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 185Creation of the VM will fail if the requested IPA size (whether it is 186implicit or explicit) is unsupported on the host. 187 188Please note that configuring the IPA size does not affect the capability 189exposed by the guest CPUs in ID_AA64MMFR0_EL1[PARange]. It only affects 190size of the address translated by the stage2 level (guest physical to 191host physical address translations). 192 193 1944.3 KVM_GET_MSR_INDEX_LIST, KVM_GET_MSR_FEATURE_INDEX_LIST 195---------------------------------------------------------- 196 197:Capability: basic, KVM_CAP_GET_MSR_FEATURES for KVM_GET_MSR_FEATURE_INDEX_LIST 198:Architectures: x86 199:Type: system ioctl 200:Parameters: struct kvm_msr_list (in/out) 201:Returns: 0 on success; -1 on error 202 203Errors: 204 205 ====== ============================================================ 206 EFAULT the msr index list cannot be read from or written to 207 E2BIG the msr index list is to be to fit in the array specified by 208 the user. 209 ====== ============================================================ 210 211:: 212 213 struct kvm_msr_list { 214 __u32 nmsrs; /* number of msrs in entries */ 215 __u32 indices[0]; 216 }; 217 218The user fills in the size of the indices array in nmsrs, and in return 219kvm adjusts nmsrs to reflect the actual number of msrs and fills in the 220indices array with their numbers. 221 222KVM_GET_MSR_INDEX_LIST returns the guest msrs that are supported. The list 223varies by kvm version and host processor, but does not change otherwise. 224 225Note: if kvm indicates supports MCE (KVM_CAP_MCE), then the MCE bank MSRs are 226not returned in the MSR list, as different vcpus can have a different number 227of banks, as set via the KVM_X86_SETUP_MCE ioctl. 228 229KVM_GET_MSR_FEATURE_INDEX_LIST returns the list of MSRs that can be passed 230to the KVM_GET_MSRS system ioctl. This lets userspace probe host capabilities 231and processor features that are exposed via MSRs (e.g., VMX capabilities). 232This list also varies by kvm version and host processor, but does not change 233otherwise. 234 235 2364.4 KVM_CHECK_EXTENSION 237----------------------- 238 239:Capability: basic, KVM_CAP_CHECK_EXTENSION_VM for vm ioctl 240:Architectures: all 241:Type: system ioctl, vm ioctl 242:Parameters: extension identifier (KVM_CAP_*) 243:Returns: 0 if unsupported; 1 (or some other positive integer) if supported 244 245The API allows the application to query about extensions to the core 246kvm API. Userspace passes an extension identifier (an integer) and 247receives an integer that describes the extension availability. 248Generally 0 means no and 1 means yes, but some extensions may report 249additional information in the integer return value. 250 251Based on their initialization different VMs may have different capabilities. 252It is thus encouraged to use the vm ioctl to query for capabilities (available 253with KVM_CAP_CHECK_EXTENSION_VM on the vm fd) 254 2554.5 KVM_GET_VCPU_MMAP_SIZE 256-------------------------- 257 258:Capability: basic 259:Architectures: all 260:Type: system ioctl 261:Parameters: none 262:Returns: size of vcpu mmap area, in bytes 263 264The KVM_RUN ioctl (cf.) communicates with userspace via a shared 265memory region. This ioctl returns the size of that region. See the 266KVM_RUN documentation for details. 267 268 2694.6 KVM_SET_MEMORY_REGION 270------------------------- 271 272:Capability: basic 273:Architectures: all 274:Type: vm ioctl 275:Parameters: struct kvm_memory_region (in) 276:Returns: 0 on success, -1 on error 277 278This ioctl is obsolete and has been removed. 279 280 2814.7 KVM_CREATE_VCPU 282------------------- 283 284:Capability: basic 285:Architectures: all 286:Type: vm ioctl 287:Parameters: vcpu id (apic id on x86) 288:Returns: vcpu fd on success, -1 on error 289 290This API adds a vcpu to a virtual machine. No more than max_vcpus may be added. 291The vcpu id is an integer in the range [0, max_vcpu_id). 292 293The recommended max_vcpus value can be retrieved using the KVM_CAP_NR_VCPUS of 294the KVM_CHECK_EXTENSION ioctl() at run-time. 295The maximum possible value for max_vcpus can be retrieved using the 296KVM_CAP_MAX_VCPUS of the KVM_CHECK_EXTENSION ioctl() at run-time. 297 298If the KVM_CAP_NR_VCPUS does not exist, you should assume that max_vcpus is 4 299cpus max. 300If the KVM_CAP_MAX_VCPUS does not exist, you should assume that max_vcpus is 301same as the value returned from KVM_CAP_NR_VCPUS. 302 303The maximum possible value for max_vcpu_id can be retrieved using the 304KVM_CAP_MAX_VCPU_ID of the KVM_CHECK_EXTENSION ioctl() at run-time. 305 306If the KVM_CAP_MAX_VCPU_ID does not exist, you should assume that max_vcpu_id 307is the same as the value returned from KVM_CAP_MAX_VCPUS. 308 309On powerpc using book3s_hv mode, the vcpus are mapped onto virtual 310threads in one or more virtual CPU cores. (This is because the 311hardware requires all the hardware threads in a CPU core to be in the 312same partition.) The KVM_CAP_PPC_SMT capability indicates the number 313of vcpus per virtual core (vcore). The vcore id is obtained by 314dividing the vcpu id by the number of vcpus per vcore. The vcpus in a 315given vcore will always be in the same physical core as each other 316(though that might be a different physical core from time to time). 317Userspace can control the threading (SMT) mode of the guest by its 318allocation of vcpu ids. For example, if userspace wants 319single-threaded guest vcpus, it should make all vcpu ids be a multiple 320of the number of vcpus per vcore. 321 322For virtual cpus that have been created with S390 user controlled virtual 323machines, the resulting vcpu fd can be memory mapped at page offset 324KVM_S390_SIE_PAGE_OFFSET in order to obtain a memory map of the virtual 325cpu's hardware control block. 326 327 3284.8 KVM_GET_DIRTY_LOG (vm ioctl) 329-------------------------------- 330 331:Capability: basic 332:Architectures: all 333:Type: vm ioctl 334:Parameters: struct kvm_dirty_log (in/out) 335:Returns: 0 on success, -1 on error 336 337:: 338 339 /* for KVM_GET_DIRTY_LOG */ 340 struct kvm_dirty_log { 341 __u32 slot; 342 __u32 padding; 343 union { 344 void __user *dirty_bitmap; /* one bit per page */ 345 __u64 padding; 346 }; 347 }; 348 349Given a memory slot, return a bitmap containing any pages dirtied 350since the last call to this ioctl. Bit 0 is the first page in the 351memory slot. Ensure the entire structure is cleared to avoid padding 352issues. 353 354If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 specifies 355the address space for which you want to return the dirty bitmap. 356They must be less than the value that KVM_CHECK_EXTENSION returns for 357the KVM_CAP_MULTI_ADDRESS_SPACE capability. 358 359The bits in the dirty bitmap are cleared before the ioctl returns, unless 360KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is enabled. For more information, 361see the description of the capability. 362 3634.9 KVM_SET_MEMORY_ALIAS 364------------------------ 365 366:Capability: basic 367:Architectures: x86 368:Type: vm ioctl 369:Parameters: struct kvm_memory_alias (in) 370:Returns: 0 (success), -1 (error) 371 372This ioctl is obsolete and has been removed. 373 374 3754.10 KVM_RUN 376------------ 377 378:Capability: basic 379:Architectures: all 380:Type: vcpu ioctl 381:Parameters: none 382:Returns: 0 on success, -1 on error 383 384Errors: 385 386 ======= ============================================================== 387 EINTR an unmasked signal is pending 388 ENOEXEC the vcpu hasn't been initialized or the guest tried to execute 389 instructions from device memory (arm64) 390 ENOSYS data abort outside memslots with no syndrome info and 391 KVM_CAP_ARM_NISV_TO_USER not enabled (arm64) 392 EPERM SVE feature set but not finalized (arm64) 393 ======= ============================================================== 394 395This ioctl is used to run a guest virtual cpu. While there are no 396explicit parameters, there is an implicit parameter block that can be 397obtained by mmap()ing the vcpu fd at offset 0, with the size given by 398KVM_GET_VCPU_MMAP_SIZE. The parameter block is formatted as a 'struct 399kvm_run' (see below). 400 401 4024.11 KVM_GET_REGS 403----------------- 404 405:Capability: basic 406:Architectures: all except ARM, arm64 407:Type: vcpu ioctl 408:Parameters: struct kvm_regs (out) 409:Returns: 0 on success, -1 on error 410 411Reads the general purpose registers from the vcpu. 412 413:: 414 415 /* x86 */ 416 struct kvm_regs { 417 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */ 418 __u64 rax, rbx, rcx, rdx; 419 __u64 rsi, rdi, rsp, rbp; 420 __u64 r8, r9, r10, r11; 421 __u64 r12, r13, r14, r15; 422 __u64 rip, rflags; 423 }; 424 425 /* mips */ 426 struct kvm_regs { 427 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */ 428 __u64 gpr[32]; 429 __u64 hi; 430 __u64 lo; 431 __u64 pc; 432 }; 433 434 4354.12 KVM_SET_REGS 436----------------- 437 438:Capability: basic 439:Architectures: all except ARM, arm64 440:Type: vcpu ioctl 441:Parameters: struct kvm_regs (in) 442:Returns: 0 on success, -1 on error 443 444Writes the general purpose registers into the vcpu. 445 446See KVM_GET_REGS for the data structure. 447 448 4494.13 KVM_GET_SREGS 450------------------ 451 452:Capability: basic 453:Architectures: x86, ppc 454:Type: vcpu ioctl 455:Parameters: struct kvm_sregs (out) 456:Returns: 0 on success, -1 on error 457 458Reads special registers from the vcpu. 459 460:: 461 462 /* x86 */ 463 struct kvm_sregs { 464 struct kvm_segment cs, ds, es, fs, gs, ss; 465 struct kvm_segment tr, ldt; 466 struct kvm_dtable gdt, idt; 467 __u64 cr0, cr2, cr3, cr4, cr8; 468 __u64 efer; 469 __u64 apic_base; 470 __u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64]; 471 }; 472 473 /* ppc -- see arch/powerpc/include/uapi/asm/kvm.h */ 474 475interrupt_bitmap is a bitmap of pending external interrupts. At most 476one bit may be set. This interrupt has been acknowledged by the APIC 477but not yet injected into the cpu core. 478 479 4804.14 KVM_SET_SREGS 481------------------ 482 483:Capability: basic 484:Architectures: x86, ppc 485:Type: vcpu ioctl 486:Parameters: struct kvm_sregs (in) 487:Returns: 0 on success, -1 on error 488 489Writes special registers into the vcpu. See KVM_GET_SREGS for the 490data structures. 491 492 4934.15 KVM_TRANSLATE 494------------------ 495 496:Capability: basic 497:Architectures: x86 498:Type: vcpu ioctl 499:Parameters: struct kvm_translation (in/out) 500:Returns: 0 on success, -1 on error 501 502Translates a virtual address according to the vcpu's current address 503translation mode. 504 505:: 506 507 struct kvm_translation { 508 /* in */ 509 __u64 linear_address; 510 511 /* out */ 512 __u64 physical_address; 513 __u8 valid; 514 __u8 writeable; 515 __u8 usermode; 516 __u8 pad[5]; 517 }; 518 519 5204.16 KVM_INTERRUPT 521------------------ 522 523:Capability: basic 524:Architectures: x86, ppc, mips 525:Type: vcpu ioctl 526:Parameters: struct kvm_interrupt (in) 527:Returns: 0 on success, negative on failure. 528 529Queues a hardware interrupt vector to be injected. 530 531:: 532 533 /* for KVM_INTERRUPT */ 534 struct kvm_interrupt { 535 /* in */ 536 __u32 irq; 537 }; 538 539X86: 540^^^^ 541 542:Returns: 543 544 ========= =================================== 545 0 on success, 546 -EEXIST if an interrupt is already enqueued 547 -EINVAL the irq number is invalid 548 -ENXIO if the PIC is in the kernel 549 -EFAULT if the pointer is invalid 550 ========= =================================== 551 552Note 'irq' is an interrupt vector, not an interrupt pin or line. This 553ioctl is useful if the in-kernel PIC is not used. 554 555PPC: 556^^^^ 557 558Queues an external interrupt to be injected. This ioctl is overleaded 559with 3 different irq values: 560 561a) KVM_INTERRUPT_SET 562 563 This injects an edge type external interrupt into the guest once it's ready 564 to receive interrupts. When injected, the interrupt is done. 565 566b) KVM_INTERRUPT_UNSET 567 568 This unsets any pending interrupt. 569 570 Only available with KVM_CAP_PPC_UNSET_IRQ. 571 572c) KVM_INTERRUPT_SET_LEVEL 573 574 This injects a level type external interrupt into the guest context. The 575 interrupt stays pending until a specific ioctl with KVM_INTERRUPT_UNSET 576 is triggered. 577 578 Only available with KVM_CAP_PPC_IRQ_LEVEL. 579 580Note that any value for 'irq' other than the ones stated above is invalid 581and incurs unexpected behavior. 582 583This is an asynchronous vcpu ioctl and can be invoked from any thread. 584 585MIPS: 586^^^^^ 587 588Queues an external interrupt to be injected into the virtual CPU. A negative 589interrupt number dequeues the interrupt. 590 591This is an asynchronous vcpu ioctl and can be invoked from any thread. 592 593 5944.17 KVM_DEBUG_GUEST 595-------------------- 596 597:Capability: basic 598:Architectures: none 599:Type: vcpu ioctl 600:Parameters: none) 601:Returns: -1 on error 602 603Support for this has been removed. Use KVM_SET_GUEST_DEBUG instead. 604 605 6064.18 KVM_GET_MSRS 607----------------- 608 609:Capability: basic (vcpu), KVM_CAP_GET_MSR_FEATURES (system) 610:Architectures: x86 611:Type: system ioctl, vcpu ioctl 612:Parameters: struct kvm_msrs (in/out) 613:Returns: number of msrs successfully returned; 614 -1 on error 615 616When used as a system ioctl: 617Reads the values of MSR-based features that are available for the VM. This 618is similar to KVM_GET_SUPPORTED_CPUID, but it returns MSR indices and values. 619The list of msr-based features can be obtained using KVM_GET_MSR_FEATURE_INDEX_LIST 620in a system ioctl. 621 622When used as a vcpu ioctl: 623Reads model-specific registers from the vcpu. Supported msr indices can 624be obtained using KVM_GET_MSR_INDEX_LIST in a system ioctl. 625 626:: 627 628 struct kvm_msrs { 629 __u32 nmsrs; /* number of msrs in entries */ 630 __u32 pad; 631 632 struct kvm_msr_entry entries[0]; 633 }; 634 635 struct kvm_msr_entry { 636 __u32 index; 637 __u32 reserved; 638 __u64 data; 639 }; 640 641Application code should set the 'nmsrs' member (which indicates the 642size of the entries array) and the 'index' member of each array entry. 643kvm will fill in the 'data' member. 644 645 6464.19 KVM_SET_MSRS 647----------------- 648 649:Capability: basic 650:Architectures: x86 651:Type: vcpu ioctl 652:Parameters: struct kvm_msrs (in) 653:Returns: number of msrs successfully set (see below), -1 on error 654 655Writes model-specific registers to the vcpu. See KVM_GET_MSRS for the 656data structures. 657 658Application code should set the 'nmsrs' member (which indicates the 659size of the entries array), and the 'index' and 'data' members of each 660array entry. 661 662It tries to set the MSRs in array entries[] one by one. If setting an MSR 663fails, e.g., due to setting reserved bits, the MSR isn't supported/emulated 664by KVM, etc..., it stops processing the MSR list and returns the number of 665MSRs that have been set successfully. 666 667 6684.20 KVM_SET_CPUID 669------------------ 670 671:Capability: basic 672:Architectures: x86 673:Type: vcpu ioctl 674:Parameters: struct kvm_cpuid (in) 675:Returns: 0 on success, -1 on error 676 677Defines the vcpu responses to the cpuid instruction. Applications 678should use the KVM_SET_CPUID2 ioctl if available. 679 680Note, when this IOCTL fails, KVM gives no guarantees that previous valid CPUID 681configuration (if there is) is not corrupted. Userspace can get a copy of the 682resulting CPUID configuration through KVM_GET_CPUID2 in case. 683 684:: 685 686 struct kvm_cpuid_entry { 687 __u32 function; 688 __u32 eax; 689 __u32 ebx; 690 __u32 ecx; 691 __u32 edx; 692 __u32 padding; 693 }; 694 695 /* for KVM_SET_CPUID */ 696 struct kvm_cpuid { 697 __u32 nent; 698 __u32 padding; 699 struct kvm_cpuid_entry entries[0]; 700 }; 701 702 7034.21 KVM_SET_SIGNAL_MASK 704------------------------ 705 706:Capability: basic 707:Architectures: all 708:Type: vcpu ioctl 709:Parameters: struct kvm_signal_mask (in) 710:Returns: 0 on success, -1 on error 711 712Defines which signals are blocked during execution of KVM_RUN. This 713signal mask temporarily overrides the threads signal mask. Any 714unblocked signal received (except SIGKILL and SIGSTOP, which retain 715their traditional behaviour) will cause KVM_RUN to return with -EINTR. 716 717Note the signal will only be delivered if not blocked by the original 718signal mask. 719 720:: 721 722 /* for KVM_SET_SIGNAL_MASK */ 723 struct kvm_signal_mask { 724 __u32 len; 725 __u8 sigset[0]; 726 }; 727 728 7294.22 KVM_GET_FPU 730---------------- 731 732:Capability: basic 733:Architectures: x86 734:Type: vcpu ioctl 735:Parameters: struct kvm_fpu (out) 736:Returns: 0 on success, -1 on error 737 738Reads the floating point state from the vcpu. 739 740:: 741 742 /* for KVM_GET_FPU and KVM_SET_FPU */ 743 struct kvm_fpu { 744 __u8 fpr[8][16]; 745 __u16 fcw; 746 __u16 fsw; 747 __u8 ftwx; /* in fxsave format */ 748 __u8 pad1; 749 __u16 last_opcode; 750 __u64 last_ip; 751 __u64 last_dp; 752 __u8 xmm[16][16]; 753 __u32 mxcsr; 754 __u32 pad2; 755 }; 756 757 7584.23 KVM_SET_FPU 759---------------- 760 761:Capability: basic 762:Architectures: x86 763:Type: vcpu ioctl 764:Parameters: struct kvm_fpu (in) 765:Returns: 0 on success, -1 on error 766 767Writes the floating point state to the vcpu. 768 769:: 770 771 /* for KVM_GET_FPU and KVM_SET_FPU */ 772 struct kvm_fpu { 773 __u8 fpr[8][16]; 774 __u16 fcw; 775 __u16 fsw; 776 __u8 ftwx; /* in fxsave format */ 777 __u8 pad1; 778 __u16 last_opcode; 779 __u64 last_ip; 780 __u64 last_dp; 781 __u8 xmm[16][16]; 782 __u32 mxcsr; 783 __u32 pad2; 784 }; 785 786 7874.24 KVM_CREATE_IRQCHIP 788----------------------- 789 790:Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQCHIP (s390) 791:Architectures: x86, ARM, arm64, s390 792:Type: vm ioctl 793:Parameters: none 794:Returns: 0 on success, -1 on error 795 796Creates an interrupt controller model in the kernel. 797On x86, creates a virtual ioapic, a virtual PIC (two PICs, nested), and sets up 798future vcpus to have a local APIC. IRQ routing for GSIs 0-15 is set to both 799PIC and IOAPIC; GSI 16-23 only go to the IOAPIC. 800On ARM/arm64, a GICv2 is created. Any other GIC versions require the usage of 801KVM_CREATE_DEVICE, which also supports creating a GICv2. Using 802KVM_CREATE_DEVICE is preferred over KVM_CREATE_IRQCHIP for GICv2. 803On s390, a dummy irq routing table is created. 804 805Note that on s390 the KVM_CAP_S390_IRQCHIP vm capability needs to be enabled 806before KVM_CREATE_IRQCHIP can be used. 807 808 8094.25 KVM_IRQ_LINE 810----------------- 811 812:Capability: KVM_CAP_IRQCHIP 813:Architectures: x86, arm, arm64 814:Type: vm ioctl 815:Parameters: struct kvm_irq_level 816:Returns: 0 on success, -1 on error 817 818Sets the level of a GSI input to the interrupt controller model in the kernel. 819On some architectures it is required that an interrupt controller model has 820been previously created with KVM_CREATE_IRQCHIP. Note that edge-triggered 821interrupts require the level to be set to 1 and then back to 0. 822 823On real hardware, interrupt pins can be active-low or active-high. This 824does not matter for the level field of struct kvm_irq_level: 1 always 825means active (asserted), 0 means inactive (deasserted). 826 827x86 allows the operating system to program the interrupt polarity 828(active-low/active-high) for level-triggered interrupts, and KVM used 829to consider the polarity. However, due to bitrot in the handling of 830active-low interrupts, the above convention is now valid on x86 too. 831This is signaled by KVM_CAP_X86_IOAPIC_POLARITY_IGNORED. Userspace 832should not present interrupts to the guest as active-low unless this 833capability is present (or unless it is not using the in-kernel irqchip, 834of course). 835 836 837ARM/arm64 can signal an interrupt either at the CPU level, or at the 838in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to 839use PPIs designated for specific cpus. The irq field is interpreted 840like this:: 841 842 bits: | 31 ... 28 | 27 ... 24 | 23 ... 16 | 15 ... 0 | 843 field: | vcpu2_index | irq_type | vcpu_index | irq_id | 844 845The irq_type field has the following values: 846 847- irq_type[0]: 848 out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ 849- irq_type[1]: 850 in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.) 