From mboxrd@z Thu Jan 1 00:00:00 1970 From: Mark Lord Subject: Re: [PATCH] Fixup ahci suspend / resume Date: Tue, 28 Feb 2006 09:45:54 -0500 Message-ID: <44046222.4090008@rtr.ca> References: <44045FB1.5040408@suse.de> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Return-path: Received: from rtr.ca ([64.26.128.89]:31932 "EHLO mail.rtr.ca") by vger.kernel.org with ESMTP id S1751002AbWB1OqA (ORCPT ); Tue, 28 Feb 2006 09:46:00 -0500 In-Reply-To: <44045FB1.5040408@suse.de> Sender: linux-ide-owner@vger.kernel.org List-Id: linux-ide@vger.kernel.org To: Hannes Reinecke Cc: Linux Kernel , linux-ide@vger.kernel.org Hannes Reinecke wrote: > Hi all, > > the current ahci doesn't handle suspend / resume cycles too well. > On certain machines the disk just locks up and refuses to work after a > resume. > > This is due to a initialisation error after resume; ahci has fo[u]r > registers containing DMA addresses (which are obviously allocated by the > kernel). Of course there is no guarantee that these addresses are > unchanged across reboots. So ahci loads the suspended image, and after > the suspended image is started the driver is presented with different > DMA addresses, whereas the chip still uses the original ones. > > This patch rearranges the suspend / resume code to properly initialise > those registers after a resume. It also contains some initialisation > fixes to make the driver behave more spec-compliant. .. Is ahci the *only* interface afflicted with this problem?