1.. SPDX-License-Identifier: GPL-2.0 2 3==================== 4Generic vm interface 5==================== 6 7The virtual machine "device" also accepts the ioctls KVM_SET_DEVICE_ATTR, 8KVM_GET_DEVICE_ATTR, and KVM_HAS_DEVICE_ATTR. The interface uses the same 9struct kvm_device_attr as other devices, but targets VM-wide settings 10and controls. 11 12The groups and attributes per virtual machine, if any, are architecture 13specific. 14 151. GROUP: KVM_S390_VM_MEM_CTRL 16============================== 17 18:Architectures: s390 19 201.1. ATTRIBUTE: KVM_S390_VM_MEM_ENABLE_CMMA 21------------------------------------------- 22 23:Parameters: none 24:Returns: -EBUSY if a vcpu is already defined, otherwise 0 25 26Enables Collaborative Memory Management Assist (CMMA) for the virtual machine. 27 281.2. ATTRIBUTE: KVM_S390_VM_MEM_CLR_CMMA 29---------------------------------------- 30 31:Parameters: none 32:Returns: -EINVAL if CMMA was not enabled; 33 0 otherwise 34 35Clear the CMMA status for all guest pages, so any pages the guest marked 36as unused are again used any may not be reclaimed by the host. 37 381.3. ATTRIBUTE KVM_S390_VM_MEM_LIMIT_SIZE 39----------------------------------------- 40 41:Parameters: in attr->addr the address for the new limit of guest memory 42:Returns: -EFAULT if the given address is not accessible; 43 -EINVAL if the virtual machine is of type UCONTROL; 44 -E2BIG if the given guest memory is to big for that machine; 45 -EBUSY if a vcpu is already defined; 46 -ENOMEM if not enough memory is available for a new shadow guest mapping; 47 0 otherwise. 48 49Allows userspace to query the actual limit and set a new limit for 50the maximum guest memory size. The limit will be rounded up to 512048 MB, 4096 GB, 8192 TB respectively, as this limit is governed by 52the number of page table levels. In the case that there is no limit we will set 53the limit to KVM_S390_NO_MEM_LIMIT (U64_MAX). 54 552. GROUP: KVM_S390_VM_CPU_MODEL 56=============================== 57 58:Architectures: s390 59 602.1. ATTRIBUTE: KVM_S390_VM_CPU_MACHINE (r/o) 61--------------------------------------------- 62 63Allows user space to retrieve machine and kvm specific cpu related information:: 64 65 struct kvm_s390_vm_cpu_machine { 66 __u64 cpuid; # CPUID of host 67 __u32 ibc; # IBC level range offered by host 68 __u8 pad[4]; 69 __u64 fac_mask[256]; # set of cpu facilities enabled by KVM 70 __u64 fac_list[256]; # set of cpu facilities offered by host 71 } 72 73:Parameters: address of buffer to store the machine related cpu data 74 of type struct kvm_s390_vm_cpu_machine* 75:Returns: -EFAULT if the given address is not accessible from kernel space; 76 -ENOMEM if not enough memory is available to process the ioctl; 77 0 in case of success. 78 792.2. ATTRIBUTE: KVM_S390_VM_CPU_PROCESSOR (r/w) 80=============================================== 81 82Allows user space to retrieve or request to change cpu related information for a vcpu:: 83 84 struct kvm_s390_vm_cpu_processor { 85 __u64 cpuid; # CPUID currently (to be) used by this vcpu 86 __u16 ibc; # IBC level currently (to be) used by this vcpu 87 __u8 pad[6]; 88 __u64 fac_list[256]; # set of cpu facilities currently (to be) used 89 # by this vcpu 90 } 91 92KVM does not enforce or limit the cpu model data in any form. Take the information 93retrieved by means of KVM_S390_VM_CPU_MACHINE as hint for reasonable configuration 94setups. Instruction interceptions triggered by additionally set facility bits that 95are not handled by KVM need to by implemented in the VM driver code. 96 97:Parameters: address of buffer to store/set the processor related cpu 98 data of type struct kvm_s390_vm_cpu_processor*. 99:Returns: -EBUSY in case 1 or more vcpus are already activated (only in write case); 100 -EFAULT if the given address is not accessible from kernel space; 101 -ENOMEM if not enough memory is available to process the ioctl; 102 0 in case of success. 