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 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 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 672Note, when this IOCTL fails, KVM gives no guarantees that previous valid CPUID 673configuration (if there is) is not corrupted. Userspace can get a copy of the 674resulting CPUID configuration through KVM_GET_CPUID2 in case. 675 676:: 677 678 struct kvm_cpuid_entry { 679 __u32 function; 680 __u32 eax; 681 __u32 ebx; 682 __u32 ecx; 683 __u32 edx; 684 __u32 padding; 685 }; 686 687 /* for KVM_SET_CPUID */ 688 struct kvm_cpuid { 689 __u32 nent; 690 __u32 padding; 691 struct kvm_cpuid_entry entries[0]; 692 }; 693 694 6954.21 KVM_SET_SIGNAL_MASK 696------------------------ 697 698:Capability: basic 699:Architectures: all 700:Type: vcpu ioctl 701:Parameters: struct kvm_signal_mask (in) 702:Returns: 0 on success, -1 on error 703 704Defines which signals are blocked during execution of KVM_RUN. This 705signal mask temporarily overrides the threads signal mask. Any 706unblocked signal received (except SIGKILL and SIGSTOP, which retain 707their traditional behaviour) will cause KVM_RUN to return with -EINTR. 708 709Note the signal will only be delivered if not blocked by the original 710signal mask. 711 712:: 713 714 /* for KVM_SET_SIGNAL_MASK */ 715 struct kvm_signal_mask { 716 __u32 len; 717 __u8 sigset[0]; 718 }; 719 720 7214.22 KVM_GET_FPU 722---------------- 723 724:Capability: basic 725:Architectures: x86 726:Type: vcpu ioctl 727:Parameters: struct kvm_fpu (out) 728:Returns: 0 on success, -1 on error 729 730Reads the floating point state from the vcpu. 731 732:: 733 734 /* for KVM_GET_FPU and KVM_SET_FPU */ 735 struct kvm_fpu { 736 __u8 fpr[8][16]; 737 __u16 fcw; 738 __u16 fsw; 739 __u8 ftwx; /* in fxsave format */ 740 __u8 pad1; 741 __u16 last_opcode; 742 __u64 last_ip; 743 __u64 last_dp; 744 __u8 xmm[16][16]; 745 __u32 mxcsr; 746 __u32 pad2; 747 }; 748 749 7504.23 KVM_SET_FPU 751---------------- 752 753:Capability: basic 754:Architectures: x86 755:Type: vcpu ioctl 756:Parameters: struct kvm_fpu (in) 757:Returns: 0 on success, -1 on error 758 759Writes the floating point state to the vcpu. 760 761:: 762 763 /* for KVM_GET_FPU and KVM_SET_FPU */ 764 struct kvm_fpu { 765 __u8 fpr[8][16]; 766 __u16 fcw; 767 __u16 fsw; 768 __u8 ftwx; /* in fxsave format */ 769 __u8 pad1; 770 __u16 last_opcode; 771 __u64 last_ip; 772 __u64 last_dp; 773 __u8 xmm[16][16]; 774 __u32 mxcsr; 775 __u32 pad2; 776 }; 777 778 7794.24 KVM_CREATE_IRQCHIP 780----------------------- 781 782:Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQCHIP (s390) 783:Architectures: x86, ARM, arm64, s390 784:Type: vm ioctl 785:Parameters: none 786:Returns: 0 on success, -1 on error 787 788Creates an interrupt controller model in the kernel. 789On x86, creates a virtual ioapic, a virtual PIC (two PICs, nested), and sets up 790future vcpus to have a local APIC. IRQ routing for GSIs 0-15 is set to both 791PIC and IOAPIC; GSI 16-23 only go to the IOAPIC. 792On ARM/arm64, a GICv2 is created. Any other GIC versions require the usage of 793KVM_CREATE_DEVICE, which also supports creating a GICv2. Using 794KVM_CREATE_DEVICE is preferred over KVM_CREATE_IRQCHIP for GICv2. 795On s390, a dummy irq routing table is created. 796 797Note that on s390 the KVM_CAP_S390_IRQCHIP vm capability needs to be enabled 798before KVM_CREATE_IRQCHIP can be used. 799 800 8014.25 KVM_IRQ_LINE 802----------------- 803 804:Capability: KVM_CAP_IRQCHIP 805:Architectures: x86, arm, arm64 806:Type: vm ioctl 807:Parameters: struct kvm_irq_level 808:Returns: 0 on success, -1 on error 809 810Sets the level of a GSI input to the interrupt controller model in the kernel. 811On some architectures it is required that an interrupt controller model has 812been previously created with KVM_CREATE_IRQCHIP. Note that edge-triggered 813interrupts require the level to be set to 1 and then back to 0. 814 815On real hardware, interrupt pins can be active-low or active-high. This 816does not matter for the level field of struct kvm_irq_level: 1 always 817means active (asserted), 0 means inactive (deasserted). 818 819x86 allows the operating system to program the interrupt polarity 820(active-low/active-high) for level-triggered interrupts, and KVM used 821to consider the polarity. However, due to bitrot in the handling of 822active-low interrupts, the above convention is now valid on x86 too. 823This is signaled by KVM_CAP_X86_IOAPIC_POLARITY_IGNORED. Userspace 824should not present interrupts to the guest as active-low unless this 825capability is present (or unless it is not using the in-kernel irqchip, 826of course). 827 828 829ARM/arm64 can signal an interrupt either at the CPU level, or at the 830in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to 831use PPIs designated for specific cpus. The irq field is interpreted 832like this:: 833 834 bits: | 31 ... 28 | 27 ... 24 | 23 ... 16 | 15 ... 0 | 835 field: | vcpu2_index | irq_type | vcpu_index | irq_id | 836 837The irq_type field has the following values: 838 839- irq_type[0]: 840 out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ 841- irq_type[1]: 842 in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.) 843 (the vcpu_index field is ignored) 844- irq_type[2]: 845 in-kernel GIC: PPI, irq_id between 16 and 31 (incl.) 846 847(The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs) 848 849In both cases, level is used to assert/deassert the line. 850 851When KVM_CAP_ARM_IRQ_LINE_LAYOUT_2 is supported, the target vcpu is 852identified as (256 * vcpu2_index + vcpu_index). Otherwise, vcpu2_index 853must be zero. 854 855Note that on arm/arm64, the KVM_CAP_IRQCHIP capability only conditions 856injection of interrupts for the in-kernel irqchip. KVM_IRQ_LINE can always 857be used for a userspace interrupt controller. 858 859:: 860 861 struct kvm_irq_level { 862 union { 863 __u32 irq; /* GSI */ 864 __s32 status; /* not used for KVM_IRQ_LEVEL */ 865 }; 866 __u32 level; /* 0 or 1 */ 867 }; 868 869 8704.26 KVM_GET_IRQCHIP 871-------------------- 872 873:Capability: KVM_CAP_IRQCHIP 874:Architectures: x86 875:Type: vm ioctl 876:Parameters: struct kvm_irqchip (in/out) 877:Returns: 0 on success, -1 on error 878 879Reads the state of a kernel interrupt controller created with 880KVM_CREATE_IRQCHIP into a buffer provided by the caller. 881 882:: 883 884 struct kvm_irqchip { 885 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */ 886 __u32 pad; 887 union { 888 char dummy[512]; /* reserving space */ 889 struct kvm_pic_state pic; 890 struct kvm_ioapic_state ioapic; 891 } chip; 892 }; 893 894 8954.27 KVM_SET_IRQCHIP 896-------------------- 897 898:Capability: KVM_CAP_IRQCHIP 899:Architectures: x86 900:Type: vm ioctl 901:Parameters: struct kvm_irqchip (in) 902:Returns: 0 on success, -1 on error 903 904Sets the state of a kernel interrupt controller created with 905KVM_CREATE_IRQCHIP from a buffer provided by the caller. 906 907:: 908 909 struct kvm_irqchip { 910 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */ 911 __u32 pad; 912 union { 913 char dummy[512]; /* reserving space */ 914 struct kvm_pic_state pic; 915 struct kvm_ioapic_state ioapic; 916 } chip; 917 }; 918 919 9204.28 KVM_XEN_HVM_CONFIG 921----------------------- 922 923:Capability: KVM_CAP_XEN_HVM 924:Architectures: x86 925:Type: vm ioctl 926:Parameters: struct kvm_xen_hvm_config (in) 927:Returns: 0 on success, -1 on error 928 929Sets the MSR that the Xen HVM guest uses to initialize its hypercall 930page, and provides the starting address and size of the hypercall 931blobs in userspace. When the guest writes the MSR, kvm copies one 932page of a blob (32- or 64-bit, depending on the vcpu mode) to guest 933memory. 934 935:: 936 937 struct kvm_xen_hvm_config { 938 __u32 flags; 939 __u32 msr; 940 __u64 blob_addr_32; 941 __u64 blob_addr_64; 942 __u8 blob_size_32; 943 __u8 blob_size_64; 944 __u8 pad2[30]; 945 }; 946 947 9484.29 KVM_GET_CLOCK 949------------------ 950 951:Capability: KVM_CAP_ADJUST_CLOCK 952:Architectures: x86 953:Type: vm ioctl 954:Parameters: struct kvm_clock_data (out) 955:Returns: 0 on success, -1 on error 956 957Gets the current timestamp of kvmclock as seen by the current guest. In 958conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios 959such as migration. 960 961When KVM_CAP_ADJUST_CLOCK is passed to KVM_CHECK_EXTENSION, it returns the 962set of bits that KVM can return in struct kvm_clock_data's flag member. 963 964The only flag defined now is KVM_CLOCK_TSC_STABLE. If set, the returned 965value is the exact kvmclock value seen by all VCPUs at the instant 966when KVM_GET_CLOCK was called. If clear, the returned value is simply 967CLOCK_MONOTONIC plus a constant offset; the offset can be modified 968with KVM_SET_CLOCK. KVM will try to make all VCPUs follow this clock, 969but the exact value read by each VCPU could differ, because the host 970TSC is not stable. 971 972:: 973 974 struct kvm_clock_data { 975 __u64 clock; /* kvmclock current value */ 976 __u32 flags; 977 __u32 pad[9]; 978 }; 979 980 9814.30 KVM_SET_CLOCK 982------------------ 983 984:Capability: KVM_CAP_ADJUST_CLOCK 985:Architectures: x86 986:Type: vm ioctl 987:Parameters: struct kvm_clock_data (in) 988:Returns: 0 on success, -1 on error 989 990Sets the current timestamp of kvmclock to the value specified in its parameter. 991In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios 992such as migration. 993 994:: 995 996 struct kvm_clock_data { 997 __u64 clock; /* kvmclock current value */ 998 __u32 flags; 999 __u32 pad[9]; 1000 }; 1001 1002 10034.31 KVM_GET_VCPU_EVENTS 1004------------------------ 1005 1006:Capability: KVM_CAP_VCPU_EVENTS 1007:Extended by: KVM_CAP_INTR_SHADOW 1008:Architectures: x86, arm, arm64 1009:Type: vcpu ioctl 1010:Parameters: struct kvm_vcpu_event (out) 1011:Returns: 0 on success, -1 on error 1012 1013X86: 1014^^^^ 1015 1016Gets currently pending exceptions, interrupts, and NMIs as well as related 1017states of the vcpu. 1018 1019:: 1020 1021 struct kvm_vcpu_events { 1022 struct { 1023 __u8 injected; 1024 __u8 nr; 1025 __u8 has_error_code; 1026 __u8 pending; 1027 __u32 error_code; 1028 } exception; 1029 struct { 1030 __u8 injected; 1031 __u8 nr; 1032 __u8 soft; 1033 __u8 shadow; 1034 } interrupt; 1035 struct { 1036 __u8 injected; 1037 __u8 pending; 1038 __u8 masked; 1039 __u8 pad; 1040 } nmi; 1041 __u32 sipi_vector; 1042 __u32 flags; 1043 struct { 1044 __u8 smm; 1045 __u8 pending; 1046 __u8 smm_inside_nmi; 1047 __u8 latched_init; 1048 } smi; 1049 __u8 reserved[27]; 1050 __u8 exception_has_payload; 1051 __u64 exception_payload; 1052 }; 1053 1054The following bits are defined in the flags field: 1055 1056- KVM_VCPUEVENT_VALID_SHADOW may be set to signal that 1057 interrupt.shadow contains a valid state. 1058 1059- KVM_VCPUEVENT_VALID_SMM may be set to signal that smi contains a 1060 valid state. 1061 1062- KVM_VCPUEVENT_VALID_PAYLOAD may be set to signal that the 1063 exception_has_payload, exception_payload, and exception.pending 1064 fields contain a valid state. This bit will be set whenever 1065 KVM_CAP_EXCEPTION_PAYLOAD is enabled. 1066 1067ARM/ARM64: 1068^^^^^^^^^^ 1069 1070If the guest accesses a device that is being emulated by the host kernel in 1071such a way that a real device would generate a physical SError, KVM may make 1072a virtual SError pending for that VCPU. This system error interrupt remains 1073pending until the guest takes the exception by unmasking PSTATE.A. 1074 1075Running the VCPU may cause it to take a pending SError, or make an access that 1076causes an SError to become pending. The event's description is only valid while 1077the VPCU is not running. 1078 1079This API provides a way to read and write the pending 'event' state that is not 1080visible to the guest. To save, restore or migrate a VCPU the struct representing 1081the state can be read then written using this GET/SET API, along with the other 1082guest-visible registers. It is not possible to 'cancel' an SError that has been 1083made pending. 1084 1085A device being emulated in user-space may also wish to generate an SError. To do 1086this the events structure can be populated by user-space. The current state 1087should be read first, to ensure no existing SError is pending. If an existing 1088SError is pending, the architecture's 'Multiple SError interrupts' rules should 1089be followed. (2.5.3 of DDI0587.a "ARM Reliability, Availability, and 1090Serviceability (RAS) Specification"). 1091 1092SError exceptions always have an ESR value. Some CPUs have the ability to 1093specify what the virtual SError's ESR value should be. These systems will 1094advertise KVM_CAP_ARM_INJECT_SERROR_ESR. In this case exception.has_esr will 1095always have a non-zero value when read, and the agent making an SError pending 1096should specify the ISS field in the lower 24 bits of exception.serror_esr. If 1097the system supports KVM_CAP_ARM_INJECT_SERROR_ESR, but user-space sets the events 1098with exception.has_esr as zero, KVM will choose an ESR. 1099 1100Specifying exception.has_esr on a system that does not support it will return 1101-EINVAL. Setting anything other than the lower 24bits of exception.serror_esr 1102will return -EINVAL. 1103 1104It is not possible to read back a pending external abort (injected via 1105KVM_SET_VCPU_EVENTS or otherwise) because such an exception is always delivered 1106directly to the virtual CPU). 1107 1108:: 1109 1110 struct kvm_vcpu_events { 1111 struct { 1112 __u8 serror_pending; 1113 __u8 serror_has_esr; 1114 __u8 ext_dabt_pending; 1115 /* Align it to 8 bytes */ 1116 __u8 pad[5]; 1117 __u64 serror_esr; 1118 } exception; 1119 __u32 reserved[12]; 1120 }; 1121 11224.32 KVM_SET_VCPU_EVENTS 1123------------------------ 1124 1125:Capability: KVM_CAP_VCPU_EVENTS 1126:Extended by: KVM_CAP_INTR_SHADOW 1127:Architectures: x86, arm, arm64 1128:Type: vcpu ioctl 1129:Parameters: struct kvm_vcpu_event (in) 1130:Returns: 0 on success, -1 on error 1131 1132X86: 1133^^^^ 1134 1135Set pending exceptions, interrupts, and NMIs as well as related states of the 1136vcpu. 1137 1138See KVM_GET_VCPU_EVENTS for the data structure. 1139 1140Fields that may be modified asynchronously by running VCPUs can be excluded 1141from the update. These fields are nmi.pending, sipi_vector, smi.smm, 1142smi.pending. Keep the corresponding bits in the flags field cleared to 1143suppress overwriting the current in-kernel state. The bits are: 1144 1145=============================== ================================== 1146KVM_VCPUEVENT_VALID_NMI_PENDING transfer nmi.pending to the kernel 1147KVM_VCPUEVENT_VALID_SIPI_VECTOR transfer sipi_vector 1148KVM_VCPUEVENT_VALID_SMM transfer the smi sub-struct. 1149=============================== ================================== 1150 1151If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in 1152the flags field to signal that interrupt.shadow contains a valid state and 1153shall be written into the VCPU. 1154 1155KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available. 1156 1157If KVM_CAP_EXCEPTION_PAYLOAD is enabled, KVM_VCPUEVENT_VALID_PAYLOAD 1158can be set in the flags field to signal that the 1159exception_has_payload, exception_payload, and exception.pending fields 1160contain a valid state and shall be written into the VCPU. 1161 1162ARM/ARM64: 1163^^^^^^^^^^ 1164 1165User space may need to inject several types of events to the guest. 1166 1167Set the pending SError exception state for this VCPU. It is not possible to 1168'cancel' an Serror that has been made pending. 1169 1170If the guest performed an access to I/O memory which could not be handled by 1171userspace, for example because of missing instruction syndrome decode 1172information or because there is no device mapped at the accessed IPA, then 1173userspace can ask the kernel to inject an external abort using the address 1174from the exiting fault on the VCPU. It is a programming error to set 1175ext_dabt_pending after an exit which was not either KVM_EXIT_MMIO or 1176KVM_EXIT_ARM_NISV. This feature is only available if the system supports 1177KVM_CAP_ARM_INJECT_EXT_DABT. This is a helper which provides commonality in 1178how userspace reports accesses for the above cases to guests, across different 1179userspace implementations. Nevertheless, userspace can still emulate all Arm 1180exceptions by manipulating individual registers using the KVM_SET_ONE_REG API. 1181 1182See KVM_GET_VCPU_EVENTS for the data structure. 1183 1184 11854.33 KVM_GET_DEBUGREGS 1186---------------------- 1187 1188:Capability: KVM_CAP_DEBUGREGS 1189:Architectures: x86 1190:Type: vm ioctl 1191:Parameters: struct kvm_debugregs (out) 1192:Returns: 0 on success, -1 on error 1193 1194Reads debug registers from the vcpu. 1195 1196:: 1197 1198 struct kvm_debugregs { 1199 __u64 db[4]; 1200 __u64 dr6; 1201 __u64 dr7; 1202 __u64 flags; 1203 __u64 reserved[9]; 1204 }; 1205 1206 12074.34 KVM_SET_DEBUGREGS 1208---------------------- 1209 1210:Capability: KVM_CAP_DEBUGREGS 1211:Architectures: x86 1212:Type: vm ioctl 1213:Parameters: struct kvm_debugregs (in) 1214:Returns: 0 on success, -1 on error 1215 1216Writes debug registers into the vcpu. 1217 1218See KVM_GET_DEBUGREGS for the data structure. The flags field is unused 1219yet and must be cleared on entry. 1220 1221 12224.35 KVM_SET_USER_MEMORY_REGION 1223------------------------------- 1224 1225:Capability: KVM_CAP_USER_MEMORY 1226:Architectures: all 1227:Type: vm ioctl 1228:Parameters: struct kvm_userspace_memory_region (in) 1229:Returns: 0 on success, -1 on error 1230 1231:: 1232 1233 struct kvm_userspace_memory_region { 1234 __u32 slot; 1235 __u32 flags; 1236 __u64 guest_phys_addr; 1237 __u64 memory_size; /* bytes */ 1238 __u64 userspace_addr; /* start of the userspace allocated memory */ 1239 }; 1240 1241 /* for kvm_memory_region::flags */ 1242 #define KVM_MEM_LOG_DIRTY_PAGES (1UL << 0) 1243 #define KVM_MEM_READONLY (1UL << 1) 1244 1245This ioctl allows the user to create, modify or delete a guest physical 1246memory slot. Bits 0-15 of "slot" specify the slot id and this value 1247should be less than the maximum number of user memory slots supported per 1248VM. The maximum allowed slots can be queried using KVM_CAP_NR_MEMSLOTS. 1249Slots may not overlap in guest physical address space. 1250 1251If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot" 1252specifies the address space which is being modified. They must be 1253less than the value that KVM_CHECK_EXTENSION returns for the 1254KVM_CAP_MULTI_ADDRESS_SPACE capability. Slots in separate address spaces 1255are unrelated; the restriction on overlapping slots only applies within 1256each address space. 1257 1258Deleting a slot is done by passing zero for memory_size. When changing 1259an existing slot, it may be moved in the guest physical memory space, 1260or its flags may be modified, but it may not be resized. 1261 1262Memory for the region is taken starting at the address denoted by the 1263field userspace_addr, which must point at user addressable memory for 1264the entire memory slot size. Any object may back this memory, including 1265anonymous memory, ordinary files, and hugetlbfs. 1266 1267It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr 1268be identical. This allows large pages in the guest to be backed by large 1269pages in the host. 1270 1271The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and 1272KVM_MEM_READONLY. The former can be set to instruct KVM to keep track of 1273writes to memory within the slot. See KVM_GET_DIRTY_LOG ioctl to know how to 1274use it. The latter can be set, if KVM_CAP_READONLY_MEM capability allows it, 1275to make a new slot read-only. In this case, writes to this memory will be 1276posted to userspace as KVM_EXIT_MMIO exits. 1277 1278When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of 1279the memory region are automatically reflected into the guest. For example, an 1280mmap() that affects the region will be made visible immediately. Another 1281example is madvise(MADV_DROP). 1282 1283It is recommended to use this API instead of the KVM_SET_MEMORY_REGION ioctl. 1284The KVM_SET_MEMORY_REGION does not allow fine grained control over memory 1285allocation and is deprecated. 1286 1287 12884.36 KVM_SET_TSS_ADDR 1289--------------------- 1290 1291:Capability: KVM_CAP_SET_TSS_ADDR 1292:Architectures: x86 1293:Type: vm ioctl 1294:Parameters: unsigned long tss_address (in) 1295:Returns: 0 on success, -1 on error 1296 1297This ioctl defines the physical address of a three-page region in the guest 1298physical address space. The region must be within the first 4GB of the 1299guest physical address space and must not conflict with any memory slot 1300or any mmio address. The guest may malfunction if it accesses this memory 1301region. 1302 1303This ioctl is required on Intel-based hosts. This is needed on Intel hardware 1304because of a quirk in the virtualization implementation (see the internals 1305documentation when it pops into existence). 1306 1307 13084.37 KVM_ENABLE_CAP 1309------------------- 1310 1311:Capability: KVM_CAP_ENABLE_CAP 1312:Architectures: mips, ppc, s390 1313:Type: vcpu ioctl 1314:Parameters: struct kvm_enable_cap (in) 1315:Returns: 0 on success; -1 on error 1316 1317:Capability: KVM_CAP_ENABLE_CAP_VM 1318:Architectures: all 1319:Type: vcpu ioctl 1320:Parameters: struct kvm_enable_cap (in) 1321:Returns: 0 on success; -1 on error 1322 1323.. note:: 1324 1325 Not all extensions are enabled by default. Using this ioctl the application 1326 can enable an extension, making it available to the guest. 1327 1328On systems that do not support this ioctl, it always fails. On systems that 1329do support it, it only works for extensions that are supported for enablement. 1330 1331To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should 1332be used. 1333 1334:: 1335 1336 struct kvm_enable_cap { 1337 /* in */ 1338 __u32 cap; 1339 1340The capability that is supposed to get enabled. 1341 1342:: 1343 1344 __u32 flags; 1345 1346A bitfield indicating future enhancements. Has to be 0 for now. 1347 1348:: 1349 1350 __u64 args[4]; 1351 1352Arguments for enabling a feature. If a feature needs initial values to 1353function properly, this is the place to put them. 1354 1355:: 1356 1357 __u8 pad[64]; 1358 }; 1359 1360The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl 1361for vm-wide capabilities. 1362 13634.38 KVM_GET_MP_STATE 1364--------------------- 1365 1366:Capability: KVM_CAP_MP_STATE 1367:Architectures: x86, s390, arm, arm64 1368:Type: vcpu ioctl 1369:Parameters: struct kvm_mp_state (out) 1370:Returns: 0 on success; -1 on error 1371 1372:: 1373 1374 struct kvm_mp_state { 1375 __u32 mp_state; 1376 }; 1377 1378Returns the vcpu's current "multiprocessing state" (though also valid on 1379uniprocessor guests). 1380 1381Possible values are: 1382 1383 ========================== =============================================== 1384 KVM_MP_STATE_RUNNABLE the vcpu is currently running [x86,arm/arm64] 1385 KVM_MP_STATE_UNINITIALIZED the vcpu is an application processor (AP) 1386 which has not yet received an INIT signal [x86] 1387 KVM_MP_STATE_INIT_RECEIVED the vcpu has received an INIT signal, and is 1388 now ready for a SIPI [x86] 1389 KVM_MP_STATE_HALTED the vcpu has executed a HLT instruction and 1390 is waiting for an interrupt [x86] 1391 KVM_MP_STATE_SIPI_RECEIVED the vcpu has just received a SIPI (vector 1392 accessible via KVM_GET_VCPU_EVENTS) [x86] 1393 KVM_MP_STATE_STOPPED the vcpu is stopped [s390,arm/arm64] 1394 KVM_MP_STATE_CHECK_STOP the vcpu is in a special error state [s390] 1395 KVM_MP_STATE_OPERATING the vcpu is operating (running or halted) 1396 [s390] 1397 KVM_MP_STATE_LOAD the vcpu is in a special load/startup state 1398 [s390] 1399 ========================== =============================================== 1400 1401On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an 1402in-kernel irqchip, the multiprocessing state must be maintained by userspace on 1403these architectures. 1404 1405For arm/arm64: 1406^^^^^^^^^^^^^^ 1407 1408The only states that are valid are KVM_MP_STATE_STOPPED and 1409KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not. 1410 14114.39 KVM_SET_MP_STATE 1412--------------------- 1413 1414:Capability: KVM_CAP_MP_STATE 1415:Architectures: x86, s390, arm, arm64 1416:Type: vcpu ioctl 1417:Parameters: struct kvm_mp_state (in) 1418:Returns: 0 on success; -1 on error 1419 1420Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for 1421arguments. 1422 1423On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an 1424in-kernel irqchip, the multiprocessing state must be maintained by userspace on 1425these architectures. 