From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-la0-f44.google.com ([209.85.215.44]:48657 "EHLO mail-la0-f44.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754353AbaEISu4 (ORCPT ); Fri, 9 May 2014 14:50:56 -0400 Received: by mail-la0-f44.google.com with SMTP id hr17so169844lab.17 for ; Fri, 09 May 2014 11:50:55 -0700 (PDT) Message-ID: <536D2389.7060800@gmail.com> Date: Fri, 09 May 2014 22:50:49 +0400 From: "Matwey V. Kornilov" MIME-Version: 1.0 To: Lucas De Marchi CC: linux-modules , Rusty Russell , Gustavo Sverzut Barbieri Subject: Re: [PATCH] depmod: Make dependency loops be fatal References: <1399659299-31241-1-git-send-email-lucas.demarchi@intel.com> <20140509184502.GC584@intel.com> In-Reply-To: <20140509184502.GC584@intel.com> Content-Type: text/plain; charset=UTF-8 Sender: linux-modules-owner@vger.kernel.org List-ID: 09.05.2014 22:45, Lucas De Marchi пишет: >> Do I understand you correctly that your patch fails to create modules.dep > > I'd say it avoids losing the previous, valid modules.dep. When building an RPM package, we don't have the previous one. > This is what I'm arguing for... dependency loops are bugs in kernel > modules that should rather be fixed instead of papering over. > > Btw, I don't think they are that common since we never received a bug > report about this in kmod. I'm open to people chiming in, claiming > otherwise. How about introducing --force switch or something like that to alter the behavior in runtime? For instance, the default is to fail, but with --force is to omit looped modules.