851 (the vcpu_index field is ignored) 852- irq_type[2]: 853 in-kernel GIC: PPI, irq_id between 16 and 31 (incl.) 854 855(The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs) 856 857In both cases, level is used to assert/deassert the line. 858 859When KVM_CAP_ARM_IRQ_LINE_LAYOUT_2 is supported, the target vcpu is 860identified as (256 * vcpu2_index + vcpu_index). Otherwise, vcpu2_index 861must be zero. 862 863Note that on arm/arm64, the KVM_CAP_IRQCHIP capability only conditions 864injection of interrupts for the in-kernel irqchip. KVM_IRQ_LINE can always 865be used for a userspace interrupt controller. 866 867:: 868 869 struct kvm_irq_level { 870 union { 871 __u32 irq; /* GSI */ 872 __s32 status; /* not used for KVM_IRQ_LEVEL */ 873 }; 874 __u32 level; /* 0 or 1 */ 875 }; 876 877 8784.26 KVM_GET_IRQCHIP 879-------------------- 880 881:Capability: KVM_CAP_IRQCHIP 882:Architectures: x86 883:Type: vm ioctl 884:Parameters: struct kvm_irqchip (in/out) 885:Returns: 0 on success, -1 on error 886 887Reads the state of a kernel interrupt controller created with 888KVM_CREATE_IRQCHIP into a buffer provided by the caller. 889 890:: 891 892 struct kvm_irqchip { 893 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */ 894 __u32 pad; 895 union { 896 char dummy[512]; /* reserving space */ 897 struct kvm_pic_state pic; 898 struct kvm_ioapic_state ioapic; 899 } chip; 900 }; 901 902 9034.27 KVM_SET_IRQCHIP 904-------------------- 905 906:Capability: KVM_CAP_IRQCHIP 907:Architectures: x86 908:Type: vm ioctl 909:Parameters: struct kvm_irqchip (in) 910:Returns: 0 on success, -1 on error 911 912Sets the state of a kernel interrupt controller created with 913KVM_CREATE_IRQCHIP from a buffer provided by the caller. 914 915:: 916 917 struct kvm_irqchip { 918 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */ 919 __u32 pad; 920 union { 921 char dummy[512]; /* reserving space */ 922 struct kvm_pic_state pic; 923 struct kvm_ioapic_state ioapic; 924 } chip; 925 }; 926 927 9284.28 KVM_XEN_HVM_CONFIG 929----------------------- 930 931:Capability: KVM_CAP_XEN_HVM 932:Architectures: x86 933:Type: vm ioctl 934:Parameters: struct kvm_xen_hvm_config (in) 935:Returns: 0 on success, -1 on error 936 937Sets the MSR that the Xen HVM guest uses to initialize its hypercall 938page, and provides the starting address and size of the hypercall 939blobs in userspace. When the guest writes the MSR, kvm copies one 940page of a blob (32- or 64-bit, depending on the vcpu mode) to guest 941memory. 942 943:: 944 945 struct kvm_xen_hvm_config { 946 __u32 flags; 947 __u32 msr; 948 __u64 blob_addr_32; 949 __u64 blob_addr_64; 950 __u8 blob_size_32; 951 __u8 blob_size_64; 952 __u8 pad2[30]; 953 }; 954 955 9564.29 KVM_GET_CLOCK 957------------------ 958 959:Capability: KVM_CAP_ADJUST_CLOCK 960:Architectures: x86 961:Type: vm ioctl 962:Parameters: struct kvm_clock_data (out) 963:Returns: 0 on success, -1 on error 964 965Gets the current timestamp of kvmclock as seen by the current guest. In 966conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios 967such as migration. 968 969When KVM_CAP_ADJUST_CLOCK is passed to KVM_CHECK_EXTENSION, it returns the 970set of bits that KVM can return in struct kvm_clock_data's flag member. 971 972The only flag defined now is KVM_CLOCK_TSC_STABLE. If set, the returned 973value is the exact kvmclock value seen by all VCPUs at the instant 974when KVM_GET_CLOCK was called. If clear, the returned value is simply 975CLOCK_MONOTONIC plus a constant offset; the offset can be modified 976with KVM_SET_CLOCK. KVM will try to make all VCPUs follow this clock, 977but the exact value read by each VCPU could differ, because the host 978TSC is not stable. 979 980:: 981 982 struct kvm_clock_data { 983 __u64 clock; /* kvmclock current value */ 984 __u32 flags; 985 __u32 pad[9]; 986 }; 987 988 9894.30 KVM_SET_CLOCK 990------------------ 991 992:Capability: KVM_CAP_ADJUST_CLOCK 993:Architectures: x86 994:Type: vm ioctl 995:Parameters: struct kvm_clock_data (in) 996:Returns: 0 on success, -1 on error 997 998Sets the current timestamp of kvmclock to the value specified in its parameter. 999In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios 1000such as migration. 1001 1002:: 1003 1004 struct kvm_clock_data { 1005 __u64 clock; /* kvmclock current value */ 1006 __u32 flags; 1007 __u32 pad[9]; 1008 }; 1009 1010 10114.31 KVM_GET_VCPU_EVENTS 1012------------------------ 1013 1014:Capability: KVM_CAP_VCPU_EVENTS 1015:Extended by: KVM_CAP_INTR_SHADOW 1016:Architectures: x86, arm, arm64 1017:Type: vcpu ioctl 1018:Parameters: struct kvm_vcpu_event (out) 1019:Returns: 0 on success, -1 on error 1020 1021X86: 1022^^^^ 1023 1024Gets currently pending exceptions, interrupts, and NMIs as well as related 1025states of the vcpu. 1026 1027:: 1028 1029 struct kvm_vcpu_events { 1030 struct { 1031 __u8 injected; 1032 __u8 nr; 1033 __u8 has_error_code; 1034 __u8 pending; 1035 __u32 error_code; 1036 } exception; 1037 struct { 1038 __u8 injected; 1039 __u8 nr; 1040 __u8 soft; 1041 __u8 shadow; 1042 } interrupt; 1043 struct { 1044 __u8 injected; 1045 __u8 pending; 1046 __u8 masked; 1047 __u8 pad; 1048 } nmi; 1049 __u32 sipi_vector; 1050 __u32 flags; 1051 struct { 1052 __u8 smm; 1053 __u8 pending; 1054 __u8 smm_inside_nmi; 1055 __u8 latched_init; 1056 } smi; 1057 __u8 reserved[27]; 1058 __u8 exception_has_payload; 1059 __u64 exception_payload; 1060 }; 1061 1062The following bits are defined in the flags field: 1063 1064- KVM_VCPUEVENT_VALID_SHADOW may be set to signal that 1065 interrupt.shadow contains a valid state. 1066 1067- KVM_VCPUEVENT_VALID_SMM may be set to signal that smi contains a 1068 valid state. 1069 1070- KVM_VCPUEVENT_VALID_PAYLOAD may be set to signal that the 1071 exception_has_payload, exception_payload, and exception.pending 1072 fields contain a valid state. This bit will be set whenever 1073 KVM_CAP_EXCEPTION_PAYLOAD is enabled. 1074 1075ARM/ARM64: 1076^^^^^^^^^^ 1077 1078If the guest accesses a device that is being emulated by the host kernel in 1079such a way that a real device would generate a physical SError, KVM may make 1080a virtual SError pending for that VCPU. This system error interrupt remains 1081pending until the guest takes the exception by unmasking PSTATE.A. 1082 1083Running the VCPU may cause it to take a pending SError, or make an access that 1084causes an SError to become pending. The event's description is only valid while 1085the VPCU is not running. 1086 1087This API provides a way to read and write the pending 'event' state that is not 1088visible to the guest. To save, restore or migrate a VCPU the struct representing 1089the state can be read then written using this GET/SET API, along with the other 1090guest-visible registers. It is not possible to 'cancel' an SError that has been 1091made pending. 1092 1093A device being emulated in user-space may also wish to generate an SError. To do 1094this the events structure can be populated by user-space. The current state 1095should be read first, to ensure no existing SError is pending. If an existing 1096SError is pending, the architecture's 'Multiple SError interrupts' rules should 1097be followed. (2.5.3 of DDI0587.a "ARM Reliability, Availability, and 1098Serviceability (RAS) Specification"). 1099 1100SError exceptions always have an ESR value. Some CPUs have the ability to 1101specify what the virtual SError's ESR value should be. These systems will 1102advertise KVM_CAP_ARM_INJECT_SERROR_ESR. In this case exception.has_esr will 1103always have a non-zero value when read, and the agent making an SError pending 1104should specify the ISS field in the lower 24 bits of exception.serror_esr. If 1105the system supports KVM_CAP_ARM_INJECT_SERROR_ESR, but user-space sets the events 1106with exception.has_esr as zero, KVM will choose an ESR. 1107 1108Specifying exception.has_esr on a system that does not support it will return 1109-EINVAL. Setting anything other than the lower 24bits of exception.serror_esr 1110will return -EINVAL. 1111 1112It is not possible to read back a pending external abort (injected via 1113KVM_SET_VCPU_EVENTS or otherwise) because such an exception is always delivered 1114directly to the virtual CPU). 1115 1116:: 1117 1118 struct kvm_vcpu_events { 1119 struct { 1120 __u8 serror_pending; 1121 __u8 serror_has_esr; 1122 __u8 ext_dabt_pending; 1123 /* Align it to 8 bytes */ 1124 __u8 pad[5]; 1125 __u64 serror_esr; 1126 } exception; 1127 __u32 reserved[12]; 1128 }; 1129 11304.32 KVM_SET_VCPU_EVENTS 1131------------------------ 1132 1133:Capability: KVM_CAP_VCPU_EVENTS 1134:Extended by: KVM_CAP_INTR_SHADOW 1135:Architectures: x86, arm, arm64 1136:Type: vcpu ioctl 1137:Parameters: struct kvm_vcpu_event (in) 1138:Returns: 0 on success, -1 on error 1139 1140X86: 1141^^^^ 1142 1143Set pending exceptions, interrupts, and NMIs as well as related states of the 1144vcpu. 1145 1146See KVM_GET_VCPU_EVENTS for the data structure. 1147 1148Fields that may be modified asynchronously by running VCPUs can be excluded 1149from the update. These fields are nmi.pending, sipi_vector, smi.smm, 1150smi.pending. Keep the corresponding bits in the flags field cleared to 1151suppress overwriting the current in-kernel state. The bits are: 1152 1153=============================== ================================== 1154KVM_VCPUEVENT_VALID_NMI_PENDING transfer nmi.pending to the kernel 1155KVM_VCPUEVENT_VALID_SIPI_VECTOR transfer sipi_vector 1156KVM_VCPUEVENT_VALID_SMM transfer the smi sub-struct. 1157=============================== ================================== 1158 1159If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in 1160the flags field to signal that interrupt.shadow contains a valid state and 1161shall be written into the VCPU. 1162 1163KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available. 1164 1165If KVM_CAP_EXCEPTION_PAYLOAD is enabled, KVM_VCPUEVENT_VALID_PAYLOAD 1166can be set in the flags field to signal that the 1167exception_has_payload, exception_payload, and exception.pending fields 1168contain a valid state and shall be written into the VCPU. 1169 1170ARM/ARM64: 1171^^^^^^^^^^ 1172 1173User space may need to inject several types of events to the guest. 1174 1175Set the pending SError exception state for this VCPU. It is not possible to 1176'cancel' an Serror that has been made pending. 1177 1178If the guest performed an access to I/O memory which could not be handled by 1179userspace, for example because of missing instruction syndrome decode 1180information or because there is no device mapped at the accessed IPA, then 1181userspace can ask the kernel to inject an external abort using the address 1182from the exiting fault on the VCPU. It is a programming error to set 1183ext_dabt_pending after an exit which was not either KVM_EXIT_MMIO or 1184KVM_EXIT_ARM_NISV. This feature is only available if the system supports 1185KVM_CAP_ARM_INJECT_EXT_DABT. This is a helper which provides commonality in 1186how userspace reports accesses for the above cases to guests, across different 1187userspace implementations. Nevertheless, userspace can still emulate all Arm 1188exceptions by manipulating individual registers using the KVM_SET_ONE_REG API. 1189 1190See KVM_GET_VCPU_EVENTS for the data structure. 1191 1192 11934.33 KVM_GET_DEBUGREGS 1194---------------------- 1195 1196:Capability: KVM_CAP_DEBUGREGS 1197:Architectures: x86 1198:Type: vm ioctl 1199:Parameters: struct kvm_debugregs (out) 1200:Returns: 0 on success, -1 on error 1201 1202Reads debug registers from the vcpu. 1203 1204:: 1205 1206 struct kvm_debugregs { 1207 __u64 db[4]; 1208 __u64 dr6; 1209 __u64 dr7; 1210 __u64 flags; 1211 __u64 reserved[9]; 1212 }; 1213 1214 12154.34 KVM_SET_DEBUGREGS 1216---------------------- 1217 1218:Capability: KVM_CAP_DEBUGREGS 1219:Architectures: x86 1220:Type: vm ioctl 1221:Parameters: struct kvm_debugregs (in) 1222:Returns: 0 on success, -1 on error 1223 1224Writes debug registers into the vcpu. 1225 1226See KVM_GET_DEBUGREGS for the data structure. The flags field is unused 1227yet and must be cleared on entry. 1228 1229 12304.35 KVM_SET_USER_MEMORY_REGION 1231------------------------------- 1232 1233:Capability: KVM_CAP_USER_MEMORY 1234:Architectures: all 1235:Type: vm ioctl 1236:Parameters: struct kvm_userspace_memory_region (in) 1237:Returns: 0 on success, -1 on error 1238 1239:: 1240 1241 struct kvm_userspace_memory_region { 1242 __u32 slot; 1243 __u32 flags; 1244 __u64 guest_phys_addr; 1245 __u64 memory_size; /* bytes */ 1246 __u64 userspace_addr; /* start of the userspace allocated memory */ 1247 }; 1248 1249 /* for kvm_memory_region::flags */ 1250 #define KVM_MEM_LOG_DIRTY_PAGES (1UL << 0) 1251 #define KVM_MEM_READONLY (1UL << 1) 1252 1253This ioctl allows the user to create, modify or delete a guest physical 1254memory slot. Bits 0-15 of "slot" specify the slot id and this value 1255should be less than the maximum number of user memory slots supported per 1256VM. The maximum allowed slots can be queried using KVM_CAP_NR_MEMSLOTS. 1257Slots may not overlap in guest physical address space. 1258 1259If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot" 1260specifies the address space which is being modified. They must be 1261less than the value that KVM_CHECK_EXTENSION returns for the 1262KVM_CAP_MULTI_ADDRESS_SPACE capability. Slots in separate address spaces 1263are unrelated; the restriction on overlapping slots only applies within 1264each address space. 1265 1266Deleting a slot is done by passing zero for memory_size. When changing 1267an existing slot, it may be moved in the guest physical memory space, 1268or its flags may be modified, but it may not be resized. 1269 1270Memory for the region is taken starting at the address denoted by the 1271field userspace_addr, which must point at user addressable memory for 1272the entire memory slot size. Any object may back this memory, including 1273anonymous memory, ordinary files, and hugetlbfs. 1274 1275On architectures that support a form of address tagging, userspace_addr must 1276be an untagged address. 1277 1278It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr 1279be identical. This allows large pages in the guest to be backed by large 1280pages in the host. 1281 1282The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and 1283KVM_MEM_READONLY. The former can be set to instruct KVM to keep track of 1284writes to memory within the slot. See KVM_GET_DIRTY_LOG ioctl to know how to 1285use it. The latter can be set, if KVM_CAP_READONLY_MEM capability allows it, 1286to make a new slot read-only. In this case, writes to this memory will be 1287posted to userspace as KVM_EXIT_MMIO exits. 1288 1289When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of 1290the memory region are automatically reflected into the guest. For example, an 1291mmap() that affects the region will be made visible immediately. Another 1292example is madvise(MADV_DROP). 1293 1294It is recommended to use this API instead of the KVM_SET_MEMORY_REGION ioctl. 1295The KVM_SET_MEMORY_REGION does not allow fine grained control over memory 1296allocation and is deprecated. 1297 1298 12994.36 KVM_SET_TSS_ADDR 1300--------------------- 1301 1302:Capability: KVM_CAP_SET_TSS_ADDR 1303:Architectures: x86 1304:Type: vm ioctl 1305:Parameters: unsigned long tss_address (in) 1306:Returns: 0 on success, -1 on error 1307 1308This ioctl defines the physical address of a three-page region in the guest 1309physical address space. The region must be within the first 4GB of the 1310guest physical address space and must not conflict with any memory slot 1311or any mmio address. The guest may malfunction if it accesses this memory 1312region. 1313 1314This ioctl is required on Intel-based hosts. This is needed on Intel hardware 1315because of a quirk in the virtualization implementation (see the internals 1316documentation when it pops into existence). 1317 1318 13194.37 KVM_ENABLE_CAP 1320------------------- 1321 1322:Capability: KVM_CAP_ENABLE_CAP 1323:Architectures: mips, ppc, s390 1324:Type: vcpu ioctl 1325:Parameters: struct kvm_enable_cap (in) 1326:Returns: 0 on success; -1 on error 1327 1328:Capability: KVM_CAP_ENABLE_CAP_VM 1329:Architectures: all 1330:Type: vm ioctl 1331:Parameters: struct kvm_enable_cap (in) 1332:Returns: 0 on success; -1 on error 1333 1334.. note:: 1335 1336 Not all extensions are enabled by default. Using this ioctl the application 1337 can enable an extension, making it available to the guest. 1338 1339On systems that do not support this ioctl, it always fails. On systems that 1340do support it, it only works for extensions that are supported for enablement. 1341 1342To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should 1343be used. 1344 1345:: 1346 1347 struct kvm_enable_cap { 1348 /* in */ 1349 __u32 cap; 1350 1351The capability that is supposed to get enabled. 1352 1353:: 1354 1355 __u32 flags; 1356 1357A bitfield indicating future enhancements. Has to be 0 for now. 1358 1359:: 1360 1361 __u64 args[4]; 1362 1363Arguments for enabling a feature. If a feature needs initial values to 1364function properly, this is the place to put them. 1365 1366:: 1367 1368 __u8 pad[64]; 1369 }; 1370 1371The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl 1372for vm-wide capabilities. 1373 13744.38 KVM_GET_MP_STATE 1375--------------------- 1376 1377:Capability: KVM_CAP_MP_STATE 1378:Architectures: x86, s390, arm, arm64 1379:Type: vcpu ioctl 1380:Parameters: struct kvm_mp_state (out) 1381:Returns: 0 on success; -1 on error 1382 1383:: 1384 1385 struct kvm_mp_state { 1386 __u32 mp_state; 1387 }; 1388 1389Returns the vcpu's current "multiprocessing state" (though also valid on 1390uniprocessor guests). 1391 1392Possible values are: 1393 1394 ========================== =============================================== 1395 KVM_MP_STATE_RUNNABLE the vcpu is currently running [x86,arm/arm64] 1396 KVM_MP_STATE_UNINITIALIZED the vcpu is an application processor (AP) 1397 which has not yet received an INIT signal [x86] 1398 KVM_MP_STATE_INIT_RECEIVED the vcpu has received an INIT signal, and is 1399 now ready for a SIPI [x86] 1400 KVM_MP_STATE_HALTED the vcpu has executed a HLT instruction and 1401 is waiting for an interrupt [x86] 1402 KVM_MP_STATE_SIPI_RECEIVED the vcpu has just received a SIPI (vector 1403 accessible via KVM_GET_VCPU_EVENTS) [x86] 1404 KVM_MP_STATE_STOPPED the vcpu is stopped [s390,arm/arm64] 1405 KVM_MP_STATE_CHECK_STOP the vcpu is in a special error state [s390] 1406 KVM_MP_STATE_OPERATING the vcpu is operating (running or halted) 1407 [s390] 1408 KVM_MP_STATE_LOAD the vcpu is in a special load/startup state 1409 [s390] 1410 ========================== =============================================== 1411 1412On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an 1413in-kernel irqchip, the multiprocessing state must be maintained by userspace on 1414these architectures. 1415 1416For arm/arm64: 1417^^^^^^^^^^^^^^ 1418 1419The only states that are valid are KVM_MP_STATE_STOPPED and 1420KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not. 1421 14224.39 KVM_SET_MP_STATE 1423--------------------- 1424 1425:Capability: KVM_CAP_MP_STATE 1426:Architectures: x86, s390, arm, arm64 1427:Type: vcpu ioctl 1428:Parameters: struct kvm_mp_state (in) 1429:Returns: 0 on success; -1 on error 1430 1431Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for 1432arguments. 1433 1434On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an 1435in-kernel irqchip, the multiprocessing state must be maintained by userspace on 1436these architectures. 1437 1438For arm/arm64: 1439^^^^^^^^^^^^^^ 1440 1441The only states that are valid are KVM_MP_STATE_STOPPED and 1442KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not. 1443 14444.40 KVM_SET_IDENTITY_MAP_ADDR 1445------------------------------ 1446 1447:Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR 1448:Architectures: x86 1449:Type: vm ioctl 1450:Parameters: unsigned long identity (in) 1451:Returns: 0 on success, -1 on error 1452 1453This ioctl defines the physical address of a one-page region in the guest 1454physical address space. The region must be within the first 4GB of the 1455guest physical address space and must not conflict with any memory slot 1456or any mmio address. The guest may malfunction if it accesses this memory 1457region. 1458 1459Setting the address to 0 will result in resetting the address to its default 1460(0xfffbc000). 