linux-kernel@vger.kernel.org, linux-ext4@vger.kernel.org, Andrew Morton , Kasper Sandberg , Eric Sandeen , Randy Dunlap To: Andreas Dilger Return-path: Received: from smtp5-g19.free.fr ([212.27.42.35]:45649 "EHLO smtp5-g19.free.fr" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751208AbYFWWUV (ORCPT ); Mon, 23 Jun 2008 18:20:21 -0400 In-Reply-To: <20080623203107.GD6239@webber.adilger.int> Sender: linux-ext4-owner@vger.kernel.org List-ID: Hi, (Eric and Randy added as CC to keep track of this) On Mon, 23 Jun 2008 14:31:07 -0600 Andreas Dilger wrote: > On Jun 22, 2008 20:18 +0200, FD Cami wrote: > > This patch takes a shot at replacing the direct use of printk(KERN_WARNING) > > in extX by extX_warning. We now get the device number in the warning message : > > "EXT3-fs warning (device hda1): ext3_setup_super: maximal mount count reached, running e2fsck is recommended" > > It also adds the device information on a pair of printk(KERN_ERR) (this was > > first suggested by Kasper two weeks ago). > > The one issue is that KERN_ERR != KERN_WARNING, so these errors might > not be visible on the console, or may not be saved to the syslog. Hmmm, I did not replace any KERN_ERR by ext3_warning / KERN_WARNING, but made those KERN_ERR calls more verbose. Does that change anything ? For the record, I wanted to replace KERN_ERR calls by calls to a wrapper much like ext3_warning (but obviously calling KERN_ERR), later on. > The other minor difference is that the function name is also printed, and > this makes the error message very long. Ack, not pretty. Will fix that in next version. > One suggestion is to create a separate macro that passes the KERN_* > flag and __func__ to ext3_console_msg(), and call that from > ext3_warning() and a new ext3_start_error() function. I always found > it annoying to have to specify __func__ as a parameter for every call. OK, will do. Thanks for the suggestion. > > ext3_warning(sb, __func__, > > - "updating to rev %d because of new feature flag, " > > - "running e2fsck is recommended", > > - EXT3_DYNAMIC_REV); > > + "updating to rev %d because of new feature flag, " > > + "running e2fsck is recommended", > > + EXT3_DYNAMIC_REV); > > Please don't change all of the indenting. The old indending is proper > linux coding style (aligned with previous '('), the new one is not. Then this (old) coding style is not consistent within super.c , and not consistent with the example given in Documentation/CodingStyle either. I will "fix" the indentation in a separate patch for the whole file, to be merged or not. > Note that you need to split up the patches for ext2, ext3, ext4 into > separate emails. I'd suggest just sending one of them until we agree > on what is right, then submitting the rest afterward. OK. I did ext3/ext4 at once because of what Andrew replied to Kasper : "We like to keep ext3 and ext4 in sync as much as poss, please." http://lkml.org/lkml/2008/6/9/79 But I can do ext3 first and then back / forward port to ext2 and ext4 as well, as you suggest. Thank you very much for replying. Best, Francois