From mboxrd@z Thu Jan 1 00:00:00 1970 From: Dave Kleikamp Subject: Re: [RFC PATCH v2 15/21] loop: use aio to perform io on the underlying file Date: Fri, 20 Apr 2012 10:57:12 -0500 Message-ID: <4F918758.8020001@oracle.com> References: <1333122228-13633-1-git-send-email-dave.kleikamp@oracle.com> <1333122228-13633-16-git-send-email-dave.kleikamp@oracle.com> <4F917751.9040600@parallels.com> <4F917C2F.4020805@oracle.com> <4F91862F.2030502@zabbo.net> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Cc: Jeff Moyer , "Maxim V. Patlasov" , "linux-fsdevel@vger.kernel.org" , "linux-kernel@vger.kernel.org" To: Zach Brown Return-path: In-Reply-To: <4F91862F.2030502@zabbo.net> Sender: linux-kernel-owner@vger.kernel.org List-Id: linux-fsdevel.vger.kernel.org On 04/20/2012 10:52 AM, Zach Brown wrote: > On 04/20/2012 11:20 AM, Jeff Moyer wrote: >> Dave Kleikamp writes: >> >>> On 04/20/2012 09:48 AM, Maxim V. Patlasov wrote: >>>> On 03/30/2012 07:43 PM, Dave Kleikamp wrote: >>>>> From: Zach Brown >>>>> >>>>> This uses the new kernel aio interface to process loopback IO by >>>>> submitting concurrent direct aio. Previously loop's IO was serialized >>>>> by synchronous processing in a thread. >>>>> >>>> >>>> The patch ignores REQ_FLUSH bit of bi_rw. Is it simply overlook? >>> >>> Good question. Since the loop device is sending only direct IO requests, >>> it shouldn't be necessary to explicitly flush page cache, but REQ_FLUSH >> >> REQ_FLUSH isn't about the page cache, it's about flushing the volatile >> disk write cache. You need to handle that. > > I guess O_DIRECT doesn't routinely issue flushes simply because it's too > expensive? Apps that care about consistent IO and O_DIRECT are expected > to not have writeback caching enabled? 'cause there's no way they're > issuing syncs themselves. If we weren't using aio, we might be okay, but we don't know that any prior asynchronous request has completed. > > So yeah, I'd agree that the loop code should be reworked a bit so that > both the filebacked and aio methods call vfs_sync() when they see > REQ_FLUSH. It's an easy fix. I don't anticipate that it will hurt performance too badly. > > Bleh. > > - z > (Sorry, no real time to dig into this now. Lots more time in two months!)