xref: /openbmc/qemu/docs/system/arm/virt.rst (revision 76bccf3c)
1'virt' generic virtual platform (``virt``)
2==========================================
3
4The ``virt`` board is a platform which does not correspond to any
5real hardware; it is designed for use in virtual machines.
6It is the recommended board type if you simply want to run
7a guest such as Linux and do not care about reproducing the
8idiosyncrasies and limitations of a particular bit of real-world
9hardware.
10
11This is a "versioned" board model, so as well as the ``virt`` machine
12type itself (which may have improvements, bugfixes and other minor
13changes between QEMU versions) a version is provided that guarantees
14to have the same behaviour as that of previous QEMU releases, so
15that VM migration will work between QEMU versions. For instance the
16``virt-5.0`` machine type will behave like the ``virt`` machine from
17the QEMU 5.0 release, and migration should work between ``virt-5.0``
18of the 5.0 release and ``virt-5.0`` of the 5.1 release. Migration
19is not guaranteed to work between different QEMU releases for
20the non-versioned ``virt`` machine type.
21
22Supported devices
23"""""""""""""""""
24
25The virt board supports:
26
27- PCI/PCIe devices
28- Flash memory
29- Either one or two PL011 UARTs for the NonSecure World
30- An RTC
31- The fw_cfg device that allows a guest to obtain data from QEMU
32- A PL061 GPIO controller
33- An optional SMMUv3 IOMMU
34- hotpluggable DIMMs
35- hotpluggable NVDIMMs
36- An MSI controller (GICv2M or ITS). GICv2M is selected by default along
37  with GICv2. ITS is selected by default with GICv3 (>= virt-2.7). Note
38  that ITS is not modeled in TCG mode.
39- 32 virtio-mmio transport devices
40- running guests using the KVM accelerator on aarch64 hardware
41- large amounts of RAM (at least 255GB, and more if using highmem)
42- many CPUs (up to 512 if using a GICv3 and highmem)
43- Secure-World-only devices if the CPU has TrustZone:
44
45  - A second PL011 UART
46  - A second PL061 GPIO controller, with GPIO lines for triggering
47    a system reset or system poweroff
48  - A secure flash memory
49  - 16MB of secure RAM
50
51The second NonSecure UART only exists if a backend is configured
52explicitly (e.g. with a second -serial command line option) and
53TrustZone emulation is not enabled.
54
55Supported guest CPU types:
56
57- ``cortex-a7`` (32-bit)
58- ``cortex-a15`` (32-bit; the default)
59- ``cortex-a35`` (64-bit)
60- ``cortex-a53`` (64-bit)
61- ``cortex-a55`` (64-bit)
62- ``cortex-a57`` (64-bit)
63- ``cortex-a72`` (64-bit)
64- ``cortex-a76`` (64-bit)
65- ``cortex-a710`` (64-bit)
66- ``a64fx`` (64-bit)
67- ``host`` (with KVM only)
68- ``neoverse-n1`` (64-bit)
69- ``neoverse-v1`` (64-bit)
70- ``neoverse-n2`` (64-bit)
71- ``max`` (same as ``host`` for KVM; best possible emulation with TCG)
72
73Note that the default is ``cortex-a15``, so for an AArch64 guest you must
74specify a CPU type.
75
76Also, please note that passing ``max`` CPU (i.e. ``-cpu max``) won't
77enable all the CPU features for a given ``virt`` machine. Where a CPU
78architectural feature requires support in both the CPU itself and in the
79wider system (e.g. the MTE feature), it may not be enabled by default,
80but instead requires a machine option to enable it.
81
82For example, MTE support must be enabled with ``-machine virt,mte=on``,
83as well as by selecting an MTE-capable CPU (e.g., ``max``) with the
84``-cpu`` option.
85
86See the machine-specific options below, or check them for a given machine
87by passing the ``help`` suboption, like: ``-machine virt-9.0,help``.
88
89Graphics output is available, but unlike the x86 PC machine types
90there is no default display device enabled: you should select one from
91the Display devices section of "-device help". The recommended option
92is ``virtio-gpu-pci``; this is the only one which will work correctly
93with KVM. You may also need to ensure your guest kernel is configured
94with support for this; see below.
95
96Machine-specific options
97""""""""""""""""""""""""
98
99The following machine-specific options are supported:
100
101secure
102  Set ``on``/``off`` to enable/disable emulating a guest CPU which implements the
103  Arm Security Extensions (TrustZone). The default is ``off``.
104
105virtualization
106  Set ``on``/``off`` to enable/disable emulating a guest CPU which implements the
107  Arm Virtualization Extensions. The default is ``off``.
108
109mte
110  Set ``on``/``off`` to enable/disable emulating a guest CPU which implements the
111  Arm Memory Tagging Extensions. The default is ``off``.
112
113highmem
114  Set ``on``/``off`` to enable/disable placing devices and RAM in physical
115  address space above 32 bits. The default is ``on`` for machine types
116  later than ``virt-2.12`` when the CPU supports an address space
117  bigger than 32 bits (i.e. 64-bit CPUs, and 32-bit CPUs with the
118  Large Physical Address Extension (LPAE) feature). If you want to
119  boot a 32-bit kernel which does not have ``CONFIG_LPAE`` enabled on
120  a CPU type which implements LPAE, you will need to manually set
121  this to ``off``; otherwise some devices, such as the PCI controller,
122  will not be accessible.