1426 1427For arm/arm64: 1428^^^^^^^^^^^^^^ 1429 1430The only states that are valid are KVM_MP_STATE_STOPPED and 1431KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not. 1432 14334.40 KVM_SET_IDENTITY_MAP_ADDR 1434------------------------------ 1435 1436:Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR 1437:Architectures: x86 1438:Type: vm ioctl 1439:Parameters: unsigned long identity (in) 1440:Returns: 0 on success, -1 on error 1441 1442This ioctl defines the physical address of a one-page region in the guest 1443physical address space. The region must be within the first 4GB of the 1444guest physical address space and must not conflict with any memory slot 1445or any mmio address. The guest may malfunction if it accesses this memory 1446region. 1447 1448Setting the address to 0 will result in resetting the address to its default 1449(0xfffbc000). 1450 1451This ioctl is required on Intel-based hosts. This is needed on Intel hardware 1452because of a quirk in the virtualization implementation (see the internals 1453documentation when it pops into existence). 1454 1455Fails if any VCPU has already been created. 1456 14574.41 KVM_SET_BOOT_CPU_ID 1458------------------------ 1459 1460:Capability: KVM_CAP_SET_BOOT_CPU_ID 1461:Architectures: x86 1462:Type: vm ioctl 1463:Parameters: unsigned long vcpu_id 1464:Returns: 0 on success, -1 on error 1465 1466Define which vcpu is the Bootstrap Processor (BSP). Values are the same 1467as the vcpu id in KVM_CREATE_VCPU. If this ioctl is not called, the default 1468is vcpu 0. 1469 1470 14714.42 KVM_GET_XSAVE 1472------------------ 1473 1474:Capability: KVM_CAP_XSAVE 1475:Architectures: x86 1476:Type: vcpu ioctl 1477:Parameters: struct kvm_xsave (out) 1478:Returns: 0 on success, -1 on error 1479 1480 1481:: 1482 1483 struct kvm_xsave { 1484 __u32 region[1024]; 1485 }; 1486 1487This ioctl would copy current vcpu's xsave struct to the userspace. 1488 1489 14904.43 KVM_SET_XSAVE 1491------------------ 1492 1493:Capability: KVM_CAP_XSAVE 1494:Architectures: x86 1495:Type: vcpu ioctl 1496:Parameters: struct kvm_xsave (in) 1497:Returns: 0 on success, -1 on error 1498 1499:: 1500 1501 1502 struct kvm_xsave { 1503 __u32 region[1024]; 1504 }; 1505 1506This ioctl would copy userspace's xsave struct to the kernel. 1507 1508 15094.44 KVM_GET_XCRS 1510----------------- 1511 1512:Capability: KVM_CAP_XCRS 1513:Architectures: x86 1514:Type: vcpu ioctl 1515:Parameters: struct kvm_xcrs (out) 1516:Returns: 0 on success, -1 on error 1517 1518:: 1519 1520 struct kvm_xcr { 1521 __u32 xcr; 1522 __u32 reserved; 1523 __u64 value; 1524 }; 1525 1526 struct kvm_xcrs { 1527 __u32 nr_xcrs; 1528 __u32 flags; 1529 struct kvm_xcr xcrs[KVM_MAX_XCRS]; 1530 __u64 padding[16]; 1531 }; 1532 1533This ioctl would copy current vcpu's xcrs to the userspace. 1534 1535 15364.45 KVM_SET_XCRS 1537----------------- 1538 1539:Capability: KVM_CAP_XCRS 1540:Architectures: x86 1541:Type: vcpu ioctl 1542:Parameters: struct kvm_xcrs (in) 1543:Returns: 0 on success, -1 on error 1544 1545:: 1546 1547 struct kvm_xcr { 1548 __u32 xcr; 1549 __u32 reserved; 1550 __u64 value; 1551 }; 1552 1553 struct kvm_xcrs { 1554 __u32 nr_xcrs; 1555 __u32 flags; 1556 struct kvm_xcr xcrs[KVM_MAX_XCRS]; 1557 __u64 padding[16]; 1558 }; 1559 1560This ioctl would set vcpu's xcr to the value userspace specified. 1561 1562 15634.46 KVM_GET_SUPPORTED_CPUID 1564---------------------------- 1565 1566:Capability: KVM_CAP_EXT_CPUID 1567:Architectures: x86 1568:Type: system ioctl 1569:Parameters: struct kvm_cpuid2 (in/out) 1570:Returns: 0 on success, -1 on error 1571 1572:: 1573 1574 struct kvm_cpuid2 { 1575 __u32 nent; 1576 __u32 padding; 1577 struct kvm_cpuid_entry2 entries[0]; 1578 }; 1579 1580 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0) 1581 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */ 1582 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */ 1583 1584 struct kvm_cpuid_entry2 { 1585 __u32 function; 1586 __u32 index; 1587 __u32 flags; 1588 __u32 eax; 1589 __u32 ebx; 1590 __u32 ecx; 1591 __u32 edx; 1592 __u32 padding[3]; 1593 }; 1594 1595This ioctl returns x86 cpuid features which are supported by both the 1596hardware and kvm in its default configuration. Userspace can use the 1597information returned by this ioctl to construct cpuid information (for 1598KVM_SET_CPUID2) that is consistent with hardware, kernel, and 1599userspace capabilities, and with user requirements (for example, the 1600user may wish to constrain cpuid to emulate older hardware, or for 1601feature consistency across a cluster). 1602 1603Note that certain capabilities, such as KVM_CAP_X86_DISABLE_EXITS, may 1604expose cpuid features (e.g. MONITOR) which are not supported by kvm in 1605its default configuration. If userspace enables such capabilities, it 1606is responsible for modifying the results of this ioctl appropriately. 1607 1608Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure 1609with the 'nent' field indicating the number of entries in the variable-size 1610array 'entries'. If the number of entries is too low to describe the cpu 1611capabilities, an error (E2BIG) is returned. If the number is too high, 1612the 'nent' field is adjusted and an error (ENOMEM) is returned. If the 1613number is just right, the 'nent' field is adjusted to the number of valid 1614entries in the 'entries' array, which is then filled. 1615 1616The entries returned are the host cpuid as returned by the cpuid instruction, 1617with unknown or unsupported features masked out. Some features (for example, 1618x2apic), may not be present in the host cpu, but are exposed by kvm if it can 1619emulate them efficiently. The fields in each entry are defined as follows: 1620 1621 function: 1622 the eax value used to obtain the entry 1623 1624 index: 1625 the ecx value used to obtain the entry (for entries that are 1626 affected by ecx) 1627 1628 flags: 1629 an OR of zero or more of the following: 1630 1631 KVM_CPUID_FLAG_SIGNIFCANT_INDEX: 1632 if the index field is valid 1633 1634 eax, ebx, ecx, edx: 1635 the values returned by the cpuid instruction for 1636 this function/index combination 1637 1638The TSC deadline timer feature (CPUID leaf 1, ecx[24]) is always returned 1639as false, since the feature depends on KVM_CREATE_IRQCHIP for local APIC 1640support. Instead it is reported via:: 1641 1642 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER) 1643 1644if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the 1645feature in userspace, then you can enable the feature for KVM_SET_CPUID2. 1646 1647 16484.47 KVM_PPC_GET_PVINFO 1649----------------------- 1650 1651:Capability: KVM_CAP_PPC_GET_PVINFO 1652:Architectures: ppc 1653:Type: vm ioctl 1654:Parameters: struct kvm_ppc_pvinfo (out) 1655:Returns: 0 on success, !0 on error 1656 1657:: 1658 1659 struct kvm_ppc_pvinfo { 1660 __u32 flags; 1661 __u32 hcall[4]; 1662 __u8 pad[108]; 1663 }; 1664 1665This ioctl fetches PV specific information that need to be passed to the guest 1666using the device tree or other means from vm context. 1667 1668The hcall array defines 4 instructions that make up a hypercall. 1669 1670If any additional field gets added to this structure later on, a bit for that 1671additional piece of information will be set in the flags bitmap. 1672 1673The flags bitmap is defined as:: 1674 1675 /* the host supports the ePAPR idle hcall 1676 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1<<0) 1677 16784.52 KVM_SET_GSI_ROUTING 1679------------------------ 1680 1681:Capability: KVM_CAP_IRQ_ROUTING 1682:Architectures: x86 s390 arm arm64 1683:Type: vm ioctl 1684:Parameters: struct kvm_irq_routing (in) 1685:Returns: 0 on success, -1 on error 1686 1687Sets the GSI routing table entries, overwriting any previously set entries. 1688 1689On arm/arm64, GSI routing has the following limitation: 1690 1691- GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD. 1692 1693:: 1694 1695 struct kvm_irq_routing { 1696 __u32 nr; 1697 __u32 flags; 1698 struct kvm_irq_routing_entry entries[0]; 1699 }; 1700 1701No flags are specified so far, the corresponding field must be set to zero. 1702 1703:: 1704 1705 struct kvm_irq_routing_entry { 1706 __u32 gsi; 1707 __u32 type; 1708 __u32 flags; 1709 __u32 pad; 1710 union { 1711 struct kvm_irq_routing_irqchip irqchip; 1712 struct kvm_irq_routing_msi msi; 1713 struct kvm_irq_routing_s390_adapter adapter; 1714 struct kvm_irq_routing_hv_sint hv_sint; 1715 __u32 pad[8]; 1716 } u; 1717 }; 1718 1719 /* gsi routing entry types */ 1720 #define KVM_IRQ_ROUTING_IRQCHIP 1 1721 #define KVM_IRQ_ROUTING_MSI 2 1722 #define KVM_IRQ_ROUTING_S390_ADAPTER 3 1723 #define KVM_IRQ_ROUTING_HV_SINT 4 1724 1725flags: 1726 1727- KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry 1728 type, specifies that the devid field contains a valid value. The per-VM 1729 KVM_CAP_MSI_DEVID capability advertises the requirement to provide 1730 the device ID. If this capability is not available, userspace should 1731 never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail. 1732- zero otherwise 1733 1734:: 1735 1736 struct kvm_irq_routing_irqchip { 1737 __u32 irqchip; 1738 __u32 pin; 1739 }; 1740 1741 struct kvm_irq_routing_msi { 1742 __u32 address_lo; 1743 __u32 address_hi; 1744 __u32 data; 1745 union { 1746 __u32 pad; 1747 __u32 devid; 1748 }; 1749 }; 1750 1751If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier 1752for the device that wrote the MSI message. For PCI, this is usually a 1753BFD identifier in the lower 16 bits. 1754 1755On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS 1756feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled, 1757address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of 1758address_hi must be zero. 1759 1760:: 1761 1762 struct kvm_irq_routing_s390_adapter { 1763 __u64 ind_addr; 1764 __u64 summary_addr; 1765 __u64 ind_offset; 1766 __u32 summary_offset; 1767 __u32 adapter_id; 1768 }; 1769 1770 struct kvm_irq_routing_hv_sint { 1771 __u32 vcpu; 1772 __u32 sint; 1773 }; 1774 1775 17764.55 KVM_SET_TSC_KHZ 1777-------------------- 1778 1779:Capability: KVM_CAP_TSC_CONTROL 1780:Architectures: x86 1781:Type: vcpu ioctl 1782:Parameters: virtual tsc_khz 1783:Returns: 0 on success, -1 on error 1784 1785Specifies the tsc frequency for the virtual machine. The unit of the 1786frequency is KHz. 1787 1788 17894.56 KVM_GET_TSC_KHZ 1790-------------------- 1791 1792:Capability: KVM_CAP_GET_TSC_KHZ 1793:Architectures: x86 1794:Type: vcpu ioctl 1795:Parameters: none 1796:Returns: virtual tsc-khz on success, negative value on error 1797 1798Returns the tsc frequency of the guest. The unit of the return value is 1799KHz. If the host has unstable tsc this ioctl returns -EIO instead as an 1800error. 1801 1802 18034.57 KVM_GET_LAPIC 1804------------------ 1805 1806:Capability: KVM_CAP_IRQCHIP 1807:Architectures: x86 1808:Type: vcpu ioctl 1809:Parameters: struct kvm_lapic_state (out) 1810:Returns: 0 on success, -1 on error 1811 1812:: 1813 1814 #define KVM_APIC_REG_SIZE 0x400 1815 struct kvm_lapic_state { 1816 char regs[KVM_APIC_REG_SIZE]; 1817 }; 1818 1819Reads the Local APIC registers and copies them into the input argument. The 1820data format and layout are the same as documented in the architecture manual. 1821 1822If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is 1823enabled, then the format of APIC_ID register depends on the APIC mode 1824(reported by MSR_IA32_APICBASE) of its VCPU. x2APIC stores APIC ID in 1825the APIC_ID register (bytes 32-35). xAPIC only allows an 8-bit APIC ID 1826which is stored in bits 31-24 of the APIC register, or equivalently in 1827byte 35 of struct kvm_lapic_state's regs field. KVM_GET_LAPIC must then 1828be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR. 1829 1830If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state 1831always uses xAPIC format. 1832 1833 18344.58 KVM_SET_LAPIC 1835------------------ 1836 1837:Capability: KVM_CAP_IRQCHIP 1838:Architectures: x86 1839:Type: vcpu ioctl 1840:Parameters: struct kvm_lapic_state (in) 1841:Returns: 0 on success, -1 on error 1842 1843:: 1844 1845 #define KVM_APIC_REG_SIZE 0x400 1846 struct kvm_lapic_state { 1847 char regs[KVM_APIC_REG_SIZE]; 1848 }; 1849 1850Copies the input argument into the Local APIC registers. The data format 1851and layout are the same as documented in the architecture manual. 1852 1853The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's 1854regs field) depends on the state of the KVM_CAP_X2APIC_API capability. 1855See the note in KVM_GET_LAPIC. 1856 1857 18584.59 KVM_IOEVENTFD 1859------------------ 1860 1861:Capability: KVM_CAP_IOEVENTFD 1862:Architectures: all 1863:Type: vm ioctl 1864:Parameters: struct kvm_ioeventfd (in) 1865:Returns: 0 on success, !0 on error 1866 1867This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address 1868within the guest. A guest write in the registered address will signal the 1869provided event instead of triggering an exit. 1870 1871:: 1872 1873 struct kvm_ioeventfd { 1874 __u64 datamatch; 1875 __u64 addr; /* legal pio/mmio address */ 1876 __u32 len; /* 0, 1, 2, 4, or 8 bytes */ 1877 __s32 fd; 1878 __u32 flags; 1879 __u8 pad[36]; 1880 }; 1881 1882For the special case of virtio-ccw devices on s390, the ioevent is matched 1883to a subchannel/virtqueue tuple instead. 1884 1885The following flags are defined:: 1886 1887 #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch) 1888 #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio) 1889 #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign) 1890 #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \ 1891 (1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify) 1892 1893If datamatch flag is set, the event will be signaled only if the written value 1894to the registered address is equal to datamatch in struct kvm_ioeventfd. 1895 1896For virtio-ccw devices, addr contains the subchannel id and datamatch the 1897virtqueue index. 1898 1899With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and 1900the kernel will ignore the length of guest write and may get a faster vmexit. 1901The speedup may only apply to specific architectures, but the ioeventfd will 1902work anyway. 1903 19044.60 KVM_DIRTY_TLB 1905------------------ 1906 1907:Capability: KVM_CAP_SW_TLB 1908:Architectures: ppc 1909:Type: vcpu ioctl 1910:Parameters: struct kvm_dirty_tlb (in) 1911:Returns: 0 on success, -1 on error 1912 1913:: 1914 1915 struct kvm_dirty_tlb { 1916 __u64 bitmap; 1917 __u32 num_dirty; 1918 }; 1919 1920This must be called whenever userspace has changed an entry in the shared 1921TLB, prior to calling KVM_RUN on the associated vcpu. 1922 1923The "bitmap" field is the userspace address of an array. This array 1924consists of a number of bits, equal to the total number of TLB entries as 1925determined by the last successful call to KVM_CONFIG_TLB, rounded up to the 1926nearest multiple of 64. 1927 1928Each bit corresponds to one TLB entry, ordered the same as in the shared TLB 1929array. 1930 1931The array is little-endian: the bit 0 is the least significant bit of the 1932first byte, bit 8 is the least significant bit of the second byte, etc. 1933This avoids any complications with differing word sizes. 1934 1935The "num_dirty" field is a performance hint for KVM to determine whether it 1936should skip processing the bitmap and just invalidate everything. It must 1937be set to the number of set bits in the bitmap. 1938 1939 19404.62 KVM_CREATE_SPAPR_TCE 1941------------------------- 1942 1943:Capability: KVM_CAP_SPAPR_TCE 1944:Architectures: powerpc 1945:Type: vm ioctl 1946:Parameters: struct kvm_create_spapr_tce (in) 1947:Returns: file descriptor for manipulating the created TCE table 1948 1949This creates a virtual TCE (translation control entry) table, which 1950is an IOMMU for PAPR-style virtual I/O. It is used to translate 1951logical addresses used in virtual I/O into guest physical addresses, 1952and provides a scatter/gather capability for PAPR virtual I/O. 1953 1954:: 1955 1956 /* for KVM_CAP_SPAPR_TCE */ 1957 struct kvm_create_spapr_tce { 1958 __u64 liobn; 1959 __u32 window_size; 1960 }; 1961 1962The liobn field gives the logical IO bus number for which to create a 1963TCE table. The window_size field specifies the size of the DMA window 1964which this TCE table will translate - the table will contain one 64 1965bit TCE entry for every 4kiB of the DMA window. 1966 1967When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE 1968table has been created using this ioctl(), the kernel will handle it 1969in real mode, updating the TCE table. H_PUT_TCE calls for other 1970liobns will cause a vm exit and must be handled by userspace. 1971 1972The return value is a file descriptor which can be passed to mmap(2) 1973to map the created TCE table into userspace. This lets userspace read 1974the entries written by kernel-handled H_PUT_TCE calls, and also lets 1975userspace update the TCE table directly which is useful in some 1976circumstances. 1977 1978 19794.63 KVM_ALLOCATE_RMA 1980--------------------- 1981 1982:Capability: KVM_CAP_PPC_RMA 1983:Architectures: powerpc 1984:Type: vm ioctl 1985:Parameters: struct kvm_allocate_rma (out) 1986:Returns: file descriptor for mapping the allocated RMA 1987 1988This allocates a Real Mode Area (RMA) from the pool allocated at boot 1989time by the kernel. An RMA is a physically-contiguous, aligned region 1990of memory used on older POWER processors to provide the memory which 1991will be accessed by real-mode (MMU off) accesses in a KVM guest. 1992POWER processors support a set of sizes for the RMA that usually 1993includes 64MB, 128MB, 256MB and some larger powers of two. 1994 1995:: 1996 1997 /* for KVM_ALLOCATE_RMA */ 1998 struct kvm_allocate_rma { 1999 __u64 rma_size; 2000 }; 2001 2002The return value is a file descriptor which can be passed to mmap(2) 2003to map the allocated RMA into userspace. The mapped area can then be 2004passed to the KVM_SET_USER_MEMORY_REGION ioctl to establish it as the 2005RMA for a virtual machine. The size of the RMA in bytes (which is 2006fixed at host kernel boot time) is returned in the rma_size field of 2007the argument structure. 2008 2009The KVM_CAP_PPC_RMA capability is 1 or 2 if the KVM_ALLOCATE_RMA ioctl 2010is supported; 2 if the processor requires all virtual machines to have 2011an RMA, or 1 if the processor can use an RMA but doesn't require it, 2012because it supports the Virtual RMA (VRMA) facility. 2013 2014 20154.64 KVM_NMI 2016------------ 2017 2018:Capability: KVM_CAP_USER_NMI 2019:Architectures: x86 2020:Type: vcpu ioctl 2021:Parameters: none 2022:Returns: 0 on success, -1 on error 2023 2024Queues an NMI on the thread's vcpu. Note this is well defined only 2025when KVM_CREATE_IRQCHIP has not been called, since this is an interface 2026between the virtual cpu core and virtual local APIC. After KVM_CREATE_IRQCHIP 2027has been called, this interface is completely emulated within the kernel. 2028 2029To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the 2030following algorithm: 2031 2032 - pause the vcpu 2033 - read the local APIC's state (KVM_GET_LAPIC) 2034 - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1) 2035 - if so, issue KVM_NMI 2036 - resume the vcpu 2037 2038Some guests configure the LINT1 NMI input to cause a panic, aiding in 2039debugging. 2040 2041 20424.65 KVM_S390_UCAS_MAP 2043---------------------- 2044 2045:Capability: KVM_CAP_S390_UCONTROL 2046:Architectures: s390 2047:Type: vcpu ioctl 2048:Parameters: struct kvm_s390_ucas_mapping (in) 2049:Returns: 0 in case of success 2050 2051The parameter is defined like this:: 2052 2053 struct kvm_s390_ucas_mapping { 2054 __u64 user_addr; 2055 __u64 vcpu_addr; 2056 __u64 length; 2057 }; 2058 2059This ioctl maps the memory at "user_addr" with the length "length" to 2060the vcpu's address space starting at "vcpu_addr". All parameters need to 2061be aligned by 1 megabyte. 2062 2063 20644.66 KVM_S390_UCAS_UNMAP 2065------------------------ 2066 2067:Capability: KVM_CAP_S390_UCONTROL 2068:Architectures: s390 2069:Type: vcpu ioctl 2070:Parameters: struct kvm_s390_ucas_mapping (in) 2071:Returns: 0 in case of success 2072 2073The parameter is defined like this:: 2074 2075 struct kvm_s390_ucas_mapping { 2076 __u64 user_addr; 2077 __u64 vcpu_addr; 2078 __u64 length; 2079 }; 2080 2081This ioctl unmaps the memory in the vcpu's address space starting at 2082"vcpu_addr" with the length "length". The field "user_addr" is ignored. 2083All parameters need to be aligned by 1 megabyte. 2084 2085 20864.67 KVM_S390_VCPU_FAULT 2087------------------------ 2088 2089:Capability: KVM_CAP_S390_UCONTROL 2090:Architectures: s390 2091:Type: vcpu ioctl 2092:Parameters: vcpu absolute address (in) 2093:Returns: 0 in case of success 2094 2095This call creates a page table entry on the virtual cpu's address space 2096(for user controlled virtual machines) or the virtual machine's address 2097space (for regular virtual machines). This only works for minor faults, 2098thus it's recommended to access subject memory page via the user page 2099table upfront. This is useful to handle validity intercepts for user 2100controlled virtual machines to fault in the virtual cpu's lowcore pages 2101prior to calling the KVM_RUN ioctl. 2102 2103 21044.68 KVM_SET_ONE_REG 2105-------------------- 2106 2107:Capability: KVM_CAP_ONE_REG 2108:Architectures: all 2109:Type: vcpu ioctl 2110:Parameters: struct kvm_one_reg (in) 2111:Returns: 0 on success, negative value on failure 2112 2113Errors: 2114 2115 ====== ============================================================ 2116 ENOENT no such register 2117 EINVAL invalid register ID, or no such register or used with VMs in 2118 protected virtualization mode on s390 2119 EPERM (arm64) register access not allowed before vcpu finalization 2120 ====== ============================================================ 2121 2122(These error codes are indicative only: do not rely on a specific error 2123code being returned in a specific situation.) 2124 2125:: 2126 2127 struct kvm_one_reg { 2128 __u64 id; 2129 __u64 addr; 2130 }; 2131 2132Using this ioctl, a single vcpu register can be set to a specific value 2133defined by user space with the passed in struct kvm_one_reg, where id 2134refers to the register identifier as described below and addr is a pointer 2135to a variable with the respective size. There can be architecture agnostic 2136and architecture specific registers. Each have their own range of operation 2137and their own constants and width. To keep track of the implemented 2138registers, find a list below: 2139 2140 ======= =============================== ============ 2141 Arch Register Width (bits) 2142 ======= =============================== ============ 2143 PPC KVM_REG_PPC_HIOR 64 2144 PPC KVM_REG_PPC_IAC1 64 2145 PPC KVM_REG_PPC_IAC2 64 2146 PPC KVM_REG_PPC_IAC3 64 2147 PPC KVM_REG_PPC_IAC4 64 2148 PPC KVM_REG_PPC_DAC1 64 2149 PPC KVM_REG_PPC_DAC2 64 2150 PPC KVM_REG_PPC_DABR 64 2151 PPC KVM_REG_PPC_DSCR 64 2152 PPC KVM_REG_PPC_PURR 64 2153 PPC KVM_REG_PPC_SPURR 64 2154 PPC KVM_REG_PPC_DAR 64 2155 PPC KVM_REG_PPC_DSISR 32 2156 PPC KVM_REG_PPC_AMR 64 2157 PPC KVM_REG_PPC_UAMOR 64 2158 PPC KVM_REG_PPC_MMCR0 64 2159 PPC KVM_REG_PPC_MMCR1 64 2160 PPC KVM_REG_PPC_MMCRA 64 2161 PPC KVM_REG_PPC_MMCR2 64 2162 PPC KVM_REG_PPC_MMCRS 64 2163 PPC KVM_REG_PPC_SIAR 64 2164 PPC KVM_REG_PPC_SDAR 64 2165 PPC KVM_REG_PPC_SIER 64 2166 PPC KVM_REG_PPC_PMC1 32 2167 PPC KVM_REG_PPC_PMC2 32 2168 PPC KVM_REG_PPC_PMC3 32 2169 PPC KVM_REG_PPC_PMC4 32 2170 PPC KVM_REG_PPC_PMC5 32 2171 PPC KVM_REG_PPC_PMC6 32 2172 PPC KVM_REG_PPC_PMC7 32 2173 PPC KVM_REG_PPC_PMC8 32 2174 PPC KVM_REG_PPC_FPR0 64 2175 ... 2176 PPC KVM_REG_PPC_FPR31 64 2177 PPC KVM_REG_PPC_VR0 128 2178 ... 2179 PPC KVM_REG_PPC_VR31 128 2180 PPC KVM_REG_PPC_VSR0 128 2181 ... 