103 104.. _KVM_S390_VM_CPU_MACHINE_FEAT: 105 1062.3. ATTRIBUTE: KVM_S390_VM_CPU_MACHINE_FEAT (r/o) 107-------------------------------------------------- 108 109Allows user space to retrieve available cpu features. A feature is available if 110provided by the hardware and supported by kvm. In theory, cpu features could 111even be completely emulated by kvm. 112 113:: 114 115 struct kvm_s390_vm_cpu_feat { 116 __u64 feat[16]; # Bitmap (1 = feature available), MSB 0 bit numbering 117 }; 118 119:Parameters: address of a buffer to load the feature list from. 120:Returns: -EFAULT if the given address is not accessible from kernel space; 121 0 in case of success. 122 1232.4. ATTRIBUTE: KVM_S390_VM_CPU_PROCESSOR_FEAT (r/w) 124---------------------------------------------------- 125 126Allows user space to retrieve or change enabled cpu features for all VCPUs of a 127VM. Features that are not available cannot be enabled. 128 129See :ref:`KVM_S390_VM_CPU_MACHINE_FEAT` for 130a description of the parameter struct. 131 132:Parameters: address of a buffer to store/load the feature list from. 133:Returns: -EFAULT if the given address is not accessible from kernel space; 134 -EINVAL if a cpu feature that is not available is to be enabled; 135 -EBUSY if at least one VCPU has already been defined; 136 0 in case of success. 137 138.. _KVM_S390_VM_CPU_MACHINE_SUBFUNC: 139 1402.5. ATTRIBUTE: KVM_S390_VM_CPU_MACHINE_SUBFUNC (r/o) 141----------------------------------------------------- 142 143Allows user space to retrieve available cpu subfunctions without any filtering 144done by a set IBC. These subfunctions are indicated to the guest VCPU via 145query or "test bit" subfunctions and used e.g. by cpacf functions, plo and ptff. 146 147A subfunction block is only valid if KVM_S390_VM_CPU_MACHINE contains the 148STFL(E) bit introducing the affected instruction. If the affected instruction 149indicates subfunctions via a "query subfunction", the response block is 150contained in the returned struct. If the affected instruction 151indicates subfunctions via a "test bit" mechanism, the subfunction codes are 152contained in the returned struct in MSB 0 bit numbering. 153 154:: 155 156 struct kvm_s390_vm_cpu_subfunc { 157 u8 plo[32]; # always valid (ESA/390 feature) 158 u8 ptff[16]; # valid with TOD-clock steering 159 u8 kmac[16]; # valid with Message-Security-Assist 160 u8 kmc[16]; # valid with Message-Security-Assist 161 u8 km[16]; # valid with Message-Security-Assist 162 u8 kimd[16]; # valid with Message-Security-Assist 163 u8 klmd[16]; # valid with Message-Security-Assist 164 u8 pckmo[16]; # valid with Message-Security-Assist-Extension 3 165 u8 kmctr[16]; # valid with Message-Security-Assist-Extension 4 166 u8 kmf[16]; # valid with Message-Security-Assist-Extension 4 167 u8 kmo[16]; # valid with Message-Security-Assist-Extension 4 168 u8 pcc[16]; # valid with Message-Security-Assist-Extension 4 169 u8 ppno[16]; # valid with Message-Security-Assist-Extension 5 170 u8 kma[16]; # valid with Message-Security-Assist-Extension 8 171 u8 kdsa[16]; # valid with Message-Security-Assist-Extension 9 172 u8 reserved[1792]; # reserved for future instructions 173 }; 174 175:Parameters: address of a buffer to load the subfunction blocks from. 176:Returns: -EFAULT if the given address is not accessible from kernel space; 177 0 in case of success. 178 1792.6. ATTRIBUTE: KVM_S390_VM_CPU_PROCESSOR_SUBFUNC (r/w) 180------------------------------------------------------- 181 182Allows user space to retrieve or change cpu subfunctions to be indicated for 183all VCPUs of a VM. This attribute will only be available if kernel and 184hardware support are in place. 185 186The kernel uses the configured subfunction blocks for indication to 187the guest. A subfunction block will only be used if the associated STFL(E) bit 188has not been disabled by user space (so the instruction to be queried is 189actually available for the guest). 