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 althought 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 via 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 185Please note that configuring the IPA size does not affect the capability 186exposed by the guest CPUs in ID_AA64MMFR0_EL1[PARange]. It only affects 187size of the address translated by the stage2 level (guest physical to 188host physical address translations). 189 190 1914.3 KVM_GET_MSR_INDEX_LIST, KVM_GET_MSR_FEATURE_INDEX_LIST 192---------------------------------------------------------- 193 194:Capability: basic, KVM_CAP_GET_MSR_FEATURES for KVM_GET_MSR_FEATURE_INDEX_LIST 195:Architectures: x86 196:Type: system ioctl 197:Parameters: struct kvm_msr_list (in/out) 198:Returns: 0 on success; -1 on error 199 200Errors: 201 202 ====== ============================================================ 203 EFAULT the msr index list cannot be read from or written to 204 E2BIG the msr index list is to be to fit in the array specified by 205 the user. 206 ====== ============================================================ 207 208:: 209 210 struct kvm_msr_list { 211 __u32 nmsrs; /* number of msrs in entries */ 212 __u32 indices[0]; 213 }; 214 215The user fills in the size of the indices array in nmsrs, and in return 216kvm adjusts nmsrs to reflect the actual number of msrs and fills in the 217indices array with their numbers. 218 219KVM_GET_MSR_INDEX_LIST returns the guest msrs that are supported. The list 220varies by kvm version and host processor, but does not change otherwise. 221 222Note: if kvm indicates supports MCE (KVM_CAP_MCE), then the MCE bank MSRs are 223not returned in the MSR list, as different vcpus can have a different number 224of banks, as set via the KVM_X86_SETUP_MCE ioctl. 225 226KVM_GET_MSR_FEATURE_INDEX_LIST returns the list of MSRs that can be passed 227to the KVM_GET_MSRS system ioctl. This lets userspace probe host capabilities 228and processor features that are exposed via MSRs (e.g., VMX capabilities). 229This list also varies by kvm version and host processor, but does not change 230otherwise. 231 232 2334.4 KVM_CHECK_EXTENSION 234----------------------- 235 236:Capability: basic, KVM_CAP_CHECK_EXTENSION_VM for vm ioctl 237:Architectures: all 238:Type: system ioctl, vm ioctl 239:Parameters: extension identifier (KVM_CAP_*) 240:Returns: 0 if unsupported; 1 (or some other positive integer) if supported 241 242The API allows the application to query about extensions to the core 243kvm API. Userspace passes an extension identifier (an integer) and 244receives an integer that describes the extension availability. 245Generally 0 means no and 1 means yes, but some extensions may report 246additional information in the integer return value. 247 248Based on their initialization different VMs may have different capabilities. 249It is thus encouraged to use the vm ioctl to query for capabilities (available 250with KVM_CAP_CHECK_EXTENSION_VM on the vm fd) 251 2524.5 KVM_GET_VCPU_MMAP_SIZE 253-------------------------- 254 255:Capability: basic 256:Architectures: all 257:Type: system ioctl 258:Parameters: none 259:Returns: size of vcpu mmap area, in bytes 260 261The KVM_RUN ioctl (cf.) communicates with userspace via a shared 262memory region. This ioctl returns the size of that region. See the 263KVM_RUN documentation for details. 264 265 2664.6 KVM_SET_MEMORY_REGION 267------------------------- 268 269:Capability: basic 270:Architectures: all 271:Type: vm ioctl 272:Parameters: struct kvm_memory_region (in) 273:Returns: 0 on success, -1 on error 274 275This ioctl is obsolete and has been removed. 276 277 2784.7 KVM_CREATE_VCPU 279------------------- 280 281:Capability: basic 282:Architectures: all 283:Type: vm ioctl 284:Parameters: vcpu id (apic id on x86) 285:Returns: vcpu fd on success, -1 on error 286 287This API adds a vcpu to a virtual machine. No more than max_vcpus may be added. 288The vcpu id is an integer in the range [0, max_vcpu_id). 289 290The recommended max_vcpus value can be retrieved using the KVM_CAP_NR_VCPUS of 291the KVM_CHECK_EXTENSION ioctl() at run-time. 292The maximum possible value for max_vcpus can be retrieved using the 293KVM_CAP_MAX_VCPUS of the KVM_CHECK_EXTENSION ioctl() at run-time. 294 295If the KVM_CAP_NR_VCPUS does not exist, you should assume that max_vcpus is 4 296cpus max. 297If the KVM_CAP_MAX_VCPUS does not exist, you should assume that max_vcpus is 298same as the value returned from KVM_CAP_NR_VCPUS. 299 300The maximum possible value for max_vcpu_id can be retrieved using the 301KVM_CAP_MAX_VCPU_ID of the KVM_CHECK_EXTENSION ioctl() at run-time. 302 303If the KVM_CAP_MAX_VCPU_ID does not exist, you should assume that max_vcpu_id 304is the same as the value returned from KVM_CAP_MAX_VCPUS. 305 306On powerpc using book3s_hv mode, the vcpus are mapped onto virtual 307threads in one or more virtual CPU cores. (This is because the 308hardware requires all the hardware threads in a CPU core to be in the 309same partition.) The KVM_CAP_PPC_SMT capability indicates the number 310of vcpus per virtual core (vcore). The vcore id is obtained by 311dividing the vcpu id by the number of vcpus per vcore. The vcpus in a 312given vcore will always be in the same physical core as each other 313(though that might be a different physical core from time to time). 314Userspace can control the threading (SMT) mode of the guest by its 315allocation of vcpu ids. For example, if userspace wants 316single-threaded guest vcpus, it should make all vcpu ids be a multiple 317of the number of vcpus per vcore. 318 319For virtual cpus that have been created with S390 user controlled virtual 320machines, the resulting vcpu fd can be memory mapped at page offset 321KVM_S390_SIE_PAGE_OFFSET in order to obtain a memory map of the virtual 322cpu's hardware control block. 323 324 3254.8 KVM_GET_DIRTY_LOG (vm ioctl) 326-------------------------------- 327 328:Capability: basic 329:Architectures: all 330:Type: vm ioctl 331:Parameters: struct kvm_dirty_log (in/out) 332:Returns: 0 on success, -1 on error 333 334:: 335 336 /* for KVM_GET_DIRTY_LOG */ 337 struct kvm_dirty_log { 338 __u32 slot; 339 __u32 padding; 340 union { 341 void __user *dirty_bitmap; /* one bit per page */ 342 __u64 padding; 343 }; 344 }; 345 346Given a memory slot, return a bitmap containing any pages dirtied 347since the last call to this ioctl. Bit 0 is the first page in the 348memory slot. Ensure the entire structure is cleared to avoid padding 349issues. 350 351If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 specifies 352the address space for which you want to return the dirty bitmap. 353They must be less than the value that KVM_CHECK_EXTENSION returns for 354the KVM_CAP_MULTI_ADDRESS_SPACE capability. 355 356The bits in the dirty bitmap are cleared before the ioctl returns, unless 357KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is enabled. For more information, 358see the description of the capability. 359 3604.9 KVM_SET_MEMORY_ALIAS 361------------------------ 362 363:Capability: basic 364:Architectures: x86 365:Type: vm ioctl 366:Parameters: struct kvm_memory_alias (in) 367:Returns: 0 (success), -1 (error) 368 369This ioctl is obsolete and has been removed. 370 371 3724.10 KVM_RUN 373------------ 374 375:Capability: basic 376:Architectures: all 377:Type: vcpu ioctl 378:Parameters: none 379:Returns: 0 on success, -1 on error 380 381Errors: 382 383 ===== ============================= 384 EINTR an unmasked signal is pending 385 ===== ============================= 386 387This ioctl is used to run a guest virtual cpu. While there are no 388explicit parameters, there is an implicit parameter block that can be 389obtained by mmap()ing the vcpu fd at offset 0, with the size given by 390KVM_GET_VCPU_MMAP_SIZE. The parameter block is formatted as a 'struct 391kvm_run' (see below). 392 393 3944.11 KVM_GET_REGS 395----------------- 396 397:Capability: basic 398:Architectures: all except ARM, arm64 399:Type: vcpu ioctl 400:Parameters: struct kvm_regs (out) 401:Returns: 0 on success, -1 on error 402 403Reads the general purpose registers from the vcpu. 404 405:: 406 407 /* x86 */ 408 struct kvm_regs { 409 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */ 410 __u64 rax, rbx, rcx, rdx; 411 __u64 rsi, rdi, rsp, rbp; 412 __u64 r8, r9, r10, r11; 413 __u64 r12, r13, r14, r15; 414 __u64 rip, rflags; 415 }; 416 417 /* mips */ 418 struct kvm_regs { 419 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */ 420 __u64 gpr[32]; 421 __u64 hi; 422 __u64 lo; 423 __u64 pc; 424 }; 425 426 4274.12 KVM_SET_REGS 428----------------- 429 430:Capability: basic 431:Architectures: all except ARM, arm64 432:Type: vcpu ioctl 433:Parameters: struct kvm_regs (in) 434:Returns: 0 on success, -1 on error 435 436Writes the general purpose registers into the vcpu. 437 438See KVM_GET_REGS for the data structure. 439 440 4414.13 KVM_GET_SREGS 442------------------ 443 444:Capability: basic 445:Architectures: x86, ppc 446:Type: vcpu ioctl 447:Parameters: struct kvm_sregs (out) 448:Returns: 0 on success, -1 on error 449 450Reads special registers from the vcpu. 451 452:: 453 454 /* x86 */ 455 struct kvm_sregs { 456 struct kvm_segment cs, ds, es, fs, gs, ss; 457 struct kvm_segment tr, ldt; 458 struct kvm_dtable gdt, idt; 459 __u64 cr0, cr2, cr3, cr4, cr8; 460 __u64 efer; 461 __u64 apic_base; 462 __u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64]; 463 }; 464 465 /* ppc -- see arch/powerpc/include/uapi/asm/kvm.h */ 466 467interrupt_bitmap is a bitmap of pending external interrupts. At most 468one bit may be set. This interrupt has been acknowledged by the APIC 469but not yet injected into the cpu core. 470 471 4724.14 KVM_SET_SREGS 473------------------ 474 475:Capability: basic 476:Architectures: x86, ppc 477:Type: vcpu ioctl 478:Parameters: struct kvm_sregs (in) 479:Returns: 0 on success, -1 on error 480 481Writes special registers into the vcpu. See KVM_GET_SREGS for the 482data structures. 483 484 4854.15 KVM_TRANSLATE 486------------------ 487 488:Capability: basic 489:Architectures: x86 490:Type: vcpu ioctl 491:Parameters: struct kvm_translation (in/out) 492:Returns: 0 on success, -1 on error 493 494Translates a virtual address according to the vcpu's current address 495translation mode. 496 497:: 498 499 struct kvm_translation { 500 /* in */ 501 __u64 linear_address; 502 503 /* out */ 504 __u64 physical_address; 505 __u8 valid; 506 __u8 writeable; 507 __u8 usermode; 508 __u8 pad[5]; 509 }; 510 511 5124.16 KVM_INTERRUPT 513------------------ 514 515:Capability: basic 516:Architectures: x86, ppc, mips 517:Type: vcpu ioctl 518:Parameters: struct kvm_interrupt (in) 519:Returns: 0 on success, negative on failure. 520 521Queues a hardware interrupt vector to be injected. 522 523:: 524 525 /* for KVM_INTERRUPT */ 526 struct kvm_interrupt { 527 /* in */ 528 __u32 irq; 529 }; 530 531X86: 532^^^^ 533 534:Returns: 535 536 ========= =================================== 537 0 on success, 538 -EEXIST if an interrupt is already enqueued 539 -EINVAL the the irq number is invalid 540 -ENXIO if the PIC is in the kernel 541 -EFAULT if the pointer is invalid 542 ========= =================================== 543 544Note 'irq' is an interrupt vector, not an interrupt pin or line. This 545ioctl is useful if the in-kernel PIC is not used. 546 547PPC: 548^^^^ 549 550Queues an external interrupt to be injected. This ioctl is overleaded 551with 3 different irq values: 552 553a) KVM_INTERRUPT_SET 554 555 This injects an edge type external interrupt into the guest once it's ready 556 to receive interrupts. When injected, the interrupt is done. 557 558b) KVM_INTERRUPT_UNSET 559 560 This unsets any pending interrupt. 561 562 Only available with KVM_CAP_PPC_UNSET_IRQ. 563 564c) KVM_INTERRUPT_SET_LEVEL 565 566 This injects a level type external interrupt into the guest context. The 567 interrupt stays pending until a specific ioctl with KVM_INTERRUPT_UNSET 568 is triggered. 569 570 Only available with KVM_CAP_PPC_IRQ_LEVEL. 571 572Note that any value for 'irq' other than the ones stated above is invalid 573and incurs unexpected behavior. 574 575This is an asynchronous vcpu ioctl and can be invoked from any thread. 576 577MIPS: 578^^^^^ 579 580Queues an external interrupt to be injected into the virtual CPU. A negative 581interrupt number dequeues the interrupt. 582 583This is an asynchronous vcpu ioctl and can be invoked from any thread. 584 585 5864.17 KVM_DEBUG_GUEST 587-------------------- 588 589:Capability: basic 590:Architectures: none 591:Type: vcpu ioctl 592:Parameters: none) 593:Returns: -1 on error 594 595Support for this has been removed. Use KVM_SET_GUEST_DEBUG instead. 596 597 5984.18 KVM_GET_MSRS 599----------------- 600 601:Capability: basic (vcpu), KVM_CAP_GET_MSR_FEATURES (system) 602:Architectures: x86 603:Type: system ioctl, vcpu ioctl 604:Parameters: struct kvm_msrs (in/out) 605:Returns: number of msrs successfully returned; 606 -1 on error 607 608When used as a system ioctl: 609Reads the values of MSR-based features that are available for the VM. This 610is similar to KVM_GET_SUPPORTED_CPUID, but it returns MSR indices and values. 611The list of msr-based features can be obtained using KVM_GET_MSR_FEATURE_INDEX_LIST 612in a system ioctl. 613 614When used as a vcpu ioctl: 615Reads model-specific registers from the vcpu. Supported msr indices can 616be obtained using KVM_GET_MSR_INDEX_LIST in a system ioctl. 617 618:: 619 620 struct kvm_msrs { 621 __u32 nmsrs; /* number of msrs in entries */ 622 __u32 pad; 623 624 struct kvm_msr_entry entries[0]; 625 }; 626 627 struct kvm_msr_entry { 628 __u32 index; 629 __u32 reserved; 630 __u64 data; 631 }; 632 633Application code should set the 'nmsrs' member (which indicates the 634size of the entries array) and the 'index' member of each array entry. 635kvm will fill in the 'data' member. 636 637 6384.19 KVM_SET_MSRS 639----------------- 640 641:Capability: basic 642:Architectures: x86 643:Type: vcpu ioctl 644:Parameters: struct kvm_msrs (in) 645:Returns: number of msrs successfully set (see below), -1 on error 646 647Writes model-specific registers to the vcpu. See KVM_GET_MSRS for the 648data structures. 649 650Application code should set the 'nmsrs' member (which indicates the 651size of the entries array), and the 'index' and 'data' members of each 652array entry. 653 654It tries to set the MSRs in array entries[] one by one. If setting an MSR 655fails, e.g., due to setting reserved bits, the MSR isn't supported/emulated 656by KVM, etc..., it stops processing the MSR list and returns the number of 657MSRs that have been set successfully. 658 659 6604.20 KVM_SET_CPUID 661------------------ 662 663:Capability: basic 664:Architectures: x86 665:Type: vcpu ioctl 666:Parameters: struct kvm_cpuid (in) 667:Returns: 0 on success, -1 on error 668 669Defines the vcpu responses to the cpuid instruction. Applications 670should use the KVM_SET_CPUID2 ioctl if available. 671 672:: 673 674 struct kvm_cpuid_entry { 675 __u32 function; 676 __u32 eax; 677 __u32 ebx; 678 __u32 ecx; 679 __u32 edx; 680 __u32 padding; 681 }; 682 683 /* for KVM_SET_CPUID */ 684 struct kvm_cpuid { 685 __u32 nent; 686 __u32 padding; 687 struct kvm_cpuid_entry entries[0]; 688 }; 689 690 6914.21 KVM_SET_SIGNAL_MASK 692------------------------ 693 694:Capability: basic 695:Architectures: all 696:Type: vcpu ioctl 697:Parameters: struct kvm_signal_mask (in) 698:Returns: 0 on success, -1 on error 699 700Defines which signals are blocked during execution of KVM_RUN. This 701signal mask temporarily overrides the threads signal mask. Any 702unblocked signal received (except SIGKILL and SIGSTOP, which retain 703their traditional behaviour) will cause KVM_RUN to return with -EINTR. 704 705Note the signal will only be delivered if not blocked by the original 706signal mask. 707 708:: 709 710 /* for KVM_SET_SIGNAL_MASK */ 711 struct kvm_signal_mask { 712 __u32 len; 713 __u8 sigset[0]; 714 }; 715 716 7174.22 KVM_GET_FPU 718---------------- 719 720:Capability: basic 721:Architectures: x86 722:Type: vcpu ioctl 723:Parameters: struct kvm_fpu (out) 724:Returns: 0 on success, -1 on error 725 726Reads the floating point state from the vcpu. 727 728:: 729 730 /* for KVM_GET_FPU and KVM_SET_FPU */ 731 struct kvm_fpu { 732 __u8 fpr[8][16]; 733 __u16 fcw; 734 __u16 fsw; 735 __u8 ftwx; /* in fxsave format */ 736 __u8 pad1; 737 __u16 last_opcode; 738 __u64 last_ip; 739 __u64 last_dp; 740 __u8 xmm[16][16]; 741 __u32 mxcsr; 742 __u32 pad2; 743 }; 744 745 7464.23 KVM_SET_FPU 747---------------- 748 749:Capability: basic 750:Architectures: x86 751:Type: vcpu ioctl 752:Parameters: struct kvm_fpu (in) 753:Returns: 0 on success, -1 on error 754 755Writes the floating point state to the vcpu. 756 757:: 758 759 /* for KVM_GET_FPU and KVM_SET_FPU */ 760 struct kvm_fpu { 761 __u8 fpr[8][16]; 762 __u16 fcw; 763 __u16 fsw; 764 __u8 ftwx; /* in fxsave format */ 765 __u8 pad1; 766 __u16 last_opcode; 767 __u64 last_ip; 768 __u64 last_dp; 769 __u8 xmm[16][16]; 770 __u32 mxcsr; 771 __u32 pad2; 772 }; 773 774 7754.24 KVM_CREATE_IRQCHIP 776----------------------- 777 778:Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQCHIP (s390) 779:Architectures: x86, ARM, arm64, s390 780:Type: vm ioctl 781:Parameters: none 782:Returns: 0 on success, -1 on error 783 784Creates an interrupt controller model in the kernel. 785On x86, creates a virtual ioapic, a virtual PIC (two PICs, nested), and sets up 786future vcpus to have a local APIC. IRQ routing for GSIs 0-15 is set to both 787PIC and IOAPIC; GSI 16-23 only go to the IOAPIC. 788On ARM/arm64, a GICv2 is created. Any other GIC versions require the usage of 789KVM_CREATE_DEVICE, which also supports creating a GICv2. Using 790KVM_CREATE_DEVICE is preferred over KVM_CREATE_IRQCHIP for GICv2. 791On s390, a dummy irq routing table is created. 792 793Note that on s390 the KVM_CAP_S390_IRQCHIP vm capability needs to be enabled 794before KVM_CREATE_IRQCHIP can be used. 795 796 7974.25 KVM_IRQ_LINE 798----------------- 799 800:Capability: KVM_CAP_IRQCHIP 801:Architectures: x86, arm, arm64 802:Type: vm ioctl 803:Parameters: struct kvm_irq_level 804:Returns: 0 on success, -1 on error 805 806Sets the level of a GSI input to the interrupt controller model in the kernel. 807On some architectures it is required that an interrupt controller model has 808been previously created with KVM_CREATE_IRQCHIP. Note that edge-triggered 809interrupts require the level to be set to 1 and then back to 0. 810 811On real hardware, interrupt pins can be active-low or active-high. This 812does not matter for the level field of struct kvm_irq_level: 1 always 813means active (asserted), 0 means inactive (deasserted). 814 815x86 allows the operating system to program the interrupt polarity 816(active-low/active-high) for level-triggered interrupts, and KVM used 817to consider the polarity. However, due to bitrot in the handling of 818active-low interrupts, the above convention is now valid on x86 too. 819This is signaled by KVM_CAP_X86_IOAPIC_POLARITY_IGNORED. Userspace 820should not present interrupts to the guest as active-low unless this 821capability is present (or unless it is not using the in-kernel irqchip, 822of course). 823 824 825ARM/arm64 can signal an interrupt either at the CPU level, or at the 826in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to 827use PPIs designated for specific cpus. The irq field is interpreted 828like this:: 829 830 bits: | 31 ... 28 | 27 ... 24 | 23 ... 16 | 15 ... 0 | 831 field: | vcpu2_index | irq_type | vcpu_index | irq_id | 832 833The irq_type field has the following values: 834 835- irq_type[0]: 836 out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ 837- irq_type[1]: 838 in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.) 839 (the vcpu_index field is ignored) 840- irq_type[2]: 841 in-kernel GIC: PPI, irq_id between 16 and 31 (incl.) 842 843(The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs) 844 845In both cases, level is used to assert/deassert the line. 846 847When KVM_CAP_ARM_IRQ_LINE_LAYOUT_2 is supported, the target vcpu is 848identified as (256 * vcpu2_index + vcpu_index). Otherwise, vcpu2_index 849must be zero. 850 851Note that on arm/arm64, the KVM_CAP_IRQCHIP capability only conditions 852injection of interrupts for the in-kernel irqchip. KVM_IRQ_LINE can always 853be used for a userspace interrupt controller. 854 855:: 856 857 struct kvm_irq_level { 858 union { 859 __u32 irq; /* GSI */ 860 __s32 status; /* not used for KVM_IRQ_LEVEL */ 861 }; 862 __u32 level; /* 0 or 1 */ 863 }; 864 865 8664.26 KVM_GET_IRQCHIP 867-------------------- 868 869:Capability: KVM_CAP_IRQCHIP 870:Architectures: x86 871:Type: vm ioctl 872:Parameters: struct kvm_irqchip (in/out) 873:Returns: 0 on success, -1 on error 874 875Reads the state of a kernel interrupt controller created with 876KVM_CREATE_IRQCHIP into a buffer provided by the caller. 877 878:: 879 880 struct kvm_irqchip { 881 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */ 882 __u32 pad; 883 union { 884 char dummy[512]; /* reserving space */ 885 struct kvm_pic_state pic; 886 struct kvm_ioapic_state ioapic; 887 } chip; 888 }; 889 890 8914.27 KVM_SET_IRQCHIP 892-------------------- 893 894:Capability: KVM_CAP_IRQCHIP 895:Architectures: x86 896:Type: vm ioctl 897:Parameters: struct kvm_irqchip (in) 898:Returns: 0 on success, -1 on error 899 900Sets the state of a kernel interrupt controller created with 901KVM_CREATE_IRQCHIP from a buffer provided by the caller. 902 903:: 904 905 struct kvm_irqchip { 906 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */ 907 __u32 pad; 908 union { 909 char dummy[512]; /* reserving space */ 910 struct kvm_pic_state pic; 911 struct kvm_ioapic_state ioapic; 912 } chip; 913 }; 914 915 9164.28 KVM_XEN_HVM_CONFIG 917----------------------- 918 919:Capability: KVM_CAP_XEN_HVM 920:Architectures: x86 921:Type: vm ioctl 922:Parameters: struct kvm_xen_hvm_config (in) 923:Returns: 0 on success, -1 on error 924 925Sets the MSR that the Xen HVM guest uses to initialize its hypercall 926page, and provides the starting address and size of the hypercall 927blobs in userspace. When the guest writes the MSR, kvm copies one 928page of a blob (32- or 64-bit, depending on the vcpu mode) to guest 929memory. 930 931:: 932 933 struct kvm_xen_hvm_config { 934 __u32 flags; 935 __u32 msr; 936 __u64 blob_addr_32; 937 __u64 blob_addr_64; 938 __u8 blob_size_32; 939 __u8 blob_size_64; 940 __u8 pad2[30]; 941 }; 942 943 9444.29 KVM_GET_CLOCK 945------------------ 946 947:Capability: KVM_CAP_ADJUST_CLOCK 948:Architectures: x86 949:Type: vm ioctl 950:Parameters: struct kvm_clock_data (out) 951:Returns: 0 on success, -1 on error 952 953Gets the current timestamp of kvmclock as seen by the current guest. In 954conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios 955such as migration. 956 957When KVM_CAP_ADJUST_CLOCK is passed to KVM_CHECK_EXTENSION, it returns the 958set of bits that KVM can return in struct kvm_clock_data's flag member. 959 960The only flag defined now is KVM_CLOCK_TSC_STABLE. If set, the returned 961value is the exact kvmclock value seen by all VCPUs at the instant 962when KVM_GET_CLOCK was called. If clear, the returned value is simply 963CLOCK_MONOTONIC plus a constant offset; the offset can be modified 964with KVM_SET_CLOCK. KVM will try to make all VCPUs follow this clock, 965but the exact value read by each VCPU could differ, because the host 966TSC is not stable. 967 968:: 969 970 struct kvm_clock_data { 971 __u64 clock; /* kvmclock current value */ 972 __u32 flags; 973 __u32 pad[9]; 974 }; 975 976 9774.30 KVM_SET_CLOCK 978------------------ 979 980:Capability: KVM_CAP_ADJUST_CLOCK 981:Architectures: x86 982:Type: vm ioctl 983:Parameters: struct kvm_clock_data (in) 984:Returns: 0 on success, -1 on error 985 986Sets the current timestamp of kvmclock to the value specified in its parameter. 987In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios 988such as migration. 989 990:: 991 992 struct kvm_clock_data { 993 __u64 clock; /* kvmclock current value */ 994 __u32 flags; 995 __u32 pad[9]; 996 }; 997 998 9994.31 KVM_GET_VCPU_EVENTS 1000------------------------ 1001 1002:Capability: KVM_CAP_VCPU_EVENTS 1003:Extended by: KVM_CAP_INTR_SHADOW 1004:Architectures: x86, arm, arm64 1005:Type: vcpu ioctl 1006:Parameters: struct kvm_vcpu_event (out) 1007:Returns: 0 on success, -1 on error 1008 1009X86: 1010^^^^ 1011 1012Gets currently pending exceptions, interrupts, and NMIs as well as related 1013states of the vcpu. 1014 1015:: 1016 1017 struct kvm_vcpu_events { 1018 struct { 1019 __u8 injected; 1020 __u8 nr; 1021 __u8 has_error_code; 1022 __u8 pending; 1023 __u32 error_code; 1024 } exception; 1025 struct { 1026 __u8 injected; 1027 __u8 nr; 1028 __u8 soft; 1029 __u8 shadow; 1030 } interrupt; 1031 struct { 1032 __u8 injected; 1033 __u8 pending; 1034 __u8 masked; 1035 __u8 pad; 1036 } nmi; 1037 __u32 sipi_vector; 1038 __u32 flags; 1039 struct { 1040 __u8 smm; 1041 __u8 pending; 1042 __u8 smm_inside_nmi; 1043 __u8 latched_init; 1044 } smi; 1045 __u8 reserved[27]; 1046 __u8 exception_has_payload; 1047 __u64 exception_payload; 1048 }; 1049 1050The following bits are defined in the flags field: 1051 1052- KVM_VCPUEVENT_VALID_SHADOW may be set to signal that 1053 interrupt.shadow contains a valid state. 1054 1055- KVM_VCPUEVENT_VALID_SMM may be set to signal that smi contains a 1056 valid state. 1057 1058- KVM_VCPUEVENT_VALID_PAYLOAD may be set to signal that the 1059 exception_has_payload, exception_payload, and exception.pending 1060 fields contain a valid state. This bit will be set whenever 1061 KVM_CAP_EXCEPTION_PAYLOAD is enabled. 1062 1063ARM/ARM64: 1064^^^^^^^^^^ 1065 1066If the guest accesses a device that is being emulated by the host kernel in 1067such a way that a real device would generate a physical SError, KVM may make 1068a virtual SError pending for that VCPU. This system error interrupt remains 1069pending until the guest takes the exception by unmasking PSTATE.A. 1070 1071Running the VCPU may cause it to take a pending SError, or make an access that 1072causes an SError to become pending. The event's description is only valid while 1073the VPCU is not running. 1074 1075This API provides a way to read and write the pending 'event' state that is not 1076visible to the guest. To save, restore or migrate a VCPU the struct representing 1077the state can be read then written using this GET/SET API, along with the other 1078guest-visible registers. It is not possible to 'cancel' an SError that has been 1079made pending. 