2182 PPC KVM_REG_PPC_VSR31 128 2183 PPC KVM_REG_PPC_FPSCR 64 2184 PPC KVM_REG_PPC_VSCR 32 2185 PPC KVM_REG_PPC_VPA_ADDR 64 2186 PPC KVM_REG_PPC_VPA_SLB 128 2187 PPC KVM_REG_PPC_VPA_DTL 128 2188 PPC KVM_REG_PPC_EPCR 32 2189 PPC KVM_REG_PPC_EPR 32 2190 PPC KVM_REG_PPC_TCR 32 2191 PPC KVM_REG_PPC_TSR 32 2192 PPC KVM_REG_PPC_OR_TSR 32 2193 PPC KVM_REG_PPC_CLEAR_TSR 32 2194 PPC KVM_REG_PPC_MAS0 32 2195 PPC KVM_REG_PPC_MAS1 32 2196 PPC KVM_REG_PPC_MAS2 64 2197 PPC KVM_REG_PPC_MAS7_3 64 2198 PPC KVM_REG_PPC_MAS4 32 2199 PPC KVM_REG_PPC_MAS6 32 2200 PPC KVM_REG_PPC_MMUCFG 32 2201 PPC KVM_REG_PPC_TLB0CFG 32 2202 PPC KVM_REG_PPC_TLB1CFG 32 2203 PPC KVM_REG_PPC_TLB2CFG 32 2204 PPC KVM_REG_PPC_TLB3CFG 32 2205 PPC KVM_REG_PPC_TLB0PS 32 2206 PPC KVM_REG_PPC_TLB1PS 32 2207 PPC KVM_REG_PPC_TLB2PS 32 2208 PPC KVM_REG_PPC_TLB3PS 32 2209 PPC KVM_REG_PPC_EPTCFG 32 2210 PPC KVM_REG_PPC_ICP_STATE 64 2211 PPC KVM_REG_PPC_VP_STATE 128 2212 PPC KVM_REG_PPC_TB_OFFSET 64 2213 PPC KVM_REG_PPC_SPMC1 32 2214 PPC KVM_REG_PPC_SPMC2 32 2215 PPC KVM_REG_PPC_IAMR 64 2216 PPC KVM_REG_PPC_TFHAR 64 2217 PPC KVM_REG_PPC_TFIAR 64 2218 PPC KVM_REG_PPC_TEXASR 64 2219 PPC KVM_REG_PPC_FSCR 64 2220 PPC KVM_REG_PPC_PSPB 32 2221 PPC KVM_REG_PPC_EBBHR 64 2222 PPC KVM_REG_PPC_EBBRR 64 2223 PPC KVM_REG_PPC_BESCR 64 2224 PPC KVM_REG_PPC_TAR 64 2225 PPC KVM_REG_PPC_DPDES 64 2226 PPC KVM_REG_PPC_DAWR 64 2227 PPC KVM_REG_PPC_DAWRX 64 2228 PPC KVM_REG_PPC_CIABR 64 2229 PPC KVM_REG_PPC_IC 64 2230 PPC KVM_REG_PPC_VTB 64 2231 PPC KVM_REG_PPC_CSIGR 64 2232 PPC KVM_REG_PPC_TACR 64 2233 PPC KVM_REG_PPC_TCSCR 64 2234 PPC KVM_REG_PPC_PID 64 2235 PPC KVM_REG_PPC_ACOP 64 2236 PPC KVM_REG_PPC_VRSAVE 32 2237 PPC KVM_REG_PPC_LPCR 32 2238 PPC KVM_REG_PPC_LPCR_64 64 2239 PPC KVM_REG_PPC_PPR 64 2240 PPC KVM_REG_PPC_ARCH_COMPAT 32 2241 PPC KVM_REG_PPC_DABRX 32 2242 PPC KVM_REG_PPC_WORT 64 2243 PPC KVM_REG_PPC_SPRG9 64 2244 PPC KVM_REG_PPC_DBSR 32 2245 PPC KVM_REG_PPC_TIDR 64 2246 PPC KVM_REG_PPC_PSSCR 64 2247 PPC KVM_REG_PPC_DEC_EXPIRY 64 2248 PPC KVM_REG_PPC_PTCR 64 2249 PPC KVM_REG_PPC_TM_GPR0 64 2250 ... 2251 PPC KVM_REG_PPC_TM_GPR31 64 2252 PPC KVM_REG_PPC_TM_VSR0 128 2253 ... 2254 PPC KVM_REG_PPC_TM_VSR63 128 2255 PPC KVM_REG_PPC_TM_CR 64 2256 PPC KVM_REG_PPC_TM_LR 64 2257 PPC KVM_REG_PPC_TM_CTR 64 2258 PPC KVM_REG_PPC_TM_FPSCR 64 2259 PPC KVM_REG_PPC_TM_AMR 64 2260 PPC KVM_REG_PPC_TM_PPR 64 2261 PPC KVM_REG_PPC_TM_VRSAVE 64 2262 PPC KVM_REG_PPC_TM_VSCR 32 2263 PPC KVM_REG_PPC_TM_DSCR 64 2264 PPC KVM_REG_PPC_TM_TAR 64 2265 PPC KVM_REG_PPC_TM_XER 64 2266 2267 MIPS KVM_REG_MIPS_R0 64 2268 ... 2269 MIPS KVM_REG_MIPS_R31 64 2270 MIPS KVM_REG_MIPS_HI 64 2271 MIPS KVM_REG_MIPS_LO 64 2272 MIPS KVM_REG_MIPS_PC 64 2273 MIPS KVM_REG_MIPS_CP0_INDEX 32 2274 MIPS KVM_REG_MIPS_CP0_ENTRYLO0 64 2275 MIPS KVM_REG_MIPS_CP0_ENTRYLO1 64 2276 MIPS KVM_REG_MIPS_CP0_CONTEXT 64 2277 MIPS KVM_REG_MIPS_CP0_CONTEXTCONFIG 32 2278 MIPS KVM_REG_MIPS_CP0_USERLOCAL 64 2279 MIPS KVM_REG_MIPS_CP0_XCONTEXTCONFIG 64 2280 MIPS KVM_REG_MIPS_CP0_PAGEMASK 32 2281 MIPS KVM_REG_MIPS_CP0_PAGEGRAIN 32 2282 MIPS KVM_REG_MIPS_CP0_SEGCTL0 64 2283 MIPS KVM_REG_MIPS_CP0_SEGCTL1 64 2284 MIPS KVM_REG_MIPS_CP0_SEGCTL2 64 2285 MIPS KVM_REG_MIPS_CP0_PWBASE 64 2286 MIPS KVM_REG_MIPS_CP0_PWFIELD 64 2287 MIPS KVM_REG_MIPS_CP0_PWSIZE 64 2288 MIPS KVM_REG_MIPS_CP0_WIRED 32 2289 MIPS KVM_REG_MIPS_CP0_PWCTL 32 2290 MIPS KVM_REG_MIPS_CP0_HWRENA 32 2291 MIPS KVM_REG_MIPS_CP0_BADVADDR 64 2292 MIPS KVM_REG_MIPS_CP0_BADINSTR 32 2293 MIPS KVM_REG_MIPS_CP0_BADINSTRP 32 2294 MIPS KVM_REG_MIPS_CP0_COUNT 32 2295 MIPS KVM_REG_MIPS_CP0_ENTRYHI 64 2296 MIPS KVM_REG_MIPS_CP0_COMPARE 32 2297 MIPS KVM_REG_MIPS_CP0_STATUS 32 2298 MIPS KVM_REG_MIPS_CP0_INTCTL 32 2299 MIPS KVM_REG_MIPS_CP0_CAUSE 32 2300 MIPS KVM_REG_MIPS_CP0_EPC 64 2301 MIPS KVM_REG_MIPS_CP0_PRID 32 2302 MIPS KVM_REG_MIPS_CP0_EBASE 64 2303 MIPS KVM_REG_MIPS_CP0_CONFIG 32 2304 MIPS KVM_REG_MIPS_CP0_CONFIG1 32 2305 MIPS KVM_REG_MIPS_CP0_CONFIG2 32 2306 MIPS KVM_REG_MIPS_CP0_CONFIG3 32 2307 MIPS KVM_REG_MIPS_CP0_CONFIG4 32 2308 MIPS KVM_REG_MIPS_CP0_CONFIG5 32 2309 MIPS KVM_REG_MIPS_CP0_CONFIG7 32 2310 MIPS KVM_REG_MIPS_CP0_XCONTEXT 64 2311 MIPS KVM_REG_MIPS_CP0_ERROREPC 64 2312 MIPS KVM_REG_MIPS_CP0_KSCRATCH1 64 2313 MIPS KVM_REG_MIPS_CP0_KSCRATCH2 64 2314 MIPS KVM_REG_MIPS_CP0_KSCRATCH3 64 2315 MIPS KVM_REG_MIPS_CP0_KSCRATCH4 64 2316 MIPS KVM_REG_MIPS_CP0_KSCRATCH5 64 2317 MIPS KVM_REG_MIPS_CP0_KSCRATCH6 64 2318 MIPS KVM_REG_MIPS_CP0_MAAR(0..63) 64 2319 MIPS KVM_REG_MIPS_COUNT_CTL 64 2320 MIPS KVM_REG_MIPS_COUNT_RESUME 64 2321 MIPS KVM_REG_MIPS_COUNT_HZ 64 2322 MIPS KVM_REG_MIPS_FPR_32(0..31) 32 2323 MIPS KVM_REG_MIPS_FPR_64(0..31) 64 2324 MIPS KVM_REG_MIPS_VEC_128(0..31) 128 2325 MIPS KVM_REG_MIPS_FCR_IR 32 2326 MIPS KVM_REG_MIPS_FCR_CSR 32 2327 MIPS KVM_REG_MIPS_MSA_IR 32 2328 MIPS KVM_REG_MIPS_MSA_CSR 32 2329 ======= =============================== ============ 2330 2331ARM registers are mapped using the lower 32 bits. The upper 16 of that 2332is the register group type, or coprocessor number: 2333 2334ARM core registers have the following id bit patterns:: 2335 2336 0x4020 0000 0010 <index into the kvm_regs struct:16> 2337 2338ARM 32-bit CP15 registers have the following id bit patterns:: 2339 2340 0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3> 2341 2342ARM 64-bit CP15 registers have the following id bit patterns:: 2343 2344 0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3> 2345 2346ARM CCSIDR registers are demultiplexed by CSSELR value:: 2347 2348 0x4020 0000 0011 00 <csselr:8> 2349 2350ARM 32-bit VFP control registers have the following id bit patterns:: 2351 2352 0x4020 0000 0012 1 <regno:12> 2353 2354ARM 64-bit FP registers have the following id bit patterns:: 2355 2356 0x4030 0000 0012 0 <regno:12> 2357 2358ARM firmware pseudo-registers have the following bit pattern:: 2359 2360 0x4030 0000 0014 <regno:16> 2361 2362 2363arm64 registers are mapped using the lower 32 bits. The upper 16 of 2364that is the register group type, or coprocessor number: 2365 2366arm64 core/FP-SIMD registers have the following id bit patterns. Note 2367that the size of the access is variable, as the kvm_regs structure 2368contains elements ranging from 32 to 128 bits. The index is a 32bit 2369value in the kvm_regs structure seen as a 32bit array:: 2370 2371 0x60x0 0000 0010 <index into the kvm_regs struct:16> 2372 2373Specifically: 2374 2375======================= ========= ===== ======================================= 2376 Encoding Register Bits kvm_regs member 2377======================= ========= ===== ======================================= 2378 0x6030 0000 0010 0000 X0 64 regs.regs[0] 2379 0x6030 0000 0010 0002 X1 64 regs.regs[1] 2380 ... 2381 0x6030 0000 0010 003c X30 64 regs.regs[30] 2382 0x6030 0000 0010 003e SP 64 regs.sp 2383 0x6030 0000 0010 0040 PC 64 regs.pc 2384 0x6030 0000 0010 0042 PSTATE 64 regs.pstate 2385 0x6030 0000 0010 0044 SP_EL1 64 sp_el1 2386 0x6030 0000 0010 0046 ELR_EL1 64 elr_el1 2387 0x6030 0000 0010 0048 SPSR_EL1 64 spsr[KVM_SPSR_EL1] (alias SPSR_SVC) 2388 0x6030 0000 0010 004a SPSR_ABT 64 spsr[KVM_SPSR_ABT] 2389 0x6030 0000 0010 004c SPSR_UND 64 spsr[KVM_SPSR_UND] 2390 0x6030 0000 0010 004e SPSR_IRQ 64 spsr[KVM_SPSR_IRQ] 2391 0x6060 0000 0010 0050 SPSR_FIQ 64 spsr[KVM_SPSR_FIQ] 2392 0x6040 0000 0010 0054 V0 128 fp_regs.vregs[0] [1]_ 2393 0x6040 0000 0010 0058 V1 128 fp_regs.vregs[1] [1]_ 2394 ... 2395 0x6040 0000 0010 00d0 V31 128 fp_regs.vregs[31] [1]_ 2396 0x6020 0000 0010 00d4 FPSR 32 fp_regs.fpsr 2397 0x6020 0000 0010 00d5 FPCR 32 fp_regs.fpcr 2398======================= ========= ===== ======================================= 2399 2400.. [1] These encodings are not accepted for SVE-enabled vcpus. See 2401 KVM_ARM_VCPU_INIT. 2402 2403 The equivalent register content can be accessed via bits [127:0] of 2404 the corresponding SVE Zn registers instead for vcpus that have SVE 2405 enabled (see below). 2406 2407arm64 CCSIDR registers are demultiplexed by CSSELR value:: 2408 2409 0x6020 0000 0011 00 <csselr:8> 2410 2411arm64 system registers have the following id bit patterns:: 2412 2413 0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3> 2414 2415.. warning:: 2416 2417 Two system register IDs do not follow the specified pattern. These 2418 are KVM_REG_ARM_TIMER_CVAL and KVM_REG_ARM_TIMER_CNT, which map to 2419 system registers CNTV_CVAL_EL0 and CNTVCT_EL0 respectively. These 2420 two had their values accidentally swapped, which means TIMER_CVAL is 2421 derived from the register encoding for CNTVCT_EL0 and TIMER_CNT is 2422 derived from the register encoding for CNTV_CVAL_EL0. As this is 2423 API, it must remain this way. 2424 2425arm64 firmware pseudo-registers have the following bit pattern:: 2426 2427 0x6030 0000 0014 <regno:16> 2428 2429arm64 SVE registers have the following bit patterns:: 2430 2431 0x6080 0000 0015 00 <n:5> <slice:5> Zn bits[2048*slice + 2047 : 2048*slice] 2432 0x6050 0000 0015 04 <n:4> <slice:5> Pn bits[256*slice + 255 : 256*slice] 2433 0x6050 0000 0015 060 <slice:5> FFR bits[256*slice + 255 : 256*slice] 2434 0x6060 0000 0015 ffff KVM_REG_ARM64_SVE_VLS pseudo-register 2435 2436Access to register IDs where 2048 * slice >= 128 * max_vq will fail with 2437ENOENT. max_vq is the vcpu's maximum supported vector length in 128-bit 2438quadwords: see [2]_ below. 2439 2440These registers are only accessible on vcpus for which SVE is enabled. 2441See KVM_ARM_VCPU_INIT for details. 2442 2443In addition, except for KVM_REG_ARM64_SVE_VLS, these registers are not 2444accessible until the vcpu's SVE configuration has been finalized 2445using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). See KVM_ARM_VCPU_INIT 2446and KVM_ARM_VCPU_FINALIZE for more information about this procedure. 2447 2448KVM_REG_ARM64_SVE_VLS is a pseudo-register that allows the set of vector 2449lengths supported by the vcpu to be discovered and configured by 2450userspace. When transferred to or from user memory via KVM_GET_ONE_REG 2451or KVM_SET_ONE_REG, the value of this register is of type 2452__u64[KVM_ARM64_SVE_VLS_WORDS], and encodes the set of vector lengths as 2453follows:: 2454 2455 __u64 vector_lengths[KVM_ARM64_SVE_VLS_WORDS]; 2456 2457 if (vq >= SVE_VQ_MIN && vq <= SVE_VQ_MAX && 2458 ((vector_lengths[(vq - KVM_ARM64_SVE_VQ_MIN) / 64] >> 2459 ((vq - KVM_ARM64_SVE_VQ_MIN) % 64)) & 1)) 2460 /* Vector length vq * 16 bytes supported */ 2461 else 2462 /* Vector length vq * 16 bytes not supported */ 2463 2464.. [2] The maximum value vq for which the above condition is true is 2465 max_vq. This is the maximum vector length available to the guest on 2466 this vcpu, and determines which register slices are visible through 2467 this ioctl interface. 2468 2469(See Documentation/arm64/sve.rst for an explanation of the "vq" 2470nomenclature.) 2471 2472KVM_REG_ARM64_SVE_VLS is only accessible after KVM_ARM_VCPU_INIT. 2473KVM_ARM_VCPU_INIT initialises it to the best set of vector lengths that 2474the host supports. 2475 2476Userspace may subsequently modify it if desired until the vcpu's SVE 2477configuration is finalized using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). 2478 2479Apart from simply removing all vector lengths from the host set that 2480exceed some value, support for arbitrarily chosen sets of vector lengths 2481is hardware-dependent and may not be available. Attempting to configure 2482an invalid set of vector lengths via KVM_SET_ONE_REG will fail with 2483EINVAL. 2484 2485After the vcpu's SVE configuration is finalized, further attempts to 2486write this register will fail with EPERM. 2487 2488 2489MIPS registers are mapped using the lower 32 bits. The upper 16 of that is 2490the register group type: 2491 2492MIPS core registers (see above) have the following id bit patterns:: 2493 2494 0x7030 0000 0000 <reg:16> 2495 2496MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit 2497patterns depending on whether they're 32-bit or 64-bit registers:: 2498 2499 0x7020 0000 0001 00 <reg:5> <sel:3> (32-bit) 2500 0x7030 0000 0001 00 <reg:5> <sel:3> (64-bit) 2501 2502Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_MIPS_CP0_ENTRYLO1 are the MIPS64 2503versions of the EntryLo registers regardless of the word size of the host 2504hardware, host kernel, guest, and whether XPA is present in the guest, i.e. 2505with the RI and XI bits (if they exist) in bits 63 and 62 respectively, and 2506the PFNX field starting at bit 30. 2507 2508MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) above) have the following id bit 2509patterns:: 2510 2511 0x7030 0000 0001 01 <reg:8> 2512 2513MIPS KVM control registers (see above) have the following id bit patterns:: 2514 2515 0x7030 0000 0002 <reg:16> 2516 2517MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following 2518id bit patterns depending on the size of the register being accessed. They are 2519always accessed according to the current guest FPU mode (Status.FR and 2520Config5.FRE), i.e. as the guest would see them, and they become unpredictable 2521if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector 2522registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they 2523overlap the FPU registers:: 2524 2525 0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers) 2526 0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers) 2527 0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers) 2528 2529MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the 2530following id bit patterns:: 2531 2532 0x7020 0000 0003 01 <0:3> <reg:5> 2533 2534MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the 2535following id bit patterns:: 2536 2537 0x7020 0000 0003 02 <0:3> <reg:5> 2538 2539 25404.69 KVM_GET_ONE_REG 2541-------------------- 2542 2543:Capability: KVM_CAP_ONE_REG 2544:Architectures: all 2545:Type: vcpu ioctl 2546:Parameters: struct kvm_one_reg (in and out) 2547:Returns: 0 on success, negative value on failure 2548 2549Errors include: 2550 2551 ======== ============================================================ 2552 ENOENT no such register 2553 EINVAL invalid register ID, or no such register or used with VMs in 2554 protected virtualization mode on s390 2555 EPERM (arm64) register access not allowed before vcpu finalization 2556 ======== ============================================================ 2557 2558(These error codes are indicative only: do not rely on a specific error 2559code being returned in a specific situation.) 2560 2561This ioctl allows to receive the value of a single register implemented 2562in a vcpu. The register to read is indicated by the "id" field of the 2563kvm_one_reg struct passed in. On success, the register value can be found 2564at the memory location pointed to by "addr". 2565 2566The list of registers accessible using this interface is identical to the 2567list in 4.68. 2568 2569 25704.70 KVM_KVMCLOCK_CTRL 2571---------------------- 2572 2573:Capability: KVM_CAP_KVMCLOCK_CTRL 2574:Architectures: Any that implement pvclocks (currently x86 only) 2575:Type: vcpu ioctl 2576:Parameters: None 2577:Returns: 0 on success, -1 on error 2578 2579This ioctl sets a flag accessible to the guest indicating that the specified 2580vCPU has been paused by the host userspace. 2581 2582The host will set a flag in the pvclock structure that is checked from the 2583soft lockup watchdog. The flag is part of the pvclock structure that is 2584shared between guest and host, specifically the second bit of the flags 2585field of the pvclock_vcpu_time_info structure. It will be set exclusively by 2586the host and read/cleared exclusively by the guest. The guest operation of 2587checking and clearing the flag must be an atomic operation so 2588load-link/store-conditional, or equivalent must be used. There are two cases 2589where the guest will clear the flag: when the soft lockup watchdog timer resets 2590itself or when a soft lockup is detected. This ioctl can be called any time 2591after pausing the vcpu, but before it is resumed. 2592 2593 25944.71 KVM_SIGNAL_MSI 2595------------------- 2596 2597:Capability: KVM_CAP_SIGNAL_MSI 2598:Architectures: x86 arm arm64 2599:Type: vm ioctl 2600:Parameters: struct kvm_msi (in) 2601:Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error 2602 2603Directly inject a MSI message. Only valid with in-kernel irqchip that handles 2604MSI messages. 2605 2606:: 2607 2608 struct kvm_msi { 2609 __u32 address_lo; 2610 __u32 address_hi; 2611 __u32 data; 2612 __u32 flags; 2613 __u32 devid; 2614 __u8 pad[12]; 2615 }; 2616 2617flags: 2618 KVM_MSI_VALID_DEVID: devid contains a valid value. The per-VM 2619 KVM_CAP_MSI_DEVID capability advertises the requirement to provide 2620 the device ID. If this capability is not available, userspace 2621 should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail. 2622 2623If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier 2624for the device that wrote the MSI message. For PCI, this is usually a 2625BFD identifier in the lower 16 bits. 2626 2627On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS 2628feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled, 2629address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of 2630address_hi must be zero. 2631 2632 26334.71 KVM_CREATE_PIT2 2634-------------------- 2635 2636:Capability: KVM_CAP_PIT2 2637:Architectures: x86 2638:Type: vm ioctl 2639:Parameters: struct kvm_pit_config (in) 2640:Returns: 0 on success, -1 on error 2641 2642Creates an in-kernel device model for the i8254 PIT. This call is only valid 2643after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following 2644parameters have to be passed:: 2645 2646 struct kvm_pit_config { 2647 __u32 flags; 2648 __u32 pad[15]; 2649 }; 2650 2651Valid flags are:: 2652 2653 #define KVM_PIT_SPEAKER_DUMMY 1 /* emulate speaker port stub */ 2654 2655PIT timer interrupts may use a per-VM kernel thread for injection. If it 2656exists, this thread will have a name of the following pattern:: 2657 2658 kvm-pit/<owner-process-pid> 2659 2660When running a guest with elevated priorities, the scheduling parameters of 2661this thread may have to be adjusted accordingly. 2662 2663This IOCTL replaces the obsolete KVM_CREATE_PIT. 2664 2665 26664.72 KVM_GET_PIT2 2667----------------- 2668 2669:Capability: KVM_CAP_PIT_STATE2 2670:Architectures: x86 2671:Type: vm ioctl 2672:Parameters: struct kvm_pit_state2 (out) 2673:Returns: 0 on success, -1 on error 2674 2675Retrieves the state of the in-kernel PIT model. Only valid after 2676KVM_CREATE_PIT2. The state is returned in the following structure:: 2677 2678 struct kvm_pit_state2 { 2679 struct kvm_pit_channel_state channels[3]; 2680 __u32 flags; 2681 __u32 reserved[9]; 2682 }; 2683 2684Valid flags are:: 2685 2686 /* disable PIT in HPET legacy mode */ 2687 #define KVM_PIT_FLAGS_HPET_LEGACY 0x00000001 2688 2689This IOCTL replaces the obsolete KVM_GET_PIT. 2690 2691 26924.73 KVM_SET_PIT2 2693----------------- 2694 2695:Capability: KVM_CAP_PIT_STATE2 2696:Architectures: x86 2697:Type: vm ioctl 2698:Parameters: struct kvm_pit_state2 (in) 2699:Returns: 0 on success, -1 on error 2700 2701Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2. 2702See KVM_GET_PIT2 for details on struct kvm_pit_state2. 2703 2704This IOCTL replaces the obsolete KVM_SET_PIT. 2705 2706 27074.74 KVM_PPC_GET_SMMU_INFO 2708-------------------------- 2709 2710:Capability: KVM_CAP_PPC_GET_SMMU_INFO 2711:Architectures: powerpc 2712:Type: vm ioctl 2713:Parameters: None 2714:Returns: 0 on success, -1 on error 2715 2716This populates and returns a structure describing the features of 2717the "Server" class MMU emulation supported by KVM. 2718This can in turn be used by userspace to generate the appropriate 2719device-tree properties for the guest operating system. 2720 2721The structure contains some global information, followed by an 2722array of supported segment page sizes:: 2723 2724 struct kvm_ppc_smmu_info { 2725 __u64 flags; 2726 __u32 slb_size; 2727 __u32 pad; 2728 struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ]; 2729 }; 2730 2731The supported flags are: 2732 2733 - KVM_PPC_PAGE_SIZES_REAL: 2734 When that flag is set, guest page sizes must "fit" the backing 2735 store page sizes. When not set, any page size in the list can 2736 be used regardless of how they are backed by userspace. 2737 2738 - KVM_PPC_1T_SEGMENTS 2739 The emulated MMU supports 1T segments in addition to the 2740 standard 256M ones. 2741 2742 - KVM_PPC_NO_HASH 2743 This flag indicates that HPT guests are not supported by KVM, 2744 thus all guests must use radix MMU mode. 2745 2746The "slb_size" field indicates how many SLB entries are supported 2747 2748The "sps" array contains 8 entries indicating the supported base 2749page sizes for a segment in increasing order. Each entry is defined 2750as follow:: 2751 2752 struct kvm_ppc_one_seg_page_size { 2753 __u32 page_shift; /* Base page shift of segment (or 0) */ 2754 __u32 slb_enc; /* SLB encoding for BookS */ 2755 struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ]; 2756 }; 2757 2758An entry with a "page_shift" of 0 is unused. Because the array is 2759organized in increasing order, a lookup can stop when encoutering 2760such an entry. 2761 2762The "slb_enc" field provides the encoding to use in the SLB for the 2763page size. The bits are in positions such as the value can directly 2764be OR'ed into the "vsid" argument of the slbmte instruction. 2765 2766The "enc" array is a list which for each of those segment base page 2767size provides the list of supported actual page sizes (which can be 2768only larger or equal to the base page size), along with the 2769corresponding encoding in the hash PTE. Similarly, the array is 27708 entries sorted by increasing sizes and an entry with a "0" shift 2771is an empty entry and a terminator:: 2772 2773 struct kvm_ppc_one_page_size { 2774 __u32 page_shift; /* Page shift (or 0) */ 2775 __u32 pte_enc; /* Encoding in the HPTE (>>12) */ 2776 }; 2777 2778The "pte_enc" field provides a value that can OR'ed into the hash 2779PTE's RPN field (ie, it needs to be shifted left by 12 to OR it 2780into the hash PTE second double word). 2781 27824.75 KVM_IRQFD 2783-------------- 2784 2785:Capability: KVM_CAP_IRQFD 2786:Architectures: x86 s390 arm arm64 2787:Type: vm ioctl 2788:Parameters: struct kvm_irqfd (in) 2789:Returns: 0 on success, -1 on error 2790 2791Allows setting an eventfd to directly trigger a guest interrupt. 2792kvm_irqfd.fd specifies the file descriptor to use as the eventfd and 2793kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When 2794an event is triggered on the eventfd, an interrupt is injected into 2795the guest using the specified gsi pin. The irqfd is removed using 2796the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd 2797and kvm_irqfd.gsi. 2798 2799With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify 2800mechanism allowing emulation of level-triggered, irqfd-based 2801interrupts. When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an 2802additional eventfd in the kvm_irqfd.resamplefd field. When operating 2803in resample mode, posting of an interrupt through kvm_irq.fd asserts 2804the specified gsi in the irqchip. When the irqchip is resampled, such 2805as from an EOI, the gsi is de-asserted and the user is notified via 2806kvm_irqfd.resamplefd. It is the user's responsibility to re-queue 2807the interrupt if the device making use of it still requires service. 2808Note that closing the resamplefd is not sufficient to disable the 2809irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment 2810and need not be specified with KVM_IRQFD_FLAG_DEASSIGN. 2811 2812On arm/arm64, gsi routing being supported, the following can happen: 2813 2814- in case no routing entry is associated to this gsi, injection fails 2815- in case the gsi is associated to an irqchip routing entry, 2816 irqchip.pin + 32 corresponds to the injected SPI ID. 2817- in case the gsi is associated to an MSI routing entry, the MSI 2818 message and device ID are translated into an LPI (support restricted 2819 to GICv3 ITS in-kernel emulation). 2820 28214.76 KVM_PPC_ALLOCATE_HTAB 2822-------------------------- 2823 2824:Capability: KVM_CAP_PPC_ALLOC_HTAB 2825:Architectures: powerpc 2826:Type: vm ioctl 2827:Parameters: Pointer to u32 containing hash table order (in/out) 2828:Returns: 0 on success, -1 on error 2829 2830This requests the host kernel to allocate an MMU hash table for a 2831guest using the PAPR paravirtualization interface. This only does 2832anything if the kernel is configured to use the Book 3S HV style of 2833virtualization. Otherwise the capability doesn't exist and the ioctl 2834returns an ENOTTY error. The rest of this description assumes Book 3S 2835HV. 2836 2837There must be no vcpus running when this ioctl is called; if there 2838are, it will do nothing and return an EBUSY error. 2839 2840The parameter is a pointer to a 32-bit unsigned integer variable 2841containing the order (log base 2) of the desired size of the hash 2842table, which must be between 18 and 46. On successful return from the 2843ioctl, the value will not be changed by the kernel. 2844 2845If no hash table has been allocated when any vcpu is asked to run 2846(with the KVM_RUN ioctl), the host kernel will allocate a 2847default-sized hash table (16 MB). 