From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:53319) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1aR3To-0002br-J9 for qemu-devel@nongnu.org; Wed, 03 Feb 2016 14:52:48 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1aR3Tj-00058S-IS for qemu-devel@nongnu.org; Wed, 03 Feb 2016 14:52:44 -0500 Received: from mx1.redhat.com ([209.132.183.28]:56270) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1aR3Tj-00058M-DD for qemu-devel@nongnu.org; Wed, 03 Feb 2016 14:52:39 -0500 Received: from int-mx11.intmail.prod.int.phx2.redhat.com (int-mx11.intmail.prod.int.phx2.redhat.com [10.5.11.24]) by mx1.redhat.com (Postfix) with ESMTPS id 8ECC5C0B2E12 for ; Wed, 3 Feb 2016 19:52:38 +0000 (UTC) Message-ID: <1454529157.18969.12.camel@redhat.com> From: Alex Williamson Date: Wed, 03 Feb 2016 12:52:37 -0700 In-Reply-To: <1454491768.4967.56.camel@redhat.com> References: <20160202033735.29810.41508.stgit@gimli.home> <20160202200845.5260.64183.stgit@gimli.home> <1454491768.4967.56.camel@redhat.com> Content-Type: text/plain; charset="UTF-8" Mime-Version: 1.0 Content-Transfer-Encoding: quoted-printable Subject: Re: [Qemu-devel] [RFC PATCH v2 7/7] vfio/pci: Find and expose Intel IGD OpRegion List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Gerd Hoffmann Cc: qemu-devel@nongnu.org On Wed, 2016-02-03 at 10:29 +0100, Gerd Hoffmann wrote: > On Di, 2016-02-02 at 13:09 -0700, Alex Williamson wrote: > > This is provided via a device specific region, look for it on Intel > > VGA class devices.=C2=A0=C2=A0Our default mechanism to expose this to= the BIOS > > is via fw_cfg where it's expected that the BIOS will copy the data > > into a reserved RAM area and update the ASL Storage register to > > reference the GPA of that buffer. >=C2=A0 > > =C2=A0 We also support directly mapping > > the OpRegion through to the host in response to the ASL Storage > > register write, which makes the data "live" and potentially provides > > write access should the underlying vfio region support writes. >=C2=A0 > This should better be splitted into a separate patch. Ok.=C2=A0=C2=A0Perhaps I'll drop it for now and make the kernel-level vfi= o code expose the OpRegion as read-only as a start.=C2=A0=C2=A0It's easy enough = to dig out of the mail archives and add back later if we want. > > This > > option is automatically enabled if we somehow don't support fw_cfg (i= s > > this a good idea?). >=C2=A0 > I think this can't happen.=C2=A0=C2=A0And even in case it can: we have = bigger > problems than the opregion then. A few of the other fw_cfg_find() users seem to handle the error case, which is why I ask.=C2=A0=C2=A0Thanks, Alex