Device Model Parameters and Launch Script¶

Hypervisor Device Model (DM) is a QEMU-like application in the Service VM responsible for creating a User VM and then performing devices emulation based on command line configurations, as introduced in the Device Model High-Level Design. ACRN Configurator generates launch scripts for Post-launched VMs that include a call to the acrn-dm command with parameter values that were set in the configurator. Generally, you should not edit these launch scripts and change the parameters manually. Any edits you make would be overwritten if you run the configurator again and save the configuration and launch scripts.

The rest of this document provides details about the acrn-dm parameters as a reference, and should help you understand what the generated launch scripts are doing. We also include information useful to ACRN contributors about how setting in the scenario file, created by the configurator, are transformed into the launch script.

Device Model Parameters¶

Here are descriptions for each of these acrn-dm command line parameters:

-B, --bootargs <bootargs>

Set the User VM kernel command-line arguments. The maximum length is 1023. The bootargs string will be passed to the kernel as its cmdline.

Example:

-B "loglevel=7"

specifies the kernel log level at 7

--debugexit: Enable guest to write io port 0xf4 to exit guest. It’s mainly used by guest unit test.

-E, --elf_file <elf image path>: This option is to define a static elf binary which could be loaded by DM. DM will run elf as guest of ACRN.

--enable_trusty

Enable trusty for guest. For Android guest OS, ACRN provides a VM environment with two worlds: normal world and trusty world. The Android OS runs in the normal world. The trusty OS and security sensitive applications runs in the trusty world. The trusty world can see the memory of normal world but not vice versa. See Trusty TEE for more information.

By default, the trusty world is disabled. Use this option to enable it.

-h, --help: Show a summary of commands.

--intr_monitor <intr_monitor_params>

Enable interrupt storm monitor for User VM. Use this option to prevent an interrupt storm from the User VM.

usage: --intr_monitor threshold/s probe-period(s) delay_time(ms) delay_duration(ms)

Example:

--intr_monitor 10000,10,1,100

10000: interrupt rate larger than 10000/s will be treated as interrupt storm
10: use the last 10s of interrupt data to detect an interrupt storm
1: when interrupts are identified as a storm, the next interrupt will be delayed 1ms before being injected to the guest
100: after 100ms, we will cancel the interrupt injection delay and restore to normal.

-k, --kernel <kernel_image_path>

Set the kernel (full path) for the User VM kernel. The maximum path length is 1023 characters. The DM handles bzImage image format.

usage: -k /path/to/your/kernel_image

-l, --lpc <lpc_device_configuration>

-m, --memsize <memory_size>

Setup total memory size for User VM.

memory_size format is: “<size>{K/k, B/b, M/m, G/g}”, and size is an integer.

usage: -m 4g: set User VM memory to 4 gigabytes.

-r, --ramdisk <ramdisk_image_path>

Set the ramdisk (full path) for the User VM. The maximum length is 1023. The supported ramdisk format depends on your User VM kernel configuration.

usage: -r /path/to/your/ramdisk_image

-s, --pci_slot <slot_config>

Setup PCI device configuration.

slot_config format is:

<bus>:<slot>:<func>,<emul>[,<config>]
<slot>[:<func>],<emul>[,<config>]

Where:

slot is 0..31
func is 0..7
emul is a string describing the type of PCI device, e.g. virtio-net
config is an optional device-dependent string, used for configuration.

Examples:

-s 7,xhci,1-2,2-2

This configuration means the virtual xHCI will appear in PCI slot 7 in User VM. Any physical USB device attached on 1-2 (bus 1, port 2) or 2-2 (bus 2, port 2) will be detected by User VM and be used as expected. To determine which bus and port a USB device is attached, you could run lsusb -t in Service VM.

-s 9,virtio-blk,/root/test.img

This adds virtual block in PCI slot 9 and uses /root/test.img as the disk image.

For more information about emulated device types, see :ref:’emul_config’.

-v, --version: Show Device Model version.

--ovmf [w,]<ovmf_file_path> --ovmf [w,]code=<ovmf_code_file>,vars=<ovmf_vars_file>

Open Virtual Machine Firmware (OVMF) is an EDK II based project to enable UEFI support for Virtual Machines.