2848 2849If this ioctl is called when a hash table has already been allocated, 2850with a different order from the existing hash table, the existing hash 2851table will be freed and a new one allocated. If this is ioctl is 2852called when a hash table has already been allocated of the same order 2853as specified, the kernel will clear out the existing hash table (zero 2854all HPTEs). In either case, if the guest is using the virtualized 2855real-mode area (VRMA) facility, the kernel will re-create the VMRA 2856HPTEs on the next KVM_RUN of any vcpu. 2857 28584.77 KVM_S390_INTERRUPT 2859----------------------- 2860 2861:Capability: basic 2862:Architectures: s390 2863:Type: vm ioctl, vcpu ioctl 2864:Parameters: struct kvm_s390_interrupt (in) 2865:Returns: 0 on success, -1 on error 2866 2867Allows to inject an interrupt to the guest. Interrupts can be floating 2868(vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type. 2869 2870Interrupt parameters are passed via kvm_s390_interrupt:: 2871 2872 struct kvm_s390_interrupt { 2873 __u32 type; 2874 __u32 parm; 2875 __u64 parm64; 2876 }; 2877 2878type can be one of the following: 2879 2880KVM_S390_SIGP_STOP (vcpu) 2881 - sigp stop; optional flags in parm 2882KVM_S390_PROGRAM_INT (vcpu) 2883 - program check; code in parm 2884KVM_S390_SIGP_SET_PREFIX (vcpu) 2885 - sigp set prefix; prefix address in parm 2886KVM_S390_RESTART (vcpu) 2887 - restart 2888KVM_S390_INT_CLOCK_COMP (vcpu) 2889 - clock comparator interrupt 2890KVM_S390_INT_CPU_TIMER (vcpu) 2891 - CPU timer interrupt 2892KVM_S390_INT_VIRTIO (vm) 2893 - virtio external interrupt; external interrupt 2894 parameters in parm and parm64 2895KVM_S390_INT_SERVICE (vm) 2896 - sclp external interrupt; sclp parameter in parm 2897KVM_S390_INT_EMERGENCY (vcpu) 2898 - sigp emergency; source cpu in parm 2899KVM_S390_INT_EXTERNAL_CALL (vcpu) 2900 - sigp external call; source cpu in parm 2901KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm) 2902 - compound value to indicate an 2903 I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel); 2904 I/O interruption parameters in parm (subchannel) and parm64 (intparm, 2905 interruption subclass) 2906KVM_S390_MCHK (vm, vcpu) 2907 - machine check interrupt; cr 14 bits in parm, machine check interrupt 2908 code in parm64 (note that machine checks needing further payload are not 2909 supported by this ioctl) 2910 2911This is an asynchronous vcpu ioctl and can be invoked from any thread. 2912 29134.78 KVM_PPC_GET_HTAB_FD 2914------------------------ 2915 2916:Capability: KVM_CAP_PPC_HTAB_FD 2917:Architectures: powerpc 2918:Type: vm ioctl 2919:Parameters: Pointer to struct kvm_get_htab_fd (in) 2920:Returns: file descriptor number (>= 0) on success, -1 on error 2921 2922This returns a file descriptor that can be used either to read out the 2923entries in the guest's hashed page table (HPT), or to write entries to 2924initialize the HPT. The returned fd can only be written to if the 2925KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and 2926can only be read if that bit is clear. The argument struct looks like 2927this:: 2928 2929 /* For KVM_PPC_GET_HTAB_FD */ 2930 struct kvm_get_htab_fd { 2931 __u64 flags; 2932 __u64 start_index; 2933 __u64 reserved[2]; 2934 }; 2935 2936 /* Values for kvm_get_htab_fd.flags */ 2937 #define KVM_GET_HTAB_BOLTED_ONLY ((__u64)0x1) 2938 #define KVM_GET_HTAB_WRITE ((__u64)0x2) 2939 2940The 'start_index' field gives the index in the HPT of the entry at 2941which to start reading. It is ignored when writing. 2942 2943Reads on the fd will initially supply information about all 2944"interesting" HPT entries. Interesting entries are those with the 2945bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise 2946all entries. When the end of the HPT is reached, the read() will 2947return. If read() is called again on the fd, it will start again from 2948the beginning of the HPT, but will only return HPT entries that have 2949changed since they were last read. 2950 2951Data read or written is structured as a header (8 bytes) followed by a 2952series of valid HPT entries (16 bytes) each. The header indicates how 2953many valid HPT entries there are and how many invalid entries follow 2954the valid entries. The invalid entries are not represented explicitly 2955in the stream. The header format is:: 2956 2957 struct kvm_get_htab_header { 2958 __u32 index; 2959 __u16 n_valid; 2960 __u16 n_invalid; 2961 }; 2962 2963Writes to the fd create HPT entries starting at the index given in the 2964header; first 'n_valid' valid entries with contents from the data 2965written, then 'n_invalid' invalid entries, invalidating any previously 2966valid entries found. 2967 29684.79 KVM_CREATE_DEVICE 2969---------------------- 2970 2971:Capability: KVM_CAP_DEVICE_CTRL 2972:Type: vm ioctl 2973:Parameters: struct kvm_create_device (in/out) 2974:Returns: 0 on success, -1 on error 2975 2976Errors: 2977 2978 ====== ======================================================= 2979 ENODEV The device type is unknown or unsupported 2980 EEXIST Device already created, and this type of device may not 2981 be instantiated multiple times 2982 ====== ======================================================= 2983 2984 Other error conditions may be defined by individual device types or 2985 have their standard meanings. 2986 2987Creates an emulated device in the kernel. The file descriptor returned 2988in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR. 2989 2990If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the 2991device type is supported (not necessarily whether it can be created 2992in the current vm). 2993 2994Individual devices should not define flags. Attributes should be used 2995for specifying any behavior that is not implied by the device type 2996number. 2997 2998:: 2999 3000 struct kvm_create_device { 3001 __u32 type; /* in: KVM_DEV_TYPE_xxx */ 3002 __u32 fd; /* out: device handle */ 3003 __u32 flags; /* in: KVM_CREATE_DEVICE_xxx */ 3004 }; 3005 30064.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR 3007-------------------------------------------- 3008 3009:Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device, 3010 KVM_CAP_VCPU_ATTRIBUTES for vcpu device 3011:Type: device ioctl, vm ioctl, vcpu ioctl 3012:Parameters: struct kvm_device_attr 3013:Returns: 0 on success, -1 on error 3014 3015Errors: 3016 3017 ===== ============================================================= 3018 ENXIO The group or attribute is unknown/unsupported for this device 3019 or hardware support is missing. 3020 EPERM The attribute cannot (currently) be accessed this way 3021 (e.g. read-only attribute, or attribute that only makes 3022 sense when the device is in a different state) 3023 ===== ============================================================= 3024 3025 Other error conditions may be defined by individual device types. 3026 3027Gets/sets a specified piece of device configuration and/or state. The 3028semantics are device-specific. See individual device documentation in 3029the "devices" directory. As with ONE_REG, the size of the data 3030transferred is defined by the particular attribute. 3031 3032:: 3033 3034 struct kvm_device_attr { 3035 __u32 flags; /* no flags currently defined */ 3036 __u32 group; /* device-defined */ 3037 __u64 attr; /* group-defined */ 3038 __u64 addr; /* userspace address of attr data */ 3039 }; 3040 30414.81 KVM_HAS_DEVICE_ATTR 3042------------------------ 3043 3044:Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device, 3045 KVM_CAP_VCPU_ATTRIBUTES for vcpu device 3046:Type: device ioctl, vm ioctl, vcpu ioctl 3047:Parameters: struct kvm_device_attr 3048:Returns: 0 on success, -1 on error 3049 3050Errors: 3051 3052 ===== ============================================================= 3053 ENXIO The group or attribute is unknown/unsupported for this device 3054 or hardware support is missing. 3055 ===== ============================================================= 3056 3057Tests whether a device supports a particular attribute. A successful 3058return indicates the attribute is implemented. It does not necessarily 3059indicate that the attribute can be read or written in the device's 3060current state. "addr" is ignored. 3061 30624.82 KVM_ARM_VCPU_INIT 3063---------------------- 3064 3065:Capability: basic 3066:Architectures: arm, arm64 3067:Type: vcpu ioctl 3068:Parameters: struct kvm_vcpu_init (in) 3069:Returns: 0 on success; -1 on error 3070 3071Errors: 3072 3073 ====== ================================================================= 3074 EINVAL the target is unknown, or the combination of features is invalid. 3075 ENOENT a features bit specified is unknown. 3076 ====== ================================================================= 3077 3078This tells KVM what type of CPU to present to the guest, and what 3079optional features it should have. This will cause a reset of the cpu 3080registers to their initial values. If this is not called, KVM_RUN will 3081return ENOEXEC for that vcpu. 3082 3083Note that because some registers reflect machine topology, all vcpus 3084should be created before this ioctl is invoked. 3085 3086Userspace can call this function multiple times for a given vcpu, including 3087after the vcpu has been run. This will reset the vcpu to its initial 3088state. All calls to this function after the initial call must use the same 3089target and same set of feature flags, otherwise EINVAL will be returned. 3090 3091Possible features: 3092 3093 - KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state. 3094 Depends on KVM_CAP_ARM_PSCI. If not set, the CPU will be powered on 3095 and execute guest code when KVM_RUN is called. 3096 - KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode. 3097 Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only). 3098 - KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 (or a future revision 3099 backward compatible with v0.2) for the CPU. 3100 Depends on KVM_CAP_ARM_PSCI_0_2. 3101 - KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU. 3102 Depends on KVM_CAP_ARM_PMU_V3. 3103 3104 - KVM_ARM_VCPU_PTRAUTH_ADDRESS: Enables Address Pointer authentication 3105 for arm64 only. 3106 Depends on KVM_CAP_ARM_PTRAUTH_ADDRESS. 3107 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are 3108 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and 3109 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be 3110 requested. 3111 3112 - KVM_ARM_VCPU_PTRAUTH_GENERIC: Enables Generic Pointer authentication 3113 for arm64 only. 3114 Depends on KVM_CAP_ARM_PTRAUTH_GENERIC. 3115 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are 3116 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and 3117 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be 3118 requested. 3119 3120 - KVM_ARM_VCPU_SVE: Enables SVE for the CPU (arm64 only). 3121 Depends on KVM_CAP_ARM_SVE. 3122 Requires KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3123 3124 * After KVM_ARM_VCPU_INIT: 3125 3126 - KVM_REG_ARM64_SVE_VLS may be read using KVM_GET_ONE_REG: the 3127 initial value of this pseudo-register indicates the best set of 3128 vector lengths possible for a vcpu on this host. 3129 3130 * Before KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3131 3132 - KVM_RUN and KVM_GET_REG_LIST are not available; 3133 3134 - KVM_GET_ONE_REG and KVM_SET_ONE_REG cannot be used to access 3135 the scalable archietctural SVE registers 3136 KVM_REG_ARM64_SVE_ZREG(), KVM_REG_ARM64_SVE_PREG() or 3137 KVM_REG_ARM64_SVE_FFR; 3138 3139 - KVM_REG_ARM64_SVE_VLS may optionally be written using 3140 KVM_SET_ONE_REG, to modify the set of vector lengths available 3141 for the vcpu. 3142 3143 * After KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3144 3145 - the KVM_REG_ARM64_SVE_VLS pseudo-register is immutable, and can 3146 no longer be written using KVM_SET_ONE_REG. 3147 31484.83 KVM_ARM_PREFERRED_TARGET 3149----------------------------- 3150 3151:Capability: basic 3152:Architectures: arm, arm64 3153:Type: vm ioctl 3154:Parameters: struct kvm_vcpu_init (out) 3155:Returns: 0 on success; -1 on error 3156 3157Errors: 3158 3159 ====== ========================================== 3160 ENODEV no preferred target available for the host 3161 ====== ========================================== 3162 3163This queries KVM for preferred CPU target type which can be emulated 3164by KVM on underlying host. 3165 3166The ioctl returns struct kvm_vcpu_init instance containing information 3167about preferred CPU target type and recommended features for it. The 3168kvm_vcpu_init->features bitmap returned will have feature bits set if 3169the preferred target recommends setting these features, but this is 3170not mandatory. 3171 3172The information returned by this ioctl can be used to prepare an instance 3173of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in 3174VCPU matching underlying host. 3175 3176 31774.84 KVM_GET_REG_LIST 3178--------------------- 3179 3180:Capability: basic 3181:Architectures: arm, arm64, mips 3182:Type: vcpu ioctl 3183:Parameters: struct kvm_reg_list (in/out) 3184:Returns: 0 on success; -1 on error 3185 3186Errors: 3187 3188 ===== ============================================================== 3189 E2BIG the reg index list is too big to fit in the array specified by 3190 the user (the number required will be written into n). 3191 ===== ============================================================== 3192 3193:: 3194 3195 struct kvm_reg_list { 3196 __u64 n; /* number of registers in reg[] */ 3197 __u64 reg[0]; 3198 }; 3199 3200This ioctl returns the guest registers that are supported for the 3201KVM_GET_ONE_REG/KVM_SET_ONE_REG calls. 3202 3203 32044.85 KVM_ARM_SET_DEVICE_ADDR (deprecated) 3205----------------------------------------- 3206 3207:Capability: KVM_CAP_ARM_SET_DEVICE_ADDR 3208:Architectures: arm, arm64 3209:Type: vm ioctl 3210:Parameters: struct kvm_arm_device_address (in) 3211:Returns: 0 on success, -1 on error 3212 3213Errors: 3214 3215 ====== ============================================ 3216 ENODEV The device id is unknown 3217 ENXIO Device not supported on current system 3218 EEXIST Address already set 3219 E2BIG Address outside guest physical address space 3220 EBUSY Address overlaps with other device range 3221 ====== ============================================ 3222 3223:: 3224 3225 struct kvm_arm_device_addr { 3226 __u64 id; 3227 __u64 addr; 3228 }; 3229 3230Specify a device address in the guest's physical address space where guests 3231can access emulated or directly exposed devices, which the host kernel needs 3232to know about. The id field is an architecture specific identifier for a 3233specific device. 3234 3235ARM/arm64 divides the id field into two parts, a device id and an 3236address type id specific to the individual device:: 3237 3238 bits: | 63 ... 32 | 31 ... 16 | 15 ... 0 | 3239 field: | 0x00000000 | device id | addr type id | 3240 3241ARM/arm64 currently only require this when using the in-kernel GIC 3242support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2 3243as the device id. When setting the base address for the guest's 3244mapping of the VGIC virtual CPU and distributor interface, the ioctl 3245must be called after calling KVM_CREATE_IRQCHIP, but before calling 3246KVM_RUN on any of the VCPUs. Calling this ioctl twice for any of the 3247base addresses will return -EEXIST. 3248 3249Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API 3250should be used instead. 3251 3252 32534.86 KVM_PPC_RTAS_DEFINE_TOKEN 3254------------------------------ 3255 3256:Capability: KVM_CAP_PPC_RTAS 3257:Architectures: ppc 3258:Type: vm ioctl 3259:Parameters: struct kvm_rtas_token_args 3260:Returns: 0 on success, -1 on error 3261 3262Defines a token value for a RTAS (Run Time Abstraction Services) 3263service in order to allow it to be handled in the kernel. The 3264argument struct gives the name of the service, which must be the name 3265of a service that has a kernel-side implementation. If the token 3266value is non-zero, it will be associated with that service, and 3267subsequent RTAS calls by the guest specifying that token will be 3268handled by the kernel. If the token value is 0, then any token 3269associated with the service will be forgotten, and subsequent RTAS 3270calls by the guest for that service will be passed to userspace to be 3271handled. 3272 32734.87 KVM_SET_GUEST_DEBUG 3274------------------------ 3275 3276:Capability: KVM_CAP_SET_GUEST_DEBUG 3277:Architectures: x86, s390, ppc, arm64 3278:Type: vcpu ioctl 3279:Parameters: struct kvm_guest_debug (in) 3280:Returns: 0 on success; -1 on error 3281 3282:: 3283 3284 struct kvm_guest_debug { 3285 __u32 control; 3286 __u32 pad; 3287 struct kvm_guest_debug_arch arch; 3288 }; 3289 3290Set up the processor specific debug registers and configure vcpu for 3291handling guest debug events. There are two parts to the structure, the 3292first a control bitfield indicates the type of debug events to handle 3293when running. Common control bits are: 3294 3295 - KVM_GUESTDBG_ENABLE: guest debugging is enabled 3296 - KVM_GUESTDBG_SINGLESTEP: the next run should single-step 3297 3298The top 16 bits of the control field are architecture specific control 3299flags which can include the following: 3300 3301 - KVM_GUESTDBG_USE_SW_BP: using software breakpoints [x86, arm64] 3302 - KVM_GUESTDBG_USE_HW_BP: using hardware breakpoints [x86, s390, arm64] 3303 - KVM_GUESTDBG_INJECT_DB: inject DB type exception [x86] 3304 - KVM_GUESTDBG_INJECT_BP: inject BP type exception [x86] 3305 - KVM_GUESTDBG_EXIT_PENDING: trigger an immediate guest exit [s390] 3306 3307For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints 3308are enabled in memory so we need to ensure breakpoint exceptions are 3309correctly trapped and the KVM run loop exits at the breakpoint and not 3310running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP 3311we need to ensure the guest vCPUs architecture specific registers are 3312updated to the correct (supplied) values. 3313 3314The second part of the structure is architecture specific and 3315typically contains a set of debug registers. 3316 3317For arm64 the number of debug registers is implementation defined and 3318can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and 3319KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number 3320indicating the number of supported registers. 3321 3322For ppc, the KVM_CAP_PPC_GUEST_DEBUG_SSTEP capability indicates whether 3323the single-step debug event (KVM_GUESTDBG_SINGLESTEP) is supported. 3324 3325When debug events exit the main run loop with the reason 3326KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run 3327structure containing architecture specific debug information. 3328 33294.88 KVM_GET_EMULATED_CPUID 3330--------------------------- 3331 3332:Capability: KVM_CAP_EXT_EMUL_CPUID 3333:Architectures: x86 3334:Type: system ioctl 3335:Parameters: struct kvm_cpuid2 (in/out) 3336:Returns: 0 on success, -1 on error 3337 3338:: 3339 3340 struct kvm_cpuid2 { 3341 __u32 nent; 3342 __u32 flags; 3343 struct kvm_cpuid_entry2 entries[0]; 3344 }; 3345 3346The member 'flags' is used for passing flags from userspace. 3347 3348:: 3349 3350 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0) 3351 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */ 3352 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */ 3353 3354 struct kvm_cpuid_entry2 { 3355 __u32 function; 3356 __u32 index; 3357 __u32 flags; 3358 __u32 eax; 3359 __u32 ebx; 3360 __u32 ecx; 3361 __u32 edx; 3362 __u32 padding[3]; 3363 }; 3364 3365This ioctl returns x86 cpuid features which are emulated by 3366kvm.Userspace can use the information returned by this ioctl to query 3367which features are emulated by kvm instead of being present natively. 3368 3369Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2 3370structure with the 'nent' field indicating the number of entries in 3371the variable-size array 'entries'. If the number of entries is too low 3372to describe the cpu capabilities, an error (E2BIG) is returned. If the 3373number is too high, the 'nent' field is adjusted and an error (ENOMEM) 3374is returned. If the number is just right, the 'nent' field is adjusted 3375to the number of valid entries in the 'entries' array, which is then 3376filled. 3377 3378The entries returned are the set CPUID bits of the respective features 3379which kvm emulates, as returned by the CPUID instruction, with unknown 3380or unsupported feature bits cleared. 3381 3382Features like x2apic, for example, may not be present in the host cpu 3383but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be 3384emulated efficiently and thus not included here. 3385 3386The fields in each entry are defined as follows: 3387 3388 function: 3389 the eax value used to obtain the entry 3390 index: 3391 the ecx value used to obtain the entry (for entries that are 3392 affected by ecx) 3393 flags: 3394 an OR of zero or more of the following: 3395 3396 KVM_CPUID_FLAG_SIGNIFCANT_INDEX: 3397 if the index field is valid 3398 3399 eax, ebx, ecx, edx: 3400 3401 the values returned by the cpuid instruction for 3402 this function/index combination 3403 34044.89 KVM_S390_MEM_OP 3405-------------------- 3406 3407:Capability: KVM_CAP_S390_MEM_OP 3408:Architectures: s390 3409:Type: vcpu ioctl 3410:Parameters: struct kvm_s390_mem_op (in) 3411:Returns: = 0 on success, 3412 < 0 on generic error (e.g. -EFAULT or -ENOMEM), 3413 > 0 if an exception occurred while walking the page tables 3414 3415Read or write data from/to the logical (virtual) memory of a VCPU. 3416 3417Parameters are specified via the following structure:: 3418 3419 struct kvm_s390_mem_op { 3420 __u64 gaddr; /* the guest address */ 3421 __u64 flags; /* flags */ 3422 __u32 size; /* amount of bytes */ 3423 __u32 op; /* type of operation */ 3424 __u64 buf; /* buffer in userspace */ 3425 __u8 ar; /* the access register number */ 3426 __u8 reserved[31]; /* should be set to 0 */ 3427 }; 3428 3429The type of operation is specified in the "op" field. It is either 3430KVM_S390_MEMOP_LOGICAL_READ for reading from logical memory space or 3431KVM_S390_MEMOP_LOGICAL_WRITE for writing to logical memory space. The 3432KVM_S390_MEMOP_F_CHECK_ONLY flag can be set in the "flags" field to check 3433whether the corresponding memory access would create an access exception 3434(without touching the data in the memory at the destination). In case an 3435access exception occurred while walking the MMU tables of the guest, the 3436ioctl returns a positive error number to indicate the type of exception. 3437This exception is also raised directly at the corresponding VCPU if the 3438flag KVM_S390_MEMOP_F_INJECT_EXCEPTION is set in the "flags" field. 3439 3440The start address of the memory region has to be specified in the "gaddr" 3441field, and the length of the region in the "size" field (which must not 3442be 0). The maximum value for "size" can be obtained by checking the 3443KVM_CAP_S390_MEM_OP capability. "buf" is the buffer supplied by the 3444userspace application where the read data should be written to for 3445KVM_S390_MEMOP_LOGICAL_READ, or where the data that should be written is 3446stored for a KVM_S390_MEMOP_LOGICAL_WRITE. When KVM_S390_MEMOP_F_CHECK_ONLY 3447is specified, "buf" is unused and can be NULL. "ar" designates the access 3448register number to be used; the valid range is 0..15. 3449 3450The "reserved" field is meant for future extensions. It is not used by 3451KVM with the currently defined set of flags. 3452 34534.90 KVM_S390_GET_SKEYS 3454----------------------- 3455 3456:Capability: KVM_CAP_S390_SKEYS 3457:Architectures: s390 3458:Type: vm ioctl 3459:Parameters: struct kvm_s390_skeys 3460:Returns: 0 on success, KVM_S390_GET_KEYS_NONE if guest is not using storage 3461 keys, negative value on error 3462 3463This ioctl is used to get guest storage key values on the s390 3464architecture. The ioctl takes parameters via the kvm_s390_skeys struct:: 3465 3466 struct kvm_s390_skeys { 3467 __u64 start_gfn; 3468 __u64 count; 3469 __u64 skeydata_addr; 3470 __u32 flags; 3471 __u32 reserved[9]; 3472 }; 3473 3474The start_gfn field is the number of the first guest frame whose storage keys 3475you want to get. 3476 3477The count field is the number of consecutive frames (starting from start_gfn) 3478whose storage keys to get. The count field must be at least 1 and the maximum 3479allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range 3480will cause the ioctl to return -EINVAL. 3481 3482The skeydata_addr field is the address to a buffer large enough to hold count 3483bytes. This buffer will be filled with storage key data by the ioctl. 3484 34854.91 KVM_S390_SET_SKEYS 3486----------------------- 3487 3488:Capability: KVM_CAP_S390_SKEYS 3489:Architectures: s390 3490:Type: vm ioctl 3491:Parameters: struct kvm_s390_skeys 3492:Returns: 0 on success, negative value on error 3493 3494This ioctl is used to set guest storage key values on the s390 3495architecture. The ioctl takes parameters via the kvm_s390_skeys struct. 