1.. SPDX-License-Identifier: GPL-2.0 2 3=================================================================== 4The Definitive KVM (Kernel-based Virtual Machine) API Documentation 5=================================================================== 6 71. General description 8====================== 9 10The kvm API is a set of ioctls that are issued to control various aspects 11of a virtual machine. The ioctls belong to the following classes: 12 13 - System ioctls: These query and set global attributes which affect the 14 whole kvm subsystem. In addition a system ioctl is used to create 15 virtual machines. 16 17 - VM ioctls: These query and set attributes that affect an entire virtual 18 machine, for example memory layout. In addition a VM ioctl is used to 19 create virtual cpus (vcpus) and devices. 20 21 VM ioctls must be issued from the same process (address space) that was 22 used to create the VM. 23 24 - vcpu ioctls: These query and set attributes that control the operation 25 of a single virtual cpu. 26 27 vcpu ioctls should be issued from the same thread that was used to create 28 the vcpu, except for asynchronous vcpu ioctl that are marked as such in 29 the documentation. Otherwise, the first ioctl after switching threads 30 could see a performance impact. 31 32 - device ioctls: These query and set attributes that control the operation 33 of a single device. 34 35 device ioctls must be issued from the same process (address space) that 36 was used to create the VM. 37 382. File descriptors 39=================== 40 41The kvm API is centered around file descriptors. An initial 42open("/dev/kvm") obtains a handle to the kvm subsystem; this handle 43can be used to issue system ioctls. A KVM_CREATE_VM ioctl on this 44handle will create a VM file descriptor which can be used to issue VM 45ioctls. A KVM_CREATE_VCPU or KVM_CREATE_DEVICE ioctl on a VM fd will 46create a virtual cpu or device and return a file descriptor pointing to 47the new resource. Finally, ioctls on a vcpu or device fd can be used 48to control the vcpu or device. For vcpus, this includes the important 49task of actually running guest code. 50 51In general file descriptors can be migrated among processes by means 52of fork() and the SCM_RIGHTS facility of unix domain socket. These 53kinds of tricks are explicitly not supported by kvm. While they will 54not cause harm to the host, their actual behavior is not guaranteed by 55the API. See "General description" for details on the ioctl usage 56model that is supported by KVM. 57 58It is important to note that although VM ioctls may only be issued from 59the process that created the VM, a VM's lifecycle is associated with its 60file descriptor, not its creator (process). In other words, the VM and 61its resources, *including the associated address space*, are not freed 62until the last reference to the VM's file descriptor has been released. 63For example, if fork() is issued after ioctl(KVM_CREATE_VM), the VM will 64not be freed until both the parent (original) process and its child have 65put their references to the VM's file descriptor. 66 67Because a VM's resources are not freed until the last reference to its 68file descriptor is released, creating additional references to a VM 69via fork(), dup(), etc... without careful consideration is strongly 70discouraged and may have unwanted side effects, e.g. memory allocated 71by and on behalf of the VM's process may not be freed/unaccounted when 72the VM is shut down. 73 74 753. Extensions 76============= 77 78As of Linux 2.6.22, the KVM ABI has been stabilized: no backward 79incompatible change are allowed. However, there is an extension 80facility that allows backward-compatible extensions to the API to be 81queried and used. 82 83The extension mechanism is not based on the Linux version number. 84Instead, kvm defines extension identifiers and a facility to query 85whether a particular extension identifier is available. If it is, a 86set of ioctls is available for application use. 87 88 894. API description 90================== 91 92This section describes ioctls that can be used to control kvm guests. 93For each ioctl, the following information is provided along with a 94description: 95 96 Capability: 97 which KVM extension provides this ioctl. Can be 'basic', 98 which means that is will be provided by any kernel that supports 99 API version 12 (see section 4.1), a KVM_CAP_xyz constant, which 100 means availability needs to be checked with KVM_CHECK_EXTENSION 101 (see section 4.4), or 'none' which means that while not all kernels 102 support this ioctl, there's no capability bit to check its 103 availability: for kernels that don't support the ioctl, 104 the ioctl returns -ENOTTY. 105 106 Architectures: 107 which instruction set architectures provide this ioctl. 108 x86 includes both i386 and x86_64. 109 110 Type: 111 system, vm, or vcpu. 112 113 Parameters: 114 what parameters are accepted by the ioctl. 115 116 Returns: 117 the return value. General error numbers (EBADF, ENOMEM, EINVAL) 118 are not detailed, but errors with specific meanings are. 119 120 1214.1 KVM_GET_API_VERSION 122----------------------- 123 124:Capability: basic 125:Architectures: all 126:Type: system ioctl 127:Parameters: none 128:Returns: the constant KVM_API_VERSION (=12) 129 130This identifies the API version as the stable kvm API. It is not 131expected that this number will change. However, Linux 2.6.20 and 1322.6.21 report earlier versions; these are not documented and not 133supported. Applications should refuse to run if KVM_GET_API_VERSION 134returns a value other than 12. If this check passes, all ioctls 135described as 'basic' will be available. 136 137 1384.2 KVM_CREATE_VM 139----------------- 140 141:Capability: basic 142:Architectures: all 143:Type: system ioctl 144:Parameters: machine type identifier (KVM_VM_*) 145:Returns: a VM fd that can be used to control the new virtual machine. 146 147The new VM has no virtual cpus and no memory. 148You probably want to use 0 as machine type. 149 150In order to create user controlled virtual machines on S390, check 151KVM_CAP_S390_UCONTROL and use the flag KVM_VM_S390_UCONTROL as 152privileged user (CAP_SYS_ADMIN). 153 154To use hardware assisted virtualization on MIPS (VZ ASE) rather than 155the default trap & emulate implementation (which changes the virtual 156memory layout to fit in user mode), check KVM_CAP_MIPS_VZ and use the 157flag KVM_VM_MIPS_VZ. 158 159 160On arm64, the physical address size for a VM (IPA Size limit) is limited 161to 40bits by default. The limit can be configured if the host supports the 162extension KVM_CAP_ARM_VM_IPA_SIZE. When supported, use 163KVM_VM_TYPE_ARM_IPA_SIZE(IPA_Bits) to set the size in the machine type 164identifier, where IPA_Bits is the maximum width of any physical 165address used by the VM. The IPA_Bits is encoded in bits[7-0] of the 166machine type identifier. 167 168e.g, to configure a guest to use 48bit physical address size:: 169 170 vm_fd = ioctl(dev_fd, KVM_CREATE_VM, KVM_VM_TYPE_ARM_IPA_SIZE(48)); 171 172The requested size (IPA_Bits) must be: 173 174 == ========================================================= 175 0 Implies default size, 40bits (for backward compatibility) 176 N Implies N bits, where N is a positive integer such that, 177 32 <= N <= Host_IPA_Limit 178 == ========================================================= 179 180Host_IPA_Limit is the maximum possible value for IPA_Bits on the host and 181is dependent on the CPU capability and the kernel configuration. The limit can 182be retrieved using KVM_CAP_ARM_VM_IPA_SIZE of the KVM_CHECK_EXTENSION 183ioctl() at run-time. 184 185Creation of the VM will fail if the requested IPA size (whether it is 186implicit or explicit) is unsupported on the host. 187 188Please note that configuring the IPA size does not affect the capability 189exposed by the guest CPUs in ID_AA64MMFR0_EL1[PARange]. It only affects 190size of the address translated by the stage2 level (guest physical to 191host physical address translations). 192 193 1944.3 KVM_GET_MSR_INDEX_LIST, KVM_GET_MSR_FEATURE_INDEX_LIST 195---------------------------------------------------------- 196 197:Capability: basic, KVM_CAP_GET_MSR_FEATURES for KVM_GET_MSR_FEATURE_INDEX_LIST 198:Architectures: x86 199:Type: system ioctl 200:Parameters: struct kvm_msr_list (in/out) 201:Returns: 0 on success; -1 on error 202 203Errors: 204 205 ====== ============================================================ 206 EFAULT the msr index list cannot be read from or written to 207 E2BIG the msr index list is too big to fit in the array specified by 208 the user. 209 ====== ============================================================ 210 211:: 212 213 struct kvm_msr_list { 214 __u32 nmsrs; /* number of msrs in entries */ 215 __u32 indices[0]; 216 }; 217 218The user fills in the size of the indices array in nmsrs, and in return 219kvm adjusts nmsrs to reflect the actual number of msrs and fills in the 220indices array with their numbers. 221 222KVM_GET_MSR_INDEX_LIST returns the guest msrs that are supported. The list 223varies by kvm version and host processor, but does not change otherwise. 224 225Note: if kvm indicates supports MCE (KVM_CAP_MCE), then the MCE bank MSRs are 226not returned in the MSR list, as different vcpus can have a different number 227of banks, as set via the KVM_X86_SETUP_MCE ioctl. 228 229KVM_GET_MSR_FEATURE_INDEX_LIST returns the list of MSRs that can be passed 230to the KVM_GET_MSRS system ioctl. This lets userspace probe host capabilities 231and processor features that are exposed via MSRs (e.g., VMX capabilities). 232This list also varies by kvm version and host processor, but does not change 233otherwise. 234 235 2364.4 KVM_CHECK_EXTENSION 237----------------------- 238 239:Capability: basic, KVM_CAP_CHECK_EXTENSION_VM for vm ioctl 240:Architectures: all 241:Type: system ioctl, vm ioctl 242:Parameters: extension identifier (KVM_CAP_*) 243:Returns: 0 if unsupported; 1 (or some other positive integer) if supported 244 245The API allows the application to query about extensions to the core 246kvm API. Userspace passes an extension identifier (an integer) and 247receives an integer that describes the extension availability. 248Generally 0 means no and 1 means yes, but some extensions may report 249additional information in the integer return value. 250 251Based on their initialization different VMs may have different capabilities. 252It is thus encouraged to use the vm ioctl to query for capabilities (available 253with KVM_CAP_CHECK_EXTENSION_VM on the vm fd) 254 2554.5 KVM_GET_VCPU_MMAP_SIZE 256-------------------------- 257 258:Capability: basic 259:Architectures: all 260:Type: system ioctl 261:Parameters: none 262:Returns: size of vcpu mmap area, in bytes 263 264The KVM_RUN ioctl (cf.) communicates with userspace via a shared 265memory region. This ioctl returns the size of that region. See the 266KVM_RUN documentation for details. 267 268Besides the size of the KVM_RUN communication region, other areas of 269the VCPU file descriptor can be mmap-ed, including: 270 271- if KVM_CAP_COALESCED_MMIO is available, a page at 272 KVM_COALESCED_MMIO_PAGE_OFFSET * PAGE_SIZE; for historical reasons, 273 this page is included in the result of KVM_GET_VCPU_MMAP_SIZE. 274 KVM_CAP_COALESCED_MMIO is not documented yet. 275 276- if KVM_CAP_DIRTY_LOG_RING is available, a number of pages at 277 KVM_DIRTY_LOG_PAGE_OFFSET * PAGE_SIZE. For more information on 278 KVM_CAP_DIRTY_LOG_RING, see section 8.3. 279 280 2814.6 KVM_SET_MEMORY_REGION 282------------------------- 283 284:Capability: basic 285:Architectures: all 286:Type: vm ioctl 287:Parameters: struct kvm_memory_region (in) 288:Returns: 0 on success, -1 on error 289 290This ioctl is obsolete and has been removed. 291 292 2934.7 KVM_CREATE_VCPU 294------------------- 295 296:Capability: basic 297:Architectures: all 298:Type: vm ioctl 299:Parameters: vcpu id (apic id on x86) 300:Returns: vcpu fd on success, -1 on error 301 302This API adds a vcpu to a virtual machine. No more than max_vcpus may be added. 303The vcpu id is an integer in the range [0, max_vcpu_id). 304 305The recommended max_vcpus value can be retrieved using the KVM_CAP_NR_VCPUS of 306the KVM_CHECK_EXTENSION ioctl() at run-time. 307The maximum possible value for max_vcpus can be retrieved using the 308KVM_CAP_MAX_VCPUS of the KVM_CHECK_EXTENSION ioctl() at run-time. 309 310If the KVM_CAP_NR_VCPUS does not exist, you should assume that max_vcpus is 4 311cpus max. 312If the KVM_CAP_MAX_VCPUS does not exist, you should assume that max_vcpus is 313same as the value returned from KVM_CAP_NR_VCPUS. 314 315The maximum possible value for max_vcpu_id can be retrieved using the 316KVM_CAP_MAX_VCPU_ID of the KVM_CHECK_EXTENSION ioctl() at run-time. 317 318If the KVM_CAP_MAX_VCPU_ID does not exist, you should assume that max_vcpu_id 319is the same as the value returned from KVM_CAP_MAX_VCPUS. 320 321On powerpc using book3s_hv mode, the vcpus are mapped onto virtual 322threads in one or more virtual CPU cores. (This is because the 323hardware requires all the hardware threads in a CPU core to be in the 324same partition.) The KVM_CAP_PPC_SMT capability indicates the number 325of vcpus per virtual core (vcore). The vcore id is obtained by 326dividing the vcpu id by the number of vcpus per vcore. The vcpus in a 327given vcore will always be in the same physical core as each other 328(though that might be a different physical core from time to time). 329Userspace can control the threading (SMT) mode of the guest by its 330allocation of vcpu ids. For example, if userspace wants 331single-threaded guest vcpus, it should make all vcpu ids be a multiple 332of the number of vcpus per vcore. 333 334For virtual cpus that have been created with S390 user controlled virtual 335machines, the resulting vcpu fd can be memory mapped at page offset 336KVM_S390_SIE_PAGE_OFFSET in order to obtain a memory map of the virtual 337cpu's hardware control block. 338 339 3404.8 KVM_GET_DIRTY_LOG (vm ioctl) 341-------------------------------- 342 343:Capability: basic 344:Architectures: all 345:Type: vm ioctl 346:Parameters: struct kvm_dirty_log (in/out) 347:Returns: 0 on success, -1 on error 348 349:: 350 351 /* for KVM_GET_DIRTY_LOG */ 352 struct kvm_dirty_log { 353 __u32 slot; 354 __u32 padding; 355 union { 356 void __user *dirty_bitmap; /* one bit per page */ 357 __u64 padding; 358 }; 359 }; 360 361Given a memory slot, return a bitmap containing any pages dirtied 362since the last call to this ioctl. Bit 0 is the first page in the 363memory slot. Ensure the entire structure is cleared to avoid padding 364issues. 365 366If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies 367the address space for which you want to return the dirty bitmap. See 368KVM_SET_USER_MEMORY_REGION for details on the usage of slot field. 369 370The bits in the dirty bitmap are cleared before the ioctl returns, unless 371KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is enabled. For more information, 372see the description of the capability. 373 3744.9 KVM_SET_MEMORY_ALIAS 375------------------------ 376 377:Capability: basic 378:Architectures: x86 379:Type: vm ioctl 380:Parameters: struct kvm_memory_alias (in) 381:Returns: 0 (success), -1 (error) 382 383This ioctl is obsolete and has been removed. 384 385 3864.10 KVM_RUN 387------------ 388 389:Capability: basic 390:Architectures: all 391:Type: vcpu ioctl 392:Parameters: none 393:Returns: 0 on success, -1 on error 394 395Errors: 396 397 ======= ============================================================== 398 EINTR an unmasked signal is pending 399 ENOEXEC the vcpu hasn't been initialized or the guest tried to execute 400 instructions from device memory (arm64) 401 ENOSYS data abort outside memslots with no syndrome info and 402 KVM_CAP_ARM_NISV_TO_USER not enabled (arm64) 403 EPERM SVE feature set but not finalized (arm64) 404 ======= ============================================================== 405 406This ioctl is used to run a guest virtual cpu. While there are no 407explicit parameters, there is an implicit parameter block that can be 408obtained by mmap()ing the vcpu fd at offset 0, with the size given by 409KVM_GET_VCPU_MMAP_SIZE. The parameter block is formatted as a 'struct 410kvm_run' (see below). 411 412 4134.11 KVM_GET_REGS 414----------------- 415 416:Capability: basic 417:Architectures: all except ARM, arm64 418:Type: vcpu ioctl 419:Parameters: struct kvm_regs (out) 420:Returns: 0 on success, -1 on error 421 422Reads the general purpose registers from the vcpu. 423 424:: 425 426 /* x86 */ 427 struct kvm_regs { 428 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */ 429 __u64 rax, rbx, rcx, rdx; 430 __u64 rsi, rdi, rsp, rbp; 431 __u64 r8, r9, r10, r11; 432 __u64 r12, r13, r14, r15; 433 __u64 rip, rflags; 434 }; 435 436 /* mips */ 437 struct kvm_regs { 438 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */ 439 __u64 gpr[32]; 440 __u64 hi; 441 __u64 lo; 442 __u64 pc; 443 }; 444 445 4464.12 KVM_SET_REGS 447----------------- 448 449:Capability: basic 450:Architectures: all except ARM, arm64 451:Type: vcpu ioctl 452:Parameters: struct kvm_regs (in) 453:Returns: 0 on success, -1 on error 454 455Writes the general purpose registers into the vcpu. 456 457See KVM_GET_REGS for the data structure. 458 459 4604.13 KVM_GET_SREGS 461------------------ 462 463:Capability: basic 464:Architectures: x86, ppc 465:Type: vcpu ioctl 466:Parameters: struct kvm_sregs (out) 467:Returns: 0 on success, -1 on error 468 469Reads special registers from the vcpu. 470 471:: 472 473 /* x86 */ 474 struct kvm_sregs { 475 struct kvm_segment cs, ds, es, fs, gs, ss; 476 struct kvm_segment tr, ldt; 477 struct kvm_dtable gdt, idt; 478 __u64 cr0, cr2, cr3, cr4, cr8; 479 __u64 efer; 480 __u64 apic_base; 481 __u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64]; 482 }; 483 484 /* ppc -- see arch/powerpc/include/uapi/asm/kvm.h */ 485 486interrupt_bitmap is a bitmap of pending external interrupts. At most 487one bit may be set. This interrupt has been acknowledged by the APIC 488but not yet injected into the cpu core. 489 490 4914.14 KVM_SET_SREGS 492------------------ 493 494:Capability: basic 495:Architectures: x86, ppc 496:Type: vcpu ioctl 497:Parameters: struct kvm_sregs (in) 498:Returns: 0 on success, -1 on error 499 500Writes special registers into the vcpu. See KVM_GET_SREGS for the 501data structures. 502 503 5044.15 KVM_TRANSLATE 505------------------ 506 507:Capability: basic 508:Architectures: x86 509:Type: vcpu ioctl 510:Parameters: struct kvm_translation (in/out) 511:Returns: 0 on success, -1 on error 512 513Translates a virtual address according to the vcpu's current address 514translation mode. 515 516:: 517 518 struct kvm_translation { 519 /* in */ 520 __u64 linear_address; 521 522 /* out */ 523 __u64 physical_address; 524 __u8 valid; 525 __u8 writeable; 526 __u8 usermode; 527 __u8 pad[5]; 528 }; 529 530 5314.16 KVM_INTERRUPT 532------------------ 533 534:Capability: basic 535:Architectures: x86, ppc, mips 536:Type: vcpu ioctl 537:Parameters: struct kvm_interrupt (in) 538:Returns: 0 on success, negative on failure. 539 540Queues a hardware interrupt vector to be injected. 541 542:: 543 544 /* for KVM_INTERRUPT */ 545 struct kvm_interrupt { 546 /* in */ 547 __u32 irq; 548 }; 549 550X86: 551^^^^ 552 553:Returns: 554 555 ========= =================================== 556 0 on success, 557 -EEXIST if an interrupt is already enqueued 558 -EINVAL the irq number is invalid 559 -ENXIO if the PIC is in the kernel 560 -EFAULT if the pointer is invalid 561 ========= =================================== 562 563Note 'irq' is an interrupt vector, not an interrupt pin or line. This 564ioctl is useful if the in-kernel PIC is not used. 565 566PPC: 567^^^^ 568 569Queues an external interrupt to be injected. This ioctl is overleaded 570with 3 different irq values: 571 572a) KVM_INTERRUPT_SET 573 574 This injects an edge type external interrupt into the guest once it's ready 575 to receive interrupts. When injected, the interrupt is done. 576 577b) KVM_INTERRUPT_UNSET 578 579 This unsets any pending interrupt. 580 581 Only available with KVM_CAP_PPC_UNSET_IRQ. 582 583c) KVM_INTERRUPT_SET_LEVEL 584 585 This injects a level type external interrupt into the guest context. The 586 interrupt stays pending until a specific ioctl with KVM_INTERRUPT_UNSET 587 is triggered. 588 589 Only available with KVM_CAP_PPC_IRQ_LEVEL. 590 591Note that any value for 'irq' other than the ones stated above is invalid 592and incurs unexpected behavior. 593 594This is an asynchronous vcpu ioctl and can be invoked from any thread. 595 596MIPS: 597^^^^^ 598 599Queues an external interrupt to be injected into the virtual CPU. A negative 600interrupt number dequeues the interrupt. 601 602This is an asynchronous vcpu ioctl and can be invoked from any thread. 603 604 6054.17 KVM_DEBUG_GUEST 606-------------------- 607 608:Capability: basic 609:Architectures: none 610:Type: vcpu ioctl 611:Parameters: none) 612:Returns: -1 on error 613 614Support for this has been removed. Use KVM_SET_GUEST_DEBUG instead. 615 616 6174.18 KVM_GET_MSRS 618----------------- 619 620:Capability: basic (vcpu), KVM_CAP_GET_MSR_FEATURES (system) 621:Architectures: x86 622:Type: system ioctl, vcpu ioctl 623:Parameters: struct kvm_msrs (in/out) 624:Returns: number of msrs successfully returned; 625 -1 on error 626 627When used as a system ioctl: 628Reads the values of MSR-based features that are available for the VM. This 629is similar to KVM_GET_SUPPORTED_CPUID, but it returns MSR indices and values. 630The list of msr-based features can be obtained using KVM_GET_MSR_FEATURE_INDEX_LIST 631in a system ioctl. 632 633When used as a vcpu ioctl: 634Reads model-specific registers from the vcpu. Supported msr indices can 635be obtained using KVM_GET_MSR_INDEX_LIST in a system ioctl. 636 637:: 638 639 struct kvm_msrs { 640 __u32 nmsrs; /* number of msrs in entries */ 641 __u32 pad; 642 643 struct kvm_msr_entry entries[0]; 644 }; 645 646 struct kvm_msr_entry { 647 __u32 index; 648 __u32 reserved; 649 __u64 data; 650 }; 651 652Application code should set the 'nmsrs' member (which indicates the 653size of the entries array) and the 'index' member of each array entry. 654kvm will fill in the 'data' member. 655 656 6574.19 KVM_SET_MSRS 658----------------- 659 660:Capability: basic 661:Architectures: x86 662:Type: vcpu ioctl 663:Parameters: struct kvm_msrs (in) 664:Returns: number of msrs successfully set (see below), -1 on error 665 666Writes model-specific registers to the vcpu. See KVM_GET_MSRS for the 667data structures. 668 669Application code should set the 'nmsrs' member (which indicates the 670size of the entries array), and the 'index' and 'data' members of each 671array entry. 672 673It tries to set the MSRs in array entries[] one by one. If setting an MSR 674fails, e.g., due to setting reserved bits, the MSR isn't supported/emulated 675by KVM, etc..., it stops processing the MSR list and returns the number of 676MSRs that have been set successfully. 677 678 6794.20 KVM_SET_CPUID 680------------------ 681 682:Capability: basic 683:Architectures: x86 684:Type: vcpu ioctl 685:Parameters: struct kvm_cpuid (in) 686:Returns: 0 on success, -1 on error 687 688Defines the vcpu responses to the cpuid instruction. Applications 689should use the KVM_SET_CPUID2 ioctl if available. 690 691Caveat emptor: 692 - If this IOCTL fails, KVM gives no guarantees that previous valid CPUID 693 configuration (if there is) is not corrupted. Userspace can get a copy 694 of the resulting CPUID configuration through KVM_GET_CPUID2 in case. 695 - Using KVM_SET_CPUID{,2} after KVM_RUN, i.e. changing the guest vCPU model 696 after running the guest, may cause guest instability. 697 - Using heterogeneous CPUID configurations, modulo APIC IDs, topology, etc... 698 may cause guest instability. 699 700:: 701 702 struct kvm_cpuid_entry { 703 __u32 function; 704 __u32 eax; 705 __u32 ebx; 706 __u32 ecx; 707 __u32 edx; 708 __u32 padding; 709 }; 710 711 /* for KVM_SET_CPUID */ 712 struct kvm_cpuid { 713 __u32 nent; 714 __u32 padding; 715 struct kvm_cpuid_entry entries[0]; 716 }; 717 718 7194.21 KVM_SET_SIGNAL_MASK 720------------------------ 721 722:Capability: basic 723:Architectures: all 724:Type: vcpu ioctl 725:Parameters: struct kvm_signal_mask (in) 726:Returns: 0 on success, -1 on error 727 728Defines which signals are blocked during execution of KVM_RUN. This 729signal mask temporarily overrides the threads signal mask. Any 730unblocked signal received (except SIGKILL and SIGSTOP, which retain 731their traditional behaviour) will cause KVM_RUN to return with -EINTR. 732 733Note the signal will only be delivered if not blocked by the original 734signal mask. 735 736:: 737 738 /* for KVM_SET_SIGNAL_MASK */ 739 struct kvm_signal_mask { 740 __u32 len; 741 __u8 sigset[0]; 742 }; 743 744 7454.22 KVM_GET_FPU 746---------------- 747 748:Capability: basic 749:Architectures: x86 750:Type: vcpu ioctl 751:Parameters: struct kvm_fpu (out) 752:Returns: 0 on success, -1 on error 753 754Reads the floating point state from the vcpu. 755 756:: 757 758 /* for KVM_GET_FPU and KVM_SET_FPU */ 759 struct kvm_fpu { 760 __u8 fpr[8][16]; 761 __u16 fcw; 762 __u16 fsw; 763 __u8 ftwx; /* in fxsave format */ 764 __u8 pad1; 765 __u16 last_opcode; 766 __u64 last_ip; 767 __u64 last_dp; 768 __u8 xmm[16][16]; 769 __u32 mxcsr; 770 __u32 pad2; 771 }; 772 773 7744.23 KVM_SET_FPU 775---------------- 776 777:Capability: basic 778:Architectures: x86 779:Type: vcpu ioctl 780:Parameters: struct kvm_fpu (in) 781:Returns: 0 on success, -1 on error 782 783Writes the floating point state to the vcpu. 784 785:: 786 787 /* for KVM_GET_FPU and KVM_SET_FPU */ 788 struct kvm_fpu { 789 __u8 fpr[8][16]; 790 __u16 fcw; 791 __u16 fsw; 792 __u8 ftwx; /* in fxsave format */ 793 __u8 pad1; 794 __u16 last_opcode; 795 __u64 last_ip; 796 __u64 last_dp; 797 __u8 xmm[16][16]; 798 __u32 mxcsr; 799 __u32 pad2; 800 }; 801 802 8034.24 KVM_CREATE_IRQCHIP 804----------------------- 805 806:Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQCHIP (s390) 807:Architectures: x86, ARM, arm64, s390 808:Type: vm ioctl 809:Parameters: none 810:Returns: 0 on success, -1 on error 811 812Creates an interrupt controller model in the kernel. 813On x86, creates a virtual ioapic, a virtual PIC (two PICs, nested), and sets up 814future vcpus to have a local APIC. IRQ routing for GSIs 0-15 is set to both 815PIC and IOAPIC; GSI 16-23 only go to the IOAPIC. 816On ARM/arm64, a GICv2 is created. Any other GIC versions require the usage of 817KVM_CREATE_DEVICE, which also supports creating a GICv2. Using 818KVM_CREATE_DEVICE is preferred over KVM_CREATE_IRQCHIP for GICv2. 819On s390, a dummy irq routing table is created. 820 821Note that on s390 the KVM_CAP_S390_IRQCHIP vm capability needs to be enabled 822before KVM_CREATE_IRQCHIP can be used. 823 824 8254.25 KVM_IRQ_LINE 826----------------- 827 828:Capability: KVM_CAP_IRQCHIP 829:Architectures: x86, arm, arm64 830:Type: vm ioctl 831:Parameters: struct kvm_irq_level 832:Returns: 0 on success, -1 on error 833 834Sets the level of a GSI input to the interrupt controller model in the kernel. 835On some architectures it is required that an interrupt controller model has 836been previously created with KVM_CREATE_IRQCHIP. Note that edge-triggered 837interrupts require the level to be set to 1 and then back to 0. 838 839On real hardware, interrupt pins can be active-low or active-high. This 840does not matter for the level field of struct kvm_irq_level: 1 always 841means active (asserted), 0 means inactive (deasserted). 842 843x86 allows the operating system to program the interrupt polarity 844(active-low/active-high) for level-triggered interrupts, and KVM used 845to consider the polarity. However, due to bitrot in the handling of 846active-low interrupts, the above convention is now valid on x86 too. 847This is signaled by KVM_CAP_X86_IOAPIC_POLARITY_IGNORED. Userspace 848should not present interrupts to the guest as active-low unless this 849capability is present (or unless it is not using the in-kernel irqchip, 850of course). 851 852 853ARM/arm64 can signal an interrupt either at the CPU level, or at the 854in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to 855use PPIs designated for specific cpus. The irq field is interpreted 856like this:: 857 858 bits: | 31 ... 28 | 27 ... 24 | 23 ... 16 | 15 ... 0 | 859 field: | vcpu2_index | irq_type | vcpu_index | irq_id | 860 861The irq_type field has the following values: 862 863- irq_type[0]: 864 out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ 865- irq_type[1]: 866 in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.) 867 (the vcpu_index field is ignored) 868- irq_type[2]: 869 in-kernel GIC: PPI, irq_id between 16 and 31 (incl.) 870 871(The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs) 872 873In both cases, level is used to assert/deassert the line. 874 875When KVM_CAP_ARM_IRQ_LINE_LAYOUT_2 is supported, the target vcpu is 876identified as (256 * vcpu2_index + vcpu_index). Otherwise, vcpu2_index 877must be zero. 878 879Note that on arm/arm64, the KVM_CAP_IRQCHIP capability only conditions 880injection of interrupts for the in-kernel irqchip. KVM_IRQ_LINE can always 881be used for a userspace interrupt controller. 882 883:: 884 885 struct kvm_irq_level { 886 union { 887 __u32 irq; /* GSI */ 888 __s32 status; /* not used for KVM_IRQ_LEVEL */ 889 }; 890 __u32 level; /* 0 or 1 */ 891 }; 892 893 8944.26 KVM_GET_IRQCHIP 895-------------------- 896 897:Capability: KVM_CAP_IRQCHIP 898:Architectures: x86 899:Type: vm ioctl 900:Parameters: struct kvm_irqchip (in/out) 901:Returns: 0 on success, -1 on error 902 903Reads the state of a kernel interrupt controller created with 904KVM_CREATE_IRQCHIP into a buffer provided by the caller. 905 906:: 907 908 struct kvm_irqchip { 909 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */ 910 __u32 pad; 911 union { 912 char dummy[512]; /* reserving space */ 913 struct kvm_pic_state pic; 914 struct kvm_ioapic_state ioapic; 915 } chip; 916 }; 917 918 9194.27 KVM_SET_IRQCHIP 920-------------------- 921 922:Capability: KVM_CAP_IRQCHIP 923:Architectures: x86 924:Type: vm ioctl 925:Parameters: struct kvm_irqchip (in) 926:Returns: 0 on success, -1 on error 927 928Sets the state of a kernel interrupt controller created with 929KVM_CREATE_IRQCHIP from a buffer provided by the caller. 930 931:: 932 933 struct kvm_irqchip { 934 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */ 935 __u32 pad; 936 union { 937 char dummy[512]; /* reserving space */ 938 struct kvm_pic_state pic; 939 struct kvm_ioapic_state ioapic; 940 } chip; 941 }; 942 943 9444.28 KVM_XEN_HVM_CONFIG 945----------------------- 946 947:Capability: KVM_CAP_XEN_HVM 948:Architectures: x86 949:Type: vm ioctl 950:Parameters: struct kvm_xen_hvm_config (in) 951:Returns: 0 on success, -1 on error 952 953Sets the MSR that the Xen HVM guest uses to initialize its hypercall 954page, and provides the starting address and size of the hypercall 955blobs in userspace. When the guest writes the MSR, kvm copies one 956page of a blob (32- or 64-bit, depending on the vcpu mode) to guest 957memory. 958 959:: 960 961 struct kvm_xen_hvm_config { 962 __u32 flags; 963 __u32 msr; 964 __u64 blob_addr_32; 965 __u64 blob_addr_64; 966 __u8 blob_size_32; 967 __u8 blob_size_64; 968 __u8 pad2[30]; 969 }; 970 971If the KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL flag is returned from the 972KVM_CAP_XEN_HVM check, it may be set in the flags field of this ioctl. 973This requests KVM to generate the contents of the hypercall page 974automatically; hypercalls will be intercepted and passed to userspace 975through KVM_EXIT_XEN. In this case, all of the blob size and address 976fields must be zero. 977 978No other flags are currently valid in the struct kvm_xen_hvm_config. 979 9804.29 KVM_GET_CLOCK 981------------------ 982 983:Capability: KVM_CAP_ADJUST_CLOCK 984:Architectures: x86 985:Type: vm ioctl 986:Parameters: struct kvm_clock_data (out) 987:Returns: 0 on success, -1 on error 988 989Gets the current timestamp of kvmclock as seen by the current guest. In 990conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios 991such as migration. 992 993When KVM_CAP_ADJUST_CLOCK is passed to KVM_CHECK_EXTENSION, it returns the 994set of bits that KVM can return in struct kvm_clock_data's flag member. 995 996The only flag defined now is KVM_CLOCK_TSC_STABLE. If set, the returned 997value is the exact kvmclock value seen by all VCPUs at the instant 998when KVM_GET_CLOCK was called. If clear, the returned value is simply 999CLOCK_MONOTONIC plus a constant offset; the offset can be modified 1000with KVM_SET_CLOCK. KVM will try to make all VCPUs follow this clock, 1001but the exact value read by each VCPU could differ, because the host 1002TSC is not stable. 1003 1004:: 1005 1006 struct kvm_clock_data { 1007 __u64 clock; /* kvmclock current value */ 1008 __u32 flags; 1009 __u32 pad[9]; 1010 }; 1011 1012 10134.30 KVM_SET_CLOCK 1014------------------ 1015 1016:Capability: KVM_CAP_ADJUST_CLOCK 1017:Architectures: x86 1018:Type: vm ioctl 1019:Parameters: struct kvm_clock_data (in) 1020:Returns: 0 on success, -1 on error 1021 1022Sets the current timestamp of kvmclock to the value specified in its parameter. 