1080 1081A device being emulated in user-space may also wish to generate an SError. To do 1082this the events structure can be populated by user-space. The current state 1083should be read first, to ensure no existing SError is pending. If an existing 1084SError is pending, the architecture's 'Multiple SError interrupts' rules should 1085be followed. (2.5.3 of DDI0587.a "ARM Reliability, Availability, and 1086Serviceability (RAS) Specification"). 1087 1088SError exceptions always have an ESR value. Some CPUs have the ability to 1089specify what the virtual SError's ESR value should be. These systems will 1090advertise KVM_CAP_ARM_INJECT_SERROR_ESR. In this case exception.has_esr will 1091always have a non-zero value when read, and the agent making an SError pending 1092should specify the ISS field in the lower 24 bits of exception.serror_esr. If 1093the system supports KVM_CAP_ARM_INJECT_SERROR_ESR, but user-space sets the events 1094with exception.has_esr as zero, KVM will choose an ESR. 1095 1096Specifying exception.has_esr on a system that does not support it will return 1097-EINVAL. Setting anything other than the lower 24bits of exception.serror_esr 1098will return -EINVAL. 1099 1100It is not possible to read back a pending external abort (injected via 1101KVM_SET_VCPU_EVENTS or otherwise) because such an exception is always delivered 1102directly to the virtual CPU). 1103 1104:: 1105 1106 struct kvm_vcpu_events { 1107 struct { 1108 __u8 serror_pending; 1109 __u8 serror_has_esr; 1110 __u8 ext_dabt_pending; 1111 /* Align it to 8 bytes */ 1112 __u8 pad[5]; 1113 __u64 serror_esr; 1114 } exception; 1115 __u32 reserved[12]; 1116 }; 1117 11184.32 KVM_SET_VCPU_EVENTS 1119------------------------ 1120 1121:Capability: KVM_CAP_VCPU_EVENTS 1122:Extended by: KVM_CAP_INTR_SHADOW 1123:Architectures: x86, arm, arm64 1124:Type: vcpu ioctl 1125:Parameters: struct kvm_vcpu_event (in) 1126:Returns: 0 on success, -1 on error 1127 1128X86: 1129^^^^ 1130 1131Set pending exceptions, interrupts, and NMIs as well as related states of the 1132vcpu. 1133 1134See KVM_GET_VCPU_EVENTS for the data structure. 1135 1136Fields that may be modified asynchronously by running VCPUs can be excluded 1137from the update. These fields are nmi.pending, sipi_vector, smi.smm, 1138smi.pending. Keep the corresponding bits in the flags field cleared to 1139suppress overwriting the current in-kernel state. The bits are: 1140 1141=============================== ================================== 1142KVM_VCPUEVENT_VALID_NMI_PENDING transfer nmi.pending to the kernel 1143KVM_VCPUEVENT_VALID_SIPI_VECTOR transfer sipi_vector 1144KVM_VCPUEVENT_VALID_SMM transfer the smi sub-struct. 1145=============================== ================================== 1146 1147If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in 1148the flags field to signal that interrupt.shadow contains a valid state and 1149shall be written into the VCPU. 1150 1151KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available. 1152 1153If KVM_CAP_EXCEPTION_PAYLOAD is enabled, KVM_VCPUEVENT_VALID_PAYLOAD 1154can be set in the flags field to signal that the 1155exception_has_payload, exception_payload, and exception.pending fields 1156contain a valid state and shall be written into the VCPU. 1157 1158ARM/ARM64: 1159^^^^^^^^^^ 1160 1161User space may need to inject several types of events to the guest. 1162 1163Set the pending SError exception state for this VCPU. It is not possible to 1164'cancel' an Serror that has been made pending. 1165 1166If the guest performed an access to I/O memory which could not be handled by 1167userspace, for example because of missing instruction syndrome decode 1168information or because there is no device mapped at the accessed IPA, then 1169userspace can ask the kernel to inject an external abort using the address 1170from the exiting fault on the VCPU. It is a programming error to set 1171ext_dabt_pending after an exit which was not either KVM_EXIT_MMIO or 1172KVM_EXIT_ARM_NISV. This feature is only available if the system supports 1173KVM_CAP_ARM_INJECT_EXT_DABT. This is a helper which provides commonality in 1174how userspace reports accesses for the above cases to guests, across different 1175userspace implementations. Nevertheless, userspace can still emulate all Arm 1176exceptions by manipulating individual registers using the KVM_SET_ONE_REG API. 1177 1178See KVM_GET_VCPU_EVENTS for the data structure. 1179 1180 11814.33 KVM_GET_DEBUGREGS 1182---------------------- 1183 1184:Capability: KVM_CAP_DEBUGREGS 1185:Architectures: x86 1186:Type: vm ioctl 1187:Parameters: struct kvm_debugregs (out) 1188:Returns: 0 on success, -1 on error 1189 1190Reads debug registers from the vcpu. 1191 1192:: 1193 1194 struct kvm_debugregs { 1195 __u64 db[4]; 1196 __u64 dr6; 1197 __u64 dr7; 1198 __u64 flags; 1199 __u64 reserved[9]; 1200 }; 1201 1202 12034.34 KVM_SET_DEBUGREGS 1204---------------------- 1205 1206:Capability: KVM_CAP_DEBUGREGS 1207:Architectures: x86 1208:Type: vm ioctl 1209:Parameters: struct kvm_debugregs (in) 1210:Returns: 0 on success, -1 on error 1211 1212Writes debug registers into the vcpu. 1213 1214See KVM_GET_DEBUGREGS for the data structure. The flags field is unused 1215yet and must be cleared on entry. 1216 1217 12184.35 KVM_SET_USER_MEMORY_REGION 1219------------------------------- 1220 1221:Capability: KVM_CAP_USER_MEMORY 1222:Architectures: all 1223:Type: vm ioctl 1224:Parameters: struct kvm_userspace_memory_region (in) 1225:Returns: 0 on success, -1 on error 1226 1227:: 1228 1229 struct kvm_userspace_memory_region { 1230 __u32 slot; 1231 __u32 flags; 1232 __u64 guest_phys_addr; 1233 __u64 memory_size; /* bytes */ 1234 __u64 userspace_addr; /* start of the userspace allocated memory */ 1235 }; 1236 1237 /* for kvm_memory_region::flags */ 1238 #define KVM_MEM_LOG_DIRTY_PAGES (1UL << 0) 1239 #define KVM_MEM_READONLY (1UL << 1) 1240 1241This ioctl allows the user to create, modify or delete a guest physical 1242memory slot. Bits 0-15 of "slot" specify the slot id and this value 1243should be less than the maximum number of user memory slots supported per 1244VM. The maximum allowed slots can be queried using KVM_CAP_NR_MEMSLOTS. 1245Slots may not overlap in guest physical address space. 1246 1247If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot" 1248specifies the address space which is being modified. They must be 1249less than the value that KVM_CHECK_EXTENSION returns for the 1250KVM_CAP_MULTI_ADDRESS_SPACE capability. Slots in separate address spaces 1251are unrelated; the restriction on overlapping slots only applies within 1252each address space. 1253 1254Deleting a slot is done by passing zero for memory_size. When changing 1255an existing slot, it may be moved in the guest physical memory space, 1256or its flags may be modified, but it may not be resized. 1257 1258Memory for the region is taken starting at the address denoted by the 1259field userspace_addr, which must point at user addressable memory for 1260the entire memory slot size. Any object may back this memory, including 1261anonymous memory, ordinary files, and hugetlbfs. 1262 1263It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr 1264be identical. This allows large pages in the guest to be backed by large 1265pages in the host. 1266 1267The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and 1268KVM_MEM_READONLY. The former can be set to instruct KVM to keep track of 1269writes to memory within the slot. See KVM_GET_DIRTY_LOG ioctl to know how to 1270use it. The latter can be set, if KVM_CAP_READONLY_MEM capability allows it, 1271to make a new slot read-only. In this case, writes to this memory will be 1272posted to userspace as KVM_EXIT_MMIO exits. 1273 1274When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of 1275the memory region are automatically reflected into the guest. For example, an 1276mmap() that affects the region will be made visible immediately. Another 1277example is madvise(MADV_DROP). 1278 1279It is recommended to use this API instead of the KVM_SET_MEMORY_REGION ioctl. 1280The KVM_SET_MEMORY_REGION does not allow fine grained control over memory 1281allocation and is deprecated. 1282 1283 12844.36 KVM_SET_TSS_ADDR 1285--------------------- 1286 1287:Capability: KVM_CAP_SET_TSS_ADDR 1288:Architectures: x86 1289:Type: vm ioctl 1290:Parameters: unsigned long tss_address (in) 1291:Returns: 0 on success, -1 on error 1292 1293This ioctl defines the physical address of a three-page region in the guest 1294physical address space. The region must be within the first 4GB of the 1295guest physical address space and must not conflict with any memory slot 1296or any mmio address. The guest may malfunction if it accesses this memory 1297region. 1298 1299This ioctl is required on Intel-based hosts. This is needed on Intel hardware 1300because of a quirk in the virtualization implementation (see the internals 1301documentation when it pops into existence). 1302 1303 13044.37 KVM_ENABLE_CAP 1305------------------- 1306 1307:Capability: KVM_CAP_ENABLE_CAP 1308:Architectures: mips, ppc, s390 1309:Type: vcpu ioctl 1310:Parameters: struct kvm_enable_cap (in) 1311:Returns: 0 on success; -1 on error 1312 1313:Capability: KVM_CAP_ENABLE_CAP_VM 1314:Architectures: all 1315:Type: vcpu ioctl 1316:Parameters: struct kvm_enable_cap (in) 1317:Returns: 0 on success; -1 on error 1318 1319.. note:: 1320 1321 Not all extensions are enabled by default. Using this ioctl the application 1322 can enable an extension, making it available to the guest. 1323 1324On systems that do not support this ioctl, it always fails. On systems that 1325do support it, it only works for extensions that are supported for enablement. 1326 1327To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should 1328be used. 1329 1330:: 1331 1332 struct kvm_enable_cap { 1333 /* in */ 1334 __u32 cap; 1335 1336The capability that is supposed to get enabled. 1337 1338:: 1339 1340 __u32 flags; 1341 1342A bitfield indicating future enhancements. Has to be 0 for now. 1343 1344:: 1345 1346 __u64 args[4]; 1347 1348Arguments for enabling a feature. If a feature needs initial values to 1349function properly, this is the place to put them. 1350 1351:: 1352 1353 __u8 pad[64]; 1354 }; 1355 1356The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl 1357for vm-wide capabilities. 1358 13594.38 KVM_GET_MP_STATE 1360--------------------- 1361 1362:Capability: KVM_CAP_MP_STATE 1363:Architectures: x86, s390, arm, arm64 1364:Type: vcpu ioctl 1365:Parameters: struct kvm_mp_state (out) 1366:Returns: 0 on success; -1 on error 1367 1368:: 1369 1370 struct kvm_mp_state { 1371 __u32 mp_state; 1372 }; 1373 1374Returns the vcpu's current "multiprocessing state" (though also valid on 1375uniprocessor guests). 1376 1377Possible values are: 1378 1379 ========================== =============================================== 1380 KVM_MP_STATE_RUNNABLE the vcpu is currently running [x86,arm/arm64] 1381 KVM_MP_STATE_UNINITIALIZED the vcpu is an application processor (AP) 1382 which has not yet received an INIT signal [x86] 1383 KVM_MP_STATE_INIT_RECEIVED the vcpu has received an INIT signal, and is 1384 now ready for a SIPI [x86] 1385 KVM_MP_STATE_HALTED the vcpu has executed a HLT instruction and 1386 is waiting for an interrupt [x86] 1387 KVM_MP_STATE_SIPI_RECEIVED the vcpu has just received a SIPI (vector 1388 accessible via KVM_GET_VCPU_EVENTS) [x86] 1389 KVM_MP_STATE_STOPPED the vcpu is stopped [s390,arm/arm64] 1390 KVM_MP_STATE_CHECK_STOP the vcpu is in a special error state [s390] 1391 KVM_MP_STATE_OPERATING the vcpu is operating (running or halted) 1392 [s390] 1393 KVM_MP_STATE_LOAD the vcpu is in a special load/startup state 1394 [s390] 1395 ========================== =============================================== 1396 1397On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an 1398in-kernel irqchip, the multiprocessing state must be maintained by userspace on 1399these architectures. 1400 1401For arm/arm64: 1402^^^^^^^^^^^^^^ 1403 1404The only states that are valid are KVM_MP_STATE_STOPPED and 1405KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not. 1406 14074.39 KVM_SET_MP_STATE 1408--------------------- 1409 1410:Capability: KVM_CAP_MP_STATE 1411:Architectures: x86, s390, arm, arm64 1412:Type: vcpu ioctl 1413:Parameters: struct kvm_mp_state (in) 1414:Returns: 0 on success; -1 on error 1415 1416Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for 1417arguments. 1418 1419On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an 1420in-kernel irqchip, the multiprocessing state must be maintained by userspace on 1421these architectures. 1422 1423For arm/arm64: 1424^^^^^^^^^^^^^^ 1425 1426The only states that are valid are KVM_MP_STATE_STOPPED and 1427KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not. 1428 14294.40 KVM_SET_IDENTITY_MAP_ADDR 1430------------------------------ 1431 1432:Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR 1433:Architectures: x86 1434:Type: vm ioctl 1435:Parameters: unsigned long identity (in) 1436:Returns: 0 on success, -1 on error 1437 1438This ioctl defines the physical address of a one-page region in the guest 1439physical address space. The region must be within the first 4GB of the 1440guest physical address space and must not conflict with any memory slot 1441or any mmio address. The guest may malfunction if it accesses this memory 1442region. 1443 1444Setting the address to 0 will result in resetting the address to its default 1445(0xfffbc000). 1446 1447This ioctl is required on Intel-based hosts. This is needed on Intel hardware 1448because of a quirk in the virtualization implementation (see the internals 1449documentation when it pops into existence). 1450 1451Fails if any VCPU has already been created. 1452 14534.41 KVM_SET_BOOT_CPU_ID 1454------------------------ 1455 1456:Capability: KVM_CAP_SET_BOOT_CPU_ID 1457:Architectures: x86 1458:Type: vm ioctl 1459:Parameters: unsigned long vcpu_id 1460:Returns: 0 on success, -1 on error 1461 1462Define which vcpu is the Bootstrap Processor (BSP). Values are the same 1463as the vcpu id in KVM_CREATE_VCPU. If this ioctl is not called, the default 1464is vcpu 0. 1465 1466 14674.42 KVM_GET_XSAVE 1468------------------ 1469 1470:Capability: KVM_CAP_XSAVE 1471:Architectures: x86 1472:Type: vcpu ioctl 1473:Parameters: struct kvm_xsave (out) 1474:Returns: 0 on success, -1 on error 1475 1476 1477:: 1478 1479 struct kvm_xsave { 1480 __u32 region[1024]; 1481 }; 1482 1483This ioctl would copy current vcpu's xsave struct to the userspace. 1484 1485 14864.43 KVM_SET_XSAVE 1487------------------ 1488 1489:Capability: KVM_CAP_XSAVE 1490:Architectures: x86 1491:Type: vcpu ioctl 1492:Parameters: struct kvm_xsave (in) 1493:Returns: 0 on success, -1 on error 1494 1495:: 1496 1497 1498 struct kvm_xsave { 1499 __u32 region[1024]; 1500 }; 1501 1502This ioctl would copy userspace's xsave struct to the kernel. 1503 1504 15054.44 KVM_GET_XCRS 1506----------------- 1507 1508:Capability: KVM_CAP_XCRS 1509:Architectures: x86 1510:Type: vcpu ioctl 1511:Parameters: struct kvm_xcrs (out) 1512:Returns: 0 on success, -1 on error 1513 1514:: 1515 1516 struct kvm_xcr { 1517 __u32 xcr; 1518 __u32 reserved; 1519 __u64 value; 1520 }; 1521 1522 struct kvm_xcrs { 1523 __u32 nr_xcrs; 1524 __u32 flags; 1525 struct kvm_xcr xcrs[KVM_MAX_XCRS]; 1526 __u64 padding[16]; 1527 }; 1528 1529This ioctl would copy current vcpu's xcrs to the userspace. 1530 1531 15324.45 KVM_SET_XCRS 1533----------------- 1534 1535:Capability: KVM_CAP_XCRS 1536:Architectures: x86 1537:Type: vcpu ioctl 1538:Parameters: struct kvm_xcrs (in) 1539:Returns: 0 on success, -1 on error 1540 1541:: 1542 1543 struct kvm_xcr { 1544 __u32 xcr; 1545 __u32 reserved; 1546 __u64 value; 1547 }; 1548 1549 struct kvm_xcrs { 1550 __u32 nr_xcrs; 1551 __u32 flags; 1552 struct kvm_xcr xcrs[KVM_MAX_XCRS]; 1553 __u64 padding[16]; 1554 }; 1555 1556This ioctl would set vcpu's xcr to the value userspace specified. 1557 1558 15594.46 KVM_GET_SUPPORTED_CPUID 1560---------------------------- 1561 1562:Capability: KVM_CAP_EXT_CPUID 1563:Architectures: x86 1564:Type: system ioctl 1565:Parameters: struct kvm_cpuid2 (in/out) 1566:Returns: 0 on success, -1 on error 1567 1568:: 1569 1570 struct kvm_cpuid2 { 1571 __u32 nent; 1572 __u32 padding; 1573 struct kvm_cpuid_entry2 entries[0]; 1574 }; 1575 1576 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0) 1577 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */ 1578 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */ 1579 1580 struct kvm_cpuid_entry2 { 1581 __u32 function; 1582 __u32 index; 1583 __u32 flags; 1584 __u32 eax; 1585 __u32 ebx; 1586 __u32 ecx; 1587 __u32 edx; 1588 __u32 padding[3]; 1589 }; 1590 1591This ioctl returns x86 cpuid features which are supported by both the 1592hardware and kvm in its default configuration. Userspace can use the 1593information returned by this ioctl to construct cpuid information (for 1594KVM_SET_CPUID2) that is consistent with hardware, kernel, and 1595userspace capabilities, and with user requirements (for example, the 1596user may wish to constrain cpuid to emulate older hardware, or for 1597feature consistency across a cluster). 1598 1599Note that certain capabilities, such as KVM_CAP_X86_DISABLE_EXITS, may 1600expose cpuid features (e.g. MONITOR) which are not supported by kvm in 1601its default configuration. If userspace enables such capabilities, it 1602is responsible for modifying the results of this ioctl appropriately. 1603 1604Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure 1605with the 'nent' field indicating the number of entries in the variable-size 1606array 'entries'. If the number of entries is too low to describe the cpu 1607capabilities, an error (E2BIG) is returned. If the number is too high, 1608the 'nent' field is adjusted and an error (ENOMEM) is returned. If the 1609number is just right, the 'nent' field is adjusted to the number of valid 1610entries in the 'entries' array, which is then filled. 1611 1612The entries returned are the host cpuid as returned by the cpuid instruction, 1613with unknown or unsupported features masked out. Some features (for example, 1614x2apic), may not be present in the host cpu, but are exposed by kvm if it can 1615emulate them efficiently. The fields in each entry are defined as follows: 1616 1617 function: 1618 the eax value used to obtain the entry 1619 1620 index: 1621 the ecx value used to obtain the entry (for entries that are 1622 affected by ecx) 1623 1624 flags: 1625 an OR of zero or more of the following: 1626 1627 KVM_CPUID_FLAG_SIGNIFCANT_INDEX: 1628 if the index field is valid 1629 1630 eax, ebx, ecx, edx: 1631 the values returned by the cpuid instruction for 1632 this function/index combination 1633 1634The TSC deadline timer feature (CPUID leaf 1, ecx[24]) is always returned 1635as false, since the feature depends on KVM_CREATE_IRQCHIP for local APIC 1636support. Instead it is reported via:: 1637 1638 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER) 1639 1640if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the 1641feature in userspace, then you can enable the feature for KVM_SET_CPUID2. 1642 1643 16444.47 KVM_PPC_GET_PVINFO 1645----------------------- 1646 1647:Capability: KVM_CAP_PPC_GET_PVINFO 1648:Architectures: ppc 1649:Type: vm ioctl 1650:Parameters: struct kvm_ppc_pvinfo (out) 1651:Returns: 0 on success, !0 on error 1652 1653:: 1654 1655 struct kvm_ppc_pvinfo { 1656 __u32 flags; 1657 __u32 hcall[4]; 1658 __u8 pad[108]; 1659 }; 1660 1661This ioctl fetches PV specific information that need to be passed to the guest 1662using the device tree or other means from vm context. 1663 1664The hcall array defines 4 instructions that make up a hypercall. 1665 1666If any additional field gets added to this structure later on, a bit for that 1667additional piece of information will be set in the flags bitmap. 1668 1669The flags bitmap is defined as:: 1670 1671 /* the host supports the ePAPR idle hcall 1672 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1<<0) 1673 16744.52 KVM_SET_GSI_ROUTING 1675------------------------ 1676 1677:Capability: KVM_CAP_IRQ_ROUTING 1678:Architectures: x86 s390 arm arm64 1679:Type: vm ioctl 1680:Parameters: struct kvm_irq_routing (in) 1681:Returns: 0 on success, -1 on error 1682 1683Sets the GSI routing table entries, overwriting any previously set entries. 1684 1685On arm/arm64, GSI routing has the following limitation: 1686 1687- GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD. 1688 1689:: 1690 1691 struct kvm_irq_routing { 1692 __u32 nr; 1693 __u32 flags; 1694 struct kvm_irq_routing_entry entries[0]; 1695 }; 1696 1697No flags are specified so far, the corresponding field must be set to zero. 1698 1699:: 1700 1701 struct kvm_irq_routing_entry { 1702 __u32 gsi; 1703 __u32 type; 1704 __u32 flags; 1705 __u32 pad; 1706 union { 1707 struct kvm_irq_routing_irqchip irqchip; 1708 struct kvm_irq_routing_msi msi; 1709 struct kvm_irq_routing_s390_adapter adapter; 1710 struct kvm_irq_routing_hv_sint hv_sint; 1711 __u32 pad[8]; 1712 } u; 1713 }; 1714 1715 /* gsi routing entry types */ 1716 #define KVM_IRQ_ROUTING_IRQCHIP 1 1717 #define KVM_IRQ_ROUTING_MSI 2 1718 #define KVM_IRQ_ROUTING_S390_ADAPTER 3 1719 #define KVM_IRQ_ROUTING_HV_SINT 4 1720 1721flags: 1722 1723- KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry 1724 type, specifies that the devid field contains a valid value. The per-VM 1725 KVM_CAP_MSI_DEVID capability advertises the requirement to provide 1726 the device ID. If this capability is not available, userspace should 1727 never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail. 1728- zero otherwise 1729 1730:: 1731 1732 struct kvm_irq_routing_irqchip { 1733 __u32 irqchip; 1734 __u32 pin; 1735 }; 1736 1737 struct kvm_irq_routing_msi { 1738 __u32 address_lo; 1739 __u32 address_hi; 1740 __u32 data; 1741 union { 1742 __u32 pad; 1743 __u32 devid; 1744 }; 1745 }; 1746 1747If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier 1748for the device that wrote the MSI message. For PCI, this is usually a 1749BFD identifier in the lower 16 bits. 1750 1751On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS 1752feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled, 1753address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of 1754address_hi must be zero. 1755 1756:: 1757 1758 struct kvm_irq_routing_s390_adapter { 1759 __u64 ind_addr; 1760 __u64 summary_addr; 1761 __u64 ind_offset; 1762 __u32 summary_offset; 1763 __u32 adapter_id; 1764 }; 1765 1766 struct kvm_irq_routing_hv_sint { 1767 __u32 vcpu; 1768 __u32 sint; 1769 }; 1770 1771 17724.55 KVM_SET_TSC_KHZ 1773-------------------- 1774 1775:Capability: KVM_CAP_TSC_CONTROL 1776:Architectures: x86 1777:Type: vcpu ioctl 1778:Parameters: virtual tsc_khz 1779:Returns: 0 on success, -1 on error 1780 1781Specifies the tsc frequency for the virtual machine. The unit of the 1782frequency is KHz. 1783 1784 17854.56 KVM_GET_TSC_KHZ 1786-------------------- 1787 1788:Capability: KVM_CAP_GET_TSC_KHZ 1789:Architectures: x86 1790:Type: vcpu ioctl 1791:Parameters: none 1792:Returns: virtual tsc-khz on success, negative value on error 1793 1794Returns the tsc frequency of the guest. The unit of the return value is 1795KHz. If the host has unstable tsc this ioctl returns -EIO instead as an 1796error. 1797 1798 17994.57 KVM_GET_LAPIC 1800------------------ 1801 1802:Capability: KVM_CAP_IRQCHIP 1803:Architectures: x86 1804:Type: vcpu ioctl 1805:Parameters: struct kvm_lapic_state (out) 1806:Returns: 0 on success, -1 on error 1807 1808:: 1809 1810 #define KVM_APIC_REG_SIZE 0x400 1811 struct kvm_lapic_state { 1812 char regs[KVM_APIC_REG_SIZE]; 1813 }; 1814 1815Reads the Local APIC registers and copies them into the input argument. The 1816data format and layout are the same as documented in the architecture manual. 1817 1818If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is 1819enabled, then the format of APIC_ID register depends on the APIC mode 1820(reported by MSR_IA32_APICBASE) of its VCPU. x2APIC stores APIC ID in 1821the APIC_ID register (bytes 32-35). xAPIC only allows an 8-bit APIC ID 1822which is stored in bits 31-24 of the APIC register, or equivalently in 1823byte 35 of struct kvm_lapic_state's regs field. KVM_GET_LAPIC must then 1824be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR. 1825 1826If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state 1827always uses xAPIC format. 1828 1829 18304.58 KVM_SET_LAPIC 1831------------------ 1832 1833:Capability: KVM_CAP_IRQCHIP 1834:Architectures: x86 1835:Type: vcpu ioctl 1836:Parameters: struct kvm_lapic_state (in) 1837:Returns: 0 on success, -1 on error 1838 1839:: 1840 1841 #define KVM_APIC_REG_SIZE 0x400 1842 struct kvm_lapic_state { 1843 char regs[KVM_APIC_REG_SIZE]; 1844 }; 1845 1846Copies the input argument into the Local APIC registers. The data format 1847and layout are the same as documented in the architecture manual. 1848 1849The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's 1850regs field) depends on the state of the KVM_CAP_X2APIC_API capability. 1851See the note in KVM_GET_LAPIC. 1852 1853 18544.59 KVM_IOEVENTFD 1855------------------ 1856 1857:Capability: KVM_CAP_IOEVENTFD 1858:Architectures: all 1859:Type: vm ioctl 1860:Parameters: struct kvm_ioeventfd (in) 1861:Returns: 0 on success, !0 on error 1862 1863This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address 1864within the guest. A guest write in the registered address will signal the 1865provided event instead of triggering an exit. 1866 1867:: 1868 1869 struct kvm_ioeventfd { 1870 __u64 datamatch; 1871 __u64 addr; /* legal pio/mmio address */ 1872 __u32 len; /* 0, 1, 2, 4, or 8 bytes */ 1873 __s32 fd; 1874 __u32 flags; 1875 __u8 pad[36]; 1876 }; 1877 1878For the special case of virtio-ccw devices on s390, the ioevent is matched 1879to a subchannel/virtqueue tuple instead. 1880 1881The following flags are defined:: 1882 1883 #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch) 1884 #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio) 1885 #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign) 1886 #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \ 1887 (1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify) 1888 1889If datamatch flag is set, the event will be signaled only if the written value 1890to the registered address is equal to datamatch in struct kvm_ioeventfd. 