190 191As long as no data has been written, a read will fail. The IBC will be used 192to determine available subfunctions in this case, this will guarantee backward 193compatibility. 194 195See :ref:`KVM_S390_VM_CPU_MACHINE_SUBFUNC` for a 196description of the parameter struct. 197 198:Parameters: address of a buffer to store/load the subfunction blocks from. 199:Returns: -EFAULT if the given address is not accessible from kernel space; 200 -EINVAL when reading, if there was no write yet; 201 -EBUSY if at least one VCPU has already been defined; 202 0 in case of success. 203 2043. GROUP: KVM_S390_VM_TOD 205========================= 206 207:Architectures: s390 208 2093.1. ATTRIBUTE: KVM_S390_VM_TOD_HIGH 210------------------------------------ 211 212Allows user space to set/get the TOD clock extension (u8) (superseded by 213KVM_S390_VM_TOD_EXT). 214 215:Parameters: address of a buffer in user space to store the data (u8) to 216:Returns: -EFAULT if the given address is not accessible from kernel space; 217 -EINVAL if setting the TOD clock extension to != 0 is not supported 218 -EOPNOTSUPP for a PV guest (TOD managed by the ultravisor) 219 2203.2. ATTRIBUTE: KVM_S390_VM_TOD_LOW 221----------------------------------- 222 223Allows user space to set/get bits 0-63 of the TOD clock register as defined in 224the POP (u64). 225 226:Parameters: address of a buffer in user space to store the data (u64) to 227:Returns: -EFAULT if the given address is not accessible from kernel space 228 -EOPNOTSUPP for a PV guest (TOD managed by the ultravisor) 229 2303.3. ATTRIBUTE: KVM_S390_VM_TOD_EXT 231----------------------------------- 232 233Allows user space to set/get bits 0-63 of the TOD clock register as defined in 234the POP (u64). If the guest CPU model supports the TOD clock extension (u8), it 235also allows user space to get/set it. If the guest CPU model does not support 236it, it is stored as 0 and not allowed to be set to a value != 0. 237 238:Parameters: address of a buffer in user space to store the data 239 (kvm_s390_vm_tod_clock) to 240:Returns: -EFAULT if the given address is not accessible from kernel space; 241 -EINVAL if setting the TOD clock extension to != 0 is not supported 242 -EOPNOTSUPP for a PV guest (TOD managed by the ultravisor) 243 2444. GROUP: KVM_S390_VM_CRYPTO 245============================ 246 247:Architectures: s390 248 2494.1. ATTRIBUTE: KVM_S390_VM_CRYPTO_ENABLE_AES_KW (w/o) 250------------------------------------------------------ 251 252Allows user space to enable aes key wrapping, including generating a new 253wrapping key. 254 255:Parameters: none 256:Returns: 0 257 2584.2. ATTRIBUTE: KVM_S390_VM_CRYPTO_ENABLE_DEA_KW (w/o) 259------------------------------------------------------ 260 261Allows user space to enable dea key wrapping, including generating a new 262wrapping key. 263 264:Parameters: none 265:Returns: 0 266 2674.3. ATTRIBUTE: KVM_S390_VM_CRYPTO_DISABLE_AES_KW (w/o) 268------------------------------------------------------- 269 270Allows user space to disable aes key wrapping, clearing the wrapping key. 271 272:Parameters: none 273:Returns: 0 274 2754.4. ATTRIBUTE: KVM_S390_VM_CRYPTO_DISABLE_DEA_KW (w/o) 276------------------------------------------------------- 277 278Allows user space to disable dea key wrapping, clearing the wrapping key. 279 280:Parameters: none 281:Returns: 0 282 2835. GROUP: KVM_S390_VM_MIGRATION 284=============================== 285 286:Architectures: s390 287 2885.1. ATTRIBUTE: KVM_S390_VM_MIGRATION_STOP (w/o) 289------------------------------------------------ 290 291Allows userspace to stop migration mode, needed for PGSTE migration. 292Setting this attribute when migration mode is not active will have no 293effects. 