1023In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios 1024such as migration. 1025 1026:: 1027 1028 struct kvm_clock_data { 1029 __u64 clock; /* kvmclock current value */ 1030 __u32 flags; 1031 __u32 pad[9]; 1032 }; 1033 1034 10354.31 KVM_GET_VCPU_EVENTS 1036------------------------ 1037 1038:Capability: KVM_CAP_VCPU_EVENTS 1039:Extended by: KVM_CAP_INTR_SHADOW 1040:Architectures: x86, arm, arm64 1041:Type: vcpu ioctl 1042:Parameters: struct kvm_vcpu_event (out) 1043:Returns: 0 on success, -1 on error 1044 1045X86: 1046^^^^ 1047 1048Gets currently pending exceptions, interrupts, and NMIs as well as related 1049states of the vcpu. 1050 1051:: 1052 1053 struct kvm_vcpu_events { 1054 struct { 1055 __u8 injected; 1056 __u8 nr; 1057 __u8 has_error_code; 1058 __u8 pending; 1059 __u32 error_code; 1060 } exception; 1061 struct { 1062 __u8 injected; 1063 __u8 nr; 1064 __u8 soft; 1065 __u8 shadow; 1066 } interrupt; 1067 struct { 1068 __u8 injected; 1069 __u8 pending; 1070 __u8 masked; 1071 __u8 pad; 1072 } nmi; 1073 __u32 sipi_vector; 1074 __u32 flags; 1075 struct { 1076 __u8 smm; 1077 __u8 pending; 1078 __u8 smm_inside_nmi; 1079 __u8 latched_init; 1080 } smi; 1081 __u8 reserved[27]; 1082 __u8 exception_has_payload; 1083 __u64 exception_payload; 1084 }; 1085 1086The following bits are defined in the flags field: 1087 1088- KVM_VCPUEVENT_VALID_SHADOW may be set to signal that 1089 interrupt.shadow contains a valid state. 1090 1091- KVM_VCPUEVENT_VALID_SMM may be set to signal that smi contains a 1092 valid state. 1093 1094- KVM_VCPUEVENT_VALID_PAYLOAD may be set to signal that the 1095 exception_has_payload, exception_payload, and exception.pending 1096 fields contain a valid state. This bit will be set whenever 1097 KVM_CAP_EXCEPTION_PAYLOAD is enabled. 1098 1099ARM/ARM64: 1100^^^^^^^^^^ 1101 1102If the guest accesses a device that is being emulated by the host kernel in 1103such a way that a real device would generate a physical SError, KVM may make 1104a virtual SError pending for that VCPU. This system error interrupt remains 1105pending until the guest takes the exception by unmasking PSTATE.A. 1106 1107Running the VCPU may cause it to take a pending SError, or make an access that 1108causes an SError to become pending. The event's description is only valid while 1109the VPCU is not running. 1110 1111This API provides a way to read and write the pending 'event' state that is not 1112visible to the guest. To save, restore or migrate a VCPU the struct representing 1113the state can be read then written using this GET/SET API, along with the other 1114guest-visible registers. It is not possible to 'cancel' an SError that has been 1115made pending. 1116 1117A device being emulated in user-space may also wish to generate an SError. To do 1118this the events structure can be populated by user-space. The current state 1119should be read first, to ensure no existing SError is pending. If an existing 1120SError is pending, the architecture's 'Multiple SError interrupts' rules should 1121be followed. (2.5.3 of DDI0587.a "ARM Reliability, Availability, and 1122Serviceability (RAS) Specification"). 1123 1124SError exceptions always have an ESR value. Some CPUs have the ability to 1125specify what the virtual SError's ESR value should be. These systems will 1126advertise KVM_CAP_ARM_INJECT_SERROR_ESR. In this case exception.has_esr will 1127always have a non-zero value when read, and the agent making an SError pending 1128should specify the ISS field in the lower 24 bits of exception.serror_esr. If 1129the system supports KVM_CAP_ARM_INJECT_SERROR_ESR, but user-space sets the events 1130with exception.has_esr as zero, KVM will choose an ESR. 1131 1132Specifying exception.has_esr on a system that does not support it will return 1133-EINVAL. Setting anything other than the lower 24bits of exception.serror_esr 1134will return -EINVAL. 1135 1136It is not possible to read back a pending external abort (injected via 1137KVM_SET_VCPU_EVENTS or otherwise) because such an exception is always delivered 1138directly to the virtual CPU). 1139 1140:: 1141 1142 struct kvm_vcpu_events { 1143 struct { 1144 __u8 serror_pending; 1145 __u8 serror_has_esr; 1146 __u8 ext_dabt_pending; 1147 /* Align it to 8 bytes */ 1148 __u8 pad[5]; 1149 __u64 serror_esr; 1150 } exception; 1151 __u32 reserved[12]; 1152 }; 1153 11544.32 KVM_SET_VCPU_EVENTS 1155------------------------ 1156 1157:Capability: KVM_CAP_VCPU_EVENTS 1158:Extended by: KVM_CAP_INTR_SHADOW 1159:Architectures: x86, arm, arm64 1160:Type: vcpu ioctl 1161:Parameters: struct kvm_vcpu_event (in) 1162:Returns: 0 on success, -1 on error 1163 1164X86: 1165^^^^ 1166 1167Set pending exceptions, interrupts, and NMIs as well as related states of the 1168vcpu. 1169 1170See KVM_GET_VCPU_EVENTS for the data structure. 1171 1172Fields that may be modified asynchronously by running VCPUs can be excluded 1173from the update. These fields are nmi.pending, sipi_vector, smi.smm, 1174smi.pending. Keep the corresponding bits in the flags field cleared to 1175suppress overwriting the current in-kernel state. The bits are: 1176 1177=============================== ================================== 1178KVM_VCPUEVENT_VALID_NMI_PENDING transfer nmi.pending to the kernel 1179KVM_VCPUEVENT_VALID_SIPI_VECTOR transfer sipi_vector 1180KVM_VCPUEVENT_VALID_SMM transfer the smi sub-struct. 1181=============================== ================================== 1182 1183If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in 1184the flags field to signal that interrupt.shadow contains a valid state and 1185shall be written into the VCPU. 1186 1187KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available. 1188 1189If KVM_CAP_EXCEPTION_PAYLOAD is enabled, KVM_VCPUEVENT_VALID_PAYLOAD 1190can be set in the flags field to signal that the 1191exception_has_payload, exception_payload, and exception.pending fields 1192contain a valid state and shall be written into the VCPU. 1193 1194ARM/ARM64: 1195^^^^^^^^^^ 1196 1197User space may need to inject several types of events to the guest. 1198 1199Set the pending SError exception state for this VCPU. It is not possible to 1200'cancel' an Serror that has been made pending. 1201 1202If the guest performed an access to I/O memory which could not be handled by 1203userspace, for example because of missing instruction syndrome decode 1204information or because there is no device mapped at the accessed IPA, then 1205userspace can ask the kernel to inject an external abort using the address 1206from the exiting fault on the VCPU. It is a programming error to set 1207ext_dabt_pending after an exit which was not either KVM_EXIT_MMIO or 1208KVM_EXIT_ARM_NISV. This feature is only available if the system supports 1209KVM_CAP_ARM_INJECT_EXT_DABT. This is a helper which provides commonality in 1210how userspace reports accesses for the above cases to guests, across different 1211userspace implementations. Nevertheless, userspace can still emulate all Arm 1212exceptions by manipulating individual registers using the KVM_SET_ONE_REG API. 1213 1214See KVM_GET_VCPU_EVENTS for the data structure. 1215 1216 12174.33 KVM_GET_DEBUGREGS 1218---------------------- 1219 1220:Capability: KVM_CAP_DEBUGREGS 1221:Architectures: x86 1222:Type: vm ioctl 1223:Parameters: struct kvm_debugregs (out) 1224:Returns: 0 on success, -1 on error 1225 1226Reads debug registers from the vcpu. 1227 1228:: 1229 1230 struct kvm_debugregs { 1231 __u64 db[4]; 1232 __u64 dr6; 1233 __u64 dr7; 1234 __u64 flags; 1235 __u64 reserved[9]; 1236 }; 1237 1238 12394.34 KVM_SET_DEBUGREGS 1240---------------------- 1241 1242:Capability: KVM_CAP_DEBUGREGS 1243:Architectures: x86 1244:Type: vm ioctl 1245:Parameters: struct kvm_debugregs (in) 1246:Returns: 0 on success, -1 on error 1247 1248Writes debug registers into the vcpu. 1249 1250See KVM_GET_DEBUGREGS for the data structure. The flags field is unused 1251yet and must be cleared on entry. 1252 1253 12544.35 KVM_SET_USER_MEMORY_REGION 1255------------------------------- 1256 1257:Capability: KVM_CAP_USER_MEMORY 1258:Architectures: all 1259:Type: vm ioctl 1260:Parameters: struct kvm_userspace_memory_region (in) 1261:Returns: 0 on success, -1 on error 1262 1263:: 1264 1265 struct kvm_userspace_memory_region { 1266 __u32 slot; 1267 __u32 flags; 1268 __u64 guest_phys_addr; 1269 __u64 memory_size; /* bytes */ 1270 __u64 userspace_addr; /* start of the userspace allocated memory */ 1271 }; 1272 1273 /* for kvm_memory_region::flags */ 1274 #define KVM_MEM_LOG_DIRTY_PAGES (1UL << 0) 1275 #define KVM_MEM_READONLY (1UL << 1) 1276 1277This ioctl allows the user to create, modify or delete a guest physical 1278memory slot. Bits 0-15 of "slot" specify the slot id and this value 1279should be less than the maximum number of user memory slots supported per 1280VM. The maximum allowed slots can be queried using KVM_CAP_NR_MEMSLOTS. 1281Slots may not overlap in guest physical address space. 1282 1283If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot" 1284specifies the address space which is being modified. They must be 1285less than the value that KVM_CHECK_EXTENSION returns for the 1286KVM_CAP_MULTI_ADDRESS_SPACE capability. Slots in separate address spaces 1287are unrelated; the restriction on overlapping slots only applies within 1288each address space. 1289 1290Deleting a slot is done by passing zero for memory_size. When changing 1291an existing slot, it may be moved in the guest physical memory space, 1292or its flags may be modified, but it may not be resized. 1293 1294Memory for the region is taken starting at the address denoted by the 1295field userspace_addr, which must point at user addressable memory for 1296the entire memory slot size. Any object may back this memory, including 1297anonymous memory, ordinary files, and hugetlbfs. 1298 1299On architectures that support a form of address tagging, userspace_addr must 1300be an untagged address. 1301 1302It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr 1303be identical. This allows large pages in the guest to be backed by large 1304pages in the host. 1305 1306The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and 1307KVM_MEM_READONLY. The former can be set to instruct KVM to keep track of 1308writes to memory within the slot. See KVM_GET_DIRTY_LOG ioctl to know how to 1309use it. The latter can be set, if KVM_CAP_READONLY_MEM capability allows it, 1310to make a new slot read-only. In this case, writes to this memory will be 1311posted to userspace as KVM_EXIT_MMIO exits. 1312 1313When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of 1314the memory region are automatically reflected into the guest. For example, an 1315mmap() that affects the region will be made visible immediately. Another 1316example is madvise(MADV_DROP). 1317 1318It is recommended to use this API instead of the KVM_SET_MEMORY_REGION ioctl. 1319The KVM_SET_MEMORY_REGION does not allow fine grained control over memory 1320allocation and is deprecated. 1321 1322 13234.36 KVM_SET_TSS_ADDR 1324--------------------- 1325 1326:Capability: KVM_CAP_SET_TSS_ADDR 1327:Architectures: x86 1328:Type: vm ioctl 1329:Parameters: unsigned long tss_address (in) 1330:Returns: 0 on success, -1 on error 1331 1332This ioctl defines the physical address of a three-page region in the guest 1333physical address space. The region must be within the first 4GB of the 1334guest physical address space and must not conflict with any memory slot 1335or any mmio address. The guest may malfunction if it accesses this memory 1336region. 1337 1338This ioctl is required on Intel-based hosts. This is needed on Intel hardware 1339because of a quirk in the virtualization implementation (see the internals 1340documentation when it pops into existence). 1341 1342 13434.37 KVM_ENABLE_CAP 1344------------------- 1345 1346:Capability: KVM_CAP_ENABLE_CAP 1347:Architectures: mips, ppc, s390 1348:Type: vcpu ioctl 1349:Parameters: struct kvm_enable_cap (in) 1350:Returns: 0 on success; -1 on error 1351 1352:Capability: KVM_CAP_ENABLE_CAP_VM 1353:Architectures: all 1354:Type: vm ioctl 1355:Parameters: struct kvm_enable_cap (in) 1356:Returns: 0 on success; -1 on error 1357 1358.. note:: 1359 1360 Not all extensions are enabled by default. Using this ioctl the application 1361 can enable an extension, making it available to the guest. 1362 1363On systems that do not support this ioctl, it always fails. On systems that 1364do support it, it only works for extensions that are supported for enablement. 1365 1366To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should 1367be used. 1368 1369:: 1370 1371 struct kvm_enable_cap { 1372 /* in */ 1373 __u32 cap; 1374 1375The capability that is supposed to get enabled. 1376 1377:: 1378 1379 __u32 flags; 1380 1381A bitfield indicating future enhancements. Has to be 0 for now. 1382 1383:: 1384 1385 __u64 args[4]; 1386 1387Arguments for enabling a feature. If a feature needs initial values to 1388function properly, this is the place to put them. 1389 1390:: 1391 1392 __u8 pad[64]; 1393 }; 1394 1395The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl 1396for vm-wide capabilities. 1397 13984.38 KVM_GET_MP_STATE 1399--------------------- 1400 1401:Capability: KVM_CAP_MP_STATE 1402:Architectures: x86, s390, arm, arm64 1403:Type: vcpu ioctl 1404:Parameters: struct kvm_mp_state (out) 1405:Returns: 0 on success; -1 on error 1406 1407:: 1408 1409 struct kvm_mp_state { 1410 __u32 mp_state; 1411 }; 1412 1413Returns the vcpu's current "multiprocessing state" (though also valid on 1414uniprocessor guests). 1415 1416Possible values are: 1417 1418 ========================== =============================================== 1419 KVM_MP_STATE_RUNNABLE the vcpu is currently running [x86,arm/arm64] 1420 KVM_MP_STATE_UNINITIALIZED the vcpu is an application processor (AP) 1421 which has not yet received an INIT signal [x86] 1422 KVM_MP_STATE_INIT_RECEIVED the vcpu has received an INIT signal, and is 1423 now ready for a SIPI [x86] 1424 KVM_MP_STATE_HALTED the vcpu has executed a HLT instruction and 1425 is waiting for an interrupt [x86] 1426 KVM_MP_STATE_SIPI_RECEIVED the vcpu has just received a SIPI (vector 1427 accessible via KVM_GET_VCPU_EVENTS) [x86] 1428 KVM_MP_STATE_STOPPED the vcpu is stopped [s390,arm/arm64] 1429 KVM_MP_STATE_CHECK_STOP the vcpu is in a special error state [s390] 1430 KVM_MP_STATE_OPERATING the vcpu is operating (running or halted) 1431 [s390] 1432 KVM_MP_STATE_LOAD the vcpu is in a special load/startup state 1433 [s390] 1434 ========================== =============================================== 1435 1436On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an 1437in-kernel irqchip, the multiprocessing state must be maintained by userspace on 1438these architectures. 1439 1440For arm/arm64: 1441^^^^^^^^^^^^^^ 1442 1443The only states that are valid are KVM_MP_STATE_STOPPED and 1444KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not. 1445 14464.39 KVM_SET_MP_STATE 1447--------------------- 1448 1449:Capability: KVM_CAP_MP_STATE 1450:Architectures: x86, s390, arm, arm64 1451:Type: vcpu ioctl 1452:Parameters: struct kvm_mp_state (in) 1453:Returns: 0 on success; -1 on error 1454 1455Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for 1456arguments. 1457 1458On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an 1459in-kernel irqchip, the multiprocessing state must be maintained by userspace on 1460these architectures. 1461 1462For arm/arm64: 1463^^^^^^^^^^^^^^ 1464 1465The only states that are valid are KVM_MP_STATE_STOPPED and 1466KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not. 1467 14684.40 KVM_SET_IDENTITY_MAP_ADDR 1469------------------------------ 1470 1471:Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR 1472:Architectures: x86 1473:Type: vm ioctl 1474:Parameters: unsigned long identity (in) 1475:Returns: 0 on success, -1 on error 1476 1477This ioctl defines the physical address of a one-page region in the guest 1478physical address space. The region must be within the first 4GB of the 1479guest physical address space and must not conflict with any memory slot 1480or any mmio address. The guest may malfunction if it accesses this memory 1481region. 1482 1483Setting the address to 0 will result in resetting the address to its default 1484(0xfffbc000). 1485 1486This ioctl is required on Intel-based hosts. This is needed on Intel hardware 1487because of a quirk in the virtualization implementation (see the internals 1488documentation when it pops into existence). 1489 1490Fails if any VCPU has already been created. 1491 14924.41 KVM_SET_BOOT_CPU_ID 1493------------------------ 1494 1495:Capability: KVM_CAP_SET_BOOT_CPU_ID 1496:Architectures: x86 1497:Type: vm ioctl 1498:Parameters: unsigned long vcpu_id 1499:Returns: 0 on success, -1 on error 1500 1501Define which vcpu is the Bootstrap Processor (BSP). Values are the same 1502as the vcpu id in KVM_CREATE_VCPU. If this ioctl is not called, the default 1503is vcpu 0. This ioctl has to be called before vcpu creation, 1504otherwise it will return EBUSY error. 1505 1506 15074.42 KVM_GET_XSAVE 1508------------------ 1509 1510:Capability: KVM_CAP_XSAVE 1511:Architectures: x86 1512:Type: vcpu ioctl 1513:Parameters: struct kvm_xsave (out) 1514:Returns: 0 on success, -1 on error 1515 1516 1517:: 1518 1519 struct kvm_xsave { 1520 __u32 region[1024]; 1521 }; 1522 1523This ioctl would copy current vcpu's xsave struct to the userspace. 1524 1525 15264.43 KVM_SET_XSAVE 1527------------------ 1528 1529:Capability: KVM_CAP_XSAVE 1530:Architectures: x86 1531:Type: vcpu ioctl 1532:Parameters: struct kvm_xsave (in) 1533:Returns: 0 on success, -1 on error 1534 1535:: 1536 1537 1538 struct kvm_xsave { 1539 __u32 region[1024]; 1540 }; 1541 1542This ioctl would copy userspace's xsave struct to the kernel. 1543 1544 15454.44 KVM_GET_XCRS 1546----------------- 1547 1548:Capability: KVM_CAP_XCRS 1549:Architectures: x86 1550:Type: vcpu ioctl 1551:Parameters: struct kvm_xcrs (out) 1552:Returns: 0 on success, -1 on error 1553 1554:: 1555 1556 struct kvm_xcr { 1557 __u32 xcr; 1558 __u32 reserved; 1559 __u64 value; 1560 }; 1561 1562 struct kvm_xcrs { 1563 __u32 nr_xcrs; 1564 __u32 flags; 1565 struct kvm_xcr xcrs[KVM_MAX_XCRS]; 1566 __u64 padding[16]; 1567 }; 1568 1569This ioctl would copy current vcpu's xcrs to the userspace. 1570 1571 15724.45 KVM_SET_XCRS 1573----------------- 1574 1575:Capability: KVM_CAP_XCRS 1576:Architectures: x86 1577:Type: vcpu ioctl 1578:Parameters: struct kvm_xcrs (in) 1579:Returns: 0 on success, -1 on error 1580 1581:: 1582 1583 struct kvm_xcr { 1584 __u32 xcr; 1585 __u32 reserved; 1586 __u64 value; 1587 }; 1588 1589 struct kvm_xcrs { 1590 __u32 nr_xcrs; 1591 __u32 flags; 1592 struct kvm_xcr xcrs[KVM_MAX_XCRS]; 1593 __u64 padding[16]; 1594 }; 1595 1596This ioctl would set vcpu's xcr to the value userspace specified. 1597 1598 15994.46 KVM_GET_SUPPORTED_CPUID 1600---------------------------- 1601 1602:Capability: KVM_CAP_EXT_CPUID 1603:Architectures: x86 1604:Type: system ioctl 1605:Parameters: struct kvm_cpuid2 (in/out) 1606:Returns: 0 on success, -1 on error 1607 1608:: 1609 1610 struct kvm_cpuid2 { 1611 __u32 nent; 1612 __u32 padding; 1613 struct kvm_cpuid_entry2 entries[0]; 1614 }; 1615 1616 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0) 1617 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */ 1618 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */ 1619 1620 struct kvm_cpuid_entry2 { 1621 __u32 function; 1622 __u32 index; 1623 __u32 flags; 1624 __u32 eax; 1625 __u32 ebx; 1626 __u32 ecx; 1627 __u32 edx; 1628 __u32 padding[3]; 1629 }; 1630 1631This ioctl returns x86 cpuid features which are supported by both the 1632hardware and kvm in its default configuration. Userspace can use the 1633information returned by this ioctl to construct cpuid information (for 1634KVM_SET_CPUID2) that is consistent with hardware, kernel, and 1635userspace capabilities, and with user requirements (for example, the 1636user may wish to constrain cpuid to emulate older hardware, or for 1637feature consistency across a cluster). 1638 1639Note that certain capabilities, such as KVM_CAP_X86_DISABLE_EXITS, may 1640expose cpuid features (e.g. MONITOR) which are not supported by kvm in 1641its default configuration. If userspace enables such capabilities, it 1642is responsible for modifying the results of this ioctl appropriately. 1643 1644Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure 1645with the 'nent' field indicating the number of entries in the variable-size 1646array 'entries'. If the number of entries is too low to describe the cpu 1647capabilities, an error (E2BIG) is returned. If the number is too high, 1648the 'nent' field is adjusted and an error (ENOMEM) is returned. If the 1649number is just right, the 'nent' field is adjusted to the number of valid 1650entries in the 'entries' array, which is then filled. 1651 1652The entries returned are the host cpuid as returned by the cpuid instruction, 1653with unknown or unsupported features masked out. Some features (for example, 1654x2apic), may not be present in the host cpu, but are exposed by kvm if it can 1655emulate them efficiently. The fields in each entry are defined as follows: 1656 1657 function: 1658 the eax value used to obtain the entry 1659 1660 index: 1661 the ecx value used to obtain the entry (for entries that are 1662 affected by ecx) 1663 1664 flags: 1665 an OR of zero or more of the following: 1666 1667 KVM_CPUID_FLAG_SIGNIFCANT_INDEX: 1668 if the index field is valid 1669 1670 eax, ebx, ecx, edx: 1671 the values returned by the cpuid instruction for 1672 this function/index combination 1673 1674The TSC deadline timer feature (CPUID leaf 1, ecx[24]) is always returned 1675as false, since the feature depends on KVM_CREATE_IRQCHIP for local APIC 1676support. Instead it is reported via:: 1677 1678 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER) 1679 1680if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the 1681feature in userspace, then you can enable the feature for KVM_SET_CPUID2. 1682 1683 16844.47 KVM_PPC_GET_PVINFO 1685----------------------- 1686 1687:Capability: KVM_CAP_PPC_GET_PVINFO 1688:Architectures: ppc 1689:Type: vm ioctl 1690:Parameters: struct kvm_ppc_pvinfo (out) 1691:Returns: 0 on success, !0 on error 1692 1693:: 1694 1695 struct kvm_ppc_pvinfo { 1696 __u32 flags; 1697 __u32 hcall[4]; 1698 __u8 pad[108]; 1699 }; 1700 1701This ioctl fetches PV specific information that need to be passed to the guest 1702using the device tree or other means from vm context. 1703 1704The hcall array defines 4 instructions that make up a hypercall. 1705 1706If any additional field gets added to this structure later on, a bit for that 1707additional piece of information will be set in the flags bitmap. 1708 1709The flags bitmap is defined as:: 1710 1711 /* the host supports the ePAPR idle hcall 1712 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1<<0) 1713 17144.52 KVM_SET_GSI_ROUTING 1715------------------------ 1716 1717:Capability: KVM_CAP_IRQ_ROUTING 1718:Architectures: x86 s390 arm arm64 1719:Type: vm ioctl 1720:Parameters: struct kvm_irq_routing (in) 1721:Returns: 0 on success, -1 on error 1722 1723Sets the GSI routing table entries, overwriting any previously set entries. 1724 1725On arm/arm64, GSI routing has the following limitation: 1726 1727- GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD. 1728 1729:: 1730 1731 struct kvm_irq_routing { 1732 __u32 nr; 1733 __u32 flags; 1734 struct kvm_irq_routing_entry entries[0]; 1735 }; 1736 1737No flags are specified so far, the corresponding field must be set to zero. 1738 1739:: 1740 1741 struct kvm_irq_routing_entry { 1742 __u32 gsi; 1743 __u32 type; 1744 __u32 flags; 1745 __u32 pad; 1746 union { 1747 struct kvm_irq_routing_irqchip irqchip; 1748 struct kvm_irq_routing_msi msi; 1749 struct kvm_irq_routing_s390_adapter adapter; 1750 struct kvm_irq_routing_hv_sint hv_sint; 1751 __u32 pad[8]; 1752 } u; 1753 }; 1754 1755 /* gsi routing entry types */ 1756 #define KVM_IRQ_ROUTING_IRQCHIP 1 1757 #define KVM_IRQ_ROUTING_MSI 2 1758 #define KVM_IRQ_ROUTING_S390_ADAPTER 3 1759 #define KVM_IRQ_ROUTING_HV_SINT 4 1760 1761flags: 1762 1763- KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry 1764 type, specifies that the devid field contains a valid value. The per-VM 1765 KVM_CAP_MSI_DEVID capability advertises the requirement to provide 1766 the device ID. If this capability is not available, userspace should 1767 never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail. 1768- zero otherwise 1769 1770:: 1771 1772 struct kvm_irq_routing_irqchip { 1773 __u32 irqchip; 1774 __u32 pin; 1775 }; 1776 1777 struct kvm_irq_routing_msi { 1778 __u32 address_lo; 1779 __u32 address_hi; 1780 __u32 data; 1781 union { 1782 __u32 pad; 1783 __u32 devid; 1784 }; 1785 }; 1786 1787If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier 1788for the device that wrote the MSI message. For PCI, this is usually a 1789BFD identifier in the lower 16 bits. 1790 1791On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS 1792feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled, 1793address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of 1794address_hi must be zero. 1795 1796:: 1797 1798 struct kvm_irq_routing_s390_adapter { 1799 __u64 ind_addr; 1800 __u64 summary_addr; 1801 __u64 ind_offset; 1802 __u32 summary_offset; 1803 __u32 adapter_id; 1804 }; 1805 1806 struct kvm_irq_routing_hv_sint { 1807 __u32 vcpu; 1808 __u32 sint; 1809 }; 1810 1811 18124.55 KVM_SET_TSC_KHZ 1813-------------------- 1814 1815:Capability: KVM_CAP_TSC_CONTROL 1816:Architectures: x86 1817:Type: vcpu ioctl 1818:Parameters: virtual tsc_khz 1819:Returns: 0 on success, -1 on error 1820 1821Specifies the tsc frequency for the virtual machine. The unit of the 1822frequency is KHz. 1823 1824 18254.56 KVM_GET_TSC_KHZ 1826-------------------- 1827 1828:Capability: KVM_CAP_GET_TSC_KHZ 1829:Architectures: x86 1830:Type: vcpu ioctl 1831:Parameters: none 1832:Returns: virtual tsc-khz on success, negative value on error 1833 1834Returns the tsc frequency of the guest. The unit of the return value is 1835KHz. If the host has unstable tsc this ioctl returns -EIO instead as an 1836error. 1837 1838 18394.57 KVM_GET_LAPIC 1840------------------ 1841 1842:Capability: KVM_CAP_IRQCHIP 1843:Architectures: x86 1844:Type: vcpu ioctl 1845:Parameters: struct kvm_lapic_state (out) 1846:Returns: 0 on success, -1 on error 1847 1848:: 1849 1850 #define KVM_APIC_REG_SIZE 0x400 1851 struct kvm_lapic_state { 1852 char regs[KVM_APIC_REG_SIZE]; 1853 }; 1854 1855Reads the Local APIC registers and copies them into the input argument. The 1856data format and layout are the same as documented in the architecture manual. 1857 1858If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is 1859enabled, then the format of APIC_ID register depends on the APIC mode 1860(reported by MSR_IA32_APICBASE) of its VCPU. x2APIC stores APIC ID in 1861the APIC_ID register (bytes 32-35). xAPIC only allows an 8-bit APIC ID 1862which is stored in bits 31-24 of the APIC register, or equivalently in 1863byte 35 of struct kvm_lapic_state's regs field. KVM_GET_LAPIC must then 1864be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR. 1865 1866If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state 1867always uses xAPIC format. 1868 1869 18704.58 KVM_SET_LAPIC 1871------------------ 1872 1873:Capability: KVM_CAP_IRQCHIP 1874:Architectures: x86 1875:Type: vcpu ioctl 1876:Parameters: struct kvm_lapic_state (in) 1877:Returns: 0 on success, -1 on error 1878 1879:: 1880 1881 #define KVM_APIC_REG_SIZE 0x400 1882 struct kvm_lapic_state { 1883 char regs[KVM_APIC_REG_SIZE]; 1884 }; 1885 1886Copies the input argument into the Local APIC registers. The data format 1887and layout are the same as documented in the architecture manual. 1888 1889The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's 1890regs field) depends on the state of the KVM_CAP_X2APIC_API capability. 1891See the note in KVM_GET_LAPIC. 1892 1893 18944.59 KVM_IOEVENTFD 1895------------------ 1896 1897:Capability: KVM_CAP_IOEVENTFD 1898:Architectures: all 1899:Type: vm ioctl 1900:Parameters: struct kvm_ioeventfd (in) 1901:Returns: 0 on success, !0 on error 1902 1903This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address 1904within the guest. A guest write in the registered address will signal the 1905provided event instead of triggering an exit. 1906 1907:: 1908 1909 struct kvm_ioeventfd { 1910 __u64 datamatch; 1911 __u64 addr; /* legal pio/mmio address */ 1912 __u32 len; /* 0, 1, 2, 4, or 8 bytes */ 1913 __s32 fd; 1914 __u32 flags; 1915 __u8 pad[36]; 1916 }; 1917 1918For the special case of virtio-ccw devices on s390, the ioevent is matched 1919to a subchannel/virtqueue tuple instead. 1920 1921The following flags are defined:: 1922 1923 #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch) 1924 #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio) 1925 #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign) 1926 #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \ 1927 (1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify) 1928 1929If datamatch flag is set, the event will be signaled only if the written value 1930to the registered address is equal to datamatch in struct kvm_ioeventfd. 1931 1932For virtio-ccw devices, addr contains the subchannel id and datamatch the 1933virtqueue index. 1934 1935With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and 1936the kernel will ignore the length of guest write and may get a faster vmexit. 1937The speedup may only apply to specific architectures, but the ioeventfd will 1938work anyway. 1939 19404.60 KVM_DIRTY_TLB 1941------------------ 1942 1943:Capability: KVM_CAP_SW_TLB 1944:Architectures: ppc 1945:Type: vcpu ioctl 1946:Parameters: struct kvm_dirty_tlb (in) 1947:Returns: 0 on success, -1 on error 1948 1949:: 1950 1951 struct kvm_dirty_tlb { 1952 __u64 bitmap; 1953 __u32 num_dirty; 1954 }; 1955 1956This must be called whenever userspace has changed an entry in the shared 1957TLB, prior to calling KVM_RUN on the associated vcpu. 1958 1959The "bitmap" field is the userspace address of an array. This array 1960consists of a number of bits, equal to the total number of TLB entries as 1961determined by the last successful call to KVM_CONFIG_TLB, rounded up to the 1962nearest multiple of 64. 1963 1964Each bit corresponds to one TLB entry, ordered the same as in the shared TLB 1965array. 1966 1967The array is little-endian: the bit 0 is the least significant bit of the 1968first byte, bit 8 is the least significant bit of the second byte, etc. 1969This avoids any complications with differing word sizes. 1970 1971The "num_dirty" field is a performance hint for KVM to determine whether it 1972should skip processing the bitmap and just invalidate everything. It must 1973be set to the number of set bits in the bitmap. 1974 1975 19764.62 KVM_CREATE_SPAPR_TCE 1977------------------------- 1978 1979:Capability: KVM_CAP_SPAPR_TCE 1980:Architectures: powerpc 1981:Type: vm ioctl 1982:Parameters: struct kvm_create_spapr_tce (in) 1983:Returns: file descriptor for manipulating the created TCE table 1984 1985This creates a virtual TCE (translation control entry) table, which 1986is an IOMMU for PAPR-style virtual I/O. It is used to translate 1987logical addresses used in virtual I/O into guest physical addresses, 1988and provides a scatter/gather capability for PAPR virtual I/O. 1989 1990:: 1991 1992 /* for KVM_CAP_SPAPR_TCE */ 1993 struct kvm_create_spapr_tce { 1994 __u64 liobn; 1995 __u32 window_size; 1996 }; 1997 1998The liobn field gives the logical IO bus number for which to create a 1999TCE table. The window_size field specifies the size of the DMA window 2000which this TCE table will translate - the table will contain one 64 2001bit TCE entry for every 4kiB of the DMA window. 2002 2003When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE 2004table has been created using this ioctl(), the kernel will handle it 2005in real mode, updating the TCE table. H_PUT_TCE calls for other 2006liobns will cause a vm exit and must be handled by userspace. 2007 2008The return value is a file descriptor which can be passed to mmap(2) 2009to map the created TCE table into userspace. This lets userspace read 2010the entries written by kernel-handled H_PUT_TCE calls, and also lets 2011userspace update the TCE table directly which is useful in some 2012circumstances. 2013 2014 20154.63 KVM_ALLOCATE_RMA 2016--------------------- 2017 2018:Capability: KVM_CAP_PPC_RMA 2019:Architectures: powerpc 2020:Type: vm ioctl 2021:Parameters: struct kvm_allocate_rma (out) 2022:Returns: file descriptor for mapping the allocated RMA 2023 2024This allocates a Real Mode Area (RMA) from the pool allocated at boot 2025time by the kernel. An RMA is a physically-contiguous, aligned region 2026of memory used on older POWER processors to provide the memory which 2027will be accessed by real-mode (MMU off) accesses in a KVM guest. 2028POWER processors support a set of sizes for the RMA that usually 2029includes 64MB, 128MB, 256MB and some larger powers of two. 2030 2031:: 2032 2033 /* for KVM_ALLOCATE_RMA */ 2034 struct kvm_allocate_rma { 2035 __u64 rma_size; 2036 }; 2037 2038The return value is a file descriptor which can be passed to mmap(2) 2039to map the allocated RMA into userspace. The mapped area can then be 2040passed to the KVM_SET_USER_MEMORY_REGION ioctl to establish it as the 2041RMA for a virtual machine. The size of the RMA in bytes (which is 2042fixed at host kernel boot time) is returned in the rma_size field of 2043the argument structure. 