From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:58636) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1UsCBm-0004px-Cw for qemu-devel@nongnu.org; Thu, 27 Jun 2013 09:24:50 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1UsCBf-0008Ct-9o for qemu-devel@nongnu.org; Thu, 27 Jun 2013 09:24:42 -0400 Received: from 1.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.d.1.0.0.b.8.0.1.0.0.2.ip6.arpa ([2001:8b0:1d0::1]:58278 helo=mnementh.archaic.org.uk) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1UsCBf-0008CT-37 for qemu-devel@nongnu.org; Thu, 27 Jun 2013 09:24:35 -0400 From: Peter Maydell Date: Thu, 27 Jun 2013 14:04:07 +0100 Message-Id: <1372338255-3934-1-git-send-email-peter.maydell@linaro.org> Subject: [Qemu-devel] [PATCH 0/8] Add virtio-mmio and use it in vexpress List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: qemu-devel@nongnu.org Cc: Anthony Liguori , patches@linaro.org, Alexander Graf , David Gibson , kvmarm@lists.cs.columbia.edu, KONRAD Frederic This patch series adds an implementation of the virtio-mmio transport, and uses it in the vexpress-a9 and vexpress-a15 board models. The basic idea is that the board instantiates some transports, the user can create backends which automatically plug into them (via -device virtio-blk-backend and the like), and we tell the guest kernel about them using a new arm/boot hook that lets the board modify the user's device tree blob. (The last part of this represents another reluctant step into manipulating device trees; however interrupt-controller and virtio nodes should be sufficiently stable to be safe as these things go.) Is there a convenient way to say "don't create the transport if there's no backend to plug into it"? Sample command line: ./build/x86/arm-softmmu/qemu-system-arm -machine type=vexpress-a15 \ -serial stdio -display none -kernel zImage \ -dtb vexpress-v2p-ca15-tc1.dtb \ -append "root=/dev/mmcblk0 rw console=ttyAMA0 rootwait earlyprintk" \ -sd arm-wheezy.img -device virtio-serial-device \ -drive if=none,file=debian_lenny_arm_small.qcow2,id=foo \ -device virtio-blk-device,drive=foo (note that '-device virtio-blk' and '-drive if=virtio' won't work since they will result in the creation of PCI virtio devices.) The first two patches here are effectively a 'v2' of the versions I sent out earlier this week; I've updated them as per review so they no longer have the awkward "varargs don't promote" problem. The core virtio patches are based initially on the prototype versions I wrote back in 2011, which Fred Konrad has subsequently updated to work on top of his virtio refactoring. I then did some minor fixes and cleanup. When I get mach-virt into shape that will use virtio-mmio too. Peter Maydell (8): device_tree: Add qemu_devtree_setprop_sized_cells() utility functions arm/boot: Use qemu_devtree_setprop_sized_cells() virtio: Add support for guest setting of queue size virtio: Support transports which can specify the vring alignment virtio: Implement MMIO based virtio transport arm/boot: Allow boards to modify the FDT blob vexpress: Make VEDBoardInfo extend arm_boot_info vexpress: Add virtio-mmio transports device_tree.c | 31 +++ hw/arm/boot.c | 33 ++-- hw/arm/vexpress.c | 128 ++++++++++-- hw/virtio/Makefile.objs | 1 + hw/virtio/virtio-mmio.c | 424 ++++++++++++++++++++++++++++++++++++++++ hw/virtio/virtio.c | 38 +++- include/hw/arm/arm.h | 4 + include/hw/virtio/virtio-bus.h | 6 + include/hw/virtio/virtio.h | 2 + include/sysemu/device_tree.h | 62 ++++++ 10 files changed, 690 insertions(+), 39 deletions(-) create mode 100644 hw/virtio/virtio-mmio.c -- 1.7.9.5