3496See section on KVM_S390_GET_SKEYS for struct definition. 3497 3498The start_gfn field is the number of the first guest frame whose storage keys 3499you want to set. 3500 3501The count field is the number of consecutive frames (starting from start_gfn) 3502whose storage keys to get. The count field must be at least 1 and the maximum 3503allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range 3504will cause the ioctl to return -EINVAL. 3505 3506The skeydata_addr field is the address to a buffer containing count bytes of 3507storage keys. Each byte in the buffer will be set as the storage key for a 3508single frame starting at start_gfn for count frames. 3509 3510Note: If any architecturally invalid key value is found in the given data then 3511the ioctl will return -EINVAL. 3512 35134.92 KVM_S390_IRQ 3514----------------- 3515 3516:Capability: KVM_CAP_S390_INJECT_IRQ 3517:Architectures: s390 3518:Type: vcpu ioctl 3519:Parameters: struct kvm_s390_irq (in) 3520:Returns: 0 on success, -1 on error 3521 3522Errors: 3523 3524 3525 ====== ================================================================= 3526 EINVAL interrupt type is invalid 3527 type is KVM_S390_SIGP_STOP and flag parameter is invalid value, 3528 type is KVM_S390_INT_EXTERNAL_CALL and code is bigger 3529 than the maximum of VCPUs 3530 EBUSY type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped, 3531 type is KVM_S390_SIGP_STOP and a stop irq is already pending, 3532 type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt 3533 is already pending 3534 ====== ================================================================= 3535 3536Allows to inject an interrupt to the guest. 3537 3538Using struct kvm_s390_irq as a parameter allows 3539to inject additional payload which is not 3540possible via KVM_S390_INTERRUPT. 3541 3542Interrupt parameters are passed via kvm_s390_irq:: 3543 3544 struct kvm_s390_irq { 3545 __u64 type; 3546 union { 3547 struct kvm_s390_io_info io; 3548 struct kvm_s390_ext_info ext; 3549 struct kvm_s390_pgm_info pgm; 3550 struct kvm_s390_emerg_info emerg; 3551 struct kvm_s390_extcall_info extcall; 3552 struct kvm_s390_prefix_info prefix; 3553 struct kvm_s390_stop_info stop; 3554 struct kvm_s390_mchk_info mchk; 3555 char reserved[64]; 3556 } u; 3557 }; 3558 3559type can be one of the following: 3560 3561- KVM_S390_SIGP_STOP - sigp stop; parameter in .stop 3562- KVM_S390_PROGRAM_INT - program check; parameters in .pgm 3563- KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix 3564- KVM_S390_RESTART - restart; no parameters 3565- KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters 3566- KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters 3567- KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg 3568- KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall 3569- KVM_S390_MCHK - machine check interrupt; parameters in .mchk 3570 3571This is an asynchronous vcpu ioctl and can be invoked from any thread. 3572 35734.94 KVM_S390_GET_IRQ_STATE 3574--------------------------- 3575 3576:Capability: KVM_CAP_S390_IRQ_STATE 3577:Architectures: s390 3578:Type: vcpu ioctl 3579:Parameters: struct kvm_s390_irq_state (out) 3580:Returns: >= number of bytes copied into buffer, 3581 -EINVAL if buffer size is 0, 3582 -ENOBUFS if buffer size is too small to fit all pending interrupts, 3583 -EFAULT if the buffer address was invalid 3584 3585This ioctl allows userspace to retrieve the complete state of all currently 3586pending interrupts in a single buffer. Use cases include migration 3587and introspection. The parameter structure contains the address of a 3588userspace buffer and its length:: 3589 3590 struct kvm_s390_irq_state { 3591 __u64 buf; 3592 __u32 flags; /* will stay unused for compatibility reasons */ 3593 __u32 len; 3594 __u32 reserved[4]; /* will stay unused for compatibility reasons */ 3595 }; 3596 3597Userspace passes in the above struct and for each pending interrupt a 3598struct kvm_s390_irq is copied to the provided buffer. 3599 3600The structure contains a flags and a reserved field for future extensions. As 3601the kernel never checked for flags == 0 and QEMU never pre-zeroed flags and 3602reserved, these fields can not be used in the future without breaking 3603compatibility. 3604 3605If -ENOBUFS is returned the buffer provided was too small and userspace 3606may retry with a bigger buffer. 3607 36084.95 KVM_S390_SET_IRQ_STATE 3609--------------------------- 3610 3611:Capability: KVM_CAP_S390_IRQ_STATE 3612:Architectures: s390 3613:Type: vcpu ioctl 3614:Parameters: struct kvm_s390_irq_state (in) 3615:Returns: 0 on success, 3616 -EFAULT if the buffer address was invalid, 3617 -EINVAL for an invalid buffer length (see below), 3618 -EBUSY if there were already interrupts pending, 3619 errors occurring when actually injecting the 3620 interrupt. See KVM_S390_IRQ. 3621 3622This ioctl allows userspace to set the complete state of all cpu-local 3623interrupts currently pending for the vcpu. It is intended for restoring 3624interrupt state after a migration. The input parameter is a userspace buffer 3625containing a struct kvm_s390_irq_state:: 3626 3627 struct kvm_s390_irq_state { 3628 __u64 buf; 3629 __u32 flags; /* will stay unused for compatibility reasons */ 3630 __u32 len; 3631 __u32 reserved[4]; /* will stay unused for compatibility reasons */ 3632 }; 3633 3634The restrictions for flags and reserved apply as well. 3635(see KVM_S390_GET_IRQ_STATE) 3636 3637The userspace memory referenced by buf contains a struct kvm_s390_irq 3638for each interrupt to be injected into the guest. 3639If one of the interrupts could not be injected for some reason the 3640ioctl aborts. 3641 3642len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0 3643and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq), 3644which is the maximum number of possibly pending cpu-local interrupts. 3645 36464.96 KVM_SMI 3647------------ 3648 3649:Capability: KVM_CAP_X86_SMM 3650:Architectures: x86 3651:Type: vcpu ioctl 3652:Parameters: none 3653:Returns: 0 on success, -1 on error 3654 3655Queues an SMI on the thread's vcpu. 3656 36574.97 KVM_CAP_PPC_MULTITCE 3658------------------------- 3659 3660:Capability: KVM_CAP_PPC_MULTITCE 3661:Architectures: ppc 3662:Type: vm 3663 3664This capability means the kernel is capable of handling hypercalls 3665H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user 3666space. This significantly accelerates DMA operations for PPC KVM guests. 3667User space should expect that its handlers for these hypercalls 3668are not going to be called if user space previously registered LIOBN 3669in KVM (via KVM_CREATE_SPAPR_TCE or similar calls). 3670 3671In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest, 3672user space might have to advertise it for the guest. For example, 3673IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is 3674present in the "ibm,hypertas-functions" device-tree property. 3675 3676The hypercalls mentioned above may or may not be processed successfully 3677in the kernel based fast path. If they can not be handled by the kernel, 3678they will get passed on to user space. So user space still has to have 3679an implementation for these despite the in kernel acceleration. 3680 3681This capability is always enabled. 3682 36834.98 KVM_CREATE_SPAPR_TCE_64 3684---------------------------- 3685 3686:Capability: KVM_CAP_SPAPR_TCE_64 3687:Architectures: powerpc 3688:Type: vm ioctl 3689:Parameters: struct kvm_create_spapr_tce_64 (in) 3690:Returns: file descriptor for manipulating the created TCE table 3691 3692This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit 3693windows, described in 4.62 KVM_CREATE_SPAPR_TCE 3694 3695This capability uses extended struct in ioctl interface:: 3696 3697 /* for KVM_CAP_SPAPR_TCE_64 */ 3698 struct kvm_create_spapr_tce_64 { 3699 __u64 liobn; 3700 __u32 page_shift; 3701 __u32 flags; 3702 __u64 offset; /* in pages */ 3703 __u64 size; /* in pages */ 3704 }; 3705 3706The aim of extension is to support an additional bigger DMA window with 3707a variable page size. 3708KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and 3709a bus offset of the corresponding DMA window, @size and @offset are numbers 3710of IOMMU pages. 3711 3712@flags are not used at the moment. 3713 3714The rest of functionality is identical to KVM_CREATE_SPAPR_TCE. 3715 37164.99 KVM_REINJECT_CONTROL 3717------------------------- 3718 3719:Capability: KVM_CAP_REINJECT_CONTROL 3720:Architectures: x86 3721:Type: vm ioctl 3722:Parameters: struct kvm_reinject_control (in) 3723:Returns: 0 on success, 3724 -EFAULT if struct kvm_reinject_control cannot be read, 3725 -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier. 3726 3727i8254 (PIT) has two modes, reinject and !reinject. The default is reinject, 3728where KVM queues elapsed i8254 ticks and monitors completion of interrupt from 3729vector(s) that i8254 injects. Reinject mode dequeues a tick and injects its 3730interrupt whenever there isn't a pending interrupt from i8254. 3731!reinject mode injects an interrupt as soon as a tick arrives. 3732 3733:: 3734 3735 struct kvm_reinject_control { 3736 __u8 pit_reinject; 3737 __u8 reserved[31]; 3738 }; 3739 3740pit_reinject = 0 (!reinject mode) is recommended, unless running an old 3741operating system that uses the PIT for timing (e.g. Linux 2.4.x). 3742 37434.100 KVM_PPC_CONFIGURE_V3_MMU 3744------------------------------ 3745 3746:Capability: KVM_CAP_PPC_RADIX_MMU or KVM_CAP_PPC_HASH_MMU_V3 3747:Architectures: ppc 3748:Type: vm ioctl 3749:Parameters: struct kvm_ppc_mmuv3_cfg (in) 3750:Returns: 0 on success, 3751 -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read, 3752 -EINVAL if the configuration is invalid 3753 3754This ioctl controls whether the guest will use radix or HPT (hashed 3755page table) translation, and sets the pointer to the process table for 3756the guest. 3757 3758:: 3759 3760 struct kvm_ppc_mmuv3_cfg { 3761 __u64 flags; 3762 __u64 process_table; 3763 }; 3764 3765There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and 3766KVM_PPC_MMUV3_GTSE. KVM_PPC_MMUV3_RADIX, if set, configures the guest 3767to use radix tree translation, and if clear, to use HPT translation. 3768KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest 3769to be able to use the global TLB and SLB invalidation instructions; 3770if clear, the guest may not use these instructions. 3771 3772The process_table field specifies the address and size of the guest 3773process table, which is in the guest's space. This field is formatted 3774as the second doubleword of the partition table entry, as defined in 3775the Power ISA V3.00, Book III section 5.7.6.1. 3776 37774.101 KVM_PPC_GET_RMMU_INFO 3778--------------------------- 3779 3780:Capability: KVM_CAP_PPC_RADIX_MMU 3781:Architectures: ppc 3782:Type: vm ioctl 3783:Parameters: struct kvm_ppc_rmmu_info (out) 3784:Returns: 0 on success, 3785 -EFAULT if struct kvm_ppc_rmmu_info cannot be written, 3786 -EINVAL if no useful information can be returned 3787 3788This ioctl returns a structure containing two things: (a) a list 3789containing supported radix tree geometries, and (b) a list that maps 3790page sizes to put in the "AP" (actual page size) field for the tlbie 3791(TLB invalidate entry) instruction. 3792 3793:: 3794 3795 struct kvm_ppc_rmmu_info { 3796 struct kvm_ppc_radix_geom { 3797 __u8 page_shift; 3798 __u8 level_bits[4]; 3799 __u8 pad[3]; 3800 } geometries[8]; 3801 __u32 ap_encodings[8]; 3802 }; 3803 3804The geometries[] field gives up to 8 supported geometries for the 3805radix page table, in terms of the log base 2 of the smallest page 3806size, and the number of bits indexed at each level of the tree, from 3807the PTE level up to the PGD level in that order. Any unused entries 3808will have 0 in the page_shift field. 3809 3810The ap_encodings gives the supported page sizes and their AP field 3811encodings, encoded with the AP value in the top 3 bits and the log 3812base 2 of the page size in the bottom 6 bits. 3813 38144.102 KVM_PPC_RESIZE_HPT_PREPARE 3815-------------------------------- 3816 3817:Capability: KVM_CAP_SPAPR_RESIZE_HPT 3818:Architectures: powerpc 3819:Type: vm ioctl 3820:Parameters: struct kvm_ppc_resize_hpt (in) 3821:Returns: 0 on successful completion, 3822 >0 if a new HPT is being prepared, the value is an estimated 3823 number of milliseconds until preparation is complete, 3824 -EFAULT if struct kvm_reinject_control cannot be read, 3825 -EINVAL if the supplied shift or flags are invalid, 3826 -ENOMEM if unable to allocate the new HPT, 3827 -ENOSPC if there was a hash collision 3828 3829:: 3830 3831 struct kvm_ppc_rmmu_info { 3832 struct kvm_ppc_radix_geom { 3833 __u8 page_shift; 3834 __u8 level_bits[4]; 3835 __u8 pad[3]; 3836 } geometries[8]; 3837 __u32 ap_encodings[8]; 3838 }; 3839 3840The geometries[] field gives up to 8 supported geometries for the 3841radix page table, in terms of the log base 2 of the smallest page 3842size, and the number of bits indexed at each level of the tree, from 3843the PTE level up to the PGD level in that order. Any unused entries 3844will have 0 in the page_shift field. 3845 3846The ap_encodings gives the supported page sizes and their AP field 3847encodings, encoded with the AP value in the top 3 bits and the log 3848base 2 of the page size in the bottom 6 bits. 3849 38504.102 KVM_PPC_RESIZE_HPT_PREPARE 3851-------------------------------- 3852 3853:Capability: KVM_CAP_SPAPR_RESIZE_HPT 3854:Architectures: powerpc 3855:Type: vm ioctl 3856:Parameters: struct kvm_ppc_resize_hpt (in) 3857:Returns: 0 on successful completion, 3858 >0 if a new HPT is being prepared, the value is an estimated 3859 number of milliseconds until preparation is complete, 3860 -EFAULT if struct kvm_reinject_control cannot be read, 3861 -EINVAL if the supplied shift or flags are invalid,when moving existing 3862 HPT entries to the new HPT, 3863 -EIO on other error conditions 3864 3865Used to implement the PAPR extension for runtime resizing of a guest's 3866Hashed Page Table (HPT). Specifically this starts, stops or monitors 3867the preparation of a new potential HPT for the guest, essentially 3868implementing the H_RESIZE_HPT_PREPARE hypercall. 3869 3870If called with shift > 0 when there is no pending HPT for the guest, 3871this begins preparation of a new pending HPT of size 2^(shift) bytes. 3872It then returns a positive integer with the estimated number of 3873milliseconds until preparation is complete. 3874 3875If called when there is a pending HPT whose size does not match that 3876requested in the parameters, discards the existing pending HPT and 3877creates a new one as above. 3878 3879If called when there is a pending HPT of the size requested, will: 3880 3881 * If preparation of the pending HPT is already complete, return 0 3882 * If preparation of the pending HPT has failed, return an error 3883 code, then discard the pending HPT. 3884 * If preparation of the pending HPT is still in progress, return an 3885 estimated number of milliseconds until preparation is complete. 3886 3887If called with shift == 0, discards any currently pending HPT and 3888returns 0 (i.e. cancels any in-progress preparation). 3889 3890flags is reserved for future expansion, currently setting any bits in 3891flags will result in an -EINVAL. 3892 3893Normally this will be called repeatedly with the same parameters until 3894it returns <= 0. The first call will initiate preparation, subsequent 3895ones will monitor preparation until it completes or fails. 3896 3897:: 3898 3899 struct kvm_ppc_resize_hpt { 3900 __u64 flags; 3901 __u32 shift; 3902 __u32 pad; 3903 }; 3904 39054.103 KVM_PPC_RESIZE_HPT_COMMIT 3906------------------------------- 3907 3908:Capability: KVM_CAP_SPAPR_RESIZE_HPT 3909:Architectures: powerpc 3910:Type: vm ioctl 3911:Parameters: struct kvm_ppc_resize_hpt (in) 3912:Returns: 0 on successful completion, 3913 -EFAULT if struct kvm_reinject_control cannot be read, 3914 -EINVAL if the supplied shift or flags are invalid, 3915 -ENXIO is there is no pending HPT, or the pending HPT doesn't 3916 have the requested size, 3917 -EBUSY if the pending HPT is not fully prepared, 3918 -ENOSPC if there was a hash collision when moving existing 3919 HPT entries to the new HPT, 3920 -EIO on other error conditions 3921 3922Used to implement the PAPR extension for runtime resizing of a guest's 3923Hashed Page Table (HPT). Specifically this requests that the guest be 3924transferred to working with the new HPT, essentially implementing the 3925H_RESIZE_HPT_COMMIT hypercall. 3926 3927This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has 3928returned 0 with the same parameters. In other cases 3929KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or 3930-EBUSY, though others may be possible if the preparation was started, 3931but failed). 3932 3933This will have undefined effects on the guest if it has not already 3934placed itself in a quiescent state where no vcpu will make MMU enabled 3935memory accesses. 3936 3937On succsful completion, the pending HPT will become the guest's active 3938HPT and the previous HPT will be discarded. 3939 3940On failure, the guest will still be operating on its previous HPT. 3941 3942:: 3943 3944 struct kvm_ppc_resize_hpt { 3945 __u64 flags; 3946 __u32 shift; 3947 __u32 pad; 3948 }; 3949 39504.104 KVM_X86_GET_MCE_CAP_SUPPORTED 3951----------------------------------- 3952 3953:Capability: KVM_CAP_MCE 3954:Architectures: x86 3955:Type: system ioctl 3956:Parameters: u64 mce_cap (out) 3957:Returns: 0 on success, -1 on error 3958 3959Returns supported MCE capabilities. The u64 mce_cap parameter 3960has the same format as the MSR_IA32_MCG_CAP register. Supported 3961capabilities will have the corresponding bits set. 3962 39634.105 KVM_X86_SETUP_MCE 3964----------------------- 3965 3966:Capability: KVM_CAP_MCE 3967:Architectures: x86 3968:Type: vcpu ioctl 3969:Parameters: u64 mcg_cap (in) 3970:Returns: 0 on success, 3971 -EFAULT if u64 mcg_cap cannot be read, 3972 -EINVAL if the requested number of banks is invalid, 3973 -EINVAL if requested MCE capability is not supported. 3974 3975Initializes MCE support for use. The u64 mcg_cap parameter 3976has the same format as the MSR_IA32_MCG_CAP register and 3977specifies which capabilities should be enabled. The maximum 3978supported number of error-reporting banks can be retrieved when 3979checking for KVM_CAP_MCE. The supported capabilities can be 3980retrieved with KVM_X86_GET_MCE_CAP_SUPPORTED. 3981 39824.106 KVM_X86_SET_MCE 3983--------------------- 3984 3985:Capability: KVM_CAP_MCE 3986:Architectures: x86 3987:Type: vcpu ioctl 3988:Parameters: struct kvm_x86_mce (in) 3989:Returns: 0 on success, 3990 -EFAULT if struct kvm_x86_mce cannot be read, 3991 -EINVAL if the bank number is invalid, 3992 -EINVAL if VAL bit is not set in status field. 3993 3994Inject a machine check error (MCE) into the guest. The input 3995parameter is:: 3996 3997 struct kvm_x86_mce { 3998 __u64 status; 3999 __u64 addr; 4000 __u64 misc; 4001 __u64 mcg_status; 4002 __u8 bank; 4003 __u8 pad1[7]; 4004 __u64 pad2[3]; 4005 }; 4006 4007If the MCE being reported is an uncorrected error, KVM will 4008inject it as an MCE exception into the guest. If the guest 4009MCG_STATUS register reports that an MCE is in progress, KVM 4010causes an KVM_EXIT_SHUTDOWN vmexit. 4011 4012Otherwise, if the MCE is a corrected error, KVM will just 4013store it in the corresponding bank (provided this bank is 4014not holding a previously reported uncorrected error). 4015 40164.107 KVM_S390_GET_CMMA_BITS 4017---------------------------- 4018 4019:Capability: KVM_CAP_S390_CMMA_MIGRATION 4020:Architectures: s390 4021:Type: vm ioctl 4022:Parameters: struct kvm_s390_cmma_log (in, out) 4023:Returns: 0 on success, a negative value on error 4024 4025This ioctl is used to get the values of the CMMA bits on the s390 4026architecture. It is meant to be used in two scenarios: 4027 4028- During live migration to save the CMMA values. Live migration needs 4029 to be enabled via the KVM_REQ_START_MIGRATION VM property. 4030- To non-destructively peek at the CMMA values, with the flag 4031 KVM_S390_CMMA_PEEK set. 4032 4033The ioctl takes parameters via the kvm_s390_cmma_log struct. The desired 4034values are written to a buffer whose location is indicated via the "values" 4035member in the kvm_s390_cmma_log struct. The values in the input struct are 4036also updated as needed. 4037 4038Each CMMA value takes up one byte. 4039 4040:: 4041 4042 struct kvm_s390_cmma_log { 4043 __u64 start_gfn; 4044 __u32 count; 4045 __u32 flags; 4046 union { 4047 __u64 remaining; 4048 __u64 mask; 4049 }; 4050 __u64 values; 4051 }; 4052 4053start_gfn is the number of the first guest frame whose CMMA values are 4054to be retrieved, 4055 4056count is the length of the buffer in bytes, 4057 4058values points to the buffer where the result will be written to. 4059 4060If count is greater than KVM_S390_SKEYS_MAX, then it is considered to be 4061KVM_S390_SKEYS_MAX. KVM_S390_SKEYS_MAX is re-used for consistency with 4062other ioctls. 4063 4064The result is written in the buffer pointed to by the field values, and 4065the values of the input parameter are updated as follows. 4066 4067Depending on the flags, different actions are performed. The only 4068supported flag so far is KVM_S390_CMMA_PEEK. 4069 4070The default behaviour if KVM_S390_CMMA_PEEK is not set is: 4071start_gfn will indicate the first page frame whose CMMA bits were dirty. 4072It is not necessarily the same as the one passed as input, as clean pages 4073are skipped. 4074 4075count will indicate the number of bytes actually written in the buffer. 4076It can (and very often will) be smaller than the input value, since the 4077buffer is only filled until 16 bytes of clean values are found (which 4078are then not copied in the buffer). Since a CMMA migration block needs 4079the base address and the length, for a total of 16 bytes, we will send 4080back some clean data if there is some dirty data afterwards, as long as 4081the size of the clean data does not exceed the size of the header. This 4082allows to minimize the amount of data to be saved or transferred over 4083the network at the expense of more roundtrips to userspace. The next 4084invocation of the ioctl will skip over all the clean values, saving 4085potentially more than just the 16 bytes we found. 4086 4087If KVM_S390_CMMA_PEEK is set: 4088the existing storage attributes are read even when not in migration 4089mode, and no other action is performed; 4090 4091the output start_gfn will be equal to the input start_gfn, 4092 4093the output count will be equal to the input count, except if the end of 4094memory has been reached. 4095 4096In both cases: 4097the field "remaining" will indicate the total number of dirty CMMA values 4098still remaining, or 0 if KVM_S390_CMMA_PEEK is set and migration mode is 4099not enabled. 4100 4101mask is unused. 4102 4103values points to the userspace buffer where the result will be stored. 4104 4105This ioctl can fail with -ENOMEM if not enough memory can be allocated to 4106complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if 4107KVM_S390_CMMA_PEEK is not set but migration mode was not enabled, with 4108-EFAULT if the userspace address is invalid or if no page table is 4109present for the addresses (e.g. when using hugepages). 4110 41114.108 KVM_S390_SET_CMMA_BITS 4112---------------------------- 4113 4114:Capability: KVM_CAP_S390_CMMA_MIGRATION 4115:Architectures: s390 4116:Type: vm ioctl 4117:Parameters: struct kvm_s390_cmma_log (in) 4118:Returns: 0 on success, a negative value on error 4119 4120This ioctl is used to set the values of the CMMA bits on the s390 4121architecture. It is meant to be used during live migration to restore 4122the CMMA values, but there are no restrictions on its use. 4123The ioctl takes parameters via the kvm_s390_cmma_values struct. 4124Each CMMA value takes up one byte. 4125 4126:: 4127 4128 struct kvm_s390_cmma_log { 4129 __u64 start_gfn; 4130 __u32 count; 4131 __u32 flags; 4132 union { 4133 __u64 remaining; 4134 __u64 mask; 4135 }; 4136 __u64 values; 4137 }; 4138 4139start_gfn indicates the starting guest frame number, 4140 4141count indicates how many values are to be considered in the buffer, 4142 4143flags is not used and must be 0. 4144 4145mask indicates which PGSTE bits are to be considered. 4146 4147remaining is not used. 4148 4149values points to the buffer in userspace where to store the values. 