From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mx1.redhat.com ([209.132.183.28]:19935 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751948Ab3IRLI6 (ORCPT ); Wed, 18 Sep 2013 07:08:58 -0400 From: David Howells In-Reply-To: <1379459317-13046-1-git-send-email-daniel.santos@pobox.com> References: <1379459317-13046-1-git-send-email-daniel.santos@pobox.com> Subject: Re: [PATCH 0/5] Preliminary: Add error names & descrptions to printks Date: Wed, 18 Sep 2013 12:08:38 +0100 Message-ID: <22288.1379502518@warthog.procyon.org.uk> Sender: linux-kbuild-owner@vger.kernel.org List-ID: To: danielfsantos@att.net Cc: dhowells@redhat.com, linux-kbuild , LKML , Michal Marek , Andrew Morton , "Paul E. McKenney" , Thomas Gleixner , Michael Kerrisk , Dave Hansen , George Spelvin danielfsantos@att.net wrote: > Typically, we don't care about error messages or names in the kernel because > userspace will manage that. But sometimes we need to output an error number > to printks and that creates a situation where a user, system admistrator or > developer must find an error number reference to figure out what went wrong > with a particular driver or whatever. This patch adds two alternatives at > increasing memory costs: > > 1. print the number in addition to the name for 2k extra or > 2. print the number, name and description for 6k extra. I like the idea generally - and have occasionally entertained the idea of implementing it myself. However, I wouldn't bother with the "human readable" description if we're going to do this. Generally, the symbolic representation is good enough - and that's what you're going to grep the code for anyway. David