2044 2045The KVM_CAP_PPC_RMA capability is 1 or 2 if the KVM_ALLOCATE_RMA ioctl 2046is supported; 2 if the processor requires all virtual machines to have 2047an RMA, or 1 if the processor can use an RMA but doesn't require it, 2048because it supports the Virtual RMA (VRMA) facility. 2049 2050 20514.64 KVM_NMI 2052------------ 2053 2054:Capability: KVM_CAP_USER_NMI 2055:Architectures: x86 2056:Type: vcpu ioctl 2057:Parameters: none 2058:Returns: 0 on success, -1 on error 2059 2060Queues an NMI on the thread's vcpu. Note this is well defined only 2061when KVM_CREATE_IRQCHIP has not been called, since this is an interface 2062between the virtual cpu core and virtual local APIC. After KVM_CREATE_IRQCHIP 2063has been called, this interface is completely emulated within the kernel. 2064 2065To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the 2066following algorithm: 2067 2068 - pause the vcpu 2069 - read the local APIC's state (KVM_GET_LAPIC) 2070 - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1) 2071 - if so, issue KVM_NMI 2072 - resume the vcpu 2073 2074Some guests configure the LINT1 NMI input to cause a panic, aiding in 2075debugging. 2076 2077 20784.65 KVM_S390_UCAS_MAP 2079---------------------- 2080 2081:Capability: KVM_CAP_S390_UCONTROL 2082:Architectures: s390 2083:Type: vcpu ioctl 2084:Parameters: struct kvm_s390_ucas_mapping (in) 2085:Returns: 0 in case of success 2086 2087The parameter is defined like this:: 2088 2089 struct kvm_s390_ucas_mapping { 2090 __u64 user_addr; 2091 __u64 vcpu_addr; 2092 __u64 length; 2093 }; 2094 2095This ioctl maps the memory at "user_addr" with the length "length" to 2096the vcpu's address space starting at "vcpu_addr". All parameters need to 2097be aligned by 1 megabyte. 2098 2099 21004.66 KVM_S390_UCAS_UNMAP 2101------------------------ 2102 2103:Capability: KVM_CAP_S390_UCONTROL 2104:Architectures: s390 2105:Type: vcpu ioctl 2106:Parameters: struct kvm_s390_ucas_mapping (in) 2107:Returns: 0 in case of success 2108 2109The parameter is defined like this:: 2110 2111 struct kvm_s390_ucas_mapping { 2112 __u64 user_addr; 2113 __u64 vcpu_addr; 2114 __u64 length; 2115 }; 2116 2117This ioctl unmaps the memory in the vcpu's address space starting at 2118"vcpu_addr" with the length "length". The field "user_addr" is ignored. 2119All parameters need to be aligned by 1 megabyte. 2120 2121 21224.67 KVM_S390_VCPU_FAULT 2123------------------------ 2124 2125:Capability: KVM_CAP_S390_UCONTROL 2126:Architectures: s390 2127:Type: vcpu ioctl 2128:Parameters: vcpu absolute address (in) 2129:Returns: 0 in case of success 2130 2131This call creates a page table entry on the virtual cpu's address space 2132(for user controlled virtual machines) or the virtual machine's address 2133space (for regular virtual machines). This only works for minor faults, 2134thus it's recommended to access subject memory page via the user page 2135table upfront. This is useful to handle validity intercepts for user 2136controlled virtual machines to fault in the virtual cpu's lowcore pages 2137prior to calling the KVM_RUN ioctl. 2138 2139 21404.68 KVM_SET_ONE_REG 2141-------------------- 2142 2143:Capability: KVM_CAP_ONE_REG 2144:Architectures: all 2145:Type: vcpu ioctl 2146:Parameters: struct kvm_one_reg (in) 2147:Returns: 0 on success, negative value on failure 2148 2149Errors: 2150 2151 ====== ============================================================ 2152 ENOENT no such register 2153 EINVAL invalid register ID, or no such register or used with VMs in 2154 protected virtualization mode on s390 2155 EPERM (arm64) register access not allowed before vcpu finalization 2156 ====== ============================================================ 2157 2158(These error codes are indicative only: do not rely on a specific error 2159code being returned in a specific situation.) 2160 2161:: 2162 2163 struct kvm_one_reg { 2164 __u64 id; 2165 __u64 addr; 2166 }; 2167 2168Using this ioctl, a single vcpu register can be set to a specific value 2169defined by user space with the passed in struct kvm_one_reg, where id 2170refers to the register identifier as described below and addr is a pointer 2171to a variable with the respective size. There can be architecture agnostic 2172and architecture specific registers. Each have their own range of operation 2173and their own constants and width. To keep track of the implemented 2174registers, find a list below: 2175 2176 ======= =============================== ============ 2177 Arch Register Width (bits) 2178 ======= =============================== ============ 2179 PPC KVM_REG_PPC_HIOR 64 2180 PPC KVM_REG_PPC_IAC1 64 2181 PPC KVM_REG_PPC_IAC2 64 2182 PPC KVM_REG_PPC_IAC3 64 2183 PPC KVM_REG_PPC_IAC4 64 2184 PPC KVM_REG_PPC_DAC1 64 2185 PPC KVM_REG_PPC_DAC2 64 2186 PPC KVM_REG_PPC_DABR 64 2187 PPC KVM_REG_PPC_DSCR 64 2188 PPC KVM_REG_PPC_PURR 64 2189 PPC KVM_REG_PPC_SPURR 64 2190 PPC KVM_REG_PPC_DAR 64 2191 PPC KVM_REG_PPC_DSISR 32 2192 PPC KVM_REG_PPC_AMR 64 2193 PPC KVM_REG_PPC_UAMOR 64 2194 PPC KVM_REG_PPC_MMCR0 64 2195 PPC KVM_REG_PPC_MMCR1 64 2196 PPC KVM_REG_PPC_MMCRA 64 2197 PPC KVM_REG_PPC_MMCR2 64 2198 PPC KVM_REG_PPC_MMCRS 64 2199 PPC KVM_REG_PPC_MMCR3 64 2200 PPC KVM_REG_PPC_SIAR 64 2201 PPC KVM_REG_PPC_SDAR 64 2202 PPC KVM_REG_PPC_SIER 64 2203 PPC KVM_REG_PPC_SIER2 64 2204 PPC KVM_REG_PPC_SIER3 64 2205 PPC KVM_REG_PPC_PMC1 32 2206 PPC KVM_REG_PPC_PMC2 32 2207 PPC KVM_REG_PPC_PMC3 32 2208 PPC KVM_REG_PPC_PMC4 32 2209 PPC KVM_REG_PPC_PMC5 32 2210 PPC KVM_REG_PPC_PMC6 32 2211 PPC KVM_REG_PPC_PMC7 32 2212 PPC KVM_REG_PPC_PMC8 32 2213 PPC KVM_REG_PPC_FPR0 64 2214 ... 2215 PPC KVM_REG_PPC_FPR31 64 2216 PPC KVM_REG_PPC_VR0 128 2217 ... 2218 PPC KVM_REG_PPC_VR31 128 2219 PPC KVM_REG_PPC_VSR0 128 2220 ... 2221 PPC KVM_REG_PPC_VSR31 128 2222 PPC KVM_REG_PPC_FPSCR 64 2223 PPC KVM_REG_PPC_VSCR 32 2224 PPC KVM_REG_PPC_VPA_ADDR 64 2225 PPC KVM_REG_PPC_VPA_SLB 128 2226 PPC KVM_REG_PPC_VPA_DTL 128 2227 PPC KVM_REG_PPC_EPCR 32 2228 PPC KVM_REG_PPC_EPR 32 2229 PPC KVM_REG_PPC_TCR 32 2230 PPC KVM_REG_PPC_TSR 32 2231 PPC KVM_REG_PPC_OR_TSR 32 2232 PPC KVM_REG_PPC_CLEAR_TSR 32 2233 PPC KVM_REG_PPC_MAS0 32 2234 PPC KVM_REG_PPC_MAS1 32 2235 PPC KVM_REG_PPC_MAS2 64 2236 PPC KVM_REG_PPC_MAS7_3 64 2237 PPC KVM_REG_PPC_MAS4 32 2238 PPC KVM_REG_PPC_MAS6 32 2239 PPC KVM_REG_PPC_MMUCFG 32 2240 PPC KVM_REG_PPC_TLB0CFG 32 2241 PPC KVM_REG_PPC_TLB1CFG 32 2242 PPC KVM_REG_PPC_TLB2CFG 32 2243 PPC KVM_REG_PPC_TLB3CFG 32 2244 PPC KVM_REG_PPC_TLB0PS 32 2245 PPC KVM_REG_PPC_TLB1PS 32 2246 PPC KVM_REG_PPC_TLB2PS 32 2247 PPC KVM_REG_PPC_TLB3PS 32 2248 PPC KVM_REG_PPC_EPTCFG 32 2249 PPC KVM_REG_PPC_ICP_STATE 64 2250 PPC KVM_REG_PPC_VP_STATE 128 2251 PPC KVM_REG_PPC_TB_OFFSET 64 2252 PPC KVM_REG_PPC_SPMC1 32 2253 PPC KVM_REG_PPC_SPMC2 32 2254 PPC KVM_REG_PPC_IAMR 64 2255 PPC KVM_REG_PPC_TFHAR 64 2256 PPC KVM_REG_PPC_TFIAR 64 2257 PPC KVM_REG_PPC_TEXASR 64 2258 PPC KVM_REG_PPC_FSCR 64 2259 PPC KVM_REG_PPC_PSPB 32 2260 PPC KVM_REG_PPC_EBBHR 64 2261 PPC KVM_REG_PPC_EBBRR 64 2262 PPC KVM_REG_PPC_BESCR 64 2263 PPC KVM_REG_PPC_TAR 64 2264 PPC KVM_REG_PPC_DPDES 64 2265 PPC KVM_REG_PPC_DAWR 64 2266 PPC KVM_REG_PPC_DAWRX 64 2267 PPC KVM_REG_PPC_CIABR 64 2268 PPC KVM_REG_PPC_IC 64 2269 PPC KVM_REG_PPC_VTB 64 2270 PPC KVM_REG_PPC_CSIGR 64 2271 PPC KVM_REG_PPC_TACR 64 2272 PPC KVM_REG_PPC_TCSCR 64 2273 PPC KVM_REG_PPC_PID 64 2274 PPC KVM_REG_PPC_ACOP 64 2275 PPC KVM_REG_PPC_VRSAVE 32 2276 PPC KVM_REG_PPC_LPCR 32 2277 PPC KVM_REG_PPC_LPCR_64 64 2278 PPC KVM_REG_PPC_PPR 64 2279 PPC KVM_REG_PPC_ARCH_COMPAT 32 2280 PPC KVM_REG_PPC_DABRX 32 2281 PPC KVM_REG_PPC_WORT 64 2282 PPC KVM_REG_PPC_SPRG9 64 2283 PPC KVM_REG_PPC_DBSR 32 2284 PPC KVM_REG_PPC_TIDR 64 2285 PPC KVM_REG_PPC_PSSCR 64 2286 PPC KVM_REG_PPC_DEC_EXPIRY 64 2287 PPC KVM_REG_PPC_PTCR 64 2288 PPC KVM_REG_PPC_DAWR1 64 2289 PPC KVM_REG_PPC_DAWRX1 64 2290 PPC KVM_REG_PPC_TM_GPR0 64 2291 ... 2292 PPC KVM_REG_PPC_TM_GPR31 64 2293 PPC KVM_REG_PPC_TM_VSR0 128 2294 ... 2295 PPC KVM_REG_PPC_TM_VSR63 128 2296 PPC KVM_REG_PPC_TM_CR 64 2297 PPC KVM_REG_PPC_TM_LR 64 2298 PPC KVM_REG_PPC_TM_CTR 64 2299 PPC KVM_REG_PPC_TM_FPSCR 64 2300 PPC KVM_REG_PPC_TM_AMR 64 2301 PPC KVM_REG_PPC_TM_PPR 64 2302 PPC KVM_REG_PPC_TM_VRSAVE 64 2303 PPC KVM_REG_PPC_TM_VSCR 32 2304 PPC KVM_REG_PPC_TM_DSCR 64 2305 PPC KVM_REG_PPC_TM_TAR 64 2306 PPC KVM_REG_PPC_TM_XER 64 2307 2308 MIPS KVM_REG_MIPS_R0 64 2309 ... 2310 MIPS KVM_REG_MIPS_R31 64 2311 MIPS KVM_REG_MIPS_HI 64 2312 MIPS KVM_REG_MIPS_LO 64 2313 MIPS KVM_REG_MIPS_PC 64 2314 MIPS KVM_REG_MIPS_CP0_INDEX 32 2315 MIPS KVM_REG_MIPS_CP0_ENTRYLO0 64 2316 MIPS KVM_REG_MIPS_CP0_ENTRYLO1 64 2317 MIPS KVM_REG_MIPS_CP0_CONTEXT 64 2318 MIPS KVM_REG_MIPS_CP0_CONTEXTCONFIG 32 2319 MIPS KVM_REG_MIPS_CP0_USERLOCAL 64 2320 MIPS KVM_REG_MIPS_CP0_XCONTEXTCONFIG 64 2321 MIPS KVM_REG_MIPS_CP0_PAGEMASK 32 2322 MIPS KVM_REG_MIPS_CP0_PAGEGRAIN 32 2323 MIPS KVM_REG_MIPS_CP0_SEGCTL0 64 2324 MIPS KVM_REG_MIPS_CP0_SEGCTL1 64 2325 MIPS KVM_REG_MIPS_CP0_SEGCTL2 64 2326 MIPS KVM_REG_MIPS_CP0_PWBASE 64 2327 MIPS KVM_REG_MIPS_CP0_PWFIELD 64 2328 MIPS KVM_REG_MIPS_CP0_PWSIZE 64 2329 MIPS KVM_REG_MIPS_CP0_WIRED 32 2330 MIPS KVM_REG_MIPS_CP0_PWCTL 32 2331 MIPS KVM_REG_MIPS_CP0_HWRENA 32 2332 MIPS KVM_REG_MIPS_CP0_BADVADDR 64 2333 MIPS KVM_REG_MIPS_CP0_BADINSTR 32 2334 MIPS KVM_REG_MIPS_CP0_BADINSTRP 32 2335 MIPS KVM_REG_MIPS_CP0_COUNT 32 2336 MIPS KVM_REG_MIPS_CP0_ENTRYHI 64 2337 MIPS KVM_REG_MIPS_CP0_COMPARE 32 2338 MIPS KVM_REG_MIPS_CP0_STATUS 32 2339 MIPS KVM_REG_MIPS_CP0_INTCTL 32 2340 MIPS KVM_REG_MIPS_CP0_CAUSE 32 2341 MIPS KVM_REG_MIPS_CP0_EPC 64 2342 MIPS KVM_REG_MIPS_CP0_PRID 32 2343 MIPS KVM_REG_MIPS_CP0_EBASE 64 2344 MIPS KVM_REG_MIPS_CP0_CONFIG 32 2345 MIPS KVM_REG_MIPS_CP0_CONFIG1 32 2346 MIPS KVM_REG_MIPS_CP0_CONFIG2 32 2347 MIPS KVM_REG_MIPS_CP0_CONFIG3 32 2348 MIPS KVM_REG_MIPS_CP0_CONFIG4 32 2349 MIPS KVM_REG_MIPS_CP0_CONFIG5 32 2350 MIPS KVM_REG_MIPS_CP0_CONFIG7 32 2351 MIPS KVM_REG_MIPS_CP0_XCONTEXT 64 2352 MIPS KVM_REG_MIPS_CP0_ERROREPC 64 2353 MIPS KVM_REG_MIPS_CP0_KSCRATCH1 64 2354 MIPS KVM_REG_MIPS_CP0_KSCRATCH2 64 2355 MIPS KVM_REG_MIPS_CP0_KSCRATCH3 64 2356 MIPS KVM_REG_MIPS_CP0_KSCRATCH4 64 2357 MIPS KVM_REG_MIPS_CP0_KSCRATCH5 64 2358 MIPS KVM_REG_MIPS_CP0_KSCRATCH6 64 2359 MIPS KVM_REG_MIPS_CP0_MAAR(0..63) 64 2360 MIPS KVM_REG_MIPS_COUNT_CTL 64 2361 MIPS KVM_REG_MIPS_COUNT_RESUME 64 2362 MIPS KVM_REG_MIPS_COUNT_HZ 64 2363 MIPS KVM_REG_MIPS_FPR_32(0..31) 32 2364 MIPS KVM_REG_MIPS_FPR_64(0..31) 64 2365 MIPS KVM_REG_MIPS_VEC_128(0..31) 128 2366 MIPS KVM_REG_MIPS_FCR_IR 32 2367 MIPS KVM_REG_MIPS_FCR_CSR 32 2368 MIPS KVM_REG_MIPS_MSA_IR 32 2369 MIPS KVM_REG_MIPS_MSA_CSR 32 2370 ======= =============================== ============ 2371 2372ARM registers are mapped using the lower 32 bits. The upper 16 of that 2373is the register group type, or coprocessor number: 2374 2375ARM core registers have the following id bit patterns:: 2376 2377 0x4020 0000 0010 <index into the kvm_regs struct:16> 2378 2379ARM 32-bit CP15 registers have the following id bit patterns:: 2380 2381 0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3> 2382 2383ARM 64-bit CP15 registers have the following id bit patterns:: 2384 2385 0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3> 2386 2387ARM CCSIDR registers are demultiplexed by CSSELR value:: 2388 2389 0x4020 0000 0011 00 <csselr:8> 2390 2391ARM 32-bit VFP control registers have the following id bit patterns:: 2392 2393 0x4020 0000 0012 1 <regno:12> 2394 2395ARM 64-bit FP registers have the following id bit patterns:: 2396 2397 0x4030 0000 0012 0 <regno:12> 2398 2399ARM firmware pseudo-registers have the following bit pattern:: 2400 2401 0x4030 0000 0014 <regno:16> 2402 2403 2404arm64 registers are mapped using the lower 32 bits. The upper 16 of 2405that is the register group type, or coprocessor number: 2406 2407arm64 core/FP-SIMD registers have the following id bit patterns. Note 2408that the size of the access is variable, as the kvm_regs structure 2409contains elements ranging from 32 to 128 bits. The index is a 32bit 2410value in the kvm_regs structure seen as a 32bit array:: 2411 2412 0x60x0 0000 0010 <index into the kvm_regs struct:16> 2413 2414Specifically: 2415 2416======================= ========= ===== ======================================= 2417 Encoding Register Bits kvm_regs member 2418======================= ========= ===== ======================================= 2419 0x6030 0000 0010 0000 X0 64 regs.regs[0] 2420 0x6030 0000 0010 0002 X1 64 regs.regs[1] 2421 ... 2422 0x6030 0000 0010 003c X30 64 regs.regs[30] 2423 0x6030 0000 0010 003e SP 64 regs.sp 2424 0x6030 0000 0010 0040 PC 64 regs.pc 2425 0x6030 0000 0010 0042 PSTATE 64 regs.pstate 2426 0x6030 0000 0010 0044 SP_EL1 64 sp_el1 2427 0x6030 0000 0010 0046 ELR_EL1 64 elr_el1 2428 0x6030 0000 0010 0048 SPSR_EL1 64 spsr[KVM_SPSR_EL1] (alias SPSR_SVC) 2429 0x6030 0000 0010 004a SPSR_ABT 64 spsr[KVM_SPSR_ABT] 2430 0x6030 0000 0010 004c SPSR_UND 64 spsr[KVM_SPSR_UND] 2431 0x6030 0000 0010 004e SPSR_IRQ 64 spsr[KVM_SPSR_IRQ] 2432 0x6060 0000 0010 0050 SPSR_FIQ 64 spsr[KVM_SPSR_FIQ] 2433 0x6040 0000 0010 0054 V0 128 fp_regs.vregs[0] [1]_ 2434 0x6040 0000 0010 0058 V1 128 fp_regs.vregs[1] [1]_ 2435 ... 2436 0x6040 0000 0010 00d0 V31 128 fp_regs.vregs[31] [1]_ 2437 0x6020 0000 0010 00d4 FPSR 32 fp_regs.fpsr 2438 0x6020 0000 0010 00d5 FPCR 32 fp_regs.fpcr 2439======================= ========= ===== ======================================= 2440 2441.. [1] These encodings are not accepted for SVE-enabled vcpus. See 2442 KVM_ARM_VCPU_INIT. 2443 2444 The equivalent register content can be accessed via bits [127:0] of 2445 the corresponding SVE Zn registers instead for vcpus that have SVE 2446 enabled (see below). 2447 2448arm64 CCSIDR registers are demultiplexed by CSSELR value:: 2449 2450 0x6020 0000 0011 00 <csselr:8> 2451 2452arm64 system registers have the following id bit patterns:: 2453 2454 0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3> 2455 2456.. warning:: 2457 2458 Two system register IDs do not follow the specified pattern. These 2459 are KVM_REG_ARM_TIMER_CVAL and KVM_REG_ARM_TIMER_CNT, which map to 2460 system registers CNTV_CVAL_EL0 and CNTVCT_EL0 respectively. These 2461 two had their values accidentally swapped, which means TIMER_CVAL is 2462 derived from the register encoding for CNTVCT_EL0 and TIMER_CNT is 2463 derived from the register encoding for CNTV_CVAL_EL0. As this is 2464 API, it must remain this way. 2465 2466arm64 firmware pseudo-registers have the following bit pattern:: 2467 2468 0x6030 0000 0014 <regno:16> 2469 2470arm64 SVE registers have the following bit patterns:: 2471 2472 0x6080 0000 0015 00 <n:5> <slice:5> Zn bits[2048*slice + 2047 : 2048*slice] 2473 0x6050 0000 0015 04 <n:4> <slice:5> Pn bits[256*slice + 255 : 256*slice] 2474 0x6050 0000 0015 060 <slice:5> FFR bits[256*slice + 255 : 256*slice] 2475 0x6060 0000 0015 ffff KVM_REG_ARM64_SVE_VLS pseudo-register 2476 2477Access to register IDs where 2048 * slice >= 128 * max_vq will fail with 2478ENOENT. max_vq is the vcpu's maximum supported vector length in 128-bit 2479quadwords: see [2]_ below. 2480 2481These registers are only accessible on vcpus for which SVE is enabled. 2482See KVM_ARM_VCPU_INIT for details. 2483 2484In addition, except for KVM_REG_ARM64_SVE_VLS, these registers are not 2485accessible until the vcpu's SVE configuration has been finalized 2486using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). See KVM_ARM_VCPU_INIT 2487and KVM_ARM_VCPU_FINALIZE for more information about this procedure. 2488 2489KVM_REG_ARM64_SVE_VLS is a pseudo-register that allows the set of vector 2490lengths supported by the vcpu to be discovered and configured by 2491userspace. When transferred to or from user memory via KVM_GET_ONE_REG 2492or KVM_SET_ONE_REG, the value of this register is of type 2493__u64[KVM_ARM64_SVE_VLS_WORDS], and encodes the set of vector lengths as 2494follows:: 2495 2496 __u64 vector_lengths[KVM_ARM64_SVE_VLS_WORDS]; 2497 2498 if (vq >= SVE_VQ_MIN && vq <= SVE_VQ_MAX && 2499 ((vector_lengths[(vq - KVM_ARM64_SVE_VQ_MIN) / 64] >> 2500 ((vq - KVM_ARM64_SVE_VQ_MIN) % 64)) & 1)) 2501 /* Vector length vq * 16 bytes supported */ 2502 else 2503 /* Vector length vq * 16 bytes not supported */ 2504 2505.. [2] The maximum value vq for which the above condition is true is 2506 max_vq. This is the maximum vector length available to the guest on 2507 this vcpu, and determines which register slices are visible through 2508 this ioctl interface. 2509 2510(See Documentation/arm64/sve.rst for an explanation of the "vq" 2511nomenclature.) 2512 2513KVM_REG_ARM64_SVE_VLS is only accessible after KVM_ARM_VCPU_INIT. 2514KVM_ARM_VCPU_INIT initialises it to the best set of vector lengths that 2515the host supports. 2516 2517Userspace may subsequently modify it if desired until the vcpu's SVE 2518configuration is finalized using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). 2519 2520Apart from simply removing all vector lengths from the host set that 2521exceed some value, support for arbitrarily chosen sets of vector lengths 2522is hardware-dependent and may not be available. Attempting to configure 2523an invalid set of vector lengths via KVM_SET_ONE_REG will fail with 2524EINVAL. 2525 2526After the vcpu's SVE configuration is finalized, further attempts to 2527write this register will fail with EPERM. 2528 2529 2530MIPS registers are mapped using the lower 32 bits. The upper 16 of that is 2531the register group type: 2532 2533MIPS core registers (see above) have the following id bit patterns:: 2534 2535 0x7030 0000 0000 <reg:16> 2536 2537MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit 2538patterns depending on whether they're 32-bit or 64-bit registers:: 2539 2540 0x7020 0000 0001 00 <reg:5> <sel:3> (32-bit) 2541 0x7030 0000 0001 00 <reg:5> <sel:3> (64-bit) 2542 2543Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_MIPS_CP0_ENTRYLO1 are the MIPS64 2544versions of the EntryLo registers regardless of the word size of the host 2545hardware, host kernel, guest, and whether XPA is present in the guest, i.e. 2546with the RI and XI bits (if they exist) in bits 63 and 62 respectively, and 2547the PFNX field starting at bit 30. 2548 2549MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) above) have the following id bit 2550patterns:: 2551 2552 0x7030 0000 0001 01 <reg:8> 2553 2554MIPS KVM control registers (see above) have the following id bit patterns:: 2555 2556 0x7030 0000 0002 <reg:16> 2557 2558MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following 2559id bit patterns depending on the size of the register being accessed. They are 2560always accessed according to the current guest FPU mode (Status.FR and 2561Config5.FRE), i.e. as the guest would see them, and they become unpredictable 2562if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector 2563registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they 2564overlap the FPU registers:: 2565 2566 0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers) 2567 0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers) 2568 0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers) 2569 2570MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the 2571following id bit patterns:: 2572 2573 0x7020 0000 0003 01 <0:3> <reg:5> 2574 2575MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the 2576following id bit patterns:: 2577 2578 0x7020 0000 0003 02 <0:3> <reg:5> 2579 2580 25814.69 KVM_GET_ONE_REG 2582-------------------- 2583 2584:Capability: KVM_CAP_ONE_REG 2585:Architectures: all 2586:Type: vcpu ioctl 2587:Parameters: struct kvm_one_reg (in and out) 2588:Returns: 0 on success, negative value on failure 2589 2590Errors include: 2591 2592 ======== ============================================================ 2593 ENOENT no such register 2594 EINVAL invalid register ID, or no such register or used with VMs in 2595 protected virtualization mode on s390 2596 EPERM (arm64) register access not allowed before vcpu finalization 2597 ======== ============================================================ 2598 2599(These error codes are indicative only: do not rely on a specific error 2600code being returned in a specific situation.) 2601 2602This ioctl allows to receive the value of a single register implemented 2603in a vcpu. The register to read is indicated by the "id" field of the 2604kvm_one_reg struct passed in. On success, the register value can be found 2605at the memory location pointed to by "addr". 2606 2607The list of registers accessible using this interface is identical to the 2608list in 4.68. 2609 2610 26114.70 KVM_KVMCLOCK_CTRL 2612---------------------- 2613 2614:Capability: KVM_CAP_KVMCLOCK_CTRL 2615:Architectures: Any that implement pvclocks (currently x86 only) 2616:Type: vcpu ioctl 2617:Parameters: None 2618:Returns: 0 on success, -1 on error 2619 2620This ioctl sets a flag accessible to the guest indicating that the specified 2621vCPU has been paused by the host userspace. 2622 2623The host will set a flag in the pvclock structure that is checked from the 2624soft lockup watchdog. The flag is part of the pvclock structure that is 2625shared between guest and host, specifically the second bit of the flags 2626field of the pvclock_vcpu_time_info structure. It will be set exclusively by 2627the host and read/cleared exclusively by the guest. The guest operation of 2628checking and clearing the flag must be an atomic operation so 2629load-link/store-conditional, or equivalent must be used. There are two cases 2630where the guest will clear the flag: when the soft lockup watchdog timer resets 2631itself or when a soft lockup is detected. This ioctl can be called any time 2632after pausing the vcpu, but before it is resumed. 2633 2634 26354.71 KVM_SIGNAL_MSI 2636------------------- 2637 2638:Capability: KVM_CAP_SIGNAL_MSI 2639:Architectures: x86 arm arm64 2640:Type: vm ioctl 2641:Parameters: struct kvm_msi (in) 2642:Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error 2643 2644Directly inject a MSI message. Only valid with in-kernel irqchip that handles 2645MSI messages. 2646 2647:: 2648 2649 struct kvm_msi { 2650 __u32 address_lo; 2651 __u32 address_hi; 2652 __u32 data; 2653 __u32 flags; 2654 __u32 devid; 2655 __u8 pad[12]; 2656 }; 2657 2658flags: 2659 KVM_MSI_VALID_DEVID: devid contains a valid value. The per-VM 2660 KVM_CAP_MSI_DEVID capability advertises the requirement to provide 2661 the device ID. If this capability is not available, userspace 2662 should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail. 2663 2664If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier 2665for the device that wrote the MSI message. For PCI, this is usually a 2666BFD identifier in the lower 16 bits. 2667 2668On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS 2669feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled, 2670address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of 2671address_hi must be zero. 2672 2673 26744.71 KVM_CREATE_PIT2 2675-------------------- 2676 2677:Capability: KVM_CAP_PIT2 2678:Architectures: x86 2679:Type: vm ioctl 2680:Parameters: struct kvm_pit_config (in) 2681:Returns: 0 on success, -1 on error 2682 2683Creates an in-kernel device model for the i8254 PIT. This call is only valid 2684after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following 2685parameters have to be passed:: 2686 2687 struct kvm_pit_config { 2688 __u32 flags; 2689 __u32 pad[15]; 2690 }; 2691 2692Valid flags are:: 2693 2694 #define KVM_PIT_SPEAKER_DUMMY 1 /* emulate speaker port stub */ 2695 2696PIT timer interrupts may use a per-VM kernel thread for injection. If it 2697exists, this thread will have a name of the following pattern:: 2698 2699 kvm-pit/<owner-process-pid> 2700 2701When running a guest with elevated priorities, the scheduling parameters of 2702this thread may have to be adjusted accordingly. 2703 2704This IOCTL replaces the obsolete KVM_CREATE_PIT. 2705 2706 27074.72 KVM_GET_PIT2 2708----------------- 2709 2710:Capability: KVM_CAP_PIT_STATE2 2711:Architectures: x86 2712:Type: vm ioctl 2713:Parameters: struct kvm_pit_state2 (out) 2714:Returns: 0 on success, -1 on error 2715 2716Retrieves the state of the in-kernel PIT model. Only valid after 2717KVM_CREATE_PIT2. The state is returned in the following structure:: 2718 2719 struct kvm_pit_state2 { 2720 struct kvm_pit_channel_state channels[3]; 2721 __u32 flags; 2722 __u32 reserved[9]; 2723 }; 2724 2725Valid flags are:: 2726 2727 /* disable PIT in HPET legacy mode */ 2728 #define KVM_PIT_FLAGS_HPET_LEGACY 0x00000001 2729 2730This IOCTL replaces the obsolete KVM_GET_PIT. 2731 2732 27334.73 KVM_SET_PIT2 2734----------------- 2735 2736:Capability: KVM_CAP_PIT_STATE2 2737:Architectures: x86 2738:Type: vm ioctl 2739:Parameters: struct kvm_pit_state2 (in) 2740:Returns: 0 on success, -1 on error 2741 2742Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2. 2743See KVM_GET_PIT2 for details on struct kvm_pit_state2. 2744 2745This IOCTL replaces the obsolete KVM_SET_PIT. 2746 2747 27484.74 KVM_PPC_GET_SMMU_INFO 2749-------------------------- 2750 2751:Capability: KVM_CAP_PPC_GET_SMMU_INFO 2752:Architectures: powerpc 2753:Type: vm ioctl 2754:Parameters: None 2755:Returns: 0 on success, -1 on error 2756 2757This populates and returns a structure describing the features of 2758the "Server" class MMU emulation supported by KVM. 2759This can in turn be used by userspace to generate the appropriate 2760device-tree properties for the guest operating system. 2761 2762The structure contains some global information, followed by an 2763array of supported segment page sizes:: 2764 2765 struct kvm_ppc_smmu_info { 2766 __u64 flags; 2767 __u32 slb_size; 2768 __u32 pad; 2769 struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ]; 2770 }; 2771 2772The supported flags are: 2773 2774 - KVM_PPC_PAGE_SIZES_REAL: 2775 When that flag is set, guest page sizes must "fit" the backing 2776 store page sizes. When not set, any page size in the list can 2777 be used regardless of how they are backed by userspace. 2778 2779 - KVM_PPC_1T_SEGMENTS 2780 The emulated MMU supports 1T segments in addition to the 2781 standard 256M ones. 2782 2783 - KVM_PPC_NO_HASH 2784 This flag indicates that HPT guests are not supported by KVM, 2785 thus all guests must use radix MMU mode. 2786 2787The "slb_size" field indicates how many SLB entries are supported 2788 2789The "sps" array contains 8 entries indicating the supported base 2790page sizes for a segment in increasing order. Each entry is defined 2791as follow:: 2792 2793 struct kvm_ppc_one_seg_page_size { 2794 __u32 page_shift; /* Base page shift of segment (or 0) */ 2795 __u32 slb_enc; /* SLB encoding for BookS */ 2796 struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ]; 2797 }; 2798 2799An entry with a "page_shift" of 0 is unused. Because the array is 2800organized in increasing order, a lookup can stop when encoutering 2801such an entry. 2802 2803The "slb_enc" field provides the encoding to use in the SLB for the 2804page size. The bits are in positions such as the value can directly 2805be OR'ed into the "vsid" argument of the slbmte instruction. 2806 2807The "enc" array is a list which for each of those segment base page 2808size provides the list of supported actual page sizes (which can be 2809only larger or equal to the base page size), along with the 2810corresponding encoding in the hash PTE. Similarly, the array is 28118 entries sorted by increasing sizes and an entry with a "0" shift 2812is an empty entry and a terminator:: 2813 2814 struct kvm_ppc_one_page_size { 2815 __u32 page_shift; /* Page shift (or 0) */ 2816 __u32 pte_enc; /* Encoding in the HPTE (>>12) */ 2817 }; 2818 2819The "pte_enc" field provides a value that can OR'ed into the hash 2820PTE's RPN field (ie, it needs to be shifted left by 12 to OR it 2821into the hash PTE second double word). 2822 28234.75 KVM_IRQFD 2824-------------- 2825 2826:Capability: KVM_CAP_IRQFD 2827:Architectures: x86 s390 arm arm64 2828:Type: vm ioctl 2829:Parameters: struct kvm_irqfd (in) 2830:Returns: 0 on success, -1 on error 2831 2832Allows setting an eventfd to directly trigger a guest interrupt. 2833kvm_irqfd.fd specifies the file descriptor to use as the eventfd and 2834kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When 2835an event is triggered on the eventfd, an interrupt is injected into 2836the guest using the specified gsi pin. The irqfd is removed using 2837the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd 2838and kvm_irqfd.gsi. 2839 2840With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify 2841mechanism allowing emulation of level-triggered, irqfd-based 2842interrupts. When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an 2843additional eventfd in the kvm_irqfd.resamplefd field. When operating 2844in resample mode, posting of an interrupt through kvm_irq.fd asserts 2845the specified gsi in the irqchip. When the irqchip is resampled, such 2846as from an EOI, the gsi is de-asserted and the user is notified via 2847kvm_irqfd.resamplefd. It is the user's responsibility to re-queue 2848the interrupt if the device making use of it still requires service. 2849Note that closing the resamplefd is not sufficient to disable the 2850irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment 2851and need not be specified with KVM_IRQFD_FLAG_DEASSIGN. 2852 2853On arm/arm64, gsi routing being supported, the following can happen: 2854 2855- in case no routing entry is associated to this gsi, injection fails 2856- in case the gsi is associated to an irqchip routing entry, 2857 irqchip.pin + 32 corresponds to the injected SPI ID. 2858- in case the gsi is associated to an MSI routing entry, the MSI 2859 message and device ID are translated into an LPI (support restricted 2860 to GICv3 ITS in-kernel emulation). 2861 28624.76 KVM_PPC_ALLOCATE_HTAB 2863-------------------------- 2864 2865:Capability: KVM_CAP_PPC_ALLOC_HTAB 2866:Architectures: powerpc 2867:Type: vm ioctl 2868:Parameters: Pointer to u32 containing hash table order (in/out) 2869:Returns: 0 on success, -1 on error 2870 2871This requests the host kernel to allocate an MMU hash table for a 2872guest using the PAPR paravirtualization interface. This only does 2873anything if the kernel is configured to use the Book 3S HV style of 2874virtualization. Otherwise the capability doesn't exist and the ioctl 2875returns an ENOTTY error. The rest of this description assumes Book 3S 2876HV. 2877 2878There must be no vcpus running when this ioctl is called; if there 2879are, it will do nothing and return an EBUSY error. 2880 2881The parameter is a pointer to a 32-bit unsigned integer variable 2882containing the order (log base 2) of the desired size of the hash 2883table, which must be between 18 and 46. On successful return from the 2884ioctl, the value will not be changed by the kernel. 2885 2886If no hash table has been allocated when any vcpu is asked to run 2887(with the KVM_RUN ioctl), the host kernel will allocate a 2888default-sized hash table (16 MB). 2889 2890If this ioctl is called when a hash table has already been allocated, 2891with a different order from the existing hash table, the existing hash 2892table will be freed and a new one allocated. If this is ioctl is 2893called when a hash table has already been allocated of the same order 2894as specified, the kernel will clear out the existing hash table (zero 2895all HPTEs). In either case, if the guest is using the virtualized 2896real-mode area (VRMA) facility, the kernel will re-create the VMRA 2897HPTEs on the next KVM_RUN of any vcpu. 2898 28994.77 KVM_S390_INTERRUPT 2900----------------------- 2901 2902:Capability: basic 2903:Architectures: s390 2904:Type: vm ioctl, vcpu ioctl 2905:Parameters: struct kvm_s390_interrupt (in) 2906:Returns: 0 on success, -1 on error 2907 2908Allows to inject an interrupt to the guest. Interrupts can be floating 2909(vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type. 2910 2911Interrupt parameters are passed via kvm_s390_interrupt:: 2912 2913 struct kvm_s390_interrupt { 2914 __u32 type; 2915 __u32 parm; 2916 __u64 parm64; 2917 }; 2918 2919type can be one of the following: 2920 2921KVM_S390_SIGP_STOP (vcpu) 2922 - sigp stop; optional flags in parm 2923KVM_S390_PROGRAM_INT (vcpu) 2924 - program check; code in parm 2925KVM_S390_SIGP_SET_PREFIX (vcpu) 2926 - sigp set prefix; prefix address in parm 2927KVM_S390_RESTART (vcpu) 2928 - restart 2929KVM_S390_INT_CLOCK_COMP (vcpu) 2930 - clock comparator interrupt 2931KVM_S390_INT_CPU_TIMER (vcpu) 2932 - CPU timer interrupt 2933KVM_S390_INT_VIRTIO (vm) 2934 - virtio external interrupt; external interrupt 2935 parameters in parm and parm64 2936KVM_S390_INT_SERVICE (vm) 2937 - sclp external interrupt; sclp parameter in parm 2938KVM_S390_INT_EMERGENCY (vcpu) 2939 - sigp emergency; source cpu in parm 2940KVM_S390_INT_EXTERNAL_CALL (vcpu) 2941 - sigp external call; source cpu in parm 2942KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm) 2943 - compound value to indicate an 2944 I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel); 2945 I/O interruption parameters in parm (subchannel) and parm64 (intparm, 2946 interruption subclass) 2947KVM_S390_MCHK (vm, vcpu) 2948 - machine check interrupt; cr 14 bits in parm, machine check interrupt 2949 code in parm64 (note that machine checks needing further payload are not 2950 supported by this ioctl) 2951 2952This is an asynchronous vcpu ioctl and can be invoked from any thread. 