ACRN does not support off-the-shelf OVMF builds targeted for QEMU and KVM. Compatible OVMF images are included in the source tree, under devicemodel/bios/.

usage:

--ovmf /usr/share/acrn/bios/OVMF.fd

uses /usr/share/acrn/bios/OVMF.fd as the OVMF image

ACRN also supports using OVMF split images; OVMF_CODE.fd that contains the OVMF firmware executable and OVMF_VARS.fd that contains the NV data store.

usage:

--ovmf code=/usr/share/acrn/bios/OVMF_CODE.fd,vars=/usr/share/acrn/bios/OVMF_VARS.fd

ACRN supports the option “w” for OVMF. To preserve all changes in OVMF’s NV data store section, use this option to enable writeback mode.

Writeback mode is only enabled for the OVMF_VARS.fd file in case of OVMF split images, the firmware executable (OVMF_CODE.fd) remains read-only.

usage:

--ovmf w,/usr/share/acrn/bios/OVMF.fd

--cpu_affinity <list of lapic_ids>

comma-separated list of vCPUs assigned to this VM. Each CPU has a Local Programmable Interrupt Controller (LAPIC). The unique ID of the LAPIC (lapic_id) is used to identify vCPU. The lapic_id for a vCPU can be found in the service VM file /proc/cpuinfo identified as apicid.

Example:

--cpu_affinity 1,3

to assign vCPUs with lapic_id 1 and 3 to this VM.

--virtio_poll <poll_interval>

Enable virtio poll mode with poll interval xxx ns.

Example:

--virtio_poll 1000000

enable virtio poll mode with poll interval 1ms.

--acpidev_pt <HID>[,<UID>]

This option is to enable ACPI device passthrough support. The HID is a mandatory parameter for this option which is the Hardware ID of the ACPI device.

The UID is an option and used to specify a particular instance of the HID device, the default is 00.

Example:

--acpidev_pt MSFT0101,00

To pass through a TPM (which HID is MSFT0101 and UID is 00) ACPI device to a User VM.

--mmiodev_pt <MMIO_Region>

This option is to enable MMIO device passthrough support. The MMIO_Region is a mandatory parameter for this option which is the MMIO resource of the MMIO device. The MMIO_Region needs to be the base address followed by the length of the region, both separated by a comma.

Example:

--mmiodev_pt 0xFED40000,0x00005000

To pass through a MMIO device to a User VM. The MMIO device has a MMIO region. The base address of this region is 0xFED40000 and the size of the region is 0x00005000.

--vtpm2 <sock_path>: This option is to enable virtual TPM support. The sock_path is a mandatory parameter for this option which is the path of swtpm socket fd.

--virtio_msi: This option forces virtio to use single-vector MSI. By default, any virtio-based devices will use MSI-X as its interrupt method. If you want to use single-vector MSI interrupt, you can do so using this option.

--lapic_pt

This option is to create a VM with the local APIC (LAPIC) passed-through. With this option, a VM is created with LAPIC_PASSTHROUGH and IO_COMPLETION_POLLING mode. This option is typically used for hard real-time scenarios.

By default, this option is not enabled.

--rtvm

This option is used to create a VM with real-time attributes. With this option, a VM is created with GUEST_FLAG_RT and GUEST_FLAG_IO_COMPLETION_POLLING mode. This kind of VM is generally used for soft real-time scenarios (without --lapic_pt) or hard real-time scenarios (with --lapic_pt). With GUEST_FLAG_RT, the Service VM cannot interfere with this kind of VM when it is running. It can only be powered off from inside the VM itself.

By default, this option is not enabled.

--logger_setting <console,level=4;disk,level=4;kmsg,level=3>

This option sets the level of logging that is used for each log channel. The general format of this option is <log channel>,level=<log level>. Different log channels are separated by a semi-colon (;). The various log channels available are: console, disk and kmsg. The log level ranges from 1 (error) up to 5 (debug).

By default, the log severity level is set to 4 (info).

--windows

This option is used to run Windows User VMs. It supports Oracle virtio-blk, virtio-net and virtio-input devices for Windows guests with secure boot.

usage:

--windows

Note

This option is mandatory for running Windows in a User VM. If it is not used, Windows will not recognize the virtual disk.

