From mboxrd@z Thu Jan 1 00:00:00 1970 From: Anthony Wright Subject: Re: USB disk corruption on Xen 4.1.0 & Linux 2.6.38.7 Date: Fri, 29 Jul 2011 20:24:41 +0100 Message-ID: <4E3308F9.9010903@overnetdata.com> References: <20110613220625.GB23755@dumpdata.com> <30260355.6.1308134733567.JavaMail.root@zimbra.overnetdata.com> <20110616201024.GA1184@dumpdata.com> <4DFB2245.7040507@overnetdata.com> <20110620123955.GB2973@dumpdata.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <20110620123955.GB2973@dumpdata.com> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xensource.com Errors-To: xen-devel-bounces@lists.xensource.com To: xen-devel@lists.xensource.com Cc: Ian Pratt , Konrad Rzeszutek Wilk List-Id: xen-devel@lists.xenproject.org On 20/06/2011 13:39, Konrad Rzeszutek Wilk wrote: > On Fri, Jun 17, 2011 at 10:45:41AM +0100, Anthony Wright wrote: >> On 16/06/2011 21:10, Konrad Rzeszutek Wilk wrote: >>> So a couple of things popped at when looking at this: >>> a). "sata_nv 0000:00:08.1: BMDMA: failed to set dma mask, falling back to PIO" >>> >>> That is not good. Somehow it wasn't able to set a 32-bit mask. Don't know >>> if that is due to the driver or just that the kernel can't do it (b/c it >>> was compiled with some special options). >> I never saw anything in dmesg or any other logs. > Uh, I copied that from your logs. > >>> b). Are you using some special .config option? If you were to use a normal >>> distro would you see this? >> My config is attached. > Great. >>> c). You are using a 32-bit dom0. Didn't notice it until now so I wonder >>> if there are some lingering issues (for example not flushing the highmemory >>> mappings) that are at foot. Had you tried a 64-bit kernel before? >> I haven't tried 64 bit because I need a 32 bit kernel. >>> d). Does the problem exist if you try it with a 2.6.39 kernel or the 3.0-rc3? >> It happens on every kernel I've tried which is 2.6.39, 2.6.39.x & 3.0-rc3 >>> e). Can you run lspci -n ? There is a pretty big file in drivers/usb/host/pci-quirks.c >>> Want to check against your machine. >> I have the problem on two machines and have attached the lspci -n for >> both of them. > Is it possible for you provide this binary blob/USB disk somewhere so I can > test it on my hardware? > Just to let the list know, that after a lot of work by konrad we discovered that the problem while present in 3.0-rc3 was no longer present in the final 3.0 release of linux. We never quite understood the cause and I was the only person who could reproduce it reliably, but since it's now fixed we're regarding the issue as closed. thanks, Anthony.