4150 4151This ioctl can fail with -ENOMEM if not enough memory can be allocated to 4152complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if 4153the count field is too large (e.g. more than KVM_S390_CMMA_SIZE_MAX) or 4154if the flags field was not 0, with -EFAULT if the userspace address is 4155invalid, if invalid pages are written to (e.g. after the end of memory) 4156or if no page table is present for the addresses (e.g. when using 4157hugepages). 4158 41594.109 KVM_PPC_GET_CPU_CHAR 4160-------------------------- 4161 4162:Capability: KVM_CAP_PPC_GET_CPU_CHAR 4163:Architectures: powerpc 4164:Type: vm ioctl 4165:Parameters: struct kvm_ppc_cpu_char (out) 4166:Returns: 0 on successful completion, 4167 -EFAULT if struct kvm_ppc_cpu_char cannot be written 4168 4169This ioctl gives userspace information about certain characteristics 4170of the CPU relating to speculative execution of instructions and 4171possible information leakage resulting from speculative execution (see 4172CVE-2017-5715, CVE-2017-5753 and CVE-2017-5754). The information is 4173returned in struct kvm_ppc_cpu_char, which looks like this:: 4174 4175 struct kvm_ppc_cpu_char { 4176 __u64 character; /* characteristics of the CPU */ 4177 __u64 behaviour; /* recommended software behaviour */ 4178 __u64 character_mask; /* valid bits in character */ 4179 __u64 behaviour_mask; /* valid bits in behaviour */ 4180 }; 4181 4182For extensibility, the character_mask and behaviour_mask fields 4183indicate which bits of character and behaviour have been filled in by 4184the kernel. If the set of defined bits is extended in future then 4185userspace will be able to tell whether it is running on a kernel that 4186knows about the new bits. 4187 4188The character field describes attributes of the CPU which can help 4189with preventing inadvertent information disclosure - specifically, 4190whether there is an instruction to flash-invalidate the L1 data cache 4191(ori 30,30,0 or mtspr SPRN_TRIG2,rN), whether the L1 data cache is set 4192to a mode where entries can only be used by the thread that created 4193them, whether the bcctr[l] instruction prevents speculation, and 4194whether a speculation barrier instruction (ori 31,31,0) is provided. 4195 4196The behaviour field describes actions that software should take to 4197prevent inadvertent information disclosure, and thus describes which 4198vulnerabilities the hardware is subject to; specifically whether the 4199L1 data cache should be flushed when returning to user mode from the 4200kernel, and whether a speculation barrier should be placed between an 4201array bounds check and the array access. 4202 4203These fields use the same bit definitions as the new 4204H_GET_CPU_CHARACTERISTICS hypercall. 4205 42064.110 KVM_MEMORY_ENCRYPT_OP 4207--------------------------- 4208 4209:Capability: basic 4210:Architectures: x86 4211:Type: system 4212:Parameters: an opaque platform specific structure (in/out) 4213:Returns: 0 on success; -1 on error 4214 4215If the platform supports creating encrypted VMs then this ioctl can be used 4216for issuing platform-specific memory encryption commands to manage those 4217encrypted VMs. 4218 4219Currently, this ioctl is used for issuing Secure Encrypted Virtualization 4220(SEV) commands on AMD Processors. The SEV commands are defined in 4221Documentation/virt/kvm/amd-memory-encryption.rst. 4222 42234.111 KVM_MEMORY_ENCRYPT_REG_REGION 4224----------------------------------- 4225 4226:Capability: basic 4227:Architectures: x86 4228:Type: system 4229:Parameters: struct kvm_enc_region (in) 4230:Returns: 0 on success; -1 on error 4231 4232This ioctl can be used to register a guest memory region which may 4233contain encrypted data (e.g. guest RAM, SMRAM etc). 4234 4235It is used in the SEV-enabled guest. When encryption is enabled, a guest 4236memory region may contain encrypted data. The SEV memory encryption 4237engine uses a tweak such that two identical plaintext pages, each at 4238different locations will have differing ciphertexts. So swapping or 4239moving ciphertext of those pages will not result in plaintext being 4240swapped. So relocating (or migrating) physical backing pages for the SEV 4241guest will require some additional steps. 4242 4243Note: The current SEV key management spec does not provide commands to 4244swap or migrate (move) ciphertext pages. Hence, for now we pin the guest 4245memory region registered with the ioctl. 4246 42474.112 KVM_MEMORY_ENCRYPT_UNREG_REGION 4248------------------------------------- 4249 4250:Capability: basic 4251:Architectures: x86 4252:Type: system 4253:Parameters: struct kvm_enc_region (in) 4254:Returns: 0 on success; -1 on error 4255 4256This ioctl can be used to unregister the guest memory region registered 4257with KVM_MEMORY_ENCRYPT_REG_REGION ioctl above. 4258 42594.113 KVM_HYPERV_EVENTFD 4260------------------------ 4261 4262:Capability: KVM_CAP_HYPERV_EVENTFD 4263:Architectures: x86 4264:Type: vm ioctl 4265:Parameters: struct kvm_hyperv_eventfd (in) 4266 4267This ioctl (un)registers an eventfd to receive notifications from the guest on 4268the specified Hyper-V connection id through the SIGNAL_EVENT hypercall, without 4269causing a user exit. SIGNAL_EVENT hypercall with non-zero event flag number 4270(bits 24-31) still triggers a KVM_EXIT_HYPERV_HCALL user exit. 4271 4272:: 4273 4274 struct kvm_hyperv_eventfd { 4275 __u32 conn_id; 4276 __s32 fd; 4277 __u32 flags; 4278 __u32 padding[3]; 4279 }; 4280 4281The conn_id field should fit within 24 bits:: 4282 4283 #define KVM_HYPERV_CONN_ID_MASK 0x00ffffff 4284 4285The acceptable values for the flags field are:: 4286 4287 #define KVM_HYPERV_EVENTFD_DEASSIGN (1 << 0) 4288 4289:Returns: 0 on success, 4290 -EINVAL if conn_id or flags is outside the allowed range, 4291 -ENOENT on deassign if the conn_id isn't registered, 4292 -EEXIST on assign if the conn_id is already registered 4293 42944.114 KVM_GET_NESTED_STATE 4295-------------------------- 4296 4297:Capability: KVM_CAP_NESTED_STATE 4298:Architectures: x86 4299:Type: vcpu ioctl 4300:Parameters: struct kvm_nested_state (in/out) 4301:Returns: 0 on success, -1 on error 4302 4303Errors: 4304 4305 ===== ============================================================= 4306 E2BIG the total state size exceeds the value of 'size' specified by 4307 the user; the size required will be written into size. 4308 ===== ============================================================= 4309 4310:: 4311 4312 struct kvm_nested_state { 4313 __u16 flags; 4314 __u16 format; 4315 __u32 size; 4316 4317 union { 4318 struct kvm_vmx_nested_state_hdr vmx; 4319 struct kvm_svm_nested_state_hdr svm; 4320 4321 /* Pad the header to 128 bytes. */ 4322 __u8 pad[120]; 4323 } hdr; 4324 4325 union { 4326 struct kvm_vmx_nested_state_data vmx[0]; 4327 struct kvm_svm_nested_state_data svm[0]; 4328 } data; 4329 }; 4330 4331 #define KVM_STATE_NESTED_GUEST_MODE 0x00000001 4332 #define KVM_STATE_NESTED_RUN_PENDING 0x00000002 4333 #define KVM_STATE_NESTED_EVMCS 0x00000004 4334 4335 #define KVM_STATE_NESTED_FORMAT_VMX 0 4336 #define KVM_STATE_NESTED_FORMAT_SVM 1 4337 4338 #define KVM_STATE_NESTED_VMX_VMCS_SIZE 0x1000 4339 4340 #define KVM_STATE_NESTED_VMX_SMM_GUEST_MODE 0x00000001 4341 #define KVM_STATE_NESTED_VMX_SMM_VMXON 0x00000002 4342 4343#define KVM_STATE_VMX_PREEMPTION_TIMER_DEADLINE 0x00000001 4344 4345 struct kvm_vmx_nested_state_hdr { 4346 __u64 vmxon_pa; 4347 __u64 vmcs12_pa; 4348 4349 struct { 4350 __u16 flags; 4351 } smm; 4352 4353 __u32 flags; 4354 __u64 preemption_timer_deadline; 4355 }; 4356 4357 struct kvm_vmx_nested_state_data { 4358 __u8 vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE]; 4359 __u8 shadow_vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE]; 4360 }; 4361 4362This ioctl copies the vcpu's nested virtualization state from the kernel to 4363userspace. 4364 4365The maximum size of the state can be retrieved by passing KVM_CAP_NESTED_STATE 4366to the KVM_CHECK_EXTENSION ioctl(). 4367 43684.115 KVM_SET_NESTED_STATE 4369-------------------------- 4370 4371:Capability: KVM_CAP_NESTED_STATE 4372:Architectures: x86 4373:Type: vcpu ioctl 4374:Parameters: struct kvm_nested_state (in) 4375:Returns: 0 on success, -1 on error 4376 4377This copies the vcpu's kvm_nested_state struct from userspace to the kernel. 4378For the definition of struct kvm_nested_state, see KVM_GET_NESTED_STATE. 4379 43804.116 KVM_(UN)REGISTER_COALESCED_MMIO 4381------------------------------------- 4382 4383:Capability: KVM_CAP_COALESCED_MMIO (for coalesced mmio) 4384 KVM_CAP_COALESCED_PIO (for coalesced pio) 4385:Architectures: all 4386:Type: vm ioctl 4387:Parameters: struct kvm_coalesced_mmio_zone 4388:Returns: 0 on success, < 0 on error 4389 4390Coalesced I/O is a performance optimization that defers hardware 4391register write emulation so that userspace exits are avoided. It is 4392typically used to reduce the overhead of emulating frequently accessed 4393hardware registers. 4394 4395When a hardware register is configured for coalesced I/O, write accesses 4396do not exit to userspace and their value is recorded in a ring buffer 4397that is shared between kernel and userspace. 4398 4399Coalesced I/O is used if one or more write accesses to a hardware 4400register can be deferred until a read or a write to another hardware 4401register on the same device. This last access will cause a vmexit and 4402userspace will process accesses from the ring buffer before emulating 4403it. That will avoid exiting to userspace on repeated writes. 4404 4405Coalesced pio is based on coalesced mmio. There is little difference 4406between coalesced mmio and pio except that coalesced pio records accesses 4407to I/O ports. 4408 44094.117 KVM_CLEAR_DIRTY_LOG (vm ioctl) 4410------------------------------------ 4411 4412:Capability: KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 4413:Architectures: x86, arm, arm64, mips 4414:Type: vm ioctl 4415:Parameters: struct kvm_dirty_log (in) 4416:Returns: 0 on success, -1 on error 4417 4418:: 4419 4420 /* for KVM_CLEAR_DIRTY_LOG */ 4421 struct kvm_clear_dirty_log { 4422 __u32 slot; 4423 __u32 num_pages; 4424 __u64 first_page; 4425 union { 4426 void __user *dirty_bitmap; /* one bit per page */ 4427 __u64 padding; 4428 }; 4429 }; 4430 4431The ioctl clears the dirty status of pages in a memory slot, according to 4432the bitmap that is passed in struct kvm_clear_dirty_log's dirty_bitmap 4433field. Bit 0 of the bitmap corresponds to page "first_page" in the 4434memory slot, and num_pages is the size in bits of the input bitmap. 4435first_page must be a multiple of 64; num_pages must also be a multiple of 443664 unless first_page + num_pages is the size of the memory slot. For each 4437bit that is set in the input bitmap, the corresponding page is marked "clean" 4438in KVM's dirty bitmap, and dirty tracking is re-enabled for that page 4439(for example via write-protection, or by clearing the dirty bit in 4440a page table entry). 4441 4442If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 specifies 4443the address space for which you want to return the dirty bitmap. 4444They must be less than the value that KVM_CHECK_EXTENSION returns for 4445the KVM_CAP_MULTI_ADDRESS_SPACE capability. 4446 4447This ioctl is mostly useful when KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 4448is enabled; for more information, see the description of the capability. 4449However, it can always be used as long as KVM_CHECK_EXTENSION confirms 4450that KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is present. 4451 44524.118 KVM_GET_SUPPORTED_HV_CPUID 4453-------------------------------- 4454 4455:Capability: KVM_CAP_HYPERV_CPUID 4456:Architectures: x86 4457:Type: vcpu ioctl 4458:Parameters: struct kvm_cpuid2 (in/out) 4459:Returns: 0 on success, -1 on error 4460 4461:: 4462 4463 struct kvm_cpuid2 { 4464 __u32 nent; 4465 __u32 padding; 4466 struct kvm_cpuid_entry2 entries[0]; 4467 }; 4468 4469 struct kvm_cpuid_entry2 { 4470 __u32 function; 4471 __u32 index; 4472 __u32 flags; 4473 __u32 eax; 4474 __u32 ebx; 4475 __u32 ecx; 4476 __u32 edx; 4477 __u32 padding[3]; 4478 }; 4479 4480This ioctl returns x86 cpuid features leaves related to Hyper-V emulation in 4481KVM. Userspace can use the information returned by this ioctl to construct 4482cpuid information presented to guests consuming Hyper-V enlightenments (e.g. 4483Windows or Hyper-V guests). 4484 4485CPUID feature leaves returned by this ioctl are defined by Hyper-V Top Level 4486Functional Specification (TLFS). These leaves can't be obtained with 4487KVM_GET_SUPPORTED_CPUID ioctl because some of them intersect with KVM feature 4488leaves (0x40000000, 0x40000001). 4489 4490Currently, the following list of CPUID leaves are returned: 4491 - HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS 4492 - HYPERV_CPUID_INTERFACE 4493 - HYPERV_CPUID_VERSION 4494 - HYPERV_CPUID_FEATURES 4495 - HYPERV_CPUID_ENLIGHTMENT_INFO 4496 - HYPERV_CPUID_IMPLEMENT_LIMITS 4497 - HYPERV_CPUID_NESTED_FEATURES 4498 4499HYPERV_CPUID_NESTED_FEATURES leaf is only exposed when Enlightened VMCS was 4500enabled on the corresponding vCPU (KVM_CAP_HYPERV_ENLIGHTENED_VMCS). 4501 4502Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure 4503with the 'nent' field indicating the number of entries in the variable-size 4504array 'entries'. If the number of entries is too low to describe all Hyper-V 4505feature leaves, an error (E2BIG) is returned. If the number is more or equal 4506to the number of Hyper-V feature leaves, the 'nent' field is adjusted to the 4507number of valid entries in the 'entries' array, which is then filled. 4508 4509'index' and 'flags' fields in 'struct kvm_cpuid_entry2' are currently reserved, 4510userspace should not expect to get any particular value there. 4511 45124.119 KVM_ARM_VCPU_FINALIZE 4513--------------------------- 4514 4515:Architectures: arm, arm64 4516:Type: vcpu ioctl 4517:Parameters: int feature (in) 4518:Returns: 0 on success, -1 on error 4519 4520Errors: 4521 4522 ====== ============================================================== 4523 EPERM feature not enabled, needs configuration, or already finalized 4524 EINVAL feature unknown or not present 4525 ====== ============================================================== 4526 4527Recognised values for feature: 4528 4529 ===== =========================================== 4530 arm64 KVM_ARM_VCPU_SVE (requires KVM_CAP_ARM_SVE) 4531 ===== =========================================== 4532 4533Finalizes the configuration of the specified vcpu feature. 4534 4535The vcpu must already have been initialised, enabling the affected feature, by 4536means of a successful KVM_ARM_VCPU_INIT call with the appropriate flag set in 4537features[]. 4538 4539For affected vcpu features, this is a mandatory step that must be performed 4540before the vcpu is fully usable. 4541 4542Between KVM_ARM_VCPU_INIT and KVM_ARM_VCPU_FINALIZE, the feature may be 4543configured by use of ioctls such as KVM_SET_ONE_REG. The exact configuration 4544that should be performaned and how to do it are feature-dependent. 4545 4546Other calls that depend on a particular feature being finalized, such as 4547KVM_RUN, KVM_GET_REG_LIST, KVM_GET_ONE_REG and KVM_SET_ONE_REG, will fail with 4548-EPERM unless the feature has already been finalized by means of a 4549KVM_ARM_VCPU_FINALIZE call. 4550 4551See KVM_ARM_VCPU_INIT for details of vcpu features that require finalization 4552using this ioctl. 4553 45544.120 KVM_SET_PMU_EVENT_FILTER 4555------------------------------ 4556 4557:Capability: KVM_CAP_PMU_EVENT_FILTER 4558:Architectures: x86 4559:Type: vm ioctl 4560:Parameters: struct kvm_pmu_event_filter (in) 4561:Returns: 0 on success, -1 on error 4562 4563:: 4564 4565 struct kvm_pmu_event_filter { 4566 __u32 action; 4567 __u32 nevents; 4568 __u32 fixed_counter_bitmap; 4569 __u32 flags; 4570 __u32 pad[4]; 4571 __u64 events[0]; 4572 }; 4573 4574This ioctl restricts the set of PMU events that the guest can program. 4575The argument holds a list of events which will be allowed or denied. 4576The eventsel+umask of each event the guest attempts to program is compared 4577against the events field to determine whether the guest should have access. 4578The events field only controls general purpose counters; fixed purpose 4579counters are controlled by the fixed_counter_bitmap. 4580 4581No flags are defined yet, the field must be zero. 4582 4583Valid values for 'action':: 4584 4585 #define KVM_PMU_EVENT_ALLOW 0 4586 #define KVM_PMU_EVENT_DENY 1 4587 45884.121 KVM_PPC_SVM_OFF 4589--------------------- 4590 4591:Capability: basic 4592:Architectures: powerpc 4593:Type: vm ioctl 4594:Parameters: none 4595:Returns: 0 on successful completion, 4596 4597Errors: 4598 4599 ====== ================================================================ 4600 EINVAL if ultravisor failed to terminate the secure guest 4601 ENOMEM if hypervisor failed to allocate new radix page tables for guest 4602 ====== ================================================================ 4603 4604This ioctl is used to turn off the secure mode of the guest or transition 4605the guest from secure mode to normal mode. This is invoked when the guest 4606is reset. This has no effect if called for a normal guest. 4607 4608This ioctl issues an ultravisor call to terminate the secure guest, 4609unpins the VPA pages and releases all the device pages that are used to 4610track the secure pages by hypervisor. 4611 46124.122 KVM_S390_NORMAL_RESET 4613--------------------------- 4614 4615:Capability: KVM_CAP_S390_VCPU_RESETS 4616:Architectures: s390 4617:Type: vcpu ioctl 4618:Parameters: none 4619:Returns: 0 4620 4621This ioctl resets VCPU registers and control structures according to 4622the cpu reset definition in the POP (Principles Of Operation). 4623 46244.123 KVM_S390_INITIAL_RESET 4625---------------------------- 4626 4627:Capability: none 4628:Architectures: s390 4629:Type: vcpu ioctl 4630:Parameters: none 4631:Returns: 0 4632 4633This ioctl resets VCPU registers and control structures according to 4634the initial cpu reset definition in the POP. However, the cpu is not 4635put into ESA mode. This reset is a superset of the normal reset. 4636 46374.124 KVM_S390_CLEAR_RESET 4638-------------------------- 4639 4640:Capability: KVM_CAP_S390_VCPU_RESETS 4641:Architectures: s390 4642:Type: vcpu ioctl 4643:Parameters: none 4644:Returns: 0 4645 4646This ioctl resets VCPU registers and control structures according to 4647the clear cpu reset definition in the POP. However, the cpu is not put 4648into ESA mode. This reset is a superset of the initial reset. 4649 4650 46514.125 KVM_S390_PV_COMMAND 4652------------------------- 4653 4654:Capability: KVM_CAP_S390_PROTECTED 4655:Architectures: s390 4656:Type: vm ioctl 4657:Parameters: struct kvm_pv_cmd 4658:Returns: 0 on success, < 0 on error 4659 4660:: 4661 4662 struct kvm_pv_cmd { 4663 __u32 cmd; /* Command to be executed */ 4664 __u16 rc; /* Ultravisor return code */ 4665 __u16 rrc; /* Ultravisor return reason code */ 4666 __u64 data; /* Data or address */ 4667 __u32 flags; /* flags for future extensions. Must be 0 for now */ 4668 __u32 reserved[3]; 4669 }; 4670 4671cmd values: 4672 4673KVM_PV_ENABLE 4674 Allocate memory and register the VM with the Ultravisor, thereby 4675 donating memory to the Ultravisor that will become inaccessible to 4676 KVM. All existing CPUs are converted to protected ones. After this 4677 command has succeeded, any CPU added via hotplug will become 4678 protected during its creation as well. 4679 4680 Errors: 4681 4682 ===== ============================= 4683 EINTR an unmasked signal is pending 4684 ===== ============================= 4685 4686KVM_PV_DISABLE 4687 4688 Deregister the VM from the Ultravisor and reclaim the memory that 4689 had been donated to the Ultravisor, making it usable by the kernel 4690 again. All registered VCPUs are converted back to non-protected 4691 ones. 4692 4693KVM_PV_VM_SET_SEC_PARMS 4694 Pass the image header from VM memory to the Ultravisor in 4695 preparation of image unpacking and verification. 4696 4697KVM_PV_VM_UNPACK 4698 Unpack (protect and decrypt) a page of the encrypted boot image. 4699 4700KVM_PV_VM_VERIFY 4701 Verify the integrity of the unpacked image. Only if this succeeds, 4702 KVM is allowed to start protected VCPUs. 4703 4704 47055. The kvm_run structure 4706======================== 4707 4708Application code obtains a pointer to the kvm_run structure by 4709mmap()ing a vcpu fd. From that point, application code can control 4710execution by changing fields in kvm_run prior to calling the KVM_RUN 4711ioctl, and obtain information about the reason KVM_RUN returned by 4712looking up structure members. 4713 4714:: 4715 4716 struct kvm_run { 4717 /* in */ 4718 __u8 request_interrupt_window; 4719 4720Request that KVM_RUN return when it becomes possible to inject external 4721interrupts into the guest. Useful in conjunction with KVM_INTERRUPT. 4722 4723:: 4724 4725 __u8 immediate_exit; 4726 4727This field is polled once when KVM_RUN starts; if non-zero, KVM_RUN 4728exits immediately, returning -EINTR. In the common scenario where a 4729signal is used to "kick" a VCPU out of KVM_RUN, this field can be used 4730to avoid usage of KVM_SET_SIGNAL_MASK, which has worse scalability. 4731Rather than blocking the signal outside KVM_RUN, userspace can set up 4732a signal handler that sets run->immediate_exit to a non-zero value. 4733 4734This field is ignored if KVM_CAP_IMMEDIATE_EXIT is not available. 4735 4736:: 4737 4738 __u8 padding1[6]; 4739 4740 /* out */ 4741 __u32 exit_reason; 4742 4743When KVM_RUN has returned successfully (return value 0), this informs 4744application code why KVM_RUN has returned. Allowable values for this 4745field are detailed below. 4746 4747:: 4748 4749 __u8 ready_for_interrupt_injection; 4750 4751If request_interrupt_window has been specified, this field indicates 4752an interrupt can be injected now with KVM_INTERRUPT. 4753 4754:: 4755 4756 __u8 if_flag; 4757 4758The value of the current interrupt flag. Only valid if in-kernel 4759local APIC is not used. 4760 4761:: 4762 4763 __u16 flags; 4764 4765More architecture-specific flags detailing state of the VCPU that may 4766affect the device's behavior. The only currently defined flag is 4767KVM_RUN_X86_SMM, which is valid on x86 machines and is set if the 4768VCPU is in system management mode. 4769 4770:: 4771 4772 /* in (pre_kvm_run), out (post_kvm_run) */ 4773 __u64 cr8; 4774 4775The value of the cr8 register. Only valid if in-kernel local APIC is 4776not used. Both input and output. 4777 4778:: 4779 4780 __u64 apic_base; 4781 4782The value of the APIC BASE msr. Only valid if in-kernel local 4783APIC is not used. Both input and output. 4784 4785:: 4786 4787 union { 4788 /* KVM_EXIT_UNKNOWN */ 4789 struct { 4790 __u64 hardware_exit_reason; 4791 } hw; 4792 4793If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown 4794reasons. Further architecture-specific information is available in 4795hardware_exit_reason. 4796 4797:: 4798 4799 /* KVM_EXIT_FAIL_ENTRY */ 4800 struct { 4801 __u64 hardware_entry_failure_reason; 4802 __u32 cpu; /* if KVM_LAST_CPU */ 4803 } fail_entry; 4804 4805If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due 4806to unknown reasons. Further architecture-specific information is 4807available in hardware_entry_failure_reason. 4808 4809:: 4810 4811 /* KVM_EXIT_EXCEPTION */ 4812 struct { 4813 __u32 exception; 4814 __u32 error_code; 4815 } ex; 4816 4817Unused. 4818 4819:: 4820 4821 /* KVM_EXIT_IO */ 4822 struct { 4823 #define KVM_EXIT_IO_IN 0 4824 #define KVM_EXIT_IO_OUT 1 4825 __u8 direction; 4826 __u8 size; /* bytes */ 4827 __u16 port; 4828 __u32 count; 4829 __u64 data_offset; /* relative to kvm_run start */ 4830 } io; 4831 4832If exit_reason is KVM_EXIT_IO, then the vcpu has 4833executed a port I/O instruction which could not be satisfied by kvm. 4834data_offset describes where the data is located (KVM_EXIT_IO_OUT) or 4835where kvm expects application code to place the data for the next 4836KVM_RUN invocation (KVM_EXIT_IO_IN). Data format is a packed array. 4837 4838:: 4839 4840 /* KVM_EXIT_DEBUG */ 4841 struct { 4842 struct kvm_debug_exit_arch arch; 4843 } debug; 4844 4845If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event 4846for which architecture specific information is returned. 4847 4848:: 4849 4850 /* KVM_EXIT_MMIO */ 4851 struct { 4852 __u64 phys_addr; 4853 __u8 data[8]; 4854 __u32 len; 4855 __u8 is_write; 4856 } mmio; 4857 4858If exit_reason is KVM_EXIT_MMIO, then the vcpu has 4859executed a memory-mapped I/O instruction which could not be satisfied 4860by kvm. The 'data' member contains the written data if 'is_write' is 4861true, and should be filled by application code otherwise. 4862 4863The 'data' member contains, in its first 'len' bytes, the value as it would 4864appear if the VCPU performed a load or store of the appropriate width directly 4865to the byte array. 