1891 1892For virtio-ccw devices, addr contains the subchannel id and datamatch the 1893virtqueue index. 1894 1895With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and 1896the kernel will ignore the length of guest write and may get a faster vmexit. 1897The speedup may only apply to specific architectures, but the ioeventfd will 1898work anyway. 1899 19004.60 KVM_DIRTY_TLB 1901------------------ 1902 1903:Capability: KVM_CAP_SW_TLB 1904:Architectures: ppc 1905:Type: vcpu ioctl 1906:Parameters: struct kvm_dirty_tlb (in) 1907:Returns: 0 on success, -1 on error 1908 1909:: 1910 1911 struct kvm_dirty_tlb { 1912 __u64 bitmap; 1913 __u32 num_dirty; 1914 }; 1915 1916This must be called whenever userspace has changed an entry in the shared 1917TLB, prior to calling KVM_RUN on the associated vcpu. 1918 1919The "bitmap" field is the userspace address of an array. This array 1920consists of a number of bits, equal to the total number of TLB entries as 1921determined by the last successful call to KVM_CONFIG_TLB, rounded up to the 1922nearest multiple of 64. 1923 1924Each bit corresponds to one TLB entry, ordered the same as in the shared TLB 1925array. 1926 1927The array is little-endian: the bit 0 is the least significant bit of the 1928first byte, bit 8 is the least significant bit of the second byte, etc. 1929This avoids any complications with differing word sizes. 1930 1931The "num_dirty" field is a performance hint for KVM to determine whether it 1932should skip processing the bitmap and just invalidate everything. It must 1933be set to the number of set bits in the bitmap. 1934 1935 19364.62 KVM_CREATE_SPAPR_TCE 1937------------------------- 1938 1939:Capability: KVM_CAP_SPAPR_TCE 1940:Architectures: powerpc 1941:Type: vm ioctl 1942:Parameters: struct kvm_create_spapr_tce (in) 1943:Returns: file descriptor for manipulating the created TCE table 1944 1945This creates a virtual TCE (translation control entry) table, which 1946is an IOMMU for PAPR-style virtual I/O. It is used to translate 1947logical addresses used in virtual I/O into guest physical addresses, 1948and provides a scatter/gather capability for PAPR virtual I/O. 1949 1950:: 1951 1952 /* for KVM_CAP_SPAPR_TCE */ 1953 struct kvm_create_spapr_tce { 1954 __u64 liobn; 1955 __u32 window_size; 1956 }; 1957 1958The liobn field gives the logical IO bus number for which to create a 1959TCE table. The window_size field specifies the size of the DMA window 1960which this TCE table will translate - the table will contain one 64 1961bit TCE entry for every 4kiB of the DMA window. 1962 1963When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE 1964table has been created using this ioctl(), the kernel will handle it 1965in real mode, updating the TCE table. H_PUT_TCE calls for other 1966liobns will cause a vm exit and must be handled by userspace. 1967 1968The return value is a file descriptor which can be passed to mmap(2) 1969to map the created TCE table into userspace. This lets userspace read 1970the entries written by kernel-handled H_PUT_TCE calls, and also lets 1971userspace update the TCE table directly which is useful in some 1972circumstances. 1973 1974 19754.63 KVM_ALLOCATE_RMA 1976--------------------- 1977 1978:Capability: KVM_CAP_PPC_RMA 1979:Architectures: powerpc 1980:Type: vm ioctl 1981:Parameters: struct kvm_allocate_rma (out) 1982:Returns: file descriptor for mapping the allocated RMA 1983 1984This allocates a Real Mode Area (RMA) from the pool allocated at boot 1985time by the kernel. An RMA is a physically-contiguous, aligned region 1986of memory used on older POWER processors to provide the memory which 1987will be accessed by real-mode (MMU off) accesses in a KVM guest. 1988POWER processors support a set of sizes for the RMA that usually 1989includes 64MB, 128MB, 256MB and some larger powers of two. 1990 1991:: 1992 1993 /* for KVM_ALLOCATE_RMA */ 1994 struct kvm_allocate_rma { 1995 __u64 rma_size; 1996 }; 1997 1998The return value is a file descriptor which can be passed to mmap(2) 1999to map the allocated RMA into userspace. The mapped area can then be 2000passed to the KVM_SET_USER_MEMORY_REGION ioctl to establish it as the 2001RMA for a virtual machine. The size of the RMA in bytes (which is 2002fixed at host kernel boot time) is returned in the rma_size field of 2003the argument structure. 2004 2005The KVM_CAP_PPC_RMA capability is 1 or 2 if the KVM_ALLOCATE_RMA ioctl 2006is supported; 2 if the processor requires all virtual machines to have 2007an RMA, or 1 if the processor can use an RMA but doesn't require it, 2008because it supports the Virtual RMA (VRMA) facility. 2009 2010 20114.64 KVM_NMI 2012------------ 2013 2014:Capability: KVM_CAP_USER_NMI 2015:Architectures: x86 2016:Type: vcpu ioctl 2017:Parameters: none 2018:Returns: 0 on success, -1 on error 2019 2020Queues an NMI on the thread's vcpu. Note this is well defined only 2021when KVM_CREATE_IRQCHIP has not been called, since this is an interface 2022between the virtual cpu core and virtual local APIC. After KVM_CREATE_IRQCHIP 2023has been called, this interface is completely emulated within the kernel. 2024 2025To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the 2026following algorithm: 2027 2028 - pause the vcpu 2029 - read the local APIC's state (KVM_GET_LAPIC) 2030 - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1) 2031 - if so, issue KVM_NMI 2032 - resume the vcpu 2033 2034Some guests configure the LINT1 NMI input to cause a panic, aiding in 2035debugging. 2036 2037 20384.65 KVM_S390_UCAS_MAP 2039---------------------- 2040 2041:Capability: KVM_CAP_S390_UCONTROL 2042:Architectures: s390 2043:Type: vcpu ioctl 2044:Parameters: struct kvm_s390_ucas_mapping (in) 2045:Returns: 0 in case of success 2046 2047The parameter is defined like this:: 2048 2049 struct kvm_s390_ucas_mapping { 2050 __u64 user_addr; 2051 __u64 vcpu_addr; 2052 __u64 length; 2053 }; 2054 2055This ioctl maps the memory at "user_addr" with the length "length" to 2056the vcpu's address space starting at "vcpu_addr". All parameters need to 2057be aligned by 1 megabyte. 2058 2059 20604.66 KVM_S390_UCAS_UNMAP 2061------------------------ 2062 2063:Capability: KVM_CAP_S390_UCONTROL 2064:Architectures: s390 2065:Type: vcpu ioctl 2066:Parameters: struct kvm_s390_ucas_mapping (in) 2067:Returns: 0 in case of success 2068 2069The parameter is defined like this:: 2070 2071 struct kvm_s390_ucas_mapping { 2072 __u64 user_addr; 2073 __u64 vcpu_addr; 2074 __u64 length; 2075 }; 2076 2077This ioctl unmaps the memory in the vcpu's address space starting at 2078"vcpu_addr" with the length "length". The field "user_addr" is ignored. 2079All parameters need to be aligned by 1 megabyte. 2080 2081 20824.67 KVM_S390_VCPU_FAULT 2083------------------------ 2084 2085:Capability: KVM_CAP_S390_UCONTROL 2086:Architectures: s390 2087:Type: vcpu ioctl 2088:Parameters: vcpu absolute address (in) 2089:Returns: 0 in case of success 2090 2091This call creates a page table entry on the virtual cpu's address space 2092(for user controlled virtual machines) or the virtual machine's address 2093space (for regular virtual machines). This only works for minor faults, 2094thus it's recommended to access subject memory page via the user page 2095table upfront. This is useful to handle validity intercepts for user 2096controlled virtual machines to fault in the virtual cpu's lowcore pages 2097prior to calling the KVM_RUN ioctl. 2098 2099 21004.68 KVM_SET_ONE_REG 2101-------------------- 2102 2103:Capability: KVM_CAP_ONE_REG 2104:Architectures: all 2105:Type: vcpu ioctl 2106:Parameters: struct kvm_one_reg (in) 2107:Returns: 0 on success, negative value on failure 2108 2109Errors: 2110 2111 ====== ============================================================ 2112 ENOENT no such register 2113 EINVAL invalid register ID, or no such register or used with VMs in 2114 protected virtualization mode on s390 2115 EPERM (arm64) register access not allowed before vcpu finalization 2116 ====== ============================================================ 2117 2118(These error codes are indicative only: do not rely on a specific error 2119code being returned in a specific situation.) 2120 2121:: 2122 2123 struct kvm_one_reg { 2124 __u64 id; 2125 __u64 addr; 2126 }; 2127 2128Using this ioctl, a single vcpu register can be set to a specific value 2129defined by user space with the passed in struct kvm_one_reg, where id 2130refers to the register identifier as described below and addr is a pointer 2131to a variable with the respective size. There can be architecture agnostic 2132and architecture specific registers. Each have their own range of operation 2133and their own constants and width. To keep track of the implemented 2134registers, find a list below: 2135 2136 ======= =============================== ============ 2137 Arch Register Width (bits) 2138 ======= =============================== ============ 2139 PPC KVM_REG_PPC_HIOR 64 2140 PPC KVM_REG_PPC_IAC1 64 2141 PPC KVM_REG_PPC_IAC2 64 2142 PPC KVM_REG_PPC_IAC3 64 2143 PPC KVM_REG_PPC_IAC4 64 2144 PPC KVM_REG_PPC_DAC1 64 2145 PPC KVM_REG_PPC_DAC2 64 2146 PPC KVM_REG_PPC_DABR 64 2147 PPC KVM_REG_PPC_DSCR 64 2148 PPC KVM_REG_PPC_PURR 64 2149 PPC KVM_REG_PPC_SPURR 64 2150 PPC KVM_REG_PPC_DAR 64 2151 PPC KVM_REG_PPC_DSISR 32 2152 PPC KVM_REG_PPC_AMR 64 2153 PPC KVM_REG_PPC_UAMOR 64 2154 PPC KVM_REG_PPC_MMCR0 64 2155 PPC KVM_REG_PPC_MMCR1 64 2156 PPC KVM_REG_PPC_MMCRA 64 2157 PPC KVM_REG_PPC_MMCR2 64 2158 PPC KVM_REG_PPC_MMCRS 64 2159 PPC KVM_REG_PPC_SIAR 64 2160 PPC KVM_REG_PPC_SDAR 64 2161 PPC KVM_REG_PPC_SIER 64 2162 PPC KVM_REG_PPC_PMC1 32 2163 PPC KVM_REG_PPC_PMC2 32 2164 PPC KVM_REG_PPC_PMC3 32 2165 PPC KVM_REG_PPC_PMC4 32 2166 PPC KVM_REG_PPC_PMC5 32 2167 PPC KVM_REG_PPC_PMC6 32 2168 PPC KVM_REG_PPC_PMC7 32 2169 PPC KVM_REG_PPC_PMC8 32 2170 PPC KVM_REG_PPC_FPR0 64 2171 ... 2172 PPC KVM_REG_PPC_FPR31 64 2173 PPC KVM_REG_PPC_VR0 128 2174 ... 2175 PPC KVM_REG_PPC_VR31 128 2176 PPC KVM_REG_PPC_VSR0 128 2177 ... 2178 PPC KVM_REG_PPC_VSR31 128 2179 PPC KVM_REG_PPC_FPSCR 64 2180 PPC KVM_REG_PPC_VSCR 32 2181 PPC KVM_REG_PPC_VPA_ADDR 64 2182 PPC KVM_REG_PPC_VPA_SLB 128 2183 PPC KVM_REG_PPC_VPA_DTL 128 2184 PPC KVM_REG_PPC_EPCR 32 2185 PPC KVM_REG_PPC_EPR 32 2186 PPC KVM_REG_PPC_TCR 32 2187 PPC KVM_REG_PPC_TSR 32 2188 PPC KVM_REG_PPC_OR_TSR 32 2189 PPC KVM_REG_PPC_CLEAR_TSR 32 2190 PPC KVM_REG_PPC_MAS0 32 2191 PPC KVM_REG_PPC_MAS1 32 2192 PPC KVM_REG_PPC_MAS2 64 2193 PPC KVM_REG_PPC_MAS7_3 64 2194 PPC KVM_REG_PPC_MAS4 32 2195 PPC KVM_REG_PPC_MAS6 32 2196 PPC KVM_REG_PPC_MMUCFG 32 2197 PPC KVM_REG_PPC_TLB0CFG 32 2198 PPC KVM_REG_PPC_TLB1CFG 32 2199 PPC KVM_REG_PPC_TLB2CFG 32 2200 PPC KVM_REG_PPC_TLB3CFG 32 2201 PPC KVM_REG_PPC_TLB0PS 32 2202 PPC KVM_REG_PPC_TLB1PS 32 2203 PPC KVM_REG_PPC_TLB2PS 32 2204 PPC KVM_REG_PPC_TLB3PS 32 2205 PPC KVM_REG_PPC_EPTCFG 32 2206 PPC KVM_REG_PPC_ICP_STATE 64 2207 PPC KVM_REG_PPC_VP_STATE 128 2208 PPC KVM_REG_PPC_TB_OFFSET 64 2209 PPC KVM_REG_PPC_SPMC1 32 2210 PPC KVM_REG_PPC_SPMC2 32 2211 PPC KVM_REG_PPC_IAMR 64 2212 PPC KVM_REG_PPC_TFHAR 64 2213 PPC KVM_REG_PPC_TFIAR 64 2214 PPC KVM_REG_PPC_TEXASR 64 2215 PPC KVM_REG_PPC_FSCR 64 2216 PPC KVM_REG_PPC_PSPB 32 2217 PPC KVM_REG_PPC_EBBHR 64 2218 PPC KVM_REG_PPC_EBBRR 64 2219 PPC KVM_REG_PPC_BESCR 64 2220 PPC KVM_REG_PPC_TAR 64 2221 PPC KVM_REG_PPC_DPDES 64 2222 PPC KVM_REG_PPC_DAWR 64 2223 PPC KVM_REG_PPC_DAWRX 64 2224 PPC KVM_REG_PPC_CIABR 64 2225 PPC KVM_REG_PPC_IC 64 2226 PPC KVM_REG_PPC_VTB 64 2227 PPC KVM_REG_PPC_CSIGR 64 2228 PPC KVM_REG_PPC_TACR 64 2229 PPC KVM_REG_PPC_TCSCR 64 2230 PPC KVM_REG_PPC_PID 64 2231 PPC KVM_REG_PPC_ACOP 64 2232 PPC KVM_REG_PPC_VRSAVE 32 2233 PPC KVM_REG_PPC_LPCR 32 2234 PPC KVM_REG_PPC_LPCR_64 64 2235 PPC KVM_REG_PPC_PPR 64 2236 PPC KVM_REG_PPC_ARCH_COMPAT 32 2237 PPC KVM_REG_PPC_DABRX 32 2238 PPC KVM_REG_PPC_WORT 64 2239 PPC KVM_REG_PPC_SPRG9 64 2240 PPC KVM_REG_PPC_DBSR 32 2241 PPC KVM_REG_PPC_TIDR 64 2242 PPC KVM_REG_PPC_PSSCR 64 2243 PPC KVM_REG_PPC_DEC_EXPIRY 64 2244 PPC KVM_REG_PPC_PTCR 64 2245 PPC KVM_REG_PPC_TM_GPR0 64 2246 ... 2247 PPC KVM_REG_PPC_TM_GPR31 64 2248 PPC KVM_REG_PPC_TM_VSR0 128 2249 ... 2250 PPC KVM_REG_PPC_TM_VSR63 128 2251 PPC KVM_REG_PPC_TM_CR 64 2252 PPC KVM_REG_PPC_TM_LR 64 2253 PPC KVM_REG_PPC_TM_CTR 64 2254 PPC KVM_REG_PPC_TM_FPSCR 64 2255 PPC KVM_REG_PPC_TM_AMR 64 2256 PPC KVM_REG_PPC_TM_PPR 64 2257 PPC KVM_REG_PPC_TM_VRSAVE 64 2258 PPC KVM_REG_PPC_TM_VSCR 32 2259 PPC KVM_REG_PPC_TM_DSCR 64 2260 PPC KVM_REG_PPC_TM_TAR 64 2261 PPC KVM_REG_PPC_TM_XER 64 2262 2263 MIPS KVM_REG_MIPS_R0 64 2264 ... 2265 MIPS KVM_REG_MIPS_R31 64 2266 MIPS KVM_REG_MIPS_HI 64 2267 MIPS KVM_REG_MIPS_LO 64 2268 MIPS KVM_REG_MIPS_PC 64 2269 MIPS KVM_REG_MIPS_CP0_INDEX 32 2270 MIPS KVM_REG_MIPS_CP0_ENTRYLO0 64 2271 MIPS KVM_REG_MIPS_CP0_ENTRYLO1 64 2272 MIPS KVM_REG_MIPS_CP0_CONTEXT 64 2273 MIPS KVM_REG_MIPS_CP0_CONTEXTCONFIG 32 2274 MIPS KVM_REG_MIPS_CP0_USERLOCAL 64 2275 MIPS KVM_REG_MIPS_CP0_XCONTEXTCONFIG 64 2276 MIPS KVM_REG_MIPS_CP0_PAGEMASK 32 2277 MIPS KVM_REG_MIPS_CP0_PAGEGRAIN 32 2278 MIPS KVM_REG_MIPS_CP0_SEGCTL0 64 2279 MIPS KVM_REG_MIPS_CP0_SEGCTL1 64 2280 MIPS KVM_REG_MIPS_CP0_SEGCTL2 64 2281 MIPS KVM_REG_MIPS_CP0_PWBASE 64 2282 MIPS KVM_REG_MIPS_CP0_PWFIELD 64 2283 MIPS KVM_REG_MIPS_CP0_PWSIZE 64 2284 MIPS KVM_REG_MIPS_CP0_WIRED 32 2285 MIPS KVM_REG_MIPS_CP0_PWCTL 32 2286 MIPS KVM_REG_MIPS_CP0_HWRENA 32 2287 MIPS KVM_REG_MIPS_CP0_BADVADDR 64 2288 MIPS KVM_REG_MIPS_CP0_BADINSTR 32 2289 MIPS KVM_REG_MIPS_CP0_BADINSTRP 32 2290 MIPS KVM_REG_MIPS_CP0_COUNT 32 2291 MIPS KVM_REG_MIPS_CP0_ENTRYHI 64 2292 MIPS KVM_REG_MIPS_CP0_COMPARE 32 2293 MIPS KVM_REG_MIPS_CP0_STATUS 32 2294 MIPS KVM_REG_MIPS_CP0_INTCTL 32 2295 MIPS KVM_REG_MIPS_CP0_CAUSE 32 2296 MIPS KVM_REG_MIPS_CP0_EPC 64 2297 MIPS KVM_REG_MIPS_CP0_PRID 32 2298 MIPS KVM_REG_MIPS_CP0_EBASE 64 2299 MIPS KVM_REG_MIPS_CP0_CONFIG 32 2300 MIPS KVM_REG_MIPS_CP0_CONFIG1 32 2301 MIPS KVM_REG_MIPS_CP0_CONFIG2 32 2302 MIPS KVM_REG_MIPS_CP0_CONFIG3 32 2303 MIPS KVM_REG_MIPS_CP0_CONFIG4 32 2304 MIPS KVM_REG_MIPS_CP0_CONFIG5 32 2305 MIPS KVM_REG_MIPS_CP0_CONFIG7 32 2306 MIPS KVM_REG_MIPS_CP0_XCONTEXT 64 2307 MIPS KVM_REG_MIPS_CP0_ERROREPC 64 2308 MIPS KVM_REG_MIPS_CP0_KSCRATCH1 64 2309 MIPS KVM_REG_MIPS_CP0_KSCRATCH2 64 2310 MIPS KVM_REG_MIPS_CP0_KSCRATCH3 64 2311 MIPS KVM_REG_MIPS_CP0_KSCRATCH4 64 2312 MIPS KVM_REG_MIPS_CP0_KSCRATCH5 64 2313 MIPS KVM_REG_MIPS_CP0_KSCRATCH6 64 2314 MIPS KVM_REG_MIPS_CP0_MAAR(0..63) 64 2315 MIPS KVM_REG_MIPS_COUNT_CTL 64 2316 MIPS KVM_REG_MIPS_COUNT_RESUME 64 2317 MIPS KVM_REG_MIPS_COUNT_HZ 64 2318 MIPS KVM_REG_MIPS_FPR_32(0..31) 32 2319 MIPS KVM_REG_MIPS_FPR_64(0..31) 64 2320 MIPS KVM_REG_MIPS_VEC_128(0..31) 128 2321 MIPS KVM_REG_MIPS_FCR_IR 32 2322 MIPS KVM_REG_MIPS_FCR_CSR 32 2323 MIPS KVM_REG_MIPS_MSA_IR 32 2324 MIPS KVM_REG_MIPS_MSA_CSR 32 2325 ======= =============================== ============ 2326 2327ARM registers are mapped using the lower 32 bits. The upper 16 of that 2328is the register group type, or coprocessor number: 2329 2330ARM core registers have the following id bit patterns:: 2331 2332 0x4020 0000 0010 <index into the kvm_regs struct:16> 2333 2334ARM 32-bit CP15 registers have the following id bit patterns:: 2335 2336 0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3> 2337 2338ARM 64-bit CP15 registers have the following id bit patterns:: 2339 2340 0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3> 2341 2342ARM CCSIDR registers are demultiplexed by CSSELR value:: 2343 2344 0x4020 0000 0011 00 <csselr:8> 2345 2346ARM 32-bit VFP control registers have the following id bit patterns:: 2347 2348 0x4020 0000 0012 1 <regno:12> 2349 2350ARM 64-bit FP registers have the following id bit patterns:: 2351 2352 0x4030 0000 0012 0 <regno:12> 2353 2354ARM firmware pseudo-registers have the following bit pattern:: 2355 2356 0x4030 0000 0014 <regno:16> 2357 2358 2359arm64 registers are mapped using the lower 32 bits. The upper 16 of 2360that is the register group type, or coprocessor number: 2361 2362arm64 core/FP-SIMD registers have the following id bit patterns. Note 2363that the size of the access is variable, as the kvm_regs structure 2364contains elements ranging from 32 to 128 bits. The index is a 32bit 2365value in the kvm_regs structure seen as a 32bit array:: 2366 2367 0x60x0 0000 0010 <index into the kvm_regs struct:16> 2368 2369Specifically: 2370 2371======================= ========= ===== ======================================= 2372 Encoding Register Bits kvm_regs member 2373======================= ========= ===== ======================================= 2374 0x6030 0000 0010 0000 X0 64 regs.regs[0] 2375 0x6030 0000 0010 0002 X1 64 regs.regs[1] 2376 ... 2377 0x6030 0000 0010 003c X30 64 regs.regs[30] 2378 0x6030 0000 0010 003e SP 64 regs.sp 2379 0x6030 0000 0010 0040 PC 64 regs.pc 2380 0x6030 0000 0010 0042 PSTATE 64 regs.pstate 2381 0x6030 0000 0010 0044 SP_EL1 64 sp_el1 2382 0x6030 0000 0010 0046 ELR_EL1 64 elr_el1 2383 0x6030 0000 0010 0048 SPSR_EL1 64 spsr[KVM_SPSR_EL1] (alias SPSR_SVC) 2384 0x6030 0000 0010 004a SPSR_ABT 64 spsr[KVM_SPSR_ABT] 2385 0x6030 0000 0010 004c SPSR_UND 64 spsr[KVM_SPSR_UND] 2386 0x6030 0000 0010 004e SPSR_IRQ 64 spsr[KVM_SPSR_IRQ] 2387 0x6060 0000 0010 0050 SPSR_FIQ 64 spsr[KVM_SPSR_FIQ] 2388 0x6040 0000 0010 0054 V0 128 fp_regs.vregs[0] [1]_ 2389 0x6040 0000 0010 0058 V1 128 fp_regs.vregs[1] [1]_ 2390 ... 2391 0x6040 0000 0010 00d0 V31 128 fp_regs.vregs[31] [1]_ 2392 0x6020 0000 0010 00d4 FPSR 32 fp_regs.fpsr 2393 0x6020 0000 0010 00d5 FPCR 32 fp_regs.fpcr 2394======================= ========= ===== ======================================= 2395 2396.. [1] These encodings are not accepted for SVE-enabled vcpus. See 2397 KVM_ARM_VCPU_INIT. 2398 2399 The equivalent register content can be accessed via bits [127:0] of 2400 the corresponding SVE Zn registers instead for vcpus that have SVE 2401 enabled (see below). 2402 2403arm64 CCSIDR registers are demultiplexed by CSSELR value:: 2404 2405 0x6020 0000 0011 00 <csselr:8> 2406 2407arm64 system registers have the following id bit patterns:: 2408 2409 0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3> 2410 2411.. warning:: 2412 2413 Two system register IDs do not follow the specified pattern. These 2414 are KVM_REG_ARM_TIMER_CVAL and KVM_REG_ARM_TIMER_CNT, which map to 2415 system registers CNTV_CVAL_EL0 and CNTVCT_EL0 respectively. These 2416 two had their values accidentally swapped, which means TIMER_CVAL is 2417 derived from the register encoding for CNTVCT_EL0 and TIMER_CNT is 2418 derived from the register encoding for CNTV_CVAL_EL0. As this is 2419 API, it must remain this way. 2420 2421arm64 firmware pseudo-registers have the following bit pattern:: 2422 2423 0x6030 0000 0014 <regno:16> 2424 2425arm64 SVE registers have the following bit patterns:: 2426 2427 0x6080 0000 0015 00 <n:5> <slice:5> Zn bits[2048*slice + 2047 : 2048*slice] 2428 0x6050 0000 0015 04 <n:4> <slice:5> Pn bits[256*slice + 255 : 256*slice] 2429 0x6050 0000 0015 060 <slice:5> FFR bits[256*slice + 255 : 256*slice] 2430 0x6060 0000 0015 ffff KVM_REG_ARM64_SVE_VLS pseudo-register 2431 2432Access to register IDs where 2048 * slice >= 128 * max_vq will fail with 2433ENOENT. max_vq is the vcpu's maximum supported vector length in 128-bit 2434quadwords: see [2]_ below. 2435 2436These registers are only accessible on vcpus for which SVE is enabled. 2437See KVM_ARM_VCPU_INIT for details. 2438 2439In addition, except for KVM_REG_ARM64_SVE_VLS, these registers are not 2440accessible until the vcpu's SVE configuration has been finalized 2441using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). See KVM_ARM_VCPU_INIT 2442and KVM_ARM_VCPU_FINALIZE for more information about this procedure. 2443 2444KVM_REG_ARM64_SVE_VLS is a pseudo-register that allows the set of vector 2445lengths supported by the vcpu to be discovered and configured by 2446userspace. When transferred to or from user memory via KVM_GET_ONE_REG 2447or KVM_SET_ONE_REG, the value of this register is of type 2448__u64[KVM_ARM64_SVE_VLS_WORDS], and encodes the set of vector lengths as 2449follows:: 2450 2451 __u64 vector_lengths[KVM_ARM64_SVE_VLS_WORDS]; 2452 2453 if (vq >= SVE_VQ_MIN && vq <= SVE_VQ_MAX && 2454 ((vector_lengths[(vq - KVM_ARM64_SVE_VQ_MIN) / 64] >> 2455 ((vq - KVM_ARM64_SVE_VQ_MIN) % 64)) & 1)) 2456 /* Vector length vq * 16 bytes supported */ 2457 else 2458 /* Vector length vq * 16 bytes not supported */ 2459 2460.. [2] The maximum value vq for which the above condition is true is 2461 max_vq. This is the maximum vector length available to the guest on 2462 this vcpu, and determines which register slices are visible through 2463 this ioctl interface. 2464 2465(See Documentation/arm64/sve.rst for an explanation of the "vq" 2466nomenclature.) 2467 2468KVM_REG_ARM64_SVE_VLS is only accessible after KVM_ARM_VCPU_INIT. 2469KVM_ARM_VCPU_INIT initialises it to the best set of vector lengths that 2470the host supports. 2471 2472Userspace may subsequently modify it if desired until the vcpu's SVE 2473configuration is finalized using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). 2474 2475Apart from simply removing all vector lengths from the host set that 2476exceed some value, support for arbitrarily chosen sets of vector lengths 2477is hardware-dependent and may not be available. Attempting to configure 2478an invalid set of vector lengths via KVM_SET_ONE_REG will fail with 2479EINVAL. 2480 2481After the vcpu's SVE configuration is finalized, further attempts to 2482write this register will fail with EPERM. 2483 2484 2485MIPS registers are mapped using the lower 32 bits. The upper 16 of that is 2486the register group type: 2487 2488MIPS core registers (see above) have the following id bit patterns:: 2489 2490 0x7030 0000 0000 <reg:16> 2491 2492MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit 2493patterns depending on whether they're 32-bit or 64-bit registers:: 2494 2495 0x7020 0000 0001 00 <reg:5> <sel:3> (32-bit) 2496 0x7030 0000 0001 00 <reg:5> <sel:3> (64-bit) 2497 2498Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_MIPS_CP0_ENTRYLO1 are the MIPS64 2499versions of the EntryLo registers regardless of the word size of the host 2500hardware, host kernel, guest, and whether XPA is present in the guest, i.e. 2501with the RI and XI bits (if they exist) in bits 63 and 62 respectively, and 2502the PFNX field starting at bit 30. 2503 2504MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) above) have the following id bit 2505patterns:: 2506 2507 0x7030 0000 0001 01 <reg:8> 2508 2509MIPS KVM control registers (see above) have the following id bit patterns:: 2510 2511 0x7030 0000 0002 <reg:16> 2512 2513MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following 2514id bit patterns depending on the size of the register being accessed. They are 2515always accessed according to the current guest FPU mode (Status.FR and 2516Config5.FRE), i.e. as the guest would see them, and they become unpredictable 2517if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector 2518registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they 2519overlap the FPU registers:: 2520 2521 0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers) 2522 0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers) 2523 0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers) 2524 2525MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the 2526following id bit patterns:: 2527 2528 0x7020 0000 0003 01 <0:3> <reg:5> 2529 2530MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the 2531following id bit patterns:: 2532 2533 0x7020 0000 0003 02 <0:3> <reg:5> 2534 2535 25364.69 KVM_GET_ONE_REG 2537-------------------- 2538 2539:Capability: KVM_CAP_ONE_REG 2540:Architectures: all 2541:Type: vcpu ioctl 2542:Parameters: struct kvm_one_reg (in and out) 2543:Returns: 0 on success, negative value on failure 2544 2545Errors include: 2546 2547 ======== ============================================================ 2548 ENOENT no such register 2549 EINVAL invalid register ID, or no such register or used with VMs in 2550 protected virtualization mode on s390 2551 EPERM (arm64) register access not allowed before vcpu finalization 2552 ======== ============================================================ 2553 2554(These error codes are indicative only: do not rely on a specific error 2555code being returned in a specific situation.) 2556 2557This ioctl allows to receive the value of a single register implemented 2558in a vcpu. The register to read is indicated by the "id" field of the 2559kvm_one_reg struct passed in. On success, the register value can be found 2560at the memory location pointed to by "addr". 2561 2562The list of registers accessible using this interface is identical to the 2563list in 4.68. 2564 2565 25664.70 KVM_KVMCLOCK_CTRL 2567---------------------- 2568 2569:Capability: KVM_CAP_KVMCLOCK_CTRL 2570:Architectures: Any that implement pvclocks (currently x86 only) 2571:Type: vcpu ioctl 2572:Parameters: None 2573:Returns: 0 on success, -1 on error 2574 2575This ioctl sets a flag accessible to the guest indicating that the specified 2576vCPU has been paused by the host userspace. 2577 2578The host will set a flag in the pvclock structure that is checked from the 2579soft lockup watchdog. The flag is part of the pvclock structure that is 2580shared between guest and host, specifically the second bit of the flags 2581field of the pvclock_vcpu_time_info structure. It will be set exclusively by 2582the host and read/cleared exclusively by the guest. The guest operation of 2583checking and clearing the flag must be an atomic operation so 2584load-link/store-conditional, or equivalent must be used. There are two cases 2585where the guest will clear the flag: when the soft lockup watchdog timer resets 2586itself or when a soft lockup is detected. This ioctl can be called any time 2587after pausing the vcpu, but before it is resumed. 2588 2589 25904.71 KVM_SIGNAL_MSI 2591------------------- 2592 2593:Capability: KVM_CAP_SIGNAL_MSI 2594:Architectures: x86 arm arm64 2595:Type: vm ioctl 2596:Parameters: struct kvm_msi (in) 2597:Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error 2598 2599Directly inject a MSI message. Only valid with in-kernel irqchip that handles 2600MSI messages. 