--ssram

This option enables Software SRAM passthrough to the VM.

usage:

--ssram

Emulated PCI Device Types¶

In the acrn-dm -s or --pci_slot command line parameter, there is a <slot_config> argument that contains a string describing the type of emulated PCI device, along with optional device-dependent arguments used for configuration. Here is a table describing these emulated device types and arguments:

Table 1 Emulated PCI Device Types¶
PCI Device Type string	Description
`xhci`	USB controller used to support USB 3.0 devices, (also supports USB 2.0 and USB 1.0 devices). Parameter `<bus number>-<port number>` should be added. The physical USB devices attached on the specified bus and port will be detected by User VM and used as expected, e.g., `xhci,1-2,2-2`.
`lpc`	Low Pin Count (LPC) bus is used to connect low speed devices to the CPU, for example a serial port, keyboard, or mouse.
`igd-lpc`	Windows graphic driver requires this virtualized LPC device to operate the display function.
`ivshmem`	Inter-VM shared memory (ivshmem) virtualized PCI device used specifically for shared memory between VMs. Parameters should be added with the format `ivshmem,<shm_name>,<shm_size>`. `<shm-name>` specifies a shared memory name, and must be listed in `hv.FEATURES.IVSHMEM.IVSHMEM_REGION` as configured using the ACRN Configurator UI, and needs to start with a `dm:/` prefix.
`ahci`	Advanced Host Controller Interface provides advanced features to access Serial ATA (SATA) storage devices, such as a hard disk. Parameter `<type:><filepath>*` should be added: `type` could be `hd` (harddisk) or `cd` (CD-ROM). `<filepath>` is the path for the backend file and could be a partition name or a regular file, e.g., `ahci,hd:/dev/sda`.
`ahci-hd`	This is an alias for `ahci`.
`ahci-cd`	Advanced Host Controller Interface used to connect with AT Attachment Packet Interface device (for CD-ROM emulation). `ahci-cd` supports the same parameters than `ahci`.
`hostbridge`	Virtualized PCI hostbridge, a hardware bridge between the CPU’s high-speed system local bus and the Peripheral Component Interconnect (PCI) bus.
`virtio-blk`	Virtio block type device, a string could be appended with the format `virtio-blk,<filepath>[,options]` `<filepath>` specifies the path of a file or disk partition. You can also could use `nodisk` to create a virtio-blk device with a dummy backend. `nodisk` is used for hot-plugging a rootfs after the User VM has been launched. It is achieved by triggering a rescan of the `virtio-blk` device by the User VM. The empty file will be updated to valid file after rescan. `[,options]` includes: `writethru`: write operation is reported completed only when the data has been written to physical storage. `writeback`: write operation is reported completed when data is placed in the page cache. Needs to be flushed to the physical storage. `ro`: open file with readonly mode. `sectorsize`: configured as either `sectorsize=<sector size>/<physical sector size>` or `sectorsize=<sector size>`. The default values for sector size and physical sector size are 512. `range`: configured as `range=<start lba in file>/<sub file size>` meaning the virtio-blk will only access part of the file, from the `<start lba in file>` to `<start lba in file>` + `<sub file site>`.
`virtio-input`	Virtio type device to emulate input device. `evdev` char device node should be appended, e.g., `-s n,virtio-input,/dev/input/eventX[,serial]`. `serial` is an optional string used as the unique identification code of guest virtio input device.
`virtio-console`	Virtio console type device for data input and output.
`virtio-heci`	Virtio Host Embedded Controller Interface, parameters should be appended with the format `<bus>:<device>:<function>,d<0~8>`. You can find the BDF information from the Service VM.
`virtio-i2c`	Virtio i2c type device,parameters with format: `<bus>[:<client_addr>[@<node>]][,<bus>[:<client_addr>[@<node>]]` `<bus>` specifies the bus number for the native I2C adapter, e.g., `2` means `/dev/i2c-2`. `<client_addr>` specifies the address for the native client devices such as `1C` or `2F`. `@` specifies the prefix for the ACPI node. `<node>` specifies the ACPI node name supported in the `acpi_node_table[]` in the source code: only `cam1`, `cam2`, and `hdac` are supported for APL platform and are platform-specific.
