From mboxrd@z Thu Jan 1 00:00:00 1970 From: Alex Courbot Subject: Re: [PATCH 5/9] gpiolib: use gpio_chips list in gpiochip_find_base Date: Wed, 6 Feb 2013 13:48:19 +0900 Message-ID: <5111E093.4060600@nvidia.com> References: <1359822572-26009-1-git-send-email-acourbot@nvidia.com> <1359822572-26009-7-git-send-email-acourbot@nvidia.com> Mime-Version: 1.0 Content-Type: text/plain; charset="ISO-8859-1"; format=flowed Content-Transfer-Encoding: 7bit Return-path: Received: from hqemgate03.nvidia.com ([216.228.121.140]:12103 "EHLO hqemgate03.nvidia.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754462Ab3BFEs1 (ORCPT ); Tue, 5 Feb 2013 23:48:27 -0500 In-Reply-To: Sender: linux-arch-owner@vger.kernel.org List-ID: To: Linus Walleij Cc: Haojian Zhuang , Grant Likely , Arnd Bergmann , "linux-arch@vger.kernel.org" , "linux-arm-kernel@lists.infradead.org" , "linux-kernel@vger.kernel.org" , "gnurou@gmail.com" On 02/06/2013 02:21 AM, Linus Walleij wrote: > This looks like it is preserving this userspace-sensitive semantic > so that dynamically added chips will still get the same assigned > numbers. It does (it should, at least), the assigned ranges should be strictly identical to the previous version. While testing I also made sure all chips had the same GPIO range. > But I want some guarantees, so state clearly in the commit > that any dynamically assigned chip will get the same base > address after this change as it got before it, and please take > this opportunity to add a comment stating that this search > method is vital for userspace ABIs, and must be preserved. Done. I will take the blame if something goes wrong. :) Thanks, Alex.