2953 29544.78 KVM_PPC_GET_HTAB_FD 2955------------------------ 2956 2957:Capability: KVM_CAP_PPC_HTAB_FD 2958:Architectures: powerpc 2959:Type: vm ioctl 2960:Parameters: Pointer to struct kvm_get_htab_fd (in) 2961:Returns: file descriptor number (>= 0) on success, -1 on error 2962 2963This returns a file descriptor that can be used either to read out the 2964entries in the guest's hashed page table (HPT), or to write entries to 2965initialize the HPT. The returned fd can only be written to if the 2966KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and 2967can only be read if that bit is clear. The argument struct looks like 2968this:: 2969 2970 /* For KVM_PPC_GET_HTAB_FD */ 2971 struct kvm_get_htab_fd { 2972 __u64 flags; 2973 __u64 start_index; 2974 __u64 reserved[2]; 2975 }; 2976 2977 /* Values for kvm_get_htab_fd.flags */ 2978 #define KVM_GET_HTAB_BOLTED_ONLY ((__u64)0x1) 2979 #define KVM_GET_HTAB_WRITE ((__u64)0x2) 2980 2981The 'start_index' field gives the index in the HPT of the entry at 2982which to start reading. It is ignored when writing. 2983 2984Reads on the fd will initially supply information about all 2985"interesting" HPT entries. Interesting entries are those with the 2986bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise 2987all entries. When the end of the HPT is reached, the read() will 2988return. If read() is called again on the fd, it will start again from 2989the beginning of the HPT, but will only return HPT entries that have 2990changed since they were last read. 2991 2992Data read or written is structured as a header (8 bytes) followed by a 2993series of valid HPT entries (16 bytes) each. The header indicates how 2994many valid HPT entries there are and how many invalid entries follow 2995the valid entries. The invalid entries are not represented explicitly 2996in the stream. The header format is:: 2997 2998 struct kvm_get_htab_header { 2999 __u32 index; 3000 __u16 n_valid; 3001 __u16 n_invalid; 3002 }; 3003 3004Writes to the fd create HPT entries starting at the index given in the 3005header; first 'n_valid' valid entries with contents from the data 3006written, then 'n_invalid' invalid entries, invalidating any previously 3007valid entries found. 3008 30094.79 KVM_CREATE_DEVICE 3010---------------------- 3011 3012:Capability: KVM_CAP_DEVICE_CTRL 3013:Type: vm ioctl 3014:Parameters: struct kvm_create_device (in/out) 3015:Returns: 0 on success, -1 on error 3016 3017Errors: 3018 3019 ====== ======================================================= 3020 ENODEV The device type is unknown or unsupported 3021 EEXIST Device already created, and this type of device may not 3022 be instantiated multiple times 3023 ====== ======================================================= 3024 3025 Other error conditions may be defined by individual device types or 3026 have their standard meanings. 3027 3028Creates an emulated device in the kernel. The file descriptor returned 3029in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR. 3030 3031If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the 3032device type is supported (not necessarily whether it can be created 3033in the current vm). 3034 3035Individual devices should not define flags. Attributes should be used 3036for specifying any behavior that is not implied by the device type 3037number. 3038 3039:: 3040 3041 struct kvm_create_device { 3042 __u32 type; /* in: KVM_DEV_TYPE_xxx */ 3043 __u32 fd; /* out: device handle */ 3044 __u32 flags; /* in: KVM_CREATE_DEVICE_xxx */ 3045 }; 3046 30474.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR 3048-------------------------------------------- 3049 3050:Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device, 3051 KVM_CAP_VCPU_ATTRIBUTES for vcpu device 3052:Type: device ioctl, vm ioctl, vcpu ioctl 3053:Parameters: struct kvm_device_attr 3054:Returns: 0 on success, -1 on error 3055 3056Errors: 3057 3058 ===== ============================================================= 3059 ENXIO The group or attribute is unknown/unsupported for this device 3060 or hardware support is missing. 3061 EPERM The attribute cannot (currently) be accessed this way 3062 (e.g. read-only attribute, or attribute that only makes 3063 sense when the device is in a different state) 3064 ===== ============================================================= 3065 3066 Other error conditions may be defined by individual device types. 3067 3068Gets/sets a specified piece of device configuration and/or state. The 3069semantics are device-specific. See individual device documentation in 3070the "devices" directory. As with ONE_REG, the size of the data 3071transferred is defined by the particular attribute. 3072 3073:: 3074 3075 struct kvm_device_attr { 3076 __u32 flags; /* no flags currently defined */ 3077 __u32 group; /* device-defined */ 3078 __u64 attr; /* group-defined */ 3079 __u64 addr; /* userspace address of attr data */ 3080 }; 3081 30824.81 KVM_HAS_DEVICE_ATTR 3083------------------------ 3084 3085:Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device, 3086 KVM_CAP_VCPU_ATTRIBUTES for vcpu device 3087:Type: device ioctl, vm ioctl, vcpu ioctl 3088:Parameters: struct kvm_device_attr 3089:Returns: 0 on success, -1 on error 3090 3091Errors: 3092 3093 ===== ============================================================= 3094 ENXIO The group or attribute is unknown/unsupported for this device 3095 or hardware support is missing. 3096 ===== ============================================================= 3097 3098Tests whether a device supports a particular attribute. A successful 3099return indicates the attribute is implemented. It does not necessarily 3100indicate that the attribute can be read or written in the device's 3101current state. "addr" is ignored. 3102 31034.82 KVM_ARM_VCPU_INIT 3104---------------------- 3105 3106:Capability: basic 3107:Architectures: arm, arm64 3108:Type: vcpu ioctl 3109:Parameters: struct kvm_vcpu_init (in) 3110:Returns: 0 on success; -1 on error 3111 3112Errors: 3113 3114 ====== ================================================================= 3115 EINVAL the target is unknown, or the combination of features is invalid. 3116 ENOENT a features bit specified is unknown. 3117 ====== ================================================================= 3118 3119This tells KVM what type of CPU to present to the guest, and what 3120optional features it should have. This will cause a reset of the cpu 3121registers to their initial values. If this is not called, KVM_RUN will 3122return ENOEXEC for that vcpu. 3123 3124The initial values are defined as: 3125 - Processor state: 3126 * AArch64: EL1h, D, A, I and F bits set. All other bits 3127 are cleared. 3128 * AArch32: SVC, A, I and F bits set. All other bits are 3129 cleared. 3130 - General Purpose registers, including PC and SP: set to 0 3131 - FPSIMD/NEON registers: set to 0 3132 - SVE registers: set to 0 3133 - System registers: Reset to their architecturally defined 3134 values as for a warm reset to EL1 (resp. SVC) 3135 3136Note that because some registers reflect machine topology, all vcpus 3137should be created before this ioctl is invoked. 3138 3139Userspace can call this function multiple times for a given vcpu, including 3140after the vcpu has been run. This will reset the vcpu to its initial 3141state. All calls to this function after the initial call must use the same 3142target and same set of feature flags, otherwise EINVAL will be returned. 3143 3144Possible features: 3145 3146 - KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state. 3147 Depends on KVM_CAP_ARM_PSCI. If not set, the CPU will be powered on 3148 and execute guest code when KVM_RUN is called. 3149 - KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode. 3150 Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only). 3151 - KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 (or a future revision 3152 backward compatible with v0.2) for the CPU. 3153 Depends on KVM_CAP_ARM_PSCI_0_2. 3154 - KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU. 3155 Depends on KVM_CAP_ARM_PMU_V3. 3156 3157 - KVM_ARM_VCPU_PTRAUTH_ADDRESS: Enables Address Pointer authentication 3158 for arm64 only. 3159 Depends on KVM_CAP_ARM_PTRAUTH_ADDRESS. 3160 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are 3161 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and 3162 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be 3163 requested. 3164 3165 - KVM_ARM_VCPU_PTRAUTH_GENERIC: Enables Generic Pointer authentication 3166 for arm64 only. 3167 Depends on KVM_CAP_ARM_PTRAUTH_GENERIC. 3168 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are 3169 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and 3170 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be 3171 requested. 3172 3173 - KVM_ARM_VCPU_SVE: Enables SVE for the CPU (arm64 only). 3174 Depends on KVM_CAP_ARM_SVE. 3175 Requires KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3176 3177 * After KVM_ARM_VCPU_INIT: 3178 3179 - KVM_REG_ARM64_SVE_VLS may be read using KVM_GET_ONE_REG: the 3180 initial value of this pseudo-register indicates the best set of 3181 vector lengths possible for a vcpu on this host. 3182 3183 * Before KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3184 3185 - KVM_RUN and KVM_GET_REG_LIST are not available; 3186 3187 - KVM_GET_ONE_REG and KVM_SET_ONE_REG cannot be used to access 3188 the scalable archietctural SVE registers 3189 KVM_REG_ARM64_SVE_ZREG(), KVM_REG_ARM64_SVE_PREG() or 3190 KVM_REG_ARM64_SVE_FFR; 3191 3192 - KVM_REG_ARM64_SVE_VLS may optionally be written using 3193 KVM_SET_ONE_REG, to modify the set of vector lengths available 3194 for the vcpu. 3195 3196 * After KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3197 3198 - the KVM_REG_ARM64_SVE_VLS pseudo-register is immutable, and can 3199 no longer be written using KVM_SET_ONE_REG. 3200 32014.83 KVM_ARM_PREFERRED_TARGET 3202----------------------------- 3203 3204:Capability: basic 3205:Architectures: arm, arm64 3206:Type: vm ioctl 3207:Parameters: struct kvm_vcpu_init (out) 3208:Returns: 0 on success; -1 on error 3209 3210Errors: 3211 3212 ====== ========================================== 3213 ENODEV no preferred target available for the host 3214 ====== ========================================== 3215 3216This queries KVM for preferred CPU target type which can be emulated 3217by KVM on underlying host. 3218 3219The ioctl returns struct kvm_vcpu_init instance containing information 3220about preferred CPU target type and recommended features for it. The 3221kvm_vcpu_init->features bitmap returned will have feature bits set if 3222the preferred target recommends setting these features, but this is 3223not mandatory. 3224 3225The information returned by this ioctl can be used to prepare an instance 3226of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in 3227VCPU matching underlying host. 3228 3229 32304.84 KVM_GET_REG_LIST 3231--------------------- 3232 3233:Capability: basic 3234:Architectures: arm, arm64, mips 3235:Type: vcpu ioctl 3236:Parameters: struct kvm_reg_list (in/out) 3237:Returns: 0 on success; -1 on error 3238 3239Errors: 3240 3241 ===== ============================================================== 3242 E2BIG the reg index list is too big to fit in the array specified by 3243 the user (the number required will be written into n). 3244 ===== ============================================================== 3245 3246:: 3247 3248 struct kvm_reg_list { 3249 __u64 n; /* number of registers in reg[] */ 3250 __u64 reg[0]; 3251 }; 3252 3253This ioctl returns the guest registers that are supported for the 3254KVM_GET_ONE_REG/KVM_SET_ONE_REG calls. 3255 3256 32574.85 KVM_ARM_SET_DEVICE_ADDR (deprecated) 3258----------------------------------------- 3259 3260:Capability: KVM_CAP_ARM_SET_DEVICE_ADDR 3261:Architectures: arm, arm64 3262:Type: vm ioctl 3263:Parameters: struct kvm_arm_device_address (in) 3264:Returns: 0 on success, -1 on error 3265 3266Errors: 3267 3268 ====== ============================================ 3269 ENODEV The device id is unknown 3270 ENXIO Device not supported on current system 3271 EEXIST Address already set 3272 E2BIG Address outside guest physical address space 3273 EBUSY Address overlaps with other device range 3274 ====== ============================================ 3275 3276:: 3277 3278 struct kvm_arm_device_addr { 3279 __u64 id; 3280 __u64 addr; 3281 }; 3282 3283Specify a device address in the guest's physical address space where guests 3284can access emulated or directly exposed devices, which the host kernel needs 3285to know about. The id field is an architecture specific identifier for a 3286specific device. 3287 3288ARM/arm64 divides the id field into two parts, a device id and an 3289address type id specific to the individual device:: 3290 3291 bits: | 63 ... 32 | 31 ... 16 | 15 ... 0 | 3292 field: | 0x00000000 | device id | addr type id | 3293 3294ARM/arm64 currently only require this when using the in-kernel GIC 3295support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2 3296as the device id. When setting the base address for the guest's 3297mapping of the VGIC virtual CPU and distributor interface, the ioctl 3298must be called after calling KVM_CREATE_IRQCHIP, but before calling 3299KVM_RUN on any of the VCPUs. Calling this ioctl twice for any of the 3300base addresses will return -EEXIST. 3301 3302Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API 3303should be used instead. 3304 3305 33064.86 KVM_PPC_RTAS_DEFINE_TOKEN 3307------------------------------ 3308 3309:Capability: KVM_CAP_PPC_RTAS 3310:Architectures: ppc 3311:Type: vm ioctl 3312:Parameters: struct kvm_rtas_token_args 3313:Returns: 0 on success, -1 on error 3314 3315Defines a token value for a RTAS (Run Time Abstraction Services) 3316service in order to allow it to be handled in the kernel. The 3317argument struct gives the name of the service, which must be the name 3318of a service that has a kernel-side implementation. If the token 3319value is non-zero, it will be associated with that service, and 3320subsequent RTAS calls by the guest specifying that token will be 3321handled by the kernel. If the token value is 0, then any token 3322associated with the service will be forgotten, and subsequent RTAS 3323calls by the guest for that service will be passed to userspace to be 3324handled. 3325 33264.87 KVM_SET_GUEST_DEBUG 3327------------------------ 3328 3329:Capability: KVM_CAP_SET_GUEST_DEBUG 3330:Architectures: x86, s390, ppc, arm64 3331:Type: vcpu ioctl 3332:Parameters: struct kvm_guest_debug (in) 3333:Returns: 0 on success; -1 on error 3334 3335:: 3336 3337 struct kvm_guest_debug { 3338 __u32 control; 3339 __u32 pad; 3340 struct kvm_guest_debug_arch arch; 3341 }; 3342 3343Set up the processor specific debug registers and configure vcpu for 3344handling guest debug events. There are two parts to the structure, the 3345first a control bitfield indicates the type of debug events to handle 3346when running. Common control bits are: 3347 3348 - KVM_GUESTDBG_ENABLE: guest debugging is enabled 3349 - KVM_GUESTDBG_SINGLESTEP: the next run should single-step 3350 3351The top 16 bits of the control field are architecture specific control 3352flags which can include the following: 3353 3354 - KVM_GUESTDBG_USE_SW_BP: using software breakpoints [x86, arm64] 3355 - KVM_GUESTDBG_USE_HW_BP: using hardware breakpoints [x86, s390] 3356 - KVM_GUESTDBG_USE_HW: using hardware debug events [arm64] 3357 - KVM_GUESTDBG_INJECT_DB: inject DB type exception [x86] 3358 - KVM_GUESTDBG_INJECT_BP: inject BP type exception [x86] 3359 - KVM_GUESTDBG_EXIT_PENDING: trigger an immediate guest exit [s390] 3360 - KVM_GUESTDBG_BLOCKIRQ: avoid injecting interrupts/NMI/SMI [x86] 3361 3362For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints 3363are enabled in memory so we need to ensure breakpoint exceptions are 3364correctly trapped and the KVM run loop exits at the breakpoint and not 3365running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP 3366we need to ensure the guest vCPUs architecture specific registers are 3367updated to the correct (supplied) values. 3368 3369The second part of the structure is architecture specific and 3370typically contains a set of debug registers. 3371 3372For arm64 the number of debug registers is implementation defined and 3373can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and 3374KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number 3375indicating the number of supported registers. 3376 3377For ppc, the KVM_CAP_PPC_GUEST_DEBUG_SSTEP capability indicates whether 3378the single-step debug event (KVM_GUESTDBG_SINGLESTEP) is supported. 3379 3380Also when supported, KVM_CAP_SET_GUEST_DEBUG2 capability indicates the 3381supported KVM_GUESTDBG_* bits in the control field. 3382 3383When debug events exit the main run loop with the reason 3384KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run 3385structure containing architecture specific debug information. 3386 33874.88 KVM_GET_EMULATED_CPUID 3388--------------------------- 3389 3390:Capability: KVM_CAP_EXT_EMUL_CPUID 3391:Architectures: x86 3392:Type: system ioctl 3393:Parameters: struct kvm_cpuid2 (in/out) 3394:Returns: 0 on success, -1 on error 3395 3396:: 3397 3398 struct kvm_cpuid2 { 3399 __u32 nent; 3400 __u32 flags; 3401 struct kvm_cpuid_entry2 entries[0]; 3402 }; 3403 3404The member 'flags' is used for passing flags from userspace. 3405 3406:: 3407 3408 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0) 3409 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */ 3410 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */ 3411 3412 struct kvm_cpuid_entry2 { 3413 __u32 function; 3414 __u32 index; 3415 __u32 flags; 3416 __u32 eax; 3417 __u32 ebx; 3418 __u32 ecx; 3419 __u32 edx; 3420 __u32 padding[3]; 3421 }; 3422 3423This ioctl returns x86 cpuid features which are emulated by 3424kvm.Userspace can use the information returned by this ioctl to query 3425which features are emulated by kvm instead of being present natively. 3426 3427Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2 3428structure with the 'nent' field indicating the number of entries in 3429the variable-size array 'entries'. If the number of entries is too low 3430to describe the cpu capabilities, an error (E2BIG) is returned. If the 3431number is too high, the 'nent' field is adjusted and an error (ENOMEM) 3432is returned. If the number is just right, the 'nent' field is adjusted 3433to the number of valid entries in the 'entries' array, which is then 3434filled. 3435 3436The entries returned are the set CPUID bits of the respective features 3437which kvm emulates, as returned by the CPUID instruction, with unknown 3438or unsupported feature bits cleared. 3439 3440Features like x2apic, for example, may not be present in the host cpu 3441but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be 3442emulated efficiently and thus not included here. 3443 3444The fields in each entry are defined as follows: 3445 3446 function: 3447 the eax value used to obtain the entry 3448 index: 3449 the ecx value used to obtain the entry (for entries that are 3450 affected by ecx) 3451 flags: 3452 an OR of zero or more of the following: 3453 3454 KVM_CPUID_FLAG_SIGNIFCANT_INDEX: 3455 if the index field is valid 3456 3457 eax, ebx, ecx, edx: 3458 3459 the values returned by the cpuid instruction for 3460 this function/index combination 3461 34624.89 KVM_S390_MEM_OP 3463-------------------- 3464 3465:Capability: KVM_CAP_S390_MEM_OP 3466:Architectures: s390 3467:Type: vcpu ioctl 3468:Parameters: struct kvm_s390_mem_op (in) 3469:Returns: = 0 on success, 3470 < 0 on generic error (e.g. -EFAULT or -ENOMEM), 3471 > 0 if an exception occurred while walking the page tables 3472 3473Read or write data from/to the logical (virtual) memory of a VCPU. 3474 3475Parameters are specified via the following structure:: 3476 3477 struct kvm_s390_mem_op { 3478 __u64 gaddr; /* the guest address */ 3479 __u64 flags; /* flags */ 3480 __u32 size; /* amount of bytes */ 3481 __u32 op; /* type of operation */ 3482 __u64 buf; /* buffer in userspace */ 3483 __u8 ar; /* the access register number */ 3484 __u8 reserved[31]; /* should be set to 0 */ 3485 }; 3486 3487The type of operation is specified in the "op" field. It is either 3488KVM_S390_MEMOP_LOGICAL_READ for reading from logical memory space or 3489KVM_S390_MEMOP_LOGICAL_WRITE for writing to logical memory space. The 3490KVM_S390_MEMOP_F_CHECK_ONLY flag can be set in the "flags" field to check 3491whether the corresponding memory access would create an access exception 3492(without touching the data in the memory at the destination). In case an 3493access exception occurred while walking the MMU tables of the guest, the 3494ioctl returns a positive error number to indicate the type of exception. 3495This exception is also raised directly at the corresponding VCPU if the 3496flag KVM_S390_MEMOP_F_INJECT_EXCEPTION is set in the "flags" field. 3497 3498The start address of the memory region has to be specified in the "gaddr" 3499field, and the length of the region in the "size" field (which must not 3500be 0). The maximum value for "size" can be obtained by checking the 3501KVM_CAP_S390_MEM_OP capability. "buf" is the buffer supplied by the 3502userspace application where the read data should be written to for 3503KVM_S390_MEMOP_LOGICAL_READ, or where the data that should be written is 3504stored for a KVM_S390_MEMOP_LOGICAL_WRITE. When KVM_S390_MEMOP_F_CHECK_ONLY 3505is specified, "buf" is unused and can be NULL. "ar" designates the access 3506register number to be used; the valid range is 0..15. 3507 3508The "reserved" field is meant for future extensions. It is not used by 3509KVM with the currently defined set of flags. 3510 35114.90 KVM_S390_GET_SKEYS 3512----------------------- 3513 3514:Capability: KVM_CAP_S390_SKEYS 3515:Architectures: s390 3516:Type: vm ioctl 3517:Parameters: struct kvm_s390_skeys 3518:Returns: 0 on success, KVM_S390_GET_KEYS_NONE if guest is not using storage 3519 keys, negative value on error 3520 3521This ioctl is used to get guest storage key values on the s390 3522architecture. The ioctl takes parameters via the kvm_s390_skeys struct:: 3523 3524 struct kvm_s390_skeys { 3525 __u64 start_gfn; 3526 __u64 count; 3527 __u64 skeydata_addr; 3528 __u32 flags; 3529 __u32 reserved[9]; 3530 }; 3531 3532The start_gfn field is the number of the first guest frame whose storage keys 3533you want to get. 3534 3535The count field is the number of consecutive frames (starting from start_gfn) 3536whose storage keys to get. The count field must be at least 1 and the maximum 3537allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range 3538will cause the ioctl to return -EINVAL. 3539 3540The skeydata_addr field is the address to a buffer large enough to hold count 3541bytes. This buffer will be filled with storage key data by the ioctl. 3542 35434.91 KVM_S390_SET_SKEYS 3544----------------------- 3545 3546:Capability: KVM_CAP_S390_SKEYS 3547:Architectures: s390 3548:Type: vm ioctl 3549:Parameters: struct kvm_s390_skeys 3550:Returns: 0 on success, negative value on error 3551 3552This ioctl is used to set guest storage key values on the s390 3553architecture. The ioctl takes parameters via the kvm_s390_skeys struct. 3554See section on KVM_S390_GET_SKEYS for struct definition. 3555 3556The start_gfn field is the number of the first guest frame whose storage keys 3557you want to set. 3558 3559The count field is the number of consecutive frames (starting from start_gfn) 3560whose storage keys to get. The count field must be at least 1 and the maximum 3561allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range 3562will cause the ioctl to return -EINVAL. 3563 3564The skeydata_addr field is the address to a buffer containing count bytes of 3565storage keys. Each byte in the buffer will be set as the storage key for a 3566single frame starting at start_gfn for count frames. 3567 3568Note: If any architecturally invalid key value is found in the given data then 3569the ioctl will return -EINVAL. 3570 35714.92 KVM_S390_IRQ 3572----------------- 3573 3574:Capability: KVM_CAP_S390_INJECT_IRQ 3575:Architectures: s390 3576:Type: vcpu ioctl 3577:Parameters: struct kvm_s390_irq (in) 3578:Returns: 0 on success, -1 on error 3579 3580Errors: 3581 3582 3583 ====== ================================================================= 3584 EINVAL interrupt type is invalid 3585 type is KVM_S390_SIGP_STOP and flag parameter is invalid value, 3586 type is KVM_S390_INT_EXTERNAL_CALL and code is bigger 3587 than the maximum of VCPUs 3588 EBUSY type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped, 3589 type is KVM_S390_SIGP_STOP and a stop irq is already pending, 3590 type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt 3591 is already pending 3592 ====== ================================================================= 3593 3594Allows to inject an interrupt to the guest. 3595 3596Using struct kvm_s390_irq as a parameter allows 3597to inject additional payload which is not 3598possible via KVM_S390_INTERRUPT. 3599 3600Interrupt parameters are passed via kvm_s390_irq:: 3601 3602 struct kvm_s390_irq { 3603 __u64 type; 3604 union { 3605 struct kvm_s390_io_info io; 3606 struct kvm_s390_ext_info ext; 3607 struct kvm_s390_pgm_info pgm; 3608 struct kvm_s390_emerg_info emerg; 3609 struct kvm_s390_extcall_info extcall; 3610 struct kvm_s390_prefix_info prefix; 3611 struct kvm_s390_stop_info stop; 3612 struct kvm_s390_mchk_info mchk; 3613 char reserved[64]; 3614 } u; 3615 }; 3616 3617type can be one of the following: 3618 3619- KVM_S390_SIGP_STOP - sigp stop; parameter in .stop 3620- KVM_S390_PROGRAM_INT - program check; parameters in .pgm 3621- KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix 3622- KVM_S390_RESTART - restart; no parameters 3623- KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters 3624- KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters 3625- KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg 3626- KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall 3627- KVM_S390_MCHK - machine check interrupt; parameters in .mchk 3628 3629This is an asynchronous vcpu ioctl and can be invoked from any thread. 3630 36314.94 KVM_S390_GET_IRQ_STATE 3632--------------------------- 3633 3634:Capability: KVM_CAP_S390_IRQ_STATE 3635:Architectures: s390 3636:Type: vcpu ioctl 3637:Parameters: struct kvm_s390_irq_state (out) 3638:Returns: >= number of bytes copied into buffer, 3639 -EINVAL if buffer size is 0, 3640 -ENOBUFS if buffer size is too small to fit all pending interrupts, 3641 -EFAULT if the buffer address was invalid 3642 3643This ioctl allows userspace to retrieve the complete state of all currently 3644pending interrupts in a single buffer. Use cases include migration 3645and introspection. The parameter structure contains the address of a 3646userspace buffer and its length:: 3647 3648 struct kvm_s390_irq_state { 3649 __u64 buf; 3650 __u32 flags; /* will stay unused for compatibility reasons */ 3651 __u32 len; 3652 __u32 reserved[4]; /* will stay unused for compatibility reasons */ 3653 }; 3654 3655Userspace passes in the above struct and for each pending interrupt a 3656struct kvm_s390_irq is copied to the provided buffer. 3657 3658The structure contains a flags and a reserved field for future extensions. As 3659the kernel never checked for flags == 0 and QEMU never pre-zeroed flags and 3660reserved, these fields can not be used in the future without breaking 3661compatibility. 3662 3663If -ENOBUFS is returned the buffer provided was too small and userspace 3664may retry with a bigger buffer. 3665 36664.95 KVM_S390_SET_IRQ_STATE 3667--------------------------- 3668 3669:Capability: KVM_CAP_S390_IRQ_STATE 3670:Architectures: s390 3671:Type: vcpu ioctl 3672:Parameters: struct kvm_s390_irq_state (in) 3673:Returns: 0 on success, 3674 -EFAULT if the buffer address was invalid, 3675 -EINVAL for an invalid buffer length (see below), 3676 -EBUSY if there were already interrupts pending, 3677 errors occurring when actually injecting the 3678 interrupt. See KVM_S390_IRQ. 3679 3680This ioctl allows userspace to set the complete state of all cpu-local 3681interrupts currently pending for the vcpu. It is intended for restoring 3682interrupt state after a migration. The input parameter is a userspace buffer 3683containing a struct kvm_s390_irq_state:: 3684 3685 struct kvm_s390_irq_state { 3686 __u64 buf; 3687 __u32 flags; /* will stay unused for compatibility reasons */ 3688 __u32 len; 3689 __u32 reserved[4]; /* will stay unused for compatibility reasons */ 3690 }; 3691 3692The restrictions for flags and reserved apply as well. 3693(see KVM_S390_GET_IRQ_STATE) 3694 3695The userspace memory referenced by buf contains a struct kvm_s390_irq 3696for each interrupt to be injected into the guest. 3697If one of the interrupts could not be injected for some reason the 3698ioctl aborts. 3699 3700len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0 3701and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq), 3702which is the maximum number of possibly pending cpu-local interrupts. 3703 37044.96 KVM_SMI 3705------------ 3706 3707:Capability: KVM_CAP_X86_SMM 3708:Architectures: x86 3709:Type: vcpu ioctl 3710:Parameters: none 3711:Returns: 0 on success, -1 on error 3712 3713Queues an SMI on the thread's vcpu. 3714 37154.97 KVM_X86_SET_MSR_FILTER 3716---------------------------- 3717 3718:Capability: KVM_X86_SET_MSR_FILTER 3719:Architectures: x86 3720:Type: vm ioctl 3721:Parameters: struct kvm_msr_filter 3722:Returns: 0 on success, < 0 on error 3723 3724:: 3725 3726 struct kvm_msr_filter_range { 3727 #define KVM_MSR_FILTER_READ (1 << 0) 3728 #define KVM_MSR_FILTER_WRITE (1 << 1) 3729 __u32 flags; 3730 __u32 nmsrs; /* number of msrs in bitmap */ 3731 __u32 base; /* MSR index the bitmap starts at */ 3732 __u8 *bitmap; /* a 1 bit allows the operations in flags, 0 denies */ 3733 }; 3734 3735 #define KVM_MSR_FILTER_MAX_RANGES 16 3736 struct kvm_msr_filter { 3737 #define KVM_MSR_FILTER_DEFAULT_ALLOW (0 << 0) 3738 #define KVM_MSR_FILTER_DEFAULT_DENY (1 << 0) 3739 __u32 flags; 3740 struct kvm_msr_filter_range ranges[KVM_MSR_FILTER_MAX_RANGES]; 3741 }; 3742 3743flags values for ``struct kvm_msr_filter_range``: 3744 3745``KVM_MSR_FILTER_READ`` 3746 3747 Filter read accesses to MSRs using the given bitmap. A 0 in the bitmap 3748 indicates that a read should immediately fail, while a 1 indicates that 3749 a read for a particular MSR should be handled regardless of the default 3750 filter action. 3751 3752``KVM_MSR_FILTER_WRITE`` 3753 3754 Filter write accesses to MSRs using the given bitmap. A 0 in the bitmap 3755 indicates that a write should immediately fail, while a 1 indicates that 3756 a write for a particular MSR should be handled regardless of the default 3757 filter action. 3758 3759``KVM_MSR_FILTER_READ | KVM_MSR_FILTER_WRITE`` 3760 3761 Filter both read and write accesses to MSRs using the given bitmap. A 0 3762 in the bitmap indicates that both reads and writes should immediately fail, 3763 while a 1 indicates that reads and writes for a particular MSR are not 3764 filtered by this range. 3765 3766flags values for ``struct kvm_msr_filter``: 3767 3768``KVM_MSR_FILTER_DEFAULT_ALLOW`` 3769 3770 If no filter range matches an MSR index that is getting accessed, KVM will 3771 fall back to allowing access to the MSR. 3772 3773``KVM_MSR_FILTER_DEFAULT_DENY`` 3774 3775 If no filter range matches an MSR index that is getting accessed, KVM will 3776 fall back to rejecting access to the MSR. In this mode, all MSRs that should 3777 be processed by KVM need to explicitly be marked as allowed in the bitmaps. 3778 3779This ioctl allows user space to define up to 16 bitmaps of MSR ranges to 3780specify whether a certain MSR access should be explicitly filtered for or not. 3781 3782If this ioctl has never been invoked, MSR accesses are not guarded and the 3783default KVM in-kernel emulation behavior is fully preserved. 3784 3785Calling this ioctl with an empty set of ranges (all nmsrs == 0) disables MSR 3786filtering. In that mode, ``KVM_MSR_FILTER_DEFAULT_DENY`` is invalid and causes 3787an error. 3788 3789As soon as the filtering is in place, every MSR access is processed through 3790the filtering except for accesses to the x2APIC MSRs (from 0x800 to 0x8ff); 3791x2APIC MSRs are always allowed, independent of the ``default_allow`` setting, 3792and their behavior depends on the ``X2APIC_ENABLE`` bit of the APIC base 3793register. 3794 3795If a bit is within one of the defined ranges, read and write accesses are 3796guarded by the bitmap's value for the MSR index if the kind of access 3797is included in the ``struct kvm_msr_filter_range`` flags. If no range 3798cover this particular access, the behavior is determined by the flags 3799field in the kvm_msr_filter struct: ``KVM_MSR_FILTER_DEFAULT_ALLOW`` 3800and ``KVM_MSR_FILTER_DEFAULT_DENY``. 