1461 1462This ioctl is required on Intel-based hosts. This is needed on Intel hardware 1463because of a quirk in the virtualization implementation (see the internals 1464documentation when it pops into existence). 1465 1466Fails if any VCPU has already been created. 1467 14684.41 KVM_SET_BOOT_CPU_ID 1469------------------------ 1470 1471:Capability: KVM_CAP_SET_BOOT_CPU_ID 1472:Architectures: x86 1473:Type: vm ioctl 1474:Parameters: unsigned long vcpu_id 1475:Returns: 0 on success, -1 on error 1476 1477Define which vcpu is the Bootstrap Processor (BSP). Values are the same 1478as the vcpu id in KVM_CREATE_VCPU. If this ioctl is not called, the default 1479is vcpu 0. 1480 1481 14824.42 KVM_GET_XSAVE 1483------------------ 1484 1485:Capability: KVM_CAP_XSAVE 1486:Architectures: x86 1487:Type: vcpu ioctl 1488:Parameters: struct kvm_xsave (out) 1489:Returns: 0 on success, -1 on error 1490 1491 1492:: 1493 1494 struct kvm_xsave { 1495 __u32 region[1024]; 1496 }; 1497 1498This ioctl would copy current vcpu's xsave struct to the userspace. 1499 1500 15014.43 KVM_SET_XSAVE 1502------------------ 1503 1504:Capability: KVM_CAP_XSAVE 1505:Architectures: x86 1506:Type: vcpu ioctl 1507:Parameters: struct kvm_xsave (in) 1508:Returns: 0 on success, -1 on error 1509 1510:: 1511 1512 1513 struct kvm_xsave { 1514 __u32 region[1024]; 1515 }; 1516 1517This ioctl would copy userspace's xsave struct to the kernel. 1518 1519 15204.44 KVM_GET_XCRS 1521----------------- 1522 1523:Capability: KVM_CAP_XCRS 1524:Architectures: x86 1525:Type: vcpu ioctl 1526:Parameters: struct kvm_xcrs (out) 1527:Returns: 0 on success, -1 on error 1528 1529:: 1530 1531 struct kvm_xcr { 1532 __u32 xcr; 1533 __u32 reserved; 1534 __u64 value; 1535 }; 1536 1537 struct kvm_xcrs { 1538 __u32 nr_xcrs; 1539 __u32 flags; 1540 struct kvm_xcr xcrs[KVM_MAX_XCRS]; 1541 __u64 padding[16]; 1542 }; 1543 1544This ioctl would copy current vcpu's xcrs to the userspace. 1545 1546 15474.45 KVM_SET_XCRS 1548----------------- 1549 1550:Capability: KVM_CAP_XCRS 1551:Architectures: x86 1552:Type: vcpu ioctl 1553:Parameters: struct kvm_xcrs (in) 1554:Returns: 0 on success, -1 on error 1555 1556:: 1557 1558 struct kvm_xcr { 1559 __u32 xcr; 1560 __u32 reserved; 1561 __u64 value; 1562 }; 1563 1564 struct kvm_xcrs { 1565 __u32 nr_xcrs; 1566 __u32 flags; 1567 struct kvm_xcr xcrs[KVM_MAX_XCRS]; 1568 __u64 padding[16]; 1569 }; 1570 1571This ioctl would set vcpu's xcr to the value userspace specified. 1572 1573 15744.46 KVM_GET_SUPPORTED_CPUID 1575---------------------------- 1576 1577:Capability: KVM_CAP_EXT_CPUID 1578:Architectures: x86 1579:Type: system ioctl 1580:Parameters: struct kvm_cpuid2 (in/out) 1581:Returns: 0 on success, -1 on error 1582 1583:: 1584 1585 struct kvm_cpuid2 { 1586 __u32 nent; 1587 __u32 padding; 1588 struct kvm_cpuid_entry2 entries[0]; 1589 }; 1590 1591 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0) 1592 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */ 1593 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */ 1594 1595 struct kvm_cpuid_entry2 { 1596 __u32 function; 1597 __u32 index; 1598 __u32 flags; 1599 __u32 eax; 1600 __u32 ebx; 1601 __u32 ecx; 1602 __u32 edx; 1603 __u32 padding[3]; 1604 }; 1605 1606This ioctl returns x86 cpuid features which are supported by both the 1607hardware and kvm in its default configuration. Userspace can use the 1608information returned by this ioctl to construct cpuid information (for 1609KVM_SET_CPUID2) that is consistent with hardware, kernel, and 1610userspace capabilities, and with user requirements (for example, the 1611user may wish to constrain cpuid to emulate older hardware, or for 1612feature consistency across a cluster). 1613 1614Note that certain capabilities, such as KVM_CAP_X86_DISABLE_EXITS, may 1615expose cpuid features (e.g. MONITOR) which are not supported by kvm in 1616its default configuration. If userspace enables such capabilities, it 1617is responsible for modifying the results of this ioctl appropriately. 1618 1619Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure 1620with the 'nent' field indicating the number of entries in the variable-size 1621array 'entries'. If the number of entries is too low to describe the cpu 1622capabilities, an error (E2BIG) is returned. If the number is too high, 1623the 'nent' field is adjusted and an error (ENOMEM) is returned. If the 1624number is just right, the 'nent' field is adjusted to the number of valid 1625entries in the 'entries' array, which is then filled. 1626 1627The entries returned are the host cpuid as returned by the cpuid instruction, 1628with unknown or unsupported features masked out. Some features (for example, 1629x2apic), may not be present in the host cpu, but are exposed by kvm if it can 1630emulate them efficiently. The fields in each entry are defined as follows: 1631 1632 function: 1633 the eax value used to obtain the entry 1634 1635 index: 1636 the ecx value used to obtain the entry (for entries that are 1637 affected by ecx) 1638 1639 flags: 1640 an OR of zero or more of the following: 1641 1642 KVM_CPUID_FLAG_SIGNIFCANT_INDEX: 1643 if the index field is valid 1644 1645 eax, ebx, ecx, edx: 1646 the values returned by the cpuid instruction for 1647 this function/index combination 1648 1649The TSC deadline timer feature (CPUID leaf 1, ecx[24]) is always returned 1650as false, since the feature depends on KVM_CREATE_IRQCHIP for local APIC 1651support. Instead it is reported via:: 1652 1653 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER) 1654 1655if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the 1656feature in userspace, then you can enable the feature for KVM_SET_CPUID2. 1657 1658 16594.47 KVM_PPC_GET_PVINFO 1660----------------------- 1661 1662:Capability: KVM_CAP_PPC_GET_PVINFO 1663:Architectures: ppc 1664:Type: vm ioctl 1665:Parameters: struct kvm_ppc_pvinfo (out) 1666:Returns: 0 on success, !0 on error 1667 1668:: 1669 1670 struct kvm_ppc_pvinfo { 1671 __u32 flags; 1672 __u32 hcall[4]; 1673 __u8 pad[108]; 1674 }; 1675 1676This ioctl fetches PV specific information that need to be passed to the guest 1677using the device tree or other means from vm context. 1678 1679The hcall array defines 4 instructions that make up a hypercall. 1680 1681If any additional field gets added to this structure later on, a bit for that 1682additional piece of information will be set in the flags bitmap. 1683 1684The flags bitmap is defined as:: 1685 1686 /* the host supports the ePAPR idle hcall 1687 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1<<0) 1688 16894.52 KVM_SET_GSI_ROUTING 1690------------------------ 1691 1692:Capability: KVM_CAP_IRQ_ROUTING 1693:Architectures: x86 s390 arm arm64 1694:Type: vm ioctl 1695:Parameters: struct kvm_irq_routing (in) 1696:Returns: 0 on success, -1 on error 1697 1698Sets the GSI routing table entries, overwriting any previously set entries. 1699 1700On arm/arm64, GSI routing has the following limitation: 1701 1702- GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD. 1703 1704:: 1705 1706 struct kvm_irq_routing { 1707 __u32 nr; 1708 __u32 flags; 1709 struct kvm_irq_routing_entry entries[0]; 1710 }; 1711 1712No flags are specified so far, the corresponding field must be set to zero. 1713 1714:: 1715 1716 struct kvm_irq_routing_entry { 1717 __u32 gsi; 1718 __u32 type; 1719 __u32 flags; 1720 __u32 pad; 1721 union { 1722 struct kvm_irq_routing_irqchip irqchip; 1723 struct kvm_irq_routing_msi msi; 1724 struct kvm_irq_routing_s390_adapter adapter; 1725 struct kvm_irq_routing_hv_sint hv_sint; 1726 __u32 pad[8]; 1727 } u; 1728 }; 1729 1730 /* gsi routing entry types */ 1731 #define KVM_IRQ_ROUTING_IRQCHIP 1 1732 #define KVM_IRQ_ROUTING_MSI 2 1733 #define KVM_IRQ_ROUTING_S390_ADAPTER 3 1734 #define KVM_IRQ_ROUTING_HV_SINT 4 1735 1736flags: 1737 1738- KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry 1739 type, specifies that the devid field contains a valid value. The per-VM 1740 KVM_CAP_MSI_DEVID capability advertises the requirement to provide 1741 the device ID. If this capability is not available, userspace should 1742 never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail. 1743- zero otherwise 1744 1745:: 1746 1747 struct kvm_irq_routing_irqchip { 1748 __u32 irqchip; 1749 __u32 pin; 1750 }; 1751 1752 struct kvm_irq_routing_msi { 1753 __u32 address_lo; 1754 __u32 address_hi; 1755 __u32 data; 1756 union { 1757 __u32 pad; 1758 __u32 devid; 1759 }; 1760 }; 1761 1762If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier 1763for the device that wrote the MSI message. For PCI, this is usually a 1764BFD identifier in the lower 16 bits. 1765 1766On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS 1767feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled, 1768address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of 1769address_hi must be zero. 1770 1771:: 1772 1773 struct kvm_irq_routing_s390_adapter { 1774 __u64 ind_addr; 1775 __u64 summary_addr; 1776 __u64 ind_offset; 1777 __u32 summary_offset; 1778 __u32 adapter_id; 1779 }; 1780 1781 struct kvm_irq_routing_hv_sint { 1782 __u32 vcpu; 1783 __u32 sint; 1784 }; 1785 1786 17874.55 KVM_SET_TSC_KHZ 1788-------------------- 1789 1790:Capability: KVM_CAP_TSC_CONTROL 1791:Architectures: x86 1792:Type: vcpu ioctl 1793:Parameters: virtual tsc_khz 1794:Returns: 0 on success, -1 on error 1795 1796Specifies the tsc frequency for the virtual machine. The unit of the 1797frequency is KHz. 1798 1799 18004.56 KVM_GET_TSC_KHZ 1801-------------------- 1802 1803:Capability: KVM_CAP_GET_TSC_KHZ 1804:Architectures: x86 1805:Type: vcpu ioctl 1806:Parameters: none 1807:Returns: virtual tsc-khz on success, negative value on error 1808 1809Returns the tsc frequency of the guest. The unit of the return value is 1810KHz. If the host has unstable tsc this ioctl returns -EIO instead as an 1811error. 1812 1813 18144.57 KVM_GET_LAPIC 1815------------------ 1816 1817:Capability: KVM_CAP_IRQCHIP 1818:Architectures: x86 1819:Type: vcpu ioctl 1820:Parameters: struct kvm_lapic_state (out) 1821:Returns: 0 on success, -1 on error 1822 1823:: 1824 1825 #define KVM_APIC_REG_SIZE 0x400 1826 struct kvm_lapic_state { 1827 char regs[KVM_APIC_REG_SIZE]; 1828 }; 1829 1830Reads the Local APIC registers and copies them into the input argument. The 1831data format and layout are the same as documented in the architecture manual. 1832 1833If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is 1834enabled, then the format of APIC_ID register depends on the APIC mode 1835(reported by MSR_IA32_APICBASE) of its VCPU. x2APIC stores APIC ID in 1836the APIC_ID register (bytes 32-35). xAPIC only allows an 8-bit APIC ID 1837which is stored in bits 31-24 of the APIC register, or equivalently in 1838byte 35 of struct kvm_lapic_state's regs field. KVM_GET_LAPIC must then 1839be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR. 1840 1841If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state 1842always uses xAPIC format. 1843 1844 18454.58 KVM_SET_LAPIC 1846------------------ 1847 1848:Capability: KVM_CAP_IRQCHIP 1849:Architectures: x86 1850:Type: vcpu ioctl 1851:Parameters: struct kvm_lapic_state (in) 1852:Returns: 0 on success, -1 on error 1853 1854:: 1855 1856 #define KVM_APIC_REG_SIZE 0x400 1857 struct kvm_lapic_state { 1858 char regs[KVM_APIC_REG_SIZE]; 1859 }; 1860 1861Copies the input argument into the Local APIC registers. The data format 1862and layout are the same as documented in the architecture manual. 1863 1864The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's 1865regs field) depends on the state of the KVM_CAP_X2APIC_API capability. 1866See the note in KVM_GET_LAPIC. 1867 1868 18694.59 KVM_IOEVENTFD 1870------------------ 1871 1872:Capability: KVM_CAP_IOEVENTFD 1873:Architectures: all 1874:Type: vm ioctl 1875:Parameters: struct kvm_ioeventfd (in) 1876:Returns: 0 on success, !0 on error 1877 1878This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address 1879within the guest. A guest write in the registered address will signal the 1880provided event instead of triggering an exit. 1881 1882:: 1883 1884 struct kvm_ioeventfd { 1885 __u64 datamatch; 1886 __u64 addr; /* legal pio/mmio address */ 1887 __u32 len; /* 0, 1, 2, 4, or 8 bytes */ 1888 __s32 fd; 1889 __u32 flags; 1890 __u8 pad[36]; 1891 }; 1892 1893For the special case of virtio-ccw devices on s390, the ioevent is matched 1894to a subchannel/virtqueue tuple instead. 1895 1896The following flags are defined:: 1897 1898 #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch) 1899 #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio) 1900 #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign) 1901 #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \ 1902 (1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify) 1903 1904If datamatch flag is set, the event will be signaled only if the written value 1905to the registered address is equal to datamatch in struct kvm_ioeventfd. 1906 1907For virtio-ccw devices, addr contains the subchannel id and datamatch the 1908virtqueue index. 1909 1910With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and 1911the kernel will ignore the length of guest write and may get a faster vmexit. 1912The speedup may only apply to specific architectures, but the ioeventfd will 1913work anyway. 1914 19154.60 KVM_DIRTY_TLB 1916------------------ 1917 1918:Capability: KVM_CAP_SW_TLB 1919:Architectures: ppc 1920:Type: vcpu ioctl 1921:Parameters: struct kvm_dirty_tlb (in) 1922:Returns: 0 on success, -1 on error 1923 1924:: 1925 1926 struct kvm_dirty_tlb { 1927 __u64 bitmap; 1928 __u32 num_dirty; 1929 }; 1930 1931This must be called whenever userspace has changed an entry in the shared 1932TLB, prior to calling KVM_RUN on the associated vcpu. 1933 1934The "bitmap" field is the userspace address of an array. This array 1935consists of a number of bits, equal to the total number of TLB entries as 1936determined by the last successful call to KVM_CONFIG_TLB, rounded up to the 1937nearest multiple of 64. 1938 1939Each bit corresponds to one TLB entry, ordered the same as in the shared TLB 1940array. 1941 1942The array is little-endian: the bit 0 is the least significant bit of the 1943first byte, bit 8 is the least significant bit of the second byte, etc. 1944This avoids any complications with differing word sizes. 1945 1946The "num_dirty" field is a performance hint for KVM to determine whether it 1947should skip processing the bitmap and just invalidate everything. It must 1948be set to the number of set bits in the bitmap. 1949 1950 19514.62 KVM_CREATE_SPAPR_TCE 1952------------------------- 1953 1954:Capability: KVM_CAP_SPAPR_TCE 1955:Architectures: powerpc 1956:Type: vm ioctl 1957:Parameters: struct kvm_create_spapr_tce (in) 1958:Returns: file descriptor for manipulating the created TCE table 1959 1960This creates a virtual TCE (translation control entry) table, which 1961is an IOMMU for PAPR-style virtual I/O. It is used to translate 1962logical addresses used in virtual I/O into guest physical addresses, 1963and provides a scatter/gather capability for PAPR virtual I/O. 1964 1965:: 1966 1967 /* for KVM_CAP_SPAPR_TCE */ 1968 struct kvm_create_spapr_tce { 1969 __u64 liobn; 1970 __u32 window_size; 1971 }; 1972 1973The liobn field gives the logical IO bus number for which to create a 1974TCE table. The window_size field specifies the size of the DMA window 1975which this TCE table will translate - the table will contain one 64 1976bit TCE entry for every 4kiB of the DMA window. 1977 1978When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE 1979table has been created using this ioctl(), the kernel will handle it 1980in real mode, updating the TCE table. H_PUT_TCE calls for other 1981liobns will cause a vm exit and must be handled by userspace. 1982 1983The return value is a file descriptor which can be passed to mmap(2) 1984to map the created TCE table into userspace. This lets userspace read 1985the entries written by kernel-handled H_PUT_TCE calls, and also lets 1986userspace update the TCE table directly which is useful in some 1987circumstances. 1988 1989 19904.63 KVM_ALLOCATE_RMA 1991--------------------- 1992 1993:Capability: KVM_CAP_PPC_RMA 1994:Architectures: powerpc 1995:Type: vm ioctl 1996:Parameters: struct kvm_allocate_rma (out) 1997:Returns: file descriptor for mapping the allocated RMA 1998 1999This allocates a Real Mode Area (RMA) from the pool allocated at boot 2000time by the kernel. An RMA is a physically-contiguous, aligned region 2001of memory used on older POWER processors to provide the memory which 2002will be accessed by real-mode (MMU off) accesses in a KVM guest. 2003POWER processors support a set of sizes for the RMA that usually 2004includes 64MB, 128MB, 256MB and some larger powers of two. 2005 2006:: 2007 2008 /* for KVM_ALLOCATE_RMA */ 2009 struct kvm_allocate_rma { 2010 __u64 rma_size; 2011 }; 2012 2013The return value is a file descriptor which can be passed to mmap(2) 2014to map the allocated RMA into userspace. The mapped area can then be 2015passed to the KVM_SET_USER_MEMORY_REGION ioctl to establish it as the 2016RMA for a virtual machine. The size of the RMA in bytes (which is 2017fixed at host kernel boot time) is returned in the rma_size field of 2018the argument structure. 2019 2020The KVM_CAP_PPC_RMA capability is 1 or 2 if the KVM_ALLOCATE_RMA ioctl 2021is supported; 2 if the processor requires all virtual machines to have 2022an RMA, or 1 if the processor can use an RMA but doesn't require it, 2023because it supports the Virtual RMA (VRMA) facility. 2024 2025 20264.64 KVM_NMI 2027------------ 2028 2029:Capability: KVM_CAP_USER_NMI 2030:Architectures: x86 2031:Type: vcpu ioctl 2032:Parameters: none 2033:Returns: 0 on success, -1 on error 2034 2035Queues an NMI on the thread's vcpu. Note this is well defined only 2036when KVM_CREATE_IRQCHIP has not been called, since this is an interface 2037between the virtual cpu core and virtual local APIC. After KVM_CREATE_IRQCHIP 2038has been called, this interface is completely emulated within the kernel. 2039 2040To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the 2041following algorithm: 2042 2043 - pause the vcpu 2044 - read the local APIC's state (KVM_GET_LAPIC) 2045 - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1) 2046 - if so, issue KVM_NMI 2047 - resume the vcpu 2048 2049Some guests configure the LINT1 NMI input to cause a panic, aiding in 2050debugging. 2051 2052 20534.65 KVM_S390_UCAS_MAP 2054---------------------- 2055 2056:Capability: KVM_CAP_S390_UCONTROL 2057:Architectures: s390 2058:Type: vcpu ioctl 2059:Parameters: struct kvm_s390_ucas_mapping (in) 2060:Returns: 0 in case of success 2061 2062The parameter is defined like this:: 2063 2064 struct kvm_s390_ucas_mapping { 2065 __u64 user_addr; 2066 __u64 vcpu_addr; 2067 __u64 length; 2068 }; 2069 2070This ioctl maps the memory at "user_addr" with the length "length" to 2071the vcpu's address space starting at "vcpu_addr". All parameters need to 2072be aligned by 1 megabyte. 2073 2074 20754.66 KVM_S390_UCAS_UNMAP 2076------------------------ 2077 2078:Capability: KVM_CAP_S390_UCONTROL 2079:Architectures: s390 2080:Type: vcpu ioctl 2081:Parameters: struct kvm_s390_ucas_mapping (in) 2082:Returns: 0 in case of success 2083 2084The parameter is defined like this:: 2085 2086 struct kvm_s390_ucas_mapping { 2087 __u64 user_addr; 2088 __u64 vcpu_addr; 2089 __u64 length; 2090 }; 2091 2092This ioctl unmaps the memory in the vcpu's address space starting at 2093"vcpu_addr" with the length "length". The field "user_addr" is ignored. 2094All parameters need to be aligned by 1 megabyte. 2095 2096 20974.67 KVM_S390_VCPU_FAULT 2098------------------------ 2099 2100:Capability: KVM_CAP_S390_UCONTROL 2101:Architectures: s390 2102:Type: vcpu ioctl 2103:Parameters: vcpu absolute address (in) 2104:Returns: 0 in case of success 2105 2106This call creates a page table entry on the virtual cpu's address space 2107(for user controlled virtual machines) or the virtual machine's address 2108space (for regular virtual machines). This only works for minor faults, 2109thus it's recommended to access subject memory page via the user page 2110table upfront. This is useful to handle validity intercepts for user 2111controlled virtual machines to fault in the virtual cpu's lowcore pages 2112prior to calling the KVM_RUN ioctl. 