Received: from eggs.gnu.org ([209.51.188.92]:34980) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1h2NiN-0001r3-KP for qemu-devel@nongnu.org; Fri, 08 Mar 2019 17:11:40 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1h2NiM-0002p5-SO for qemu-devel@nongnu.org; Fri, 08 Mar 2019 17:11:39 -0500 References: <20190305234337.18353-1-jsnow@redhat.com> <20190305234337.18353-3-jsnow@redhat.com> <28d6183a-f654-de55-36ed-ee0bafb47739@redhat.com> From: John Snow Message-ID: <73ddcaa9-97dc-238b-2949-ac73bc863abe@redhat.com> Date: Fri, 8 Mar 2019 17:11:35 -0500 MIME-Version: 1.0 In-Reply-To: Content-Type: text/plain; charset=utf-8 Content-Language: en-US Content-Transfer-Encoding: 7bit Subject: Re: [Qemu-devel] [PATCH 2/5] block/qcow2-bitmap: Allow bitmap flushing List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Vladimir Sementsov-Ogievskiy , Eric Blake , "qemu-devel@nongnu.org" Cc: Kevin Wolf , "qemu-block@nongnu.org" , Max Reitz On 3/6/19 11:12 AM, Vladimir Sementsov-Ogievskiy wrote: > 06.03.2019 18:59, John Snow wrote: >> >> >> On 3/6/19 7:58 AM, Eric Blake wrote: >>> On 3/5/19 5:43 PM, John Snow wrote: >>>> Usually, we only write out bitmaps when we're about to close out the file, >>>> so we always remove the bitmaps after to make it easier to determine the >>>> source of truth for migration purposes. >>>> >>>> However, there may be times we want to flush bitmap data more aggressively, >>>> like after a truncate event where we need to make the metadata consistent >>>> again. >>> >>> Or, as I've mentioned in other threads, after every >>> bitmap-add/bitmap-delete operation, so that 'qemu-img info -U' can see >>> which persistent bitmaps exist. But that's one step further than this >>> series, so I won't insist on it today. >>> >> >> Yes, I was keeping that in mind as I wrote this. I think it extends to >> those cases trivially, so long as Vladimir is OK with what I'm trying to do. >> > > I still don't see the point of flushing bitmaps, the only thing is optimizing > qemu-img info --force-share, which should never used except for debugging. > > So, at least, if we'll have it, I'd prefer bitmap flushing to be optional, > and other code should not rely on it. > I don't have plans to add more uses of it personally, but I do think that resize represents a genuine case of wanting the feature.