2601 2602:: 2603 2604 struct kvm_msi { 2605 __u32 address_lo; 2606 __u32 address_hi; 2607 __u32 data; 2608 __u32 flags; 2609 __u32 devid; 2610 __u8 pad[12]; 2611 }; 2612 2613flags: 2614 KVM_MSI_VALID_DEVID: devid contains a valid value. The per-VM 2615 KVM_CAP_MSI_DEVID capability advertises the requirement to provide 2616 the device ID. If this capability is not available, userspace 2617 should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail. 2618 2619If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier 2620for the device that wrote the MSI message. For PCI, this is usually a 2621BFD identifier in the lower 16 bits. 2622 2623On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS 2624feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled, 2625address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of 2626address_hi must be zero. 2627 2628 26294.71 KVM_CREATE_PIT2 2630-------------------- 2631 2632:Capability: KVM_CAP_PIT2 2633:Architectures: x86 2634:Type: vm ioctl 2635:Parameters: struct kvm_pit_config (in) 2636:Returns: 0 on success, -1 on error 2637 2638Creates an in-kernel device model for the i8254 PIT. This call is only valid 2639after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following 2640parameters have to be passed:: 2641 2642 struct kvm_pit_config { 2643 __u32 flags; 2644 __u32 pad[15]; 2645 }; 2646 2647Valid flags are:: 2648 2649 #define KVM_PIT_SPEAKER_DUMMY 1 /* emulate speaker port stub */ 2650 2651PIT timer interrupts may use a per-VM kernel thread for injection. If it 2652exists, this thread will have a name of the following pattern:: 2653 2654 kvm-pit/<owner-process-pid> 2655 2656When running a guest with elevated priorities, the scheduling parameters of 2657this thread may have to be adjusted accordingly. 2658 2659This IOCTL replaces the obsolete KVM_CREATE_PIT. 2660 2661 26624.72 KVM_GET_PIT2 2663----------------- 2664 2665:Capability: KVM_CAP_PIT_STATE2 2666:Architectures: x86 2667:Type: vm ioctl 2668:Parameters: struct kvm_pit_state2 (out) 2669:Returns: 0 on success, -1 on error 2670 2671Retrieves the state of the in-kernel PIT model. Only valid after 2672KVM_CREATE_PIT2. The state is returned in the following structure:: 2673 2674 struct kvm_pit_state2 { 2675 struct kvm_pit_channel_state channels[3]; 2676 __u32 flags; 2677 __u32 reserved[9]; 2678 }; 2679 2680Valid flags are:: 2681 2682 /* disable PIT in HPET legacy mode */ 2683 #define KVM_PIT_FLAGS_HPET_LEGACY 0x00000001 2684 2685This IOCTL replaces the obsolete KVM_GET_PIT. 2686 2687 26884.73 KVM_SET_PIT2 2689----------------- 2690 2691:Capability: KVM_CAP_PIT_STATE2 2692:Architectures: x86 2693:Type: vm ioctl 2694:Parameters: struct kvm_pit_state2 (in) 2695:Returns: 0 on success, -1 on error 2696 2697Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2. 2698See KVM_GET_PIT2 for details on struct kvm_pit_state2. 2699 2700This IOCTL replaces the obsolete KVM_SET_PIT. 2701 2702 27034.74 KVM_PPC_GET_SMMU_INFO 2704-------------------------- 2705 2706:Capability: KVM_CAP_PPC_GET_SMMU_INFO 2707:Architectures: powerpc 2708:Type: vm ioctl 2709:Parameters: None 2710:Returns: 0 on success, -1 on error 2711 2712This populates and returns a structure describing the features of 2713the "Server" class MMU emulation supported by KVM. 2714This can in turn be used by userspace to generate the appropriate 2715device-tree properties for the guest operating system. 2716 2717The structure contains some global information, followed by an 2718array of supported segment page sizes:: 2719 2720 struct kvm_ppc_smmu_info { 2721 __u64 flags; 2722 __u32 slb_size; 2723 __u32 pad; 2724 struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ]; 2725 }; 2726 2727The supported flags are: 2728 2729 - KVM_PPC_PAGE_SIZES_REAL: 2730 When that flag is set, guest page sizes must "fit" the backing 2731 store page sizes. When not set, any page size in the list can 2732 be used regardless of how they are backed by userspace. 2733 2734 - KVM_PPC_1T_SEGMENTS 2735 The emulated MMU supports 1T segments in addition to the 2736 standard 256M ones. 2737 2738 - KVM_PPC_NO_HASH 2739 This flag indicates that HPT guests are not supported by KVM, 2740 thus all guests must use radix MMU mode. 2741 2742The "slb_size" field indicates how many SLB entries are supported 2743 2744The "sps" array contains 8 entries indicating the supported base 2745page sizes for a segment in increasing order. Each entry is defined 2746as follow:: 2747 2748 struct kvm_ppc_one_seg_page_size { 2749 __u32 page_shift; /* Base page shift of segment (or 0) */ 2750 __u32 slb_enc; /* SLB encoding for BookS */ 2751 struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ]; 2752 }; 2753 2754An entry with a "page_shift" of 0 is unused. Because the array is 2755organized in increasing order, a lookup can stop when encoutering 2756such an entry. 2757 2758The "slb_enc" field provides the encoding to use in the SLB for the 2759page size. The bits are in positions such as the value can directly 2760be OR'ed into the "vsid" argument of the slbmte instruction. 2761 2762The "enc" array is a list which for each of those segment base page 2763size provides the list of supported actual page sizes (which can be 2764only larger or equal to the base page size), along with the 2765corresponding encoding in the hash PTE. Similarly, the array is 27668 entries sorted by increasing sizes and an entry with a "0" shift 2767is an empty entry and a terminator:: 2768 2769 struct kvm_ppc_one_page_size { 2770 __u32 page_shift; /* Page shift (or 0) */ 2771 __u32 pte_enc; /* Encoding in the HPTE (>>12) */ 2772 }; 2773 2774The "pte_enc" field provides a value that can OR'ed into the hash 2775PTE's RPN field (ie, it needs to be shifted left by 12 to OR it 2776into the hash PTE second double word). 2777 27784.75 KVM_IRQFD 2779-------------- 2780 2781:Capability: KVM_CAP_IRQFD 2782:Architectures: x86 s390 arm arm64 2783:Type: vm ioctl 2784:Parameters: struct kvm_irqfd (in) 2785:Returns: 0 on success, -1 on error 2786 2787Allows setting an eventfd to directly trigger a guest interrupt. 2788kvm_irqfd.fd specifies the file descriptor to use as the eventfd and 2789kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When 2790an event is triggered on the eventfd, an interrupt is injected into 2791the guest using the specified gsi pin. The irqfd is removed using 2792the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd 2793and kvm_irqfd.gsi. 2794 2795With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify 2796mechanism allowing emulation of level-triggered, irqfd-based 2797interrupts. When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an 2798additional eventfd in the kvm_irqfd.resamplefd field. When operating 2799in resample mode, posting of an interrupt through kvm_irq.fd asserts 2800the specified gsi in the irqchip. When the irqchip is resampled, such 2801as from an EOI, the gsi is de-asserted and the user is notified via 2802kvm_irqfd.resamplefd. It is the user's responsibility to re-queue 2803the interrupt if the device making use of it still requires service. 2804Note that closing the resamplefd is not sufficient to disable the 2805irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment 2806and need not be specified with KVM_IRQFD_FLAG_DEASSIGN. 2807 2808On arm/arm64, gsi routing being supported, the following can happen: 2809 2810- in case no routing entry is associated to this gsi, injection fails 2811- in case the gsi is associated to an irqchip routing entry, 2812 irqchip.pin + 32 corresponds to the injected SPI ID. 2813- in case the gsi is associated to an MSI routing entry, the MSI 2814 message and device ID are translated into an LPI (support restricted 2815 to GICv3 ITS in-kernel emulation). 2816 28174.76 KVM_PPC_ALLOCATE_HTAB 2818-------------------------- 2819 2820:Capability: KVM_CAP_PPC_ALLOC_HTAB 2821:Architectures: powerpc 2822:Type: vm ioctl 2823:Parameters: Pointer to u32 containing hash table order (in/out) 2824:Returns: 0 on success, -1 on error 2825 2826This requests the host kernel to allocate an MMU hash table for a 2827guest using the PAPR paravirtualization interface. This only does 2828anything if the kernel is configured to use the Book 3S HV style of 2829virtualization. Otherwise the capability doesn't exist and the ioctl 2830returns an ENOTTY error. The rest of this description assumes Book 3S 2831HV. 2832 2833There must be no vcpus running when this ioctl is called; if there 2834are, it will do nothing and return an EBUSY error. 2835 2836The parameter is a pointer to a 32-bit unsigned integer variable 2837containing the order (log base 2) of the desired size of the hash 2838table, which must be between 18 and 46. On successful return from the 2839ioctl, the value will not be changed by the kernel. 2840 2841If no hash table has been allocated when any vcpu is asked to run 2842(with the KVM_RUN ioctl), the host kernel will allocate a 2843default-sized hash table (16 MB). 2844 2845If this ioctl is called when a hash table has already been allocated, 2846with a different order from the existing hash table, the existing hash 2847table will be freed and a new one allocated. If this is ioctl is 2848called when a hash table has already been allocated of the same order 2849as specified, the kernel will clear out the existing hash table (zero 2850all HPTEs). In either case, if the guest is using the virtualized 2851real-mode area (VRMA) facility, the kernel will re-create the VMRA 2852HPTEs on the next KVM_RUN of any vcpu. 2853 28544.77 KVM_S390_INTERRUPT 2855----------------------- 2856 2857:Capability: basic 2858:Architectures: s390 2859:Type: vm ioctl, vcpu ioctl 2860:Parameters: struct kvm_s390_interrupt (in) 2861:Returns: 0 on success, -1 on error 2862 2863Allows to inject an interrupt to the guest. Interrupts can be floating 2864(vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type. 2865 2866Interrupt parameters are passed via kvm_s390_interrupt:: 2867 2868 struct kvm_s390_interrupt { 2869 __u32 type; 2870 __u32 parm; 2871 __u64 parm64; 2872 }; 2873 2874type can be one of the following: 2875 2876KVM_S390_SIGP_STOP (vcpu) 2877 - sigp stop; optional flags in parm 2878KVM_S390_PROGRAM_INT (vcpu) 2879 - program check; code in parm 2880KVM_S390_SIGP_SET_PREFIX (vcpu) 2881 - sigp set prefix; prefix address in parm 2882KVM_S390_RESTART (vcpu) 2883 - restart 2884KVM_S390_INT_CLOCK_COMP (vcpu) 2885 - clock comparator interrupt 2886KVM_S390_INT_CPU_TIMER (vcpu) 2887 - CPU timer interrupt 2888KVM_S390_INT_VIRTIO (vm) 2889 - virtio external interrupt; external interrupt 2890 parameters in parm and parm64 2891KVM_S390_INT_SERVICE (vm) 2892 - sclp external interrupt; sclp parameter in parm 2893KVM_S390_INT_EMERGENCY (vcpu) 2894 - sigp emergency; source cpu in parm 2895KVM_S390_INT_EXTERNAL_CALL (vcpu) 2896 - sigp external call; source cpu in parm 2897KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm) 2898 - compound value to indicate an 2899 I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel); 2900 I/O interruption parameters in parm (subchannel) and parm64 (intparm, 2901 interruption subclass) 2902KVM_S390_MCHK (vm, vcpu) 2903 - machine check interrupt; cr 14 bits in parm, machine check interrupt 2904 code in parm64 (note that machine checks needing further payload are not 2905 supported by this ioctl) 2906 2907This is an asynchronous vcpu ioctl and can be invoked from any thread. 2908 29094.78 KVM_PPC_GET_HTAB_FD 2910------------------------ 2911 2912:Capability: KVM_CAP_PPC_HTAB_FD 2913:Architectures: powerpc 2914:Type: vm ioctl 2915:Parameters: Pointer to struct kvm_get_htab_fd (in) 2916:Returns: file descriptor number (>= 0) on success, -1 on error 2917 2918This returns a file descriptor that can be used either to read out the 2919entries in the guest's hashed page table (HPT), or to write entries to 2920initialize the HPT. The returned fd can only be written to if the 2921KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and 2922can only be read if that bit is clear. The argument struct looks like 2923this:: 2924 2925 /* For KVM_PPC_GET_HTAB_FD */ 2926 struct kvm_get_htab_fd { 2927 __u64 flags; 2928 __u64 start_index; 2929 __u64 reserved[2]; 2930 }; 2931 2932 /* Values for kvm_get_htab_fd.flags */ 2933 #define KVM_GET_HTAB_BOLTED_ONLY ((__u64)0x1) 2934 #define KVM_GET_HTAB_WRITE ((__u64)0x2) 2935 2936The 'start_index' field gives the index in the HPT of the entry at 2937which to start reading. It is ignored when writing. 2938 2939Reads on the fd will initially supply information about all 2940"interesting" HPT entries. Interesting entries are those with the 2941bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise 2942all entries. When the end of the HPT is reached, the read() will 2943return. If read() is called again on the fd, it will start again from 2944the beginning of the HPT, but will only return HPT entries that have 2945changed since they were last read. 2946 2947Data read or written is structured as a header (8 bytes) followed by a 2948series of valid HPT entries (16 bytes) each. The header indicates how 2949many valid HPT entries there are and how many invalid entries follow 2950the valid entries. The invalid entries are not represented explicitly 2951in the stream. The header format is:: 2952 2953 struct kvm_get_htab_header { 2954 __u32 index; 2955 __u16 n_valid; 2956 __u16 n_invalid; 2957 }; 2958 2959Writes to the fd create HPT entries starting at the index given in the 2960header; first 'n_valid' valid entries with contents from the data 2961written, then 'n_invalid' invalid entries, invalidating any previously 2962valid entries found. 2963 29644.79 KVM_CREATE_DEVICE 2965---------------------- 2966 2967:Capability: KVM_CAP_DEVICE_CTRL 2968:Type: vm ioctl 2969:Parameters: struct kvm_create_device (in/out) 2970:Returns: 0 on success, -1 on error 2971 2972Errors: 2973 2974 ====== ======================================================= 2975 ENODEV The device type is unknown or unsupported 2976 EEXIST Device already created, and this type of device may not 2977 be instantiated multiple times 2978 ====== ======================================================= 2979 2980 Other error conditions may be defined by individual device types or 2981 have their standard meanings. 2982 2983Creates an emulated device in the kernel. The file descriptor returned 2984in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR. 2985 2986If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the 2987device type is supported (not necessarily whether it can be created 2988in the current vm). 2989 2990Individual devices should not define flags. Attributes should be used 2991for specifying any behavior that is not implied by the device type 2992number. 2993 2994:: 2995 2996 struct kvm_create_device { 2997 __u32 type; /* in: KVM_DEV_TYPE_xxx */ 2998 __u32 fd; /* out: device handle */ 2999 __u32 flags; /* in: KVM_CREATE_DEVICE_xxx */ 3000 }; 3001 30024.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR 3003-------------------------------------------- 3004 3005:Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device, 3006 KVM_CAP_VCPU_ATTRIBUTES for vcpu device 3007:Type: device ioctl, vm ioctl, vcpu ioctl 3008:Parameters: struct kvm_device_attr 3009:Returns: 0 on success, -1 on error 3010 3011Errors: 3012 3013 ===== ============================================================= 3014 ENXIO The group or attribute is unknown/unsupported for this device 3015 or hardware support is missing. 3016 EPERM The attribute cannot (currently) be accessed this way 3017 (e.g. read-only attribute, or attribute that only makes 3018 sense when the device is in a different state) 3019 ===== ============================================================= 3020 3021 Other error conditions may be defined by individual device types. 3022 3023Gets/sets a specified piece of device configuration and/or state. The 3024semantics are device-specific. See individual device documentation in 3025the "devices" directory. As with ONE_REG, the size of the data 3026transferred is defined by the particular attribute. 3027 3028:: 3029 3030 struct kvm_device_attr { 3031 __u32 flags; /* no flags currently defined */ 3032 __u32 group; /* device-defined */ 3033 __u64 attr; /* group-defined */ 3034 __u64 addr; /* userspace address of attr data */ 3035 }; 3036 30374.81 KVM_HAS_DEVICE_ATTR 3038------------------------ 3039 3040:Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device, 3041 KVM_CAP_VCPU_ATTRIBUTES for vcpu device 3042:Type: device ioctl, vm ioctl, vcpu ioctl 3043:Parameters: struct kvm_device_attr 3044:Returns: 0 on success, -1 on error 3045 3046Errors: 3047 3048 ===== ============================================================= 3049 ENXIO The group or attribute is unknown/unsupported for this device 3050 or hardware support is missing. 3051 ===== ============================================================= 3052 3053Tests whether a device supports a particular attribute. A successful 3054return indicates the attribute is implemented. It does not necessarily 3055indicate that the attribute can be read or written in the device's 3056current state. "addr" is ignored. 3057 30584.82 KVM_ARM_VCPU_INIT 3059---------------------- 3060 3061:Capability: basic 3062:Architectures: arm, arm64 3063:Type: vcpu ioctl 3064:Parameters: struct kvm_vcpu_init (in) 3065:Returns: 0 on success; -1 on error 3066 3067Errors: 3068 3069 ====== ================================================================= 3070 EINVAL the target is unknown, or the combination of features is invalid. 3071 ENOENT a features bit specified is unknown. 3072 ====== ================================================================= 3073 3074This tells KVM what type of CPU to present to the guest, and what 3075optional features it should have. This will cause a reset of the cpu 3076registers to their initial values. If this is not called, KVM_RUN will 3077return ENOEXEC for that vcpu. 3078 3079Note that because some registers reflect machine topology, all vcpus 3080should be created before this ioctl is invoked. 3081 3082Userspace can call this function multiple times for a given vcpu, including 3083after the vcpu has been run. This will reset the vcpu to its initial 3084state. All calls to this function after the initial call must use the same 3085target and same set of feature flags, otherwise EINVAL will be returned. 3086 3087Possible features: 3088 3089 - KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state. 3090 Depends on KVM_CAP_ARM_PSCI. If not set, the CPU will be powered on 3091 and execute guest code when KVM_RUN is called. 3092 - KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode. 3093 Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only). 3094 - KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 (or a future revision 3095 backward compatible with v0.2) for the CPU. 3096 Depends on KVM_CAP_ARM_PSCI_0_2. 3097 - KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU. 3098 Depends on KVM_CAP_ARM_PMU_V3. 3099 3100 - KVM_ARM_VCPU_PTRAUTH_ADDRESS: Enables Address Pointer authentication 3101 for arm64 only. 3102 Depends on KVM_CAP_ARM_PTRAUTH_ADDRESS. 3103 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are 3104 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and 3105 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be 3106 requested. 3107 3108 - KVM_ARM_VCPU_PTRAUTH_GENERIC: Enables Generic Pointer authentication 3109 for arm64 only. 3110 Depends on KVM_CAP_ARM_PTRAUTH_GENERIC. 3111 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are 3112 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and 3113 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be 3114 requested. 3115 3116 - KVM_ARM_VCPU_SVE: Enables SVE for the CPU (arm64 only). 3117 Depends on KVM_CAP_ARM_SVE. 3118 Requires KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3119 3120 * After KVM_ARM_VCPU_INIT: 3121 3122 - KVM_REG_ARM64_SVE_VLS may be read using KVM_GET_ONE_REG: the 3123 initial value of this pseudo-register indicates the best set of 3124 vector lengths possible for a vcpu on this host. 3125 3126 * Before KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3127 3128 - KVM_RUN and KVM_GET_REG_LIST are not available; 3129 3130 - KVM_GET_ONE_REG and KVM_SET_ONE_REG cannot be used to access 3131 the scalable archietctural SVE registers 3132 KVM_REG_ARM64_SVE_ZREG(), KVM_REG_ARM64_SVE_PREG() or 3133 KVM_REG_ARM64_SVE_FFR; 3134 3135 - KVM_REG_ARM64_SVE_VLS may optionally be written using 3136 KVM_SET_ONE_REG, to modify the set of vector lengths available 3137 for the vcpu. 3138 3139 * After KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3140 3141 - the KVM_REG_ARM64_SVE_VLS pseudo-register is immutable, and can 3142 no longer be written using KVM_SET_ONE_REG. 3143 31444.83 KVM_ARM_PREFERRED_TARGET 3145----------------------------- 3146 3147:Capability: basic 3148:Architectures: arm, arm64 3149:Type: vm ioctl 3150:Parameters: struct struct kvm_vcpu_init (out) 3151:Returns: 0 on success; -1 on error 3152 3153Errors: 3154 3155 ====== ========================================== 3156 ENODEV no preferred target available for the host 3157 ====== ========================================== 3158 3159This queries KVM for preferred CPU target type which can be emulated 3160by KVM on underlying host. 3161 3162The ioctl returns struct kvm_vcpu_init instance containing information 3163about preferred CPU target type and recommended features for it. The 3164kvm_vcpu_init->features bitmap returned will have feature bits set if 3165the preferred target recommends setting these features, but this is 3166not mandatory. 3167 3168The information returned by this ioctl can be used to prepare an instance 3169of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in 3170in VCPU matching underlying host. 3171 3172 31734.84 KVM_GET_REG_LIST 3174--------------------- 3175 3176:Capability: basic 3177:Architectures: arm, arm64, mips 3178:Type: vcpu ioctl 3179:Parameters: struct kvm_reg_list (in/out) 3180:Returns: 0 on success; -1 on error 3181 3182Errors: 3183 3184 ===== ============================================================== 3185 E2BIG the reg index list is too big to fit in the array specified by 3186 the user (the number required will be written into n). 3187 ===== ============================================================== 3188 3189:: 3190 3191 struct kvm_reg_list { 3192 __u64 n; /* number of registers in reg[] */ 3193 __u64 reg[0]; 3194 }; 3195 3196This ioctl returns the guest registers that are supported for the 3197KVM_GET_ONE_REG/KVM_SET_ONE_REG calls. 3198 3199 32004.85 KVM_ARM_SET_DEVICE_ADDR (deprecated) 3201----------------------------------------- 3202 3203:Capability: KVM_CAP_ARM_SET_DEVICE_ADDR 3204:Architectures: arm, arm64 3205:Type: vm ioctl 3206:Parameters: struct kvm_arm_device_address (in) 3207:Returns: 0 on success, -1 on error 3208 3209Errors: 3210 3211 ====== ============================================ 3212 ENODEV The device id is unknown 3213 ENXIO Device not supported on current system 3214 EEXIST Address already set 3215 E2BIG Address outside guest physical address space 3216 EBUSY Address overlaps with other device range 3217 ====== ============================================ 3218 3219:: 3220 3221 struct kvm_arm_device_addr { 3222 __u64 id; 3223 __u64 addr; 3224 }; 3225 3226Specify a device address in the guest's physical address space where guests 3227can access emulated or directly exposed devices, which the host kernel needs 3228to know about. The id field is an architecture specific identifier for a 3229specific device. 3230 3231ARM/arm64 divides the id field into two parts, a device id and an 3232address type id specific to the individual device:: 3233 3234 bits: | 63 ... 32 | 31 ... 16 | 15 ... 0 | 3235 field: | 0x00000000 | device id | addr type id | 3236 3237ARM/arm64 currently only require this when using the in-kernel GIC 3238support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2 3239as the device id. When setting the base address for the guest's 3240mapping of the VGIC virtual CPU and distributor interface, the ioctl 3241must be called after calling KVM_CREATE_IRQCHIP, but before calling 3242KVM_RUN on any of the VCPUs. Calling this ioctl twice for any of the 3243base addresses will return -EEXIST. 3244 3245Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API 3246should be used instead. 3247 3248 32494.86 KVM_PPC_RTAS_DEFINE_TOKEN 3250------------------------------ 3251 3252:Capability: KVM_CAP_PPC_RTAS 3253:Architectures: ppc 3254:Type: vm ioctl 3255:Parameters: struct kvm_rtas_token_args 3256:Returns: 0 on success, -1 on error 3257 3258Defines a token value for a RTAS (Run Time Abstraction Services) 3259service in order to allow it to be handled in the kernel. The 3260argument struct gives the name of the service, which must be the name 3261of a service that has a kernel-side implementation. If the token 3262value is non-zero, it will be associated with that service, and 3263subsequent RTAS calls by the guest specifying that token will be 3264handled by the kernel. If the token value is 0, then any token 3265associated with the service will be forgotten, and subsequent RTAS 3266calls by the guest for that service will be passed to userspace to be 3267handled. 3268 32694.87 KVM_SET_GUEST_DEBUG 3270------------------------ 3271 3272:Capability: KVM_CAP_SET_GUEST_DEBUG 3273:Architectures: x86, s390, ppc, arm64 3274:Type: vcpu ioctl 3275:Parameters: struct kvm_guest_debug (in) 3276:Returns: 0 on success; -1 on error 3277 3278:: 3279 3280 struct kvm_guest_debug { 3281 __u32 control; 3282 __u32 pad; 3283 struct kvm_guest_debug_arch arch; 3284 }; 3285 3286Set up the processor specific debug registers and configure vcpu for 3287handling guest debug events. There are two parts to the structure, the 3288first a control bitfield indicates the type of debug events to handle 3289when running. Common control bits are: 3290 3291 - KVM_GUESTDBG_ENABLE: guest debugging is enabled 3292 - KVM_GUESTDBG_SINGLESTEP: the next run should single-step 3293 3294The top 16 bits of the control field are architecture specific control 3295flags which can include the following: 3296 3297 - KVM_GUESTDBG_USE_SW_BP: using software breakpoints [x86, arm64] 3298 - KVM_GUESTDBG_USE_HW_BP: using hardware breakpoints [x86, s390, arm64] 3299 - KVM_GUESTDBG_INJECT_DB: inject DB type exception [x86] 3300 - KVM_GUESTDBG_INJECT_BP: inject BP type exception [x86] 3301 - KVM_GUESTDBG_EXIT_PENDING: trigger an immediate guest exit [s390] 3302 3303For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints 3304are enabled in memory so we need to ensure breakpoint exceptions are 3305correctly trapped and the KVM run loop exits at the breakpoint and not 3306running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP 3307we need to ensure the guest vCPUs architecture specific registers are 3308updated to the correct (supplied) values. 