2113 2114 21154.68 KVM_SET_ONE_REG 2116-------------------- 2117 2118:Capability: KVM_CAP_ONE_REG 2119:Architectures: all 2120:Type: vcpu ioctl 2121:Parameters: struct kvm_one_reg (in) 2122:Returns: 0 on success, negative value on failure 2123 2124Errors: 2125 2126 ====== ============================================================ 2127 ENOENT no such register 2128 EINVAL invalid register ID, or no such register or used with VMs in 2129 protected virtualization mode on s390 2130 EPERM (arm64) register access not allowed before vcpu finalization 2131 ====== ============================================================ 2132 2133(These error codes are indicative only: do not rely on a specific error 2134code being returned in a specific situation.) 2135 2136:: 2137 2138 struct kvm_one_reg { 2139 __u64 id; 2140 __u64 addr; 2141 }; 2142 2143Using this ioctl, a single vcpu register can be set to a specific value 2144defined by user space with the passed in struct kvm_one_reg, where id 2145refers to the register identifier as described below and addr is a pointer 2146to a variable with the respective size. There can be architecture agnostic 2147and architecture specific registers. Each have their own range of operation 2148and their own constants and width. To keep track of the implemented 2149registers, find a list below: 2150 2151 ======= =============================== ============ 2152 Arch Register Width (bits) 2153 ======= =============================== ============ 2154 PPC KVM_REG_PPC_HIOR 64 2155 PPC KVM_REG_PPC_IAC1 64 2156 PPC KVM_REG_PPC_IAC2 64 2157 PPC KVM_REG_PPC_IAC3 64 2158 PPC KVM_REG_PPC_IAC4 64 2159 PPC KVM_REG_PPC_DAC1 64 2160 PPC KVM_REG_PPC_DAC2 64 2161 PPC KVM_REG_PPC_DABR 64 2162 PPC KVM_REG_PPC_DSCR 64 2163 PPC KVM_REG_PPC_PURR 64 2164 PPC KVM_REG_PPC_SPURR 64 2165 PPC KVM_REG_PPC_DAR 64 2166 PPC KVM_REG_PPC_DSISR 32 2167 PPC KVM_REG_PPC_AMR 64 2168 PPC KVM_REG_PPC_UAMOR 64 2169 PPC KVM_REG_PPC_MMCR0 64 2170 PPC KVM_REG_PPC_MMCR1 64 2171 PPC KVM_REG_PPC_MMCRA 64 2172 PPC KVM_REG_PPC_MMCR2 64 2173 PPC KVM_REG_PPC_MMCRS 64 2174 PPC KVM_REG_PPC_MMCR3 64 2175 PPC KVM_REG_PPC_SIAR 64 2176 PPC KVM_REG_PPC_SDAR 64 2177 PPC KVM_REG_PPC_SIER 64 2178 PPC KVM_REG_PPC_SIER2 64 2179 PPC KVM_REG_PPC_SIER3 64 2180 PPC KVM_REG_PPC_PMC1 32 2181 PPC KVM_REG_PPC_PMC2 32 2182 PPC KVM_REG_PPC_PMC3 32 2183 PPC KVM_REG_PPC_PMC4 32 2184 PPC KVM_REG_PPC_PMC5 32 2185 PPC KVM_REG_PPC_PMC6 32 2186 PPC KVM_REG_PPC_PMC7 32 2187 PPC KVM_REG_PPC_PMC8 32 2188 PPC KVM_REG_PPC_FPR0 64 2189 ... 2190 PPC KVM_REG_PPC_FPR31 64 2191 PPC KVM_REG_PPC_VR0 128 2192 ... 2193 PPC KVM_REG_PPC_VR31 128 2194 PPC KVM_REG_PPC_VSR0 128 2195 ... 2196 PPC KVM_REG_PPC_VSR31 128 2197 PPC KVM_REG_PPC_FPSCR 64 2198 PPC KVM_REG_PPC_VSCR 32 2199 PPC KVM_REG_PPC_VPA_ADDR 64 2200 PPC KVM_REG_PPC_VPA_SLB 128 2201 PPC KVM_REG_PPC_VPA_DTL 128 2202 PPC KVM_REG_PPC_EPCR 32 2203 PPC KVM_REG_PPC_EPR 32 2204 PPC KVM_REG_PPC_TCR 32 2205 PPC KVM_REG_PPC_TSR 32 2206 PPC KVM_REG_PPC_OR_TSR 32 2207 PPC KVM_REG_PPC_CLEAR_TSR 32 2208 PPC KVM_REG_PPC_MAS0 32 2209 PPC KVM_REG_PPC_MAS1 32 2210 PPC KVM_REG_PPC_MAS2 64 2211 PPC KVM_REG_PPC_MAS7_3 64 2212 PPC KVM_REG_PPC_MAS4 32 2213 PPC KVM_REG_PPC_MAS6 32 2214 PPC KVM_REG_PPC_MMUCFG 32 2215 PPC KVM_REG_PPC_TLB0CFG 32 2216 PPC KVM_REG_PPC_TLB1CFG 32 2217 PPC KVM_REG_PPC_TLB2CFG 32 2218 PPC KVM_REG_PPC_TLB3CFG 32 2219 PPC KVM_REG_PPC_TLB0PS 32 2220 PPC KVM_REG_PPC_TLB1PS 32 2221 PPC KVM_REG_PPC_TLB2PS 32 2222 PPC KVM_REG_PPC_TLB3PS 32 2223 PPC KVM_REG_PPC_EPTCFG 32 2224 PPC KVM_REG_PPC_ICP_STATE 64 2225 PPC KVM_REG_PPC_VP_STATE 128 2226 PPC KVM_REG_PPC_TB_OFFSET 64 2227 PPC KVM_REG_PPC_SPMC1 32 2228 PPC KVM_REG_PPC_SPMC2 32 2229 PPC KVM_REG_PPC_IAMR 64 2230 PPC KVM_REG_PPC_TFHAR 64 2231 PPC KVM_REG_PPC_TFIAR 64 2232 PPC KVM_REG_PPC_TEXASR 64 2233 PPC KVM_REG_PPC_FSCR 64 2234 PPC KVM_REG_PPC_PSPB 32 2235 PPC KVM_REG_PPC_EBBHR 64 2236 PPC KVM_REG_PPC_EBBRR 64 2237 PPC KVM_REG_PPC_BESCR 64 2238 PPC KVM_REG_PPC_TAR 64 2239 PPC KVM_REG_PPC_DPDES 64 2240 PPC KVM_REG_PPC_DAWR 64 2241 PPC KVM_REG_PPC_DAWRX 64 2242 PPC KVM_REG_PPC_CIABR 64 2243 PPC KVM_REG_PPC_IC 64 2244 PPC KVM_REG_PPC_VTB 64 2245 PPC KVM_REG_PPC_CSIGR 64 2246 PPC KVM_REG_PPC_TACR 64 2247 PPC KVM_REG_PPC_TCSCR 64 2248 PPC KVM_REG_PPC_PID 64 2249 PPC KVM_REG_PPC_ACOP 64 2250 PPC KVM_REG_PPC_VRSAVE 32 2251 PPC KVM_REG_PPC_LPCR 32 2252 PPC KVM_REG_PPC_LPCR_64 64 2253 PPC KVM_REG_PPC_PPR 64 2254 PPC KVM_REG_PPC_ARCH_COMPAT 32 2255 PPC KVM_REG_PPC_DABRX 32 2256 PPC KVM_REG_PPC_WORT 64 2257 PPC KVM_REG_PPC_SPRG9 64 2258 PPC KVM_REG_PPC_DBSR 32 2259 PPC KVM_REG_PPC_TIDR 64 2260 PPC KVM_REG_PPC_PSSCR 64 2261 PPC KVM_REG_PPC_DEC_EXPIRY 64 2262 PPC KVM_REG_PPC_PTCR 64 2263 PPC KVM_REG_PPC_TM_GPR0 64 2264 ... 2265 PPC KVM_REG_PPC_TM_GPR31 64 2266 PPC KVM_REG_PPC_TM_VSR0 128 2267 ... 2268 PPC KVM_REG_PPC_TM_VSR63 128 2269 PPC KVM_REG_PPC_TM_CR 64 2270 PPC KVM_REG_PPC_TM_LR 64 2271 PPC KVM_REG_PPC_TM_CTR 64 2272 PPC KVM_REG_PPC_TM_FPSCR 64 2273 PPC KVM_REG_PPC_TM_AMR 64 2274 PPC KVM_REG_PPC_TM_PPR 64 2275 PPC KVM_REG_PPC_TM_VRSAVE 64 2276 PPC KVM_REG_PPC_TM_VSCR 32 2277 PPC KVM_REG_PPC_TM_DSCR 64 2278 PPC KVM_REG_PPC_TM_TAR 64 2279 PPC KVM_REG_PPC_TM_XER 64 2280 2281 MIPS KVM_REG_MIPS_R0 64 2282 ... 2283 MIPS KVM_REG_MIPS_R31 64 2284 MIPS KVM_REG_MIPS_HI 64 2285 MIPS KVM_REG_MIPS_LO 64 2286 MIPS KVM_REG_MIPS_PC 64 2287 MIPS KVM_REG_MIPS_CP0_INDEX 32 2288 MIPS KVM_REG_MIPS_CP0_ENTRYLO0 64 2289 MIPS KVM_REG_MIPS_CP0_ENTRYLO1 64 2290 MIPS KVM_REG_MIPS_CP0_CONTEXT 64 2291 MIPS KVM_REG_MIPS_CP0_CONTEXTCONFIG 32 2292 MIPS KVM_REG_MIPS_CP0_USERLOCAL 64 2293 MIPS KVM_REG_MIPS_CP0_XCONTEXTCONFIG 64 2294 MIPS KVM_REG_MIPS_CP0_PAGEMASK 32 2295 MIPS KVM_REG_MIPS_CP0_PAGEGRAIN 32 2296 MIPS KVM_REG_MIPS_CP0_SEGCTL0 64 2297 MIPS KVM_REG_MIPS_CP0_SEGCTL1 64 2298 MIPS KVM_REG_MIPS_CP0_SEGCTL2 64 2299 MIPS KVM_REG_MIPS_CP0_PWBASE 64 2300 MIPS KVM_REG_MIPS_CP0_PWFIELD 64 2301 MIPS KVM_REG_MIPS_CP0_PWSIZE 64 2302 MIPS KVM_REG_MIPS_CP0_WIRED 32 2303 MIPS KVM_REG_MIPS_CP0_PWCTL 32 2304 MIPS KVM_REG_MIPS_CP0_HWRENA 32 2305 MIPS KVM_REG_MIPS_CP0_BADVADDR 64 2306 MIPS KVM_REG_MIPS_CP0_BADINSTR 32 2307 MIPS KVM_REG_MIPS_CP0_BADINSTRP 32 2308 MIPS KVM_REG_MIPS_CP0_COUNT 32 2309 MIPS KVM_REG_MIPS_CP0_ENTRYHI 64 2310 MIPS KVM_REG_MIPS_CP0_COMPARE 32 2311 MIPS KVM_REG_MIPS_CP0_STATUS 32 2312 MIPS KVM_REG_MIPS_CP0_INTCTL 32 2313 MIPS KVM_REG_MIPS_CP0_CAUSE 32 2314 MIPS KVM_REG_MIPS_CP0_EPC 64 2315 MIPS KVM_REG_MIPS_CP0_PRID 32 2316 MIPS KVM_REG_MIPS_CP0_EBASE 64 2317 MIPS KVM_REG_MIPS_CP0_CONFIG 32 2318 MIPS KVM_REG_MIPS_CP0_CONFIG1 32 2319 MIPS KVM_REG_MIPS_CP0_CONFIG2 32 2320 MIPS KVM_REG_MIPS_CP0_CONFIG3 32 2321 MIPS KVM_REG_MIPS_CP0_CONFIG4 32 2322 MIPS KVM_REG_MIPS_CP0_CONFIG5 32 2323 MIPS KVM_REG_MIPS_CP0_CONFIG7 32 2324 MIPS KVM_REG_MIPS_CP0_XCONTEXT 64 2325 MIPS KVM_REG_MIPS_CP0_ERROREPC 64 2326 MIPS KVM_REG_MIPS_CP0_KSCRATCH1 64 2327 MIPS KVM_REG_MIPS_CP0_KSCRATCH2 64 2328 MIPS KVM_REG_MIPS_CP0_KSCRATCH3 64 2329 MIPS KVM_REG_MIPS_CP0_KSCRATCH4 64 2330 MIPS KVM_REG_MIPS_CP0_KSCRATCH5 64 2331 MIPS KVM_REG_MIPS_CP0_KSCRATCH6 64 2332 MIPS KVM_REG_MIPS_CP0_MAAR(0..63) 64 2333 MIPS KVM_REG_MIPS_COUNT_CTL 64 2334 MIPS KVM_REG_MIPS_COUNT_RESUME 64 2335 MIPS KVM_REG_MIPS_COUNT_HZ 64 2336 MIPS KVM_REG_MIPS_FPR_32(0..31) 32 2337 MIPS KVM_REG_MIPS_FPR_64(0..31) 64 2338 MIPS KVM_REG_MIPS_VEC_128(0..31) 128 2339 MIPS KVM_REG_MIPS_FCR_IR 32 2340 MIPS KVM_REG_MIPS_FCR_CSR 32 2341 MIPS KVM_REG_MIPS_MSA_IR 32 2342 MIPS KVM_REG_MIPS_MSA_CSR 32 2343 ======= =============================== ============ 2344 2345ARM registers are mapped using the lower 32 bits. The upper 16 of that 2346is the register group type, or coprocessor number: 2347 2348ARM core registers have the following id bit patterns:: 2349 2350 0x4020 0000 0010 <index into the kvm_regs struct:16> 2351 2352ARM 32-bit CP15 registers have the following id bit patterns:: 2353 2354 0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3> 2355 2356ARM 64-bit CP15 registers have the following id bit patterns:: 2357 2358 0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3> 2359 2360ARM CCSIDR registers are demultiplexed by CSSELR value:: 2361 2362 0x4020 0000 0011 00 <csselr:8> 2363 2364ARM 32-bit VFP control registers have the following id bit patterns:: 2365 2366 0x4020 0000 0012 1 <regno:12> 2367 2368ARM 64-bit FP registers have the following id bit patterns:: 2369 2370 0x4030 0000 0012 0 <regno:12> 2371 2372ARM firmware pseudo-registers have the following bit pattern:: 2373 2374 0x4030 0000 0014 <regno:16> 2375 2376 2377arm64 registers are mapped using the lower 32 bits. The upper 16 of 2378that is the register group type, or coprocessor number: 2379 2380arm64 core/FP-SIMD registers have the following id bit patterns. Note 2381that the size of the access is variable, as the kvm_regs structure 2382contains elements ranging from 32 to 128 bits. The index is a 32bit 2383value in the kvm_regs structure seen as a 32bit array:: 2384 2385 0x60x0 0000 0010 <index into the kvm_regs struct:16> 2386 2387Specifically: 2388 2389======================= ========= ===== ======================================= 2390 Encoding Register Bits kvm_regs member 2391======================= ========= ===== ======================================= 2392 0x6030 0000 0010 0000 X0 64 regs.regs[0] 2393 0x6030 0000 0010 0002 X1 64 regs.regs[1] 2394 ... 2395 0x6030 0000 0010 003c X30 64 regs.regs[30] 2396 0x6030 0000 0010 003e SP 64 regs.sp 2397 0x6030 0000 0010 0040 PC 64 regs.pc 2398 0x6030 0000 0010 0042 PSTATE 64 regs.pstate 2399 0x6030 0000 0010 0044 SP_EL1 64 sp_el1 2400 0x6030 0000 0010 0046 ELR_EL1 64 elr_el1 2401 0x6030 0000 0010 0048 SPSR_EL1 64 spsr[KVM_SPSR_EL1] (alias SPSR_SVC) 2402 0x6030 0000 0010 004a SPSR_ABT 64 spsr[KVM_SPSR_ABT] 2403 0x6030 0000 0010 004c SPSR_UND 64 spsr[KVM_SPSR_UND] 2404 0x6030 0000 0010 004e SPSR_IRQ 64 spsr[KVM_SPSR_IRQ] 2405 0x6060 0000 0010 0050 SPSR_FIQ 64 spsr[KVM_SPSR_FIQ] 2406 0x6040 0000 0010 0054 V0 128 fp_regs.vregs[0] [1]_ 2407 0x6040 0000 0010 0058 V1 128 fp_regs.vregs[1] [1]_ 2408 ... 2409 0x6040 0000 0010 00d0 V31 128 fp_regs.vregs[31] [1]_ 2410 0x6020 0000 0010 00d4 FPSR 32 fp_regs.fpsr 2411 0x6020 0000 0010 00d5 FPCR 32 fp_regs.fpcr 2412======================= ========= ===== ======================================= 2413 2414.. [1] These encodings are not accepted for SVE-enabled vcpus. See 2415 KVM_ARM_VCPU_INIT. 2416 2417 The equivalent register content can be accessed via bits [127:0] of 2418 the corresponding SVE Zn registers instead for vcpus that have SVE 2419 enabled (see below). 2420 2421arm64 CCSIDR registers are demultiplexed by CSSELR value:: 2422 2423 0x6020 0000 0011 00 <csselr:8> 2424 2425arm64 system registers have the following id bit patterns:: 2426 2427 0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3> 2428 2429.. warning:: 2430 2431 Two system register IDs do not follow the specified pattern. These 2432 are KVM_REG_ARM_TIMER_CVAL and KVM_REG_ARM_TIMER_CNT, which map to 2433 system registers CNTV_CVAL_EL0 and CNTVCT_EL0 respectively. These 2434 two had their values accidentally swapped, which means TIMER_CVAL is 2435 derived from the register encoding for CNTVCT_EL0 and TIMER_CNT is 2436 derived from the register encoding for CNTV_CVAL_EL0. As this is 2437 API, it must remain this way. 2438 2439arm64 firmware pseudo-registers have the following bit pattern:: 2440 2441 0x6030 0000 0014 <regno:16> 2442 2443arm64 SVE registers have the following bit patterns:: 2444 2445 0x6080 0000 0015 00 <n:5> <slice:5> Zn bits[2048*slice + 2047 : 2048*slice] 2446 0x6050 0000 0015 04 <n:4> <slice:5> Pn bits[256*slice + 255 : 256*slice] 2447 0x6050 0000 0015 060 <slice:5> FFR bits[256*slice + 255 : 256*slice] 2448 0x6060 0000 0015 ffff KVM_REG_ARM64_SVE_VLS pseudo-register 2449 2450Access to register IDs where 2048 * slice >= 128 * max_vq will fail with 2451ENOENT. max_vq is the vcpu's maximum supported vector length in 128-bit 2452quadwords: see [2]_ below. 2453 2454These registers are only accessible on vcpus for which SVE is enabled. 2455See KVM_ARM_VCPU_INIT for details. 2456 2457In addition, except for KVM_REG_ARM64_SVE_VLS, these registers are not 2458accessible until the vcpu's SVE configuration has been finalized 2459using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). See KVM_ARM_VCPU_INIT 2460and KVM_ARM_VCPU_FINALIZE for more information about this procedure. 2461 2462KVM_REG_ARM64_SVE_VLS is a pseudo-register that allows the set of vector 2463lengths supported by the vcpu to be discovered and configured by 2464userspace. When transferred to or from user memory via KVM_GET_ONE_REG 2465or KVM_SET_ONE_REG, the value of this register is of type 2466__u64[KVM_ARM64_SVE_VLS_WORDS], and encodes the set of vector lengths as 2467follows:: 2468 2469 __u64 vector_lengths[KVM_ARM64_SVE_VLS_WORDS]; 2470 2471 if (vq >= SVE_VQ_MIN && vq <= SVE_VQ_MAX && 2472 ((vector_lengths[(vq - KVM_ARM64_SVE_VQ_MIN) / 64] >> 2473 ((vq - KVM_ARM64_SVE_VQ_MIN) % 64)) & 1)) 2474 /* Vector length vq * 16 bytes supported */ 2475 else 2476 /* Vector length vq * 16 bytes not supported */ 2477 2478.. [2] The maximum value vq for which the above condition is true is 2479 max_vq. This is the maximum vector length available to the guest on 2480 this vcpu, and determines which register slices are visible through 2481 this ioctl interface. 2482 2483(See Documentation/arm64/sve.rst for an explanation of the "vq" 2484nomenclature.) 2485 2486KVM_REG_ARM64_SVE_VLS is only accessible after KVM_ARM_VCPU_INIT. 2487KVM_ARM_VCPU_INIT initialises it to the best set of vector lengths that 2488the host supports. 2489 2490Userspace may subsequently modify it if desired until the vcpu's SVE 2491configuration is finalized using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). 2492 2493Apart from simply removing all vector lengths from the host set that 2494exceed some value, support for arbitrarily chosen sets of vector lengths 2495is hardware-dependent and may not be available. Attempting to configure 2496an invalid set of vector lengths via KVM_SET_ONE_REG will fail with 2497EINVAL. 2498 2499After the vcpu's SVE configuration is finalized, further attempts to 2500write this register will fail with EPERM. 2501 2502 2503MIPS registers are mapped using the lower 32 bits. The upper 16 of that is 2504the register group type: 2505 2506MIPS core registers (see above) have the following id bit patterns:: 2507 2508 0x7030 0000 0000 <reg:16> 2509 2510MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit 2511patterns depending on whether they're 32-bit or 64-bit registers:: 2512 2513 0x7020 0000 0001 00 <reg:5> <sel:3> (32-bit) 2514 0x7030 0000 0001 00 <reg:5> <sel:3> (64-bit) 2515 2516Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_MIPS_CP0_ENTRYLO1 are the MIPS64 2517versions of the EntryLo registers regardless of the word size of the host 2518hardware, host kernel, guest, and whether XPA is present in the guest, i.e. 2519with the RI and XI bits (if they exist) in bits 63 and 62 respectively, and 2520the PFNX field starting at bit 30. 2521 2522MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) above) have the following id bit 2523patterns:: 2524 2525 0x7030 0000 0001 01 <reg:8> 2526 2527MIPS KVM control registers (see above) have the following id bit patterns:: 2528 2529 0x7030 0000 0002 <reg:16> 2530 2531MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following 2532id bit patterns depending on the size of the register being accessed. They are 2533always accessed according to the current guest FPU mode (Status.FR and 2534Config5.FRE), i.e. as the guest would see them, and they become unpredictable 2535if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector 2536registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they 2537overlap the FPU registers:: 2538 2539 0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers) 2540 0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers) 2541 0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers) 2542 2543MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the 2544following id bit patterns:: 2545 2546 0x7020 0000 0003 01 <0:3> <reg:5> 2547 2548MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the 2549following id bit patterns:: 2550 2551 0x7020 0000 0003 02 <0:3> <reg:5> 2552 2553 25544.69 KVM_GET_ONE_REG 2555-------------------- 2556 2557:Capability: KVM_CAP_ONE_REG 2558:Architectures: all 2559:Type: vcpu ioctl 2560:Parameters: struct kvm_one_reg (in and out) 2561:Returns: 0 on success, negative value on failure 2562 2563Errors include: 2564 2565 ======== ============================================================ 2566 ENOENT no such register 2567 EINVAL invalid register ID, or no such register or used with VMs in 2568 protected virtualization mode on s390 2569 EPERM (arm64) register access not allowed before vcpu finalization 2570 ======== ============================================================ 2571 2572(These error codes are indicative only: do not rely on a specific error 2573code being returned in a specific situation.) 2574 2575This ioctl allows to receive the value of a single register implemented 2576in a vcpu. The register to read is indicated by the "id" field of the 2577kvm_one_reg struct passed in. On success, the register value can be found 2578at the memory location pointed to by "addr". 2579 2580The list of registers accessible using this interface is identical to the 2581list in 4.68. 2582 2583 25844.70 KVM_KVMCLOCK_CTRL 2585---------------------- 2586 2587:Capability: KVM_CAP_KVMCLOCK_CTRL 2588:Architectures: Any that implement pvclocks (currently x86 only) 2589:Type: vcpu ioctl 2590:Parameters: None 2591:Returns: 0 on success, -1 on error 2592 2593This ioctl sets a flag accessible to the guest indicating that the specified 2594vCPU has been paused by the host userspace. 2595 2596The host will set a flag in the pvclock structure that is checked from the 2597soft lockup watchdog. The flag is part of the pvclock structure that is 2598shared between guest and host, specifically the second bit of the flags 2599field of the pvclock_vcpu_time_info structure. It will be set exclusively by 2600the host and read/cleared exclusively by the guest. The guest operation of 2601checking and clearing the flag must be an atomic operation so 2602load-link/store-conditional, or equivalent must be used. There are two cases 2603where the guest will clear the flag: when the soft lockup watchdog timer resets 2604itself or when a soft lockup is detected. This ioctl can be called any time 2605after pausing the vcpu, but before it is resumed. 2606 2607 26084.71 KVM_SIGNAL_MSI 2609------------------- 2610 2611:Capability: KVM_CAP_SIGNAL_MSI 2612:Architectures: x86 arm arm64 2613:Type: vm ioctl 2614:Parameters: struct kvm_msi (in) 2615:Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error 2616 2617Directly inject a MSI message. Only valid with in-kernel irqchip that handles 2618MSI messages. 2619 2620:: 2621 2622 struct kvm_msi { 2623 __u32 address_lo; 2624 __u32 address_hi; 2625 __u32 data; 2626 __u32 flags; 2627 __u32 devid; 2628 __u8 pad[12]; 2629 }; 2630 2631flags: 2632 KVM_MSI_VALID_DEVID: devid contains a valid value. The per-VM 2633 KVM_CAP_MSI_DEVID capability advertises the requirement to provide 2634 the device ID. If this capability is not available, userspace 2635 should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail. 2636 2637If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier 2638for the device that wrote the MSI message. For PCI, this is usually a 2639BFD identifier in the lower 16 bits. 2640 2641On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS 2642feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled, 2643address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of 2644address_hi must be zero. 2645 2646 26474.71 KVM_CREATE_PIT2 2648-------------------- 2649 2650:Capability: KVM_CAP_PIT2 2651:Architectures: x86 2652:Type: vm ioctl 2653:Parameters: struct kvm_pit_config (in) 2654:Returns: 0 on success, -1 on error 2655 2656Creates an in-kernel device model for the i8254 PIT. This call is only valid 2657after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following 2658parameters have to be passed:: 2659 2660 struct kvm_pit_config { 2661 __u32 flags; 2662 __u32 pad[15]; 2663 }; 2664 2665Valid flags are:: 2666 2667 #define KVM_PIT_SPEAKER_DUMMY 1 /* emulate speaker port stub */ 2668 2669PIT timer interrupts may use a per-VM kernel thread for injection. If it 2670exists, this thread will have a name of the following pattern:: 2671 2672 kvm-pit/<owner-process-pid> 2673 2674When running a guest with elevated priorities, the scheduling parameters of 2675this thread may have to be adjusted accordingly. 