From mboxrd@z Thu Jan 1 00:00:00 1970 From: Krzysztof Halasa Subject: Re: sata_via: SError: { UnrecovData Proto TrStaTrns } Date: Tue, 20 Oct 2009 18:00:40 +0200 Message-ID: References: <19147.28614.505296.752363@pilspetsen.it.uu.se> <19157.6679.939371.469696@pilspetsen.it.uu.se> <20091014125344.7d0b2597@lxorguk.ukuu.org.uk> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Received: from khc.piap.pl ([195.187.100.11]:38654 "EHLO khc.piap.pl" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751599AbZJTQAi (ORCPT ); Tue, 20 Oct 2009 12:00:38 -0400 In-Reply-To: <20091014125344.7d0b2597@lxorguk.ukuu.org.uk> (Alan Cox's message of "Wed, 14 Oct 2009 12:53:44 +0100") Sender: linux-ide-owner@vger.kernel.org List-Id: linux-ide@vger.kernel.org To: Alan Cox Cc: Mikael Pettersson , linux-ide@vger.kernel.org, linux-kernel@vger.kernel.org, Robert Hancock Update: I've connected Kingston SSD instead of Transcend's. The same errors appear, but now there is a lot of them (> 30 instead of maybe 5 per my test with Transcend). Alan Cox writes: > These two may be worth playing with, especially latency. Forcing the > latency on the problem board would be interesting given that it is a DMA > transfer that is failing. Changing the latency byte doesn't seem to change anything. OTOH the VT6421A mini-PCI card is the only PCI device in system (not counting the on-CPU AHB-PCI bridge). The problem is also present (to a lesser extent) when I plug it into x86-64 machine (using mini-PCI adapter). The SSD works fine if connected to motherboard's NVidia MCP55 SATA controller. I have read the datasheet and tried some changes to PCI registers but it doesn't fix it. I'm going to test x86-64 + VT6421A + Kingston SSD combination soon. Unfortunately I don't have a (mini) PCI analyser... I will look at REQ/GNT and similar signals on PCI, though. -- Krzysztof Halasa