From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:51394) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1dcZSQ-0007HB-AJ for qemu-devel@nongnu.org; Tue, 01 Aug 2017 11:51:44 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1dcZSL-0003cw-Mp for qemu-devel@nongnu.org; Tue, 01 Aug 2017 11:51:42 -0400 Received: from mx1.redhat.com ([209.132.183.28]:52352) by eggs.gnu.org with esmtps (TLS1.0:DHE_RSA_AES_256_CBC_SHA1:32) (Exim 4.71) (envelope-from ) id 1dcZSL-0003bl-GZ for qemu-devel@nongnu.org; Tue, 01 Aug 2017 11:51:37 -0400 Date: Tue, 1 Aug 2017 18:51:18 +0300 From: "Michael S. Tsirkin" Message-ID: <20170801184904-mutt-send-email-mst@kernel.org> References: <1501285073-2215-1-git-send-email-zuban32s@gmail.com> <1501285073-2215-3-git-send-email-zuban32s@gmail.com> <20170801183208-mutt-send-email-mst@kernel.org> <789fdbb3-73b8-9308-5d94-b3944fd70d03@redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <789fdbb3-73b8-9308-5d94-b3944fd70d03@redhat.com> Subject: Re: [Qemu-devel] [PATCH v3 2/5] hw/pci: introduce pcie-pci-bridge device List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Marcel Apfelbaum Cc: Alexander Bezzubikov , qemu-devel@nongnu.org, Kevin OConnor , lersek@redhat.com, seabios@seabios.org, Gerd Hoffmann , Igor Mammedov , pbonzini@redhat.com, rth@twiddle.net, ehabkost@redhat.com On Tue, Aug 01, 2017 at 06:45:13PM +0300, Marcel Apfelbaum wrote: > On 01/08/2017 18:32, Michael S. Tsirkin wrote: > > On Mon, Jul 31, 2017 at 09:40:41PM +0300, Alexander Bezzubikov wrote: > > > > > +typedef struct PCIEPCIBridge { > > > > > + /*< private >*/ > > > > > + PCIBridge parent_obj; > > > > > + > > > > > + bool msi_enable; > > > > > > > > > > > > Please rename the msi_enable property to "msi" in order > > > > to be aligned with the existent PCIBridgeDev and > > > > consider making it OnOffAuto for the same reason. > > > > (I am not sure about the last part though, we have > > > > no meaning for "auto" here) > > > > > > > > > > Agreed about "msi", but OnOffAuto looks weird to me > > > as we always want MSI to be enabled. > > > > Hi Michael, > > > Why even have a property then? Can't you enable it unconditionally? > > > > Because of a current bug in Linux kernel: > https://www.spinics.net/lists/linux-pci/msg63052.html > msi will not work until the patch is merged. Even when > it will be merged, not all linux kernels will contain the patch. You should Cc stable to make sure they all gain it eventually. > Disabling msi is a workaround for the above case. > > Thanks, > Marcel Really enabling MSI without bus master is a bug that I'm not 100% sure it even worth working around. But I guess it's not too bad to have a work-around given it's this simple. -- MST