3309 3310The second part of the structure is architecture specific and 3311typically contains a set of debug registers. 3312 3313For arm64 the number of debug registers is implementation defined and 3314can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and 3315KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number 3316indicating the number of supported registers. 3317 3318For ppc, the KVM_CAP_PPC_GUEST_DEBUG_SSTEP capability indicates whether 3319the single-step debug event (KVM_GUESTDBG_SINGLESTEP) is supported. 3320 3321When debug events exit the main run loop with the reason 3322KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run 3323structure containing architecture specific debug information. 3324 33254.88 KVM_GET_EMULATED_CPUID 3326--------------------------- 3327 3328:Capability: KVM_CAP_EXT_EMUL_CPUID 3329:Architectures: x86 3330:Type: system ioctl 3331:Parameters: struct kvm_cpuid2 (in/out) 3332:Returns: 0 on success, -1 on error 3333 3334:: 3335 3336 struct kvm_cpuid2 { 3337 __u32 nent; 3338 __u32 flags; 3339 struct kvm_cpuid_entry2 entries[0]; 3340 }; 3341 3342The member 'flags' is used for passing flags from userspace. 3343 3344:: 3345 3346 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0) 3347 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */ 3348 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */ 3349 3350 struct kvm_cpuid_entry2 { 3351 __u32 function; 3352 __u32 index; 3353 __u32 flags; 3354 __u32 eax; 3355 __u32 ebx; 3356 __u32 ecx; 3357 __u32 edx; 3358 __u32 padding[3]; 3359 }; 3360 3361This ioctl returns x86 cpuid features which are emulated by 3362kvm.Userspace can use the information returned by this ioctl to query 3363which features are emulated by kvm instead of being present natively. 3364 3365Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2 3366structure with the 'nent' field indicating the number of entries in 3367the variable-size array 'entries'. If the number of entries is too low 3368to describe the cpu capabilities, an error (E2BIG) is returned. If the 3369number is too high, the 'nent' field is adjusted and an error (ENOMEM) 3370is returned. If the number is just right, the 'nent' field is adjusted 3371to the number of valid entries in the 'entries' array, which is then 3372filled. 3373 3374The entries returned are the set CPUID bits of the respective features 3375which kvm emulates, as returned by the CPUID instruction, with unknown 3376or unsupported feature bits cleared. 3377 3378Features like x2apic, for example, may not be present in the host cpu 3379but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be 3380emulated efficiently and thus not included here. 3381 3382The fields in each entry are defined as follows: 3383 3384 function: 3385 the eax value used to obtain the entry 3386 index: 3387 the ecx value used to obtain the entry (for entries that are 3388 affected by ecx) 3389 flags: 3390 an OR of zero or more of the following: 3391 3392 KVM_CPUID_FLAG_SIGNIFCANT_INDEX: 3393 if the index field is valid 3394 3395 eax, ebx, ecx, edx: 3396 3397 the values returned by the cpuid instruction for 3398 this function/index combination 3399 34004.89 KVM_S390_MEM_OP 3401-------------------- 3402 3403:Capability: KVM_CAP_S390_MEM_OP 3404:Architectures: s390 3405:Type: vcpu ioctl 3406:Parameters: struct kvm_s390_mem_op (in) 3407:Returns: = 0 on success, 3408 < 0 on generic error (e.g. -EFAULT or -ENOMEM), 3409 > 0 if an exception occurred while walking the page tables 3410 3411Read or write data from/to the logical (virtual) memory of a VCPU. 3412 3413Parameters are specified via the following structure:: 3414 3415 struct kvm_s390_mem_op { 3416 __u64 gaddr; /* the guest address */ 3417 __u64 flags; /* flags */ 3418 __u32 size; /* amount of bytes */ 3419 __u32 op; /* type of operation */ 3420 __u64 buf; /* buffer in userspace */ 3421 __u8 ar; /* the access register number */ 3422 __u8 reserved[31]; /* should be set to 0 */ 3423 }; 3424 3425The type of operation is specified in the "op" field. It is either 3426KVM_S390_MEMOP_LOGICAL_READ for reading from logical memory space or 3427KVM_S390_MEMOP_LOGICAL_WRITE for writing to logical memory space. The 3428KVM_S390_MEMOP_F_CHECK_ONLY flag can be set in the "flags" field to check 3429whether the corresponding memory access would create an access exception 3430(without touching the data in the memory at the destination). In case an 3431access exception occurred while walking the MMU tables of the guest, the 3432ioctl returns a positive error number to indicate the type of exception. 3433This exception is also raised directly at the corresponding VCPU if the 3434flag KVM_S390_MEMOP_F_INJECT_EXCEPTION is set in the "flags" field. 3435 3436The start address of the memory region has to be specified in the "gaddr" 3437field, and the length of the region in the "size" field (which must not 3438be 0). The maximum value for "size" can be obtained by checking the 3439KVM_CAP_S390_MEM_OP capability. "buf" is the buffer supplied by the 3440userspace application where the read data should be written to for 3441KVM_S390_MEMOP_LOGICAL_READ, or where the data that should be written is 3442stored for a KVM_S390_MEMOP_LOGICAL_WRITE. When KVM_S390_MEMOP_F_CHECK_ONLY 3443is specified, "buf" is unused and can be NULL. "ar" designates the access 3444register number to be used; the valid range is 0..15. 3445 3446The "reserved" field is meant for future extensions. It is not used by 3447KVM with the currently defined set of flags. 3448 34494.90 KVM_S390_GET_SKEYS 3450----------------------- 3451 3452:Capability: KVM_CAP_S390_SKEYS 3453:Architectures: s390 3454:Type: vm ioctl 3455:Parameters: struct kvm_s390_skeys 3456:Returns: 0 on success, KVM_S390_GET_KEYS_NONE if guest is not using storage 3457 keys, negative value on error 3458 3459This ioctl is used to get guest storage key values on the s390 3460architecture. The ioctl takes parameters via the kvm_s390_skeys struct:: 3461 3462 struct kvm_s390_skeys { 3463 __u64 start_gfn; 3464 __u64 count; 3465 __u64 skeydata_addr; 3466 __u32 flags; 3467 __u32 reserved[9]; 3468 }; 3469 3470The start_gfn field is the number of the first guest frame whose storage keys 3471you want to get. 3472 3473The count field is the number of consecutive frames (starting from start_gfn) 3474whose storage keys to get. The count field must be at least 1 and the maximum 3475allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range 3476will cause the ioctl to return -EINVAL. 3477 3478The skeydata_addr field is the address to a buffer large enough to hold count 3479bytes. This buffer will be filled with storage key data by the ioctl. 3480 34814.91 KVM_S390_SET_SKEYS 3482----------------------- 3483 3484:Capability: KVM_CAP_S390_SKEYS 3485:Architectures: s390 3486:Type: vm ioctl 3487:Parameters: struct kvm_s390_skeys 3488:Returns: 0 on success, negative value on error 3489 3490This ioctl is used to set guest storage key values on the s390 3491architecture. The ioctl takes parameters via the kvm_s390_skeys struct. 3492See section on KVM_S390_GET_SKEYS for struct definition. 3493 3494The start_gfn field is the number of the first guest frame whose storage keys 3495you want to set. 3496 3497The count field is the number of consecutive frames (starting from start_gfn) 3498whose storage keys to get. The count field must be at least 1 and the maximum 3499allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range 3500will cause the ioctl to return -EINVAL. 3501 3502The skeydata_addr field is the address to a buffer containing count bytes of 3503storage keys. Each byte in the buffer will be set as the storage key for a 3504single frame starting at start_gfn for count frames. 3505 3506Note: If any architecturally invalid key value is found in the given data then 3507the ioctl will return -EINVAL. 3508 35094.92 KVM_S390_IRQ 3510----------------- 3511 3512:Capability: KVM_CAP_S390_INJECT_IRQ 3513:Architectures: s390 3514:Type: vcpu ioctl 3515:Parameters: struct kvm_s390_irq (in) 3516:Returns: 0 on success, -1 on error 3517 3518Errors: 3519 3520 3521 ====== ================================================================= 3522 EINVAL interrupt type is invalid 3523 type is KVM_S390_SIGP_STOP and flag parameter is invalid value, 3524 type is KVM_S390_INT_EXTERNAL_CALL and code is bigger 3525 than the maximum of VCPUs 3526 EBUSY type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped, 3527 type is KVM_S390_SIGP_STOP and a stop irq is already pending, 3528 type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt 3529 is already pending 3530 ====== ================================================================= 3531 3532Allows to inject an interrupt to the guest. 3533 3534Using struct kvm_s390_irq as a parameter allows 3535to inject additional payload which is not 3536possible via KVM_S390_INTERRUPT. 3537 3538Interrupt parameters are passed via kvm_s390_irq:: 3539 3540 struct kvm_s390_irq { 3541 __u64 type; 3542 union { 3543 struct kvm_s390_io_info io; 3544 struct kvm_s390_ext_info ext; 3545 struct kvm_s390_pgm_info pgm; 3546 struct kvm_s390_emerg_info emerg; 3547 struct kvm_s390_extcall_info extcall; 3548 struct kvm_s390_prefix_info prefix; 3549 struct kvm_s390_stop_info stop; 3550 struct kvm_s390_mchk_info mchk; 3551 char reserved[64]; 3552 } u; 3553 }; 3554 3555type can be one of the following: 3556 3557- KVM_S390_SIGP_STOP - sigp stop; parameter in .stop 3558- KVM_S390_PROGRAM_INT - program check; parameters in .pgm 3559- KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix 3560- KVM_S390_RESTART - restart; no parameters 3561- KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters 3562- KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters 3563- KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg 3564- KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall 3565- KVM_S390_MCHK - machine check interrupt; parameters in .mchk 3566 3567This is an asynchronous vcpu ioctl and can be invoked from any thread. 3568 35694.94 KVM_S390_GET_IRQ_STATE 3570--------------------------- 3571 3572:Capability: KVM_CAP_S390_IRQ_STATE 3573:Architectures: s390 3574:Type: vcpu ioctl 3575:Parameters: struct kvm_s390_irq_state (out) 3576:Returns: >= number of bytes copied into buffer, 3577 -EINVAL if buffer size is 0, 3578 -ENOBUFS if buffer size is too small to fit all pending interrupts, 3579 -EFAULT if the buffer address was invalid 3580 3581This ioctl allows userspace to retrieve the complete state of all currently 3582pending interrupts in a single buffer. Use cases include migration 3583and introspection. The parameter structure contains the address of a 3584userspace buffer and its length:: 3585 3586 struct kvm_s390_irq_state { 3587 __u64 buf; 3588 __u32 flags; /* will stay unused for compatibility reasons */ 3589 __u32 len; 3590 __u32 reserved[4]; /* will stay unused for compatibility reasons */ 3591 }; 3592 3593Userspace passes in the above struct and for each pending interrupt a 3594struct kvm_s390_irq is copied to the provided buffer. 3595 3596The structure contains a flags and a reserved field for future extensions. As 3597the kernel never checked for flags == 0 and QEMU never pre-zeroed flags and 3598reserved, these fields can not be used in the future without breaking 3599compatibility. 3600 3601If -ENOBUFS is returned the buffer provided was too small and userspace 3602may retry with a bigger buffer. 3603 36044.95 KVM_S390_SET_IRQ_STATE 3605--------------------------- 3606 3607:Capability: KVM_CAP_S390_IRQ_STATE 3608:Architectures: s390 3609:Type: vcpu ioctl 3610:Parameters: struct kvm_s390_irq_state (in) 3611:Returns: 0 on success, 3612 -EFAULT if the buffer address was invalid, 3613 -EINVAL for an invalid buffer length (see below), 3614 -EBUSY if there were already interrupts pending, 3615 errors occurring when actually injecting the 3616 interrupt. See KVM_S390_IRQ. 3617 3618This ioctl allows userspace to set the complete state of all cpu-local 3619interrupts currently pending for the vcpu. It is intended for restoring 3620interrupt state after a migration. The input parameter is a userspace buffer 3621containing a struct kvm_s390_irq_state:: 3622 3623 struct kvm_s390_irq_state { 3624 __u64 buf; 3625 __u32 flags; /* will stay unused for compatibility reasons */ 3626 __u32 len; 3627 __u32 reserved[4]; /* will stay unused for compatibility reasons */ 3628 }; 3629 3630The restrictions for flags and reserved apply as well. 3631(see KVM_S390_GET_IRQ_STATE) 3632 3633The userspace memory referenced by buf contains a struct kvm_s390_irq 3634for each interrupt to be injected into the guest. 3635If one of the interrupts could not be injected for some reason the 3636ioctl aborts. 3637 3638len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0 3639and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq), 3640which is the maximum number of possibly pending cpu-local interrupts. 3641 36424.96 KVM_SMI 3643------------ 3644 3645:Capability: KVM_CAP_X86_SMM 3646:Architectures: x86 3647:Type: vcpu ioctl 3648:Parameters: none 3649:Returns: 0 on success, -1 on error 3650 3651Queues an SMI on the thread's vcpu. 3652 36534.97 KVM_CAP_PPC_MULTITCE 3654------------------------- 3655 3656:Capability: KVM_CAP_PPC_MULTITCE 3657:Architectures: ppc 3658:Type: vm 3659 3660This capability means the kernel is capable of handling hypercalls 3661H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user 3662space. This significantly accelerates DMA operations for PPC KVM guests. 3663User space should expect that its handlers for these hypercalls 3664are not going to be called if user space previously registered LIOBN 3665in KVM (via KVM_CREATE_SPAPR_TCE or similar calls). 3666 3667In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest, 3668user space might have to advertise it for the guest. For example, 3669IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is 3670present in the "ibm,hypertas-functions" device-tree property. 3671 3672The hypercalls mentioned above may or may not be processed successfully 3673in the kernel based fast path. If they can not be handled by the kernel, 3674they will get passed on to user space. So user space still has to have 3675an implementation for these despite the in kernel acceleration. 3676 3677This capability is always enabled. 3678 36794.98 KVM_CREATE_SPAPR_TCE_64 3680---------------------------- 3681 3682:Capability: KVM_CAP_SPAPR_TCE_64 3683:Architectures: powerpc 3684:Type: vm ioctl 3685:Parameters: struct kvm_create_spapr_tce_64 (in) 3686:Returns: file descriptor for manipulating the created TCE table 3687 3688This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit 3689windows, described in 4.62 KVM_CREATE_SPAPR_TCE 3690 3691This capability uses extended struct in ioctl interface:: 3692 3693 /* for KVM_CAP_SPAPR_TCE_64 */ 3694 struct kvm_create_spapr_tce_64 { 3695 __u64 liobn; 3696 __u32 page_shift; 3697 __u32 flags; 3698 __u64 offset; /* in pages */ 3699 __u64 size; /* in pages */ 3700 }; 3701 3702The aim of extension is to support an additional bigger DMA window with 3703a variable page size. 3704KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and 3705a bus offset of the corresponding DMA window, @size and @offset are numbers 3706of IOMMU pages. 3707 3708@flags are not used at the moment. 3709 3710The rest of functionality is identical to KVM_CREATE_SPAPR_TCE. 3711 37124.99 KVM_REINJECT_CONTROL 3713------------------------- 3714 3715:Capability: KVM_CAP_REINJECT_CONTROL 3716:Architectures: x86 3717:Type: vm ioctl 3718:Parameters: struct kvm_reinject_control (in) 3719:Returns: 0 on success, 3720 -EFAULT if struct kvm_reinject_control cannot be read, 3721 -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier. 3722 3723i8254 (PIT) has two modes, reinject and !reinject. The default is reinject, 3724where KVM queues elapsed i8254 ticks and monitors completion of interrupt from 3725vector(s) that i8254 injects. Reinject mode dequeues a tick and injects its 3726interrupt whenever there isn't a pending interrupt from i8254. 3727!reinject mode injects an interrupt as soon as a tick arrives. 3728 3729:: 3730 3731 struct kvm_reinject_control { 3732 __u8 pit_reinject; 3733 __u8 reserved[31]; 3734 }; 3735 3736pit_reinject = 0 (!reinject mode) is recommended, unless running an old 3737operating system that uses the PIT for timing (e.g. Linux 2.4.x). 3738 37394.100 KVM_PPC_CONFIGURE_V3_MMU 3740------------------------------ 3741 3742:Capability: KVM_CAP_PPC_RADIX_MMU or KVM_CAP_PPC_HASH_MMU_V3 3743:Architectures: ppc 3744:Type: vm ioctl 3745:Parameters: struct kvm_ppc_mmuv3_cfg (in) 3746:Returns: 0 on success, 3747 -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read, 3748 -EINVAL if the configuration is invalid 3749 3750This ioctl controls whether the guest will use radix or HPT (hashed 3751page table) translation, and sets the pointer to the process table for 3752the guest. 3753 3754:: 3755 3756 struct kvm_ppc_mmuv3_cfg { 3757 __u64 flags; 3758 __u64 process_table; 3759 }; 3760 3761There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and 3762KVM_PPC_MMUV3_GTSE. KVM_PPC_MMUV3_RADIX, if set, configures the guest 3763to use radix tree translation, and if clear, to use HPT translation. 3764KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest 3765to be able to use the global TLB and SLB invalidation instructions; 3766if clear, the guest may not use these instructions. 3767 3768The process_table field specifies the address and size of the guest 3769process table, which is in the guest's space. This field is formatted 3770as the second doubleword of the partition table entry, as defined in 3771the Power ISA V3.00, Book III section 5.7.6.1. 3772 37734.101 KVM_PPC_GET_RMMU_INFO 3774--------------------------- 3775 3776:Capability: KVM_CAP_PPC_RADIX_MMU 3777:Architectures: ppc 3778:Type: vm ioctl 3779:Parameters: struct kvm_ppc_rmmu_info (out) 3780:Returns: 0 on success, 3781 -EFAULT if struct kvm_ppc_rmmu_info cannot be written, 3782 -EINVAL if no useful information can be returned 3783 3784This ioctl returns a structure containing two things: (a) a list 3785containing supported radix tree geometries, and (b) a list that maps 3786page sizes to put in the "AP" (actual page size) field for the tlbie 3787(TLB invalidate entry) instruction. 3788 3789:: 3790 3791 struct kvm_ppc_rmmu_info { 3792 struct kvm_ppc_radix_geom { 3793 __u8 page_shift; 3794 __u8 level_bits[4]; 3795 __u8 pad[3]; 3796 } geometries[8]; 3797 __u32 ap_encodings[8]; 3798 }; 3799 3800The geometries[] field gives up to 8 supported geometries for the 3801radix page table, in terms of the log base 2 of the smallest page 3802size, and the number of bits indexed at each level of the tree, from 3803the PTE level up to the PGD level in that order. Any unused entries 3804will have 0 in the page_shift field. 3805 3806The ap_encodings gives the supported page sizes and their AP field 3807encodings, encoded with the AP value in the top 3 bits and the log 3808base 2 of the page size in the bottom 6 bits. 3809 38104.102 KVM_PPC_RESIZE_HPT_PREPARE 3811-------------------------------- 3812 3813:Capability: KVM_CAP_SPAPR_RESIZE_HPT 3814:Architectures: powerpc 3815:Type: vm ioctl 3816:Parameters: struct kvm_ppc_resize_hpt (in) 3817:Returns: 0 on successful completion, 3818 >0 if a new HPT is being prepared, the value is an estimated 3819 number of milliseconds until preparation is complete, 3820 -EFAULT if struct kvm_reinject_control cannot be read, 3821 -EINVAL if the supplied shift or flags are invalid, 3822 -ENOMEM if unable to allocate the new HPT, 3823 -ENOSPC if there was a hash collision 3824 3825:: 3826 3827 struct kvm_ppc_rmmu_info { 3828 struct kvm_ppc_radix_geom { 3829 __u8 page_shift; 3830 __u8 level_bits[4]; 3831 __u8 pad[3]; 3832 } geometries[8]; 3833 __u32 ap_encodings[8]; 3834 }; 3835 3836The geometries[] field gives up to 8 supported geometries for the 3837radix page table, in terms of the log base 2 of the smallest page 3838size, and the number of bits indexed at each level of the tree, from 3839the PTE level up to the PGD level in that order. Any unused entries 3840will have 0 in the page_shift field. 3841 3842The ap_encodings gives the supported page sizes and their AP field 3843encodings, encoded with the AP value in the top 3 bits and the log 3844base 2 of the page size in the bottom 6 bits. 3845 38464.102 KVM_PPC_RESIZE_HPT_PREPARE 3847-------------------------------- 3848 3849:Capability: KVM_CAP_SPAPR_RESIZE_HPT 3850:Architectures: powerpc 3851:Type: vm ioctl 3852:Parameters: struct kvm_ppc_resize_hpt (in) 3853:Returns: 0 on successful completion, 3854 >0 if a new HPT is being prepared, the value is an estimated 3855 number of milliseconds until preparation is complete, 3856 -EFAULT if struct kvm_reinject_control cannot be read, 3857 -EINVAL if the supplied shift or flags are invalid,when moving existing 3858 HPT entries to the new HPT, 3859 -EIO on other error conditions 3860 3861Used to implement the PAPR extension for runtime resizing of a guest's 3862Hashed Page Table (HPT). Specifically this starts, stops or monitors 3863the preparation of a new potential HPT for the guest, essentially 3864implementing the H_RESIZE_HPT_PREPARE hypercall. 3865 3866If called with shift > 0 when there is no pending HPT for the guest, 3867this begins preparation of a new pending HPT of size 2^(shift) bytes. 3868It then returns a positive integer with the estimated number of 3869milliseconds until preparation is complete. 3870 3871If called when there is a pending HPT whose size does not match that 3872requested in the parameters, discards the existing pending HPT and 3873creates a new one as above. 3874 3875If called when there is a pending HPT of the size requested, will: 3876 3877 * If preparation of the pending HPT is already complete, return 0 3878 * If preparation of the pending HPT has failed, return an error 3879 code, then discard the pending HPT. 3880 * If preparation of the pending HPT is still in progress, return an 3881 estimated number of milliseconds until preparation is complete. 3882 3883If called with shift == 0, discards any currently pending HPT and 3884returns 0 (i.e. cancels any in-progress preparation). 3885 3886flags is reserved for future expansion, currently setting any bits in 3887flags will result in an -EINVAL. 3888 3889Normally this will be called repeatedly with the same parameters until 3890it returns <= 0. The first call will initiate preparation, subsequent 3891ones will monitor preparation until it completes or fails. 3892 3893:: 3894 3895 struct kvm_ppc_resize_hpt { 3896 __u64 flags; 3897 __u32 shift; 3898 __u32 pad; 3899 }; 3900 39014.103 KVM_PPC_RESIZE_HPT_COMMIT 3902------------------------------- 3903 3904:Capability: KVM_CAP_SPAPR_RESIZE_HPT 3905:Architectures: powerpc 3906:Type: vm ioctl 3907:Parameters: struct kvm_ppc_resize_hpt (in) 3908:Returns: 0 on successful completion, 3909 -EFAULT if struct kvm_reinject_control cannot be read, 3910 -EINVAL if the supplied shift or flags are invalid, 3911 -ENXIO is there is no pending HPT, or the pending HPT doesn't 3912 have the requested size, 3913 -EBUSY if the pending HPT is not fully prepared, 3914 -ENOSPC if there was a hash collision when moving existing 3915 HPT entries to the new HPT, 3916 -EIO on other error conditions 3917 3918Used to implement the PAPR extension for runtime resizing of a guest's 3919Hashed Page Table (HPT). Specifically this requests that the guest be 3920transferred to working with the new HPT, essentially implementing the 3921H_RESIZE_HPT_COMMIT hypercall. 3922 3923This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has 3924returned 0 with the same parameters. In other cases 3925KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or 3926-EBUSY, though others may be possible if the preparation was started, 3927but failed). 3928 3929This will have undefined effects on the guest if it has not already 3930placed itself in a quiescent state where no vcpu will make MMU enabled 3931memory accesses. 3932 3933On succsful completion, the pending HPT will become the guest's active 3934HPT and the previous HPT will be discarded. 3935 3936On failure, the guest will still be operating on its previous HPT. 3937 3938:: 3939 3940 struct kvm_ppc_resize_hpt { 3941 __u64 flags; 3942 __u32 shift; 3943 __u32 pad; 3944 }; 3945 39464.104 KVM_X86_GET_MCE_CAP_SUPPORTED 3947----------------------------------- 3948 3949:Capability: KVM_CAP_MCE 3950:Architectures: x86 3951:Type: system ioctl 3952:Parameters: u64 mce_cap (out) 3953:Returns: 0 on success, -1 on error 3954 3955Returns supported MCE capabilities. The u64 mce_cap parameter 3956has the same format as the MSR_IA32_MCG_CAP register. Supported 3957capabilities will have the corresponding bits set. 3958 39594.105 KVM_X86_SETUP_MCE 3960----------------------- 3961 3962:Capability: KVM_CAP_MCE 3963:Architectures: x86 3964:Type: vcpu ioctl 3965:Parameters: u64 mcg_cap (in) 3966:Returns: 0 on success, 3967 -EFAULT if u64 mcg_cap cannot be read, 3968 -EINVAL if the requested number of banks is invalid, 3969 -EINVAL if requested MCE capability is not supported. 3970 3971Initializes MCE support for use. The u64 mcg_cap parameter 3972has the same format as the MSR_IA32_MCG_CAP register and 3973specifies which capabilities should be enabled. The maximum 3974supported number of error-reporting banks can be retrieved when 3975checking for KVM_CAP_MCE. The supported capabilities can be 3976retrieved with KVM_X86_GET_MCE_CAP_SUPPORTED. 3977 39784.106 KVM_X86_SET_MCE 3979--------------------- 3980 3981:Capability: KVM_CAP_MCE 3982:Architectures: x86 3983:Type: vcpu ioctl 3984:Parameters: struct kvm_x86_mce (in) 3985:Returns: 0 on success, 3986 -EFAULT if struct kvm_x86_mce cannot be read, 3987 -EINVAL if the bank number is invalid, 3988 -EINVAL if VAL bit is not set in status field. 3989 3990Inject a machine check error (MCE) into the guest. The input 3991parameter is:: 3992 3993 struct kvm_x86_mce { 3994 __u64 status; 3995 __u64 addr; 3996 __u64 misc; 3997 __u64 mcg_status; 3998 __u8 bank; 3999 __u8 pad1[7]; 4000 __u64 pad2[3]; 4001 }; 4002 4003If the MCE being reported is an uncorrected error, KVM will 4004inject it as an MCE exception into the guest. If the guest 4005MCG_STATUS register reports that an MCE is in progress, KVM 4006causes an KVM_EXIT_SHUTDOWN vmexit. 