From mboxrd@z Thu Jan 1 00:00:00 1970 From: Viresh Kumar Subject: Re: [PATCH V4 1/2] cpuidle: simplify multiple driver support Date: Thu, 26 Sep 2013 10:27:31 +0530 Message-ID: References: <1370641990-13884-1-git-send-email-daniel.lezcano@linaro.org> <5242FAFC.8010905@linaro.org> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Return-path: Received: from mail-qc0-f177.google.com ([209.85.216.177]:45415 "EHLO mail-qc0-f177.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750795Ab3IZE5c (ORCPT ); Thu, 26 Sep 2013 00:57:32 -0400 Received: by mail-qc0-f177.google.com with SMTP id x12so431152qcv.36 for ; Wed, 25 Sep 2013 21:57:31 -0700 (PDT) In-Reply-To: <5242FAFC.8010905@linaro.org> Sender: linux-pm-owner@vger.kernel.org List-Id: linux-pm@vger.kernel.org To: Daniel Lezcano Cc: "Rafael J. Wysocki" , "linux-pm@vger.kernel.org" , Lists linaro-kernel , Patch Tracking , Len Brown On 25 September 2013 20:32, Daniel Lezcano wrote: > Where is it wrong in design ? If the multiple driver support is enabled > in the kernel but the driver handles all the cpu, it works. Yeah It works but I don't really support two separate code paths based on how kernel is compiled.. For example: Consider any ARM system which doesn't need multiple drivers support and so must be using the non-per-cpu variable for holding drivers pointer.. Now the same driver is included as part of a multiplatform kernel and now the same driver/platform is using the per-cpu variables for storing driver pointer.. Even sysfs will have files with name "driver" to print drivers name for CPUs.. I know it doesn't break anything but it didn't looked good design wise to me..