4866 4867.. note:: 4868 4869 For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR and 4870 KVM_EXIT_EPR the corresponding 4871 4872operations are complete (and guest state is consistent) only after userspace 4873has re-entered the kernel with KVM_RUN. The kernel side will first finish 4874incomplete operations and then check for pending signals. Userspace 4875can re-enter the guest with an unmasked signal pending to complete 4876pending operations. 4877 4878:: 4879 4880 /* KVM_EXIT_HYPERCALL */ 4881 struct { 4882 __u64 nr; 4883 __u64 args[6]; 4884 __u64 ret; 4885 __u32 longmode; 4886 __u32 pad; 4887 } hypercall; 4888 4889Unused. This was once used for 'hypercall to userspace'. To implement 4890such functionality, use KVM_EXIT_IO (x86) or KVM_EXIT_MMIO (all except s390). 4891 4892.. note:: KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO. 4893 4894:: 4895 4896 /* KVM_EXIT_TPR_ACCESS */ 4897 struct { 4898 __u64 rip; 4899 __u32 is_write; 4900 __u32 pad; 4901 } tpr_access; 4902 4903To be documented (KVM_TPR_ACCESS_REPORTING). 4904 4905:: 4906 4907 /* KVM_EXIT_S390_SIEIC */ 4908 struct { 4909 __u8 icptcode; 4910 __u64 mask; /* psw upper half */ 4911 __u64 addr; /* psw lower half */ 4912 __u16 ipa; 4913 __u32 ipb; 4914 } s390_sieic; 4915 4916s390 specific. 4917 4918:: 4919 4920 /* KVM_EXIT_S390_RESET */ 4921 #define KVM_S390_RESET_POR 1 4922 #define KVM_S390_RESET_CLEAR 2 4923 #define KVM_S390_RESET_SUBSYSTEM 4 4924 #define KVM_S390_RESET_CPU_INIT 8 4925 #define KVM_S390_RESET_IPL 16 4926 __u64 s390_reset_flags; 4927 4928s390 specific. 4929 4930:: 4931 4932 /* KVM_EXIT_S390_UCONTROL */ 4933 struct { 4934 __u64 trans_exc_code; 4935 __u32 pgm_code; 4936 } s390_ucontrol; 4937 4938s390 specific. A page fault has occurred for a user controlled virtual 4939machine (KVM_VM_S390_UNCONTROL) on it's host page table that cannot be 4940resolved by the kernel. 4941The program code and the translation exception code that were placed 4942in the cpu's lowcore are presented here as defined by the z Architecture 4943Principles of Operation Book in the Chapter for Dynamic Address Translation 4944(DAT) 4945 4946:: 4947 4948 /* KVM_EXIT_DCR */ 4949 struct { 4950 __u32 dcrn; 4951 __u32 data; 4952 __u8 is_write; 4953 } dcr; 4954 4955Deprecated - was used for 440 KVM. 4956 4957:: 4958 4959 /* KVM_EXIT_OSI */ 4960 struct { 4961 __u64 gprs[32]; 4962 } osi; 4963 4964MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch 4965hypercalls and exit with this exit struct that contains all the guest gprs. 4966 4967If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall. 4968Userspace can now handle the hypercall and when it's done modify the gprs as 4969necessary. Upon guest entry all guest GPRs will then be replaced by the values 4970in this struct. 4971 4972:: 4973 4974 /* KVM_EXIT_PAPR_HCALL */ 4975 struct { 4976 __u64 nr; 4977 __u64 ret; 4978 __u64 args[9]; 4979 } papr_hcall; 4980 4981This is used on 64-bit PowerPC when emulating a pSeries partition, 4982e.g. with the 'pseries' machine type in qemu. It occurs when the 4983guest does a hypercall using the 'sc 1' instruction. The 'nr' field 4984contains the hypercall number (from the guest R3), and 'args' contains 4985the arguments (from the guest R4 - R12). Userspace should put the 4986return code in 'ret' and any extra returned values in args[]. 4987The possible hypercalls are defined in the Power Architecture Platform 4988Requirements (PAPR) document available from www.power.org (free 4989developer registration required to access it). 4990 4991:: 4992 4993 /* KVM_EXIT_S390_TSCH */ 4994 struct { 4995 __u16 subchannel_id; 4996 __u16 subchannel_nr; 4997 __u32 io_int_parm; 4998 __u32 io_int_word; 4999 __u32 ipb; 5000 __u8 dequeued; 5001 } s390_tsch; 5002 5003s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled 5004and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O 5005interrupt for the target subchannel has been dequeued and subchannel_id, 5006subchannel_nr, io_int_parm and io_int_word contain the parameters for that 5007interrupt. ipb is needed for instruction parameter decoding. 5008 5009:: 5010 5011 /* KVM_EXIT_EPR */ 5012 struct { 5013 __u32 epr; 5014 } epr; 5015 5016On FSL BookE PowerPC chips, the interrupt controller has a fast patch 5017interrupt acknowledge path to the core. When the core successfully 5018delivers an interrupt, it automatically populates the EPR register with 5019the interrupt vector number and acknowledges the interrupt inside 5020the interrupt controller. 5021 5022In case the interrupt controller lives in user space, we need to do 5023the interrupt acknowledge cycle through it to fetch the next to be 5024delivered interrupt vector using this exit. 5025 5026It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an 5027external interrupt has just been delivered into the guest. User space 5028should put the acknowledged interrupt vector into the 'epr' field. 5029 5030:: 5031 5032 /* KVM_EXIT_SYSTEM_EVENT */ 5033 struct { 5034 #define KVM_SYSTEM_EVENT_SHUTDOWN 1 5035 #define KVM_SYSTEM_EVENT_RESET 2 5036 #define KVM_SYSTEM_EVENT_CRASH 3 5037 __u32 type; 5038 __u64 flags; 5039 } system_event; 5040 5041If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered 5042a system-level event using some architecture specific mechanism (hypercall 5043or some special instruction). In case of ARM/ARM64, this is triggered using 5044HVC instruction based PSCI call from the vcpu. The 'type' field describes 5045the system-level event type. The 'flags' field describes architecture 5046specific flags for the system-level event. 5047 5048Valid values for 'type' are: 5049 5050 - KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the 5051 VM. Userspace is not obliged to honour this, and if it does honour 5052 this does not need to destroy the VM synchronously (ie it may call 5053 KVM_RUN again before shutdown finally occurs). 5054 - KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM. 5055 As with SHUTDOWN, userspace can choose to ignore the request, or 5056 to schedule the reset to occur in the future and may call KVM_RUN again. 5057 - KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest 5058 has requested a crash condition maintenance. Userspace can choose 5059 to ignore the request, or to gather VM memory core dump and/or 5060 reset/shutdown of the VM. 5061 5062:: 5063 5064 /* KVM_EXIT_IOAPIC_EOI */ 5065 struct { 5066 __u8 vector; 5067 } eoi; 5068 5069Indicates that the VCPU's in-kernel local APIC received an EOI for a 5070level-triggered IOAPIC interrupt. This exit only triggers when the 5071IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled); 5072the userspace IOAPIC should process the EOI and retrigger the interrupt if 5073it is still asserted. Vector is the LAPIC interrupt vector for which the 5074EOI was received. 5075 5076:: 5077 5078 struct kvm_hyperv_exit { 5079 #define KVM_EXIT_HYPERV_SYNIC 1 5080 #define KVM_EXIT_HYPERV_HCALL 2 5081 #define KVM_EXIT_HYPERV_SYNDBG 3 5082 __u32 type; 5083 __u32 pad1; 5084 union { 5085 struct { 5086 __u32 msr; 5087 __u32 pad2; 5088 __u64 control; 5089 __u64 evt_page; 5090 __u64 msg_page; 5091 } synic; 5092 struct { 5093 __u64 input; 5094 __u64 result; 5095 __u64 params[2]; 5096 } hcall; 5097 struct { 5098 __u32 msr; 5099 __u32 pad2; 5100 __u64 control; 5101 __u64 status; 5102 __u64 send_page; 5103 __u64 recv_page; 5104 __u64 pending_page; 5105 } syndbg; 5106 } u; 5107 }; 5108 /* KVM_EXIT_HYPERV */ 5109 struct kvm_hyperv_exit hyperv; 5110 5111Indicates that the VCPU exits into userspace to process some tasks 5112related to Hyper-V emulation. 5113 5114Valid values for 'type' are: 5115 5116 - KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about 5117 5118Hyper-V SynIC state change. Notification is used to remap SynIC 5119event/message pages and to enable/disable SynIC messages/events processing 5120in userspace. 5121 5122 - KVM_EXIT_HYPERV_SYNDBG -- synchronously notify user-space about 5123 5124Hyper-V Synthetic debugger state change. Notification is used to either update 5125the pending_page location or to send a control command (send the buffer located 5126in send_page or recv a buffer to recv_page). 5127 5128:: 5129 5130 /* KVM_EXIT_ARM_NISV */ 5131 struct { 5132 __u64 esr_iss; 5133 __u64 fault_ipa; 5134 } arm_nisv; 5135 5136Used on arm and arm64 systems. If a guest accesses memory not in a memslot, 5137KVM will typically return to userspace and ask it to do MMIO emulation on its 5138behalf. However, for certain classes of instructions, no instruction decode 5139(direction, length of memory access) is provided, and fetching and decoding 5140the instruction from the VM is overly complicated to live in the kernel. 5141 5142Historically, when this situation occurred, KVM would print a warning and kill 5143the VM. KVM assumed that if the guest accessed non-memslot memory, it was 5144trying to do I/O, which just couldn't be emulated, and the warning message was 5145phrased accordingly. However, what happened more often was that a guest bug 5146caused access outside the guest memory areas which should lead to a more 5147meaningful warning message and an external abort in the guest, if the access 5148did not fall within an I/O window. 5149 5150Userspace implementations can query for KVM_CAP_ARM_NISV_TO_USER, and enable 5151this capability at VM creation. Once this is done, these types of errors will 5152instead return to userspace with KVM_EXIT_ARM_NISV, with the valid bits from 5153the HSR (arm) and ESR_EL2 (arm64) in the esr_iss field, and the faulting IPA 5154in the fault_ipa field. Userspace can either fix up the access if it's 5155actually an I/O access by decoding the instruction from guest memory (if it's 5156very brave) and continue executing the guest, or it can decide to suspend, 5157dump, or restart the guest. 5158 5159Note that KVM does not skip the faulting instruction as it does for 5160KVM_EXIT_MMIO, but userspace has to emulate any change to the processing state 5161if it decides to decode and emulate the instruction. 5162 5163:: 5164 5165 /* Fix the size of the union. */ 5166 char padding[256]; 5167 }; 5168 5169 /* 5170 * shared registers between kvm and userspace. 5171 * kvm_valid_regs specifies the register classes set by the host 5172 * kvm_dirty_regs specified the register classes dirtied by userspace 5173 * struct kvm_sync_regs is architecture specific, as well as the 5174 * bits for kvm_valid_regs and kvm_dirty_regs 5175 */ 5176 __u64 kvm_valid_regs; 5177 __u64 kvm_dirty_regs; 5178 union { 5179 struct kvm_sync_regs regs; 5180 char padding[SYNC_REGS_SIZE_BYTES]; 5181 } s; 5182 5183If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access 5184certain guest registers without having to call SET/GET_*REGS. Thus we can 5185avoid some system call overhead if userspace has to handle the exit. 5186Userspace can query the validity of the structure by checking 5187kvm_valid_regs for specific bits. These bits are architecture specific 5188and usually define the validity of a groups of registers. (e.g. one bit 5189for general purpose registers) 5190 5191Please note that the kernel is allowed to use the kvm_run structure as the 5192primary storage for certain register types. Therefore, the kernel may use the 5193values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set. 5194 5195:: 5196 5197 }; 5198 5199 5200 52016. Capabilities that can be enabled on vCPUs 5202============================================ 5203 5204There are certain capabilities that change the behavior of the virtual CPU or 5205the virtual machine when enabled. To enable them, please see section 4.37. 5206Below you can find a list of capabilities and what their effect on the vCPU or 5207the virtual machine is when enabling them. 5208 5209The following information is provided along with the description: 5210 5211 Architectures: 5212 which instruction set architectures provide this ioctl. 5213 x86 includes both i386 and x86_64. 5214 5215 Target: 5216 whether this is a per-vcpu or per-vm capability. 5217 5218 Parameters: 5219 what parameters are accepted by the capability. 5220 5221 Returns: 5222 the return value. General error numbers (EBADF, ENOMEM, EINVAL) 5223 are not detailed, but errors with specific meanings are. 5224 5225 52266.1 KVM_CAP_PPC_OSI 5227------------------- 5228 5229:Architectures: ppc 5230:Target: vcpu 5231:Parameters: none 5232:Returns: 0 on success; -1 on error 5233 5234This capability enables interception of OSI hypercalls that otherwise would 5235be treated as normal system calls to be injected into the guest. OSI hypercalls 5236were invented by Mac-on-Linux to have a standardized communication mechanism 5237between the guest and the host. 5238 5239When this capability is enabled, KVM_EXIT_OSI can occur. 5240 5241 52426.2 KVM_CAP_PPC_PAPR 5243-------------------- 5244 5245:Architectures: ppc 5246:Target: vcpu 5247:Parameters: none 5248:Returns: 0 on success; -1 on error 5249 5250This capability enables interception of PAPR hypercalls. PAPR hypercalls are 5251done using the hypercall instruction "sc 1". 5252 5253It also sets the guest privilege level to "supervisor" mode. Usually the guest 5254runs in "hypervisor" privilege mode with a few missing features. 5255 5256In addition to the above, it changes the semantics of SDR1. In this mode, the 5257HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the 5258HTAB invisible to the guest. 5259 5260When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur. 5261 5262 52636.3 KVM_CAP_SW_TLB 5264------------------ 5265 5266:Architectures: ppc 5267:Target: vcpu 5268:Parameters: args[0] is the address of a struct kvm_config_tlb 5269:Returns: 0 on success; -1 on error 5270 5271:: 5272 5273 struct kvm_config_tlb { 5274 __u64 params; 5275 __u64 array; 5276 __u32 mmu_type; 5277 __u32 array_len; 5278 }; 5279 5280Configures the virtual CPU's TLB array, establishing a shared memory area 5281between userspace and KVM. The "params" and "array" fields are userspace 5282addresses of mmu-type-specific data structures. The "array_len" field is an 5283safety mechanism, and should be set to the size in bytes of the memory that 5284userspace has reserved for the array. It must be at least the size dictated 5285by "mmu_type" and "params". 5286 5287While KVM_RUN is active, the shared region is under control of KVM. Its 5288contents are undefined, and any modification by userspace results in 5289boundedly undefined behavior. 5290 5291On return from KVM_RUN, the shared region will reflect the current state of 5292the guest's TLB. If userspace makes any changes, it must call KVM_DIRTY_TLB 5293to tell KVM which entries have been changed, prior to calling KVM_RUN again 5294on this vcpu. 5295 5296For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV: 5297 5298 - The "params" field is of type "struct kvm_book3e_206_tlb_params". 5299 - The "array" field points to an array of type "struct 5300 kvm_book3e_206_tlb_entry". 5301 - The array consists of all entries in the first TLB, followed by all 5302 entries in the second TLB. 5303 - Within a TLB, entries are ordered first by increasing set number. Within a 5304 set, entries are ordered by way (increasing ESEL). 5305 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1) 5306 where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value. 5307 - The tsize field of mas1 shall be set to 4K on TLB0, even though the 5308 hardware ignores this value for TLB0. 5309 53106.4 KVM_CAP_S390_CSS_SUPPORT 5311---------------------------- 5312 5313:Architectures: s390 5314:Target: vcpu 5315:Parameters: none 5316:Returns: 0 on success; -1 on error 5317 5318This capability enables support for handling of channel I/O instructions. 5319 5320TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are 5321handled in-kernel, while the other I/O instructions are passed to userspace. 5322 5323When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST 5324SUBCHANNEL intercepts. 5325 5326Note that even though this capability is enabled per-vcpu, the complete 5327virtual machine is affected. 5328 53296.5 KVM_CAP_PPC_EPR 5330------------------- 5331 5332:Architectures: ppc 5333:Target: vcpu 5334:Parameters: args[0] defines whether the proxy facility is active 5335:Returns: 0 on success; -1 on error 5336 5337This capability enables or disables the delivery of interrupts through the 5338external proxy facility. 5339 5340When enabled (args[0] != 0), every time the guest gets an external interrupt 5341delivered, it automatically exits into user space with a KVM_EXIT_EPR exit 5342to receive the topmost interrupt vector. 5343 5344When disabled (args[0] == 0), behavior is as if this facility is unsupported. 5345 5346When this capability is enabled, KVM_EXIT_EPR can occur. 5347 53486.6 KVM_CAP_IRQ_MPIC 5349-------------------- 5350 5351:Architectures: ppc 5352:Parameters: args[0] is the MPIC device fd; 5353 args[1] is the MPIC CPU number for this vcpu 5354 5355This capability connects the vcpu to an in-kernel MPIC device. 5356 53576.7 KVM_CAP_IRQ_XICS 5358-------------------- 5359 5360:Architectures: ppc 5361:Target: vcpu 5362:Parameters: args[0] is the XICS device fd; 5363 args[1] is the XICS CPU number (server ID) for this vcpu 5364 5365This capability connects the vcpu to an in-kernel XICS device. 5366 53676.8 KVM_CAP_S390_IRQCHIP 5368------------------------ 5369 5370:Architectures: s390 5371:Target: vm 5372:Parameters: none 5373 5374This capability enables the in-kernel irqchip for s390. Please refer to 5375"4.24 KVM_CREATE_IRQCHIP" for details. 5376 53776.9 KVM_CAP_MIPS_FPU 5378-------------------- 5379 5380:Architectures: mips 5381:Target: vcpu 5382:Parameters: args[0] is reserved for future use (should be 0). 5383 5384This capability allows the use of the host Floating Point Unit by the guest. It 5385allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is 5386done the ``KVM_REG_MIPS_FPR_*`` and ``KVM_REG_MIPS_FCR_*`` registers can be 5387accessed (depending on the current guest FPU register mode), and the Status.FR, 5388Config5.FRE bits are accessible via the KVM API and also from the guest, 5389depending on them being supported by the FPU. 5390 53916.10 KVM_CAP_MIPS_MSA 5392--------------------- 5393 5394:Architectures: mips 5395:Target: vcpu 5396:Parameters: args[0] is reserved for future use (should be 0). 5397 5398This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest. 5399It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest. 5400Once this is done the ``KVM_REG_MIPS_VEC_*`` and ``KVM_REG_MIPS_MSA_*`` 5401registers can be accessed, and the Config5.MSAEn bit is accessible via the 5402KVM API and also from the guest. 5403 54046.74 KVM_CAP_SYNC_REGS 5405---------------------- 5406 5407:Architectures: s390, x86 5408:Target: s390: always enabled, x86: vcpu 5409:Parameters: none 5410:Returns: x86: KVM_CHECK_EXTENSION returns a bit-array indicating which register 5411 sets are supported 5412 (bitfields defined in arch/x86/include/uapi/asm/kvm.h). 5413 5414As described above in the kvm_sync_regs struct info in section 5 (kvm_run): 5415KVM_CAP_SYNC_REGS "allow[s] userspace to access certain guest registers 5416without having to call SET/GET_*REGS". This reduces overhead by eliminating 5417repeated ioctl calls for setting and/or getting register values. This is 5418particularly important when userspace is making synchronous guest state 5419modifications, e.g. when emulating and/or intercepting instructions in 5420userspace. 5421 5422For s390 specifics, please refer to the source code. 5423 5424For x86: 5425 5426- the register sets to be copied out to kvm_run are selectable 5427 by userspace (rather that all sets being copied out for every exit). 5428- vcpu_events are available in addition to regs and sregs. 5429 5430For x86, the 'kvm_valid_regs' field of struct kvm_run is overloaded to 5431function as an input bit-array field set by userspace to indicate the 5432specific register sets to be copied out on the next exit. 5433 5434To indicate when userspace has modified values that should be copied into 5435the vCPU, the all architecture bitarray field, 'kvm_dirty_regs' must be set. 5436This is done using the same bitflags as for the 'kvm_valid_regs' field. 5437If the dirty bit is not set, then the register set values will not be copied 5438into the vCPU even if they've been modified. 5439 5440Unused bitfields in the bitarrays must be set to zero. 5441 5442:: 5443 5444 struct kvm_sync_regs { 5445 struct kvm_regs regs; 5446 struct kvm_sregs sregs; 5447 struct kvm_vcpu_events events; 5448 }; 5449 54506.75 KVM_CAP_PPC_IRQ_XIVE 5451------------------------- 5452 5453:Architectures: ppc 5454:Target: vcpu 5455:Parameters: args[0] is the XIVE device fd; 5456 args[1] is the XIVE CPU number (server ID) for this vcpu 5457 5458This capability connects the vcpu to an in-kernel XIVE device. 5459 54607. Capabilities that can be enabled on VMs 5461========================================== 5462 5463There are certain capabilities that change the behavior of the virtual 5464machine when enabled. To enable them, please see section 4.37. Below 5465you can find a list of capabilities and what their effect on the VM 5466is when enabling them. 5467 5468The following information is provided along with the description: 5469 5470 Architectures: 5471 which instruction set architectures provide this ioctl. 5472 x86 includes both i386 and x86_64. 5473 5474 Parameters: 5475 what parameters are accepted by the capability. 5476 5477 Returns: 5478 the return value. General error numbers (EBADF, ENOMEM, EINVAL) 5479 are not detailed, but errors with specific meanings are. 5480 5481 54827.1 KVM_CAP_PPC_ENABLE_HCALL 5483---------------------------- 5484 5485:Architectures: ppc 5486:Parameters: args[0] is the sPAPR hcall number; 5487 args[1] is 0 to disable, 1 to enable in-kernel handling 5488 5489This capability controls whether individual sPAPR hypercalls (hcalls) 5490get handled by the kernel or not. Enabling or disabling in-kernel 5491handling of an hcall is effective across the VM. On creation, an 5492initial set of hcalls are enabled for in-kernel handling, which 5493consists of those hcalls for which in-kernel handlers were implemented 5494before this capability was implemented. If disabled, the kernel will 5495not to attempt to handle the hcall, but will always exit to userspace 5496to handle it. Note that it may not make sense to enable some and 5497disable others of a group of related hcalls, but KVM does not prevent 5498userspace from doing that. 5499 5500If the hcall number specified is not one that has an in-kernel 5501implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL 5502error. 5503 55047.2 KVM_CAP_S390_USER_SIGP 5505-------------------------- 5506 5507:Architectures: s390 5508:Parameters: none 5509 5510This capability controls which SIGP orders will be handled completely in user 5511space. With this capability enabled, all fast orders will be handled completely 5512in the kernel: 5513 5514- SENSE 5515- SENSE RUNNING 5516- EXTERNAL CALL 5517- EMERGENCY SIGNAL 5518- CONDITIONAL EMERGENCY SIGNAL 5519 5520All other orders will be handled completely in user space. 5521 5522Only privileged operation exceptions will be checked for in the kernel (or even 5523in the hardware prior to interception). If this capability is not enabled, the 5524old way of handling SIGP orders is used (partially in kernel and user space). 5525 55267.3 KVM_CAP_S390_VECTOR_REGISTERS 5527--------------------------------- 5528 5529:Architectures: s390 5530:Parameters: none 5531:Returns: 0 on success, negative value on error 5532 5533Allows use of the vector registers introduced with z13 processor, and 5534provides for the synchronization between host and user space. Will 5535return -EINVAL if the machine does not support vectors. 5536 55377.4 KVM_CAP_S390_USER_STSI 5538-------------------------- 5539 5540:Architectures: s390 5541:Parameters: none 5542 5543This capability allows post-handlers for the STSI instruction. After 5544initial handling in the kernel, KVM exits to user space with 5545KVM_EXIT_S390_STSI to allow user space to insert further data. 5546 5547Before exiting to userspace, kvm handlers should fill in s390_stsi field of 5548vcpu->run:: 5549 5550 struct { 5551 __u64 addr; 5552 __u8 ar; 5553 __u8 reserved; 5554 __u8 fc; 5555 __u8 sel1; 5556 __u16 sel2; 5557 } s390_stsi; 5558 5559 @addr - guest address of STSI SYSIB 5560 @fc - function code 5561 @sel1 - selector 1 5562 @sel2 - selector 2 5563 @ar - access register number 5564 5565KVM handlers should exit to userspace with rc = -EREMOTE. 