From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([140.186.70.92]:43225) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1QD0RN-0005jS-59 for qemu-devel@nongnu.org; Thu, 21 Apr 2011 16:25:29 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1QD0RM-00063Z-6f for qemu-devel@nongnu.org; Thu, 21 Apr 2011 16:25:29 -0400 Received: from mail-ww0-f41.google.com ([74.125.82.41]:33861) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1QD0RM-00063U-2L for qemu-devel@nongnu.org; Thu, 21 Apr 2011 16:25:28 -0400 Received: by wwi18 with SMTP id 18so141442wwi.4 for ; Thu, 21 Apr 2011 13:25:27 -0700 (PDT) MIME-Version: 1.0 In-Reply-To: <20110421124938.GB27317@lst.de> References: <1303375391-26679-1-git-send-email-ronniesahlberg@gmail.com> <20110421085026.GA4817@lst.de> <20110421090935.GA5381@lst.de> <20110421124938.GB27317@lst.de> Date: Fri, 22 Apr 2011 06:25:26 +1000 Message-ID: From: ronnie sahlberg Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Subject: Re: [Qemu-devel] iSCSI support for QEMU List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Christoph Hellwig Cc: Stefan Hajnoczi , qemu-devel@nongnu.org, stefanha@linux.vnet.ibm.com On Thu, Apr 21, 2011 at 10:49 PM, Christoph Hellwig wrote: > On Thu, Apr 21, 2011 at 09:36:12PM +1000, ronnie sahlberg wrote: >> There are some high-volume arrays that advertise support but fail any >> cdb with FUA, FUA_NV bits set with sense, so it needs to be made optiona= l. > > Which on would that be? =A0Linux uses the FUA bit if the device advertise= s support > via the DPOFUA bit since 2005, and prints a warning if a FUA write fails, > and since last year even fails the request hard. =A0I'd be really surpris= ed > if a common device fails FUA writes and we didn't know about it by now. > > Either way you'll have to still guarantee data made it to non-volatile > storage for cache=3Dwritethrough mode, either by disabling the WCE bit > using MODE SELECT, or by flushing the cache after every write. > Thankyou for your patience. I understand and I have updated the patch to set FUA when the _WB flag is not set. regards ronnie sahlberg