From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:43112) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1VJf3h-0003WS-3K for qemu-devel@nongnu.org; Wed, 11 Sep 2013 03:42:01 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1VJf3Y-0008CG-LK for qemu-devel@nongnu.org; Wed, 11 Sep 2013 03:41:53 -0400 Received: from mail-ee0-x234.google.com ([2a00:1450:4013:c00::234]:33978) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1VJf3Y-0008C9-DW for qemu-devel@nongnu.org; Wed, 11 Sep 2013 03:41:44 -0400 Received: by mail-ee0-f52.google.com with SMTP id c41so4337332eek.25 for ; Wed, 11 Sep 2013 00:41:43 -0700 (PDT) Sender: Paolo Bonzini Message-ID: <52301EBD.5070209@redhat.com> Date: Wed, 11 Sep 2013 09:41:49 +0200 From: Paolo Bonzini MIME-Version: 1.0 References: <1378877909-28518-1-git-send-email-famz@redhat.com> <1378877909-28518-6-git-send-email-famz@redhat.com> <52301B4E.7000308@redhat.com> In-Reply-To: <52301B4E.7000308@redhat.com> Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Subject: Re: [Qemu-devel] [RFC PATCH v5 5/6] configure: introduce --enable-modules List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Gerd Hoffmann Cc: peter.maydell@linaro.org, Fam Zheng , mjt@tls.msk.ru, qemu-devel@nongnu.org, stefanha@redhat.com, vilanova@ac.upc.edu, rth@twiddle.net Il 11/09/2013 09:27, Paolo Bonzini ha scritto: > > There are a couple of things that can be improved still (I don't like > obj-save-y for example), but things are taking shape and all of this > looks like something that can be fixed on top. If you look at > converting more parts to modules (e.g. rbd or spice), you can drop that > RFC! :) Talking about spice, a question for Gerd. With Fam's work to enable shared modules, hw/display/qxl* would have to be placed in a module as well because they depend on ui/spice-core.c. Right now, modularization is limited to files that are built once for all of QEMU, which is not the case for qxl. It looks like TARGET_PAGE_SIZE is the only reason why qxl is built per-target, and in qxl_ram_set_dirty it should be enough to do qxl_set_dirty(&qxl->vga.vram, offset, offset + 1); Would it be fine to use a generic 4096 constant everywhere else? Paolo