From mboxrd@z Thu Jan 1 00:00:00 1970 From: Oliver Hartkopp Subject: Re: [PATCH can-utils v2 1/3] headers: use include/linux only for Linux kernel headers Date: Mon, 13 Jan 2014 19:25:20 +0100 Message-ID: <52D42F90.7080005@hartkopp.net> References: <1389481823-8379-1-git-send-email-u.kleine-koenig@pengutronix.de> <1389561879-16729-1-git-send-email-u.kleine-koenig@pengutronix.de> <1389561879-16729-2-git-send-email-u.kleine-koenig@pengutronix.de> <52D3867B.6090108@hartkopp.net> <20140113072049.GB29475@pengutronix.de> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: QUOTED-PRINTABLE Return-path: Received: from mo4-p00-ob.smtp.rzone.de ([81.169.146.221]:36578 "EHLO mo4-p00-ob.smtp.rzone.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751316AbaAMSZX (ORCPT ); Mon, 13 Jan 2014 13:25:23 -0500 In-Reply-To: <20140113072049.GB29475@pengutronix.de> Sender: linux-can-owner@vger.kernel.org List-ID: To: =?ISO-8859-1?Q?Uwe_Kleine-K=F6nig?= Cc: linux-can@vger.kernel.org, Marc Kleine-Budde , Yegor Yefremov On 13.01.2014 08:20, Uwe Kleine-K=F6nig wrote: > Hello Oliver, >=20 > On Mon, Jan 13, 2014 at 07:23:55AM +0100, Oliver Hartkopp wrote: >> On 12.01.2014 22:24, Uwe Kleine-K=F6nig wrote: >> >>> - use __kernel_sa_family_t instead of sa_family_t which requires >>> >> can-utils FTBFS on e.g. Debian squeeze. If this is an issue this >>> could be fixed by adding something like: >> >> This was exactly the idea to have a separate set of include files to= make sure >> the 'newest and greatest' tools can be build even on 2.6.24 environm= ents where >> there was no SocketCAN in mainline Linux. > Hmm, the newest and greatest tools are unusable if your kernel doesn'= t > also include the newest and greatest SocketCAN stuff. =46eatures that are not available in the kernel come back with an error= =2E But there's no reason not to have the latest can-utils with bug fixes a= nd functional enhancements for older kernels, right? > And then it is > easy to also have updated kernel headers, isn't it? Well, anyhow I ju= st > found a system here at pengutronix with old enough kernel headers, so= I > will do a fix. ok. tnx. > =20 >> And e.g. this >> >>> #include >>> -#include >>> +#include "isotp.h" >>> #include >> >> should be avoided too. > Why? When isotp hits the kernel, there's no need to adapt all userland progr= ams. I don't think the current idea was that bad. AFAIK the iproute2 package has a similar approach and contains always u= pdated kernel headers to be able to build even on older systems. Best regards, Oliver