2676 2677This IOCTL replaces the obsolete KVM_CREATE_PIT. 2678 2679 26804.72 KVM_GET_PIT2 2681----------------- 2682 2683:Capability: KVM_CAP_PIT_STATE2 2684:Architectures: x86 2685:Type: vm ioctl 2686:Parameters: struct kvm_pit_state2 (out) 2687:Returns: 0 on success, -1 on error 2688 2689Retrieves the state of the in-kernel PIT model. Only valid after 2690KVM_CREATE_PIT2. The state is returned in the following structure:: 2691 2692 struct kvm_pit_state2 { 2693 struct kvm_pit_channel_state channels[3]; 2694 __u32 flags; 2695 __u32 reserved[9]; 2696 }; 2697 2698Valid flags are:: 2699 2700 /* disable PIT in HPET legacy mode */ 2701 #define KVM_PIT_FLAGS_HPET_LEGACY 0x00000001 2702 2703This IOCTL replaces the obsolete KVM_GET_PIT. 2704 2705 27064.73 KVM_SET_PIT2 2707----------------- 2708 2709:Capability: KVM_CAP_PIT_STATE2 2710:Architectures: x86 2711:Type: vm ioctl 2712:Parameters: struct kvm_pit_state2 (in) 2713:Returns: 0 on success, -1 on error 2714 2715Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2. 2716See KVM_GET_PIT2 for details on struct kvm_pit_state2. 2717 2718This IOCTL replaces the obsolete KVM_SET_PIT. 2719 2720 27214.74 KVM_PPC_GET_SMMU_INFO 2722-------------------------- 2723 2724:Capability: KVM_CAP_PPC_GET_SMMU_INFO 2725:Architectures: powerpc 2726:Type: vm ioctl 2727:Parameters: None 2728:Returns: 0 on success, -1 on error 2729 2730This populates and returns a structure describing the features of 2731the "Server" class MMU emulation supported by KVM. 2732This can in turn be used by userspace to generate the appropriate 2733device-tree properties for the guest operating system. 2734 2735The structure contains some global information, followed by an 2736array of supported segment page sizes:: 2737 2738 struct kvm_ppc_smmu_info { 2739 __u64 flags; 2740 __u32 slb_size; 2741 __u32 pad; 2742 struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ]; 2743 }; 2744 2745The supported flags are: 2746 2747 - KVM_PPC_PAGE_SIZES_REAL: 2748 When that flag is set, guest page sizes must "fit" the backing 2749 store page sizes. When not set, any page size in the list can 2750 be used regardless of how they are backed by userspace. 2751 2752 - KVM_PPC_1T_SEGMENTS 2753 The emulated MMU supports 1T segments in addition to the 2754 standard 256M ones. 2755 2756 - KVM_PPC_NO_HASH 2757 This flag indicates that HPT guests are not supported by KVM, 2758 thus all guests must use radix MMU mode. 2759 2760The "slb_size" field indicates how many SLB entries are supported 2761 2762The "sps" array contains 8 entries indicating the supported base 2763page sizes for a segment in increasing order. Each entry is defined 2764as follow:: 2765 2766 struct kvm_ppc_one_seg_page_size { 2767 __u32 page_shift; /* Base page shift of segment (or 0) */ 2768 __u32 slb_enc; /* SLB encoding for BookS */ 2769 struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ]; 2770 }; 2771 2772An entry with a "page_shift" of 0 is unused. Because the array is 2773organized in increasing order, a lookup can stop when encoutering 2774such an entry. 2775 2776The "slb_enc" field provides the encoding to use in the SLB for the 2777page size. The bits are in positions such as the value can directly 2778be OR'ed into the "vsid" argument of the slbmte instruction. 2779 2780The "enc" array is a list which for each of those segment base page 2781size provides the list of supported actual page sizes (which can be 2782only larger or equal to the base page size), along with the 2783corresponding encoding in the hash PTE. Similarly, the array is 27848 entries sorted by increasing sizes and an entry with a "0" shift 2785is an empty entry and a terminator:: 2786 2787 struct kvm_ppc_one_page_size { 2788 __u32 page_shift; /* Page shift (or 0) */ 2789 __u32 pte_enc; /* Encoding in the HPTE (>>12) */ 2790 }; 2791 2792The "pte_enc" field provides a value that can OR'ed into the hash 2793PTE's RPN field (ie, it needs to be shifted left by 12 to OR it 2794into the hash PTE second double word). 2795 27964.75 KVM_IRQFD 2797-------------- 2798 2799:Capability: KVM_CAP_IRQFD 2800:Architectures: x86 s390 arm arm64 2801:Type: vm ioctl 2802:Parameters: struct kvm_irqfd (in) 2803:Returns: 0 on success, -1 on error 2804 2805Allows setting an eventfd to directly trigger a guest interrupt. 2806kvm_irqfd.fd specifies the file descriptor to use as the eventfd and 2807kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When 2808an event is triggered on the eventfd, an interrupt is injected into 2809the guest using the specified gsi pin. The irqfd is removed using 2810the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd 2811and kvm_irqfd.gsi. 2812 2813With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify 2814mechanism allowing emulation of level-triggered, irqfd-based 2815interrupts. When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an 2816additional eventfd in the kvm_irqfd.resamplefd field. When operating 2817in resample mode, posting of an interrupt through kvm_irq.fd asserts 2818the specified gsi in the irqchip. When the irqchip is resampled, such 2819as from an EOI, the gsi is de-asserted and the user is notified via 2820kvm_irqfd.resamplefd. It is the user's responsibility to re-queue 2821the interrupt if the device making use of it still requires service. 2822Note that closing the resamplefd is not sufficient to disable the 2823irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment 2824and need not be specified with KVM_IRQFD_FLAG_DEASSIGN. 2825 2826On arm/arm64, gsi routing being supported, the following can happen: 2827 2828- in case no routing entry is associated to this gsi, injection fails 2829- in case the gsi is associated to an irqchip routing entry, 2830 irqchip.pin + 32 corresponds to the injected SPI ID. 2831- in case the gsi is associated to an MSI routing entry, the MSI 2832 message and device ID are translated into an LPI (support restricted 2833 to GICv3 ITS in-kernel emulation). 2834 28354.76 KVM_PPC_ALLOCATE_HTAB 2836-------------------------- 2837 2838:Capability: KVM_CAP_PPC_ALLOC_HTAB 2839:Architectures: powerpc 2840:Type: vm ioctl 2841:Parameters: Pointer to u32 containing hash table order (in/out) 2842:Returns: 0 on success, -1 on error 2843 2844This requests the host kernel to allocate an MMU hash table for a 2845guest using the PAPR paravirtualization interface. This only does 2846anything if the kernel is configured to use the Book 3S HV style of 2847virtualization. Otherwise the capability doesn't exist and the ioctl 2848returns an ENOTTY error. The rest of this description assumes Book 3S 2849HV. 2850 2851There must be no vcpus running when this ioctl is called; if there 2852are, it will do nothing and return an EBUSY error. 2853 2854The parameter is a pointer to a 32-bit unsigned integer variable 2855containing the order (log base 2) of the desired size of the hash 2856table, which must be between 18 and 46. On successful return from the 2857ioctl, the value will not be changed by the kernel. 2858 2859If no hash table has been allocated when any vcpu is asked to run 2860(with the KVM_RUN ioctl), the host kernel will allocate a 2861default-sized hash table (16 MB). 2862 2863If this ioctl is called when a hash table has already been allocated, 2864with a different order from the existing hash table, the existing hash 2865table will be freed and a new one allocated. If this is ioctl is 2866called when a hash table has already been allocated of the same order 2867as specified, the kernel will clear out the existing hash table (zero 2868all HPTEs). In either case, if the guest is using the virtualized 2869real-mode area (VRMA) facility, the kernel will re-create the VMRA 2870HPTEs on the next KVM_RUN of any vcpu. 2871 28724.77 KVM_S390_INTERRUPT 2873----------------------- 2874 2875:Capability: basic 2876:Architectures: s390 2877:Type: vm ioctl, vcpu ioctl 2878:Parameters: struct kvm_s390_interrupt (in) 2879:Returns: 0 on success, -1 on error 2880 2881Allows to inject an interrupt to the guest. Interrupts can be floating 2882(vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type. 2883 2884Interrupt parameters are passed via kvm_s390_interrupt:: 2885 2886 struct kvm_s390_interrupt { 2887 __u32 type; 2888 __u32 parm; 2889 __u64 parm64; 2890 }; 2891 2892type can be one of the following: 2893 2894KVM_S390_SIGP_STOP (vcpu) 2895 - sigp stop; optional flags in parm 2896KVM_S390_PROGRAM_INT (vcpu) 2897 - program check; code in parm 2898KVM_S390_SIGP_SET_PREFIX (vcpu) 2899 - sigp set prefix; prefix address in parm 2900KVM_S390_RESTART (vcpu) 2901 - restart 2902KVM_S390_INT_CLOCK_COMP (vcpu) 2903 - clock comparator interrupt 2904KVM_S390_INT_CPU_TIMER (vcpu) 2905 - CPU timer interrupt 2906KVM_S390_INT_VIRTIO (vm) 2907 - virtio external interrupt; external interrupt 2908 parameters in parm and parm64 2909KVM_S390_INT_SERVICE (vm) 2910 - sclp external interrupt; sclp parameter in parm 2911KVM_S390_INT_EMERGENCY (vcpu) 2912 - sigp emergency; source cpu in parm 2913KVM_S390_INT_EXTERNAL_CALL (vcpu) 2914 - sigp external call; source cpu in parm 2915KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm) 2916 - compound value to indicate an 2917 I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel); 2918 I/O interruption parameters in parm (subchannel) and parm64 (intparm, 2919 interruption subclass) 2920KVM_S390_MCHK (vm, vcpu) 2921 - machine check interrupt; cr 14 bits in parm, machine check interrupt 2922 code in parm64 (note that machine checks needing further payload are not 2923 supported by this ioctl) 2924 2925This is an asynchronous vcpu ioctl and can be invoked from any thread. 2926 29274.78 KVM_PPC_GET_HTAB_FD 2928------------------------ 2929 2930:Capability: KVM_CAP_PPC_HTAB_FD 2931:Architectures: powerpc 2932:Type: vm ioctl 2933:Parameters: Pointer to struct kvm_get_htab_fd (in) 2934:Returns: file descriptor number (>= 0) on success, -1 on error 2935 2936This returns a file descriptor that can be used either to read out the 2937entries in the guest's hashed page table (HPT), or to write entries to 2938initialize the HPT. The returned fd can only be written to if the 2939KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and 2940can only be read if that bit is clear. The argument struct looks like 2941this:: 2942 2943 /* For KVM_PPC_GET_HTAB_FD */ 2944 struct kvm_get_htab_fd { 2945 __u64 flags; 2946 __u64 start_index; 2947 __u64 reserved[2]; 2948 }; 2949 2950 /* Values for kvm_get_htab_fd.flags */ 2951 #define KVM_GET_HTAB_BOLTED_ONLY ((__u64)0x1) 2952 #define KVM_GET_HTAB_WRITE ((__u64)0x2) 2953 2954The 'start_index' field gives the index in the HPT of the entry at 2955which to start reading. It is ignored when writing. 2956 2957Reads on the fd will initially supply information about all 2958"interesting" HPT entries. Interesting entries are those with the 2959bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise 2960all entries. When the end of the HPT is reached, the read() will 2961return. If read() is called again on the fd, it will start again from 2962the beginning of the HPT, but will only return HPT entries that have 2963changed since they were last read. 2964 2965Data read or written is structured as a header (8 bytes) followed by a 2966series of valid HPT entries (16 bytes) each. The header indicates how 2967many valid HPT entries there are and how many invalid entries follow 2968the valid entries. The invalid entries are not represented explicitly 2969in the stream. The header format is:: 2970 2971 struct kvm_get_htab_header { 2972 __u32 index; 2973 __u16 n_valid; 2974 __u16 n_invalid; 2975 }; 2976 2977Writes to the fd create HPT entries starting at the index given in the 2978header; first 'n_valid' valid entries with contents from the data 2979written, then 'n_invalid' invalid entries, invalidating any previously 2980valid entries found. 2981 29824.79 KVM_CREATE_DEVICE 2983---------------------- 2984 2985:Capability: KVM_CAP_DEVICE_CTRL 2986:Type: vm ioctl 2987:Parameters: struct kvm_create_device (in/out) 2988:Returns: 0 on success, -1 on error 2989 2990Errors: 2991 2992 ====== ======================================================= 2993 ENODEV The device type is unknown or unsupported 2994 EEXIST Device already created, and this type of device may not 2995 be instantiated multiple times 2996 ====== ======================================================= 2997 2998 Other error conditions may be defined by individual device types or 2999 have their standard meanings. 3000 3001Creates an emulated device in the kernel. The file descriptor returned 3002in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR. 3003 3004If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the 3005device type is supported (not necessarily whether it can be created 3006in the current vm). 3007 3008Individual devices should not define flags. Attributes should be used 3009for specifying any behavior that is not implied by the device type 3010number. 3011 3012:: 3013 3014 struct kvm_create_device { 3015 __u32 type; /* in: KVM_DEV_TYPE_xxx */ 3016 __u32 fd; /* out: device handle */ 3017 __u32 flags; /* in: KVM_CREATE_DEVICE_xxx */ 3018 }; 3019 30204.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR 3021-------------------------------------------- 3022 3023:Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device, 3024 KVM_CAP_VCPU_ATTRIBUTES for vcpu device 3025:Type: device ioctl, vm ioctl, vcpu ioctl 3026:Parameters: struct kvm_device_attr 3027:Returns: 0 on success, -1 on error 3028 3029Errors: 3030 3031 ===== ============================================================= 3032 ENXIO The group or attribute is unknown/unsupported for this device 3033 or hardware support is missing. 3034 EPERM The attribute cannot (currently) be accessed this way 3035 (e.g. read-only attribute, or attribute that only makes 3036 sense when the device is in a different state) 3037 ===== ============================================================= 3038 3039 Other error conditions may be defined by individual device types. 3040 3041Gets/sets a specified piece of device configuration and/or state. The 3042semantics are device-specific. See individual device documentation in 3043the "devices" directory. As with ONE_REG, the size of the data 3044transferred is defined by the particular attribute. 3045 3046:: 3047 3048 struct kvm_device_attr { 3049 __u32 flags; /* no flags currently defined */ 3050 __u32 group; /* device-defined */ 3051 __u64 attr; /* group-defined */ 3052 __u64 addr; /* userspace address of attr data */ 3053 }; 3054 30554.81 KVM_HAS_DEVICE_ATTR 3056------------------------ 3057 3058:Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device, 3059 KVM_CAP_VCPU_ATTRIBUTES for vcpu device 3060:Type: device ioctl, vm ioctl, vcpu ioctl 3061:Parameters: struct kvm_device_attr 3062:Returns: 0 on success, -1 on error 3063 3064Errors: 3065 3066 ===== ============================================================= 3067 ENXIO The group or attribute is unknown/unsupported for this device 3068 or hardware support is missing. 3069 ===== ============================================================= 3070 3071Tests whether a device supports a particular attribute. A successful 3072return indicates the attribute is implemented. It does not necessarily 3073indicate that the attribute can be read or written in the device's 3074current state. "addr" is ignored. 3075 30764.82 KVM_ARM_VCPU_INIT 3077---------------------- 3078 3079:Capability: basic 3080:Architectures: arm, arm64 3081:Type: vcpu ioctl 3082:Parameters: struct kvm_vcpu_init (in) 3083:Returns: 0 on success; -1 on error 3084 3085Errors: 3086 3087 ====== ================================================================= 3088 EINVAL the target is unknown, or the combination of features is invalid. 3089 ENOENT a features bit specified is unknown. 3090 ====== ================================================================= 3091 3092This tells KVM what type of CPU to present to the guest, and what 3093optional features it should have. This will cause a reset of the cpu 3094registers to their initial values. If this is not called, KVM_RUN will 3095return ENOEXEC for that vcpu. 3096 3097Note that because some registers reflect machine topology, all vcpus 3098should be created before this ioctl is invoked. 3099 3100Userspace can call this function multiple times for a given vcpu, including 3101after the vcpu has been run. This will reset the vcpu to its initial 3102state. All calls to this function after the initial call must use the same 3103target and same set of feature flags, otherwise EINVAL will be returned. 3104 3105Possible features: 3106 3107 - KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state. 3108 Depends on KVM_CAP_ARM_PSCI. If not set, the CPU will be powered on 3109 and execute guest code when KVM_RUN is called. 3110 - KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode. 3111 Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only). 3112 - KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 (or a future revision 3113 backward compatible with v0.2) for the CPU. 3114 Depends on KVM_CAP_ARM_PSCI_0_2. 3115 - KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU. 3116 Depends on KVM_CAP_ARM_PMU_V3. 3117 3118 - KVM_ARM_VCPU_PTRAUTH_ADDRESS: Enables Address Pointer authentication 3119 for arm64 only. 3120 Depends on KVM_CAP_ARM_PTRAUTH_ADDRESS. 3121 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are 3122 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and 3123 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be 3124 requested. 3125 3126 - KVM_ARM_VCPU_PTRAUTH_GENERIC: Enables Generic Pointer authentication 3127 for arm64 only. 3128 Depends on KVM_CAP_ARM_PTRAUTH_GENERIC. 3129 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are 3130 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and 3131 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be 3132 requested. 3133 3134 - KVM_ARM_VCPU_SVE: Enables SVE for the CPU (arm64 only). 3135 Depends on KVM_CAP_ARM_SVE. 3136 Requires KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3137 3138 * After KVM_ARM_VCPU_INIT: 3139 3140 - KVM_REG_ARM64_SVE_VLS may be read using KVM_GET_ONE_REG: the 3141 initial value of this pseudo-register indicates the best set of 3142 vector lengths possible for a vcpu on this host. 3143 3144 * Before KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3145 3146 - KVM_RUN and KVM_GET_REG_LIST are not available; 3147 3148 - KVM_GET_ONE_REG and KVM_SET_ONE_REG cannot be used to access 3149 the scalable archietctural SVE registers 3150 KVM_REG_ARM64_SVE_ZREG(), KVM_REG_ARM64_SVE_PREG() or 3151 KVM_REG_ARM64_SVE_FFR; 3152 3153 - KVM_REG_ARM64_SVE_VLS may optionally be written using 3154 KVM_SET_ONE_REG, to modify the set of vector lengths available 3155 for the vcpu. 3156 3157 * After KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3158 3159 - the KVM_REG_ARM64_SVE_VLS pseudo-register is immutable, and can 3160 no longer be written using KVM_SET_ONE_REG. 3161 31624.83 KVM_ARM_PREFERRED_TARGET 3163----------------------------- 3164 3165:Capability: basic 3166:Architectures: arm, arm64 3167:Type: vm ioctl 3168:Parameters: struct kvm_vcpu_init (out) 3169:Returns: 0 on success; -1 on error 3170 3171Errors: 3172 3173 ====== ========================================== 3174 ENODEV no preferred target available for the host 3175 ====== ========================================== 3176 3177This queries KVM for preferred CPU target type which can be emulated 3178by KVM on underlying host. 3179 3180The ioctl returns struct kvm_vcpu_init instance containing information 3181about preferred CPU target type and recommended features for it. The 3182kvm_vcpu_init->features bitmap returned will have feature bits set if 3183the preferred target recommends setting these features, but this is 3184not mandatory. 