From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from stat16.steeleye.com ([209.192.50.48]:50049 "EHLO hancock.sc.steeleye.com") by vger.kernel.org with ESMTP id S267685AbUIOWjO (ORCPT ); Wed, 15 Sep 2004 18:39:14 -0400 Subject: Re: RFC: being more anal about iospace accesses.. From: James Bottomley In-Reply-To: References: Content-Type: text/plain Content-Transfer-Encoding: 7bit Date: 15 Sep 2004 18:38:50 -0400 Message-Id: <1095287935.1688.4.camel@mulgrave> Mime-Version: 1.0 To: Linus Torvalds Cc: Geert Uytterhoeven , Linux Arch list , Al Viro , Andrew Morton , Alan Cox , "David S. Miller" , Jeff Garzik List-ID: On Wed, 2004-09-15 at 15:16, Linus Torvalds wrote: > What this new interface is doing is to slowlt move "ioremap()" users over > to "pci_iomap()" (and by extension, maybe "sbus_ioremap()" etc if somebody > ever starts doing that too), which allows you to do the bus- and device- > specific translation just _once_. Please can we not do it this way, that was the old pci_map_single et al which we've just spent quite a bit of time converting to dma_map_single. It might not matter to the PCI centric people, but the generic DMA API was a wonderful simplification to the rest of us. Can we start iomap off on the right foot and simply do dma_iomap (or dev_iomap since dma_ may end up to be confusing), taking a struct device as an argument and let the platform, which knows where the buses are, map this to the correct operations? James