From mboxrd@z Thu Jan 1 00:00:00 1970 From: Steve Ralston Subject: Re: [TRIVIAL] warning cleanup for drivers_scsi_qla1280.c Date: Wed, 07 Aug 2002 06:12:04 -0500 Sender: linux-scsi-owner@vger.kernel.org Message-ID: <3D510084.3040708@cox.net> References: <20020807043057.395C74BD6@lists.samba.org> <1028720225.18156.246.camel@irongate.swansea.linux.org.uk> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Return-path: List-Id: linux-scsi@vger.kernel.org To: Alan Cox Cc: linux-scsi@vger.kernel.org On 08/07/2002 06:37 AM, Alan Cox wrote: >General policy - rip out version ifdefs unless the driver has a >maintainer who actively objects ? > Hi Alan, When I worked for LSI, and was actively maintaining drivers/message/fusion/... code, we had a single set of driver source code that compiled and ran across lots of linux kernels (2.2.5 thru 2.4.19). Most of the version-specific #ifdef's were isolated to one header file (linux_compat.h). We then also had a set of patch files (each pretty small) that could be applied against any explicit kernel (in the supported range). A customer could then (in theory:-) download one set of driver source (including the patch file sets), untar/overlay it, apply a single patch file, and then compile+run the drivers on any of those kernels. So just curious, are you saying that you'd rather have driver maintainers maintain a separate set of driver source for each explicit kernel (with no version-specific #ifdef's) vs. the way we did it above? Thanks, -SteveR