3185 3186The information returned by this ioctl can be used to prepare an instance 3187of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in 3188VCPU matching underlying host. 3189 3190 31914.84 KVM_GET_REG_LIST 3192--------------------- 3193 3194:Capability: basic 3195:Architectures: arm, arm64, mips 3196:Type: vcpu ioctl 3197:Parameters: struct kvm_reg_list (in/out) 3198:Returns: 0 on success; -1 on error 3199 3200Errors: 3201 3202 ===== ============================================================== 3203 E2BIG the reg index list is too big to fit in the array specified by 3204 the user (the number required will be written into n). 3205 ===== ============================================================== 3206 3207:: 3208 3209 struct kvm_reg_list { 3210 __u64 n; /* number of registers in reg[] */ 3211 __u64 reg[0]; 3212 }; 3213 3214This ioctl returns the guest registers that are supported for the 3215KVM_GET_ONE_REG/KVM_SET_ONE_REG calls. 3216 3217 32184.85 KVM_ARM_SET_DEVICE_ADDR (deprecated) 3219----------------------------------------- 3220 3221:Capability: KVM_CAP_ARM_SET_DEVICE_ADDR 3222:Architectures: arm, arm64 3223:Type: vm ioctl 3224:Parameters: struct kvm_arm_device_address (in) 3225:Returns: 0 on success, -1 on error 3226 3227Errors: 3228 3229 ====== ============================================ 3230 ENODEV The device id is unknown 3231 ENXIO Device not supported on current system 3232 EEXIST Address already set 3233 E2BIG Address outside guest physical address space 3234 EBUSY Address overlaps with other device range 3235 ====== ============================================ 3236 3237:: 3238 3239 struct kvm_arm_device_addr { 3240 __u64 id; 3241 __u64 addr; 3242 }; 3243 3244Specify a device address in the guest's physical address space where guests 3245can access emulated or directly exposed devices, which the host kernel needs 3246to know about. The id field is an architecture specific identifier for a 3247specific device. 3248 3249ARM/arm64 divides the id field into two parts, a device id and an 3250address type id specific to the individual device:: 3251 3252 bits: | 63 ... 32 | 31 ... 16 | 15 ... 0 | 3253 field: | 0x00000000 | device id | addr type id | 3254 3255ARM/arm64 currently only require this when using the in-kernel GIC 3256support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2 3257as the device id. When setting the base address for the guest's 3258mapping of the VGIC virtual CPU and distributor interface, the ioctl 3259must be called after calling KVM_CREATE_IRQCHIP, but before calling 3260KVM_RUN on any of the VCPUs. Calling this ioctl twice for any of the 3261base addresses will return -EEXIST. 3262 3263Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API 3264should be used instead. 3265 3266 32674.86 KVM_PPC_RTAS_DEFINE_TOKEN 3268------------------------------ 3269 3270:Capability: KVM_CAP_PPC_RTAS 3271:Architectures: ppc 3272:Type: vm ioctl 3273:Parameters: struct kvm_rtas_token_args 3274:Returns: 0 on success, -1 on error 3275 3276Defines a token value for a RTAS (Run Time Abstraction Services) 3277service in order to allow it to be handled in the kernel. The 3278argument struct gives the name of the service, which must be the name 3279of a service that has a kernel-side implementation. If the token 3280value is non-zero, it will be associated with that service, and 3281subsequent RTAS calls by the guest specifying that token will be 3282handled by the kernel. If the token value is 0, then any token 3283associated with the service will be forgotten, and subsequent RTAS 3284calls by the guest for that service will be passed to userspace to be 3285handled. 3286 32874.87 KVM_SET_GUEST_DEBUG 3288------------------------ 3289 3290:Capability: KVM_CAP_SET_GUEST_DEBUG 3291:Architectures: x86, s390, ppc, arm64 3292:Type: vcpu ioctl 3293:Parameters: struct kvm_guest_debug (in) 3294:Returns: 0 on success; -1 on error 3295 3296:: 3297 3298 struct kvm_guest_debug { 3299 __u32 control; 3300 __u32 pad; 3301 struct kvm_guest_debug_arch arch; 3302 }; 3303 3304Set up the processor specific debug registers and configure vcpu for 3305handling guest debug events. There are two parts to the structure, the 3306first a control bitfield indicates the type of debug events to handle 3307when running. Common control bits are: 3308 3309 - KVM_GUESTDBG_ENABLE: guest debugging is enabled 3310 - KVM_GUESTDBG_SINGLESTEP: the next run should single-step 3311 3312The top 16 bits of the control field are architecture specific control 3313flags which can include the following: 3314 3315 - KVM_GUESTDBG_USE_SW_BP: using software breakpoints [x86, arm64] 3316 - KVM_GUESTDBG_USE_HW_BP: using hardware breakpoints [x86, s390, arm64] 3317 - KVM_GUESTDBG_INJECT_DB: inject DB type exception [x86] 3318 - KVM_GUESTDBG_INJECT_BP: inject BP type exception [x86] 3319 - KVM_GUESTDBG_EXIT_PENDING: trigger an immediate guest exit [s390] 3320 3321For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints 3322are enabled in memory so we need to ensure breakpoint exceptions are 3323correctly trapped and the KVM run loop exits at the breakpoint and not 3324running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP 3325we need to ensure the guest vCPUs architecture specific registers are 3326updated to the correct (supplied) values. 3327 3328The second part of the structure is architecture specific and 3329typically contains a set of debug registers. 3330 3331For arm64 the number of debug registers is implementation defined and 3332can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and 3333KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number 3334indicating the number of supported registers. 3335 3336For ppc, the KVM_CAP_PPC_GUEST_DEBUG_SSTEP capability indicates whether 3337the single-step debug event (KVM_GUESTDBG_SINGLESTEP) is supported. 3338 3339When debug events exit the main run loop with the reason 3340KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run 3341structure containing architecture specific debug information. 3342 33434.88 KVM_GET_EMULATED_CPUID 3344--------------------------- 3345 3346:Capability: KVM_CAP_EXT_EMUL_CPUID 3347:Architectures: x86 3348:Type: system ioctl 3349:Parameters: struct kvm_cpuid2 (in/out) 3350:Returns: 0 on success, -1 on error 3351 3352:: 3353 3354 struct kvm_cpuid2 { 3355 __u32 nent; 3356 __u32 flags; 3357 struct kvm_cpuid_entry2 entries[0]; 3358 }; 3359 3360The member 'flags' is used for passing flags from userspace. 3361 3362:: 3363 3364 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0) 3365 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */ 3366 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */ 3367 3368 struct kvm_cpuid_entry2 { 3369 __u32 function; 3370 __u32 index; 3371 __u32 flags; 3372 __u32 eax; 3373 __u32 ebx; 3374 __u32 ecx; 3375 __u32 edx; 3376 __u32 padding[3]; 3377 }; 3378 3379This ioctl returns x86 cpuid features which are emulated by 3380kvm.Userspace can use the information returned by this ioctl to query 3381which features are emulated by kvm instead of being present natively. 3382 3383Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2 3384structure with the 'nent' field indicating the number of entries in 3385the variable-size array 'entries'. If the number of entries is too low 3386to describe the cpu capabilities, an error (E2BIG) is returned. If the 3387number is too high, the 'nent' field is adjusted and an error (ENOMEM) 3388is returned. If the number is just right, the 'nent' field is adjusted 3389to the number of valid entries in the 'entries' array, which is then 3390filled. 3391 3392The entries returned are the set CPUID bits of the respective features 3393which kvm emulates, as returned by the CPUID instruction, with unknown 3394or unsupported feature bits cleared. 3395 3396Features like x2apic, for example, may not be present in the host cpu 3397but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be 3398emulated efficiently and thus not included here. 3399 3400The fields in each entry are defined as follows: 3401 3402 function: 3403 the eax value used to obtain the entry 3404 index: 3405 the ecx value used to obtain the entry (for entries that are 3406 affected by ecx) 3407 flags: 3408 an OR of zero or more of the following: 3409 3410 KVM_CPUID_FLAG_SIGNIFCANT_INDEX: 3411 if the index field is valid 3412 3413 eax, ebx, ecx, edx: 3414 3415 the values returned by the cpuid instruction for 3416 this function/index combination 3417 34184.89 KVM_S390_MEM_OP 3419-------------------- 3420 3421:Capability: KVM_CAP_S390_MEM_OP 3422:Architectures: s390 3423:Type: vcpu ioctl 3424:Parameters: struct kvm_s390_mem_op (in) 3425:Returns: = 0 on success, 3426 < 0 on generic error (e.g. -EFAULT or -ENOMEM), 3427 > 0 if an exception occurred while walking the page tables 3428 3429Read or write data from/to the logical (virtual) memory of a VCPU. 3430 3431Parameters are specified via the following structure:: 3432 3433 struct kvm_s390_mem_op { 3434 __u64 gaddr; /* the guest address */ 3435 __u64 flags; /* flags */ 3436 __u32 size; /* amount of bytes */ 3437 __u32 op; /* type of operation */ 3438 __u64 buf; /* buffer in userspace */ 3439 __u8 ar; /* the access register number */ 3440 __u8 reserved[31]; /* should be set to 0 */ 3441 }; 3442 3443The type of operation is specified in the "op" field. It is either 3444KVM_S390_MEMOP_LOGICAL_READ for reading from logical memory space or 3445KVM_S390_MEMOP_LOGICAL_WRITE for writing to logical memory space. The 3446KVM_S390_MEMOP_F_CHECK_ONLY flag can be set in the "flags" field to check 3447whether the corresponding memory access would create an access exception 3448(without touching the data in the memory at the destination). In case an 3449access exception occurred while walking the MMU tables of the guest, the 3450ioctl returns a positive error number to indicate the type of exception. 3451This exception is also raised directly at the corresponding VCPU if the 3452flag KVM_S390_MEMOP_F_INJECT_EXCEPTION is set in the "flags" field. 3453 3454The start address of the memory region has to be specified in the "gaddr" 3455field, and the length of the region in the "size" field (which must not 3456be 0). The maximum value for "size" can be obtained by checking the 3457KVM_CAP_S390_MEM_OP capability. "buf" is the buffer supplied by the 3458userspace application where the read data should be written to for 3459KVM_S390_MEMOP_LOGICAL_READ, or where the data that should be written is 3460stored for a KVM_S390_MEMOP_LOGICAL_WRITE. When KVM_S390_MEMOP_F_CHECK_ONLY 3461is specified, "buf" is unused and can be NULL. "ar" designates the access 3462register number to be used; the valid range is 0..15. 3463 3464The "reserved" field is meant for future extensions. It is not used by 3465KVM with the currently defined set of flags. 3466 34674.90 KVM_S390_GET_SKEYS 3468----------------------- 3469 3470:Capability: KVM_CAP_S390_SKEYS 3471:Architectures: s390 3472:Type: vm ioctl 3473:Parameters: struct kvm_s390_skeys 3474:Returns: 0 on success, KVM_S390_GET_KEYS_NONE if guest is not using storage 3475 keys, negative value on error 3476 3477This ioctl is used to get guest storage key values on the s390 3478architecture. The ioctl takes parameters via the kvm_s390_skeys struct:: 3479 3480 struct kvm_s390_skeys { 3481 __u64 start_gfn; 3482 __u64 count; 3483 __u64 skeydata_addr; 3484 __u32 flags; 3485 __u32 reserved[9]; 3486 }; 3487 3488The start_gfn field is the number of the first guest frame whose storage keys 3489you want to get. 3490 3491The count field is the number of consecutive frames (starting from start_gfn) 3492whose storage keys to get. The count field must be at least 1 and the maximum 3493allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range 3494will cause the ioctl to return -EINVAL. 3495 3496The skeydata_addr field is the address to a buffer large enough to hold count 3497bytes. This buffer will be filled with storage key data by the ioctl. 3498 34994.91 KVM_S390_SET_SKEYS 3500----------------------- 3501 3502:Capability: KVM_CAP_S390_SKEYS 3503:Architectures: s390 3504:Type: vm ioctl 3505:Parameters: struct kvm_s390_skeys 3506:Returns: 0 on success, negative value on error 3507 3508This ioctl is used to set guest storage key values on the s390 3509architecture. The ioctl takes parameters via the kvm_s390_skeys struct. 3510See section on KVM_S390_GET_SKEYS for struct definition. 3511 3512The start_gfn field is the number of the first guest frame whose storage keys 3513you want to set. 3514 3515The count field is the number of consecutive frames (starting from start_gfn) 3516whose storage keys to get. The count field must be at least 1 and the maximum 3517allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range 3518will cause the ioctl to return -EINVAL. 3519 3520The skeydata_addr field is the address to a buffer containing count bytes of 3521storage keys. Each byte in the buffer will be set as the storage key for a 3522single frame starting at start_gfn for count frames. 3523 3524Note: If any architecturally invalid key value is found in the given data then 3525the ioctl will return -EINVAL. 3526 35274.92 KVM_S390_IRQ 3528----------------- 3529 3530:Capability: KVM_CAP_S390_INJECT_IRQ 3531:Architectures: s390 3532:Type: vcpu ioctl 3533:Parameters: struct kvm_s390_irq (in) 3534:Returns: 0 on success, -1 on error 3535 3536Errors: 3537 3538 3539 ====== ================================================================= 3540 EINVAL interrupt type is invalid 3541 type is KVM_S390_SIGP_STOP and flag parameter is invalid value, 3542 type is KVM_S390_INT_EXTERNAL_CALL and code is bigger 3543 than the maximum of VCPUs 3544 EBUSY type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped, 3545 type is KVM_S390_SIGP_STOP and a stop irq is already pending, 3546 type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt 3547 is already pending 3548 ====== ================================================================= 3549 3550Allows to inject an interrupt to the guest. 3551 3552Using struct kvm_s390_irq as a parameter allows 3553to inject additional payload which is not 3554possible via KVM_S390_INTERRUPT. 3555 3556Interrupt parameters are passed via kvm_s390_irq:: 3557 3558 struct kvm_s390_irq { 3559 __u64 type; 3560 union { 3561 struct kvm_s390_io_info io; 3562 struct kvm_s390_ext_info ext; 3563 struct kvm_s390_pgm_info pgm; 3564 struct kvm_s390_emerg_info emerg; 3565 struct kvm_s390_extcall_info extcall; 3566 struct kvm_s390_prefix_info prefix; 3567 struct kvm_s390_stop_info stop; 3568 struct kvm_s390_mchk_info mchk; 3569 char reserved[64]; 3570 } u; 3571 }; 3572 3573type can be one of the following: 3574 3575- KVM_S390_SIGP_STOP - sigp stop; parameter in .stop 3576- KVM_S390_PROGRAM_INT - program check; parameters in .pgm 3577- KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix 3578- KVM_S390_RESTART - restart; no parameters 3579- KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters 3580- KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters 3581- KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg 3582- KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall 3583- KVM_S390_MCHK - machine check interrupt; parameters in .mchk 3584 3585This is an asynchronous vcpu ioctl and can be invoked from any thread. 3586 35874.94 KVM_S390_GET_IRQ_STATE 3588--------------------------- 3589 3590:Capability: KVM_CAP_S390_IRQ_STATE 3591:Architectures: s390 3592:Type: vcpu ioctl 3593:Parameters: struct kvm_s390_irq_state (out) 3594:Returns: >= number of bytes copied into buffer, 3595 -EINVAL if buffer size is 0, 3596 -ENOBUFS if buffer size is too small to fit all pending interrupts, 3597 -EFAULT if the buffer address was invalid 3598 3599This ioctl allows userspace to retrieve the complete state of all currently 3600pending interrupts in a single buffer. Use cases include migration 3601and introspection. The parameter structure contains the address of a 3602userspace buffer and its length:: 3603 3604 struct kvm_s390_irq_state { 3605 __u64 buf; 3606 __u32 flags; /* will stay unused for compatibility reasons */ 3607 __u32 len; 3608 __u32 reserved[4]; /* will stay unused for compatibility reasons */ 3609 }; 3610 3611Userspace passes in the above struct and for each pending interrupt a 3612struct kvm_s390_irq is copied to the provided buffer. 3613 3614The structure contains a flags and a reserved field for future extensions. As 3615the kernel never checked for flags == 0 and QEMU never pre-zeroed flags and 3616reserved, these fields can not be used in the future without breaking 3617compatibility. 3618 3619If -ENOBUFS is returned the buffer provided was too small and userspace 3620may retry with a bigger buffer. 3621 36224.95 KVM_S390_SET_IRQ_STATE 3623--------------------------- 3624 3625:Capability: KVM_CAP_S390_IRQ_STATE 3626:Architectures: s390 3627:Type: vcpu ioctl 3628:Parameters: struct kvm_s390_irq_state (in) 3629:Returns: 0 on success, 3630 -EFAULT if the buffer address was invalid, 3631 -EINVAL for an invalid buffer length (see below), 3632 -EBUSY if there were already interrupts pending, 3633 errors occurring when actually injecting the 3634 interrupt. See KVM_S390_IRQ. 3635 3636This ioctl allows userspace to set the complete state of all cpu-local 3637interrupts currently pending for the vcpu. It is intended for restoring 3638interrupt state after a migration. The input parameter is a userspace buffer 3639containing a struct kvm_s390_irq_state:: 3640 3641 struct kvm_s390_irq_state { 3642 __u64 buf; 3643 __u32 flags; /* will stay unused for compatibility reasons */ 3644 __u32 len; 3645 __u32 reserved[4]; /* will stay unused for compatibility reasons */ 3646 }; 3647 3648The restrictions for flags and reserved apply as well. 3649(see KVM_S390_GET_IRQ_STATE) 3650 3651The userspace memory referenced by buf contains a struct kvm_s390_irq 3652for each interrupt to be injected into the guest. 3653If one of the interrupts could not be injected for some reason the 3654ioctl aborts. 3655 3656len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0 3657and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq), 3658which is the maximum number of possibly pending cpu-local interrupts. 3659 36604.96 KVM_SMI 3661------------ 3662 3663:Capability: KVM_CAP_X86_SMM 3664:Architectures: x86 3665:Type: vcpu ioctl 3666:Parameters: none 3667:Returns: 0 on success, -1 on error 3668 3669Queues an SMI on the thread's vcpu. 3670 36714.97 KVM_CAP_PPC_MULTITCE 3672------------------------- 3673 3674:Capability: KVM_CAP_PPC_MULTITCE 3675:Architectures: ppc 3676:Type: vm 3677 3678This capability means the kernel is capable of handling hypercalls 3679H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user 3680space. This significantly accelerates DMA operations for PPC KVM guests. 3681User space should expect that its handlers for these hypercalls 3682are not going to be called if user space previously registered LIOBN 3683in KVM (via KVM_CREATE_SPAPR_TCE or similar calls). 3684 3685In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest, 3686user space might have to advertise it for the guest. For example, 3687IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is 3688present in the "ibm,hypertas-functions" device-tree property. 3689 3690The hypercalls mentioned above may or may not be processed successfully 3691in the kernel based fast path. If they can not be handled by the kernel, 3692they will get passed on to user space. So user space still has to have 3693an implementation for these despite the in kernel acceleration. 3694 3695This capability is always enabled. 3696 36974.98 KVM_CREATE_SPAPR_TCE_64 3698---------------------------- 3699 3700:Capability: KVM_CAP_SPAPR_TCE_64 3701:Architectures: powerpc 3702:Type: vm ioctl 3703:Parameters: struct kvm_create_spapr_tce_64 (in) 3704:Returns: file descriptor for manipulating the created TCE table 3705 3706This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit 3707windows, described in 4.62 KVM_CREATE_SPAPR_TCE 3708 3709This capability uses extended struct in ioctl interface:: 3710 3711 /* for KVM_CAP_SPAPR_TCE_64 */ 3712 struct kvm_create_spapr_tce_64 { 3713 __u64 liobn; 3714 __u32 page_shift; 3715 __u32 flags; 3716 __u64 offset; /* in pages */ 3717 __u64 size; /* in pages */ 3718 }; 3719 3720The aim of extension is to support an additional bigger DMA window with 3721a variable page size. 3722KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and 3723a bus offset of the corresponding DMA window, @size and @offset are numbers 3724of IOMMU pages. 3725 3726@flags are not used at the moment. 