3801 3802Each bitmap range specifies a range of MSRs to potentially allow access on. 3803The range goes from MSR index [base .. base+nmsrs]. The flags field 3804indicates whether reads, writes or both reads and writes are filtered 3805by setting a 1 bit in the bitmap for the corresponding MSR index. 3806 3807If an MSR access is not permitted through the filtering, it generates a 3808#GP inside the guest. When combined with KVM_CAP_X86_USER_SPACE_MSR, that 3809allows user space to deflect and potentially handle various MSR accesses 3810into user space. 3811 3812If a vCPU is in running state while this ioctl is invoked, the vCPU may 3813experience inconsistent filtering behavior on MSR accesses. 3814 38154.98 KVM_CREATE_SPAPR_TCE_64 3816---------------------------- 3817 3818:Capability: KVM_CAP_SPAPR_TCE_64 3819:Architectures: powerpc 3820:Type: vm ioctl 3821:Parameters: struct kvm_create_spapr_tce_64 (in) 3822:Returns: file descriptor for manipulating the created TCE table 3823 3824This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit 3825windows, described in 4.62 KVM_CREATE_SPAPR_TCE 3826 3827This capability uses extended struct in ioctl interface:: 3828 3829 /* for KVM_CAP_SPAPR_TCE_64 */ 3830 struct kvm_create_spapr_tce_64 { 3831 __u64 liobn; 3832 __u32 page_shift; 3833 __u32 flags; 3834 __u64 offset; /* in pages */ 3835 __u64 size; /* in pages */ 3836 }; 3837 3838The aim of extension is to support an additional bigger DMA window with 3839a variable page size. 3840KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and 3841a bus offset of the corresponding DMA window, @size and @offset are numbers 3842of IOMMU pages. 3843 3844@flags are not used at the moment. 3845 3846The rest of functionality is identical to KVM_CREATE_SPAPR_TCE. 3847 38484.99 KVM_REINJECT_CONTROL 3849------------------------- 3850 3851:Capability: KVM_CAP_REINJECT_CONTROL 3852:Architectures: x86 3853:Type: vm ioctl 3854:Parameters: struct kvm_reinject_control (in) 3855:Returns: 0 on success, 3856 -EFAULT if struct kvm_reinject_control cannot be read, 3857 -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier. 3858 3859i8254 (PIT) has two modes, reinject and !reinject. The default is reinject, 3860where KVM queues elapsed i8254 ticks and monitors completion of interrupt from 3861vector(s) that i8254 injects. Reinject mode dequeues a tick and injects its 3862interrupt whenever there isn't a pending interrupt from i8254. 3863!reinject mode injects an interrupt as soon as a tick arrives. 3864 3865:: 3866 3867 struct kvm_reinject_control { 3868 __u8 pit_reinject; 3869 __u8 reserved[31]; 3870 }; 3871 3872pit_reinject = 0 (!reinject mode) is recommended, unless running an old 3873operating system that uses the PIT for timing (e.g. Linux 2.4.x). 3874 38754.100 KVM_PPC_CONFIGURE_V3_MMU 3876------------------------------ 3877 3878:Capability: KVM_CAP_PPC_RADIX_MMU or KVM_CAP_PPC_HASH_MMU_V3 3879:Architectures: ppc 3880:Type: vm ioctl 3881:Parameters: struct kvm_ppc_mmuv3_cfg (in) 3882:Returns: 0 on success, 3883 -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read, 3884 -EINVAL if the configuration is invalid 3885 3886This ioctl controls whether the guest will use radix or HPT (hashed 3887page table) translation, and sets the pointer to the process table for 3888the guest. 3889 3890:: 3891 3892 struct kvm_ppc_mmuv3_cfg { 3893 __u64 flags; 3894 __u64 process_table; 3895 }; 3896 3897There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and 3898KVM_PPC_MMUV3_GTSE. KVM_PPC_MMUV3_RADIX, if set, configures the guest 3899to use radix tree translation, and if clear, to use HPT translation. 3900KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest 3901to be able to use the global TLB and SLB invalidation instructions; 3902if clear, the guest may not use these instructions. 3903 3904The process_table field specifies the address and size of the guest 3905process table, which is in the guest's space. This field is formatted 3906as the second doubleword of the partition table entry, as defined in 3907the Power ISA V3.00, Book III section 5.7.6.1. 3908 39094.101 KVM_PPC_GET_RMMU_INFO 3910--------------------------- 3911 3912:Capability: KVM_CAP_PPC_RADIX_MMU 3913:Architectures: ppc 3914:Type: vm ioctl 3915:Parameters: struct kvm_ppc_rmmu_info (out) 3916:Returns: 0 on success, 3917 -EFAULT if struct kvm_ppc_rmmu_info cannot be written, 3918 -EINVAL if no useful information can be returned 3919 3920This ioctl returns a structure containing two things: (a) a list 3921containing supported radix tree geometries, and (b) a list that maps 3922page sizes to put in the "AP" (actual page size) field for the tlbie 3923(TLB invalidate entry) instruction. 3924 3925:: 3926 3927 struct kvm_ppc_rmmu_info { 3928 struct kvm_ppc_radix_geom { 3929 __u8 page_shift; 3930 __u8 level_bits[4]; 3931 __u8 pad[3]; 3932 } geometries[8]; 3933 __u32 ap_encodings[8]; 3934 }; 3935 3936The geometries[] field gives up to 8 supported geometries for the 3937radix page table, in terms of the log base 2 of the smallest page 3938size, and the number of bits indexed at each level of the tree, from 3939the PTE level up to the PGD level in that order. Any unused entries 3940will have 0 in the page_shift field. 3941 3942The ap_encodings gives the supported page sizes and their AP field 3943encodings, encoded with the AP value in the top 3 bits and the log 3944base 2 of the page size in the bottom 6 bits. 3945 39464.102 KVM_PPC_RESIZE_HPT_PREPARE 3947-------------------------------- 3948 3949:Capability: KVM_CAP_SPAPR_RESIZE_HPT 3950:Architectures: powerpc 3951:Type: vm ioctl 3952:Parameters: struct kvm_ppc_resize_hpt (in) 3953:Returns: 0 on successful completion, 3954 >0 if a new HPT is being prepared, the value is an estimated 3955 number of milliseconds until preparation is complete, 3956 -EFAULT if struct kvm_reinject_control cannot be read, 3957 -EINVAL if the supplied shift or flags are invalid, 3958 -ENOMEM if unable to allocate the new HPT, 3959 3960Used to implement the PAPR extension for runtime resizing of a guest's 3961Hashed Page Table (HPT). Specifically this starts, stops or monitors 3962the preparation of a new potential HPT for the guest, essentially 3963implementing the H_RESIZE_HPT_PREPARE hypercall. 3964 3965:: 3966 3967 struct kvm_ppc_resize_hpt { 3968 __u64 flags; 3969 __u32 shift; 3970 __u32 pad; 3971 }; 3972 3973If called with shift > 0 when there is no pending HPT for the guest, 3974this begins preparation of a new pending HPT of size 2^(shift) bytes. 3975It then returns a positive integer with the estimated number of 3976milliseconds until preparation is complete. 3977 3978If called when there is a pending HPT whose size does not match that 3979requested in the parameters, discards the existing pending HPT and 3980creates a new one as above. 3981 3982If called when there is a pending HPT of the size requested, will: 3983 3984 * If preparation of the pending HPT is already complete, return 0 3985 * If preparation of the pending HPT has failed, return an error 3986 code, then discard the pending HPT. 3987 * If preparation of the pending HPT is still in progress, return an 3988 estimated number of milliseconds until preparation is complete. 3989 3990If called with shift == 0, discards any currently pending HPT and 3991returns 0 (i.e. cancels any in-progress preparation). 3992 3993flags is reserved for future expansion, currently setting any bits in 3994flags will result in an -EINVAL. 3995 3996Normally this will be called repeatedly with the same parameters until 3997it returns <= 0. The first call will initiate preparation, subsequent 3998ones will monitor preparation until it completes or fails. 3999 40004.103 KVM_PPC_RESIZE_HPT_COMMIT 4001------------------------------- 4002 4003:Capability: KVM_CAP_SPAPR_RESIZE_HPT 4004:Architectures: powerpc 4005:Type: vm ioctl 4006:Parameters: struct kvm_ppc_resize_hpt (in) 4007:Returns: 0 on successful completion, 4008 -EFAULT if struct kvm_reinject_control cannot be read, 4009 -EINVAL if the supplied shift or flags are invalid, 4010 -ENXIO is there is no pending HPT, or the pending HPT doesn't 4011 have the requested size, 4012 -EBUSY if the pending HPT is not fully prepared, 4013 -ENOSPC if there was a hash collision when moving existing 4014 HPT entries to the new HPT, 4015 -EIO on other error conditions 4016 4017Used to implement the PAPR extension for runtime resizing of a guest's 4018Hashed Page Table (HPT). Specifically this requests that the guest be 4019transferred to working with the new HPT, essentially implementing the 4020H_RESIZE_HPT_COMMIT hypercall. 4021 4022:: 4023 4024 struct kvm_ppc_resize_hpt { 4025 __u64 flags; 4026 __u32 shift; 4027 __u32 pad; 4028 }; 4029 4030This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has 4031returned 0 with the same parameters. In other cases 4032KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or 4033-EBUSY, though others may be possible if the preparation was started, 4034but failed). 4035 4036This will have undefined effects on the guest if it has not already 4037placed itself in a quiescent state where no vcpu will make MMU enabled 4038memory accesses. 4039 4040On succsful completion, the pending HPT will become the guest's active 4041HPT and the previous HPT will be discarded. 4042 4043On failure, the guest will still be operating on its previous HPT. 4044 40454.104 KVM_X86_GET_MCE_CAP_SUPPORTED 4046----------------------------------- 4047 4048:Capability: KVM_CAP_MCE 4049:Architectures: x86 4050:Type: system ioctl 4051:Parameters: u64 mce_cap (out) 4052:Returns: 0 on success, -1 on error 4053 4054Returns supported MCE capabilities. The u64 mce_cap parameter 4055has the same format as the MSR_IA32_MCG_CAP register. Supported 4056capabilities will have the corresponding bits set. 4057 40584.105 KVM_X86_SETUP_MCE 4059----------------------- 4060 4061:Capability: KVM_CAP_MCE 4062:Architectures: x86 4063:Type: vcpu ioctl 4064:Parameters: u64 mcg_cap (in) 4065:Returns: 0 on success, 4066 -EFAULT if u64 mcg_cap cannot be read, 4067 -EINVAL if the requested number of banks is invalid, 4068 -EINVAL if requested MCE capability is not supported. 4069 4070Initializes MCE support for use. The u64 mcg_cap parameter 4071has the same format as the MSR_IA32_MCG_CAP register and 4072specifies which capabilities should be enabled. The maximum 4073supported number of error-reporting banks can be retrieved when 4074checking for KVM_CAP_MCE. The supported capabilities can be 4075retrieved with KVM_X86_GET_MCE_CAP_SUPPORTED. 4076 40774.106 KVM_X86_SET_MCE 4078--------------------- 4079 4080:Capability: KVM_CAP_MCE 4081:Architectures: x86 4082:Type: vcpu ioctl 4083:Parameters: struct kvm_x86_mce (in) 4084:Returns: 0 on success, 4085 -EFAULT if struct kvm_x86_mce cannot be read, 4086 -EINVAL if the bank number is invalid, 4087 -EINVAL if VAL bit is not set in status field. 4088 4089Inject a machine check error (MCE) into the guest. The input 4090parameter is:: 4091 4092 struct kvm_x86_mce { 4093 __u64 status; 4094 __u64 addr; 4095 __u64 misc; 4096 __u64 mcg_status; 4097 __u8 bank; 4098 __u8 pad1[7]; 4099 __u64 pad2[3]; 4100 }; 4101 4102If the MCE being reported is an uncorrected error, KVM will 4103inject it as an MCE exception into the guest. If the guest 4104MCG_STATUS register reports that an MCE is in progress, KVM 4105causes an KVM_EXIT_SHUTDOWN vmexit. 4106 4107Otherwise, if the MCE is a corrected error, KVM will just 4108store it in the corresponding bank (provided this bank is 4109not holding a previously reported uncorrected error). 4110 41114.107 KVM_S390_GET_CMMA_BITS 4112---------------------------- 4113 4114:Capability: KVM_CAP_S390_CMMA_MIGRATION 4115:Architectures: s390 4116:Type: vm ioctl 4117:Parameters: struct kvm_s390_cmma_log (in, out) 4118:Returns: 0 on success, a negative value on error 4119 4120This ioctl is used to get the values of the CMMA bits on the s390 4121architecture. It is meant to be used in two scenarios: 4122 4123- During live migration to save the CMMA values. Live migration needs 4124 to be enabled via the KVM_REQ_START_MIGRATION VM property. 4125- To non-destructively peek at the CMMA values, with the flag 4126 KVM_S390_CMMA_PEEK set. 4127 4128The ioctl takes parameters via the kvm_s390_cmma_log struct. The desired 4129values are written to a buffer whose location is indicated via the "values" 4130member in the kvm_s390_cmma_log struct. The values in the input struct are 4131also updated as needed. 4132 4133Each CMMA value takes up one byte. 4134 4135:: 4136 4137 struct kvm_s390_cmma_log { 4138 __u64 start_gfn; 4139 __u32 count; 4140 __u32 flags; 4141 union { 4142 __u64 remaining; 4143 __u64 mask; 4144 }; 4145 __u64 values; 4146 }; 4147 4148start_gfn is the number of the first guest frame whose CMMA values are 4149to be retrieved, 4150 4151count is the length of the buffer in bytes, 4152 4153values points to the buffer where the result will be written to. 4154 4155If count is greater than KVM_S390_SKEYS_MAX, then it is considered to be 4156KVM_S390_SKEYS_MAX. KVM_S390_SKEYS_MAX is re-used for consistency with 4157other ioctls. 4158 4159The result is written in the buffer pointed to by the field values, and 4160the values of the input parameter are updated as follows. 4161 4162Depending on the flags, different actions are performed. The only 4163supported flag so far is KVM_S390_CMMA_PEEK. 4164 4165The default behaviour if KVM_S390_CMMA_PEEK is not set is: 4166start_gfn will indicate the first page frame whose CMMA bits were dirty. 4167It is not necessarily the same as the one passed as input, as clean pages 4168are skipped. 4169 4170count will indicate the number of bytes actually written in the buffer. 4171It can (and very often will) be smaller than the input value, since the 4172buffer is only filled until 16 bytes of clean values are found (which 4173are then not copied in the buffer). Since a CMMA migration block needs 4174the base address and the length, for a total of 16 bytes, we will send 4175back some clean data if there is some dirty data afterwards, as long as 4176the size of the clean data does not exceed the size of the header. This 4177allows to minimize the amount of data to be saved or transferred over 4178the network at the expense of more roundtrips to userspace. The next 4179invocation of the ioctl will skip over all the clean values, saving 4180potentially more than just the 16 bytes we found. 4181 4182If KVM_S390_CMMA_PEEK is set: 4183the existing storage attributes are read even when not in migration 4184mode, and no other action is performed; 4185 4186the output start_gfn will be equal to the input start_gfn, 4187 4188the output count will be equal to the input count, except if the end of 4189memory has been reached. 4190 4191In both cases: 4192the field "remaining" will indicate the total number of dirty CMMA values 4193still remaining, or 0 if KVM_S390_CMMA_PEEK is set and migration mode is 4194not enabled. 4195 4196mask is unused. 4197 4198values points to the userspace buffer where the result will be stored. 4199 4200This ioctl can fail with -ENOMEM if not enough memory can be allocated to 4201complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if 4202KVM_S390_CMMA_PEEK is not set but migration mode was not enabled, with 4203-EFAULT if the userspace address is invalid or if no page table is 4204present for the addresses (e.g. when using hugepages). 4205 42064.108 KVM_S390_SET_CMMA_BITS 4207---------------------------- 4208 4209:Capability: KVM_CAP_S390_CMMA_MIGRATION 4210:Architectures: s390 4211:Type: vm ioctl 4212:Parameters: struct kvm_s390_cmma_log (in) 4213:Returns: 0 on success, a negative value on error 4214 4215This ioctl is used to set the values of the CMMA bits on the s390 4216architecture. It is meant to be used during live migration to restore 4217the CMMA values, but there are no restrictions on its use. 4218The ioctl takes parameters via the kvm_s390_cmma_values struct. 4219Each CMMA value takes up one byte. 4220 4221:: 4222 4223 struct kvm_s390_cmma_log { 4224 __u64 start_gfn; 4225 __u32 count; 4226 __u32 flags; 4227 union { 4228 __u64 remaining; 4229 __u64 mask; 4230 }; 4231 __u64 values; 4232 }; 4233 4234start_gfn indicates the starting guest frame number, 4235 4236count indicates how many values are to be considered in the buffer, 4237 4238flags is not used and must be 0. 4239 4240mask indicates which PGSTE bits are to be considered. 4241 4242remaining is not used. 4243 4244values points to the buffer in userspace where to store the values. 4245 4246This ioctl can fail with -ENOMEM if not enough memory can be allocated to 4247complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if 4248the count field is too large (e.g. more than KVM_S390_CMMA_SIZE_MAX) or 4249if the flags field was not 0, with -EFAULT if the userspace address is 4250invalid, if invalid pages are written to (e.g. after the end of memory) 4251or if no page table is present for the addresses (e.g. when using 4252hugepages). 4253 42544.109 KVM_PPC_GET_CPU_CHAR 4255-------------------------- 4256 4257:Capability: KVM_CAP_PPC_GET_CPU_CHAR 4258:Architectures: powerpc 4259:Type: vm ioctl 4260:Parameters: struct kvm_ppc_cpu_char (out) 4261:Returns: 0 on successful completion, 4262 -EFAULT if struct kvm_ppc_cpu_char cannot be written 4263 4264This ioctl gives userspace information about certain characteristics 4265of the CPU relating to speculative execution of instructions and 4266possible information leakage resulting from speculative execution (see 4267CVE-2017-5715, CVE-2017-5753 and CVE-2017-5754). The information is 4268returned in struct kvm_ppc_cpu_char, which looks like this:: 4269 4270 struct kvm_ppc_cpu_char { 4271 __u64 character; /* characteristics of the CPU */ 4272 __u64 behaviour; /* recommended software behaviour */ 4273 __u64 character_mask; /* valid bits in character */ 4274 __u64 behaviour_mask; /* valid bits in behaviour */ 4275 }; 4276 4277For extensibility, the character_mask and behaviour_mask fields 4278indicate which bits of character and behaviour have been filled in by 4279the kernel. If the set of defined bits is extended in future then 4280userspace will be able to tell whether it is running on a kernel that 4281knows about the new bits. 4282 4283The character field describes attributes of the CPU which can help 4284with preventing inadvertent information disclosure - specifically, 4285whether there is an instruction to flash-invalidate the L1 data cache 4286(ori 30,30,0 or mtspr SPRN_TRIG2,rN), whether the L1 data cache is set 4287to a mode where entries can only be used by the thread that created 4288them, whether the bcctr[l] instruction prevents speculation, and 4289whether a speculation barrier instruction (ori 31,31,0) is provided. 4290 4291The behaviour field describes actions that software should take to 4292prevent inadvertent information disclosure, and thus describes which 4293vulnerabilities the hardware is subject to; specifically whether the 4294L1 data cache should be flushed when returning to user mode from the 4295kernel, and whether a speculation barrier should be placed between an 4296array bounds check and the array access. 4297 4298These fields use the same bit definitions as the new 4299H_GET_CPU_CHARACTERISTICS hypercall. 4300 43014.110 KVM_MEMORY_ENCRYPT_OP 4302--------------------------- 4303 4304:Capability: basic 4305:Architectures: x86 4306:Type: vm 4307:Parameters: an opaque platform specific structure (in/out) 4308:Returns: 0 on success; -1 on error 4309 4310If the platform supports creating encrypted VMs then this ioctl can be used 4311for issuing platform-specific memory encryption commands to manage those 4312encrypted VMs. 4313 4314Currently, this ioctl is used for issuing Secure Encrypted Virtualization 4315(SEV) commands on AMD Processors. The SEV commands are defined in 4316Documentation/virt/kvm/amd-memory-encryption.rst. 4317 43184.111 KVM_MEMORY_ENCRYPT_REG_REGION 4319----------------------------------- 4320 4321:Capability: basic 4322:Architectures: x86 4323:Type: system 4324:Parameters: struct kvm_enc_region (in) 4325:Returns: 0 on success; -1 on error 4326 4327This ioctl can be used to register a guest memory region which may 4328contain encrypted data (e.g. guest RAM, SMRAM etc). 4329 4330It is used in the SEV-enabled guest. When encryption is enabled, a guest 4331memory region may contain encrypted data. The SEV memory encryption 4332engine uses a tweak such that two identical plaintext pages, each at 4333different locations will have differing ciphertexts. So swapping or 4334moving ciphertext of those pages will not result in plaintext being 4335swapped. So relocating (or migrating) physical backing pages for the SEV 4336guest will require some additional steps. 4337 4338Note: The current SEV key management spec does not provide commands to 4339swap or migrate (move) ciphertext pages. Hence, for now we pin the guest 4340memory region registered with the ioctl. 4341 43424.112 KVM_MEMORY_ENCRYPT_UNREG_REGION 4343------------------------------------- 4344 4345:Capability: basic 4346:Architectures: x86 4347:Type: system 4348:Parameters: struct kvm_enc_region (in) 4349:Returns: 0 on success; -1 on error 4350 4351This ioctl can be used to unregister the guest memory region registered 4352with KVM_MEMORY_ENCRYPT_REG_REGION ioctl above. 4353 43544.113 KVM_HYPERV_EVENTFD 4355------------------------ 4356 4357:Capability: KVM_CAP_HYPERV_EVENTFD 4358:Architectures: x86 4359:Type: vm ioctl 4360:Parameters: struct kvm_hyperv_eventfd (in) 4361 4362This ioctl (un)registers an eventfd to receive notifications from the guest on 4363the specified Hyper-V connection id through the SIGNAL_EVENT hypercall, without 4364causing a user exit. SIGNAL_EVENT hypercall with non-zero event flag number 4365(bits 24-31) still triggers a KVM_EXIT_HYPERV_HCALL user exit. 4366 4367:: 4368 4369 struct kvm_hyperv_eventfd { 4370 __u32 conn_id; 4371 __s32 fd; 4372 __u32 flags; 4373 __u32 padding[3]; 4374 }; 4375 4376The conn_id field should fit within 24 bits:: 4377 4378 #define KVM_HYPERV_CONN_ID_MASK 0x00ffffff 4379 4380The acceptable values for the flags field are:: 4381 4382 #define KVM_HYPERV_EVENTFD_DEASSIGN (1 << 0) 4383 4384:Returns: 0 on success, 4385 -EINVAL if conn_id or flags is outside the allowed range, 4386 -ENOENT on deassign if the conn_id isn't registered, 4387 -EEXIST on assign if the conn_id is already registered 4388 43894.114 KVM_GET_NESTED_STATE 4390-------------------------- 4391 4392:Capability: KVM_CAP_NESTED_STATE 4393:Architectures: x86 4394:Type: vcpu ioctl 4395:Parameters: struct kvm_nested_state (in/out) 4396:Returns: 0 on success, -1 on error 4397 4398Errors: 4399 4400 ===== ============================================================= 4401 E2BIG the total state size exceeds the value of 'size' specified by 4402 the user; the size required will be written into size. 4403 ===== ============================================================= 4404 4405:: 4406 4407 struct kvm_nested_state { 4408 __u16 flags; 4409 __u16 format; 4410 __u32 size; 4411 4412 union { 4413 struct kvm_vmx_nested_state_hdr vmx; 4414 struct kvm_svm_nested_state_hdr svm; 4415 4416 /* Pad the header to 128 bytes. */ 4417 __u8 pad[120]; 4418 } hdr; 4419 4420 union { 4421 struct kvm_vmx_nested_state_data vmx[0]; 4422 struct kvm_svm_nested_state_data svm[0]; 4423 } data; 4424 }; 4425 4426 #define KVM_STATE_NESTED_GUEST_MODE 0x00000001 4427 #define KVM_STATE_NESTED_RUN_PENDING 0x00000002 4428 #define KVM_STATE_NESTED_EVMCS 0x00000004 4429 4430 #define KVM_STATE_NESTED_FORMAT_VMX 0 4431 #define KVM_STATE_NESTED_FORMAT_SVM 1 4432 4433 #define KVM_STATE_NESTED_VMX_VMCS_SIZE 0x1000 4434 4435 #define KVM_STATE_NESTED_VMX_SMM_GUEST_MODE 0x00000001 4436 #define KVM_STATE_NESTED_VMX_SMM_VMXON 0x00000002 4437 4438 #define KVM_STATE_VMX_PREEMPTION_TIMER_DEADLINE 0x00000001 4439 4440 struct kvm_vmx_nested_state_hdr { 4441 __u64 vmxon_pa; 4442 __u64 vmcs12_pa; 4443 4444 struct { 4445 __u16 flags; 4446 } smm; 4447 4448 __u32 flags; 4449 __u64 preemption_timer_deadline; 4450 }; 4451 4452 struct kvm_vmx_nested_state_data { 4453 __u8 vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE]; 4454 __u8 shadow_vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE]; 4455 }; 4456 4457This ioctl copies the vcpu's nested virtualization state from the kernel to 4458userspace. 4459 4460The maximum size of the state can be retrieved by passing KVM_CAP_NESTED_STATE 4461to the KVM_CHECK_EXTENSION ioctl(). 4462 44634.115 KVM_SET_NESTED_STATE 4464-------------------------- 4465 4466:Capability: KVM_CAP_NESTED_STATE 4467:Architectures: x86 4468:Type: vcpu ioctl 4469:Parameters: struct kvm_nested_state (in) 4470:Returns: 0 on success, -1 on error 4471 4472This copies the vcpu's kvm_nested_state struct from userspace to the kernel. 4473For the definition of struct kvm_nested_state, see KVM_GET_NESTED_STATE. 4474 44754.116 KVM_(UN)REGISTER_COALESCED_MMIO 4476------------------------------------- 4477 4478:Capability: KVM_CAP_COALESCED_MMIO (for coalesced mmio) 4479 KVM_CAP_COALESCED_PIO (for coalesced pio) 4480:Architectures: all 4481:Type: vm ioctl 4482:Parameters: struct kvm_coalesced_mmio_zone 4483:Returns: 0 on success, < 0 on error 4484 4485Coalesced I/O is a performance optimization that defers hardware 4486register write emulation so that userspace exits are avoided. It is 4487typically used to reduce the overhead of emulating frequently accessed 4488hardware registers. 4489 4490When a hardware register is configured for coalesced I/O, write accesses 4491do not exit to userspace and their value is recorded in a ring buffer 4492that is shared between kernel and userspace. 4493 4494Coalesced I/O is used if one or more write accesses to a hardware 4495register can be deferred until a read or a write to another hardware 4496register on the same device. This last access will cause a vmexit and 4497userspace will process accesses from the ring buffer before emulating 4498it. That will avoid exiting to userspace on repeated writes. 4499 4500Coalesced pio is based on coalesced mmio. There is little difference 4501between coalesced mmio and pio except that coalesced pio records accesses 4502to I/O ports. 4503 45044.117 KVM_CLEAR_DIRTY_LOG (vm ioctl) 4505------------------------------------ 4506 4507:Capability: KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 4508:Architectures: x86, arm, arm64, mips 4509:Type: vm ioctl 4510:Parameters: struct kvm_clear_dirty_log (in) 4511:Returns: 0 on success, -1 on error 4512 4513:: 4514 4515 /* for KVM_CLEAR_DIRTY_LOG */ 4516 struct kvm_clear_dirty_log { 4517 __u32 slot; 4518 __u32 num_pages; 4519 __u64 first_page; 4520 union { 4521 void __user *dirty_bitmap; /* one bit per page */ 4522 __u64 padding; 4523 }; 4524 }; 4525 4526The ioctl clears the dirty status of pages in a memory slot, according to 4527the bitmap that is passed in struct kvm_clear_dirty_log's dirty_bitmap 4528field. Bit 0 of the bitmap corresponds to page "first_page" in the 4529memory slot, and num_pages is the size in bits of the input bitmap. 4530first_page must be a multiple of 64; num_pages must also be a multiple of 453164 unless first_page + num_pages is the size of the memory slot. For each 4532bit that is set in the input bitmap, the corresponding page is marked "clean" 4533in KVM's dirty bitmap, and dirty tracking is re-enabled for that page 4534(for example via write-protection, or by clearing the dirty bit in 4535a page table entry). 4536 4537If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies 4538the address space for which you want to clear the dirty status. See 4539KVM_SET_USER_MEMORY_REGION for details on the usage of slot field. 4540 4541This ioctl is mostly useful when KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 4542is enabled; for more information, see the description of the capability. 4543However, it can always be used as long as KVM_CHECK_EXTENSION confirms 4544that KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is present. 4545 45464.118 KVM_GET_SUPPORTED_HV_CPUID 4547-------------------------------- 4548 4549:Capability: KVM_CAP_HYPERV_CPUID (vcpu), KVM_CAP_SYS_HYPERV_CPUID (system) 4550:Architectures: x86 4551:Type: system ioctl, vcpu ioctl 4552:Parameters: struct kvm_cpuid2 (in/out) 4553:Returns: 0 on success, -1 on error 4554 4555:: 4556 4557 struct kvm_cpuid2 { 4558 __u32 nent; 4559 __u32 padding; 4560 struct kvm_cpuid_entry2 entries[0]; 4561 }; 4562 4563 struct kvm_cpuid_entry2 { 4564 __u32 function; 4565 __u32 index; 4566 __u32 flags; 4567 __u32 eax; 4568 __u32 ebx; 4569 __u32 ecx; 4570 __u32 edx; 4571 __u32 padding[3]; 4572 }; 4573 4574This ioctl returns x86 cpuid features leaves related to Hyper-V emulation in 4575KVM. Userspace can use the information returned by this ioctl to construct 4576cpuid information presented to guests consuming Hyper-V enlightenments (e.g. 4577Windows or Hyper-V guests). 4578 4579CPUID feature leaves returned by this ioctl are defined by Hyper-V Top Level 4580Functional Specification (TLFS). These leaves can't be obtained with 4581KVM_GET_SUPPORTED_CPUID ioctl because some of them intersect with KVM feature 4582leaves (0x40000000, 0x40000001). 4583 4584Currently, the following list of CPUID leaves are returned: 4585 4586 - HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS 4587 - HYPERV_CPUID_INTERFACE 4588 - HYPERV_CPUID_VERSION 4589 - HYPERV_CPUID_FEATURES 4590 - HYPERV_CPUID_ENLIGHTMENT_INFO 4591 - HYPERV_CPUID_IMPLEMENT_LIMITS 4592 - HYPERV_CPUID_NESTED_FEATURES 4593 - HYPERV_CPUID_SYNDBG_VENDOR_AND_MAX_FUNCTIONS 4594 - HYPERV_CPUID_SYNDBG_INTERFACE 4595 - HYPERV_CPUID_SYNDBG_PLATFORM_CAPABILITIES 4596 4597Userspace invokes KVM_GET_SUPPORTED_HV_CPUID by passing a kvm_cpuid2 structure 4598with the 'nent' field indicating the number of entries in the variable-size 4599array 'entries'. If the number of entries is too low to describe all Hyper-V 4600feature leaves, an error (E2BIG) is returned. If the number is more or equal 4601to the number of Hyper-V feature leaves, the 'nent' field is adjusted to the 4602number of valid entries in the 'entries' array, which is then filled. 4603 4604'index' and 'flags' fields in 'struct kvm_cpuid_entry2' are currently reserved, 4605userspace should not expect to get any particular value there. 4606 4607Note, vcpu version of KVM_GET_SUPPORTED_HV_CPUID is currently deprecated. Unlike 4608system ioctl which exposes all supported feature bits unconditionally, vcpu 4609version has the following quirks: 4610 4611- HYPERV_CPUID_NESTED_FEATURES leaf and HV_X64_ENLIGHTENED_VMCS_RECOMMENDED 4612 feature bit are only exposed when Enlightened VMCS was previously enabled 4613 on the corresponding vCPU (KVM_CAP_HYPERV_ENLIGHTENED_VMCS). 4614- HV_STIMER_DIRECT_MODE_AVAILABLE bit is only exposed with in-kernel LAPIC. 4615 (presumes KVM_CREATE_IRQCHIP has already been called). 4616 46174.119 KVM_ARM_VCPU_FINALIZE 4618--------------------------- 4619 4620:Architectures: arm, arm64 4621:Type: vcpu ioctl 4622:Parameters: int feature (in) 4623:Returns: 0 on success, -1 on error 4624 4625Errors: 4626 4627 ====== ============================================================== 4628 EPERM feature not enabled, needs configuration, or already finalized 4629 EINVAL feature unknown or not present 4630 ====== ============================================================== 4631 4632Recognised values for feature: 4633 4634 ===== =========================================== 4635 arm64 KVM_ARM_VCPU_SVE (requires KVM_CAP_ARM_SVE) 4636 ===== =========================================== 4637 4638Finalizes the configuration of the specified vcpu feature. 4639 4640The vcpu must already have been initialised, enabling the affected feature, by 4641means of a successful KVM_ARM_VCPU_INIT call with the appropriate flag set in 4642features[]. 4643 4644For affected vcpu features, this is a mandatory step that must be performed 4645before the vcpu is fully usable. 4646 4647Between KVM_ARM_VCPU_INIT and KVM_ARM_VCPU_FINALIZE, the feature may be 4648configured by use of ioctls such as KVM_SET_ONE_REG. The exact configuration 4649that should be performaned and how to do it are feature-dependent. 4650 4651Other calls that depend on a particular feature being finalized, such as 4652KVM_RUN, KVM_GET_REG_LIST, KVM_GET_ONE_REG and KVM_SET_ONE_REG, will fail with 4653-EPERM unless the feature has already been finalized by means of a 4654KVM_ARM_VCPU_FINALIZE call. 4655 4656See KVM_ARM_VCPU_INIT for details of vcpu features that require finalization 4657using this ioctl. 4658 46594.120 KVM_SET_PMU_EVENT_FILTER 4660------------------------------ 4661 4662:Capability: KVM_CAP_PMU_EVENT_FILTER 4663:Architectures: x86 4664:Type: vm ioctl 4665:Parameters: struct kvm_pmu_event_filter (in) 4666:Returns: 0 on success, -1 on error 4667 4668:: 4669 4670 struct kvm_pmu_event_filter { 4671 __u32 action; 4672 __u32 nevents; 4673 __u32 fixed_counter_bitmap; 4674 __u32 flags; 4675 __u32 pad[4]; 4676 __u64 events[0]; 4677 }; 4678 4679This ioctl restricts the set of PMU events that the guest can program. 4680The argument holds a list of events which will be allowed or denied. 4681The eventsel+umask of each event the guest attempts to program is compared 4682against the events field to determine whether the guest should have access. 4683The events field only controls general purpose counters; fixed purpose 4684counters are controlled by the fixed_counter_bitmap. 