Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1030613AbcCQOZx (ORCPT ); Thu, 17 Mar 2016 10:25:53 -0400 Received: from slow1-d.mail.gandi.net ([217.70.178.86]:41792 "EHLO slow1-d.mail.gandi.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751428AbcCQOZu (ORCPT ); Thu, 17 Mar 2016 10:25:50 -0400 X-Originating-IP: 88.159.34.112 From: Maarten ter Huurne To: Mark Brown Cc: Wenyou Yang , Zubair Lutfullah Kakakhel , Liam Girdwood , linux-kernel@vger.kernel.org Subject: Re: [PATCH 7/7] regulator: act8865: Init at subsys level Date: Thu, 17 Mar 2016 15:17:42 +0100 Message-ID: <8580166.enKcLbBhVL@hyperion> User-Agent: KMail/4.14.10 (Linux/4.1.15-8-default; KDE/4.14.16; x86_64; ; ) In-Reply-To: <20160229113537.GZ18327@sirena.org.uk> References: <1456674809-16422-1-git-send-email-maarten@treewalker.org> <1456674809-16422-7-git-send-email-maarten@treewalker.org> <20160229113537.GZ18327@sirena.org.uk> MIME-Version: 1.0 Content-Transfer-Encoding: 7Bit Content-Type: text/plain; charset="us-ascii" Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 1340 Lines: 28 On Monday 29 February 2016 20:35:37 Mark Brown wrote: > On Sun, Feb 28, 2016 at 04:53:29PM +0100, Maarten ter Huurne wrote: > > Since the defined regulators are used in other drivers, we can avoid > > deferred probing by registering this driver sooner. > > No, don't play silly link order games. We are never going to get a > link order that avoids deferred probes, faffing about changing things > to try to make one just leads to lots of noise. If you want to avoid > noise from deferred probe then work on something that communicates > dependencies to the driver core so we don't have to brute force > things, Raphael has some ideas there he's been looking at. On our device the LCD panel is powered by one of the ACT8600 outputs. Avoiding the deferred probe means the screen turns on a couple of seconds earlier. I agree that the current hardcoded priority system is not ideal and a dependency based system would be the right solution in the long term, but I was looking for a short term solution with this patch. In any case, I excluded this patch from the v2 series. I cleaned up the register configuration (debugfs) patch, split the of_node patch into two patches and resubmitted the redundant dev lookups patch. Thanks for reviewing and please let me know if there is anything more that needs changing. Bye, Maarten