3727 3728The rest of functionality is identical to KVM_CREATE_SPAPR_TCE. 3729 37304.99 KVM_REINJECT_CONTROL 3731------------------------- 3732 3733:Capability: KVM_CAP_REINJECT_CONTROL 3734:Architectures: x86 3735:Type: vm ioctl 3736:Parameters: struct kvm_reinject_control (in) 3737:Returns: 0 on success, 3738 -EFAULT if struct kvm_reinject_control cannot be read, 3739 -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier. 3740 3741i8254 (PIT) has two modes, reinject and !reinject. The default is reinject, 3742where KVM queues elapsed i8254 ticks and monitors completion of interrupt from 3743vector(s) that i8254 injects. Reinject mode dequeues a tick and injects its 3744interrupt whenever there isn't a pending interrupt from i8254. 3745!reinject mode injects an interrupt as soon as a tick arrives. 3746 3747:: 3748 3749 struct kvm_reinject_control { 3750 __u8 pit_reinject; 3751 __u8 reserved[31]; 3752 }; 3753 3754pit_reinject = 0 (!reinject mode) is recommended, unless running an old 3755operating system that uses the PIT for timing (e.g. Linux 2.4.x). 3756 37574.100 KVM_PPC_CONFIGURE_V3_MMU 3758------------------------------ 3759 3760:Capability: KVM_CAP_PPC_RADIX_MMU or KVM_CAP_PPC_HASH_MMU_V3 3761:Architectures: ppc 3762:Type: vm ioctl 3763:Parameters: struct kvm_ppc_mmuv3_cfg (in) 3764:Returns: 0 on success, 3765 -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read, 3766 -EINVAL if the configuration is invalid 3767 3768This ioctl controls whether the guest will use radix or HPT (hashed 3769page table) translation, and sets the pointer to the process table for 3770the guest. 3771 3772:: 3773 3774 struct kvm_ppc_mmuv3_cfg { 3775 __u64 flags; 3776 __u64 process_table; 3777 }; 3778 3779There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and 3780KVM_PPC_MMUV3_GTSE. KVM_PPC_MMUV3_RADIX, if set, configures the guest 3781to use radix tree translation, and if clear, to use HPT translation. 3782KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest 3783to be able to use the global TLB and SLB invalidation instructions; 3784if clear, the guest may not use these instructions. 3785 3786The process_table field specifies the address and size of the guest 3787process table, which is in the guest's space. This field is formatted 3788as the second doubleword of the partition table entry, as defined in 3789the Power ISA V3.00, Book III section 5.7.6.1. 3790 37914.101 KVM_PPC_GET_RMMU_INFO 3792--------------------------- 3793 3794:Capability: KVM_CAP_PPC_RADIX_MMU 3795:Architectures: ppc 3796:Type: vm ioctl 3797:Parameters: struct kvm_ppc_rmmu_info (out) 3798:Returns: 0 on success, 3799 -EFAULT if struct kvm_ppc_rmmu_info cannot be written, 3800 -EINVAL if no useful information can be returned 3801 3802This ioctl returns a structure containing two things: (a) a list 3803containing supported radix tree geometries, and (b) a list that maps 3804page sizes to put in the "AP" (actual page size) field for the tlbie 3805(TLB invalidate entry) instruction. 3806 3807:: 3808 3809 struct kvm_ppc_rmmu_info { 3810 struct kvm_ppc_radix_geom { 3811 __u8 page_shift; 3812 __u8 level_bits[4]; 3813 __u8 pad[3]; 3814 } geometries[8]; 3815 __u32 ap_encodings[8]; 3816 }; 3817 3818The geometries[] field gives up to 8 supported geometries for the 3819radix page table, in terms of the log base 2 of the smallest page 3820size, and the number of bits indexed at each level of the tree, from 3821the PTE level up to the PGD level in that order. Any unused entries 3822will have 0 in the page_shift field. 3823 3824The ap_encodings gives the supported page sizes and their AP field 3825encodings, encoded with the AP value in the top 3 bits and the log 3826base 2 of the page size in the bottom 6 bits. 3827 38284.102 KVM_PPC_RESIZE_HPT_PREPARE 3829-------------------------------- 3830 3831:Capability: KVM_CAP_SPAPR_RESIZE_HPT 3832:Architectures: powerpc 3833:Type: vm ioctl 3834:Parameters: struct kvm_ppc_resize_hpt (in) 3835:Returns: 0 on successful completion, 3836 >0 if a new HPT is being prepared, the value is an estimated 3837 number of milliseconds until preparation is complete, 3838 -EFAULT if struct kvm_reinject_control cannot be read, 3839 -EINVAL if the supplied shift or flags are invalid, 3840 -ENOMEM if unable to allocate the new HPT, 3841 -ENOSPC if there was a hash collision 3842 3843:: 3844 3845 struct kvm_ppc_rmmu_info { 3846 struct kvm_ppc_radix_geom { 3847 __u8 page_shift; 3848 __u8 level_bits[4]; 3849 __u8 pad[3]; 3850 } geometries[8]; 3851 __u32 ap_encodings[8]; 3852 }; 3853 3854The geometries[] field gives up to 8 supported geometries for the 3855radix page table, in terms of the log base 2 of the smallest page 3856size, and the number of bits indexed at each level of the tree, from 3857the PTE level up to the PGD level in that order. Any unused entries 3858will have 0 in the page_shift field. 3859 3860The ap_encodings gives the supported page sizes and their AP field 3861encodings, encoded with the AP value in the top 3 bits and the log 3862base 2 of the page size in the bottom 6 bits. 3863 38644.102 KVM_PPC_RESIZE_HPT_PREPARE 3865-------------------------------- 3866 3867:Capability: KVM_CAP_SPAPR_RESIZE_HPT 3868:Architectures: powerpc 3869:Type: vm ioctl 3870:Parameters: struct kvm_ppc_resize_hpt (in) 3871:Returns: 0 on successful completion, 3872 >0 if a new HPT is being prepared, the value is an estimated 3873 number of milliseconds until preparation is complete, 3874 -EFAULT if struct kvm_reinject_control cannot be read, 3875 -EINVAL if the supplied shift or flags are invalid,when moving existing 3876 HPT entries to the new HPT, 3877 -EIO on other error conditions 3878 3879Used to implement the PAPR extension for runtime resizing of a guest's 3880Hashed Page Table (HPT). Specifically this starts, stops or monitors 3881the preparation of a new potential HPT for the guest, essentially 3882implementing the H_RESIZE_HPT_PREPARE hypercall. 3883 3884If called with shift > 0 when there is no pending HPT for the guest, 3885this begins preparation of a new pending HPT of size 2^(shift) bytes. 3886It then returns a positive integer with the estimated number of 3887milliseconds until preparation is complete. 3888 3889If called when there is a pending HPT whose size does not match that 3890requested in the parameters, discards the existing pending HPT and 3891creates a new one as above. 3892 3893If called when there is a pending HPT of the size requested, will: 3894 3895 * If preparation of the pending HPT is already complete, return 0 3896 * If preparation of the pending HPT has failed, return an error 3897 code, then discard the pending HPT. 3898 * If preparation of the pending HPT is still in progress, return an 3899 estimated number of milliseconds until preparation is complete. 3900 3901If called with shift == 0, discards any currently pending HPT and 3902returns 0 (i.e. cancels any in-progress preparation). 3903 3904flags is reserved for future expansion, currently setting any bits in 3905flags will result in an -EINVAL. 3906 3907Normally this will be called repeatedly with the same parameters until 3908it returns <= 0. The first call will initiate preparation, subsequent 3909ones will monitor preparation until it completes or fails. 3910 3911:: 3912 3913 struct kvm_ppc_resize_hpt { 3914 __u64 flags; 3915 __u32 shift; 3916 __u32 pad; 3917 }; 3918 39194.103 KVM_PPC_RESIZE_HPT_COMMIT 3920------------------------------- 3921 3922:Capability: KVM_CAP_SPAPR_RESIZE_HPT 3923:Architectures: powerpc 3924:Type: vm ioctl 3925:Parameters: struct kvm_ppc_resize_hpt (in) 3926:Returns: 0 on successful completion, 3927 -EFAULT if struct kvm_reinject_control cannot be read, 3928 -EINVAL if the supplied shift or flags are invalid, 3929 -ENXIO is there is no pending HPT, or the pending HPT doesn't 3930 have the requested size, 3931 -EBUSY if the pending HPT is not fully prepared, 3932 -ENOSPC if there was a hash collision when moving existing 3933 HPT entries to the new HPT, 3934 -EIO on other error conditions 3935 3936Used to implement the PAPR extension for runtime resizing of a guest's 3937Hashed Page Table (HPT). Specifically this requests that the guest be 3938transferred to working with the new HPT, essentially implementing the 3939H_RESIZE_HPT_COMMIT hypercall. 3940 3941This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has 3942returned 0 with the same parameters. In other cases 3943KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or 3944-EBUSY, though others may be possible if the preparation was started, 3945but failed). 3946 3947This will have undefined effects on the guest if it has not already 3948placed itself in a quiescent state where no vcpu will make MMU enabled 3949memory accesses. 3950 3951On succsful completion, the pending HPT will become the guest's active 3952HPT and the previous HPT will be discarded. 3953 3954On failure, the guest will still be operating on its previous HPT. 3955 3956:: 3957 3958 struct kvm_ppc_resize_hpt { 3959 __u64 flags; 3960 __u32 shift; 3961 __u32 pad; 3962 }; 3963 39644.104 KVM_X86_GET_MCE_CAP_SUPPORTED 3965----------------------------------- 3966 3967:Capability: KVM_CAP_MCE 3968:Architectures: x86 3969:Type: system ioctl 3970:Parameters: u64 mce_cap (out) 3971:Returns: 0 on success, -1 on error 3972 3973Returns supported MCE capabilities. The u64 mce_cap parameter 3974has the same format as the MSR_IA32_MCG_CAP register. Supported 3975capabilities will have the corresponding bits set. 3976 39774.105 KVM_X86_SETUP_MCE 3978----------------------- 3979 3980:Capability: KVM_CAP_MCE 3981:Architectures: x86 3982:Type: vcpu ioctl 3983:Parameters: u64 mcg_cap (in) 3984:Returns: 0 on success, 3985 -EFAULT if u64 mcg_cap cannot be read, 3986 -EINVAL if the requested number of banks is invalid, 3987 -EINVAL if requested MCE capability is not supported. 3988 3989Initializes MCE support for use. The u64 mcg_cap parameter 3990has the same format as the MSR_IA32_MCG_CAP register and 3991specifies which capabilities should be enabled. The maximum 3992supported number of error-reporting banks can be retrieved when 3993checking for KVM_CAP_MCE. The supported capabilities can be 3994retrieved with KVM_X86_GET_MCE_CAP_SUPPORTED. 3995 39964.106 KVM_X86_SET_MCE 3997--------------------- 3998 3999:Capability: KVM_CAP_MCE 4000:Architectures: x86 4001:Type: vcpu ioctl 4002:Parameters: struct kvm_x86_mce (in) 4003:Returns: 0 on success, 4004 -EFAULT if struct kvm_x86_mce cannot be read, 4005 -EINVAL if the bank number is invalid, 4006 -EINVAL if VAL bit is not set in status field. 4007 4008Inject a machine check error (MCE) into the guest. The input 4009parameter is:: 4010 4011 struct kvm_x86_mce { 4012 __u64 status; 4013 __u64 addr; 4014 __u64 misc; 4015 __u64 mcg_status; 4016 __u8 bank; 4017 __u8 pad1[7]; 4018 __u64 pad2[3]; 4019 }; 4020 4021If the MCE being reported is an uncorrected error, KVM will 4022inject it as an MCE exception into the guest. If the guest 4023MCG_STATUS register reports that an MCE is in progress, KVM 4024causes an KVM_EXIT_SHUTDOWN vmexit. 4025 4026Otherwise, if the MCE is a corrected error, KVM will just 4027store it in the corresponding bank (provided this bank is 4028not holding a previously reported uncorrected error). 4029 40304.107 KVM_S390_GET_CMMA_BITS 4031---------------------------- 4032 4033:Capability: KVM_CAP_S390_CMMA_MIGRATION 4034:Architectures: s390 4035:Type: vm ioctl 4036:Parameters: struct kvm_s390_cmma_log (in, out) 4037:Returns: 0 on success, a negative value on error 4038 4039Errors: 4040 4041 ====== ============================================================= 4042 ENOMEM not enough memory can be allocated to complete the task 4043 ENXIO if CMMA is not enabled 4044 EINVAL if KVM_S390_CMMA_PEEK is not set but migration mode was not enabled 4045 EINVAL if KVM_S390_CMMA_PEEK is not set but dirty tracking has been 4046 disabled (and thus migration mode was automatically disabled) 4047 EFAULT if the userspace address is invalid or if no page table is 4048 present for the addresses (e.g. when using hugepages). 4049 ====== ============================================================= 4050 4051This ioctl is used to get the values of the CMMA bits on the s390 4052architecture. It is meant to be used in two scenarios: 4053 4054- During live migration to save the CMMA values. Live migration needs 4055 to be enabled via the KVM_REQ_START_MIGRATION VM property. 4056- To non-destructively peek at the CMMA values, with the flag 4057 KVM_S390_CMMA_PEEK set. 4058 4059The ioctl takes parameters via the kvm_s390_cmma_log struct. The desired 4060values are written to a buffer whose location is indicated via the "values" 4061member in the kvm_s390_cmma_log struct. The values in the input struct are 4062also updated as needed. 4063 4064Each CMMA value takes up one byte. 4065 4066:: 4067 4068 struct kvm_s390_cmma_log { 4069 __u64 start_gfn; 4070 __u32 count; 4071 __u32 flags; 4072 union { 4073 __u64 remaining; 4074 __u64 mask; 4075 }; 4076 __u64 values; 4077 }; 4078 4079start_gfn is the number of the first guest frame whose CMMA values are 4080to be retrieved, 4081 4082count is the length of the buffer in bytes, 4083 4084values points to the buffer where the result will be written to. 4085 4086If count is greater than KVM_S390_SKEYS_MAX, then it is considered to be 4087KVM_S390_SKEYS_MAX. KVM_S390_SKEYS_MAX is re-used for consistency with 4088other ioctls. 4089 4090The result is written in the buffer pointed to by the field values, and 4091the values of the input parameter are updated as follows. 4092 4093Depending on the flags, different actions are performed. The only 4094supported flag so far is KVM_S390_CMMA_PEEK. 4095 4096The default behaviour if KVM_S390_CMMA_PEEK is not set is: 4097start_gfn will indicate the first page frame whose CMMA bits were dirty. 4098It is not necessarily the same as the one passed as input, as clean pages 4099are skipped. 4100 4101count will indicate the number of bytes actually written in the buffer. 4102It can (and very often will) be smaller than the input value, since the 4103buffer is only filled until 16 bytes of clean values are found (which 4104are then not copied in the buffer). Since a CMMA migration block needs 4105the base address and the length, for a total of 16 bytes, we will send 4106back some clean data if there is some dirty data afterwards, as long as 4107the size of the clean data does not exceed the size of the header. This 4108allows to minimize the amount of data to be saved or transferred over 4109the network at the expense of more roundtrips to userspace. The next 4110invocation of the ioctl will skip over all the clean values, saving 4111potentially more than just the 16 bytes we found. 4112 4113If KVM_S390_CMMA_PEEK is set: 4114the existing storage attributes are read even when not in migration 4115mode, and no other action is performed; 4116 4117the output start_gfn will be equal to the input start_gfn, 4118 4119the output count will be equal to the input count, except if the end of 4120memory has been reached. 4121 4122In both cases: 4123the field "remaining" will indicate the total number of dirty CMMA values 4124still remaining, or 0 if KVM_S390_CMMA_PEEK is set and migration mode is 4125not enabled. 4126 4127mask is unused. 4128 4129values points to the userspace buffer where the result will be stored. 4130 41314.108 KVM_S390_SET_CMMA_BITS 4132---------------------------- 4133 4134:Capability: KVM_CAP_S390_CMMA_MIGRATION 4135:Architectures: s390 4136:Type: vm ioctl 4137:Parameters: struct kvm_s390_cmma_log (in) 4138:Returns: 0 on success, a negative value on error 4139 4140This ioctl is used to set the values of the CMMA bits on the s390 4141architecture. It is meant to be used during live migration to restore 4142the CMMA values, but there are no restrictions on its use. 4143The ioctl takes parameters via the kvm_s390_cmma_values struct. 4144Each CMMA value takes up one byte. 4145 4146:: 4147 4148 struct kvm_s390_cmma_log { 4149 __u64 start_gfn; 4150 __u32 count; 4151 __u32 flags; 4152 union { 4153 __u64 remaining; 4154 __u64 mask; 4155 }; 4156 __u64 values; 4157 }; 4158 4159start_gfn indicates the starting guest frame number, 4160 4161count indicates how many values are to be considered in the buffer, 4162 4163flags is not used and must be 0. 4164 4165mask indicates which PGSTE bits are to be considered. 4166 4167remaining is not used. 4168 4169values points to the buffer in userspace where to store the values. 4170 4171This ioctl can fail with -ENOMEM if not enough memory can be allocated to 4172complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if 4173the count field is too large (e.g. more than KVM_S390_CMMA_SIZE_MAX) or 4174if the flags field was not 0, with -EFAULT if the userspace address is 4175invalid, if invalid pages are written to (e.g. after the end of memory) 4176or if no page table is present for the addresses (e.g. when using 4177hugepages). 4178 41794.109 KVM_PPC_GET_CPU_CHAR 4180-------------------------- 4181 4182:Capability: KVM_CAP_PPC_GET_CPU_CHAR 4183:Architectures: powerpc 4184:Type: vm ioctl 4185:Parameters: struct kvm_ppc_cpu_char (out) 4186:Returns: 0 on successful completion, 4187 -EFAULT if struct kvm_ppc_cpu_char cannot be written 4188 4189This ioctl gives userspace information about certain characteristics 4190of the CPU relating to speculative execution of instructions and 4191possible information leakage resulting from speculative execution (see 4192CVE-2017-5715, CVE-2017-5753 and CVE-2017-5754). The information is 4193returned in struct kvm_ppc_cpu_char, which looks like this:: 4194 4195 struct kvm_ppc_cpu_char { 4196 __u64 character; /* characteristics of the CPU */ 4197 __u64 behaviour; /* recommended software behaviour */ 4198 __u64 character_mask; /* valid bits in character */ 4199 __u64 behaviour_mask; /* valid bits in behaviour */ 4200 }; 4201 4202For extensibility, the character_mask and behaviour_mask fields 4203indicate which bits of character and behaviour have been filled in by 4204the kernel. If the set of defined bits is extended in future then 4205userspace will be able to tell whether it is running on a kernel that 4206knows about the new bits. 4207 4208The character field describes attributes of the CPU which can help 4209with preventing inadvertent information disclosure - specifically, 4210whether there is an instruction to flash-invalidate the L1 data cache 4211(ori 30,30,0 or mtspr SPRN_TRIG2,rN), whether the L1 data cache is set 4212to a mode where entries can only be used by the thread that created 4213them, whether the bcctr[l] instruction prevents speculation, and 4214whether a speculation barrier instruction (ori 31,31,0) is provided. 4215 4216The behaviour field describes actions that software should take to 4217prevent inadvertent information disclosure, and thus describes which 4218vulnerabilities the hardware is subject to; specifically whether the 4219L1 data cache should be flushed when returning to user mode from the 4220kernel, and whether a speculation barrier should be placed between an 4221array bounds check and the array access. 4222 4223These fields use the same bit definitions as the new 4224H_GET_CPU_CHARACTERISTICS hypercall. 4225 42264.110 KVM_MEMORY_ENCRYPT_OP 4227--------------------------- 4228 4229:Capability: basic 4230:Architectures: x86 4231:Type: vm 4232:Parameters: an opaque platform specific structure (in/out) 4233:Returns: 0 on success; -1 on error 4234 4235If the platform supports creating encrypted VMs then this ioctl can be used 4236for issuing platform-specific memory encryption commands to manage those 4237encrypted VMs. 4238 4239Currently, this ioctl is used for issuing Secure Encrypted Virtualization 4240(SEV) commands on AMD Processors. The SEV commands are defined in 4241Documentation/virt/kvm/amd-memory-encryption.rst. 4242 42434.111 KVM_MEMORY_ENCRYPT_REG_REGION 4244----------------------------------- 4245 4246:Capability: basic 4247:Architectures: x86 4248:Type: system 4249:Parameters: struct kvm_enc_region (in) 4250:Returns: 0 on success; -1 on error 4251 4252This ioctl can be used to register a guest memory region which may 4253contain encrypted data (e.g. guest RAM, SMRAM etc). 