4685 4686No flags are defined yet, the field must be zero. 4687 4688Valid values for 'action':: 4689 4690 #define KVM_PMU_EVENT_ALLOW 0 4691 #define KVM_PMU_EVENT_DENY 1 4692 46934.121 KVM_PPC_SVM_OFF 4694--------------------- 4695 4696:Capability: basic 4697:Architectures: powerpc 4698:Type: vm ioctl 4699:Parameters: none 4700:Returns: 0 on successful completion, 4701 4702Errors: 4703 4704 ====== ================================================================ 4705 EINVAL if ultravisor failed to terminate the secure guest 4706 ENOMEM if hypervisor failed to allocate new radix page tables for guest 4707 ====== ================================================================ 4708 4709This ioctl is used to turn off the secure mode of the guest or transition 4710the guest from secure mode to normal mode. This is invoked when the guest 4711is reset. This has no effect if called for a normal guest. 4712 4713This ioctl issues an ultravisor call to terminate the secure guest, 4714unpins the VPA pages and releases all the device pages that are used to 4715track the secure pages by hypervisor. 4716 47174.122 KVM_S390_NORMAL_RESET 4718--------------------------- 4719 4720:Capability: KVM_CAP_S390_VCPU_RESETS 4721:Architectures: s390 4722:Type: vcpu ioctl 4723:Parameters: none 4724:Returns: 0 4725 4726This ioctl resets VCPU registers and control structures according to 4727the cpu reset definition in the POP (Principles Of Operation). 4728 47294.123 KVM_S390_INITIAL_RESET 4730---------------------------- 4731 4732:Capability: none 4733:Architectures: s390 4734:Type: vcpu ioctl 4735:Parameters: none 4736:Returns: 0 4737 4738This ioctl resets VCPU registers and control structures according to 4739the initial cpu reset definition in the POP. However, the cpu is not 4740put into ESA mode. This reset is a superset of the normal reset. 4741 47424.124 KVM_S390_CLEAR_RESET 4743-------------------------- 4744 4745:Capability: KVM_CAP_S390_VCPU_RESETS 4746:Architectures: s390 4747:Type: vcpu ioctl 4748:Parameters: none 4749:Returns: 0 4750 4751This ioctl resets VCPU registers and control structures according to 4752the clear cpu reset definition in the POP. However, the cpu is not put 4753into ESA mode. This reset is a superset of the initial reset. 4754 4755 47564.125 KVM_S390_PV_COMMAND 4757------------------------- 4758 4759:Capability: KVM_CAP_S390_PROTECTED 4760:Architectures: s390 4761:Type: vm ioctl 4762:Parameters: struct kvm_pv_cmd 4763:Returns: 0 on success, < 0 on error 4764 4765:: 4766 4767 struct kvm_pv_cmd { 4768 __u32 cmd; /* Command to be executed */ 4769 __u16 rc; /* Ultravisor return code */ 4770 __u16 rrc; /* Ultravisor return reason code */ 4771 __u64 data; /* Data or address */ 4772 __u32 flags; /* flags for future extensions. Must be 0 for now */ 4773 __u32 reserved[3]; 4774 }; 4775 4776cmd values: 4777 4778KVM_PV_ENABLE 4779 Allocate memory and register the VM with the Ultravisor, thereby 4780 donating memory to the Ultravisor that will become inaccessible to 4781 KVM. All existing CPUs are converted to protected ones. After this 4782 command has succeeded, any CPU added via hotplug will become 4783 protected during its creation as well. 4784 4785 Errors: 4786 4787 ===== ============================= 4788 EINTR an unmasked signal is pending 4789 ===== ============================= 4790 4791KVM_PV_DISABLE 4792 4793 Deregister the VM from the Ultravisor and reclaim the memory that 4794 had been donated to the Ultravisor, making it usable by the kernel 4795 again. All registered VCPUs are converted back to non-protected 4796 ones. 4797 4798KVM_PV_VM_SET_SEC_PARMS 4799 Pass the image header from VM memory to the Ultravisor in 4800 preparation of image unpacking and verification. 4801 4802KVM_PV_VM_UNPACK 4803 Unpack (protect and decrypt) a page of the encrypted boot image. 4804 4805KVM_PV_VM_VERIFY 4806 Verify the integrity of the unpacked image. Only if this succeeds, 4807 KVM is allowed to start protected VCPUs. 4808 48094.126 KVM_X86_SET_MSR_FILTER 4810---------------------------- 4811 4812:Capability: KVM_CAP_X86_MSR_FILTER 4813:Architectures: x86 4814:Type: vm ioctl 4815:Parameters: struct kvm_msr_filter 4816:Returns: 0 on success, < 0 on error 4817 4818:: 4819 4820 struct kvm_msr_filter_range { 4821 #define KVM_MSR_FILTER_READ (1 << 0) 4822 #define KVM_MSR_FILTER_WRITE (1 << 1) 4823 __u32 flags; 4824 __u32 nmsrs; /* number of msrs in bitmap */ 4825 __u32 base; /* MSR index the bitmap starts at */ 4826 __u8 *bitmap; /* a 1 bit allows the operations in flags, 0 denies */ 4827 }; 4828 4829 #define KVM_MSR_FILTER_MAX_RANGES 16 4830 struct kvm_msr_filter { 4831 #define KVM_MSR_FILTER_DEFAULT_ALLOW (0 << 0) 4832 #define KVM_MSR_FILTER_DEFAULT_DENY (1 << 0) 4833 __u32 flags; 4834 struct kvm_msr_filter_range ranges[KVM_MSR_FILTER_MAX_RANGES]; 4835 }; 4836 4837flags values for ``struct kvm_msr_filter_range``: 4838 4839``KVM_MSR_FILTER_READ`` 4840 4841 Filter read accesses to MSRs using the given bitmap. A 0 in the bitmap 4842 indicates that a read should immediately fail, while a 1 indicates that 4843 a read for a particular MSR should be handled regardless of the default 4844 filter action. 4845 4846``KVM_MSR_FILTER_WRITE`` 4847 4848 Filter write accesses to MSRs using the given bitmap. A 0 in the bitmap 4849 indicates that a write should immediately fail, while a 1 indicates that 4850 a write for a particular MSR should be handled regardless of the default 4851 filter action. 4852 4853``KVM_MSR_FILTER_READ | KVM_MSR_FILTER_WRITE`` 4854 4855 Filter both read and write accesses to MSRs using the given bitmap. A 0 4856 in the bitmap indicates that both reads and writes should immediately fail, 4857 while a 1 indicates that reads and writes for a particular MSR are not 4858 filtered by this range. 4859 4860flags values for ``struct kvm_msr_filter``: 4861 4862``KVM_MSR_FILTER_DEFAULT_ALLOW`` 4863 4864 If no filter range matches an MSR index that is getting accessed, KVM will 4865 fall back to allowing access to the MSR. 4866 4867``KVM_MSR_FILTER_DEFAULT_DENY`` 4868 4869 If no filter range matches an MSR index that is getting accessed, KVM will 4870 fall back to rejecting access to the MSR. In this mode, all MSRs that should 4871 be processed by KVM need to explicitly be marked as allowed in the bitmaps. 4872 4873This ioctl allows user space to define up to 16 bitmaps of MSR ranges to 4874specify whether a certain MSR access should be explicitly filtered for or not. 4875 4876If this ioctl has never been invoked, MSR accesses are not guarded and the 4877default KVM in-kernel emulation behavior is fully preserved. 4878 4879Calling this ioctl with an empty set of ranges (all nmsrs == 0) disables MSR 4880filtering. In that mode, ``KVM_MSR_FILTER_DEFAULT_DENY`` is invalid and causes 4881an error. 4882 4883As soon as the filtering is in place, every MSR access is processed through 4884the filtering except for accesses to the x2APIC MSRs (from 0x800 to 0x8ff); 4885x2APIC MSRs are always allowed, independent of the ``default_allow`` setting, 4886and their behavior depends on the ``X2APIC_ENABLE`` bit of the APIC base 4887register. 4888 4889If a bit is within one of the defined ranges, read and write accesses are 4890guarded by the bitmap's value for the MSR index if the kind of access 4891is included in the ``struct kvm_msr_filter_range`` flags. If no range 4892cover this particular access, the behavior is determined by the flags 4893field in the kvm_msr_filter struct: ``KVM_MSR_FILTER_DEFAULT_ALLOW`` 4894and ``KVM_MSR_FILTER_DEFAULT_DENY``. 4895 4896Each bitmap range specifies a range of MSRs to potentially allow access on. 4897The range goes from MSR index [base .. base+nmsrs]. The flags field 4898indicates whether reads, writes or both reads and writes are filtered 4899by setting a 1 bit in the bitmap for the corresponding MSR index. 4900 4901If an MSR access is not permitted through the filtering, it generates a 4902#GP inside the guest. When combined with KVM_CAP_X86_USER_SPACE_MSR, that 4903allows user space to deflect and potentially handle various MSR accesses 4904into user space. 4905 4906Note, invoking this ioctl with a vCPU is running is inherently racy. However, 4907KVM does guarantee that vCPUs will see either the previous filter or the new 4908filter, e.g. MSRs with identical settings in both the old and new filter will 4909have deterministic behavior. 4910 49114.127 KVM_XEN_HVM_SET_ATTR 4912-------------------------- 4913 4914:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 4915:Architectures: x86 4916:Type: vm ioctl 4917:Parameters: struct kvm_xen_hvm_attr 4918:Returns: 0 on success, < 0 on error 4919 4920:: 4921 4922 struct kvm_xen_hvm_attr { 4923 __u16 type; 4924 __u16 pad[3]; 4925 union { 4926 __u8 long_mode; 4927 __u8 vector; 4928 struct { 4929 __u64 gfn; 4930 } shared_info; 4931 __u64 pad[4]; 4932 } u; 4933 }; 4934 4935type values: 4936 4937KVM_XEN_ATTR_TYPE_LONG_MODE 4938 Sets the ABI mode of the VM to 32-bit or 64-bit (long mode). This 4939 determines the layout of the shared info pages exposed to the VM. 4940 4941KVM_XEN_ATTR_TYPE_SHARED_INFO 4942 Sets the guest physical frame number at which the Xen "shared info" 4943 page resides. Note that although Xen places vcpu_info for the first 4944 32 vCPUs in the shared_info page, KVM does not automatically do so 4945 and instead requires that KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO be used 4946 explicitly even when the vcpu_info for a given vCPU resides at the 4947 "default" location in the shared_info page. This is because KVM is 4948 not aware of the Xen CPU id which is used as the index into the 4949 vcpu_info[] array, so cannot know the correct default location. 4950 4951KVM_XEN_ATTR_TYPE_UPCALL_VECTOR 4952 Sets the exception vector used to deliver Xen event channel upcalls. 4953 49544.127 KVM_XEN_HVM_GET_ATTR 4955-------------------------- 4956 4957:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 4958:Architectures: x86 4959:Type: vm ioctl 4960:Parameters: struct kvm_xen_hvm_attr 4961:Returns: 0 on success, < 0 on error 4962 4963Allows Xen VM attributes to be read. For the structure and types, 4964see KVM_XEN_HVM_SET_ATTR above. 4965 49664.128 KVM_XEN_VCPU_SET_ATTR 4967--------------------------- 4968 4969:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 4970:Architectures: x86 4971:Type: vcpu ioctl 4972:Parameters: struct kvm_xen_vcpu_attr 4973:Returns: 0 on success, < 0 on error 4974 4975:: 4976 4977 struct kvm_xen_vcpu_attr { 4978 __u16 type; 4979 __u16 pad[3]; 4980 union { 4981 __u64 gpa; 4982 __u64 pad[4]; 4983 struct { 4984 __u64 state; 4985 __u64 state_entry_time; 4986 __u64 time_running; 4987 __u64 time_runnable; 4988 __u64 time_blocked; 4989 __u64 time_offline; 4990 } runstate; 4991 } u; 4992 }; 4993 4994type values: 4995 4996KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO 4997 Sets the guest physical address of the vcpu_info for a given vCPU. 4998 4999KVM_XEN_VCPU_ATTR_TYPE_VCPU_TIME_INFO 5000 Sets the guest physical address of an additional pvclock structure 5001 for a given vCPU. This is typically used for guest vsyscall support. 5002 5003KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR 5004 Sets the guest physical address of the vcpu_runstate_info for a given 5005 vCPU. This is how a Xen guest tracks CPU state such as steal time. 5006 5007KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_CURRENT 5008 Sets the runstate (RUNSTATE_running/_runnable/_blocked/_offline) of 5009 the given vCPU from the .u.runstate.state member of the structure. 5010 KVM automatically accounts running and runnable time but blocked 5011 and offline states are only entered explicitly. 5012 5013KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_DATA 5014 Sets all fields of the vCPU runstate data from the .u.runstate member 5015 of the structure, including the current runstate. The state_entry_time 5016 must equal the sum of the other four times. 5017 5018KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST 5019 This *adds* the contents of the .u.runstate members of the structure 5020 to the corresponding members of the given vCPU's runstate data, thus 5021 permitting atomic adjustments to the runstate times. The adjustment 5022 to the state_entry_time must equal the sum of the adjustments to the 5023 other four times. The state field must be set to -1, or to a valid 5024 runstate value (RUNSTATE_running, RUNSTATE_runnable, RUNSTATE_blocked 5025 or RUNSTATE_offline) to set the current accounted state as of the 5026 adjusted state_entry_time. 5027 50284.129 KVM_XEN_VCPU_GET_ATTR 5029--------------------------- 5030 5031:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 5032:Architectures: x86 5033:Type: vcpu ioctl 5034:Parameters: struct kvm_xen_vcpu_attr 5035:Returns: 0 on success, < 0 on error 5036 5037Allows Xen vCPU attributes to be read. For the structure and types, 5038see KVM_XEN_VCPU_SET_ATTR above. 5039 5040The KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST type may not be used 5041with the KVM_XEN_VCPU_GET_ATTR ioctl. 5042 50434.130 KVM_ARM_MTE_COPY_TAGS 5044--------------------------- 5045 5046:Capability: KVM_CAP_ARM_MTE 5047:Architectures: arm64 5048:Type: vm ioctl 5049:Parameters: struct kvm_arm_copy_mte_tags 5050:Returns: number of bytes copied, < 0 on error (-EINVAL for incorrect 5051 arguments, -EFAULT if memory cannot be accessed). 5052 5053:: 5054 5055 struct kvm_arm_copy_mte_tags { 5056 __u64 guest_ipa; 5057 __u64 length; 5058 void __user *addr; 5059 __u64 flags; 5060 __u64 reserved[2]; 5061 }; 5062 5063Copies Memory Tagging Extension (MTE) tags to/from guest tag memory. The 5064``guest_ipa`` and ``length`` fields must be ``PAGE_SIZE`` aligned. The ``addr`` 5065field must point to a buffer which the tags will be copied to or from. 5066 5067``flags`` specifies the direction of copy, either ``KVM_ARM_TAGS_TO_GUEST`` or 5068``KVM_ARM_TAGS_FROM_GUEST``. 5069 5070The size of the buffer to store the tags is ``(length / 16)`` bytes 5071(granules in MTE are 16 bytes long). Each byte contains a single tag 5072value. This matches the format of ``PTRACE_PEEKMTETAGS`` and 5073``PTRACE_POKEMTETAGS``. 5074 5075If an error occurs before any data is copied then a negative error code is 5076returned. If some tags have been copied before an error occurs then the number 5077of bytes successfully copied is returned. If the call completes successfully 5078then ``length`` is returned. 5079 50804.131 KVM_GET_SREGS2 5081-------------------- 5082 5083:Capability: KVM_CAP_SREGS2 5084:Architectures: x86 5085:Type: vcpu ioctl 5086:Parameters: struct kvm_sregs2 (out) 5087:Returns: 0 on success, -1 on error 5088 5089Reads special registers from the vcpu. 5090This ioctl (when supported) replaces the KVM_GET_SREGS. 5091 5092:: 5093 5094 struct kvm_sregs2 { 5095 /* out (KVM_GET_SREGS2) / in (KVM_SET_SREGS2) */ 5096 struct kvm_segment cs, ds, es, fs, gs, ss; 5097 struct kvm_segment tr, ldt; 5098 struct kvm_dtable gdt, idt; 5099 __u64 cr0, cr2, cr3, cr4, cr8; 5100 __u64 efer; 5101 __u64 apic_base; 5102 __u64 flags; 5103 __u64 pdptrs[4]; 5104 }; 5105 5106flags values for ``kvm_sregs2``: 5107 5108``KVM_SREGS2_FLAGS_PDPTRS_VALID`` 5109 5110 Indicates thats the struct contain valid PDPTR values. 5111 5112 51134.132 KVM_SET_SREGS2 5114-------------------- 5115 5116:Capability: KVM_CAP_SREGS2 5117:Architectures: x86 5118:Type: vcpu ioctl 5119:Parameters: struct kvm_sregs2 (in) 5120:Returns: 0 on success, -1 on error 5121 5122Writes special registers into the vcpu. 5123See KVM_GET_SREGS2 for the data structures. 5124This ioctl (when supported) replaces the KVM_SET_SREGS. 5125 51264.133 KVM_GET_STATS_FD 5127---------------------- 5128 5129:Capability: KVM_CAP_STATS_BINARY_FD 5130:Architectures: all 5131:Type: vm ioctl, vcpu ioctl 5132:Parameters: none 5133:Returns: statistics file descriptor on success, < 0 on error 5134 5135Errors: 5136 5137 ====== ====================================================== 5138 ENOMEM if the fd could not be created due to lack of memory 5139 EMFILE if the number of opened files exceeds the limit 5140 ====== ====================================================== 5141 5142The returned file descriptor can be used to read VM/vCPU statistics data in 5143binary format. The data in the file descriptor consists of four blocks 5144organized as follows: 5145 5146+-------------+ 5147| Header | 5148+-------------+ 5149| id string | 5150+-------------+ 5151| Descriptors | 5152+-------------+ 5153| Stats Data | 5154+-------------+ 5155 5156Apart from the header starting at offset 0, please be aware that it is 5157not guaranteed that the four blocks are adjacent or in the above order; 5158the offsets of the id, descriptors and data blocks are found in the 5159header. However, all four blocks are aligned to 64 bit offsets in the 5160file and they do not overlap. 5161 5162All blocks except the data block are immutable. Userspace can read them 5163only one time after retrieving the file descriptor, and then use ``pread`` or 5164``lseek`` to read the statistics repeatedly. 5165 5166All data is in system endianness. 5167 5168The format of the header is as follows:: 5169 5170 struct kvm_stats_header { 5171 __u32 flags; 5172 __u32 name_size; 5173 __u32 num_desc; 5174 __u32 id_offset; 5175 __u32 desc_offset; 5176 __u32 data_offset; 5177 }; 5178 5179The ``flags`` field is not used at the moment. It is always read as 0. 5180 5181The ``name_size`` field is the size (in byte) of the statistics name string 5182(including trailing '\0') which is contained in the "id string" block and 5183appended at the end of every descriptor. 5184 5185The ``num_desc`` field is the number of descriptors that are included in the 5186descriptor block. (The actual number of values in the data block may be 5187larger, since each descriptor may comprise more than one value). 5188 5189The ``id_offset`` field is the offset of the id string from the start of the 5190file indicated by the file descriptor. It is a multiple of 8. 5191 5192The ``desc_offset`` field is the offset of the Descriptors block from the start 5193of the file indicated by the file descriptor. It is a multiple of 8. 5194 5195The ``data_offset`` field is the offset of the Stats Data block from the start 5196of the file indicated by the file descriptor. It is a multiple of 8. 5197 5198The id string block contains a string which identifies the file descriptor on 5199which KVM_GET_STATS_FD was invoked. The size of the block, including the 5200trailing ``'\0'``, is indicated by the ``name_size`` field in the header. 5201 5202The descriptors block is only needed to be read once for the lifetime of the 5203file descriptor contains a sequence of ``struct kvm_stats_desc``, each followed 5204by a string of size ``name_size``. 5205:: 5206 5207 #define KVM_STATS_TYPE_SHIFT 0 5208 #define KVM_STATS_TYPE_MASK (0xF << KVM_STATS_TYPE_SHIFT) 5209 #define KVM_STATS_TYPE_CUMULATIVE (0x0 << KVM_STATS_TYPE_SHIFT) 5210 #define KVM_STATS_TYPE_INSTANT (0x1 << KVM_STATS_TYPE_SHIFT) 5211 #define KVM_STATS_TYPE_PEAK (0x2 << KVM_STATS_TYPE_SHIFT) 5212 #define KVM_STATS_TYPE_LINEAR_HIST (0x3 << KVM_STATS_TYPE_SHIFT) 5213 #define KVM_STATS_TYPE_LOG_HIST (0x4 << KVM_STATS_TYPE_SHIFT) 5214 #define KVM_STATS_TYPE_MAX KVM_STATS_TYPE_LOG_HIST 5215 5216 #define KVM_STATS_UNIT_SHIFT 4 5217 #define KVM_STATS_UNIT_MASK (0xF << KVM_STATS_UNIT_SHIFT) 5218 #define KVM_STATS_UNIT_NONE (0x0 << KVM_STATS_UNIT_SHIFT) 5219 #define KVM_STATS_UNIT_BYTES (0x1 << KVM_STATS_UNIT_SHIFT) 5220 #define KVM_STATS_UNIT_SECONDS (0x2 << KVM_STATS_UNIT_SHIFT) 5221 #define KVM_STATS_UNIT_CYCLES (0x3 << KVM_STATS_UNIT_SHIFT) 5222 #define KVM_STATS_UNIT_MAX KVM_STATS_UNIT_CYCLES 5223 5224 #define KVM_STATS_BASE_SHIFT 8 5225 #define KVM_STATS_BASE_MASK (0xF << KVM_STATS_BASE_SHIFT) 5226 #define KVM_STATS_BASE_POW10 (0x0 << KVM_STATS_BASE_SHIFT) 5227 #define KVM_STATS_BASE_POW2 (0x1 << KVM_STATS_BASE_SHIFT) 5228 #define KVM_STATS_BASE_MAX KVM_STATS_BASE_POW2 5229 5230 struct kvm_stats_desc { 5231 __u32 flags; 5232 __s16 exponent; 5233 __u16 size; 5234 __u32 offset; 5235 __u32 bucket_size; 5236 char name[]; 5237 }; 5238 5239The ``flags`` field contains the type and unit of the statistics data described 5240by this descriptor. Its endianness is CPU native. 5241The following flags are supported: 5242 5243Bits 0-3 of ``flags`` encode the type: 5244 5245 * ``KVM_STATS_TYPE_CUMULATIVE`` 5246 The statistics reports a cumulative count. The value of data can only be increased. 5247 Most of the counters used in KVM are of this type. 5248 The corresponding ``size`` field for this type is always 1. 5249 All cumulative statistics data are read/write. 5250 * ``KVM_STATS_TYPE_INSTANT`` 5251 The statistics reports an instantaneous value. Its value can be increased or 5252 decreased. This type is usually used as a measurement of some resources, 5253 like the number of dirty pages, the number of large pages, etc. 5254 All instant statistics are read only. 5255 The corresponding ``size`` field for this type is always 1. 5256 * ``KVM_STATS_TYPE_PEAK`` 5257 The statistics data reports a peak value, for example the maximum number 5258 of items in a hash table bucket, the longest time waited and so on. 5259 The value of data can only be increased. 5260 The corresponding ``size`` field for this type is always 1. 5261 * ``KVM_STATS_TYPE_LINEAR_HIST`` 5262 The statistic is reported as a linear histogram. The number of 5263 buckets is specified by the ``size`` field. The size of buckets is specified 5264 by the ``hist_param`` field. The range of the Nth bucket (1 <= N < ``size``) 5265 is [``hist_param``*(N-1), ``hist_param``*N), while the range of the last 5266 bucket is [``hist_param``*(``size``-1), +INF). (+INF means positive infinity 5267 value.) The bucket value indicates how many samples fell in the bucket's range. 5268 * ``KVM_STATS_TYPE_LOG_HIST`` 5269 The statistic is reported as a logarithmic histogram. The number of 5270 buckets is specified by the ``size`` field. The range of the first bucket is 5271 [0, 1), while the range of the last bucket is [pow(2, ``size``-2), +INF). 5272 Otherwise, The Nth bucket (1 < N < ``size``) covers 5273 [pow(2, N-2), pow(2, N-1)). The bucket value indicates how many samples fell 5274 in the bucket's range. 5275 5276Bits 4-7 of ``flags`` encode the unit: 5277 5278 * ``KVM_STATS_UNIT_NONE`` 5279 There is no unit for the value of statistics data. This usually means that 5280 the value is a simple counter of an event. 5281 * ``KVM_STATS_UNIT_BYTES`` 5282 It indicates that the statistics data is used to measure memory size, in the 5283 unit of Byte, KiByte, MiByte, GiByte, etc. The unit of the data is 5284 determined by the ``exponent`` field in the descriptor. 5285 * ``KVM_STATS_UNIT_SECONDS`` 5286 It indicates that the statistics data is used to measure time or latency. 5287 * ``KVM_STATS_UNIT_CYCLES`` 5288 It indicates that the statistics data is used to measure CPU clock cycles. 5289 5290Bits 8-11 of ``flags``, together with ``exponent``, encode the scale of the 5291unit: 5292 5293 * ``KVM_STATS_BASE_POW10`` 5294 The scale is based on power of 10. It is used for measurement of time and 5295 CPU clock cycles. For example, an exponent of -9 can be used with 5296 ``KVM_STATS_UNIT_SECONDS`` to express that the unit is nanoseconds. 5297 * ``KVM_STATS_BASE_POW2`` 5298 The scale is based on power of 2. It is used for measurement of memory size. 5299 For example, an exponent of 20 can be used with ``KVM_STATS_UNIT_BYTES`` to 5300 express that the unit is MiB. 5301 5302The ``size`` field is the number of values of this statistics data. Its 5303value is usually 1 for most of simple statistics. 1 means it contains an 5304unsigned 64bit data. 5305 5306The ``offset`` field is the offset from the start of Data Block to the start of 5307the corresponding statistics data. 5308 5309The ``bucket_size`` field is used as a parameter for histogram statistics data. 5310It is only used by linear histogram statistics data, specifying the size of a 5311bucket. 5312 5313The ``name`` field is the name string of the statistics data. The name string 5314starts at the end of ``struct kvm_stats_desc``. The maximum length including 5315the trailing ``'\0'``, is indicated by ``name_size`` in the header. 5316 5317The Stats Data block contains an array of 64-bit values in the same order 5318as the descriptors in Descriptors block. 5319 53205. The kvm_run structure 5321======================== 5322 5323Application code obtains a pointer to the kvm_run structure by 5324mmap()ing a vcpu fd. From that point, application code can control 5325execution by changing fields in kvm_run prior to calling the KVM_RUN 5326ioctl, and obtain information about the reason KVM_RUN returned by 5327looking up structure members. 5328 5329:: 5330 5331 struct kvm_run { 5332 /* in */ 5333 __u8 request_interrupt_window; 5334 5335Request that KVM_RUN return when it becomes possible to inject external 5336interrupts into the guest. Useful in conjunction with KVM_INTERRUPT. 5337 5338:: 5339 5340 __u8 immediate_exit; 5341 5342This field is polled once when KVM_RUN starts; if non-zero, KVM_RUN 5343exits immediately, returning -EINTR. In the common scenario where a 5344signal is used to "kick" a VCPU out of KVM_RUN, this field can be used 5345to avoid usage of KVM_SET_SIGNAL_MASK, which has worse scalability. 5346Rather than blocking the signal outside KVM_RUN, userspace can set up 5347a signal handler that sets run->immediate_exit to a non-zero value. 5348 5349This field is ignored if KVM_CAP_IMMEDIATE_EXIT is not available. 5350 5351:: 5352 5353 __u8 padding1[6]; 5354 5355 /* out */ 5356 __u32 exit_reason; 5357 5358When KVM_RUN has returned successfully (return value 0), this informs 5359application code why KVM_RUN has returned. Allowable values for this 5360field are detailed below. 5361 5362:: 5363 5364 __u8 ready_for_interrupt_injection; 5365 5366If request_interrupt_window has been specified, this field indicates 5367an interrupt can be injected now with KVM_INTERRUPT. 5368 5369:: 5370 5371 __u8 if_flag; 5372 5373The value of the current interrupt flag. Only valid if in-kernel 5374local APIC is not used. 5375 5376:: 5377 5378 __u16 flags; 5379 5380More architecture-specific flags detailing state of the VCPU that may 5381affect the device's behavior. Current defined flags:: 5382 5383 /* x86, set if the VCPU is in system management mode */ 5384 #define KVM_RUN_X86_SMM (1 << 0) 5385 /* x86, set if bus lock detected in VM */ 5386 #define KVM_RUN_BUS_LOCK (1 << 1) 5387 5388:: 5389 5390 /* in (pre_kvm_run), out (post_kvm_run) */ 5391 __u64 cr8; 5392 5393The value of the cr8 register. Only valid if in-kernel local APIC is 5394not used. Both input and output. 5395 5396:: 5397 5398 __u64 apic_base; 5399 5400The value of the APIC BASE msr. Only valid if in-kernel local 5401APIC is not used. Both input and output. 5402 5403:: 5404 5405 union { 5406 /* KVM_EXIT_UNKNOWN */ 5407 struct { 5408 __u64 hardware_exit_reason; 5409 } hw; 5410 5411If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown 5412reasons. Further architecture-specific information is available in 5413hardware_exit_reason. 5414 5415:: 5416 5417 /* KVM_EXIT_FAIL_ENTRY */ 5418 struct { 5419 __u64 hardware_entry_failure_reason; 5420 __u32 cpu; /* if KVM_LAST_CPU */ 5421 } fail_entry; 5422 5423If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due 5424to unknown reasons. Further architecture-specific information is 5425available in hardware_entry_failure_reason. 5426 5427:: 5428 5429 /* KVM_EXIT_EXCEPTION */ 5430 struct { 5431 __u32 exception; 5432 __u32 error_code; 5433 } ex; 5434 5435Unused. 5436 5437:: 5438 5439 /* KVM_EXIT_IO */ 5440 struct { 5441 #define KVM_EXIT_IO_IN 0 5442 #define KVM_EXIT_IO_OUT 1 5443 __u8 direction; 5444 __u8 size; /* bytes */ 5445 __u16 port; 5446 __u32 count; 5447 __u64 data_offset; /* relative to kvm_run start */ 5448 } io; 5449 5450If exit_reason is KVM_EXIT_IO, then the vcpu has 5451executed a port I/O instruction which could not be satisfied by kvm. 5452data_offset describes where the data is located (KVM_EXIT_IO_OUT) or 5453where kvm expects application code to place the data for the next 5454KVM_RUN invocation (KVM_EXIT_IO_IN). Data format is a packed array. 5455 5456:: 5457 5458 /* KVM_EXIT_DEBUG */ 5459 struct { 5460 struct kvm_debug_exit_arch arch; 5461 } debug; 5462 5463If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event 5464for which architecture specific information is returned. 5465 5466:: 5467 5468 /* KVM_EXIT_MMIO */ 5469 struct { 5470 __u64 phys_addr; 5471 __u8 data[8]; 5472 __u32 len; 5473 __u8 is_write; 5474 } mmio; 5475 5476If exit_reason is KVM_EXIT_MMIO, then the vcpu has 5477executed a memory-mapped I/O instruction which could not be satisfied 5478by kvm. The 'data' member contains the written data if 'is_write' is 5479true, and should be filled by application code otherwise. 5480 5481The 'data' member contains, in its first 'len' bytes, the value as it would 5482appear if the VCPU performed a load or store of the appropriate width directly 5483to the byte array. 5484 5485.. note:: 5486 5487 For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR, KVM_EXIT_XEN, 5488 KVM_EXIT_EPR, KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR the corresponding 5489 operations are complete (and guest state is consistent) only after userspace 5490 has re-entered the kernel with KVM_RUN. The kernel side will first finish 5491 incomplete operations and then check for pending signals. 5492 5493 The pending state of the operation is not preserved in state which is 5494 visible to userspace, thus userspace should ensure that the operation is 5495 completed before performing a live migration. Userspace can re-enter the 5496 guest with an unmasked signal pending or with the immediate_exit field set 5497 to complete pending operations without allowing any further instructions 5498 to be executed. 5499 5500:: 5501 5502 /* KVM_EXIT_HYPERCALL */ 5503 struct { 5504 __u64 nr; 5505 __u64 args[6]; 5506 __u64 ret; 5507 __u32 longmode; 5508 __u32 pad; 5509 } hypercall; 5510 5511Unused. This was once used for 'hypercall to userspace'. To implement 5512such functionality, use KVM_EXIT_IO (x86) or KVM_EXIT_MMIO (all except s390). 5513 5514.. note:: KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO. 5515 5516:: 5517 5518 /* KVM_EXIT_TPR_ACCESS */ 5519 struct { 5520 __u64 rip; 5521 __u32 is_write; 5522 __u32 pad; 5523 } tpr_access; 5524 5525To be documented (KVM_TPR_ACCESS_REPORTING). 5526 5527:: 5528 5529 /* KVM_EXIT_S390_SIEIC */ 5530 struct { 5531 __u8 icptcode; 5532 __u64 mask; /* psw upper half */ 5533 __u64 addr; /* psw lower half */ 5534 __u16 ipa; 5535 __u32 ipb; 5536 } s390_sieic; 5537 5538s390 specific. 5539 5540:: 5541 5542 /* KVM_EXIT_S390_RESET */ 5543 #define KVM_S390_RESET_POR 1 5544 #define KVM_S390_RESET_CLEAR 2 5545 #define KVM_S390_RESET_SUBSYSTEM 4 5546 #define KVM_S390_RESET_CPU_INIT 8 5547 #define KVM_S390_RESET_IPL 16 5548 __u64 s390_reset_flags; 5549 5550s390 specific. 5551 5552:: 5553 5554 /* KVM_EXIT_S390_UCONTROL */ 5555 struct { 5556 __u64 trans_exc_code; 5557 __u32 pgm_code; 5558 } s390_ucontrol; 5559 5560s390 specific. A page fault has occurred for a user controlled virtual 5561machine (KVM_VM_S390_UNCONTROL) on it's host page table that cannot be 5562resolved by the kernel. 5563The program code and the translation exception code that were placed 5564in the cpu's lowcore are presented here as defined by the z Architecture 5565Principles of Operation Book in the Chapter for Dynamic Address Translation 5566(DAT) 5567 5568:: 5569 5570 /* KVM_EXIT_DCR */ 5571 struct { 5572 __u32 dcrn; 5573 __u32 data; 5574 __u8 is_write; 5575 } dcr; 5576 5577Deprecated - was used for 440 KVM. 5578 5579:: 5580 5581 /* KVM_EXIT_OSI */ 5582 struct { 5583 __u64 gprs[32]; 5584 } osi; 5585 5586MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch 5587hypercalls and exit with this exit struct that contains all the guest gprs. 5588 5589If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall. 5590Userspace can now handle the hypercall and when it's done modify the gprs as 5591necessary. Upon guest entry all guest GPRs will then be replaced by the values 5592in this struct. 5593 5594:: 5595 5596 /* KVM_EXIT_PAPR_HCALL */ 5597 struct { 5598 __u64 nr; 5599 __u64 ret; 5600 __u64 args[9]; 5601 } papr_hcall; 5602 5603This is used on 64-bit PowerPC when emulating a pSeries partition, 5604e.g. with the 'pseries' machine type in qemu. It occurs when the 5605guest does a hypercall using the 'sc 1' instruction. The 'nr' field 5606contains the hypercall number (from the guest R3), and 'args' contains 5607the arguments (from the guest R4 - R12). Userspace should put the 5608return code in 'ret' and any extra returned values in args[]. 5609The possible hypercalls are defined in the Power Architecture Platform 5610Requirements (PAPR) document available from www.power.org (free 5611developer registration required to access it). 5612 5613:: 5614 5615 /* KVM_EXIT_S390_TSCH */ 5616 struct { 5617 __u16 subchannel_id; 5618 __u16 subchannel_nr; 5619 __u32 io_int_parm; 5620 __u32 io_int_word; 5621 __u32 ipb; 5622 __u8 dequeued; 5623 } s390_tsch; 5624 5625s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled 5626and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O 5627interrupt for the target subchannel has been dequeued and subchannel_id, 5628subchannel_nr, io_int_parm and io_int_word contain the parameters for that 5629interrupt. ipb is needed for instruction parameter decoding. 