16a0b7505SPeter Maydell'virt' generic virtual platform (``virt``) 26a0b7505SPeter Maydell========================================== 36a0b7505SPeter Maydell 46df743dcSPeter MaydellThe ``virt`` board is a platform which does not correspond to any 56a0b7505SPeter Maydellreal hardware; it is designed for use in virtual machines. 66a0b7505SPeter MaydellIt is the recommended board type if you simply want to run 76a0b7505SPeter Maydella guest such as Linux and do not care about reproducing the 86a0b7505SPeter Maydellidiosyncrasies and limitations of a particular bit of real-world 96a0b7505SPeter Maydellhardware. 106a0b7505SPeter Maydell 116a0b7505SPeter MaydellThis is a "versioned" board model, so as well as the ``virt`` machine 126a0b7505SPeter Maydelltype itself (which may have improvements, bugfixes and other minor 136a0b7505SPeter Maydellchanges between QEMU versions) a version is provided that guarantees 146a0b7505SPeter Maydellto have the same behaviour as that of previous QEMU releases, so 156a0b7505SPeter Maydellthat VM migration will work between QEMU versions. For instance the 166a0b7505SPeter Maydell``virt-5.0`` machine type will behave like the ``virt`` machine from 176a0b7505SPeter Maydellthe QEMU 5.0 release, and migration should work between ``virt-5.0`` 186a0b7505SPeter Maydellof the 5.0 release and ``virt-5.0`` of the 5.1 release. Migration 196a0b7505SPeter Maydellis not guaranteed to work between different QEMU releases for 206a0b7505SPeter Maydellthe non-versioned ``virt`` machine type. 216a0b7505SPeter Maydell 226a0b7505SPeter MaydellSupported devices 236a0b7505SPeter Maydell""""""""""""""""" 246a0b7505SPeter Maydell 256a0b7505SPeter MaydellThe virt board supports: 266a0b7505SPeter Maydell 276a0b7505SPeter Maydell- PCI/PCIe devices 286a0b7505SPeter Maydell- Flash memory 296a0b7505SPeter Maydell- One PL011 UART 306a0b7505SPeter Maydell- An RTC 316a0b7505SPeter Maydell- The fw_cfg device that allows a guest to obtain data from QEMU 326a0b7505SPeter Maydell- A PL061 GPIO controller 336a0b7505SPeter Maydell- An optional SMMUv3 IOMMU 346a0b7505SPeter Maydell- hotpluggable DIMMs 356a0b7505SPeter Maydell- hotpluggable NVDIMMs 366a0b7505SPeter Maydell- An MSI controller (GICv2M or ITS). GICv2M is selected by default along 376a0b7505SPeter Maydell with GICv2. ITS is selected by default with GICv3 (>= virt-2.7). Note 386a0b7505SPeter Maydell that ITS is not modeled in TCG mode. 396a0b7505SPeter Maydell- 32 virtio-mmio transport devices 406a0b7505SPeter Maydell- running guests using the KVM accelerator on aarch64 hardware 416a0b7505SPeter Maydell- large amounts of RAM (at least 255GB, and more if using highmem) 426a0b7505SPeter Maydell- many CPUs (up to 512 if using a GICv3 and highmem) 436a0b7505SPeter Maydell- Secure-World-only devices if the CPU has TrustZone: 446a0b7505SPeter Maydell 456a0b7505SPeter Maydell - A second PL011 UART 46daa726d9SMaxim Uvarov - A second PL061 GPIO controller, with GPIO lines for triggering 47daa726d9SMaxim Uvarov a system reset or system poweroff 486a0b7505SPeter Maydell - A secure flash memory 496a0b7505SPeter Maydell - 16MB of secure RAM 506a0b7505SPeter Maydell 516a0b7505SPeter MaydellSupported guest CPU types: 526a0b7505SPeter Maydell 536a0b7505SPeter Maydell- ``cortex-a7`` (32-bit) 546a0b7505SPeter Maydell- ``cortex-a15`` (32-bit; the default) 553b16766bSHao Wu- ``cortex-a35`` (64-bit) 566a0b7505SPeter Maydell- ``cortex-a53`` (64-bit) 5794bc3b06STimofey Kutergin- ``cortex-a55`` (64-bit) 586a0b7505SPeter Maydell- ``cortex-a57`` (64-bit) 