`virtio-gpio`	Virtio GPIO type device, parameters format is: `virtio-gpio,<@controller_name{offset\|name[=mapping_name]:offset\|name[=mapping_name]:...}@controller_name{...}...]>` `controller_name`: use the command `ls /sys/bus/gpio/devices` to check the native GPIO controller information. Usually, the devices represent the `controller_name` that you can use. You can also use the command `cat /sys/bus/gpio/device/XXX/dev` to get the device id that can be used to match `/dev/XXX`, and then use `XXX` as the `controller_name`. On Intel platforms, `controller_name` may be `gpiochip0`, `gpiochip1`, `gpiochip2`, and `gpiochip3`. `offset\|name`: use GPIO offset or its name to locate one native GPIO within the GPIO controller. `mapping_name`: is optional. If you want to use a customized name for a FE GPIO, you can set a new name here.
`virtio-rnd`	Virtio random generator type device, the VBSU virtio backend is used by default.
`virtio-rpmb`	Virtio Replay Protected Memory Block (RPMB) type device, with `physical_rpmb` to specify RPMB in physical mode, otherwise RPMB is in simulated mode.
`virtio-net`	Virtio network type device, parameter should be appended with the format: `virtio-net,<device_type>=<name>[,vhost][,mac=<XX:XX:XX:XX:XX:XX> \| mac_seed=<seed_string>]`. The only supported `device_type` parameter is `tap`. The `mac` address is optional and `name` is the name of the TAP (or MacVTap) device. `vhost` specifies vhost backend, otherwise the VBSU backend is used. `mac_seed=<seed_string>` sets a platform-unique string as a seed to generate the MAC address. Each VM should have a different `seed_string`. The `seed_string` can be generated by the following method where `$(vm_name)` contains the name of the VM you are going to launch. mac=$(cat /sys/class/net/e*/address) seed_string=${mac:9:8}-${vm_name} Note `mac` and `mac_seed` are mutually exclusive, when both are set the latter is ignored and the MAC address is set to the `mac` value. `mac_seed` will only be used when `mac` is not set.
`virtio-gpu`	Virtio GPU type device, parameters format is: `virtio-gpu[,geometry=<width>x<height>+<x_off>+<y_off> \| fullscreen]` `geometry` specifies the mode of virtual display, windowed or fullscreen. If it is not set, the virtual display will use 1280x720 resolution in windowed mode. `width` specifies the width of the virtual display window in pixels. `height` specifies the height of the virtual display window in pixels. `x_off` specifies the x offset of the virtual display window from the upper-left corner of the screen. `y_off` specifies the y offset of the virtual display window from the upper-left corner of the screen. For example: `geometry=1280x720+100+50` specifies a window 1280 pixels wide by 720 high, with the top left corner 100 pixels right and 50 pixels down from the top left corner of the screen.
`passthru`	Indicates a passthrough device. Use the parameter with the format `passthru,<bus>/<device>/<function>,<optional parameter>` Optional parameters include: `keep_gsi`: keep vGSI for MSI capable passthrough device. `no_reset`: passthrough PCI devices are reset by default when assigning them to a post-launched VM. This parameter prevents this reset for debugging purposes. `d3hot_reset`: when launching a Windows post-launched VM, this parameter should be appended to enable a Windows UEFI ACPI bug fix. `gpu`: create the dedicated `igd-lpc` on `00:1f.0` for IGD passthrough. `vmsix_on_msi,<bar_id>`: enables vMSI-X emulation based on MSI capability. The specific virtual bar will be allocated. `enable_ptm`: enable PCIE precise time measurement mechanism for the passthrough device.
`uart`	Emulated PCI UART. Use the parameter with the format `uart,vuart_idx:<0~9>` to specify hypervisor-emulated PCI vUART index.
`wdt-i6300esb`	Emulated i6300ESB PCI Watch Dog Timer (WDT) Intel processors use to monitor User VMs.

