From mboxrd@z Thu Jan 1 00:00:00 1970 From: Mark Langsdorf Subject: Re: [PATCH 2/2] ahci_platform: add custom hard reset for Calxeda ahci ctrlr Date: Fri, 17 Aug 2012 17:25:31 -0500 Message-ID: <502EC4DB.6020707@calxeda.com> References: <1345215111-11051-1-git-send-email-mark.langsdorf@calxeda.com> <1345215111-11051-3-git-send-email-mark.langsdorf@calxeda.com> <502E7E7E.80101@pobox.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Return-path: Received: from smtp141.dfw.emailsrvr.com ([67.192.241.141]:38268 "EHLO smtp141.dfw.emailsrvr.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1757934Ab2HQWZX (ORCPT ); Fri, 17 Aug 2012 18:25:23 -0400 In-Reply-To: <502E7E7E.80101@pobox.com> Sender: linux-ide-owner@vger.kernel.org List-Id: linux-ide@vger.kernel.org To: Jeff Garzik Cc: "linux-ide@vger.kernel.org" , "probinson@gmail.com" , "jonathon@jonmasters.org" , "dmarlin@redhat.com" , Rob Herring On 08/17/2012 12:25 PM, Jeff Garzik wrote: > On 08/17/2012 10:51 AM, Mark Langsdorf wrote: >> Calxeda highbank SATA phy has intermittent problems bringing up a link >> with Gen3 drives. Retrying the phy hard reset can work-around this issue, >> but each reset also disables spread spectrum support. The reset function >> also needs to reprogram the phy to enable spread spectrum support. >> >> Signed-off-by: Mark Langsdorf >> Signed-off-by: Rob Herring >> --- >> .../devicetree/bindings/arm/calxeda/combophy.txt | 17 ++ >> .../devicetree/bindings/ata/ahci-platform.txt | 8 + >> arch/arm/boot/dts/highbank.dts | 17 ++ >> drivers/ata/Makefile | 1 + >> drivers/ata/ahci.h | 17 ++ >> drivers/ata/ahci_platform.c | 88 ++++++++- >> drivers/ata/sata_highbank.c | 199 ++++++++++++++++++++ >> 7 files changed, 339 insertions(+), 8 deletions(-) >> create mode 100644 Documentation/devicetree/bindings/arm/calxeda/combophy.txt >> create mode 100644 drivers/ata/sata_highbank.c > > And the biggest issue... > > 4) if you need all this code for workarounds, just create your own > libahci driver at that point. make sata_highbank a real driver. The frustrating bit is that we're nominally an AHCI compatible controller except for the reset function and some kind of initialization for the combophy iomem regions. The ata_port_operations already lets me override just the reset function, and the ahci_probe() function has a hook for the iomem initialization. Is there any way I can inherit most of the existing ahci_platform functions for suspend, resume, etc.? Because otherwise I'm just going to duplicate the code and that feels unsatisfactory and seems unlikely to get accepted anyway. Would it be permissible to remove the statics from ahci_suspend, ahci_resume, and ahci_remove so I could use them in my libahci driver? --Mark Langsdorf Calxeda, Inc.