596a0b7505SPeter Maydell- ``cortex-a72`` (64-bit) 602f6283fcSRichard Henderson- ``cortex-a76`` (64-bit) 61e3d45c0aSRichard Henderson- ``cortex-a710`` (64-bit) 624d39fcd8SShuuichirou Ishii- ``a64fx`` (64-bit) 636a0b7505SPeter Maydell- ``host`` (with KVM only) 645db6de80SRichard Henderson- ``neoverse-n1`` (64-bit) 65c74138c6SPeter Maydell- ``neoverse-v1`` (64-bit) 66dfff1000SPeter Maydell- ``neoverse-n2`` (64-bit) 676a0b7505SPeter Maydell- ``max`` (same as ``host`` for KVM; best possible emulation with TCG) 686a0b7505SPeter Maydell 696a0b7505SPeter MaydellNote that the default is ``cortex-a15``, so for an AArch64 guest you must 706a0b7505SPeter Maydellspecify a CPU type. 716a0b7505SPeter Maydell 721acf2159SGustavo RomeroAlso, please note that passing ``max`` CPU (i.e. ``-cpu max``) won't 731acf2159SGustavo Romeroenable all the CPU features for a given ``virt`` machine. Where a CPU 741acf2159SGustavo Romeroarchitectural feature requires support in both the CPU itself and in the 751acf2159SGustavo Romerowider system (e.g. the MTE feature), it may not be enabled by default, 761acf2159SGustavo Romerobut instead requires a machine option to enable it. 771acf2159SGustavo Romero 781acf2159SGustavo RomeroFor example, MTE support must be enabled with ``-machine virt,mte=on``, 791acf2159SGustavo Romeroas well as by selecting an MTE-capable CPU (e.g., ``max``) with the 801acf2159SGustavo Romero``-cpu`` option. 811acf2159SGustavo Romero 821acf2159SGustavo RomeroSee the machine-specific options below, or check them for a given machine 831acf2159SGustavo Romeroby passing the ``help`` suboption, like: ``-machine virt-9.0,help``. 841acf2159SGustavo Romero 856a0b7505SPeter MaydellGraphics output is available, but unlike the x86 PC machine types 866a0b7505SPeter Maydellthere is no default display device enabled: you should select one from 876a0b7505SPeter Maydellthe Display devices section of "-device help". The recommended option 886a0b7505SPeter Maydellis ``virtio-gpu-pci``; this is the only one which will work correctly 896a0b7505SPeter Maydellwith KVM. You may also need to ensure your guest kernel is configured 906a0b7505SPeter Maydellwith support for this; see below. 916a0b7505SPeter Maydell 926a0b7505SPeter MaydellMachine-specific options 936a0b7505SPeter Maydell"""""""""""""""""""""""" 946a0b7505SPeter Maydell 956a0b7505SPeter MaydellThe following machine-specific options are supported: 966a0b7505SPeter Maydell 976a0b7505SPeter Maydellsecure 986a0b7505SPeter Maydell Set ``on``/``off`` to enable/disable emulating a guest CPU which implements the 996a0b7505SPeter Maydell Arm Security Extensions (TrustZone). The default is ``off``. 1006a0b7505SPeter Maydell 1016a0b7505SPeter Maydellvirtualization 1026a0b7505SPeter Maydell Set ``on``/``off`` to enable/disable emulating a guest CPU which implements the 1036a0b7505SPeter Maydell Arm Virtualization Extensions. The default is ``off``. 1046a0b7505SPeter Maydell 105222f45b7SPeter Maydellmte 106222f45b7SPeter Maydell Set ``on``/``off`` to enable/disable emulating a guest CPU which implements the 107222f45b7SPeter Maydell Arm Memory Tagging Extensions. The default is ``off``. 