QEMU Arm how to passthrough a PCI Card?

Question

I'm trying to access a PCI WiFi adapter on a QEMU, simulating vexpress-a9 board.

I'm following is this tutorial which shares a passthrough PCI card of the host machine to a guest machine running on qemu-kvm. But I think its not applicable for qemu-system-arm.

• Is it possible to passthrough PCI using qemu-system-arm for vexpress-a9?
• Is there any way to access a PCI card as a USB device in the guest OS? or a host USB device in the guest OS?

Here's what I'm doing:

$ qemu-system-arm \
-kernel ./zImage \
-M vexpress-a9 \
-cpu cortex-a9 \
-m 1024 \
-smp 2 \
-serial stdio \
-append "root=/dev/mmcblk0p1 rootfstype=ext4 rw raid=noautodetect console=ttyAMA0,38400n8 rootwait devtmpfs.mount=0 vmalloc=256MB mem=1024M" \
-sd ./vexpress-quantal.img

Solution

Make sure you go through the steps outlined in this section, 13.3.4. PCI Pass-Through, of the docs, titled: Chapter 13. Running Virtual Machines with qemu-kvm.

excerpt

Procedure 13.1. Configuring PCI Pass-Through

1. Make sure that CONFIG_DMAR_DEFAULT_ON is set in the host's running Kernel:

   $ grep CONFIG_DMAR_DEFAULT_ON /boot/config-`uname -r`
   

   If this option is not set, edit your boot loader configuration and add intel_iommu=on (Intel machines) or iommu=pt iommu=1 (AMD machines). Then reboot the host machine.

2. Check that IOMMU is actively enabled and recognized on the host. Run dmesg | grep -e DMAR -e IOMMU on Intel machines, or dmesg | grep AMD-Vi on AMD machines. If you get no output, check carefully if your hardware supports IOMMU (VT-d) and check that it has been enabled in BIOS.

3. Identify the host PCI device to assign to the guest.

   tux@vmhost:~> lspci -nn
   [...] 00:1b.0 Audio device [0403]: Intel Corporation 82801H (ICH8 Family) \
   HD Audio Controller [8086:284b] (rev 02) [...]
   Note down the device (00:1b.0) and vendor (8086:284b) ID.
   

4. Unbind the device from host Kernel driver and bind it to the PCI stub driver.

   tux@vmhost:~> modprobe pci_stub
   tux@vmhost:~> echo "8086 284b" > /sys/bus/pci/drivers/pci-stub/new_id
   tux@vmhost:~> echo "0000:00:1b.0" > /sys/bus/pci/devices/0000:00:1b.0/driver/unbind
   tux@vmhost:~> echo "0000:00:1b.0" > /sys/bus/pci/drivers/pci-stub/bind
   

5. Now run the VM Guest with the PCI device assigned.

   qemu-kvm [...] -device pci-assign,host=00:1b.0
   

QEMU/KVM ARM support

According to the changelog for version 1.5 & 1.6 ARM is supported in the following ways:

1.5

• This release has initial support of KVM for the ARM architecture. This requires a 3.9 or better Linux kernel and a Cortex-A15 CPU.
• The Zynq board provides a SD host controller interface.
• The PCI controller model for VersatilePB and Realview boards has been significantly improved in functionality (including MMIO BAR support and a correct PCI IRQ mapping). Note that Linux kernels currently do not take advantage of this; however we autodetect kernels that expect the behaviour of an old broken QEMU and fall back to that. (The old IRQ behaviour can be forced with "-global versatile_pci.broken-irq-mapping=1" if absolutely necessary.)
• Incorrect handling of the SRS instruction in Thumb mode has been fixed.
• Performance of TCG emulation of ARM targets is improved over previous releases.
• Migration and vm save/load now works correctly on the vexpress-a15 and vexpress-a9 models.

1.6

• The 32-bit ARMv8 LDA/STL instructions for load-acquire/store-release are supported (only with -cpu any); the remainder of the new v8 instructions will follow in a future release.
• The vexpress-a9 and vexpress-a15 boards have experimental support for virtio devices via a virtio-mmio transport. Note that the virtio command lines used with x86 systems won't work as they will create PCI virtio devices; use the devices "virtio-blk-device", "virtio-net-device", etc instead. Note that this functionality may change in future releases.
• The -initrd option now accepts ramdisks with a U-Boot header.
• A model of the Calxeda ECX-2000 / Midway system is now supported ("midway").