From mboxrd@z Thu Jan 1 00:00:00 1970 From: Christoph Hellwig Subject: Re: [PATCH 3/4] cciss 2.4: adds 2 ioctls for ia64 based systems Date: Mon, 15 Aug 2005 22:49:03 +0100 Message-ID: <20050815214903.GA12701@infradead.org> References: <20050815212224.GD12760@beardog.cca.cpqcorp.net> <1124141573.5089.55.camel@mulgrave> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Received: from pentafluge.infradead.org ([213.146.154.40]:40093 "EHLO pentafluge.infradead.org") by vger.kernel.org with ESMTP id S964991AbVHOVtr (ORCPT ); Mon, 15 Aug 2005 17:49:47 -0400 Content-Disposition: inline In-Reply-To: <1124141573.5089.55.camel@mulgrave> Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: James Bottomley Cc: mikem , marcelo.tosatti@cyclades.com, Jens Axboe , Linux Kernel , SCSI Mailing List On Mon, Aug 15, 2005 at 04:32:53PM -0500, James Bottomley wrote: > On Mon, 2005-08-15 at 16:22 -0500, mikem wrote: > > +#ifdef CONFIG_IA64 > > + case BLKGETLASTSECT: > > + case BLKSETLASTSECT: > > +#endif > > return blk_ioctl(inode->i_rdev, cmd, arg); > > What makes these two ioctls IA64 specific? I think they're completely > general in 2.4, so there's no need for the #ifdef. They don't exist in 2.4 mainline. The ia64 patch and then many distributions introduced it because the EFI partitions spec contains some braindamage that requries accessing the last sector. In 2.6 the block device nodes can do that, so the ioctls aren't needed either. In short this patch should not go into mainline 2.4, it doesn't support ia64 anyway and if it did it wouldn't compile because the ioctls aren't defined.