108222f45b7SPeter Maydell 1096a0b7505SPeter Maydellhighmem 1106a0b7505SPeter Maydell Set ``on``/``off`` to enable/disable placing devices and RAM in physical 1116a0b7505SPeter Maydell address space above 32 bits. The default is ``on`` for machine types 1128cf1da72SPeter Maydell later than ``virt-2.12`` when the CPU supports an address space 1138cf1da72SPeter Maydell bigger than 32 bits (i.e. 64-bit CPUs, and 32-bit CPUs with the 1148cf1da72SPeter Maydell Large Physical Address Extension (LPAE) feature). If you want to 1158cf1da72SPeter Maydell boot a 32-bit kernel which does not have ``CONFIG_LPAE`` enabled on 1168cf1da72SPeter Maydell a CPU type which implements LPAE, you will need to manually set 1178cf1da72SPeter Maydell this to ``off``; otherwise some devices, such as the PCI controller, 1188cf1da72SPeter Maydell will not be accessible. 1196a0b7505SPeter Maydell 120f40408a9SGavin Shancompact-highmem 121f40408a9SGavin Shan Set ``on``/``off`` to enable/disable the compact layout for high memory regions. 122f40408a9SGavin Shan The default is ``on`` for machine types later than ``virt-7.2``. 123f40408a9SGavin Shan 1246a48c64eSGavin Shanhighmem-redists 1256a48c64eSGavin Shan Set ``on``/``off`` to enable/disable the high memory region for GICv3 or 1266a48c64eSGavin Shan GICv4 redistributor. The default is ``on``. Setting this to ``off`` will 1276a48c64eSGavin Shan limit the maximum number of CPUs when GICv3 or GICv4 is used. 1286a48c64eSGavin Shan 1296a48c64eSGavin Shanhighmem-ecam 1306a48c64eSGavin Shan Set ``on``/``off`` to enable/disable the high memory region for PCI ECAM. 1316a48c64eSGavin Shan The default is ``on`` for machine types later than ``virt-3.0``. 1326a48c64eSGavin Shan 1336a48c64eSGavin Shanhighmem-mmio 1346a48c64eSGavin Shan Set ``on``/``off`` to enable/disable the high memory region for PCI MMIO. 1356a48c64eSGavin Shan The default is ``on``. 1366a48c64eSGavin Shan 1376a0b7505SPeter Maydellgic-version 1386a0b7505SPeter Maydell Specify the version of the Generic Interrupt Controller (GIC) to provide. 1396a0b7505SPeter Maydell Valid values are: 1406a0b7505SPeter Maydell 1416a0b7505SPeter Maydell ``2`` 14201130173SHeinrich Schuchardt GICv2. Note that this limits the number of CPUs to 8. 1436a0b7505SPeter Maydell ``3`` 14401130173SHeinrich Schuchardt GICv3. This allows up to 512 CPUs. 1457cf3f8d2SPeter Maydell ``4`` 1467cf3f8d2SPeter Maydell GICv4. Requires ``virtualization`` to be ``on``; allows up to 317 CPUs. 1476a0b7505SPeter Maydell ``host`` 1486a0b7505SPeter Maydell Use the same GIC version the host provides, when using KVM 1496a0b7505SPeter Maydell ``max`` 1506a0b7505SPeter Maydell Use the best GIC version possible (same as host when using KVM; 1517cf3f8d2SPeter Maydell with TCG this is currently ``3`` if ``virtualization`` is ``off`` and 1527cf3f8d2SPeter Maydell ``4`` if ``virtualization`` is ``on``, but this may change in future) 1536a0b7505SPeter Maydell 1546a0b7505SPeter Maydellits 1556a0b7505SPeter Maydell Set ``on``/``off`` to enable/disable ITS instantiation. The default is ``on`` 1566a0b7505SPeter Maydell for machine types later than ``virt-2.7``. 1576a0b7505SPeter Maydell 1586a0b7505SPeter Maydelliommu 1596a0b7505SPeter Maydell Set the IOMMU type to create for the guest. Valid values are: 1606a0b7505SPeter Maydell 1616a0b7505SPeter Maydell ``none`` 1626a0b7505SPeter Maydell Don't create an IOMMU (the default) 1636a0b7505SPeter Maydell ``smmuv3`` 1646a0b7505SPeter Maydell Create an SMMUv3 1656a0b7505SPeter Maydell 1666a0b7505SPeter Maydellras 1676a0b7505SPeter Maydell Set ``on``/``off`` to enable/disable reporting host memory errors to a guest 1686a0b7505SPeter Maydell using ACPI and guest external abort exceptions. The default is off. 1696a0b7505SPeter Maydell 1705242876fSJason A. Donenfelddtb-randomness 1715242876fSJason A. Donenfeld Set ``on``/``off`` to pass random seeds via the guest DTB 1725242876fSJason A. Donenfeld