From mboxrd@z Thu Jan 1 00:00:00 1970 From: Tejun Heo Subject: Re: sata-nv and 790i Date: Sun, 16 Nov 2008 15:11:20 +0900 Message-ID: <491FB988.50601@kernel.org> References: <20081115004722.5158e125@werewolf.home> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: QUOTED-PRINTABLE Return-path: Received: from hera.kernel.org ([140.211.167.34]:43350 "EHLO hera.kernel.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751071AbYKPGLO (ORCPT ); Sun, 16 Nov 2008 01:11:14 -0500 In-Reply-To: <20081115004722.5158e125@werewolf.home> Sender: linux-ide-owner@vger.kernel.org List-Id: linux-ide@vger.kernel.org To: =?ISO-8859-1?Q?=22J=2EA=2E_Magall=F3n=22?= Cc: Linux-Kernel , Linux-IDE , Mdv Cooker J.A. Magall=F3n wrote: > Hi all... >=20 > I recently put my hands on a system with an ASUS P5N64 WS Pro mobo: >=20 > http://www.asus.com/products.aspx?l1=3D3&l2=3D11&l3=3D653&l4=3D0&mode= l=3D2182&modelmenu=3D1 >=20 > It has a nForce 790i Ultra SLI chipset, the part for the SATA control= ler > is MCP55. >=20 > The problem is that linux seems to fail to read the hard drive on boo= t. > I will explain. >=20 > I have tried to install both CentOS 5.2 and Mandriva 2009.0. > For CentOS (has a 2.6.18 kernel), the install seems ok, but when I re= boot > into the just intalled system the kernel hangs just after printing > the name of the initrd. Mandriva (2.6.27-rc8) just hungs half the ins= tall process. > Live versions of both hang on boot saying they can not find the root > device. >=20 > I have found some messages in the net about the MCP55 in that board n= ot > being completeley suppported as AHCI in sata_nv, and that some other > boards work puttin the SATA controller in 'compatible' mode (with > loss of performance). > But I can't find any setting in the P5N64 to put SATA in Compat mode. >=20 > Can I do this via some kernel parameters ? > Somabody has this board working ? Do current kernels have better supp= ort > for this chipset ? All three flavors of sata_nv's - generic (yours), nf2/3 and ck804 - had various detection problems from 2.6.27-rc1. The problem was that their hardresets all worked differently. 2.6.27-rc1 started to favor hardresets over softresets and exposed this problem. While trying to fix the problem, because all three flavors unexpectedly showed differen= t hardreset behaviors, there were some mistakes. A proper fix is include= d in upstream and is queued for -stable. Sorry about the mess. Thanks. --=20 tejun