5630 5631:: 5632 5633 /* KVM_EXIT_EPR */ 5634 struct { 5635 __u32 epr; 5636 } epr; 5637 5638On FSL BookE PowerPC chips, the interrupt controller has a fast patch 5639interrupt acknowledge path to the core. When the core successfully 5640delivers an interrupt, it automatically populates the EPR register with 5641the interrupt vector number and acknowledges the interrupt inside 5642the interrupt controller. 5643 5644In case the interrupt controller lives in user space, we need to do 5645the interrupt acknowledge cycle through it to fetch the next to be 5646delivered interrupt vector using this exit. 5647 5648It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an 5649external interrupt has just been delivered into the guest. User space 5650should put the acknowledged interrupt vector into the 'epr' field. 5651 5652:: 5653 5654 /* KVM_EXIT_SYSTEM_EVENT */ 5655 struct { 5656 #define KVM_SYSTEM_EVENT_SHUTDOWN 1 5657 #define KVM_SYSTEM_EVENT_RESET 2 5658 #define KVM_SYSTEM_EVENT_CRASH 3 5659 __u32 type; 5660 __u64 flags; 5661 } system_event; 5662 5663If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered 5664a system-level event using some architecture specific mechanism (hypercall 5665or some special instruction). In case of ARM/ARM64, this is triggered using 5666HVC instruction based PSCI call from the vcpu. The 'type' field describes 5667the system-level event type. The 'flags' field describes architecture 5668specific flags for the system-level event. 5669 5670Valid values for 'type' are: 5671 5672 - KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the 5673 VM. Userspace is not obliged to honour this, and if it does honour 5674 this does not need to destroy the VM synchronously (ie it may call 5675 KVM_RUN again before shutdown finally occurs). 5676 - KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM. 5677 As with SHUTDOWN, userspace can choose to ignore the request, or 5678 to schedule the reset to occur in the future and may call KVM_RUN again. 5679 - KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest 5680 has requested a crash condition maintenance. Userspace can choose 5681 to ignore the request, or to gather VM memory core dump and/or 5682 reset/shutdown of the VM. 5683 5684:: 5685 5686 /* KVM_EXIT_IOAPIC_EOI */ 5687 struct { 5688 __u8 vector; 5689 } eoi; 5690 5691Indicates that the VCPU's in-kernel local APIC received an EOI for a 5692level-triggered IOAPIC interrupt. This exit only triggers when the 5693IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled); 5694the userspace IOAPIC should process the EOI and retrigger the interrupt if 5695it is still asserted. Vector is the LAPIC interrupt vector for which the 5696EOI was received. 5697 5698:: 5699 5700 struct kvm_hyperv_exit { 5701 #define KVM_EXIT_HYPERV_SYNIC 1 5702 #define KVM_EXIT_HYPERV_HCALL 2 5703 #define KVM_EXIT_HYPERV_SYNDBG 3 5704 __u32 type; 5705 __u32 pad1; 5706 union { 5707 struct { 5708 __u32 msr; 5709 __u32 pad2; 5710 __u64 control; 5711 __u64 evt_page; 5712 __u64 msg_page; 5713 } synic; 5714 struct { 5715 __u64 input; 5716 __u64 result; 5717 __u64 params[2]; 5718 } hcall; 5719 struct { 5720 __u32 msr; 5721 __u32 pad2; 5722 __u64 control; 5723 __u64 status; 5724 __u64 send_page; 5725 __u64 recv_page; 5726 __u64 pending_page; 5727 } syndbg; 5728 } u; 5729 }; 5730 /* KVM_EXIT_HYPERV */ 5731 struct kvm_hyperv_exit hyperv; 5732 5733Indicates that the VCPU exits into userspace to process some tasks 5734related to Hyper-V emulation. 5735 5736Valid values for 'type' are: 5737 5738 - KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about 5739 5740Hyper-V SynIC state change. Notification is used to remap SynIC 5741event/message pages and to enable/disable SynIC messages/events processing 5742in userspace. 5743 5744 - KVM_EXIT_HYPERV_SYNDBG -- synchronously notify user-space about 5745 5746Hyper-V Synthetic debugger state change. Notification is used to either update 5747the pending_page location or to send a control command (send the buffer located 5748in send_page or recv a buffer to recv_page). 5749 5750:: 5751 5752 /* KVM_EXIT_ARM_NISV */ 5753 struct { 5754 __u64 esr_iss; 5755 __u64 fault_ipa; 5756 } arm_nisv; 5757 5758Used on arm and arm64 systems. If a guest accesses memory not in a memslot, 5759KVM will typically return to userspace and ask it to do MMIO emulation on its 5760behalf. However, for certain classes of instructions, no instruction decode 5761(direction, length of memory access) is provided, and fetching and decoding 5762the instruction from the VM is overly complicated to live in the kernel. 5763 5764Historically, when this situation occurred, KVM would print a warning and kill 5765the VM. KVM assumed that if the guest accessed non-memslot memory, it was 5766trying to do I/O, which just couldn't be emulated, and the warning message was 5767phrased accordingly. However, what happened more often was that a guest bug 5768caused access outside the guest memory areas which should lead to a more 5769meaningful warning message and an external abort in the guest, if the access 5770did not fall within an I/O window. 5771 5772Userspace implementations can query for KVM_CAP_ARM_NISV_TO_USER, and enable 5773this capability at VM creation. Once this is done, these types of errors will 5774instead return to userspace with KVM_EXIT_ARM_NISV, with the valid bits from 5775the HSR (arm) and ESR_EL2 (arm64) in the esr_iss field, and the faulting IPA 5776in the fault_ipa field. Userspace can either fix up the access if it's 5777actually an I/O access by decoding the instruction from guest memory (if it's 5778very brave) and continue executing the guest, or it can decide to suspend, 5779dump, or restart the guest. 5780 5781Note that KVM does not skip the faulting instruction as it does for 5782KVM_EXIT_MMIO, but userspace has to emulate any change to the processing state 5783if it decides to decode and emulate the instruction. 5784 5785:: 5786 5787 /* KVM_EXIT_X86_RDMSR / KVM_EXIT_X86_WRMSR */ 5788 struct { 5789 __u8 error; /* user -> kernel */ 5790 __u8 pad[7]; 5791 __u32 reason; /* kernel -> user */ 5792 __u32 index; /* kernel -> user */ 5793 __u64 data; /* kernel <-> user */ 5794 } msr; 5795 5796Used on x86 systems. When the VM capability KVM_CAP_X86_USER_SPACE_MSR is 5797enabled, MSR accesses to registers that would invoke a #GP by KVM kernel code 5798will instead trigger a KVM_EXIT_X86_RDMSR exit for reads and KVM_EXIT_X86_WRMSR 5799exit for writes. 5800 5801The "reason" field specifies why the MSR trap occurred. User space will only 5802receive MSR exit traps when a particular reason was requested during through 5803ENABLE_CAP. Currently valid exit reasons are: 5804 5805 KVM_MSR_EXIT_REASON_UNKNOWN - access to MSR that is unknown to KVM 5806 KVM_MSR_EXIT_REASON_INVAL - access to invalid MSRs or reserved bits 5807 KVM_MSR_EXIT_REASON_FILTER - access blocked by KVM_X86_SET_MSR_FILTER 5808 5809For KVM_EXIT_X86_RDMSR, the "index" field tells user space which MSR the guest 5810wants to read. To respond to this request with a successful read, user space 5811writes the respective data into the "data" field and must continue guest 5812execution to ensure the read data is transferred into guest register state. 5813 5814If the RDMSR request was unsuccessful, user space indicates that with a "1" in 5815the "error" field. This will inject a #GP into the guest when the VCPU is 5816executed again. 5817 5818For KVM_EXIT_X86_WRMSR, the "index" field tells user space which MSR the guest 5819wants to write. Once finished processing the event, user space must continue 5820vCPU execution. If the MSR write was unsuccessful, user space also sets the 5821"error" field to "1". 5822 5823:: 5824 5825 5826 struct kvm_xen_exit { 5827 #define KVM_EXIT_XEN_HCALL 1 5828 __u32 type; 5829 union { 5830 struct { 5831 __u32 longmode; 5832 __u32 cpl; 5833 __u64 input; 5834 __u64 result; 5835 __u64 params[6]; 5836 } hcall; 5837 } u; 5838 }; 5839 /* KVM_EXIT_XEN */ 5840 struct kvm_hyperv_exit xen; 5841 5842Indicates that the VCPU exits into userspace to process some tasks 5843related to Xen emulation. 5844 5845Valid values for 'type' are: 5846 5847 - KVM_EXIT_XEN_HCALL -- synchronously notify user-space about Xen hypercall. 5848 Userspace is expected to place the hypercall result into the appropriate 5849 field before invoking KVM_RUN again. 5850 5851:: 5852 5853 /* Fix the size of the union. */ 5854 char padding[256]; 5855 }; 5856 5857 /* 5858 * shared registers between kvm and userspace. 5859 * kvm_valid_regs specifies the register classes set by the host 5860 * kvm_dirty_regs specified the register classes dirtied by userspace 5861 * struct kvm_sync_regs is architecture specific, as well as the 5862 * bits for kvm_valid_regs and kvm_dirty_regs 5863 */ 5864 __u64 kvm_valid_regs; 5865 __u64 kvm_dirty_regs; 5866 union { 5867 struct kvm_sync_regs regs; 5868 char padding[SYNC_REGS_SIZE_BYTES]; 5869 } s; 5870 5871If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access 5872certain guest registers without having to call SET/GET_*REGS. Thus we can 5873avoid some system call overhead if userspace has to handle the exit. 5874Userspace can query the validity of the structure by checking 5875kvm_valid_regs for specific bits. These bits are architecture specific 5876and usually define the validity of a groups of registers. (e.g. one bit 5877for general purpose registers) 5878 5879Please note that the kernel is allowed to use the kvm_run structure as the 5880primary storage for certain register types. Therefore, the kernel may use the 5881values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set. 5882 5883:: 5884 5885 }; 5886 5887 5888 58896. Capabilities that can be enabled on vCPUs 5890============================================ 5891 5892There are certain capabilities that change the behavior of the virtual CPU or 5893the virtual machine when enabled. To enable them, please see section 4.37. 5894Below you can find a list of capabilities and what their effect on the vCPU or 5895the virtual machine is when enabling them. 5896 5897The following information is provided along with the description: 5898 5899 Architectures: 5900 which instruction set architectures provide this ioctl. 5901 x86 includes both i386 and x86_64. 5902 5903 Target: 5904 whether this is a per-vcpu or per-vm capability. 5905 5906 Parameters: 5907 what parameters are accepted by the capability. 5908 5909 Returns: 5910 the return value. General error numbers (EBADF, ENOMEM, EINVAL) 5911 are not detailed, but errors with specific meanings are. 5912 5913 59146.1 KVM_CAP_PPC_OSI 5915------------------- 5916 5917:Architectures: ppc 5918:Target: vcpu 5919:Parameters: none 5920:Returns: 0 on success; -1 on error 5921 5922This capability enables interception of OSI hypercalls that otherwise would 5923be treated as normal system calls to be injected into the guest. OSI hypercalls 5924were invented by Mac-on-Linux to have a standardized communication mechanism 5925between the guest and the host. 5926 5927When this capability is enabled, KVM_EXIT_OSI can occur. 5928 5929 59306.2 KVM_CAP_PPC_PAPR 5931-------------------- 5932 5933:Architectures: ppc 5934:Target: vcpu 5935:Parameters: none 5936:Returns: 0 on success; -1 on error 5937 5938This capability enables interception of PAPR hypercalls. PAPR hypercalls are 5939done using the hypercall instruction "sc 1". 5940 5941It also sets the guest privilege level to "supervisor" mode. Usually the guest 5942runs in "hypervisor" privilege mode with a few missing features. 5943 5944In addition to the above, it changes the semantics of SDR1. In this mode, the 5945HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the 5946HTAB invisible to the guest. 5947 5948When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur. 5949 5950 59516.3 KVM_CAP_SW_TLB 5952------------------ 5953 5954:Architectures: ppc 5955:Target: vcpu 5956:Parameters: args[0] is the address of a struct kvm_config_tlb 5957:Returns: 0 on success; -1 on error 5958 5959:: 5960 5961 struct kvm_config_tlb { 5962 __u64 params; 5963 __u64 array; 5964 __u32 mmu_type; 5965 __u32 array_len; 5966 }; 5967 5968Configures the virtual CPU's TLB array, establishing a shared memory area 5969between userspace and KVM. The "params" and "array" fields are userspace 5970addresses of mmu-type-specific data structures. The "array_len" field is an 5971safety mechanism, and should be set to the size in bytes of the memory that 5972userspace has reserved for the array. It must be at least the size dictated 5973by "mmu_type" and "params". 5974 5975While KVM_RUN is active, the shared region is under control of KVM. Its 5976contents are undefined, and any modification by userspace results in 5977boundedly undefined behavior. 5978 5979On return from KVM_RUN, the shared region will reflect the current state of 5980the guest's TLB. If userspace makes any changes, it must call KVM_DIRTY_TLB 5981to tell KVM which entries have been changed, prior to calling KVM_RUN again 5982on this vcpu. 5983 5984For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV: 5985 5986 - The "params" field is of type "struct kvm_book3e_206_tlb_params". 5987 - The "array" field points to an array of type "struct 5988 kvm_book3e_206_tlb_entry". 5989 - The array consists of all entries in the first TLB, followed by all 5990 entries in the second TLB. 5991 - Within a TLB, entries are ordered first by increasing set number. Within a 5992 set, entries are ordered by way (increasing ESEL). 5993 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1) 5994 where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value. 5995 - The tsize field of mas1 shall be set to 4K on TLB0, even though the 5996 hardware ignores this value for TLB0. 5997 59986.4 KVM_CAP_S390_CSS_SUPPORT 5999---------------------------- 6000 6001:Architectures: s390 6002:Target: vcpu 6003:Parameters: none 6004:Returns: 0 on success; -1 on error 6005 6006This capability enables support for handling of channel I/O instructions. 6007 6008TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are 6009handled in-kernel, while the other I/O instructions are passed to userspace. 6010 6011When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST 6012SUBCHANNEL intercepts. 6013 6014Note that even though this capability is enabled per-vcpu, the complete 6015virtual machine is affected. 6016 60176.5 KVM_CAP_PPC_EPR 6018------------------- 6019 6020:Architectures: ppc 6021:Target: vcpu 6022:Parameters: args[0] defines whether the proxy facility is active 6023:Returns: 0 on success; -1 on error 6024 6025This capability enables or disables the delivery of interrupts through the 6026external proxy facility. 6027 6028When enabled (args[0] != 0), every time the guest gets an external interrupt 6029delivered, it automatically exits into user space with a KVM_EXIT_EPR exit 6030to receive the topmost interrupt vector. 6031 6032When disabled (args[0] == 0), behavior is as if this facility is unsupported. 6033 6034When this capability is enabled, KVM_EXIT_EPR can occur. 6035 60366.6 KVM_CAP_IRQ_MPIC 6037-------------------- 6038 6039:Architectures: ppc 6040:Parameters: args[0] is the MPIC device fd; 6041 args[1] is the MPIC CPU number for this vcpu 6042 6043This capability connects the vcpu to an in-kernel MPIC device. 6044 60456.7 KVM_CAP_IRQ_XICS 6046-------------------- 6047 6048:Architectures: ppc 6049:Target: vcpu 6050:Parameters: args[0] is the XICS device fd; 6051 args[1] is the XICS CPU number (server ID) for this vcpu 6052 6053This capability connects the vcpu to an in-kernel XICS device. 6054 60556.8 KVM_CAP_S390_IRQCHIP 6056------------------------ 6057 6058:Architectures: s390 6059:Target: vm 6060:Parameters: none 6061 6062This capability enables the in-kernel irqchip for s390. Please refer to 6063"4.24 KVM_CREATE_IRQCHIP" for details. 6064 60656.9 KVM_CAP_MIPS_FPU 6066-------------------- 6067 6068:Architectures: mips 6069:Target: vcpu 6070:Parameters: args[0] is reserved for future use (should be 0). 6071 6072This capability allows the use of the host Floating Point Unit by the guest. It 6073allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is 6074done the ``KVM_REG_MIPS_FPR_*`` and ``KVM_REG_MIPS_FCR_*`` registers can be 6075accessed (depending on the current guest FPU register mode), and the Status.FR, 6076Config5.FRE bits are accessible via the KVM API and also from the guest, 6077depending on them being supported by the FPU. 6078 60796.10 KVM_CAP_MIPS_MSA 6080--------------------- 6081 6082:Architectures: mips 6083:Target: vcpu 6084:Parameters: args[0] is reserved for future use (should be 0). 6085 6086This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest. 6087It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest. 6088Once this is done the ``KVM_REG_MIPS_VEC_*`` and ``KVM_REG_MIPS_MSA_*`` 6089registers can be accessed, and the Config5.MSAEn bit is accessible via the 6090KVM API and also from the guest. 6091 60926.74 KVM_CAP_SYNC_REGS 6093---------------------- 6094 6095:Architectures: s390, x86 6096:Target: s390: always enabled, x86: vcpu 6097:Parameters: none 6098:Returns: x86: KVM_CHECK_EXTENSION returns a bit-array indicating which register 6099 sets are supported 6100 (bitfields defined in arch/x86/include/uapi/asm/kvm.h). 6101 6102As described above in the kvm_sync_regs struct info in section 5 (kvm_run): 6103KVM_CAP_SYNC_REGS "allow[s] userspace to access certain guest registers 6104without having to call SET/GET_*REGS". This reduces overhead by eliminating 6105repeated ioctl calls for setting and/or getting register values. This is 6106particularly important when userspace is making synchronous guest state 6107modifications, e.g. when emulating and/or intercepting instructions in 6108userspace. 6109 6110For s390 specifics, please refer to the source code. 6111 6112For x86: 6113 6114- the register sets to be copied out to kvm_run are selectable 6115 by userspace (rather that all sets being copied out for every exit). 6116- vcpu_events are available in addition to regs and sregs. 6117 6118For x86, the 'kvm_valid_regs' field of struct kvm_run is overloaded to 6119function as an input bit-array field set by userspace to indicate the 6120specific register sets to be copied out on the next exit. 6121 6122To indicate when userspace has modified values that should be copied into 6123the vCPU, the all architecture bitarray field, 'kvm_dirty_regs' must be set. 6124This is done using the same bitflags as for the 'kvm_valid_regs' field. 6125If the dirty bit is not set, then the register set values will not be copied 6126into the vCPU even if they've been modified. 6127 6128Unused bitfields in the bitarrays must be set to zero. 6129 6130:: 6131 6132 struct kvm_sync_regs { 6133 struct kvm_regs regs; 6134 struct kvm_sregs sregs; 6135 struct kvm_vcpu_events events; 6136 }; 6137 61386.75 KVM_CAP_PPC_IRQ_XIVE 6139------------------------- 6140 6141:Architectures: ppc 6142:Target: vcpu 6143:Parameters: args[0] is the XIVE device fd; 6144 args[1] is the XIVE CPU number (server ID) for this vcpu 6145 6146This capability connects the vcpu to an in-kernel XIVE device. 6147 61487. Capabilities that can be enabled on VMs 6149========================================== 6150 6151There are certain capabilities that change the behavior of the virtual 6152machine when enabled. To enable them, please see section 4.37. Below 6153you can find a list of capabilities and what their effect on the VM 6154is when enabling them. 6155 6156The following information is provided along with the description: 6157 6158 Architectures: 6159 which instruction set architectures provide this ioctl. 6160 x86 includes both i386 and x86_64. 6161 6162 Parameters: 6163 what parameters are accepted by the capability. 6164 6165 Returns: 6166 the return value. General error numbers (EBADF, ENOMEM, EINVAL) 6167 are not detailed, but errors with specific meanings are. 6168 6169 61707.1 KVM_CAP_PPC_ENABLE_HCALL 6171---------------------------- 6172 6173:Architectures: ppc 6174:Parameters: args[0] is the sPAPR hcall number; 6175 args[1] is 0 to disable, 1 to enable in-kernel handling 6176 6177This capability controls whether individual sPAPR hypercalls (hcalls) 6178get handled by the kernel or not. Enabling or disabling in-kernel 6179handling of an hcall is effective across the VM. On creation, an 6180initial set of hcalls are enabled for in-kernel handling, which 6181consists of those hcalls for which in-kernel handlers were implemented 6182before this capability was implemented. If disabled, the kernel will 6183not to attempt to handle the hcall, but will always exit to userspace 6184to handle it. Note that it may not make sense to enable some and 6185disable others of a group of related hcalls, but KVM does not prevent 6186userspace from doing that. 6187 6188If the hcall number specified is not one that has an in-kernel 6189implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL 6190error. 6191 61927.2 KVM_CAP_S390_USER_SIGP 6193-------------------------- 6194 6195:Architectures: s390 6196:Parameters: none 6197 6198This capability controls which SIGP orders will be handled completely in user 6199space. With this capability enabled, all fast orders will be handled completely 6200in the kernel: 6201 6202- SENSE 6203- SENSE RUNNING 6204- EXTERNAL CALL 6205- EMERGENCY SIGNAL 6206- CONDITIONAL EMERGENCY SIGNAL 6207 6208All other orders will be handled completely in user space. 6209 6210Only privileged operation exceptions will be checked for in the kernel (or even 6211in the hardware prior to interception). If this capability is not enabled, the 6212old way of handling SIGP orders is used (partially in kernel and user space). 6213 62147.3 KVM_CAP_S390_VECTOR_REGISTERS 6215--------------------------------- 6216 6217:Architectures: s390 6218:Parameters: none 6219:Returns: 0 on success, negative value on error 6220 6221Allows use of the vector registers introduced with z13 processor, and 6222provides for the synchronization between host and user space. Will 6223return -EINVAL if the machine does not support vectors. 6224 62257.4 KVM_CAP_S390_USER_STSI 6226-------------------------- 6227 6228:Architectures: s390 6229:Parameters: none 6230 6231This capability allows post-handlers for the STSI instruction. After 6232initial handling in the kernel, KVM exits to user space with 6233KVM_EXIT_S390_STSI to allow user space to insert further data. 6234 6235Before exiting to userspace, kvm handlers should fill in s390_stsi field of 6236vcpu->run:: 6237 6238 struct { 6239 __u64 addr; 6240 __u8 ar; 6241 __u8 reserved; 6242 __u8 fc; 6243 __u8 sel1; 6244 __u16 sel2; 6245 } s390_stsi; 6246 6247 @addr - guest address of STSI SYSIB 6248 @fc - function code 6249 @sel1 - selector 1 6250 @sel2 - selector 2 6251 @ar - access register number 6252 6253KVM handlers should exit to userspace with rc = -EREMOTE. 6254 62557.5 KVM_CAP_SPLIT_IRQCHIP 6256------------------------- 6257 6258:Architectures: x86 6259:Parameters: args[0] - number of routes reserved for userspace IOAPICs 6260:Returns: 0 on success, -1 on error 6261 6262Create a local apic for each processor in the kernel. This can be used 6263instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the 6264IOAPIC and PIC (and also the PIT, even though this has to be enabled 6265separately). 6266 6267This capability also enables in kernel routing of interrupt requests; 6268when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are 6269used in the IRQ routing table. The first args[0] MSI routes are reserved 6270for the IOAPIC pins. Whenever the LAPIC receives an EOI for these routes, 6271a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace. 6272 6273Fails if VCPU has already been created, or if the irqchip is already in the 6274kernel (i.e. KVM_CREATE_IRQCHIP has already been called). 6275 62767.6 KVM_CAP_S390_RI 6277------------------- 6278 6279:Architectures: s390 6280:Parameters: none 6281 6282Allows use of runtime-instrumentation introduced with zEC12 processor. 6283Will return -EINVAL if the machine does not support runtime-instrumentation. 6284Will return -EBUSY if a VCPU has already been created. 6285 62867.7 KVM_CAP_X2APIC_API 6287---------------------- 6288 6289:Architectures: x86 6290:Parameters: args[0] - features that should be enabled 6291:Returns: 0 on success, -EINVAL when args[0] contains invalid features 6292 6293Valid feature flags in args[0] are:: 6294 6295 #define KVM_X2APIC_API_USE_32BIT_IDS (1ULL << 0) 6296 #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK (1ULL << 1) 6297 6298Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of 6299KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC, 6300allowing the use of 32-bit APIC IDs. See KVM_CAP_X2APIC_API in their 6301respective sections. 6302 6303KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work 6304in logical mode or with more than 255 VCPUs. Otherwise, KVM treats 0xff 6305as a broadcast even in x2APIC mode in order to support physical x2APIC 6306without interrupt remapping. This is undesirable in logical mode, 6307where 0xff represents CPUs 0-7 in cluster 0. 6308 63097.8 KVM_CAP_S390_USER_INSTR0 6310---------------------------- 6311 6312:Architectures: s390 6313:Parameters: none 6314 6315With this capability enabled, all illegal instructions 0x0000 (2 bytes) will 6316be intercepted and forwarded to user space. User space can use this 6317mechanism e.g. to realize 2-byte software breakpoints. The kernel will 6318not inject an operating exception for these instructions, user space has 6319to take care of that. 6320 6321This capability can be enabled dynamically even if VCPUs were already 6322created and are running. 6323 63247.9 KVM_CAP_S390_GS 6325------------------- 6326 6327:Architectures: s390 6328:Parameters: none 6329:Returns: 0 on success; -EINVAL if the machine does not support 6330 guarded storage; -EBUSY if a VCPU has already been created. 6331 6332Allows use of guarded storage for the KVM guest. 6333 63347.10 KVM_CAP_S390_AIS 6335--------------------- 6336 6337:Architectures: s390 6338:Parameters: none 6339 6340Allow use of adapter-interruption suppression. 6341:Returns: 0 on success; -EBUSY if a VCPU has already been created. 6342 63437.11 KVM_CAP_PPC_SMT 6344-------------------- 6345 6346:Architectures: ppc 6347:Parameters: vsmt_mode, flags 6348 6349Enabling this capability on a VM provides userspace with a way to set 6350the desired virtual SMT mode (i.e. the number of virtual CPUs per 6351virtual core). The virtual SMT mode, vsmt_mode, must be a power of 2 6352between 1 and 8. On POWER8, vsmt_mode must also be no greater than 6353the number of threads per subcore for the host. Currently flags must 6354be 0. A successful call to enable this capability will result in 6355vsmt_mode being returned when the KVM_CAP_PPC_SMT capability is 6356subsequently queried for the VM. This capability is only supported by 6357HV KVM, and can only be set before any VCPUs have been created. 6358The KVM_CAP_PPC_SMT_POSSIBLE capability indicates which virtual SMT 6359modes are available. 6360 63617.12 KVM_CAP_PPC_FWNMI 6362---------------------- 6363 6364:Architectures: ppc 6365:Parameters: none 6366 6367With this capability a machine check exception in the guest address 6368space will cause KVM to exit the guest with NMI exit reason. This 6369enables QEMU to build error log and branch to guest kernel registered 6370machine check handling routine. Without this capability KVM will 6371branch to guests' 0x200 interrupt vector. 6372 63737.13 KVM_CAP_X86_DISABLE_EXITS 6374------------------------------ 6375 6376:Architectures: x86 6377:Parameters: args[0] defines which exits are disabled 6378:Returns: 0 on success, -EINVAL when args[0] contains invalid exits 6379 6380Valid bits in args[0] are:: 6381 6382 #define KVM_X86_DISABLE_EXITS_MWAIT (1 << 0) 6383 #define KVM_X86_DISABLE_EXITS_HLT (1 << 1) 6384 #define KVM_X86_DISABLE_EXITS_PAUSE (1 << 2) 6385 #define KVM_X86_DISABLE_EXITS_CSTATE (1 << 3) 6386 6387Enabling this capability on a VM provides userspace with a way to no 6388longer intercept some instructions for improved latency in some 6389workloads, and is suggested when vCPUs are associated to dedicated 6390physical CPUs. More bits can be added in the future; userspace can 6391just pass the KVM_CHECK_EXTENSION result to KVM_ENABLE_CAP to disable 6392all such vmexits. 6393 6394Do not enable KVM_FEATURE_PV_UNHALT if you disable HLT exits. 6395 63967.14 KVM_CAP_S390_HPAGE_1M 6397-------------------------- 6398 6399:Architectures: s390 6400:Parameters: none 6401:Returns: 0 on success, -EINVAL if hpage module parameter was not set 6402 or cmma is enabled, or the VM has the KVM_VM_S390_UCONTROL 6403 flag set 6404 6405With this capability the KVM support for memory backing with 1m pages 6406through hugetlbfs can be enabled for a VM. After the capability is 6407enabled, cmma can't be enabled anymore and pfmfi and the storage key 6408interpretation are disabled. If cmma has already been enabled or the 6409hpage module parameter is not set to 1, -EINVAL is returned. 6410 6411While it is generally possible to create a huge page backed VM without 6412this capability, the VM will not be able to run. 6413 64147.15 KVM_CAP_MSR_PLATFORM_INFO 6415------------------------------ 6416 6417:Architectures: x86 6418:Parameters: args[0] whether feature should be enabled or not 6419 6420With this capability, a guest may read the MSR_PLATFORM_INFO MSR. Otherwise, 6421a #GP would be raised when the guest tries to access. Currently, this 6422capability does not enable write permissions of this MSR for the guest. 6423 64247.16 KVM_CAP_PPC_NESTED_HV 6425-------------------------- 6426 6427:Architectures: ppc 6428:Parameters: none 6429:Returns: 0 on success, -EINVAL when the implementation doesn't support 6430 nested-HV virtualization. 6431 6432HV-KVM on POWER9 and later systems allows for "nested-HV" 6433virtualization, which provides a way for a guest VM to run guests that 6434can run using the CPU's supervisor mode (privileged non-hypervisor 6435state). Enabling this capability on a VM depends on the CPU having 6436the necessary functionality and on the facility being enabled with a 6437kvm-hv module parameter. 6438 64397.17 KVM_CAP_EXCEPTION_PAYLOAD 6440------------------------------ 6441 6442:Architectures: x86 6443:Parameters: args[0] whether feature should be enabled or not 6444 6445With this capability enabled, CR2 will not be modified prior to the 6446emulated VM-exit when L1 intercepts a #PF exception that occurs in 6447L2. Similarly, for kvm-intel only, DR6 will not be modified prior to 6448the emulated VM-exit when L1 intercepts a #DB exception that occurs in 6449L2. As a result, when KVM_GET_VCPU_EVENTS reports a pending #PF (or 6450#DB) exception for L2, exception.has_payload will be set and the 6451faulting address (or the new DR6 bits*) will be reported in the 6452exception_payload field. Similarly, when userspace injects a #PF (or 6453#DB) into L2 using KVM_SET_VCPU_EVENTS, it is expected to set 6454exception.has_payload and to put the faulting address - or the new DR6 6455bits\ [#]_ - in the exception_payload field. 6456 6457This capability also enables exception.pending in struct 6458kvm_vcpu_events, which allows userspace to distinguish between pending 6459and injected exceptions. 6460 6461 6462.. [#] For the new DR6 bits, note that bit 16 is set iff the #DB exception 6463 will clear DR6.RTM. 6464 64657.18 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 6466 6467:Architectures: x86, arm, arm64, mips 6468:Parameters: args[0] whether feature should be enabled or not 6469 6470Valid flags are:: 6471 6472 #define KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (1 << 0) 6473 #define KVM_DIRTY_LOG_INITIALLY_SET (1 << 1) 6474 6475With KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE is set, KVM_GET_DIRTY_LOG will not 6476automatically clear and write-protect all pages that are returned as dirty. 6477Rather, userspace will have to do this operation separately using 6478KVM_CLEAR_DIRTY_LOG. 6479 6480At the cost of a slightly more complicated operation, this provides better 6481scalability and responsiveness for two reasons. First, 6482KVM_CLEAR_DIRTY_LOG ioctl can operate on a 64-page granularity rather 6483than requiring to sync a full memslot; this ensures that KVM does not 6484take spinlocks for an extended period of time. Second, in some cases a 6485large amount of time can pass between a call to KVM_GET_DIRTY_LOG and 6486userspace actually using the data in the page. Pages can be modified 6487during this time, which is inefficient for both the guest and userspace: 6488the guest will incur a higher penalty due to write protection faults, 6489while userspace can see false reports of dirty pages. Manual reprotection 6490helps reducing this time, improving guest performance and reducing the 6491number of dirty log false positives. 6492 6493With KVM_DIRTY_LOG_INITIALLY_SET set, all the bits of the dirty bitmap 6494will be initialized to 1 when created. This also improves performance because 6495dirty logging can be enabled gradually in small chunks on the first call 6496to KVM_CLEAR_DIRTY_LOG. KVM_DIRTY_LOG_INITIALLY_SET depends on 6497KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (it is also only available on 6498x86 and arm64 for now). 