From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:58818) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1dX4Fl-000416-VY for qemu-devel@nongnu.org; Mon, 17 Jul 2017 07:31:54 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1dX4Fl-0002n6-8g for qemu-devel@nongnu.org; Mon, 17 Jul 2017 07:31:53 -0400 Date: Mon, 17 Jul 2017 13:31:42 +0200 From: Kevin Wolf Message-ID: <20170717113142.GF5301@noname.redhat.com> References: <1499946095-8224-1-git-send-email-kwolf@redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <1499946095-8224-1-git-send-email-kwolf@redhat.com> Subject: Re: [Qemu-devel] [PATCH v2 0/9] block: -device drive= fixes List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: qemu-block@nongnu.org Cc: mreitz@redhat.com, eblake@redhat.com, jsnow@redhat.com, pbonzini@redhat.com, armbru@redhat.com, qemu-devel@nongnu.org Am 13.07.2017 um 13:41 hat Kevin Wolf geschrieben: > Since 2.9 it is possible for a user to care only about nodes and qdev devices, > but not about BlockBackends, by defining with -blockdev and using their node > name in -device. Devices create an anonymous BlockBackend internally then. > > One of the major problems in this setup is that such devices are not listed in > query-block or HMP 'info block' because the BBs are not monitor-owned. A recent > discussion concluded that we really want to replace query-block and friends > with something that matches QEMU's internals better, but there's no way to do > that in 2.10 (and I wouldn't bet on it to be completed for 2.11). So for now, > just make the devices appear in query-block and add the qdev ID there in order > to make the devices identifiable (they have 'device': '' because the BB is > anonymous). > > With the additional qdev field, it turned out that with empty drives, ide-cd > and scsi-cd weren't even properly attached to the BlockBackend. If you do it > right, this even results in crashes. (Floppy already gets it right, in case you > wondered.) > > After this series, -device drive= should be reasonably usable > without a major loss of functionality like before. Applied to the block branch. Kevin