123
124compact-highmem
125  Set ``on``/``off`` to enable/disable the compact layout for high memory regions.
126  The default is ``on`` for machine types later than ``virt-7.2``.
127
128highmem-redists
129  Set ``on``/``off`` to enable/disable the high memory region for GICv3 or
130  GICv4 redistributor. The default is ``on``. Setting this to ``off`` will
131  limit the maximum number of CPUs when GICv3 or GICv4 is used.
132
133highmem-ecam
134  Set ``on``/``off`` to enable/disable the high memory region for PCI ECAM.
135  The default is ``on`` for machine types later than ``virt-3.0``.
136
137highmem-mmio
138  Set ``on``/``off`` to enable/disable the high memory region for PCI MMIO.
139  The default is ``on``.
140
141gic-version
142  Specify the version of the Generic Interrupt Controller (GIC) to provide.
143  Valid values are:
144
145  ``2``
146    GICv2. Note that this limits the number of CPUs to 8.
147  ``3``
148    GICv3. This allows up to 512 CPUs.
149  ``4``
150    GICv4. Requires ``virtualization`` to be ``on``; allows up to 317 CPUs.
151  ``host``
152    Use the same GIC version the host provides, when using KVM
153  ``max``
154    Use the best GIC version possible (same as host when using KVM;
155    with TCG this is currently ``3`` if ``virtualization`` is ``off`` and
156    ``4`` if ``virtualization`` is ``on``, but this may change in future)
157
158its
159  Set ``on``/``off`` to enable/disable ITS instantiation. The default is ``on``
160  for machine types later than ``virt-2.7``.
161
162iommu
163  Set the IOMMU type to create for the guest. Valid values are:
164
165  ``none``
166    Don't create an IOMMU (the default)
167  ``smmuv3``
168    Create an SMMUv3
169
170ras
171  Set ``on``/``off`` to enable/disable reporting host memory errors to a guest
172  using ACPI and guest external abort exceptions. The default is off.
173
174dtb-randomness
175  Set ``on``/``off`` to pass random seeds via the guest DTB
176  rng-seed and kaslr-seed nodes (in both "/chosen" and
177  "/secure-chosen") to use for features like the random number
178  generator and address space randomisation. The default is
179  ``on``. You will want to disable it if your trusted boot chain
180  will verify the DTB it is passed, since this option causes the
181  DTB to be non-deterministic. It would be the responsibility of
182  the firmware to come up with a seed and pass it on if it wants to.
183
184dtb-kaslr-seed
185  A deprecated synonym for dtb-randomness.
186
187Linux guest kernel configuration
188""""""""""""""""""""""""""""""""
189
190The 'defconfig' for Linux arm and arm64 kernels should include the
191right device drivers for virtio and the PCI controller; however some older
192kernel versions, especially for 32-bit Arm, did not have everything
193enabled by default. If you're not seeing PCI devices that you expect,
194then check that your guest config has::
195
196  CONFIG_PCI=y
197  CONFIG_VIRTIO_PCI=y
198  CONFIG_PCI_HOST_GENERIC=y
199
200If you want to use the ``virtio-gpu-pci`` graphics device you will also
201need::
202
203  CONFIG_DRM=y
204  CONFIG_DRM_VIRTIO_GPU=y
205
206Hardware configuration information for bare-metal programming
207"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""
208
209The ``virt`` board automatically generates a device tree blob ("dtb")
210which it passes to the guest. This provides information about the
211addresses, interrupt lines and other configuration of the various devices
212in the system. Guest code can rely on and hard-code the following
213addresses:
214
215- Flash memory starts at address 0x0000_0000
216
217- RAM starts at 0x4000_0000
218
219All other information about device locations may change between
220QEMU versions, so guest code must look in the DTB.
221
222QEMU supports two types of guest image boot for ``virt``, and
223the way for the guest code to locate the dtb binary differs:
224
225- For guests using the Linux kernel boot protocol (this means any
226  non-ELF file passed to the QEMU ``-kernel`` option) the address
227  of the DTB is passed in a register (``r2`` for 32-bit guests,
228  or ``x0`` for 64-bit guests)
229
230- For guests booting as "bare-metal" (any other kind of boot),
231  the DTB is at the start of RAM (0x4000_0000)
232