From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:48821) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1coVwu-0001sS-7J for qemu-devel@nongnu.org; Thu, 16 Mar 2017 10:00:17 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1coVwq-0000pI-F1 for qemu-devel@nongnu.org; Thu, 16 Mar 2017 10:00:16 -0400 Received: from mx1.redhat.com ([209.132.183.28]:34746) by eggs.gnu.org with esmtps (TLS1.0:DHE_RSA_AES_256_CBC_SHA1:32) (Exim 4.71) (envelope-from ) id 1coVwq-0000oR-9F for qemu-devel@nongnu.org; Thu, 16 Mar 2017 10:00:12 -0400 Message-ID: <1489672808.15659.90.camel@redhat.com> From: Gerd Hoffmann Date: Thu, 16 Mar 2017 15:00:08 +0100 In-Reply-To: <143C0AFC63FC204CB0C55BB88F3A8ABB3336BB2E@EX02.corp.qihoo.net> References: <1489656642-12925-1-git-send-email-kraxel@redhat.com> <1489656642-12925-5-git-send-email-kraxel@redhat.com> <143C0AFC63FC204CB0C55BB88F3A8ABB3336BB2E@EX02.corp.qihoo.net> Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable Mime-Version: 1.0 Subject: Re: [Qemu-devel] [PULL for-2.9 4/7] cirrus: add option to disable blitter List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: =?UTF-8?Q?=E6=9D=8E=E5=BC=BA?= Cc: "qemu-devel@nongnu.org" Hi, > > + DEFINE_PROP_BOOL("blitter", struct PCICirrusVGAState, >=20 > > + cirrus_vga.enable_blitter, true), >=20 >=20 > The default is 'ENABLE'? Yes. > I think there should be 'false'. Just flipping it to false isn't an option. At minimum we would have to keep it enabled by default for older machine types, for backward compatibility with older qemu versions. Beside that I don't want rush it. When flipping the default I'd do it early in the 2.10 development cycle, so we have a few months to the next release to see what the fallout is and decide which default we want. But I think the most sensible approach would be to have upper management layer handle this. We have libosinfo, which records capabilities of guests, and we could add a field there telling whenever the guest os in question uses the blitter (and therefore breaks when we turn it off) or whenever it is safe to turn it off. cheers, Gerd