6499 6500KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 was previously available under the name 6501KVM_CAP_MANUAL_DIRTY_LOG_PROTECT, but the implementation had bugs that make 6502it hard or impossible to use it correctly. The availability of 6503KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 signals that those bugs are fixed. 6504Userspace should not try to use KVM_CAP_MANUAL_DIRTY_LOG_PROTECT. 6505 65067.19 KVM_CAP_PPC_SECURE_GUEST 6507------------------------------ 6508 6509:Architectures: ppc 6510 6511This capability indicates that KVM is running on a host that has 6512ultravisor firmware and thus can support a secure guest. On such a 6513system, a guest can ask the ultravisor to make it a secure guest, 6514one whose memory is inaccessible to the host except for pages which 6515are explicitly requested to be shared with the host. The ultravisor 6516notifies KVM when a guest requests to become a secure guest, and KVM 6517has the opportunity to veto the transition. 6518 6519If present, this capability can be enabled for a VM, meaning that KVM 6520will allow the transition to secure guest mode. Otherwise KVM will 6521veto the transition. 6522 65237.20 KVM_CAP_HALT_POLL 6524---------------------- 6525 6526:Architectures: all 6527:Target: VM 6528:Parameters: args[0] is the maximum poll time in nanoseconds 6529:Returns: 0 on success; -1 on error 6530 6531This capability overrides the kvm module parameter halt_poll_ns for the 6532target VM. 6533 6534VCPU polling allows a VCPU to poll for wakeup events instead of immediately 6535scheduling during guest halts. The maximum time a VCPU can spend polling is 6536controlled by the kvm module parameter halt_poll_ns. This capability allows 6537the maximum halt time to specified on a per-VM basis, effectively overriding 6538the module parameter for the target VM. 6539 65407.21 KVM_CAP_X86_USER_SPACE_MSR 6541------------------------------- 6542 6543:Architectures: x86 6544:Target: VM 6545:Parameters: args[0] contains the mask of KVM_MSR_EXIT_REASON_* events to report 6546:Returns: 0 on success; -1 on error 6547 6548This capability enables trapping of #GP invoking RDMSR and WRMSR instructions 6549into user space. 6550 6551When a guest requests to read or write an MSR, KVM may not implement all MSRs 6552that are relevant to a respective system. It also does not differentiate by 6553CPU type. 6554 6555To allow more fine grained control over MSR handling, user space may enable 6556this capability. With it enabled, MSR accesses that match the mask specified in 6557args[0] and trigger a #GP event inside the guest by KVM will instead trigger 6558KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR exit notifications which user space 6559can then handle to implement model specific MSR handling and/or user notifications 6560to inform a user that an MSR was not handled. 6561 65627.22 KVM_CAP_X86_BUS_LOCK_EXIT 6563------------------------------- 6564 6565:Architectures: x86 6566:Target: VM 6567:Parameters: args[0] defines the policy used when bus locks detected in guest 6568:Returns: 0 on success, -EINVAL when args[0] contains invalid bits 6569 6570Valid bits in args[0] are:: 6571 6572 #define KVM_BUS_LOCK_DETECTION_OFF (1 << 0) 6573 #define KVM_BUS_LOCK_DETECTION_EXIT (1 << 1) 6574 6575Enabling this capability on a VM provides userspace with a way to select 6576a policy to handle the bus locks detected in guest. Userspace can obtain 6577the supported modes from the result of KVM_CHECK_EXTENSION and define it 6578through the KVM_ENABLE_CAP. 6579 6580KVM_BUS_LOCK_DETECTION_OFF and KVM_BUS_LOCK_DETECTION_EXIT are supported 6581currently and mutually exclusive with each other. More bits can be added in 6582the future. 6583 6584With KVM_BUS_LOCK_DETECTION_OFF set, bus locks in guest will not cause vm exits 6585so that no additional actions are needed. This is the default mode. 6586 6587With KVM_BUS_LOCK_DETECTION_EXIT set, vm exits happen when bus lock detected 6588in VM. KVM just exits to userspace when handling them. Userspace can enforce 6589its own throttling or other policy based mitigations. 6590 6591This capability is aimed to address the thread that VM can exploit bus locks to 6592degree the performance of the whole system. Once the userspace enable this 6593capability and select the KVM_BUS_LOCK_DETECTION_EXIT mode, KVM will set the 6594KVM_RUN_BUS_LOCK flag in vcpu-run->flags field and exit to userspace. Concerning 6595the bus lock vm exit can be preempted by a higher priority VM exit, the exit 6596notifications to userspace can be KVM_EXIT_BUS_LOCK or other reasons. 6597KVM_RUN_BUS_LOCK flag is used to distinguish between them. 6598 65997.23 KVM_CAP_PPC_DAWR1 6600---------------------- 6601 6602:Architectures: ppc 6603:Parameters: none 6604:Returns: 0 on success, -EINVAL when CPU doesn't support 2nd DAWR 6605 6606This capability can be used to check / enable 2nd DAWR feature provided 6607by POWER10 processor. 6608 6609 66107.24 KVM_CAP_VM_COPY_ENC_CONTEXT_FROM 6611------------------------------------- 6612 6613Architectures: x86 SEV enabled 6614Type: vm 6615Parameters: args[0] is the fd of the source vm 6616Returns: 0 on success; ENOTTY on error 6617 6618This capability enables userspace to copy encryption context from the vm 6619indicated by the fd to the vm this is called on. 6620 6621This is intended to support in-guest workloads scheduled by the host. This 6622allows the in-guest workload to maintain its own NPTs and keeps the two vms 6623from accidentally clobbering each other with interrupts and the like (separate 6624APIC/MSRs/etc). 6625 66267.25 KVM_CAP_SGX_ATTRIBUTE 6627-------------------------- 6628 6629:Architectures: x86 6630:Target: VM 6631:Parameters: args[0] is a file handle of a SGX attribute file in securityfs 6632:Returns: 0 on success, -EINVAL if the file handle is invalid or if a requested 6633 attribute is not supported by KVM. 6634 6635KVM_CAP_SGX_ATTRIBUTE enables a userspace VMM to grant a VM access to one or 6636more priveleged enclave attributes. args[0] must hold a file handle to a valid 6637SGX attribute file corresponding to an attribute that is supported/restricted 6638by KVM (currently only PROVISIONKEY). 6639 6640The SGX subsystem restricts access to a subset of enclave attributes to provide 6641additional security for an uncompromised kernel, e.g. use of the PROVISIONKEY 6642is restricted to deter malware from using the PROVISIONKEY to obtain a stable 6643system fingerprint. To prevent userspace from circumventing such restrictions 6644by running an enclave in a VM, KVM prevents access to privileged attributes by 6645default. 6646 6647See Documentation/x86/sgx.rst for more details. 6648 66497.26 KVM_CAP_PPC_RPT_INVALIDATE 6650------------------------------- 6651 6652:Capability: KVM_CAP_PPC_RPT_INVALIDATE 6653:Architectures: ppc 6654:Type: vm 6655 6656This capability indicates that the kernel is capable of handling 6657H_RPT_INVALIDATE hcall. 6658 6659In order to enable the use of H_RPT_INVALIDATE in the guest, 6660user space might have to advertise it for the guest. For example, 6661IBM pSeries (sPAPR) guest starts using it if "hcall-rpt-invalidate" is 6662present in the "ibm,hypertas-functions" device-tree property. 6663 6664This capability is enabled for hypervisors on platforms like POWER9 6665that support radix MMU. 6666 66677.27 KVM_CAP_EXIT_ON_EMULATION_FAILURE 6668-------------------------------------- 6669 6670:Architectures: x86 6671:Parameters: args[0] whether the feature should be enabled or not 6672 6673When this capability is enabled, an emulation failure will result in an exit 6674to userspace with KVM_INTERNAL_ERROR (except when the emulator was invoked 6675to handle a VMware backdoor instruction). Furthermore, KVM will now provide up 6676to 15 instruction bytes for any exit to userspace resulting from an emulation 6677failure. When these exits to userspace occur use the emulation_failure struct 6678instead of the internal struct. They both have the same layout, but the 6679emulation_failure struct matches the content better. It also explicitly 6680defines the 'flags' field which is used to describe the fields in the struct 6681that are valid (ie: if KVM_INTERNAL_ERROR_EMULATION_FLAG_INSTRUCTION_BYTES is 6682set in the 'flags' field then both 'insn_size' and 'insn_bytes' have valid data 6683in them.) 6684 66857.28 KVM_CAP_ARM_MTE 6686-------------------- 6687 6688:Architectures: arm64 6689:Parameters: none 6690 6691This capability indicates that KVM (and the hardware) supports exposing the 6692Memory Tagging Extensions (MTE) to the guest. It must also be enabled by the 6693VMM before creating any VCPUs to allow the guest access. Note that MTE is only 6694available to a guest running in AArch64 mode and enabling this capability will 6695cause attempts to create AArch32 VCPUs to fail. 6696 6697When enabled the guest is able to access tags associated with any memory given 6698to the guest. KVM will ensure that the tags are maintained during swap or 6699hibernation of the host; however the VMM needs to manually save/restore the 6700tags as appropriate if the VM is migrated. 6701 6702When this capability is enabled all memory in memslots must be mapped as 6703not-shareable (no MAP_SHARED), attempts to create a memslot with a 6704MAP_SHARED mmap will result in an -EINVAL return. 6705 6706When enabled the VMM may make use of the ``KVM_ARM_MTE_COPY_TAGS`` ioctl to 6707perform a bulk copy of tags to/from the guest. 6708 67098. Other capabilities. 6710====================== 6711 6712This section lists capabilities that give information about other 6713features of the KVM implementation. 6714 67158.1 KVM_CAP_PPC_HWRNG 6716--------------------- 6717 6718:Architectures: ppc 6719 6720This capability, if KVM_CHECK_EXTENSION indicates that it is 6721available, means that the kernel has an implementation of the 6722H_RANDOM hypercall backed by a hardware random-number generator. 6723If present, the kernel H_RANDOM handler can be enabled for guest use 6724with the KVM_CAP_PPC_ENABLE_HCALL capability. 6725 67268.2 KVM_CAP_HYPERV_SYNIC 6727------------------------ 6728 6729:Architectures: x86 6730 6731This capability, if KVM_CHECK_EXTENSION indicates that it is 6732available, means that the kernel has an implementation of the 6733Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is 6734used to support Windows Hyper-V based guest paravirt drivers(VMBus). 6735 6736In order to use SynIC, it has to be activated by setting this 6737capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this 6738will disable the use of APIC hardware virtualization even if supported 6739by the CPU, as it's incompatible with SynIC auto-EOI behavior. 6740 67418.3 KVM_CAP_PPC_RADIX_MMU 6742------------------------- 6743 6744:Architectures: ppc 6745 6746This capability, if KVM_CHECK_EXTENSION indicates that it is 6747available, means that the kernel can support guests using the 6748radix MMU defined in Power ISA V3.00 (as implemented in the POWER9 6749processor). 6750 67518.4 KVM_CAP_PPC_HASH_MMU_V3 6752--------------------------- 6753 6754:Architectures: ppc 6755 6756This capability, if KVM_CHECK_EXTENSION indicates that it is 6757available, means that the kernel can support guests using the 6758hashed page table MMU defined in Power ISA V3.00 (as implemented in 6759the POWER9 processor), including in-memory segment tables. 6760 67618.5 KVM_CAP_MIPS_VZ 6762------------------- 6763 6764:Architectures: mips 6765 6766This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that 6767it is available, means that full hardware assisted virtualization capabilities 6768of the hardware are available for use through KVM. An appropriate 6769KVM_VM_MIPS_* type must be passed to KVM_CREATE_VM to create a VM which 6770utilises it. 6771 6772If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is 6773available, it means that the VM is using full hardware assisted virtualization 6774capabilities of the hardware. This is useful to check after creating a VM with 6775KVM_VM_MIPS_DEFAULT. 6776 6777The value returned by KVM_CHECK_EXTENSION should be compared against known 6778values (see below). All other values are reserved. This is to allow for the 6779possibility of other hardware assisted virtualization implementations which 6780may be incompatible with the MIPS VZ ASE. 6781 6782== ========================================================================== 6783 0 The trap & emulate implementation is in use to run guest code in user 6784 mode. Guest virtual memory segments are rearranged to fit the guest in the 6785 user mode address space. 6786 6787 1 The MIPS VZ ASE is in use, providing full hardware assisted 6788 virtualization, including standard guest virtual memory segments. 6789== ========================================================================== 6790 67918.6 KVM_CAP_MIPS_TE 6792------------------- 6793 6794:Architectures: mips 6795 6796This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that 6797it is available, means that the trap & emulate implementation is available to 6798run guest code in user mode, even if KVM_CAP_MIPS_VZ indicates that hardware 6799assisted virtualisation is also available. KVM_VM_MIPS_TE (0) must be passed 6800to KVM_CREATE_VM to create a VM which utilises it. 6801 6802If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is 6803available, it means that the VM is using trap & emulate. 6804 68058.7 KVM_CAP_MIPS_64BIT 6806---------------------- 6807 6808:Architectures: mips 6809 6810This capability indicates the supported architecture type of the guest, i.e. the 6811supported register and address width. 6812 6813The values returned when this capability is checked by KVM_CHECK_EXTENSION on a 6814kvm VM handle correspond roughly to the CP0_Config.AT register field, and should 6815be checked specifically against known values (see below). All other values are 6816reserved. 6817 6818== ======================================================================== 6819 0 MIPS32 or microMIPS32. 6820 Both registers and addresses are 32-bits wide. 6821 It will only be possible to run 32-bit guest code. 6822 6823 1 MIPS64 or microMIPS64 with access only to 32-bit compatibility segments. 6824 Registers are 64-bits wide, but addresses are 32-bits wide. 6825 64-bit guest code may run but cannot access MIPS64 memory segments. 6826 It will also be possible to run 32-bit guest code. 6827 6828 2 MIPS64 or microMIPS64 with access to all address segments. 6829 Both registers and addresses are 64-bits wide. 6830 It will be possible to run 64-bit or 32-bit guest code. 6831== ======================================================================== 6832 68338.9 KVM_CAP_ARM_USER_IRQ 6834------------------------ 6835 6836:Architectures: arm, arm64 6837 6838This capability, if KVM_CHECK_EXTENSION indicates that it is available, means 6839that if userspace creates a VM without an in-kernel interrupt controller, it 6840will be notified of changes to the output level of in-kernel emulated devices, 6841which can generate virtual interrupts, presented to the VM. 6842For such VMs, on every return to userspace, the kernel 6843updates the vcpu's run->s.regs.device_irq_level field to represent the actual 6844output level of the device. 6845 6846Whenever kvm detects a change in the device output level, kvm guarantees at 6847least one return to userspace before running the VM. This exit could either 6848be a KVM_EXIT_INTR or any other exit event, like KVM_EXIT_MMIO. This way, 6849userspace can always sample the device output level and re-compute the state of 6850the userspace interrupt controller. Userspace should always check the state 6851of run->s.regs.device_irq_level on every kvm exit. 6852The value in run->s.regs.device_irq_level can represent both level and edge 6853triggered interrupt signals, depending on the device. Edge triggered interrupt 6854signals will exit to userspace with the bit in run->s.regs.device_irq_level 6855set exactly once per edge signal. 6856 6857The field run->s.regs.device_irq_level is available independent of 6858run->kvm_valid_regs or run->kvm_dirty_regs bits. 6859 6860If KVM_CAP_ARM_USER_IRQ is supported, the KVM_CHECK_EXTENSION ioctl returns a 6861number larger than 0 indicating the version of this capability is implemented 6862and thereby which bits in run->s.regs.device_irq_level can signal values. 6863 6864Currently the following bits are defined for the device_irq_level bitmap:: 6865 6866 KVM_CAP_ARM_USER_IRQ >= 1: 6867 6868 KVM_ARM_DEV_EL1_VTIMER - EL1 virtual timer 6869 KVM_ARM_DEV_EL1_PTIMER - EL1 physical timer 6870 KVM_ARM_DEV_PMU - ARM PMU overflow interrupt signal 6871 6872Future versions of kvm may implement additional events. These will get 6873indicated by returning a higher number from KVM_CHECK_EXTENSION and will be 6874listed above. 6875 68768.10 KVM_CAP_PPC_SMT_POSSIBLE 6877----------------------------- 6878 6879:Architectures: ppc 6880 6881Querying this capability returns a bitmap indicating the possible 6882virtual SMT modes that can be set using KVM_CAP_PPC_SMT. If bit N 6883(counting from the right) is set, then a virtual SMT mode of 2^N is 6884available. 6885 68868.11 KVM_CAP_HYPERV_SYNIC2 6887-------------------------- 6888 6889:Architectures: x86 6890 6891This capability enables a newer version of Hyper-V Synthetic interrupt 6892controller (SynIC). The only difference with KVM_CAP_HYPERV_SYNIC is that KVM 6893doesn't clear SynIC message and event flags pages when they are enabled by 6894writing to the respective MSRs. 6895 68968.12 KVM_CAP_HYPERV_VP_INDEX 6897---------------------------- 6898 6899:Architectures: x86 6900 6901This capability indicates that userspace can load HV_X64_MSR_VP_INDEX msr. Its 6902value is used to denote the target vcpu for a SynIC interrupt. For 6903compatibilty, KVM initializes this msr to KVM's internal vcpu index. When this 6904capability is absent, userspace can still query this msr's value. 6905 69068.13 KVM_CAP_S390_AIS_MIGRATION 6907------------------------------- 6908 6909:Architectures: s390 6910:Parameters: none 6911 6912This capability indicates if the flic device will be able to get/set the 6913AIS states for migration via the KVM_DEV_FLIC_AISM_ALL attribute and allows 6914to discover this without having to create a flic device. 6915 69168.14 KVM_CAP_S390_PSW 6917--------------------- 6918 6919:Architectures: s390 6920 6921This capability indicates that the PSW is exposed via the kvm_run structure. 6922 69238.15 KVM_CAP_S390_GMAP 6924---------------------- 6925 6926:Architectures: s390 6927 6928This capability indicates that the user space memory used as guest mapping can 6929be anywhere in the user memory address space, as long as the memory slots are 6930aligned and sized to a segment (1MB) boundary. 6931 69328.16 KVM_CAP_S390_COW 6933--------------------- 6934 6935:Architectures: s390 6936 6937This capability indicates that the user space memory used as guest mapping can 6938use copy-on-write semantics as well as dirty pages tracking via read-only page 6939tables. 6940 69418.17 KVM_CAP_S390_BPB 6942--------------------- 6943 6944:Architectures: s390 6945 6946This capability indicates that kvm will implement the interfaces to handle 6947reset, migration and nested KVM for branch prediction blocking. The stfle 6948facility 82 should not be provided to the guest without this capability. 6949 69508.18 KVM_CAP_HYPERV_TLBFLUSH 6951---------------------------- 6952 6953:Architectures: x86 6954 6955This capability indicates that KVM supports paravirtualized Hyper-V TLB Flush 6956hypercalls: 6957HvFlushVirtualAddressSpace, HvFlushVirtualAddressSpaceEx, 6958HvFlushVirtualAddressList, HvFlushVirtualAddressListEx. 6959 69608.19 KVM_CAP_ARM_INJECT_SERROR_ESR 6961---------------------------------- 6962 6963:Architectures: arm, arm64 6964 6965This capability indicates that userspace can specify (via the 6966KVM_SET_VCPU_EVENTS ioctl) the syndrome value reported to the guest when it 6967takes a virtual SError interrupt exception. 6968If KVM advertises this capability, userspace can only specify the ISS field for 6969the ESR syndrome. Other parts of the ESR, such as the EC are generated by the 6970CPU when the exception is taken. If this virtual SError is taken to EL1 using 6971AArch64, this value will be reported in the ISS field of ESR_ELx. 6972 6973See KVM_CAP_VCPU_EVENTS for more details. 6974 69758.20 KVM_CAP_HYPERV_SEND_IPI 6976---------------------------- 6977 6978:Architectures: x86 6979 6980This capability indicates that KVM supports paravirtualized Hyper-V IPI send 6981hypercalls: 6982HvCallSendSyntheticClusterIpi, HvCallSendSyntheticClusterIpiEx. 6983 69848.21 KVM_CAP_HYPERV_DIRECT_TLBFLUSH 6985----------------------------------- 6986 6987:Architectures: x86 6988 6989This capability indicates that KVM running on top of Hyper-V hypervisor 6990enables Direct TLB flush for its guests meaning that TLB flush 6991hypercalls are handled by Level 0 hypervisor (Hyper-V) bypassing KVM. 6992Due to the different ABI for hypercall parameters between Hyper-V and 6993KVM, enabling this capability effectively disables all hypercall 6994handling by KVM (as some KVM hypercall may be mistakenly treated as TLB 6995flush hypercalls by Hyper-V) so userspace should disable KVM identification 6996in CPUID and only exposes Hyper-V identification. In this case, guest 6997thinks it's running on Hyper-V and only use Hyper-V hypercalls. 6998 69998.22 KVM_CAP_S390_VCPU_RESETS 7000----------------------------- 7001 7002:Architectures: s390 7003 7004This capability indicates that the KVM_S390_NORMAL_RESET and 7005KVM_S390_CLEAR_RESET ioctls are available. 7006 70078.23 KVM_CAP_S390_PROTECTED 7008--------------------------- 7009 7010:Architectures: s390 7011 7012This capability indicates that the Ultravisor has been initialized and 7013KVM can therefore start protected VMs. 7014This capability governs the KVM_S390_PV_COMMAND ioctl and the 7015KVM_MP_STATE_LOAD MP_STATE. KVM_SET_MP_STATE can fail for protected 7016guests when the state change is invalid. 7017 70188.24 KVM_CAP_STEAL_TIME 7019----------------------- 7020 7021:Architectures: arm64, x86 7022 7023This capability indicates that KVM supports steal time accounting. 7024When steal time accounting is supported it may be enabled with 7025architecture-specific interfaces. This capability and the architecture- 7026specific interfaces must be consistent, i.e. if one says the feature 7027is supported, than the other should as well and vice versa. For arm64 7028see Documentation/virt/kvm/devices/vcpu.rst "KVM_ARM_VCPU_PVTIME_CTRL". 7029For x86 see Documentation/virt/kvm/msr.rst "MSR_KVM_STEAL_TIME". 7030 70318.25 KVM_CAP_S390_DIAG318 7032------------------------- 7033 7034:Architectures: s390 7035 7036This capability enables a guest to set information about its control program 7037(i.e. guest kernel type and version). The information is helpful during 7038system/firmware service events, providing additional data about the guest 7039environments running on the machine. 7040 7041The information is associated with the DIAGNOSE 0x318 instruction, which sets 7042an 8-byte value consisting of a one-byte Control Program Name Code (CPNC) and 7043a 7-byte Control Program Version Code (CPVC). The CPNC determines what 7044environment the control program is running in (e.g. Linux, z/VM...), and the 7045CPVC is used for information specific to OS (e.g. Linux version, Linux 7046distribution...) 7047 7048If this capability is available, then the CPNC and CPVC can be synchronized 7049between KVM and userspace via the sync regs mechanism (KVM_SYNC_DIAG318). 7050 70518.26 KVM_CAP_X86_USER_SPACE_MSR 7052------------------------------- 7053 7054:Architectures: x86 7055 7056This capability indicates that KVM supports deflection of MSR reads and 7057writes to user space. It can be enabled on a VM level. If enabled, MSR 7058accesses that would usually trigger a #GP by KVM into the guest will 7059instead get bounced to user space through the KVM_EXIT_X86_RDMSR and 7060KVM_EXIT_X86_WRMSR exit notifications. 7061 70628.27 KVM_CAP_X86_MSR_FILTER 7063--------------------------- 7064 7065:Architectures: x86 7066 7067This capability indicates that KVM supports that accesses to user defined MSRs 7068may be rejected. With this capability exposed, KVM exports new VM ioctl 7069KVM_X86_SET_MSR_FILTER which user space can call to specify bitmaps of MSR 7070ranges that KVM should reject access to. 7071 7072In combination with KVM_CAP_X86_USER_SPACE_MSR, this allows user space to 7073trap and emulate MSRs that are outside of the scope of KVM as well as 7074limit the attack surface on KVM's MSR emulation code. 7075 70768.28 KVM_CAP_ENFORCE_PV_FEATURE_CPUID 7077----------------------------- 7078 7079Architectures: x86 7080 7081When enabled, KVM will disable paravirtual features provided to the 7082guest according to the bits in the KVM_CPUID_FEATURES CPUID leaf 7083(0x40000001). Otherwise, a guest may use the paravirtual features 7084regardless of what has actually been exposed through the CPUID leaf. 7085 70868.29 KVM_CAP_DIRTY_LOG_RING 7087--------------------------- 7088 7089:Architectures: x86 7090:Parameters: args[0] - size of the dirty log ring 7091 7092KVM is capable of tracking dirty memory using ring buffers that are 7093mmaped into userspace; there is one dirty ring per vcpu. 7094 7095The dirty ring is available to userspace as an array of 7096``struct kvm_dirty_gfn``. Each dirty entry it's defined as:: 7097 7098 struct kvm_dirty_gfn { 7099 __u32 flags; 7100 __u32 slot; /* as_id | slot_id */ 7101 __u64 offset; 7102 }; 7103 7104The following values are defined for the flags field to define the 7105current state of the entry:: 7106 7107 #define KVM_DIRTY_GFN_F_DIRTY BIT(0) 7108 #define KVM_DIRTY_GFN_F_RESET BIT(1) 7109 #define KVM_DIRTY_GFN_F_MASK 0x3 7110 7111Userspace should call KVM_ENABLE_CAP ioctl right after KVM_CREATE_VM 7112ioctl to enable this capability for the new guest and set the size of 7113the rings. Enabling the capability is only allowed before creating any 7114vCPU, and the size of the ring must be a power of two. The larger the 7115ring buffer, the less likely the ring is full and the VM is forced to 7116exit to userspace. The optimal size depends on the workload, but it is 7117recommended that it be at least 64 KiB (4096 entries). 7118 7119Just like for dirty page bitmaps, the buffer tracks writes to 7120all user memory regions for which the KVM_MEM_LOG_DIRTY_PAGES flag was 7121set in KVM_SET_USER_MEMORY_REGION. Once a memory region is registered 7122with the flag set, userspace can start harvesting dirty pages from the 7123ring buffer. 7124 7125An entry in the ring buffer can be unused (flag bits ``00``), 7126dirty (flag bits ``01``) or harvested (flag bits ``1X``). The 7127state machine for the entry is as follows:: 7128 7129 dirtied harvested reset 7130 00 -----------> 01 -------------> 1X -------+ 7131 ^ | 7132 | | 7133 +------------------------------------------+ 7134 7135To harvest the dirty pages, userspace accesses the mmaped ring buffer 7136to read the dirty GFNs. If the flags has the DIRTY bit set (at this stage 7137the RESET bit must be cleared), then it means this GFN is a dirty GFN. 7138The userspace should harvest this GFN and mark the flags from state 7139``01b`` to ``1Xb`` (bit 0 will be ignored by KVM, but bit 1 must be set 7140to show that this GFN is harvested and waiting for a reset), and move 7141on to the next GFN. The userspace should continue to do this until the 7142flags of a GFN have the DIRTY bit cleared, meaning that it has harvested 7143all the dirty GFNs that were available. 7144 7145It's not necessary for userspace to harvest the all dirty GFNs at once. 7146However it must collect the dirty GFNs in sequence, i.e., the userspace 7147program cannot skip one dirty GFN to collect the one next to it. 7148 7149After processing one or more entries in the ring buffer, userspace 7150calls the VM ioctl KVM_RESET_DIRTY_RINGS to notify the kernel about 7151it, so that the kernel will reprotect those collected GFNs. 7152Therefore, the ioctl must be called *before* reading the content of 7153the dirty pages. 7154 7155The dirty ring can get full. When it happens, the KVM_RUN of the 7156vcpu will return with exit reason KVM_EXIT_DIRTY_LOG_FULL. 7157 7158The dirty ring interface has a major difference comparing to the 7159KVM_GET_DIRTY_LOG interface in that, when reading the dirty ring from 7160userspace, it's still possible that the kernel has not yet flushed the 7161processor's dirty page buffers into the kernel buffer (with dirty bitmaps, the 7162flushing is done by the KVM_GET_DIRTY_LOG ioctl). To achieve that, one 7163needs to kick the vcpu out of KVM_RUN using a signal. The resulting 7164vmexit ensures that all dirty GFNs are flushed to the dirty rings. 7165 7166NOTE: the capability KVM_CAP_DIRTY_LOG_RING and the corresponding 7167ioctl KVM_RESET_DIRTY_RINGS are mutual exclusive to the existing ioctls 7168KVM_GET_DIRTY_LOG and KVM_CLEAR_DIRTY_LOG. After enabling 7169KVM_CAP_DIRTY_LOG_RING with an acceptable dirty ring size, the virtual 7170machine will switch to ring-buffer dirty page tracking and further 7171KVM_GET_DIRTY_LOG or KVM_CLEAR_DIRTY_LOG ioctls will fail. 7172 71738.30 KVM_CAP_XEN_HVM 7174-------------------- 7175 7176:Architectures: x86 7177 7178This capability indicates the features that Xen supports for hosting Xen 7179PVHVM guests. Valid flags are:: 7180 7181 #define KVM_XEN_HVM_CONFIG_HYPERCALL_MSR (1 << 0) 7182 #define KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL (1 << 1) 7183 #define KVM_XEN_HVM_CONFIG_SHARED_INFO (1 << 2) 7184 #define KVM_XEN_HVM_CONFIG_RUNSTATE (1 << 2) 7185 7186The KVM_XEN_HVM_CONFIG_HYPERCALL_MSR flag indicates that the KVM_XEN_HVM_CONFIG 7187ioctl is available, for the guest to set its hypercall page. 7188 7189If KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL is also set, the same flag may also be 7190provided in the flags to KVM_XEN_HVM_CONFIG, without providing hypercall page 7191contents, to request that KVM generate hypercall page content automatically 7192and also enable interception of guest hypercalls with KVM_EXIT_XEN. 7193 7194The KVM_XEN_HVM_CONFIG_SHARED_INFO flag indicates the availability of the 7195KVM_XEN_HVM_SET_ATTR, KVM_XEN_HVM_GET_ATTR, KVM_XEN_VCPU_SET_ATTR and 7196KVM_XEN_VCPU_GET_ATTR ioctls, as well as the delivery of exception vectors 7197for event channel upcalls when the evtchn_upcall_pending field of a vcpu's 7198vcpu_info is set. 7199 7200The KVM_XEN_HVM_CONFIG_RUNSTATE flag indicates that the runstate-related 7201features KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR/_CURRENT/_DATA/_ADJUST are 7202supported by the KVM_XEN_VCPU_SET_ATTR/KVM_XEN_VCPU_GET_ATTR ioctls. 7203 72048.31 KVM_CAP_PPC_MULTITCE 7205------------------------- 7206 7207:Capability: KVM_CAP_PPC_MULTITCE 7208:Architectures: ppc 7209:Type: vm 7210 7211This capability means the kernel is capable of handling hypercalls 7212H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user 7213space. This significantly accelerates DMA operations for PPC KVM guests. 7214User space should expect that its handlers for these hypercalls 7215are not going to be called if user space previously registered LIOBN 7216in KVM (via KVM_CREATE_SPAPR_TCE or similar calls). 7217 7218In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest, 7219user space might have to advertise it for the guest. For example, 7220IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is 7221present in the "ibm,hypertas-functions" device-tree property. 7222 7223The hypercalls mentioned above may or may not be processed successfully 7224in the kernel based fast path. If they can not be handled by the kernel, 7225they will get passed on to user space. So user space still has to have 7226an implementation for these despite the in kernel acceleration. 7227 7228This capability is always enabled. 7229 72308.32 KVM_CAP_PTP_KVM 7231-------------------- 7232 7233:Architectures: arm64 7234 7235This capability indicates that the KVM virtual PTP service is 7236supported in the host. A VMM can check whether the service is 7237available to the guest on migration. 7238 72398.33 KVM_CAP_HYPERV_ENFORCE_CPUID 7240--------------------------------- 7241 7242Architectures: x86 7243 7244When enabled, KVM will disable emulated Hyper-V features provided to the 7245guest according to the bits Hyper-V CPUID feature leaves. Otherwise, all 7246currently implmented Hyper-V features are provided unconditionally when 7247Hyper-V identification is set in the HYPERV_CPUID_INTERFACE (0x40000001) 7248leaf. 7249 72508.34 KVM_CAP_EXIT_HYPERCALL 7251--------------------------- 7252 7253:Capability: KVM_CAP_EXIT_HYPERCALL 7254:Architectures: x86 7255:Type: vm 7256 7257This capability, if enabled, will cause KVM to exit to userspace 7258with KVM_EXIT_HYPERCALL exit reason to process some hypercalls. 7259 7260Calling KVM_CHECK_EXTENSION for this capability will return a bitmask 7261of hypercalls that can be configured to exit to userspace. 7262Right now, the only such hypercall is KVM_HC_MAP_GPA_RANGE. 7263 7264The argument to KVM_ENABLE_CAP is also a bitmask, and must be a subset 7265of the result of KVM_CHECK_EXTENSION. KVM will forward to userspace 7266the hypercalls whose corresponding bit is in the argument, and return 7267ENOSYS for the others. 7268