rng-seed and kaslr-seed nodes (in both "/chosen" and 1735242876fSJason A. Donenfeld "/secure-chosen") to use for features like the random number 1745242876fSJason A. Donenfeld generator and address space randomisation. The default is 1755242876fSJason A. Donenfeld ``on``. You will want to disable it if your trusted boot chain 1765242876fSJason A. Donenfeld will verify the DTB it is passed, since this option causes the 1775242876fSJason A. Donenfeld DTB to be non-deterministic. It would be the responsibility of 1785242876fSJason A. Donenfeld the firmware to come up with a seed and pass it on if it wants to. 1795242876fSJason A. Donenfeld 18033973e1eSAlex Bennéedtb-kaslr-seed 1815242876fSJason A. Donenfeld A deprecated synonym for dtb-randomness. 18233973e1eSAlex Bennée 1836a0b7505SPeter MaydellLinux guest kernel configuration 1846a0b7505SPeter Maydell"""""""""""""""""""""""""""""""" 1856a0b7505SPeter Maydell 1866a0b7505SPeter MaydellThe 'defconfig' for Linux arm and arm64 kernels should include the 1876a0b7505SPeter Maydellright device drivers for virtio and the PCI controller; however some older 1886a0b7505SPeter Maydellkernel versions, especially for 32-bit Arm, did not have everything 1896a0b7505SPeter Maydellenabled by default. If you're not seeing PCI devices that you expect, 1906a0b7505SPeter Maydellthen check that your guest config has:: 1916a0b7505SPeter Maydell 1926a0b7505SPeter Maydell CONFIG_PCI=y 1936a0b7505SPeter Maydell CONFIG_VIRTIO_PCI=y 1946a0b7505SPeter Maydell CONFIG_PCI_HOST_GENERIC=y 1956a0b7505SPeter Maydell 1966a0b7505SPeter MaydellIf you want to use the ``virtio-gpu-pci`` graphics device you will also 1976a0b7505SPeter Maydellneed:: 1986a0b7505SPeter Maydell 1996a0b7505SPeter Maydell CONFIG_DRM=y 2006a0b7505SPeter Maydell CONFIG_DRM_VIRTIO_GPU=y 2016a0b7505SPeter Maydell 2026a0b7505SPeter MaydellHardware configuration information for bare-metal programming 2036a0b7505SPeter Maydell""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""" 2046a0b7505SPeter Maydell 2056a0b7505SPeter MaydellThe ``virt`` board automatically generates a device tree blob ("dtb") 2066a0b7505SPeter Maydellwhich it passes to the guest. This provides information about the 2076a0b7505SPeter Maydelladdresses, interrupt lines and other configuration of the various devices 2086a0b7505SPeter Maydellin the system. Guest code can rely on and hard-code the following 2096a0b7505SPeter Maydelladdresses: 2106a0b7505SPeter Maydell 2116a0b7505SPeter Maydell- Flash memory starts at address 0x0000_0000 2126a0b7505SPeter Maydell 2136a0b7505SPeter Maydell- RAM starts at 0x4000_0000 2146a0b7505SPeter Maydell 2156a0b7505SPeter MaydellAll other information about device locations may change between 2166a0b7505SPeter MaydellQEMU versions, so guest code must look in the DTB. 2176a0b7505SPeter Maydell 2186a0b7505SPeter MaydellQEMU supports two types of guest image boot for ``virt``, and 2196a0b7505SPeter Maydellthe way for the guest code to locate the dtb binary differs: 2206a0b7505SPeter Maydell 2216a0b7505SPeter Maydell- For guests using the Linux kernel boot protocol (this means any 2226a0b7505SPeter Maydell non-ELF file passed to the QEMU ``-kernel`` option) the address 2236a0b7505SPeter Maydell of the DTB is passed in a register (``r2`` for 32-bit guests, 2246a0b7505SPeter Maydell or ``x0`` for 64-bit guests) 2256a0b7505SPeter Maydell 2266a0b7505SPeter Maydell- For guests booting as "bare-metal" (any other kind of boot), 2276a0b7505SPeter Maydell the DTB is at the start of RAM (0x4000_0000) 228