4007 4008Otherwise, if the MCE is a corrected error, KVM will just 4009store it in the corresponding bank (provided this bank is 4010not holding a previously reported uncorrected error). 4011 40124.107 KVM_S390_GET_CMMA_BITS 4013---------------------------- 4014 4015:Capability: KVM_CAP_S390_CMMA_MIGRATION 4016:Architectures: s390 4017:Type: vm ioctl 4018:Parameters: struct kvm_s390_cmma_log (in, out) 4019:Returns: 0 on success, a negative value on error 4020 4021This ioctl is used to get the values of the CMMA bits on the s390 4022architecture. It is meant to be used in two scenarios: 4023 4024- During live migration to save the CMMA values. Live migration needs 4025 to be enabled via the KVM_REQ_START_MIGRATION VM property. 4026- To non-destructively peek at the CMMA values, with the flag 4027 KVM_S390_CMMA_PEEK set. 4028 4029The ioctl takes parameters via the kvm_s390_cmma_log struct. The desired 4030values are written to a buffer whose location is indicated via the "values" 4031member in the kvm_s390_cmma_log struct. The values in the input struct are 4032also updated as needed. 4033 4034Each CMMA value takes up one byte. 4035 4036:: 4037 4038 struct kvm_s390_cmma_log { 4039 __u64 start_gfn; 4040 __u32 count; 4041 __u32 flags; 4042 union { 4043 __u64 remaining; 4044 __u64 mask; 4045 }; 4046 __u64 values; 4047 }; 4048 4049start_gfn is the number of the first guest frame whose CMMA values are 4050to be retrieved, 4051 4052count is the length of the buffer in bytes, 4053 4054values points to the buffer where the result will be written to. 4055 4056If count is greater than KVM_S390_SKEYS_MAX, then it is considered to be 4057KVM_S390_SKEYS_MAX. KVM_S390_SKEYS_MAX is re-used for consistency with 4058other ioctls. 4059 4060The result is written in the buffer pointed to by the field values, and 4061the values of the input parameter are updated as follows. 4062 4063Depending on the flags, different actions are performed. The only 4064supported flag so far is KVM_S390_CMMA_PEEK. 4065 4066The default behaviour if KVM_S390_CMMA_PEEK is not set is: 4067start_gfn will indicate the first page frame whose CMMA bits were dirty. 4068It is not necessarily the same as the one passed as input, as clean pages 4069are skipped. 4070 4071count will indicate the number of bytes actually written in the buffer. 4072It can (and very often will) be smaller than the input value, since the 4073buffer is only filled until 16 bytes of clean values are found (which 4074are then not copied in the buffer). Since a CMMA migration block needs 4075the base address and the length, for a total of 16 bytes, we will send 4076back some clean data if there is some dirty data afterwards, as long as 4077the size of the clean data does not exceed the size of the header. This 4078allows to minimize the amount of data to be saved or transferred over 4079the network at the expense of more roundtrips to userspace. The next 4080invocation of the ioctl will skip over all the clean values, saving 4081potentially more than just the 16 bytes we found. 4082 4083If KVM_S390_CMMA_PEEK is set: 4084the existing storage attributes are read even when not in migration 4085mode, and no other action is performed; 4086 4087the output start_gfn will be equal to the input start_gfn, 4088 4089the output count will be equal to the input count, except if the end of 4090memory has been reached. 4091 4092In both cases: 4093the field "remaining" will indicate the total number of dirty CMMA values 4094still remaining, or 0 if KVM_S390_CMMA_PEEK is set and migration mode is 4095not enabled. 4096 4097mask is unused. 4098 4099values points to the userspace buffer where the result will be stored. 4100 4101This ioctl can fail with -ENOMEM if not enough memory can be allocated to 4102complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if 4103KVM_S390_CMMA_PEEK is not set but migration mode was not enabled, with 4104-EFAULT if the userspace address is invalid or if no page table is 4105present for the addresses (e.g. when using hugepages). 4106 41074.108 KVM_S390_SET_CMMA_BITS 4108---------------------------- 4109 4110:Capability: KVM_CAP_S390_CMMA_MIGRATION 4111:Architectures: s390 4112:Type: vm ioctl 4113:Parameters: struct kvm_s390_cmma_log (in) 4114:Returns: 0 on success, a negative value on error 4115 4116This ioctl is used to set the values of the CMMA bits on the s390 4117architecture. It is meant to be used during live migration to restore 4118the CMMA values, but there are no restrictions on its use. 4119The ioctl takes parameters via the kvm_s390_cmma_values struct. 4120Each CMMA value takes up one byte. 4121 4122:: 4123 4124 struct kvm_s390_cmma_log { 4125 __u64 start_gfn; 4126 __u32 count; 4127 __u32 flags; 4128 union { 4129 __u64 remaining; 4130 __u64 mask; 4131 }; 4132 __u64 values; 4133 }; 4134 4135start_gfn indicates the starting guest frame number, 4136 4137count indicates how many values are to be considered in the buffer, 4138 4139flags is not used and must be 0. 4140 4141mask indicates which PGSTE bits are to be considered. 4142 4143remaining is not used. 4144 4145values points to the buffer in userspace where to store the values. 4146 4147This ioctl can fail with -ENOMEM if not enough memory can be allocated to 4148complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if 4149the count field is too large (e.g. more than KVM_S390_CMMA_SIZE_MAX) or 4150if the flags field was not 0, with -EFAULT if the userspace address is 4151invalid, if invalid pages are written to (e.g. after the end of memory) 4152or if no page table is present for the addresses (e.g. when using 4153hugepages). 4154 41554.109 KVM_PPC_GET_CPU_CHAR 4156-------------------------- 4157 4158:Capability: KVM_CAP_PPC_GET_CPU_CHAR 4159:Architectures: powerpc 4160:Type: vm ioctl 4161:Parameters: struct kvm_ppc_cpu_char (out) 4162:Returns: 0 on successful completion, 4163 -EFAULT if struct kvm_ppc_cpu_char cannot be written 4164 4165This ioctl gives userspace information about certain characteristics 4166of the CPU relating to speculative execution of instructions and 4167possible information leakage resulting from speculative execution (see 4168CVE-2017-5715, CVE-2017-5753 and CVE-2017-5754). The information is 4169returned in struct kvm_ppc_cpu_char, which looks like this:: 4170 4171 struct kvm_ppc_cpu_char { 4172 __u64 character; /* characteristics of the CPU */ 4173 __u64 behaviour; /* recommended software behaviour */ 4174 __u64 character_mask; /* valid bits in character */ 4175 __u64 behaviour_mask; /* valid bits in behaviour */ 4176 }; 4177 4178For extensibility, the character_mask and behaviour_mask fields 4179indicate which bits of character and behaviour have been filled in by 4180the kernel. If the set of defined bits is extended in future then 4181userspace will be able to tell whether it is running on a kernel that 4182knows about the new bits. 4183 4184The character field describes attributes of the CPU which can help 4185with preventing inadvertent information disclosure - specifically, 4186whether there is an instruction to flash-invalidate the L1 data cache 4187(ori 30,30,0 or mtspr SPRN_TRIG2,rN), whether the L1 data cache is set 4188to a mode where entries can only be used by the thread that created 4189them, whether the bcctr[l] instruction prevents speculation, and 4190whether a speculation barrier instruction (ori 31,31,0) is provided. 4191 4192The behaviour field describes actions that software should take to 4193prevent inadvertent information disclosure, and thus describes which 4194vulnerabilities the hardware is subject to; specifically whether the 4195L1 data cache should be flushed when returning to user mode from the 4196kernel, and whether a speculation barrier should be placed between an 4197array bounds check and the array access. 4198 4199These fields use the same bit definitions as the new 4200H_GET_CPU_CHARACTERISTICS hypercall. 4201 42024.110 KVM_MEMORY_ENCRYPT_OP 4203--------------------------- 4204 4205:Capability: basic 4206:Architectures: x86 4207:Type: system 4208:Parameters: an opaque platform specific structure (in/out) 4209:Returns: 0 on success; -1 on error 4210 4211If the platform supports creating encrypted VMs then this ioctl can be used 4212for issuing platform-specific memory encryption commands to manage those 4213encrypted VMs. 4214 4215Currently, this ioctl is used for issuing Secure Encrypted Virtualization 4216(SEV) commands on AMD Processors. The SEV commands are defined in 4217Documentation/virt/kvm/amd-memory-encryption.rst. 4218 42194.111 KVM_MEMORY_ENCRYPT_REG_REGION 4220----------------------------------- 4221 4222:Capability: basic 4223:Architectures: x86 4224:Type: system 4225:Parameters: struct kvm_enc_region (in) 4226:Returns: 0 on success; -1 on error 4227 4228This ioctl can be used to register a guest memory region which may 4229contain encrypted data (e.g. guest RAM, SMRAM etc). 4230 4231It is used in the SEV-enabled guest. When encryption is enabled, a guest 4232memory region may contain encrypted data. The SEV memory encryption 4233engine uses a tweak such that two identical plaintext pages, each at 4234different locations will have differing ciphertexts. So swapping or 4235moving ciphertext of those pages will not result in plaintext being 4236swapped. So relocating (or migrating) physical backing pages for the SEV 4237guest will require some additional steps. 4238 4239Note: The current SEV key management spec does not provide commands to 4240swap or migrate (move) ciphertext pages. Hence, for now we pin the guest 4241memory region registered with the ioctl. 4242 42434.112 KVM_MEMORY_ENCRYPT_UNREG_REGION 4244------------------------------------- 4245 4246:Capability: basic 4247:Architectures: x86 4248:Type: system 4249:Parameters: struct kvm_enc_region (in) 4250:Returns: 0 on success; -1 on error 4251 4252This ioctl can be used to unregister the guest memory region registered 4253with KVM_MEMORY_ENCRYPT_REG_REGION ioctl above. 4254 42554.113 KVM_HYPERV_EVENTFD 4256------------------------ 4257 4258:Capability: KVM_CAP_HYPERV_EVENTFD 4259:Architectures: x86 4260:Type: vm ioctl 4261:Parameters: struct kvm_hyperv_eventfd (in) 4262 4263This ioctl (un)registers an eventfd to receive notifications from the guest on 4264the specified Hyper-V connection id through the SIGNAL_EVENT hypercall, without 4265causing a user exit. SIGNAL_EVENT hypercall with non-zero event flag number 4266(bits 24-31) still triggers a KVM_EXIT_HYPERV_HCALL user exit. 4267 4268:: 4269 4270 struct kvm_hyperv_eventfd { 4271 __u32 conn_id; 4272 __s32 fd; 4273 __u32 flags; 4274 __u32 padding[3]; 4275 }; 4276 4277The conn_id field should fit within 24 bits:: 4278 4279 #define KVM_HYPERV_CONN_ID_MASK 0x00ffffff 4280 4281The acceptable values for the flags field are:: 4282 4283 #define KVM_HYPERV_EVENTFD_DEASSIGN (1 << 0) 4284 4285:Returns: 0 on success, 4286 -EINVAL if conn_id or flags is outside the allowed range, 4287 -ENOENT on deassign if the conn_id isn't registered, 4288 -EEXIST on assign if the conn_id is already registered 4289 42904.114 KVM_GET_NESTED_STATE 4291-------------------------- 4292 4293:Capability: KVM_CAP_NESTED_STATE 4294:Architectures: x86 4295:Type: vcpu ioctl 4296:Parameters: struct kvm_nested_state (in/out) 4297:Returns: 0 on success, -1 on error 4298 4299Errors: 4300 4301 ===== ============================================================= 4302 E2BIG the total state size exceeds the value of 'size' specified by 4303 the user; the size required will be written into size. 4304 ===== ============================================================= 4305 4306:: 4307 4308 struct kvm_nested_state { 4309 __u16 flags; 4310 __u16 format; 4311 __u32 size; 4312 4313 union { 4314 struct kvm_vmx_nested_state_hdr vmx; 4315 struct kvm_svm_nested_state_hdr svm; 4316 4317 /* Pad the header to 128 bytes. */ 4318 __u8 pad[120]; 4319 } hdr; 4320 4321 union { 4322 struct kvm_vmx_nested_state_data vmx[0]; 4323 struct kvm_svm_nested_state_data svm[0]; 4324 } data; 4325 }; 4326 4327 #define KVM_STATE_NESTED_GUEST_MODE 0x00000001 4328 #define KVM_STATE_NESTED_RUN_PENDING 0x00000002 4329 #define KVM_STATE_NESTED_EVMCS 0x00000004 4330 4331 #define KVM_STATE_NESTED_FORMAT_VMX 0 4332 #define KVM_STATE_NESTED_FORMAT_SVM 1 4333 4334 #define KVM_STATE_NESTED_VMX_VMCS_SIZE 0x1000 4335 4336 #define KVM_STATE_NESTED_VMX_SMM_GUEST_MODE 0x00000001 4337 #define KVM_STATE_NESTED_VMX_SMM_VMXON 0x00000002 4338 4339#define KVM_STATE_VMX_PREEMPTION_TIMER_DEADLINE 0x00000001 4340 4341 struct kvm_vmx_nested_state_hdr { 4342 __u64 vmxon_pa; 4343 __u64 vmcs12_pa; 4344 4345 struct { 4346 __u16 flags; 4347 } smm; 4348 4349 __u32 flags; 4350 __u64 preemption_timer_deadline; 4351 }; 4352 4353 struct kvm_vmx_nested_state_data { 4354 __u8 vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE]; 4355 __u8 shadow_vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE]; 4356 }; 4357 4358This ioctl copies the vcpu's nested virtualization state from the kernel to 4359userspace. 4360 4361The maximum size of the state can be retrieved by passing KVM_CAP_NESTED_STATE 4362to the KVM_CHECK_EXTENSION ioctl(). 4363 43644.115 KVM_SET_NESTED_STATE 4365-------------------------- 4366 4367:Capability: KVM_CAP_NESTED_STATE 4368:Architectures: x86 4369:Type: vcpu ioctl 4370:Parameters: struct kvm_nested_state (in) 4371:Returns: 0 on success, -1 on error 4372 4373This copies the vcpu's kvm_nested_state struct from userspace to the kernel. 4374For the definition of struct kvm_nested_state, see KVM_GET_NESTED_STATE. 4375 43764.116 KVM_(UN)REGISTER_COALESCED_MMIO 4377------------------------------------- 4378 4379:Capability: KVM_CAP_COALESCED_MMIO (for coalesced mmio) 4380 KVM_CAP_COALESCED_PIO (for coalesced pio) 4381:Architectures: all 4382:Type: vm ioctl 4383:Parameters: struct kvm_coalesced_mmio_zone 4384:Returns: 0 on success, < 0 on error 4385 4386Coalesced I/O is a performance optimization that defers hardware 4387register write emulation so that userspace exits are avoided. It is 4388typically used to reduce the overhead of emulating frequently accessed 4389hardware registers. 4390 4391When a hardware register is configured for coalesced I/O, write accesses 4392do not exit to userspace and their value is recorded in a ring buffer 4393that is shared between kernel and userspace. 4394 4395Coalesced I/O is used if one or more write accesses to a hardware 4396register can be deferred until a read or a write to another hardware 4397register on the same device. This last access will cause a vmexit and 4398userspace will process accesses from the ring buffer before emulating 4399it. That will avoid exiting to userspace on repeated writes. 4400 4401Coalesced pio is based on coalesced mmio. There is little difference 4402between coalesced mmio and pio except that coalesced pio records accesses 4403to I/O ports. 4404 44054.117 KVM_CLEAR_DIRTY_LOG (vm ioctl) 4406------------------------------------ 4407 4408:Capability: KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 4409:Architectures: x86, arm, arm64, mips 4410:Type: vm ioctl 4411:Parameters: struct kvm_dirty_log (in) 4412:Returns: 0 on success, -1 on error 4413 4414:: 4415 4416 /* for KVM_CLEAR_DIRTY_LOG */ 4417 struct kvm_clear_dirty_log { 4418 __u32 slot; 4419 __u32 num_pages; 4420 __u64 first_page; 4421 union { 4422 void __user *dirty_bitmap; /* one bit per page */ 4423 __u64 padding; 4424 }; 4425 }; 4426 4427The ioctl clears the dirty status of pages in a memory slot, according to 4428the bitmap that is passed in struct kvm_clear_dirty_log's dirty_bitmap 4429field. Bit 0 of the bitmap corresponds to page "first_page" in the 4430memory slot, and num_pages is the size in bits of the input bitmap. 4431first_page must be a multiple of 64; num_pages must also be a multiple of 443264 unless first_page + num_pages is the size of the memory slot. For each 4433bit that is set in the input bitmap, the corresponding page is marked "clean" 4434in KVM's dirty bitmap, and dirty tracking is re-enabled for that page 4435(for example via write-protection, or by clearing the dirty bit in 4436a page table entry). 4437 4438If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 specifies 4439the address space for which you want to return the dirty bitmap. 4440They must be less than the value that KVM_CHECK_EXTENSION returns for 4441the KVM_CAP_MULTI_ADDRESS_SPACE capability. 4442 4443This ioctl is mostly useful when KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 4444is enabled; for more information, see the description of the capability. 4445However, it can always be used as long as KVM_CHECK_EXTENSION confirms 4446that KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is present. 4447 44484.118 KVM_GET_SUPPORTED_HV_CPUID 4449-------------------------------- 4450 4451:Capability: KVM_CAP_HYPERV_CPUID 4452:Architectures: x86 4453:Type: vcpu ioctl 4454:Parameters: struct kvm_cpuid2 (in/out) 4455:Returns: 0 on success, -1 on error 4456 4457:: 4458 4459 struct kvm_cpuid2 { 4460 __u32 nent; 4461 __u32 padding; 4462 struct kvm_cpuid_entry2 entries[0]; 4463 }; 4464 4465 struct kvm_cpuid_entry2 { 4466 __u32 function; 4467 __u32 index; 4468 __u32 flags; 4469 __u32 eax; 4470 __u32 ebx; 4471 __u32 ecx; 4472 __u32 edx; 4473 __u32 padding[3]; 4474 }; 4475 4476This ioctl returns x86 cpuid features leaves related to Hyper-V emulation in 4477KVM. Userspace can use the information returned by this ioctl to construct 4478cpuid information presented to guests consuming Hyper-V enlightenments (e.g. 4479Windows or Hyper-V guests). 4480 4481CPUID feature leaves returned by this ioctl are defined by Hyper-V Top Level 4482Functional Specification (TLFS). These leaves can't be obtained with 4483KVM_GET_SUPPORTED_CPUID ioctl because some of them intersect with KVM feature 4484leaves (0x40000000, 0x40000001). 4485 4486Currently, the following list of CPUID leaves are returned: 4487 - HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS 4488 - HYPERV_CPUID_INTERFACE 4489 - HYPERV_CPUID_VERSION 4490 - HYPERV_CPUID_FEATURES 4491 - HYPERV_CPUID_ENLIGHTMENT_INFO 4492 - HYPERV_CPUID_IMPLEMENT_LIMITS 4493 - HYPERV_CPUID_NESTED_FEATURES 4494 4495HYPERV_CPUID_NESTED_FEATURES leaf is only exposed when Enlightened VMCS was 4496enabled on the corresponding vCPU (KVM_CAP_HYPERV_ENLIGHTENED_VMCS). 4497 4498Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure 4499with the 'nent' field indicating the number of entries in the variable-size 4500array 'entries'. If the number of entries is too low to describe all Hyper-V 4501feature leaves, an error (E2BIG) is returned. If the number is more or equal 4502to the number of Hyper-V feature leaves, the 'nent' field is adjusted to the 4503number of valid entries in the 'entries' array, which is then filled. 4504 4505'index' and 'flags' fields in 'struct kvm_cpuid_entry2' are currently reserved, 4506userspace should not expect to get any particular value there. 4507 45084.119 KVM_ARM_VCPU_FINALIZE 4509--------------------------- 4510 4511:Architectures: arm, arm64 4512:Type: vcpu ioctl 4513:Parameters: int feature (in) 4514:Returns: 0 on success, -1 on error 4515 4516Errors: 4517 4518 ====== ============================================================== 4519 EPERM feature not enabled, needs configuration, or already finalized 4520 EINVAL feature unknown or not present 4521 ====== ============================================================== 4522 4523Recognised values for feature: 4524 4525 ===== =========================================== 4526 arm64 KVM_ARM_VCPU_SVE (requires KVM_CAP_ARM_SVE) 4527 ===== =========================================== 4528 4529Finalizes the configuration of the specified vcpu feature. 4530 4531The vcpu must already have been initialised, enabling the affected feature, by 4532means of a successful KVM_ARM_VCPU_INIT call with the appropriate flag set in 4533features[]. 4534 4535For affected vcpu features, this is a mandatory step that must be performed 4536before the vcpu is fully usable. 4537 4538Between KVM_ARM_VCPU_INIT and KVM_ARM_VCPU_FINALIZE, the feature may be 4539configured by use of ioctls such as KVM_SET_ONE_REG. The exact configuration 4540that should be performaned and how to do it are feature-dependent. 4541 4542Other calls that depend on a particular feature being finalized, such as 4543KVM_RUN, KVM_GET_REG_LIST, KVM_GET_ONE_REG and KVM_SET_ONE_REG, will fail with 4544-EPERM unless the feature has already been finalized by means of a 4545KVM_ARM_VCPU_FINALIZE call. 4546 4547See KVM_ARM_VCPU_INIT for details of vcpu features that require finalization 4548using this ioctl. 4549 45504.120 KVM_SET_PMU_EVENT_FILTER 4551------------------------------ 4552 4553:Capability: KVM_CAP_PMU_EVENT_FILTER 4554:Architectures: x86 4555:Type: vm ioctl 4556:Parameters: struct kvm_pmu_event_filter (in) 4557:Returns: 0 on success, -1 on error 4558 4559:: 4560 4561 struct kvm_pmu_event_filter { 4562 __u32 action; 4563 __u32 nevents; 4564 __u32 fixed_counter_bitmap; 4565 __u32 flags; 4566 __u32 pad[4]; 4567 __u64 events[0]; 4568 }; 4569 4570This ioctl restricts the set of PMU events that the guest can program. 4571The argument holds a list of events which will be allowed or denied. 4572The eventsel+umask of each event the guest attempts to program is compared 4573against the events field to determine whether the guest should have access. 4574The events field only controls general purpose counters; fixed purpose 4575counters are controlled by the fixed_counter_bitmap. 4576 4577No flags are defined yet, the field must be zero. 4578 4579Valid values for 'action':: 4580 4581 #define KVM_PMU_EVENT_ALLOW 0 4582 #define KVM_PMU_EVENT_DENY 1 4583 45844.121 KVM_PPC_SVM_OFF 4585--------------------- 4586 4587:Capability: basic 4588:Architectures: powerpc 4589:Type: vm ioctl 4590:Parameters: none 4591:Returns: 0 on successful completion, 4592 4593Errors: 4594 4595 ====== ================================================================ 4596 EINVAL if ultravisor failed to terminate the secure guest 4597 ENOMEM if hypervisor failed to allocate new radix page tables for guest 4598 ====== ================================================================ 4599 4600This ioctl is used to turn off the secure mode of the guest or transition 4601the guest from secure mode to normal mode. This is invoked when the guest 4602is reset. This has no effect if called for a normal guest. 4603 4604This ioctl issues an ultravisor call to terminate the secure guest, 4605unpins the VPA pages and releases all the device pages that are used to 4606track the secure pages by hypervisor. 4607 46084.122 KVM_S390_NORMAL_RESET 4609--------------------------- 4610 4611:Capability: KVM_CAP_S390_VCPU_RESETS 4612:Architectures: s390 4613:Type: vcpu ioctl 4614:Parameters: none 4615:Returns: 0 4616 4617This ioctl resets VCPU registers and control structures according to 4618the cpu reset definition in the POP (Principles Of Operation). 4619 46204.123 KVM_S390_INITIAL_RESET 4621---------------------------- 4622 4623:Capability: none 4624:Architectures: s390 4625:Type: vcpu ioctl 4626:Parameters: none 4627:Returns: 0 4628 4629This ioctl resets VCPU registers and control structures according to 4630the initial cpu reset definition in the POP. However, the cpu is not 4631put into ESA mode. This reset is a superset of the normal reset. 4632 46334.124 KVM_S390_CLEAR_RESET 4634-------------------------- 4635 4636:Capability: KVM_CAP_S390_VCPU_RESETS 4637:Architectures: s390 4638:Type: vcpu ioctl 4639:Parameters: none 4640:Returns: 0 4641 4642This ioctl resets VCPU registers and control structures according to 4643the clear cpu reset definition in the POP. However, the cpu is not put 4644into ESA mode. This reset is a superset of the initial reset. 4645 4646 46474.125 KVM_S390_PV_COMMAND 4648------------------------- 4649 4650:Capability: KVM_CAP_S390_PROTECTED 4651:Architectures: s390 4652:Type: vm ioctl 4653:Parameters: struct kvm_pv_cmd 4654:Returns: 0 on success, < 0 on error 4655 4656:: 4657 4658 struct kvm_pv_cmd { 4659 __u32 cmd; /* Command to be executed */ 4660 __u16 rc; /* Ultravisor return code */ 4661 __u16 rrc; /* Ultravisor return reason code */ 4662 __u64 data; /* Data or address */ 4663 __u32 flags; /* flags for future extensions. Must be 0 for now */ 4664 __u32 reserved[3]; 4665 }; 4666 4667cmd values: 4668 4669KVM_PV_ENABLE 4670 Allocate memory and register the VM with the Ultravisor, thereby 4671 donating memory to the Ultravisor that will become inaccessible to 4672 KVM. All existing CPUs are converted to protected ones. After this 4673 command has succeeded, any CPU added via hotplug will become 4674 protected during its creation as well. 4675 4676 Errors: 4677 4678 ===== ============================= 4679 EINTR an unmasked signal is pending 4680 ===== ============================= 4681 4682KVM_PV_DISABLE 4683 4684 Deregister the VM from the Ultravisor and reclaim the memory that 4685 had been donated to the Ultravisor, making it usable by the kernel 4686 again. All registered VCPUs are converted back to non-protected 4687 ones. 4688 4689KVM_PV_VM_SET_SEC_PARMS 4690 Pass the image header from VM memory to the Ultravisor in 4691 preparation of image unpacking and verification. 4692 4693KVM_PV_VM_UNPACK 4694 Unpack (protect and decrypt) a page of the encrypted boot image. 4695 4696KVM_PV_VM_VERIFY 4697 Verify the integrity of the unpacked image. Only if this succeeds, 4698 KVM is allowed to start protected VCPUs. 4699 4700 47015. The kvm_run structure 4702======================== 4703 4704Application code obtains a pointer to the kvm_run structure by 4705mmap()ing a vcpu fd. From that point, application code can control 4706execution by changing fields in kvm_run prior to calling the KVM_RUN 4707ioctl, and obtain information about the reason KVM_RUN returned by 4708looking up structure members. 4709 4710:: 4711 4712 struct kvm_run { 4713 /* in */ 4714 __u8 request_interrupt_window; 4715 4716Request that KVM_RUN return when it becomes possible to inject external 4717interrupts into the guest. Useful in conjunction with KVM_INTERRUPT. 4718 4719:: 4720 4721 __u8 immediate_exit; 4722 4723This field is polled once when KVM_RUN starts; if non-zero, KVM_RUN 4724exits immediately, returning -EINTR. In the common scenario where a 4725signal is used to "kick" a VCPU out of KVM_RUN, this field can be used 4726to avoid usage of KVM_SET_SIGNAL_MASK, which has worse scalability. 