From mboxrd@z Thu Jan 1 00:00:00 1970 From: Sergei Shtylyov Subject: Re: [PATCH 1/2] [IDE] Platform IDE driver Date: Wed, 25 Jul 2007 23:35:00 +0400 Message-ID: <46A7A5E4.4090105@ru.mvista.com> References: <20070725165318.5331.23795.stgit@localhost.localdomain> <46A79DE0.8060405@ru.mvista.com> <46A79F14.9040409@freescale.com> <46A7A17C.8090505@ru.mvista.com> <46A7A1D5.7020003@freescale.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <46A7A1D5.7020003@freescale.com> Sender: linux-kernel-owner@vger.kernel.org To: Scott Wood Cc: Vitaly Bordug , linux-ide@vger.kernel.org, linux-kernel@vger.kernel.org, linuxppc-dev@ozlabs.org List-Id: linux-ide@vger.kernel.org Scott Wood wrote: >>> It doesn't buy us anything in here, but it's conceivable that someone >>> may want to write a driver that uses a shift in the I/O accessor >>> rather than an array of port offsets, >> It wouldn't be IDE driver then, and neither it would be libata >> which also does this another way this (despite pata_platform uses >> shifts too -- not in the accessors, so no speed loss). > The device tree is not just for Linux. Yeah, and I can't wait to see some other its users. ;-) This doesn't mean that shift is better anyway. If everyone considers it better, I give up. But be warned that shift (stride) is not the only property characterizing register accesses -- the regs might be only accessible as 16/32-bit quantities, for example (16-bit is a real world example -- from Amiga or smth of that sort, IIRC). >>> equivalent of the cntlzw innstruction, and shift makes it clear that >>> the stride must be power-of-two). Plus, using shift is consistent >>> with what we do on ns16550. >> Why the heck should we care about the UART code taling about IDE?! > Consistency? We're not obliged to be consistent with every piece of the kernel code. > -Scott MBR, Sergei