From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:44920) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1a8Xpa-0003uV-3A for qemu-devel@nongnu.org; Mon, 14 Dec 2015 13:26:42 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1a8XpW-0001pI-UB for qemu-devel@nongnu.org; Mon, 14 Dec 2015 13:26:42 -0500 Received: from mx1.redhat.com ([209.132.183.28]:33568) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1a8XpW-0001p8-PJ for qemu-devel@nongnu.org; Mon, 14 Dec 2015 13:26:38 -0500 Date: Mon, 14 Dec 2015 20:26:35 +0200 From: "Michael S. Tsirkin" Message-ID: <20151214202556-mutt-send-email-mst@redhat.com> References: <1449994112-7054-1-git-send-email-shmulik.ladkani@ravellosystems.com> <1449994112-7054-7-git-send-email-shmulik.ladkani@ravellosystems.com> <566EF753.4070605@redhat.com> <20151214190935-mutt-send-email-mst@redhat.com> <163895392.40236889.1450114130220.JavaMail.zimbra@redhat.com> <20151214193220-mutt-send-email-mst@redhat.com> <566F04B1.8070204@redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <566F04B1.8070204@redhat.com> Subject: Re: [Qemu-devel] [PATCH v2 6/6] vmw_pvscsi: Introduce 'x-disable-pcie' backword compatability property List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Paolo Bonzini Cc: Dmitry Fleytman , Marcel Apfelbaum , idan brown , Shmulik Ladkani , qemu-devel@nongnu.org On Mon, Dec 14, 2015 at 07:04:33PM +0100, Paolo Bonzini wrote: > > > On 14/12/2015 18:35, Michael S. Tsirkin wrote: > > > What distinguishes an internal from an external property? Everything > > > except links to backends would be "internal". > > > > How do you mean? We have a ton of properties e.g. > > to control which offloads are allowed for virtio net. > > Why would users set them? > > > It's set internally by machine types to avoid breaking > > migration. I don't see any reason for users to set it. > > But they do set it :) albeit only through machine types. I don't think > it's different from offloads, just much more specialized. > > Or do you mean that it could go away if we decide to remove very old > machine types? I think we would remove compat properties connected to > those machine types as well, even without "x-". > > Paolo Then we'll break users who set them directly for some reason. So x- means "not part of stable ABI". No?