4727Rather than blocking the signal outside KVM_RUN, userspace can set up 4728a signal handler that sets run->immediate_exit to a non-zero value. 4729 4730This field is ignored if KVM_CAP_IMMEDIATE_EXIT is not available. 4731 4732:: 4733 4734 __u8 padding1[6]; 4735 4736 /* out */ 4737 __u32 exit_reason; 4738 4739When KVM_RUN has returned successfully (return value 0), this informs 4740application code why KVM_RUN has returned. Allowable values for this 4741field are detailed below. 4742 4743:: 4744 4745 __u8 ready_for_interrupt_injection; 4746 4747If request_interrupt_window has been specified, this field indicates 4748an interrupt can be injected now with KVM_INTERRUPT. 4749 4750:: 4751 4752 __u8 if_flag; 4753 4754The value of the current interrupt flag. Only valid if in-kernel 4755local APIC is not used. 4756 4757:: 4758 4759 __u16 flags; 4760 4761More architecture-specific flags detailing state of the VCPU that may 4762affect the device's behavior. The only currently defined flag is 4763KVM_RUN_X86_SMM, which is valid on x86 machines and is set if the 4764VCPU is in system management mode. 4765 4766:: 4767 4768 /* in (pre_kvm_run), out (post_kvm_run) */ 4769 __u64 cr8; 4770 4771The value of the cr8 register. Only valid if in-kernel local APIC is 4772not used. Both input and output. 4773 4774:: 4775 4776 __u64 apic_base; 4777 4778The value of the APIC BASE msr. Only valid if in-kernel local 4779APIC is not used. Both input and output. 4780 4781:: 4782 4783 union { 4784 /* KVM_EXIT_UNKNOWN */ 4785 struct { 4786 __u64 hardware_exit_reason; 4787 } hw; 4788 4789If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown 4790reasons. Further architecture-specific information is available in 4791hardware_exit_reason. 4792 4793:: 4794 4795 /* KVM_EXIT_FAIL_ENTRY */ 4796 struct { 4797 __u64 hardware_entry_failure_reason; 4798 } fail_entry; 4799 4800If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due 4801to unknown reasons. Further architecture-specific information is 4802available in hardware_entry_failure_reason. 4803 4804:: 4805 4806 /* KVM_EXIT_EXCEPTION */ 4807 struct { 4808 __u32 exception; 4809 __u32 error_code; 4810 } ex; 4811 4812Unused. 4813 4814:: 4815 4816 /* KVM_EXIT_IO */ 4817 struct { 4818 #define KVM_EXIT_IO_IN 0 4819 #define KVM_EXIT_IO_OUT 1 4820 __u8 direction; 4821 __u8 size; /* bytes */ 4822 __u16 port; 4823 __u32 count; 4824 __u64 data_offset; /* relative to kvm_run start */ 4825 } io; 4826 4827If exit_reason is KVM_EXIT_IO, then the vcpu has 4828executed a port I/O instruction which could not be satisfied by kvm. 4829data_offset describes where the data is located (KVM_EXIT_IO_OUT) or 4830where kvm expects application code to place the data for the next 4831KVM_RUN invocation (KVM_EXIT_IO_IN). Data format is a packed array. 4832 4833:: 4834 4835 /* KVM_EXIT_DEBUG */ 4836 struct { 4837 struct kvm_debug_exit_arch arch; 4838 } debug; 4839 4840If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event 4841for which architecture specific information is returned. 4842 4843:: 4844 4845 /* KVM_EXIT_MMIO */ 4846 struct { 4847 __u64 phys_addr; 4848 __u8 data[8]; 4849 __u32 len; 4850 __u8 is_write; 4851 } mmio; 4852 4853If exit_reason is KVM_EXIT_MMIO, then the vcpu has 4854executed a memory-mapped I/O instruction which could not be satisfied 4855by kvm. The 'data' member contains the written data if 'is_write' is 4856true, and should be filled by application code otherwise. 4857 4858The 'data' member contains, in its first 'len' bytes, the value as it would 4859appear if the VCPU performed a load or store of the appropriate width directly 4860to the byte array. 4861 4862.. note:: 4863 4864 For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR and 4865 KVM_EXIT_EPR the corresponding 4866 4867operations are complete (and guest state is consistent) only after userspace 4868has re-entered the kernel with KVM_RUN. The kernel side will first finish 4869incomplete operations and then check for pending signals. Userspace 4870can re-enter the guest with an unmasked signal pending to complete 4871pending operations. 4872 4873:: 4874 4875 /* KVM_EXIT_HYPERCALL */ 4876 struct { 4877 __u64 nr; 4878 __u64 args[6]; 4879 __u64 ret; 4880 __u32 longmode; 4881 __u32 pad; 4882 } hypercall; 4883 4884Unused. This was once used for 'hypercall to userspace'. To implement 4885such functionality, use KVM_EXIT_IO (x86) or KVM_EXIT_MMIO (all except s390). 4886 4887.. note:: KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO. 4888 4889:: 4890 4891 /* KVM_EXIT_TPR_ACCESS */ 4892 struct { 4893 __u64 rip; 4894 __u32 is_write; 4895 __u32 pad; 4896 } tpr_access; 4897 4898To be documented (KVM_TPR_ACCESS_REPORTING). 4899 4900:: 4901 4902 /* KVM_EXIT_S390_SIEIC */ 4903 struct { 4904 __u8 icptcode; 4905 __u64 mask; /* psw upper half */ 4906 __u64 addr; /* psw lower half */ 4907 __u16 ipa; 4908 __u32 ipb; 4909 } s390_sieic; 4910 4911s390 specific. 4912 4913:: 4914 4915 /* KVM_EXIT_S390_RESET */ 4916 #define KVM_S390_RESET_POR 1 4917 #define KVM_S390_RESET_CLEAR 2 4918 #define KVM_S390_RESET_SUBSYSTEM 4 4919 #define KVM_S390_RESET_CPU_INIT 8 4920 #define KVM_S390_RESET_IPL 16 4921 __u64 s390_reset_flags; 4922 4923s390 specific. 4924 4925:: 4926 4927 /* KVM_EXIT_S390_UCONTROL */ 4928 struct { 4929 __u64 trans_exc_code; 4930 __u32 pgm_code; 4931 } s390_ucontrol; 4932 4933s390 specific. A page fault has occurred for a user controlled virtual 4934machine (KVM_VM_S390_UNCONTROL) on it's host page table that cannot be 4935resolved by the kernel. 4936The program code and the translation exception code that were placed 4937in the cpu's lowcore are presented here as defined by the z Architecture 4938Principles of Operation Book in the Chapter for Dynamic Address Translation 4939(DAT) 4940 4941:: 4942 4943 /* KVM_EXIT_DCR */ 4944 struct { 4945 __u32 dcrn; 4946 __u32 data; 4947 __u8 is_write; 4948 } dcr; 4949 4950Deprecated - was used for 440 KVM. 4951 4952:: 4953 4954 /* KVM_EXIT_OSI */ 4955 struct { 4956 __u64 gprs[32]; 4957 } osi; 4958 4959MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch 4960hypercalls and exit with this exit struct that contains all the guest gprs. 4961 4962If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall. 4963Userspace can now handle the hypercall and when it's done modify the gprs as 4964necessary. Upon guest entry all guest GPRs will then be replaced by the values 4965in this struct. 4966 4967:: 4968 4969 /* KVM_EXIT_PAPR_HCALL */ 4970 struct { 4971 __u64 nr; 4972 __u64 ret; 4973 __u64 args[9]; 4974 } papr_hcall; 4975 4976This is used on 64-bit PowerPC when emulating a pSeries partition, 4977e.g. with the 'pseries' machine type in qemu. It occurs when the 4978guest does a hypercall using the 'sc 1' instruction. The 'nr' field 4979contains the hypercall number (from the guest R3), and 'args' contains 4980the arguments (from the guest R4 - R12). Userspace should put the 4981return code in 'ret' and any extra returned values in args[]. 4982The possible hypercalls are defined in the Power Architecture Platform 4983Requirements (PAPR) document available from www.power.org (free 4984developer registration required to access it). 4985 4986:: 4987 4988 /* KVM_EXIT_S390_TSCH */ 4989 struct { 4990 __u16 subchannel_id; 4991 __u16 subchannel_nr; 4992 __u32 io_int_parm; 4993 __u32 io_int_word; 4994 __u32 ipb; 4995 __u8 dequeued; 4996 } s390_tsch; 4997 4998s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled 4999and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O 5000interrupt for the target subchannel has been dequeued and subchannel_id, 5001subchannel_nr, io_int_parm and io_int_word contain the parameters for that 5002interrupt. ipb is needed for instruction parameter decoding. 5003 5004:: 5005 5006 /* KVM_EXIT_EPR */ 5007 struct { 5008 __u32 epr; 5009 } epr; 5010 5011On FSL BookE PowerPC chips, the interrupt controller has a fast patch 5012interrupt acknowledge path to the core. When the core successfully 5013delivers an interrupt, it automatically populates the EPR register with 5014the interrupt vector number and acknowledges the interrupt inside 5015the interrupt controller. 5016 5017In case the interrupt controller lives in user space, we need to do 5018the interrupt acknowledge cycle through it to fetch the next to be 5019delivered interrupt vector using this exit. 5020 5021It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an 5022external interrupt has just been delivered into the guest. User space 5023should put the acknowledged interrupt vector into the 'epr' field. 5024 5025:: 5026 5027 /* KVM_EXIT_SYSTEM_EVENT */ 5028 struct { 5029 #define KVM_SYSTEM_EVENT_SHUTDOWN 1 5030 #define KVM_SYSTEM_EVENT_RESET 2 5031 #define KVM_SYSTEM_EVENT_CRASH 3 5032 __u32 type; 5033 __u64 flags; 5034 } system_event; 5035 5036If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered 5037a system-level event using some architecture specific mechanism (hypercall 5038or some special instruction). In case of ARM/ARM64, this is triggered using 5039HVC instruction based PSCI call from the vcpu. The 'type' field describes 5040the system-level event type. The 'flags' field describes architecture 5041specific flags for the system-level event. 5042 5043Valid values for 'type' are: 5044 5045 - KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the 5046 VM. Userspace is not obliged to honour this, and if it does honour 5047 this does not need to destroy the VM synchronously (ie it may call 5048 KVM_RUN again before shutdown finally occurs). 5049 - KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM. 5050 As with SHUTDOWN, userspace can choose to ignore the request, or 5051 to schedule the reset to occur in the future and may call KVM_RUN again. 5052 - KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest 5053 has requested a crash condition maintenance. Userspace can choose 5054 to ignore the request, or to gather VM memory core dump and/or 5055 reset/shutdown of the VM. 5056 5057:: 5058 5059 /* KVM_EXIT_IOAPIC_EOI */ 5060 struct { 5061 __u8 vector; 5062 } eoi; 5063 5064Indicates that the VCPU's in-kernel local APIC received an EOI for a 5065level-triggered IOAPIC interrupt. This exit only triggers when the 5066IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled); 5067the userspace IOAPIC should process the EOI and retrigger the interrupt if 5068it is still asserted. Vector is the LAPIC interrupt vector for which the 5069EOI was received. 5070 5071:: 5072 5073 struct kvm_hyperv_exit { 5074 #define KVM_EXIT_HYPERV_SYNIC 1 5075 #define KVM_EXIT_HYPERV_HCALL 2 5076 #define KVM_EXIT_HYPERV_SYNDBG 3 5077 __u32 type; 5078 __u32 pad1; 5079 union { 5080 struct { 5081 __u32 msr; 5082 __u32 pad2; 5083 __u64 control; 5084 __u64 evt_page; 5085 __u64 msg_page; 5086 } synic; 5087 struct { 5088 __u64 input; 5089 __u64 result; 5090 __u64 params[2]; 5091 } hcall; 5092 struct { 5093 __u32 msr; 5094 __u32 pad2; 5095 __u64 control; 5096 __u64 status; 5097 __u64 send_page; 5098 __u64 recv_page; 5099 __u64 pending_page; 5100 } syndbg; 5101 } u; 5102 }; 5103 /* KVM_EXIT_HYPERV */ 5104 struct kvm_hyperv_exit hyperv; 5105 5106Indicates that the VCPU exits into userspace to process some tasks 5107related to Hyper-V emulation. 5108 5109Valid values for 'type' are: 5110 5111 - KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about 5112 5113Hyper-V SynIC state change. Notification is used to remap SynIC 5114event/message pages and to enable/disable SynIC messages/events processing 5115in userspace. 5116 5117 - KVM_EXIT_HYPERV_SYNDBG -- synchronously notify user-space about 5118 5119Hyper-V Synthetic debugger state change. Notification is used to either update 5120the pending_page location or to send a control command (send the buffer located 5121in send_page or recv a buffer to recv_page). 5122 5123:: 5124 5125 /* KVM_EXIT_ARM_NISV */ 5126 struct { 5127 __u64 esr_iss; 5128 __u64 fault_ipa; 5129 } arm_nisv; 5130 5131Used on arm and arm64 systems. If a guest accesses memory not in a memslot, 5132KVM will typically return to userspace and ask it to do MMIO emulation on its 5133behalf. However, for certain classes of instructions, no instruction decode 5134(direction, length of memory access) is provided, and fetching and decoding 5135the instruction from the VM is overly complicated to live in the kernel. 5136 5137Historically, when this situation occurred, KVM would print a warning and kill 5138the VM. KVM assumed that if the guest accessed non-memslot memory, it was 5139trying to do I/O, which just couldn't be emulated, and the warning message was 5140phrased accordingly. However, what happened more often was that a guest bug 5141caused access outside the guest memory areas which should lead to a more 5142meaningful warning message and an external abort in the guest, if the access 5143did not fall within an I/O window. 5144 5145Userspace implementations can query for KVM_CAP_ARM_NISV_TO_USER, and enable 5146this capability at VM creation. Once this is done, these types of errors will 5147instead return to userspace with KVM_EXIT_ARM_NISV, with the valid bits from 5148the HSR (arm) and ESR_EL2 (arm64) in the esr_iss field, and the faulting IPA 5149in the fault_ipa field. Userspace can either fix up the access if it's 5150actually an I/O access by decoding the instruction from guest memory (if it's 5151very brave) and continue executing the guest, or it can decide to suspend, 5152dump, or restart the guest. 5153 5154Note that KVM does not skip the faulting instruction as it does for 5155KVM_EXIT_MMIO, but userspace has to emulate any change to the processing state 5156if it decides to decode and emulate the instruction. 5157 5158:: 5159 5160 /* Fix the size of the union. */ 5161 char padding[256]; 5162 }; 5163 5164 /* 5165 * shared registers between kvm and userspace. 5166 * kvm_valid_regs specifies the register classes set by the host 5167 * kvm_dirty_regs specified the register classes dirtied by userspace 5168 * struct kvm_sync_regs is architecture specific, as well as the 5169 * bits for kvm_valid_regs and kvm_dirty_regs 5170 */ 5171 __u64 kvm_valid_regs; 5172 __u64 kvm_dirty_regs; 5173 union { 5174 struct kvm_sync_regs regs; 5175 char padding[SYNC_REGS_SIZE_BYTES]; 5176 } s; 5177 5178If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access 5179certain guest registers without having to call SET/GET_*REGS. Thus we can 5180avoid some system call overhead if userspace has to handle the exit. 5181Userspace can query the validity of the structure by checking 5182kvm_valid_regs for specific bits. These bits are architecture specific 5183and usually define the validity of a groups of registers. (e.g. one bit 5184for general purpose registers) 5185 5186Please note that the kernel is allowed to use the kvm_run structure as the 5187primary storage for certain register types. Therefore, the kernel may use the 5188values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set. 5189 5190:: 5191 5192 }; 5193 5194 5195 51966. Capabilities that can be enabled on vCPUs 5197============================================ 5198 5199There are certain capabilities that change the behavior of the virtual CPU or 5200the virtual machine when enabled. To enable them, please see section 4.37. 5201Below you can find a list of capabilities and what their effect on the vCPU or 5202the virtual machine is when enabling them. 5203 5204The following information is provided along with the description: 5205 5206 Architectures: 5207 which instruction set architectures provide this ioctl. 5208 x86 includes both i386 and x86_64. 5209 5210 Target: 5211 whether this is a per-vcpu or per-vm capability. 5212 5213 Parameters: 5214 what parameters are accepted by the capability. 5215 5216 Returns: 5217 the return value. General error numbers (EBADF, ENOMEM, EINVAL) 5218 are not detailed, but errors with specific meanings are. 5219 5220 52216.1 KVM_CAP_PPC_OSI 5222------------------- 5223 5224:Architectures: ppc 5225:Target: vcpu 5226:Parameters: none 5227:Returns: 0 on success; -1 on error 5228 5229This capability enables interception of OSI hypercalls that otherwise would 5230be treated as normal system calls to be injected into the guest. OSI hypercalls 5231were invented by Mac-on-Linux to have a standardized communication mechanism 5232between the guest and the host. 5233 5234When this capability is enabled, KVM_EXIT_OSI can occur. 5235 5236 52376.2 KVM_CAP_PPC_PAPR 5238-------------------- 5239 5240:Architectures: ppc 5241:Target: vcpu 5242:Parameters: none 5243:Returns: 0 on success; -1 on error 5244 5245This capability enables interception of PAPR hypercalls. PAPR hypercalls are 5246done using the hypercall instruction "sc 1". 5247 5248It also sets the guest privilege level to "supervisor" mode. Usually the guest 5249runs in "hypervisor" privilege mode with a few missing features. 5250 5251In addition to the above, it changes the semantics of SDR1. In this mode, the 5252HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the 5253HTAB invisible to the guest. 5254 5255When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur. 5256 5257 52586.3 KVM_CAP_SW_TLB 5259------------------ 5260 5261:Architectures: ppc 5262:Target: vcpu 5263:Parameters: args[0] is the address of a struct kvm_config_tlb 5264:Returns: 0 on success; -1 on error 5265 5266:: 5267 5268 struct kvm_config_tlb { 5269 __u64 params; 5270 __u64 array; 5271 __u32 mmu_type; 5272 __u32 array_len; 5273 }; 5274 5275Configures the virtual CPU's TLB array, establishing a shared memory area 5276between userspace and KVM. The "params" and "array" fields are userspace 5277addresses of mmu-type-specific data structures. The "array_len" field is an 5278safety mechanism, and should be set to the size in bytes of the memory that 5279userspace has reserved for the array. It must be at least the size dictated 5280by "mmu_type" and "params". 5281 5282While KVM_RUN is active, the shared region is under control of KVM. Its 5283contents are undefined, and any modification by userspace results in 5284boundedly undefined behavior. 5285 5286On return from KVM_RUN, the shared region will reflect the current state of 5287the guest's TLB. If userspace makes any changes, it must call KVM_DIRTY_TLB 5288to tell KVM which entries have been changed, prior to calling KVM_RUN again 5289on this vcpu. 5290 5291For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV: 5292 5293 - The "params" field is of type "struct kvm_book3e_206_tlb_params". 5294 - The "array" field points to an array of type "struct 5295 kvm_book3e_206_tlb_entry". 5296 - The array consists of all entries in the first TLB, followed by all 5297 entries in the second TLB. 5298 - Within a TLB, entries are ordered first by increasing set number. Within a 5299 set, entries are ordered by way (increasing ESEL). 5300 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1) 5301 where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value. 5302 - The tsize field of mas1 shall be set to 4K on TLB0, even though the 5303 hardware ignores this value for TLB0. 5304 53056.4 KVM_CAP_S390_CSS_SUPPORT 5306---------------------------- 5307 5308:Architectures: s390 5309:Target: vcpu 5310:Parameters: none 5311:Returns: 0 on success; -1 on error 5312 5313This capability enables support for handling of channel I/O instructions. 5314 5315TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are 5316handled in-kernel, while the other I/O instructions are passed to userspace. 5317 5318When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST 5319SUBCHANNEL intercepts. 5320 5321Note that even though this capability is enabled per-vcpu, the complete 5322virtual machine is affected. 5323 53246.5 KVM_CAP_PPC_EPR 5325------------------- 5326 5327:Architectures: ppc 5328:Target: vcpu 5329:Parameters: args[0] defines whether the proxy facility is active 5330:Returns: 0 on success; -1 on error 5331 5332This capability enables or disables the delivery of interrupts through the 5333external proxy facility. 5334 5335When enabled (args[0] != 0), every time the guest gets an external interrupt 5336delivered, it automatically exits into user space with a KVM_EXIT_EPR exit 5337to receive the topmost interrupt vector. 5338 5339When disabled (args[0] == 0), behavior is as if this facility is unsupported. 5340 5341When this capability is enabled, KVM_EXIT_EPR can occur. 5342 53436.6 KVM_CAP_IRQ_MPIC 5344-------------------- 5345 5346:Architectures: ppc 5347:Parameters: args[0] is the MPIC device fd; 5348 args[1] is the MPIC CPU number for this vcpu 5349 5350This capability connects the vcpu to an in-kernel MPIC device. 5351 53526.7 KVM_CAP_IRQ_XICS 5353-------------------- 5354 5355:Architectures: ppc 5356:Target: vcpu 5357:Parameters: args[0] is the XICS device fd; 5358 args[1] is the XICS CPU number (server ID) for this vcpu 5359 5360This capability connects the vcpu to an in-kernel XICS device. 5361 53626.8 KVM_CAP_S390_IRQCHIP 5363------------------------ 5364 5365:Architectures: s390 5366:Target: vm 5367:Parameters: none 5368 5369This capability enables the in-kernel irqchip for s390. Please refer to 5370"4.24 KVM_CREATE_IRQCHIP" for details. 5371 53726.9 KVM_CAP_MIPS_FPU 5373-------------------- 5374 5375:Architectures: mips 5376:Target: vcpu 5377:Parameters: args[0] is reserved for future use (should be 0). 5378 5379This capability allows the use of the host Floating Point Unit by the guest. It 5380allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is 5381done the ``KVM_REG_MIPS_FPR_*`` and ``KVM_REG_MIPS_FCR_*`` registers can be 5382accessed (depending on the current guest FPU register mode), and the Status.FR, 5383Config5.FRE bits are accessible via the KVM API and also from the guest, 5384depending on them being supported by the FPU. 5385 53866.10 KVM_CAP_MIPS_MSA 5387--------------------- 5388 5389:Architectures: mips 5390:Target: vcpu 5391:Parameters: args[0] is reserved for future use (should be 0). 5392 5393This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest. 5394It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest. 5395Once this is done the ``KVM_REG_MIPS_VEC_*`` and ``KVM_REG_MIPS_MSA_*`` 5396registers can be accessed, and the Config5.MSAEn bit is accessible via the 5397KVM API and also from the guest. 5398 53996.74 KVM_CAP_SYNC_REGS 5400---------------------- 5401 5402:Architectures: s390, x86 5403:Target: s390: always enabled, x86: vcpu 5404:Parameters: none 5405:Returns: x86: KVM_CHECK_EXTENSION returns a bit-array indicating which register 5406 sets are supported 5407 (bitfields defined in arch/x86/include/uapi/asm/kvm.h). 5408 5409As described above in the kvm_sync_regs struct info in section 5 (kvm_run): 5410KVM_CAP_SYNC_REGS "allow[s] userspace to access certain guest registers 5411without having to call SET/GET_*REGS". This reduces overhead by eliminating 5412repeated ioctl calls for setting and/or getting register values. This is 5413particularly important when userspace is making synchronous guest state 5414modifications, e.g. when emulating and/or intercepting instructions in 5415userspace. 5416 5417For s390 specifics, please refer to the source code. 5418 5419For x86: 5420 5421- the register sets to be copied out to kvm_run are selectable 5422 by userspace (rather that all sets being copied out for every exit). 5423- vcpu_events are available in addition to regs and sregs. 5424 5425For x86, the 'kvm_valid_regs' field of struct kvm_run is overloaded to 5426function as an input bit-array field set by userspace to indicate the 5427specific register sets to be copied out on the next exit. 5428 5429To indicate when userspace has modified values that should be copied into 5430the vCPU, the all architecture bitarray field, 'kvm_dirty_regs' must be set. 5431This is done using the same bitflags as for the 'kvm_valid_regs' field. 5432If the dirty bit is not set, then the register set values will not be copied 5433into the vCPU even if they've been modified. 5434 5435Unused bitfields in the bitarrays must be set to zero. 5436 5437:: 5438 5439 struct kvm_sync_regs { 5440 struct kvm_regs regs; 5441 struct kvm_sregs sregs; 5442 struct kvm_vcpu_events events; 5443 }; 5444 54456.75 KVM_CAP_PPC_IRQ_XIVE 5446------------------------- 5447 5448:Architectures: ppc 5449:Target: vcpu 5450:Parameters: args[0] is the XIVE device fd; 5451 args[1] is the XIVE CPU number (server ID) for this vcpu 5452 5453This capability connects the vcpu to an in-kernel XIVE device. 5454 54557. Capabilities that can be enabled on VMs 5456========================================== 5457 5458There are certain capabilities that change the behavior of the virtual 5459machine when enabled. To enable them, please see section 4.37. Below 5460you can find a list of capabilities and what their effect on the VM 5461is when enabling them. 5462 5463The following information is provided along with the description: 5464 5465 Architectures: 5466 which instruction set architectures provide this ioctl. 5467 x86 includes both i386 and x86_64. 5468 5469 Parameters: 5470 what parameters are accepted by the capability. 5471 5472 Returns: 5473 the return value. General error numbers (EBADF, ENOMEM, EINVAL) 5474 are not detailed, but errors with specific meanings are. 5475 5476 54777.1 KVM_CAP_PPC_ENABLE_HCALL 5478---------------------------- 5479 5480:Architectures: ppc 5481:Parameters: args[0] is the sPAPR hcall number; 5482 args[1] is 0 to disable, 1 to enable in-kernel handling 5483 5484This capability controls whether individual sPAPR hypercalls (hcalls) 5485get handled by the kernel or not. Enabling or disabling in-kernel 5486handling of an hcall is effective across the VM. On creation, an 5487initial set of hcalls are enabled for in-kernel handling, which 5488consists of those hcalls for which in-kernel handlers were implemented 5489before this capability was implemented. If disabled, the kernel will 5490not to attempt to handle the hcall, but will always exit to userspace 5491to handle it. Note that it may not make sense to enable some and 5492disable others of a group of related hcalls, but KVM does not prevent 5493userspace from doing that. 5494 5495If the hcall number specified is not one that has an in-kernel 5496implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL 5497error. 