Launch Script¶

A launch script is used to start a User VM from the Service VM command line. It is generated by the ACRN configurator according to several settings for a User VM. Normally, you should not manually edit these generated launch scripts or change acrn-dm command line parameters. If you do so, your changes could be overwritten the next time you run the configurator.

In this section we describe how setting in the scenario file, created by the configurator, are transformed into the launch script. This information would be useful to ACRN contributors or developers interested in knowing how the launch scripts are created.

Most configurator settings for User VMs are used at launch time. When you exit the configurator, these settings are saved in the scenario.xml file and then processed by misc/config_tools/launch_config/launch_cfg_gen.py to add shell commands to create the launch script, according to the template misc/config_tools/launch_config/launch_script_template.sh. The template uses following helper functions to do system settings or to generate an acrn-dm command line parameter. For details about all acrn-dm parameters, refer to the previous section.

probe_modules

Install necessary modules before launching a Post-launched VM. For example, pci_stub is used to provide a stub pci driver that does nothing on attached pci devices. Passthrough PCIe devices will be unbound from their original driver and bound to the stub, so that they can be safely controlled by the User VM.

offline_cpus <cpu_apicid>...

This is called if we are launching an RTVM or VM whose scheduler is “SCHED_NOOP”. In both situations, CPU sharing between multiple VMs is prevented. This function will trigger taking a CPU offline (done by the Service VM kernel), and then inform the hypervisor through Hypervisor Service Module (HSM). The hypervisor will offline vCPU and freeze vCPU thread.

unbind_device <bdf>

Unbind a PCIe device with specified BDF (bus, device and function) number from its original driver and re-bind it to the pci-stub driver. After that the Service VM kernel will not operate on that device any more and it can be passed through to User VM safely.

create_tap <tap>

Create or reuse the tap interface that is attached to acrn-br0 bridge. acrn-br0 is registered to systemd-networkd.service after installing the ACRN Debian package (.deb). You also need to enable and start the service to create the bridge from the Service VM using:

sudo systemctl enable --now systemd-networkd

The bridge is used to add virtio-net interface to a User VM. virtio-net interfaces for all User VMs are virtually connected to a subnet behind the ACRN bridge.

mount_partition <partition>

Mount specified partition to a temporary directory created by mktemp -d, and return the temporary directory for later unmount. Typically this function is called to mount an image file in order to use inner rootfs file as virtio-blk backend. For example, user could set “<imgfile>:/boot/initrd.img*” in the virtio-blk input box in the ACRN Configurator. After the acrn-dm instance exits, unmount_partition will be called to unmount the image file.

unmount_partition <dir>

Unmount partition from specified directory.

add_cpus <cpu_apicid>...

Return an acrn-dm command line parameter fragment to set cpu_affinity. Refer to cpu_affinity for details. offline_cpus is called if the User VM is an RTVM or its scheduler is SCHED_NOOP.

add_interrupt_storm_monitor <threshold_per_sec> <probe_period_in_sec> <inject_delay_in_ms> <delay_duration_in_ms>

This is added if PCIe devices, other than an integrated GPU, are passed through to the User VM to monitor if interrupt storm occurred on those devices. This function and parameters are not visible in the ACRN Configurator and handled by config scripts. It returns acrn-dm command line segment to set intr_monitor.

add_logger_settings console=<n> kmsg=<n> disk=<n>

set log level of each acrn-dm logging channel: console, kmsg, disk. These settings are not exposed to user in the ACRN Configurator.

add_virtual_device <slot> <kind> <options>

Add specified kind of virtual device to the specified PCIe device slot. Some devices need options to configure further behaviors. <slot> numbers for virtual devices and passthrough devices are automatically allocated by launch_cfg_gen.py.

Typical use cases:

hostbridge
PCIe host bridge. <slot> must be 0.
uart vuart_idx:<int>
Add a PCIe vuart with specified index.
xhci <bus>-<port>[:<bus>-<port>]...
Config USB mediator. A list of USB ports each specified by <bus>-<port> will be connected to the User VM.
virtio-net tap=<tapname>[,vhost],mac_seed=<str>
The TAP should already be created by create_tap.
virtio-blk <imgfile>[,writethru|writeback|ro]
Add a virtio block device to User VM. The backend is a raw image file. Options can be specified to control access right.

For all types of virtual devices and options, refer to Emulated PCI Device Types.

add_passthrough_device <slot> <bus>/<device>/<function> <options>

Passthrough PCIe device to User VM in specified <slot>. Some kinds of devices may need extra <options> to control internal behavior. Refer to the passthru section in Emulated PCI Device Types.

These functions in the template are copied to the target launch script. Then launch_cfg_gen.py generates the following dynamic part. It first defines necessary variables such as vm_type and scheduler, and uses the functions described above to construct the dm_params parameters per the user settings in scenario.xml. Finally, acrn-dm is executed to launch a User VM with these parameters.