From mboxrd@z Thu Jan 1 00:00:00 1970 From: Alan Cox Subject: Re: Windows Bug Check 0x101 issue Date: Wed, 26 Mar 2008 12:34:47 +0000 Message-ID: <20080326123447.2397261d@core> References: <7k4pawfnxs.fsf@pingu.sky.yk.fujitsu.co.jp> <7k1w5zf50j.fsf@pingu.sky.yk.fujitsu.co.jp> <18408.57824.846687.465435@mariner.uk.xensource.com> <20080325175718.GT4411@implementation.uk.xensource.com> <7ky786dkup.fsf@pingu.sky.yk.fujitsu.co.jp> <18410.9596.488204.878357@mariner.uk.xensource.com> <20080326102744.3124cbb4@core> <18410.10932.522487.743717@mariner.uk.xensource.com> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <18410.10932.522487.743717@mariner.uk.xensource.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: Ian Jackson Cc: xen-devel@lists.xensource.com, Keir Fraser , Kouya Shimura , Samuel Thibault List-Id: xen-devel@lists.xenproject.org > Hrm. I think it would be better to return an endless sequence of > errors than to return no error at all. Since we don't know which > block failed, we could report an error flushing block 0, and then an > error flushing block 1, and so on ? On a raid volume that will give wrong results and incorrect recovery behaviour. You need to report correct information. In addition a lot of software treats repeated errors on flush as a device failure after a certain count. If a write to the underlying emulated media fails you know at least which write fails and usually the page that failed (as you get a short write) Alan