5498 54997.2 KVM_CAP_S390_USER_SIGP 5500-------------------------- 5501 5502:Architectures: s390 5503:Parameters: none 5504 5505This capability controls which SIGP orders will be handled completely in user 5506space. With this capability enabled, all fast orders will be handled completely 5507in the kernel: 5508 5509- SENSE 5510- SENSE RUNNING 5511- EXTERNAL CALL 5512- EMERGENCY SIGNAL 5513- CONDITIONAL EMERGENCY SIGNAL 5514 5515All other orders will be handled completely in user space. 5516 5517Only privileged operation exceptions will be checked for in the kernel (or even 5518in the hardware prior to interception). If this capability is not enabled, the 5519old way of handling SIGP orders is used (partially in kernel and user space). 5520 55217.3 KVM_CAP_S390_VECTOR_REGISTERS 5522--------------------------------- 5523 5524:Architectures: s390 5525:Parameters: none 5526:Returns: 0 on success, negative value on error 5527 5528Allows use of the vector registers introduced with z13 processor, and 5529provides for the synchronization between host and user space. Will 5530return -EINVAL if the machine does not support vectors. 5531 55327.4 KVM_CAP_S390_USER_STSI 5533-------------------------- 5534 5535:Architectures: s390 5536:Parameters: none 5537 5538This capability allows post-handlers for the STSI instruction. After 5539initial handling in the kernel, KVM exits to user space with 5540KVM_EXIT_S390_STSI to allow user space to insert further data. 5541 5542Before exiting to userspace, kvm handlers should fill in s390_stsi field of 5543vcpu->run:: 5544 5545 struct { 5546 __u64 addr; 5547 __u8 ar; 5548 __u8 reserved; 5549 __u8 fc; 5550 __u8 sel1; 5551 __u16 sel2; 5552 } s390_stsi; 5553 5554 @addr - guest address of STSI SYSIB 5555 @fc - function code 5556 @sel1 - selector 1 5557 @sel2 - selector 2 5558 @ar - access register number 5559 5560KVM handlers should exit to userspace with rc = -EREMOTE. 5561 55627.5 KVM_CAP_SPLIT_IRQCHIP 5563------------------------- 5564 5565:Architectures: x86 5566:Parameters: args[0] - number of routes reserved for userspace IOAPICs 5567:Returns: 0 on success, -1 on error 5568 5569Create a local apic for each processor in the kernel. This can be used 5570instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the 5571IOAPIC and PIC (and also the PIT, even though this has to be enabled 5572separately). 5573 5574This capability also enables in kernel routing of interrupt requests; 5575when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are 5576used in the IRQ routing table. The first args[0] MSI routes are reserved 5577for the IOAPIC pins. Whenever the LAPIC receives an EOI for these routes, 5578a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace. 5579 5580Fails if VCPU has already been created, or if the irqchip is already in the 5581kernel (i.e. KVM_CREATE_IRQCHIP has already been called). 5582 55837.6 KVM_CAP_S390_RI 5584------------------- 5585 5586:Architectures: s390 5587:Parameters: none 5588 5589Allows use of runtime-instrumentation introduced with zEC12 processor. 5590Will return -EINVAL if the machine does not support runtime-instrumentation. 5591Will return -EBUSY if a VCPU has already been created. 5592 55937.7 KVM_CAP_X2APIC_API 5594---------------------- 5595 5596:Architectures: x86 5597:Parameters: args[0] - features that should be enabled 5598:Returns: 0 on success, -EINVAL when args[0] contains invalid features 5599 5600Valid feature flags in args[0] are:: 5601 5602 #define KVM_X2APIC_API_USE_32BIT_IDS (1ULL << 0) 5603 #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK (1ULL << 1) 5604 5605Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of 5606KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC, 5607allowing the use of 32-bit APIC IDs. See KVM_CAP_X2APIC_API in their 5608respective sections. 5609 5610KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work 5611in logical mode or with more than 255 VCPUs. Otherwise, KVM treats 0xff 5612as a broadcast even in x2APIC mode in order to support physical x2APIC 5613without interrupt remapping. This is undesirable in logical mode, 5614where 0xff represents CPUs 0-7 in cluster 0. 5615 56167.8 KVM_CAP_S390_USER_INSTR0 5617---------------------------- 5618 5619:Architectures: s390 5620:Parameters: none 5621 5622With this capability enabled, all illegal instructions 0x0000 (2 bytes) will 5623be intercepted and forwarded to user space. User space can use this 5624mechanism e.g. to realize 2-byte software breakpoints. The kernel will 5625not inject an operating exception for these instructions, user space has 5626to take care of that. 5627 5628This capability can be enabled dynamically even if VCPUs were already 5629created and are running. 5630 56317.9 KVM_CAP_S390_GS 5632------------------- 5633 5634:Architectures: s390 5635:Parameters: none 5636:Returns: 0 on success; -EINVAL if the machine does not support 5637 guarded storage; -EBUSY if a VCPU has already been created. 5638 5639Allows use of guarded storage for the KVM guest. 5640 56417.10 KVM_CAP_S390_AIS 5642--------------------- 5643 5644:Architectures: s390 5645:Parameters: none 5646 5647Allow use of adapter-interruption suppression. 5648:Returns: 0 on success; -EBUSY if a VCPU has already been created. 5649 56507.11 KVM_CAP_PPC_SMT 5651-------------------- 5652 5653:Architectures: ppc 5654:Parameters: vsmt_mode, flags 5655 5656Enabling this capability on a VM provides userspace with a way to set 5657the desired virtual SMT mode (i.e. the number of virtual CPUs per 5658virtual core). The virtual SMT mode, vsmt_mode, must be a power of 2 5659between 1 and 8. On POWER8, vsmt_mode must also be no greater than 5660the number of threads per subcore for the host. Currently flags must 5661be 0. A successful call to enable this capability will result in 5662vsmt_mode being returned when the KVM_CAP_PPC_SMT capability is 5663subsequently queried for the VM. This capability is only supported by 5664HV KVM, and can only be set before any VCPUs have been created. 5665The KVM_CAP_PPC_SMT_POSSIBLE capability indicates which virtual SMT 5666modes are available. 5667 56687.12 KVM_CAP_PPC_FWNMI 5669---------------------- 5670 5671:Architectures: ppc 5672:Parameters: none 5673 5674With this capability a machine check exception in the guest address 5675space will cause KVM to exit the guest with NMI exit reason. This 5676enables QEMU to build error log and branch to guest kernel registered 5677machine check handling routine. Without this capability KVM will 5678branch to guests' 0x200 interrupt vector. 5679 56807.13 KVM_CAP_X86_DISABLE_EXITS 5681------------------------------ 5682 5683:Architectures: x86 5684:Parameters: args[0] defines which exits are disabled 5685:Returns: 0 on success, -EINVAL when args[0] contains invalid exits 5686 5687Valid bits in args[0] are:: 5688 5689 #define KVM_X86_DISABLE_EXITS_MWAIT (1 << 0) 5690 #define KVM_X86_DISABLE_EXITS_HLT (1 << 1) 5691 #define KVM_X86_DISABLE_EXITS_PAUSE (1 << 2) 5692 #define KVM_X86_DISABLE_EXITS_CSTATE (1 << 3) 5693 5694Enabling this capability on a VM provides userspace with a way to no 5695longer intercept some instructions for improved latency in some 5696workloads, and is suggested when vCPUs are associated to dedicated 5697physical CPUs. More bits can be added in the future; userspace can 5698just pass the KVM_CHECK_EXTENSION result to KVM_ENABLE_CAP to disable 5699all such vmexits. 5700 5701Do not enable KVM_FEATURE_PV_UNHALT if you disable HLT exits. 5702 57037.14 KVM_CAP_S390_HPAGE_1M 5704-------------------------- 5705 5706:Architectures: s390 5707:Parameters: none 5708:Returns: 0 on success, -EINVAL if hpage module parameter was not set 5709 or cmma is enabled, or the VM has the KVM_VM_S390_UCONTROL 5710 flag set 5711 5712With this capability the KVM support for memory backing with 1m pages 5713through hugetlbfs can be enabled for a VM. After the capability is 5714enabled, cmma can't be enabled anymore and pfmfi and the storage key 5715interpretation are disabled. If cmma has already been enabled or the 5716hpage module parameter is not set to 1, -EINVAL is returned. 5717 5718While it is generally possible to create a huge page backed VM without 5719this capability, the VM will not be able to run. 5720 57217.15 KVM_CAP_MSR_PLATFORM_INFO 5722------------------------------ 5723 5724:Architectures: x86 5725:Parameters: args[0] whether feature should be enabled or not 5726 5727With this capability, a guest may read the MSR_PLATFORM_INFO MSR. Otherwise, 5728a #GP would be raised when the guest tries to access. Currently, this 5729capability does not enable write permissions of this MSR for the guest. 5730 57317.16 KVM_CAP_PPC_NESTED_HV 5732-------------------------- 5733 5734:Architectures: ppc 5735:Parameters: none 5736:Returns: 0 on success, -EINVAL when the implementation doesn't support 5737 nested-HV virtualization. 5738 5739HV-KVM on POWER9 and later systems allows for "nested-HV" 5740virtualization, which provides a way for a guest VM to run guests that 5741can run using the CPU's supervisor mode (privileged non-hypervisor 5742state). Enabling this capability on a VM depends on the CPU having 5743the necessary functionality and on the facility being enabled with a 5744kvm-hv module parameter. 5745 57467.17 KVM_CAP_EXCEPTION_PAYLOAD 5747------------------------------ 5748 5749:Architectures: x86 5750:Parameters: args[0] whether feature should be enabled or not 5751 5752With this capability enabled, CR2 will not be modified prior to the 5753emulated VM-exit when L1 intercepts a #PF exception that occurs in 5754L2. Similarly, for kvm-intel only, DR6 will not be modified prior to 5755the emulated VM-exit when L1 intercepts a #DB exception that occurs in 5756L2. As a result, when KVM_GET_VCPU_EVENTS reports a pending #PF (or 5757#DB) exception for L2, exception.has_payload will be set and the 5758faulting address (or the new DR6 bits*) will be reported in the 5759exception_payload field. Similarly, when userspace injects a #PF (or 5760#DB) into L2 using KVM_SET_VCPU_EVENTS, it is expected to set 5761exception.has_payload and to put the faulting address - or the new DR6 5762bits\ [#]_ - in the exception_payload field. 5763 5764This capability also enables exception.pending in struct 5765kvm_vcpu_events, which allows userspace to distinguish between pending 5766and injected exceptions. 5767 5768 5769.. [#] For the new DR6 bits, note that bit 16 is set iff the #DB exception 5770 will clear DR6.RTM. 5771 57727.18 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 5773 5774:Architectures: x86, arm, arm64, mips 5775:Parameters: args[0] whether feature should be enabled or not 5776 5777Valid flags are:: 5778 5779 #define KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (1 << 0) 5780 #define KVM_DIRTY_LOG_INITIALLY_SET (1 << 1) 5781 5782With KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE is set, KVM_GET_DIRTY_LOG will not 5783automatically clear and write-protect all pages that are returned as dirty. 5784Rather, userspace will have to do this operation separately using 5785KVM_CLEAR_DIRTY_LOG. 5786 5787At the cost of a slightly more complicated operation, this provides better 5788scalability and responsiveness for two reasons. First, 5789KVM_CLEAR_DIRTY_LOG ioctl can operate on a 64-page granularity rather 5790than requiring to sync a full memslot; this ensures that KVM does not 5791take spinlocks for an extended period of time. Second, in some cases a 5792large amount of time can pass between a call to KVM_GET_DIRTY_LOG and 5793userspace actually using the data in the page. Pages can be modified 5794during this time, which is inefficient for both the guest and userspace: 5795the guest will incur a higher penalty due to write protection faults, 5796while userspace can see false reports of dirty pages. Manual reprotection 5797helps reducing this time, improving guest performance and reducing the 5798number of dirty log false positives. 5799 5800With KVM_DIRTY_LOG_INITIALLY_SET set, all the bits of the dirty bitmap 5801will be initialized to 1 when created. This also improves performance because 5802dirty logging can be enabled gradually in small chunks on the first call 5803to KVM_CLEAR_DIRTY_LOG. KVM_DIRTY_LOG_INITIALLY_SET depends on 5804KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (it is also only available on 5805x86 and arm64 for now). 5806 5807KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 was previously available under the name 5808KVM_CAP_MANUAL_DIRTY_LOG_PROTECT, but the implementation had bugs that make 5809it hard or impossible to use it correctly. The availability of 5810KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 signals that those bugs are fixed. 5811Userspace should not try to use KVM_CAP_MANUAL_DIRTY_LOG_PROTECT. 5812 58137.19 KVM_CAP_PPC_SECURE_GUEST 5814------------------------------ 5815 5816:Architectures: ppc 5817 5818This capability indicates that KVM is running on a host that has 5819ultravisor firmware and thus can support a secure guest. On such a 5820system, a guest can ask the ultravisor to make it a secure guest, 5821one whose memory is inaccessible to the host except for pages which 5822are explicitly requested to be shared with the host. The ultravisor 5823notifies KVM when a guest requests to become a secure guest, and KVM 5824has the opportunity to veto the transition. 5825 5826If present, this capability can be enabled for a VM, meaning that KVM 5827will allow the transition to secure guest mode. Otherwise KVM will 5828veto the transition. 5829 58307.20 KVM_CAP_HALT_POLL 5831---------------------- 5832 5833:Architectures: all 5834:Target: VM 5835:Parameters: args[0] is the maximum poll time in nanoseconds 5836:Returns: 0 on success; -1 on error 5837 5838This capability overrides the kvm module parameter halt_poll_ns for the 5839target VM. 5840 5841VCPU polling allows a VCPU to poll for wakeup events instead of immediately 5842scheduling during guest halts. The maximum time a VCPU can spend polling is 5843controlled by the kvm module parameter halt_poll_ns. This capability allows 5844the maximum halt time to specified on a per-VM basis, effectively overriding 5845the module parameter for the target VM. 5846 58478. Other capabilities. 5848====================== 5849 5850This section lists capabilities that give information about other 5851features of the KVM implementation. 5852 58538.1 KVM_CAP_PPC_HWRNG 5854--------------------- 5855 5856:Architectures: ppc 5857 5858This capability, if KVM_CHECK_EXTENSION indicates that it is 5859available, means that that the kernel has an implementation of the 5860H_RANDOM hypercall backed by a hardware random-number generator. 5861If present, the kernel H_RANDOM handler can be enabled for guest use 5862with the KVM_CAP_PPC_ENABLE_HCALL capability. 5863 58648.2 KVM_CAP_HYPERV_SYNIC 5865------------------------ 5866 5867:Architectures: x86 5868 5869This capability, if KVM_CHECK_EXTENSION indicates that it is 5870available, means that that the kernel has an implementation of the 5871Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is 5872used to support Windows Hyper-V based guest paravirt drivers(VMBus). 5873 5874In order to use SynIC, it has to be activated by setting this 5875capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this 5876will disable the use of APIC hardware virtualization even if supported 5877by the CPU, as it's incompatible with SynIC auto-EOI behavior. 5878 58798.3 KVM_CAP_PPC_RADIX_MMU 5880------------------------- 5881 5882:Architectures: ppc 5883 5884This capability, if KVM_CHECK_EXTENSION indicates that it is 5885available, means that that the kernel can support guests using the 5886radix MMU defined in Power ISA V3.00 (as implemented in the POWER9 5887processor). 5888 58898.4 KVM_CAP_PPC_HASH_MMU_V3 5890--------------------------- 5891 5892:Architectures: ppc 5893 5894This capability, if KVM_CHECK_EXTENSION indicates that it is 5895available, means that that the kernel can support guests using the 5896hashed page table MMU defined in Power ISA V3.00 (as implemented in 5897the POWER9 processor), including in-memory segment tables. 5898 58998.5 KVM_CAP_MIPS_VZ 5900------------------- 5901 5902:Architectures: mips 5903 5904This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that 5905it is available, means that full hardware assisted virtualization capabilities 5906of the hardware are available for use through KVM. An appropriate 5907KVM_VM_MIPS_* type must be passed to KVM_CREATE_VM to create a VM which 5908utilises it. 5909 5910If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is 5911available, it means that the VM is using full hardware assisted virtualization 5912capabilities of the hardware. This is useful to check after creating a VM with 5913KVM_VM_MIPS_DEFAULT. 5914 5915The value returned by KVM_CHECK_EXTENSION should be compared against known 5916values (see below). All other values are reserved. This is to allow for the 5917possibility of other hardware assisted virtualization implementations which 5918may be incompatible with the MIPS VZ ASE. 5919 5920== ========================================================================== 5921 0 The trap & emulate implementation is in use to run guest code in user 5922 mode. Guest virtual memory segments are rearranged to fit the guest in the 5923 user mode address space. 5924 5925 1 The MIPS VZ ASE is in use, providing full hardware assisted 5926 virtualization, including standard guest virtual memory segments. 5927== ========================================================================== 5928 59298.6 KVM_CAP_MIPS_TE 5930------------------- 5931 5932:Architectures: mips 5933 5934This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that 5935it is available, means that the trap & emulate implementation is available to 5936run guest code in user mode, even if KVM_CAP_MIPS_VZ indicates that hardware 5937assisted virtualisation is also available. KVM_VM_MIPS_TE (0) must be passed 5938to KVM_CREATE_VM to create a VM which utilises it. 5939 5940If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is 5941available, it means that the VM is using trap & emulate. 5942 59438.7 KVM_CAP_MIPS_64BIT 5944---------------------- 5945 5946:Architectures: mips 5947 5948This capability indicates the supported architecture type of the guest, i.e. the 5949supported register and address width. 5950 5951The values returned when this capability is checked by KVM_CHECK_EXTENSION on a 5952kvm VM handle correspond roughly to the CP0_Config.AT register field, and should 5953be checked specifically against known values (see below). All other values are 5954reserved. 5955 5956== ======================================================================== 5957 0 MIPS32 or microMIPS32. 5958 Both registers and addresses are 32-bits wide. 5959 It will only be possible to run 32-bit guest code. 5960 5961 1 MIPS64 or microMIPS64 with access only to 32-bit compatibility segments. 5962 Registers are 64-bits wide, but addresses are 32-bits wide. 5963 64-bit guest code may run but cannot access MIPS64 memory segments. 5964 It will also be possible to run 32-bit guest code. 5965 5966 2 MIPS64 or microMIPS64 with access to all address segments. 5967 Both registers and addresses are 64-bits wide. 5968 It will be possible to run 64-bit or 32-bit guest code. 5969== ======================================================================== 5970 59718.9 KVM_CAP_ARM_USER_IRQ 5972------------------------ 5973 5974:Architectures: arm, arm64 5975 5976This capability, if KVM_CHECK_EXTENSION indicates that it is available, means 5977that if userspace creates a VM without an in-kernel interrupt controller, it 5978will be notified of changes to the output level of in-kernel emulated devices, 5979which can generate virtual interrupts, presented to the VM. 5980For such VMs, on every return to userspace, the kernel 5981updates the vcpu's run->s.regs.device_irq_level field to represent the actual 5982output level of the device. 5983 5984Whenever kvm detects a change in the device output level, kvm guarantees at 5985least one return to userspace before running the VM. This exit could either 5986be a KVM_EXIT_INTR or any other exit event, like KVM_EXIT_MMIO. This way, 5987userspace can always sample the device output level and re-compute the state of 5988the userspace interrupt controller. Userspace should always check the state 5989of run->s.regs.device_irq_level on every kvm exit. 5990The value in run->s.regs.device_irq_level can represent both level and edge 5991triggered interrupt signals, depending on the device. Edge triggered interrupt 5992signals will exit to userspace with the bit in run->s.regs.device_irq_level 5993set exactly once per edge signal. 5994 5995The field run->s.regs.device_irq_level is available independent of 5996run->kvm_valid_regs or run->kvm_dirty_regs bits. 5997 5998If KVM_CAP_ARM_USER_IRQ is supported, the KVM_CHECK_EXTENSION ioctl returns a 5999number larger than 0 indicating the version of this capability is implemented 6000and thereby which bits in in run->s.regs.device_irq_level can signal values. 6001 6002Currently the following bits are defined for the device_irq_level bitmap:: 6003 6004 KVM_CAP_ARM_USER_IRQ >= 1: 6005 6006 KVM_ARM_DEV_EL1_VTIMER - EL1 virtual timer 6007 KVM_ARM_DEV_EL1_PTIMER - EL1 physical timer 6008 KVM_ARM_DEV_PMU - ARM PMU overflow interrupt signal 6009 6010Future versions of kvm may implement additional events. These will get 6011indicated by returning a higher number from KVM_CHECK_EXTENSION and will be 6012listed above. 6013 60148.10 KVM_CAP_PPC_SMT_POSSIBLE 6015----------------------------- 6016 6017:Architectures: ppc 6018 6019Querying this capability returns a bitmap indicating the possible 6020virtual SMT modes that can be set using KVM_CAP_PPC_SMT. If bit N 6021(counting from the right) is set, then a virtual SMT mode of 2^N is 6022available. 6023 60248.11 KVM_CAP_HYPERV_SYNIC2 6025-------------------------- 6026 6027:Architectures: x86 6028 6029This capability enables a newer version of Hyper-V Synthetic interrupt 6030controller (SynIC). The only difference with KVM_CAP_HYPERV_SYNIC is that KVM 6031doesn't clear SynIC message and event flags pages when they are enabled by 6032writing to the respective MSRs. 6033 60348.12 KVM_CAP_HYPERV_VP_INDEX 6035---------------------------- 6036 6037:Architectures: x86 6038 6039This capability indicates that userspace can load HV_X64_MSR_VP_INDEX msr. Its 6040value is used to denote the target vcpu for a SynIC interrupt. For 6041compatibilty, KVM initializes this msr to KVM's internal vcpu index. When this 6042capability is absent, userspace can still query this msr's value. 6043 60448.13 KVM_CAP_S390_AIS_MIGRATION 6045------------------------------- 6046 6047:Architectures: s390 6048:Parameters: none 6049 6050This capability indicates if the flic device will be able to get/set the 6051AIS states for migration via the KVM_DEV_FLIC_AISM_ALL attribute and allows 6052to discover this without having to create a flic device. 6053 60548.14 KVM_CAP_S390_PSW 6055--------------------- 6056 6057:Architectures: s390 6058 6059This capability indicates that the PSW is exposed via the kvm_run structure. 6060 60618.15 KVM_CAP_S390_GMAP 6062---------------------- 6063 6064:Architectures: s390 6065 6066This capability indicates that the user space memory used as guest mapping can 6067be anywhere in the user memory address space, as long as the memory slots are 6068aligned and sized to a segment (1MB) boundary. 6069 60708.16 KVM_CAP_S390_COW 6071--------------------- 6072 6073:Architectures: s390 6074 6075This capability indicates that the user space memory used as guest mapping can 6076use copy-on-write semantics as well as dirty pages tracking via read-only page 6077tables. 6078 60798.17 KVM_CAP_S390_BPB 6080--------------------- 6081 6082:Architectures: s390 6083 6084This capability indicates that kvm will implement the interfaces to handle 6085reset, migration and nested KVM for branch prediction blocking. The stfle 6086facility 82 should not be provided to the guest without this capability. 6087 60888.18 KVM_CAP_HYPERV_TLBFLUSH 6089---------------------------- 6090 6091:Architectures: x86 6092 6093This capability indicates that KVM supports paravirtualized Hyper-V TLB Flush 6094hypercalls: 6095HvFlushVirtualAddressSpace, HvFlushVirtualAddressSpaceEx, 6096HvFlushVirtualAddressList, HvFlushVirtualAddressListEx. 6097 60988.19 KVM_CAP_ARM_INJECT_SERROR_ESR 6099---------------------------------- 6100 6101:Architectures: arm, arm64 6102 6103This capability indicates that userspace can specify (via the 6104KVM_SET_VCPU_EVENTS ioctl) the syndrome value reported to the guest when it 6105takes a virtual SError interrupt exception. 6106If KVM advertises this capability, userspace can only specify the ISS field for 6107the ESR syndrome. Other parts of the ESR, such as the EC are generated by the 6108CPU when the exception is taken. If this virtual SError is taken to EL1 using 6109AArch64, this value will be reported in the ISS field of ESR_ELx. 6110 6111See KVM_CAP_VCPU_EVENTS for more details. 6112 61138.20 KVM_CAP_HYPERV_SEND_IPI 6114---------------------------- 6115 6116:Architectures: x86 6117 6118This capability indicates that KVM supports paravirtualized Hyper-V IPI send 6119hypercalls: 6120HvCallSendSyntheticClusterIpi, HvCallSendSyntheticClusterIpiEx. 6121 61228.21 KVM_CAP_HYPERV_DIRECT_TLBFLUSH 6123----------------------------------- 6124 6125:Architecture: x86 6126 6127This capability indicates that KVM running on top of Hyper-V hypervisor 6128enables Direct TLB flush for its guests meaning that TLB flush 6129hypercalls are handled by Level 0 hypervisor (Hyper-V) bypassing KVM. 6130Due to the different ABI for hypercall parameters between Hyper-V and 6131KVM, enabling this capability effectively disables all hypercall 6132handling by KVM (as some KVM hypercall may be mistakenly treated as TLB 6133flush hypercalls by Hyper-V) so userspace should disable KVM identification 6134in CPUID and only exposes Hyper-V identification. In this case, guest 6135thinks it's running on Hyper-V and only use Hyper-V hypercalls. 6136 61378.22 KVM_CAP_S390_VCPU_RESETS 6138 6139Architectures: s390 6140 6141This capability indicates that the KVM_S390_NORMAL_RESET and 6142KVM_S390_CLEAR_RESET ioctls are available. 6143 61448.23 KVM_CAP_S390_PROTECTED 6145 6146Architecture: s390 6147 6148 6149This capability indicates that the Ultravisor has been initialized and 6150KVM can therefore start protected VMs. 6151This capability governs the KVM_S390_PV_COMMAND ioctl and the 6152KVM_MP_STATE_LOAD MP_STATE. KVM_SET_MP_STATE can fail for protected 6153guests when the state change is invalid. 6154