[Qemu-devel] [PATCH 0/8] Add virtio-mmio and use it in vexpress

[Peter Maydell]

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