From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932775Ab2KXHE4 (ORCPT ); Sat, 24 Nov 2012 02:04:56 -0500 Received: from mail2.gnudd.com ([213.203.150.91]:52646 "EHLO mail.gnudd.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932718Ab2KXHEz (ORCPT ); Sat, 24 Nov 2012 02:04:55 -0500 Date: Sat, 24 Nov 2012 08:03:37 +0100 From: Alessandro Rubini To: jaswinder.singh@linaro.org Cc: ciminaghi@gnudd.com, viresh.linux@gmail.com, giancarlo.asnaghi@st.com, rajeev-dlh.kumar@st.com, linux@arm.linux.org.uk, nico@fluxnic.net, vinod.koul@intel.com, linus.walleij@linaro.org, hpa@zytor.com, linux-kernel@vger.kernel.org, grant.likely@secretlab.ca, wim@iguana.be, mingo@redhat.com, ben-linux@fluff.org, djbw@fb.com, tglx@linutronix.de, cjb@laptop.org, linux-arm-kernel@lists.infradead.org, davidb@codeaurora.org Subject: Re: [PATCH 1/8 v3] DMA: PL330: use prefix in reg names to build under x86 Message-ID: <20121124070337.GA8030@mail.gnudd.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Organization: GnuDD, Device Drivers, Embedded Systems, Courses In-Reply-To: References: <1353718436-20194-1-git-send-email-ciminaghi@gnudd.com> Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org My patch: >> This driver would not compile if ARM_AMBA is selected under x86, >> because "CS" and "DS" are already defined there. But AMBA >> is used in the x86 world by a PCI-to-AMBA bridge, to be submitted. >> >> The patch just adds the "PL330_" prefix to all register and bit fields, >> so it can be built by randomconfig after ARM_AMBA appears within x86. Jassy Brar: > Prefixing only CS and DS should be do the job. > Why do we have to make every symbol noisy with PL330_ ? For internal consistency. I attacked this problem in July. This is what you wrote: I fully agree with your point and IIRC I always add some prefix to definitions in header files. Private defines in a .c file, without redundant prefixes, sounded like safe to me at the time, but perhaps I was wrong. (references: http://lkml.org/lkml/2012/7/1/56) So I made the changes overall. I dislike needless long patches, but picking symbols to work around the conflict of the day while introducing inconsistent naming doesn't look good to me. The other option is making the driver depend on CONFIG_ARM, like other patches in the series. That would be a smaller patch: we have a dozen AMBA peripherals under X86, and the PL330 was not thre. But I got the impression that we don't want to single-out every driver that doesn't compile: an overall more portable system is better. I'm fine either way, but your "perhaps I was wrong" above made me choose the overall-prefixing approach. Thank you all for your patience in this (especially Davide for the upstreaming effort) /alessandro