294 295:Parameters: none 296:Returns: 0 297 2985.2. ATTRIBUTE: KVM_S390_VM_MIGRATION_START (w/o) 299------------------------------------------------- 300 301Allows userspace to start migration mode, needed for PGSTE migration. 302Setting this attribute when migration mode is already active will have 303no effects. 304 305Dirty tracking must be enabled on all memslots, else -EINVAL is returned. When 306dirty tracking is disabled on any memslot, migration mode is automatically 307stopped. 308 309:Parameters: none 310:Returns: -ENOMEM if there is not enough free memory to start migration mode; 311 -EINVAL if the state of the VM is invalid (e.g. no memory defined); 312 0 in case of success. 313 3145.3. ATTRIBUTE: KVM_S390_VM_MIGRATION_STATUS (r/o) 315-------------------------------------------------- 316 317Allows userspace to query the status of migration mode. 318 319:Parameters: address of a buffer in user space to store the data (u64) to; 320 the data itself is either 0 if migration mode is disabled or 1 321 if it is enabled 322:Returns: -EFAULT if the given address is not accessible from kernel space; 323 0 in case of success. 324 3256. GROUP: KVM_ARM_VM_SMCCC_CTRL 326=============================== 327 328:Architectures: arm64 329 3306.1. ATTRIBUTE: KVM_ARM_VM_SMCCC_FILTER (w/o) 331--------------------------------------------- 332 333:Parameters: Pointer to a ``struct kvm_smccc_filter`` 334 335:Returns: 336 337 ====== =========================================== 338 EEXIST Range intersects with a previously inserted 339 or reserved range 340 EBUSY A vCPU in the VM has already run 341 EINVAL Invalid filter configuration 342 ENOMEM Failed to allocate memory for the in-kernel 343 representation of the SMCCC filter 344 ====== =========================================== 345 346Requests the installation of an SMCCC call filter described as follows:: 347 348 enum kvm_smccc_filter_action { 349 KVM_SMCCC_FILTER_HANDLE = 0, 350 KVM_SMCCC_FILTER_DENY, 351 KVM_SMCCC_FILTER_FWD_TO_USER, 352 }; 353 354 struct kvm_smccc_filter { 355 __u32 base; 356 __u32 nr_functions; 357 __u8 action; 358 __u8 pad[15]; 359 }; 360 361The filter is defined as a set of non-overlapping ranges. Each 362range defines an action to be applied to SMCCC calls within the range. 363Userspace can insert multiple ranges into the filter by using 364successive calls to this attribute. 365 366The default configuration of KVM is such that all implemented SMCCC 367calls are allowed. Thus, the SMCCC filter can be defined sparsely 368by userspace, only describing ranges that modify the default behavior. 369 370The range expressed by ``struct kvm_smccc_filter`` is 371[``base``, ``base + nr_functions``). The range is not allowed to wrap, 372i.e. userspace cannot rely on ``base + nr_functions`` overflowing. 373 374The SMCCC filter applies to both SMC and HVC calls initiated by the 375guest. The SMCCC filter gates the in-kernel emulation of SMCCC calls 376and as such takes effect before other interfaces that interact with 377SMCCC calls (e.g. hypercall bitmap registers). 378 379Actions: 380 381 - ``KVM_SMCCC_FILTER_HANDLE``: Allows the guest SMCCC call to be 382 handled in-kernel. It is strongly recommended that userspace *not* 383 explicitly describe the allowed SMCCC call ranges. 384 385 - ``KVM_SMCCC_FILTER_DENY``: Rejects the guest SMCCC call in-kernel 386 and returns to the guest. 387 388 - ``KVM_SMCCC_FILTER_FWD_TO_USER``: The guest SMCCC call is forwarded 389 to userspace with an exit reason of ``KVM_EXIT_HYPERCALL``. 390 391The ``pad`` field is reserved for future use and must be zero. KVM may 392return ``-EINVAL`` if the field is nonzero. 393 394KVM reserves the 'Arm Architecture Calls' range of function IDs and 395will reject attempts to define a filter for any portion of these ranges: 396 397 =========== =============== 398 Start End (inclusive) 399 =========== =============== 400 0x8000_0000 0x8000_FFFF 401 0xC000_0000 0xC000_FFFF 402 =========== =============== 403