5566 55677.5 KVM_CAP_SPLIT_IRQCHIP 5568------------------------- 5569 5570:Architectures: x86 5571:Parameters: args[0] - number of routes reserved for userspace IOAPICs 5572:Returns: 0 on success, -1 on error 5573 5574Create a local apic for each processor in the kernel. This can be used 5575instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the 5576IOAPIC and PIC (and also the PIT, even though this has to be enabled 5577separately). 5578 5579This capability also enables in kernel routing of interrupt requests; 5580when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are 5581used in the IRQ routing table. The first args[0] MSI routes are reserved 5582for the IOAPIC pins. Whenever the LAPIC receives an EOI for these routes, 5583a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace. 5584 5585Fails if VCPU has already been created, or if the irqchip is already in the 5586kernel (i.e. KVM_CREATE_IRQCHIP has already been called). 5587 55887.6 KVM_CAP_S390_RI 5589------------------- 5590 5591:Architectures: s390 5592:Parameters: none 5593 5594Allows use of runtime-instrumentation introduced with zEC12 processor. 5595Will return -EINVAL if the machine does not support runtime-instrumentation. 5596Will return -EBUSY if a VCPU has already been created. 5597 55987.7 KVM_CAP_X2APIC_API 5599---------------------- 5600 5601:Architectures: x86 5602:Parameters: args[0] - features that should be enabled 5603:Returns: 0 on success, -EINVAL when args[0] contains invalid features 5604 5605Valid feature flags in args[0] are:: 5606 5607 #define KVM_X2APIC_API_USE_32BIT_IDS (1ULL << 0) 5608 #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK (1ULL << 1) 5609 5610Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of 5611KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC, 5612allowing the use of 32-bit APIC IDs. See KVM_CAP_X2APIC_API in their 5613respective sections. 5614 5615KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work 5616in logical mode or with more than 255 VCPUs. Otherwise, KVM treats 0xff 5617as a broadcast even in x2APIC mode in order to support physical x2APIC 5618without interrupt remapping. This is undesirable in logical mode, 5619where 0xff represents CPUs 0-7 in cluster 0. 5620 56217.8 KVM_CAP_S390_USER_INSTR0 5622---------------------------- 5623 5624:Architectures: s390 5625:Parameters: none 5626 5627With this capability enabled, all illegal instructions 0x0000 (2 bytes) will 5628be intercepted and forwarded to user space. User space can use this 5629mechanism e.g. to realize 2-byte software breakpoints. The kernel will 5630not inject an operating exception for these instructions, user space has 5631to take care of that. 5632 5633This capability can be enabled dynamically even if VCPUs were already 5634created and are running. 5635 56367.9 KVM_CAP_S390_GS 5637------------------- 5638 5639:Architectures: s390 5640:Parameters: none 5641:Returns: 0 on success; -EINVAL if the machine does not support 5642 guarded storage; -EBUSY if a VCPU has already been created. 5643 5644Allows use of guarded storage for the KVM guest. 5645 56467.10 KVM_CAP_S390_AIS 5647--------------------- 5648 5649:Architectures: s390 5650:Parameters: none 5651 5652Allow use of adapter-interruption suppression. 5653:Returns: 0 on success; -EBUSY if a VCPU has already been created. 5654 56557.11 KVM_CAP_PPC_SMT 5656-------------------- 5657 5658:Architectures: ppc 5659:Parameters: vsmt_mode, flags 5660 5661Enabling this capability on a VM provides userspace with a way to set 5662the desired virtual SMT mode (i.e. the number of virtual CPUs per 5663virtual core). The virtual SMT mode, vsmt_mode, must be a power of 2 5664between 1 and 8. On POWER8, vsmt_mode must also be no greater than 5665the number of threads per subcore for the host. Currently flags must 5666be 0. A successful call to enable this capability will result in 5667vsmt_mode being returned when the KVM_CAP_PPC_SMT capability is 5668subsequently queried for the VM. This capability is only supported by 5669HV KVM, and can only be set before any VCPUs have been created. 5670The KVM_CAP_PPC_SMT_POSSIBLE capability indicates which virtual SMT 5671modes are available. 5672 56737.12 KVM_CAP_PPC_FWNMI 5674---------------------- 5675 5676:Architectures: ppc 5677:Parameters: none 5678 5679With this capability a machine check exception in the guest address 5680space will cause KVM to exit the guest with NMI exit reason. This 5681enables QEMU to build error log and branch to guest kernel registered 5682machine check handling routine. Without this capability KVM will 5683branch to guests' 0x200 interrupt vector. 5684 56857.13 KVM_CAP_X86_DISABLE_EXITS 5686------------------------------ 5687 5688:Architectures: x86 5689:Parameters: args[0] defines which exits are disabled 5690:Returns: 0 on success, -EINVAL when args[0] contains invalid exits 5691 5692Valid bits in args[0] are:: 5693 5694 #define KVM_X86_DISABLE_EXITS_MWAIT (1 << 0) 5695 #define KVM_X86_DISABLE_EXITS_HLT (1 << 1) 5696 #define KVM_X86_DISABLE_EXITS_PAUSE (1 << 2) 5697 #define KVM_X86_DISABLE_EXITS_CSTATE (1 << 3) 5698 5699Enabling this capability on a VM provides userspace with a way to no 5700longer intercept some instructions for improved latency in some 5701workloads, and is suggested when vCPUs are associated to dedicated 5702physical CPUs. More bits can be added in the future; userspace can 5703just pass the KVM_CHECK_EXTENSION result to KVM_ENABLE_CAP to disable 5704all such vmexits. 5705 5706Do not enable KVM_FEATURE_PV_UNHALT if you disable HLT exits. 5707 57087.14 KVM_CAP_S390_HPAGE_1M 5709-------------------------- 5710 5711:Architectures: s390 5712:Parameters: none 5713:Returns: 0 on success, -EINVAL if hpage module parameter was not set 5714 or cmma is enabled, or the VM has the KVM_VM_S390_UCONTROL 5715 flag set 5716 5717With this capability the KVM support for memory backing with 1m pages 5718through hugetlbfs can be enabled for a VM. After the capability is 5719enabled, cmma can't be enabled anymore and pfmfi and the storage key 5720interpretation are disabled. If cmma has already been enabled or the 5721hpage module parameter is not set to 1, -EINVAL is returned. 5722 5723While it is generally possible to create a huge page backed VM without 5724this capability, the VM will not be able to run. 5725 57267.15 KVM_CAP_MSR_PLATFORM_INFO 5727------------------------------ 5728 5729:Architectures: x86 5730:Parameters: args[0] whether feature should be enabled or not 5731 5732With this capability, a guest may read the MSR_PLATFORM_INFO MSR. Otherwise, 5733a #GP would be raised when the guest tries to access. Currently, this 5734capability does not enable write permissions of this MSR for the guest. 5735 57367.16 KVM_CAP_PPC_NESTED_HV 5737-------------------------- 5738 5739:Architectures: ppc 5740:Parameters: none 5741:Returns: 0 on success, -EINVAL when the implementation doesn't support 5742 nested-HV virtualization. 5743 5744HV-KVM on POWER9 and later systems allows for "nested-HV" 5745virtualization, which provides a way for a guest VM to run guests that 5746can run using the CPU's supervisor mode (privileged non-hypervisor 5747state). Enabling this capability on a VM depends on the CPU having 5748the necessary functionality and on the facility being enabled with a 5749kvm-hv module parameter. 5750 57517.17 KVM_CAP_EXCEPTION_PAYLOAD 5752------------------------------ 5753 5754:Architectures: x86 5755:Parameters: args[0] whether feature should be enabled or not 5756 5757With this capability enabled, CR2 will not be modified prior to the 5758emulated VM-exit when L1 intercepts a #PF exception that occurs in 5759L2. Similarly, for kvm-intel only, DR6 will not be modified prior to 5760the emulated VM-exit when L1 intercepts a #DB exception that occurs in 5761L2. As a result, when KVM_GET_VCPU_EVENTS reports a pending #PF (or 5762#DB) exception for L2, exception.has_payload will be set and the 5763faulting address (or the new DR6 bits*) will be reported in the 5764exception_payload field. Similarly, when userspace injects a #PF (or 5765#DB) into L2 using KVM_SET_VCPU_EVENTS, it is expected to set 5766exception.has_payload and to put the faulting address - or the new DR6 5767bits\ [#]_ - in the exception_payload field. 5768 5769This capability also enables exception.pending in struct 5770kvm_vcpu_events, which allows userspace to distinguish between pending 5771and injected exceptions. 5772 5773 5774.. [#] For the new DR6 bits, note that bit 16 is set iff the #DB exception 5775 will clear DR6.RTM. 5776 57777.18 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 5778 5779:Architectures: x86, arm, arm64, mips 5780:Parameters: args[0] whether feature should be enabled or not 5781 5782Valid flags are:: 5783 5784 #define KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (1 << 0) 5785 #define KVM_DIRTY_LOG_INITIALLY_SET (1 << 1) 5786 5787With KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE is set, KVM_GET_DIRTY_LOG will not 5788automatically clear and write-protect all pages that are returned as dirty. 5789Rather, userspace will have to do this operation separately using 5790KVM_CLEAR_DIRTY_LOG. 5791 5792At the cost of a slightly more complicated operation, this provides better 5793scalability and responsiveness for two reasons. First, 5794KVM_CLEAR_DIRTY_LOG ioctl can operate on a 64-page granularity rather 5795than requiring to sync a full memslot; this ensures that KVM does not 5796take spinlocks for an extended period of time. Second, in some cases a 5797large amount of time can pass between a call to KVM_GET_DIRTY_LOG and 5798userspace actually using the data in the page. Pages can be modified 5799during this time, which is inefficient for both the guest and userspace: 5800the guest will incur a higher penalty due to write protection faults, 5801while userspace can see false reports of dirty pages. Manual reprotection 5802helps reducing this time, improving guest performance and reducing the 5803number of dirty log false positives. 5804 5805With KVM_DIRTY_LOG_INITIALLY_SET set, all the bits of the dirty bitmap 5806will be initialized to 1 when created. This also improves performance because 5807dirty logging can be enabled gradually in small chunks on the first call 5808to KVM_CLEAR_DIRTY_LOG. KVM_DIRTY_LOG_INITIALLY_SET depends on 5809KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (it is also only available on 5810x86 and arm64 for now). 5811 5812KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 was previously available under the name 5813KVM_CAP_MANUAL_DIRTY_LOG_PROTECT, but the implementation had bugs that make 5814it hard or impossible to use it correctly. The availability of 5815KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 signals that those bugs are fixed. 5816Userspace should not try to use KVM_CAP_MANUAL_DIRTY_LOG_PROTECT. 5817 58187.19 KVM_CAP_PPC_SECURE_GUEST 5819------------------------------ 5820 5821:Architectures: ppc 5822 5823This capability indicates that KVM is running on a host that has 5824ultravisor firmware and thus can support a secure guest. On such a 5825system, a guest can ask the ultravisor to make it a secure guest, 5826one whose memory is inaccessible to the host except for pages which 5827are explicitly requested to be shared with the host. The ultravisor 5828notifies KVM when a guest requests to become a secure guest, and KVM 5829has the opportunity to veto the transition. 5830 5831If present, this capability can be enabled for a VM, meaning that KVM 5832will allow the transition to secure guest mode. Otherwise KVM will 5833veto the transition. 5834 58357.20 KVM_CAP_HALT_POLL 5836---------------------- 5837 5838:Architectures: all 5839:Target: VM 5840:Parameters: args[0] is the maximum poll time in nanoseconds 5841:Returns: 0 on success; -1 on error 5842 5843This capability overrides the kvm module parameter halt_poll_ns for the 5844target VM. 5845 5846VCPU polling allows a VCPU to poll for wakeup events instead of immediately 5847scheduling during guest halts. The maximum time a VCPU can spend polling is 5848controlled by the kvm module parameter halt_poll_ns. This capability allows 5849the maximum halt time to specified on a per-VM basis, effectively overriding 5850the module parameter for the target VM. 5851 58528. Other capabilities. 5853====================== 5854 5855This section lists capabilities that give information about other 5856features of the KVM implementation. 5857 58588.1 KVM_CAP_PPC_HWRNG 5859--------------------- 5860 5861:Architectures: ppc 5862 5863This capability, if KVM_CHECK_EXTENSION indicates that it is 5864available, means that the kernel has an implementation of the 5865H_RANDOM hypercall backed by a hardware random-number generator. 5866If present, the kernel H_RANDOM handler can be enabled for guest use 5867with the KVM_CAP_PPC_ENABLE_HCALL capability. 5868 58698.2 KVM_CAP_HYPERV_SYNIC 5870------------------------ 5871 5872:Architectures: x86 5873 5874This capability, if KVM_CHECK_EXTENSION indicates that it is 5875available, means that the kernel has an implementation of the 5876Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is 5877used to support Windows Hyper-V based guest paravirt drivers(VMBus). 5878 5879In order to use SynIC, it has to be activated by setting this 5880capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this 5881will disable the use of APIC hardware virtualization even if supported 5882by the CPU, as it's incompatible with SynIC auto-EOI behavior. 5883 58848.3 KVM_CAP_PPC_RADIX_MMU 5885------------------------- 5886 5887:Architectures: ppc 5888 5889This capability, if KVM_CHECK_EXTENSION indicates that it is 5890available, means that the kernel can support guests using the 5891radix MMU defined in Power ISA V3.00 (as implemented in the POWER9 5892processor). 5893 58948.4 KVM_CAP_PPC_HASH_MMU_V3 5895--------------------------- 5896 5897:Architectures: ppc 5898 5899This capability, if KVM_CHECK_EXTENSION indicates that it is 5900available, means that the kernel can support guests using the 5901hashed page table MMU defined in Power ISA V3.00 (as implemented in 5902the POWER9 processor), including in-memory segment tables. 5903 59048.5 KVM_CAP_MIPS_VZ 5905------------------- 5906 5907:Architectures: mips 5908 5909This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that 5910it is available, means that full hardware assisted virtualization capabilities 5911of the hardware are available for use through KVM. An appropriate 5912KVM_VM_MIPS_* type must be passed to KVM_CREATE_VM to create a VM which 5913utilises it. 5914 5915If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is 5916available, it means that the VM is using full hardware assisted virtualization 5917capabilities of the hardware. This is useful to check after creating a VM with 5918KVM_VM_MIPS_DEFAULT. 5919 5920The value returned by KVM_CHECK_EXTENSION should be compared against known 5921values (see below). All other values are reserved. This is to allow for the 5922possibility of other hardware assisted virtualization implementations which 5923may be incompatible with the MIPS VZ ASE. 5924 5925== ========================================================================== 5926 0 The trap & emulate implementation is in use to run guest code in user 5927 mode. Guest virtual memory segments are rearranged to fit the guest in the 5928 user mode address space. 5929 5930 1 The MIPS VZ ASE is in use, providing full hardware assisted 5931 virtualization, including standard guest virtual memory segments. 5932== ========================================================================== 5933 59348.6 KVM_CAP_MIPS_TE 5935------------------- 5936 5937:Architectures: mips 5938 5939This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that 5940it is available, means that the trap & emulate implementation is available to 5941run guest code in user mode, even if KVM_CAP_MIPS_VZ indicates that hardware 5942assisted virtualisation is also available. KVM_VM_MIPS_TE (0) must be passed 5943to KVM_CREATE_VM to create a VM which utilises it. 5944 5945If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is 5946available, it means that the VM is using trap & emulate. 5947 59488.7 KVM_CAP_MIPS_64BIT 5949---------------------- 5950 5951:Architectures: mips 5952 5953This capability indicates the supported architecture type of the guest, i.e. the 5954supported register and address width. 5955 5956The values returned when this capability is checked by KVM_CHECK_EXTENSION on a 5957kvm VM handle correspond roughly to the CP0_Config.AT register field, and should 5958be checked specifically against known values (see below). All other values are 5959reserved. 5960 5961== ======================================================================== 5962 0 MIPS32 or microMIPS32. 5963 Both registers and addresses are 32-bits wide. 5964 It will only be possible to run 32-bit guest code. 5965 5966 1 MIPS64 or microMIPS64 with access only to 32-bit compatibility segments. 5967 Registers are 64-bits wide, but addresses are 32-bits wide. 5968 64-bit guest code may run but cannot access MIPS64 memory segments. 5969 It will also be possible to run 32-bit guest code. 5970 5971 2 MIPS64 or microMIPS64 with access to all address segments. 5972 Both registers and addresses are 64-bits wide. 5973 It will be possible to run 64-bit or 32-bit guest code. 5974== ======================================================================== 5975 59768.9 KVM_CAP_ARM_USER_IRQ 5977------------------------ 5978 5979:Architectures: arm, arm64 5980 5981This capability, if KVM_CHECK_EXTENSION indicates that it is available, means 5982that if userspace creates a VM without an in-kernel interrupt controller, it 5983will be notified of changes to the output level of in-kernel emulated devices, 5984which can generate virtual interrupts, presented to the VM. 5985For such VMs, on every return to userspace, the kernel 5986updates the vcpu's run->s.regs.device_irq_level field to represent the actual 5987output level of the device. 5988 5989Whenever kvm detects a change in the device output level, kvm guarantees at 5990least one return to userspace before running the VM. This exit could either 5991be a KVM_EXIT_INTR or any other exit event, like KVM_EXIT_MMIO. This way, 5992userspace can always sample the device output level and re-compute the state of 5993the userspace interrupt controller. Userspace should always check the state 5994of run->s.regs.device_irq_level on every kvm exit. 5995The value in run->s.regs.device_irq_level can represent both level and edge 5996triggered interrupt signals, depending on the device. Edge triggered interrupt 5997signals will exit to userspace with the bit in run->s.regs.device_irq_level 5998set exactly once per edge signal. 5999 6000The field run->s.regs.device_irq_level is available independent of 6001run->kvm_valid_regs or run->kvm_dirty_regs bits. 6002 6003If KVM_CAP_ARM_USER_IRQ is supported, the KVM_CHECK_EXTENSION ioctl returns a 6004number larger than 0 indicating the version of this capability is implemented 6005and thereby which bits in run->s.regs.device_irq_level can signal values. 6006 6007Currently the following bits are defined for the device_irq_level bitmap:: 6008 6009 KVM_CAP_ARM_USER_IRQ >= 1: 6010 6011 KVM_ARM_DEV_EL1_VTIMER - EL1 virtual timer 6012 KVM_ARM_DEV_EL1_PTIMER - EL1 physical timer 6013 KVM_ARM_DEV_PMU - ARM PMU overflow interrupt signal 6014 6015Future versions of kvm may implement additional events. These will get 6016indicated by returning a higher number from KVM_CHECK_EXTENSION and will be 6017listed above. 6018 60198.10 KVM_CAP_PPC_SMT_POSSIBLE 6020----------------------------- 6021 6022:Architectures: ppc 6023 6024Querying this capability returns a bitmap indicating the possible 6025virtual SMT modes that can be set using KVM_CAP_PPC_SMT. If bit N 6026(counting from the right) is set, then a virtual SMT mode of 2^N is 6027available. 6028 60298.11 KVM_CAP_HYPERV_SYNIC2 6030-------------------------- 6031 6032:Architectures: x86 6033 6034This capability enables a newer version of Hyper-V Synthetic interrupt 6035controller (SynIC). The only difference with KVM_CAP_HYPERV_SYNIC is that KVM 6036doesn't clear SynIC message and event flags pages when they are enabled by 6037writing to the respective MSRs. 6038 60398.12 KVM_CAP_HYPERV_VP_INDEX 6040---------------------------- 6041 6042:Architectures: x86 6043 6044This capability indicates that userspace can load HV_X64_MSR_VP_INDEX msr. Its 6045value is used to denote the target vcpu for a SynIC interrupt. For 6046compatibilty, KVM initializes this msr to KVM's internal vcpu index. When this 6047capability is absent, userspace can still query this msr's value. 6048 60498.13 KVM_CAP_S390_AIS_MIGRATION 6050------------------------------- 6051 6052:Architectures: s390 6053:Parameters: none 6054 6055This capability indicates if the flic device will be able to get/set the 6056AIS states for migration via the KVM_DEV_FLIC_AISM_ALL attribute and allows 6057to discover this without having to create a flic device. 6058 60598.14 KVM_CAP_S390_PSW 6060--------------------- 6061 6062:Architectures: s390 6063 6064This capability indicates that the PSW is exposed via the kvm_run structure. 6065 60668.15 KVM_CAP_S390_GMAP 6067---------------------- 6068 6069:Architectures: s390 6070 6071This capability indicates that the user space memory used as guest mapping can 6072be anywhere in the user memory address space, as long as the memory slots are 6073aligned and sized to a segment (1MB) boundary. 6074 60758.16 KVM_CAP_S390_COW 6076--------------------- 6077 6078:Architectures: s390 6079 6080This capability indicates that the user space memory used as guest mapping can 6081use copy-on-write semantics as well as dirty pages tracking via read-only page 6082tables. 6083 60848.17 KVM_CAP_S390_BPB 6085--------------------- 6086 6087:Architectures: s390 6088 6089This capability indicates that kvm will implement the interfaces to handle 6090reset, migration and nested KVM for branch prediction blocking. The stfle 6091facility 82 should not be provided to the guest without this capability. 6092 60938.18 KVM_CAP_HYPERV_TLBFLUSH 6094---------------------------- 6095 6096:Architectures: x86 6097 6098This capability indicates that KVM supports paravirtualized Hyper-V TLB Flush 6099hypercalls: 6100HvFlushVirtualAddressSpace, HvFlushVirtualAddressSpaceEx, 6101HvFlushVirtualAddressList, HvFlushVirtualAddressListEx. 6102 61038.19 KVM_CAP_ARM_INJECT_SERROR_ESR 6104---------------------------------- 6105 6106:Architectures: arm, arm64 6107 6108This capability indicates that userspace can specify (via the 6109KVM_SET_VCPU_EVENTS ioctl) the syndrome value reported to the guest when it 6110takes a virtual SError interrupt exception. 6111If KVM advertises this capability, userspace can only specify the ISS field for 6112the ESR syndrome. Other parts of the ESR, such as the EC are generated by the 6113CPU when the exception is taken. If this virtual SError is taken to EL1 using 6114AArch64, this value will be reported in the ISS field of ESR_ELx. 6115 6116See KVM_CAP_VCPU_EVENTS for more details. 6117 61188.20 KVM_CAP_HYPERV_SEND_IPI 6119---------------------------- 6120 6121:Architectures: x86 6122 6123This capability indicates that KVM supports paravirtualized Hyper-V IPI send 6124hypercalls: 6125HvCallSendSyntheticClusterIpi, HvCallSendSyntheticClusterIpiEx. 6126 61278.21 KVM_CAP_HYPERV_DIRECT_TLBFLUSH 6128----------------------------------- 6129 6130:Architecture: x86 6131 6132This capability indicates that KVM running on top of Hyper-V hypervisor 6133enables Direct TLB flush for its guests meaning that TLB flush 6134hypercalls are handled by Level 0 hypervisor (Hyper-V) bypassing KVM. 6135Due to the different ABI for hypercall parameters between Hyper-V and 6136KVM, enabling this capability effectively disables all hypercall 6137handling by KVM (as some KVM hypercall may be mistakenly treated as TLB 6138flush hypercalls by Hyper-V) so userspace should disable KVM identification 6139in CPUID and only exposes Hyper-V identification. In this case, guest 6140thinks it's running on Hyper-V and only use Hyper-V hypercalls. 6141 61428.22 KVM_CAP_S390_VCPU_RESETS 6143 6144Architectures: s390 6145 6146This capability indicates that the KVM_S390_NORMAL_RESET and 6147KVM_S390_CLEAR_RESET ioctls are available. 6148 61498.23 KVM_CAP_S390_PROTECTED 6150 6151Architecture: s390 6152 6153 6154This capability indicates that the Ultravisor has been initialized and 6155KVM can therefore start protected VMs. 6156This capability governs the KVM_S390_PV_COMMAND ioctl and the 6157KVM_MP_STATE_LOAD MP_STATE. KVM_SET_MP_STATE can fail for protected 6158guests when the state change is invalid. 6159