From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from motgate8.mot.com ([129.188.136.8]) by pentafluge.infradead.org with esmtp (Exim 4.30 #5 (Red Hat Linux)) id 1Anuya-0003Oi-Ag for linux-mtd@lists.infradead.org; Tue, 03 Feb 2004 07:28:04 +0000 Received: from il06exr02.mot.com (il06exr02.mot.com [129.188.137.132]) by motgate8.mot.com (Motorola/Motgate3) with ESMTP id i137S15w016422 for ; Tue, 3 Feb 2004 00:28:01 -0700 (MST) Received: from zch07exm04.corp.mot.com (ZCH07EXM04.corp.mot.com [200.3.1.210]) by il06exr02.mot.com (Motorola/il06exr02) with ESMTP id i137Qdup014981 for ; Tue, 3 Feb 2004 01:26:41 -0600 From: Li Yang To: David Woodhouse In-Reply-To: <1075789858.19654.5.camel@imladris.demon.co.uk> References: <1075788182.20041.38.camel@Gundam> <1075789858.19654.5.camel@imladris.demon.co.uk> Content-Type: text/plain Message-Id: <1075793438.20041.51.camel@Gundam> Mime-Version: 1.0 Date: 03 Feb 2004 15:30:39 +0800 Content-Transfer-Encoding: 7bit cc: linux-mtd@lists.infradead.org Subject: Re: Mtd compile fails on ppc arch Reply-To: leoli@motorola.com List-Id: Linux MTD discussion mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , On Tue, 2004-02-03 at 14:30, David Woodhouse wrote: > On Tue, 2004-02-03 at 14:03 +0800, Li Yang wrote: > > Hi, > > > > I updated the mtd driver in kernel through the latest snapshot. I have > > tested kernel 2.4.20 and 2.4.22. The same parse error occurred for ppc > > arch. I have tried two cross toolchains, the results are the same. So > > it's unlikely to be cross compiler problem. Btw: i386 compiles well. I > > looked into some code, the problem seems to be related to headers. But > > I can't figure it out. Does anyone have a clue? > > #include > > Evidently it gets included somehow on i386 via one of the other headers > we _do_ include -- but not on ppc. But sharp.c is deprecated and you Yes, sharp.c need linux/slab.h and linux/init.h to compile successfully which are removed from headers in linux/mtd like linux/mtd/map.h. After adding these two includes, it can compile well. > should be using jedec_probe and cfi_cmdset_0002 instead. -- Thank you, Leo