4254 4255It is used in the SEV-enabled guest. When encryption is enabled, a guest 4256memory region may contain encrypted data. The SEV memory encryption 4257engine uses a tweak such that two identical plaintext pages, each at 4258different locations will have differing ciphertexts. So swapping or 4259moving ciphertext of those pages will not result in plaintext being 4260swapped. So relocating (or migrating) physical backing pages for the SEV 4261guest will require some additional steps. 4262 4263Note: The current SEV key management spec does not provide commands to 4264swap or migrate (move) ciphertext pages. Hence, for now we pin the guest 4265memory region registered with the ioctl. 4266 42674.112 KVM_MEMORY_ENCRYPT_UNREG_REGION 4268------------------------------------- 4269 4270:Capability: basic 4271:Architectures: x86 4272:Type: system 4273:Parameters: struct kvm_enc_region (in) 4274:Returns: 0 on success; -1 on error 4275 4276This ioctl can be used to unregister the guest memory region registered 4277with KVM_MEMORY_ENCRYPT_REG_REGION ioctl above. 4278 42794.113 KVM_HYPERV_EVENTFD 4280------------------------ 4281 4282:Capability: KVM_CAP_HYPERV_EVENTFD 4283:Architectures: x86 4284:Type: vm ioctl 4285:Parameters: struct kvm_hyperv_eventfd (in) 4286 4287This ioctl (un)registers an eventfd to receive notifications from the guest on 4288the specified Hyper-V connection id through the SIGNAL_EVENT hypercall, without 4289causing a user exit. SIGNAL_EVENT hypercall with non-zero event flag number 4290(bits 24-31) still triggers a KVM_EXIT_HYPERV_HCALL user exit. 4291 4292:: 4293 4294 struct kvm_hyperv_eventfd { 4295 __u32 conn_id; 4296 __s32 fd; 4297 __u32 flags; 4298 __u32 padding[3]; 4299 }; 4300 4301The conn_id field should fit within 24 bits:: 4302 4303 #define KVM_HYPERV_CONN_ID_MASK 0x00ffffff 4304 4305The acceptable values for the flags field are:: 4306 4307 #define KVM_HYPERV_EVENTFD_DEASSIGN (1 << 0) 4308 4309:Returns: 0 on success, 4310 -EINVAL if conn_id or flags is outside the allowed range, 4311 -ENOENT on deassign if the conn_id isn't registered, 4312 -EEXIST on assign if the conn_id is already registered 4313 43144.114 KVM_GET_NESTED_STATE 4315-------------------------- 4316 4317:Capability: KVM_CAP_NESTED_STATE 4318:Architectures: x86 4319:Type: vcpu ioctl 4320:Parameters: struct kvm_nested_state (in/out) 4321:Returns: 0 on success, -1 on error 4322 4323Errors: 4324 4325 ===== ============================================================= 4326 E2BIG the total state size exceeds the value of 'size' specified by 4327 the user; the size required will be written into size. 4328 ===== ============================================================= 4329 4330:: 4331 4332 struct kvm_nested_state { 4333 __u16 flags; 4334 __u16 format; 4335 __u32 size; 4336 4337 union { 4338 struct kvm_vmx_nested_state_hdr vmx; 4339 struct kvm_svm_nested_state_hdr svm; 4340 4341 /* Pad the header to 128 bytes. */ 4342 __u8 pad[120]; 4343 } hdr; 4344 4345 union { 4346 struct kvm_vmx_nested_state_data vmx[0]; 4347 struct kvm_svm_nested_state_data svm[0]; 4348 } data; 4349 }; 4350 4351 #define KVM_STATE_NESTED_GUEST_MODE 0x00000001 4352 #define KVM_STATE_NESTED_RUN_PENDING 0x00000002 4353 #define KVM_STATE_NESTED_EVMCS 0x00000004 4354 4355 #define KVM_STATE_NESTED_FORMAT_VMX 0 4356 #define KVM_STATE_NESTED_FORMAT_SVM 1 4357 4358 #define KVM_STATE_NESTED_VMX_VMCS_SIZE 0x1000 4359 4360 #define KVM_STATE_NESTED_VMX_SMM_GUEST_MODE 0x00000001 4361 #define KVM_STATE_NESTED_VMX_SMM_VMXON 0x00000002 4362 4363 #define KVM_STATE_VMX_PREEMPTION_TIMER_DEADLINE 0x00000001 4364 4365 struct kvm_vmx_nested_state_hdr { 4366 __u64 vmxon_pa; 4367 __u64 vmcs12_pa; 4368 4369 struct { 4370 __u16 flags; 4371 } smm; 4372 4373 __u32 flags; 4374 __u64 preemption_timer_deadline; 4375 }; 4376 4377 struct kvm_vmx_nested_state_data { 4378 __u8 vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE]; 4379 __u8 shadow_vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE]; 4380 }; 4381 4382This ioctl copies the vcpu's nested virtualization state from the kernel to 4383userspace. 4384 4385The maximum size of the state can be retrieved by passing KVM_CAP_NESTED_STATE 4386to the KVM_CHECK_EXTENSION ioctl(). 4387 43884.115 KVM_SET_NESTED_STATE 4389-------------------------- 4390 4391:Capability: KVM_CAP_NESTED_STATE 4392:Architectures: x86 4393:Type: vcpu ioctl 4394:Parameters: struct kvm_nested_state (in) 4395:Returns: 0 on success, -1 on error 4396 4397This copies the vcpu's kvm_nested_state struct from userspace to the kernel. 4398For the definition of struct kvm_nested_state, see KVM_GET_NESTED_STATE. 4399 44004.116 KVM_(UN)REGISTER_COALESCED_MMIO 4401------------------------------------- 4402 4403:Capability: KVM_CAP_COALESCED_MMIO (for coalesced mmio) 4404 KVM_CAP_COALESCED_PIO (for coalesced pio) 4405:Architectures: all 4406:Type: vm ioctl 4407:Parameters: struct kvm_coalesced_mmio_zone 4408:Returns: 0 on success, < 0 on error 4409 4410Coalesced I/O is a performance optimization that defers hardware 4411register write emulation so that userspace exits are avoided. It is 4412typically used to reduce the overhead of emulating frequently accessed 4413hardware registers. 4414 4415When a hardware register is configured for coalesced I/O, write accesses 4416do not exit to userspace and their value is recorded in a ring buffer 4417that is shared between kernel and userspace. 4418 4419Coalesced I/O is used if one or more write accesses to a hardware 4420register can be deferred until a read or a write to another hardware 4421register on the same device. This last access will cause a vmexit and 4422userspace will process accesses from the ring buffer before emulating 4423it. That will avoid exiting to userspace on repeated writes. 4424 4425Coalesced pio is based on coalesced mmio. There is little difference 4426between coalesced mmio and pio except that coalesced pio records accesses 4427to I/O ports. 4428 44294.117 KVM_CLEAR_DIRTY_LOG (vm ioctl) 4430------------------------------------ 4431 4432:Capability: KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 4433:Architectures: x86, arm, arm64, mips 4434:Type: vm ioctl 4435:Parameters: struct kvm_dirty_log (in) 4436:Returns: 0 on success, -1 on error 4437 4438:: 4439 4440 /* for KVM_CLEAR_DIRTY_LOG */ 4441 struct kvm_clear_dirty_log { 4442 __u32 slot; 4443 __u32 num_pages; 4444 __u64 first_page; 4445 union { 4446 void __user *dirty_bitmap; /* one bit per page */ 4447 __u64 padding; 4448 }; 4449 }; 4450 4451The ioctl clears the dirty status of pages in a memory slot, according to 4452the bitmap that is passed in struct kvm_clear_dirty_log's dirty_bitmap 4453field. Bit 0 of the bitmap corresponds to page "first_page" in the 4454memory slot, and num_pages is the size in bits of the input bitmap. 4455first_page must be a multiple of 64; num_pages must also be a multiple of 445664 unless first_page + num_pages is the size of the memory slot. For each 4457bit that is set in the input bitmap, the corresponding page is marked "clean" 4458in KVM's dirty bitmap, and dirty tracking is re-enabled for that page 4459(for example via write-protection, or by clearing the dirty bit in 4460a page table entry). 4461 4462If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 specifies 4463the address space for which you want to return the dirty bitmap. 4464They must be less than the value that KVM_CHECK_EXTENSION returns for 4465the KVM_CAP_MULTI_ADDRESS_SPACE capability. 4466 4467This ioctl is mostly useful when KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 4468is enabled; for more information, see the description of the capability. 4469However, it can always be used as long as KVM_CHECK_EXTENSION confirms 4470that KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is present. 4471 44724.118 KVM_GET_SUPPORTED_HV_CPUID 4473-------------------------------- 4474 4475:Capability: KVM_CAP_HYPERV_CPUID 4476:Architectures: x86 4477:Type: vcpu ioctl 4478:Parameters: struct kvm_cpuid2 (in/out) 4479:Returns: 0 on success, -1 on error 4480 4481:: 4482 4483 struct kvm_cpuid2 { 4484 __u32 nent; 4485 __u32 padding; 4486 struct kvm_cpuid_entry2 entries[0]; 4487 }; 4488 4489 struct kvm_cpuid_entry2 { 4490 __u32 function; 4491 __u32 index; 4492 __u32 flags; 4493 __u32 eax; 4494 __u32 ebx; 4495 __u32 ecx; 4496 __u32 edx; 4497 __u32 padding[3]; 4498 }; 4499 4500This ioctl returns x86 cpuid features leaves related to Hyper-V emulation in 4501KVM. Userspace can use the information returned by this ioctl to construct 4502cpuid information presented to guests consuming Hyper-V enlightenments (e.g. 4503Windows or Hyper-V guests). 4504 4505CPUID feature leaves returned by this ioctl are defined by Hyper-V Top Level 4506Functional Specification (TLFS). These leaves can't be obtained with 4507KVM_GET_SUPPORTED_CPUID ioctl because some of them intersect with KVM feature 4508leaves (0x40000000, 0x40000001). 4509 4510Currently, the following list of CPUID leaves are returned: 4511 - HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS 4512 - HYPERV_CPUID_INTERFACE 4513 - HYPERV_CPUID_VERSION 4514 - HYPERV_CPUID_FEATURES 4515 - HYPERV_CPUID_ENLIGHTMENT_INFO 4516 - HYPERV_CPUID_IMPLEMENT_LIMITS 4517 - HYPERV_CPUID_NESTED_FEATURES 4518 - HYPERV_CPUID_SYNDBG_VENDOR_AND_MAX_FUNCTIONS 4519 - HYPERV_CPUID_SYNDBG_INTERFACE 4520 - HYPERV_CPUID_SYNDBG_PLATFORM_CAPABILITIES 4521 4522HYPERV_CPUID_NESTED_FEATURES leaf is only exposed when Enlightened VMCS was 4523enabled on the corresponding vCPU (KVM_CAP_HYPERV_ENLIGHTENED_VMCS). 4524 4525Userspace invokes KVM_GET_SUPPORTED_HV_CPUID by passing a kvm_cpuid2 structure 4526with the 'nent' field indicating the number of entries in the variable-size 4527array 'entries'. If the number of entries is too low to describe all Hyper-V 4528feature leaves, an error (E2BIG) is returned. If the number is more or equal 4529to the number of Hyper-V feature leaves, the 'nent' field is adjusted to the 4530number of valid entries in the 'entries' array, which is then filled. 4531 4532'index' and 'flags' fields in 'struct kvm_cpuid_entry2' are currently reserved, 4533userspace should not expect to get any particular value there. 4534 45354.119 KVM_ARM_VCPU_FINALIZE 4536--------------------------- 4537 4538:Architectures: arm, arm64 4539:Type: vcpu ioctl 4540:Parameters: int feature (in) 4541:Returns: 0 on success, -1 on error 4542 4543Errors: 4544 4545 ====== ============================================================== 4546 EPERM feature not enabled, needs configuration, or already finalized 4547 EINVAL feature unknown or not present 4548 ====== ============================================================== 4549 4550Recognised values for feature: 4551 4552 ===== =========================================== 4553 arm64 KVM_ARM_VCPU_SVE (requires KVM_CAP_ARM_SVE) 4554 ===== =========================================== 4555 4556Finalizes the configuration of the specified vcpu feature. 4557 4558The vcpu must already have been initialised, enabling the affected feature, by 4559means of a successful KVM_ARM_VCPU_INIT call with the appropriate flag set in 4560features[]. 4561 4562For affected vcpu features, this is a mandatory step that must be performed 4563before the vcpu is fully usable. 4564 4565Between KVM_ARM_VCPU_INIT and KVM_ARM_VCPU_FINALIZE, the feature may be 4566configured by use of ioctls such as KVM_SET_ONE_REG. The exact configuration 4567that should be performaned and how to do it are feature-dependent. 4568 4569Other calls that depend on a particular feature being finalized, such as 4570KVM_RUN, KVM_GET_REG_LIST, KVM_GET_ONE_REG and KVM_SET_ONE_REG, will fail with 4571-EPERM unless the feature has already been finalized by means of a 4572KVM_ARM_VCPU_FINALIZE call. 4573 4574See KVM_ARM_VCPU_INIT for details of vcpu features that require finalization 4575using this ioctl. 4576 45774.120 KVM_SET_PMU_EVENT_FILTER 4578------------------------------ 4579 4580:Capability: KVM_CAP_PMU_EVENT_FILTER 4581:Architectures: x86 4582:Type: vm ioctl 4583:Parameters: struct kvm_pmu_event_filter (in) 4584:Returns: 0 on success, -1 on error 4585 4586:: 4587 4588 struct kvm_pmu_event_filter { 4589 __u32 action; 4590 __u32 nevents; 4591 __u32 fixed_counter_bitmap; 4592 __u32 flags; 4593 __u32 pad[4]; 4594 __u64 events[0]; 4595 }; 4596 4597This ioctl restricts the set of PMU events that the guest can program. 4598The argument holds a list of events which will be allowed or denied. 4599The eventsel+umask of each event the guest attempts to program is compared 4600against the events field to determine whether the guest should have access. 4601The events field only controls general purpose counters; fixed purpose 4602counters are controlled by the fixed_counter_bitmap. 4603 4604No flags are defined yet, the field must be zero. 4605 4606Valid values for 'action':: 4607 4608 #define KVM_PMU_EVENT_ALLOW 0 4609 #define KVM_PMU_EVENT_DENY 1 4610 46114.121 KVM_PPC_SVM_OFF 4612--------------------- 4613 4614:Capability: basic 4615:Architectures: powerpc 4616:Type: vm ioctl 4617:Parameters: none 4618:Returns: 0 on successful completion, 4619 4620Errors: 4621 4622 ====== ================================================================ 4623 EINVAL if ultravisor failed to terminate the secure guest 4624 ENOMEM if hypervisor failed to allocate new radix page tables for guest 4625 ====== ================================================================ 4626 4627This ioctl is used to turn off the secure mode of the guest or transition 4628the guest from secure mode to normal mode. This is invoked when the guest 4629is reset. This has no effect if called for a normal guest. 4630 4631This ioctl issues an ultravisor call to terminate the secure guest, 4632unpins the VPA pages and releases all the device pages that are used to 4633track the secure pages by hypervisor. 4634 46354.122 KVM_S390_NORMAL_RESET 4636--------------------------- 4637 4638:Capability: KVM_CAP_S390_VCPU_RESETS 4639:Architectures: s390 4640:Type: vcpu ioctl 4641:Parameters: none 4642:Returns: 0 4643 4644This ioctl resets VCPU registers and control structures according to 4645the cpu reset definition in the POP (Principles Of Operation). 4646 46474.123 KVM_S390_INITIAL_RESET 4648---------------------------- 4649 4650:Capability: none 4651:Architectures: s390 4652:Type: vcpu ioctl 4653:Parameters: none 4654:Returns: 0 4655 4656This ioctl resets VCPU registers and control structures according to 4657the initial cpu reset definition in the POP. However, the cpu is not 4658put into ESA mode. This reset is a superset of the normal reset. 4659 46604.124 KVM_S390_CLEAR_RESET 4661-------------------------- 4662 4663:Capability: KVM_CAP_S390_VCPU_RESETS 4664:Architectures: s390 4665:Type: vcpu ioctl 4666:Parameters: none 4667:Returns: 0 4668 4669This ioctl resets VCPU registers and control structures according to 4670the clear cpu reset definition in the POP. However, the cpu is not put 4671into ESA mode. This reset is a superset of the initial reset. 4672 4673 46744.125 KVM_S390_PV_COMMAND 4675------------------------- 4676 4677:Capability: KVM_CAP_S390_PROTECTED 4678:Architectures: s390 4679:Type: vm ioctl 4680:Parameters: struct kvm_pv_cmd 4681:Returns: 0 on success, < 0 on error 4682 4683:: 4684 4685 struct kvm_pv_cmd { 4686 __u32 cmd; /* Command to be executed */ 4687 __u16 rc; /* Ultravisor return code */ 4688 __u16 rrc; /* Ultravisor return reason code */ 4689 __u64 data; /* Data or address */ 4690 __u32 flags; /* flags for future extensions. Must be 0 for now */ 4691 __u32 reserved[3]; 4692 }; 4693 4694cmd values: 4695 4696KVM_PV_ENABLE 4697 Allocate memory and register the VM with the Ultravisor, thereby 4698 donating memory to the Ultravisor that will become inaccessible to 4699 KVM. All existing CPUs are converted to protected ones. After this 4700 command has succeeded, any CPU added via hotplug will become 4701 protected during its creation as well. 4702 4703 Errors: 4704 4705 ===== ============================= 4706 EINTR an unmasked signal is pending 4707 ===== ============================= 4708 4709KVM_PV_DISABLE 4710 4711 Deregister the VM from the Ultravisor and reclaim the memory that 4712 had been donated to the Ultravisor, making it usable by the kernel 4713 again. All registered VCPUs are converted back to non-protected 4714 ones. 4715 4716KVM_PV_VM_SET_SEC_PARMS 4717 Pass the image header from VM memory to the Ultravisor in 4718 preparation of image unpacking and verification. 4719 4720KVM_PV_VM_UNPACK 4721 Unpack (protect and decrypt) a page of the encrypted boot image. 4722 4723KVM_PV_VM_VERIFY 4724 Verify the integrity of the unpacked image. Only if this succeeds, 4725 KVM is allowed to start protected VCPUs. 4726 47274.126 KVM_X86_SET_MSR_FILTER 4728---------------------------- 4729 4730:Capability: KVM_X86_SET_MSR_FILTER 4731:Architectures: x86 4732:Type: vm ioctl 4733:Parameters: struct kvm_msr_filter 4734:Returns: 0 on success, < 0 on error 4735 4736:: 4737 4738 struct kvm_msr_filter_range { 4739 #define KVM_MSR_FILTER_READ (1 << 0) 4740 #define KVM_MSR_FILTER_WRITE (1 << 1) 4741 __u32 flags; 4742 __u32 nmsrs; /* number of msrs in bitmap */ 4743 __u32 base; /* MSR index the bitmap starts at */ 4744 __u8 *bitmap; /* a 1 bit allows the operations in flags, 0 denies */ 4745 }; 4746 4747 #define KVM_MSR_FILTER_MAX_RANGES 16 4748 struct kvm_msr_filter { 4749 #define KVM_MSR_FILTER_DEFAULT_ALLOW (0 << 0) 4750 #define KVM_MSR_FILTER_DEFAULT_DENY (1 << 0) 4751 __u32 flags; 4752 struct kvm_msr_filter_range ranges[KVM_MSR_FILTER_MAX_RANGES]; 4753 }; 4754 4755flags values for ``struct kvm_msr_filter_range``: 4756 4757``KVM_MSR_FILTER_READ`` 4758 4759 Filter read accesses to MSRs using the given bitmap. A 0 in the bitmap 4760 indicates that a read should immediately fail, while a 1 indicates that 4761 a read for a particular MSR should be handled regardless of the default 4762 filter action. 4763 4764``KVM_MSR_FILTER_WRITE`` 4765 4766 Filter write accesses to MSRs using the given bitmap. A 0 in the bitmap 4767 indicates that a write should immediately fail, while a 1 indicates that 4768 a write for a particular MSR should be handled regardless of the default 4769 filter action. 4770 4771``KVM_MSR_FILTER_READ | KVM_MSR_FILTER_WRITE`` 4772 4773 Filter both read and write accesses to MSRs using the given bitmap. A 0 4774 in the bitmap indicates that both reads and writes should immediately fail, 4775 while a 1 indicates that reads and writes for a particular MSR are not 4776 filtered by this range. 4777 4778flags values for ``struct kvm_msr_filter``: 4779 4780``KVM_MSR_FILTER_DEFAULT_ALLOW`` 4781 4782 If no filter range matches an MSR index that is getting accessed, KVM will 4783 fall back to allowing access to the MSR. 4784 4785``KVM_MSR_FILTER_DEFAULT_DENY`` 4786 4787 If no filter range matches an MSR index that is getting accessed, KVM will 4788 fall back to rejecting access to the MSR. In this mode, all MSRs that should 4789 be processed by KVM need to explicitly be marked as allowed in the bitmaps. 4790 4791This ioctl allows user space to define up to 16 bitmaps of MSR ranges to 4792specify whether a certain MSR access should be explicitly filtered for or not. 4793 4794If this ioctl has never been invoked, MSR accesses are not guarded and the 4795default KVM in-kernel emulation behavior is fully preserved. 4796 4797Calling this ioctl with an empty set of ranges (all nmsrs == 0) disables MSR 4798filtering. In that mode, ``KVM_MSR_FILTER_DEFAULT_DENY`` is invalid and causes 4799an error. 4800 4801As soon as the filtering is in place, every MSR access is processed through 4802the filtering except for accesses to the x2APIC MSRs (from 0x800 to 0x8ff); 4803x2APIC MSRs are always allowed, independent of the ``default_allow`` setting, 4804and their behavior depends on the ``X2APIC_ENABLE`` bit of the APIC base 4805register. 4806 4807If a bit is within one of the defined ranges, read and write accesses are 4808guarded by the bitmap's value for the MSR index if the kind of access 4809is included in the ``struct kvm_msr_filter_range`` flags. If no range 4810cover this particular access, the behavior is determined by the flags 4811field in the kvm_msr_filter struct: ``KVM_MSR_FILTER_DEFAULT_ALLOW`` 4812and ``KVM_MSR_FILTER_DEFAULT_DENY``. 4813 4814Each bitmap range specifies a range of MSRs to potentially allow access on. 4815The range goes from MSR index [base .. base+nmsrs]. The flags field 4816indicates whether reads, writes or both reads and writes are filtered 4817by setting a 1 bit in the bitmap for the corresponding MSR index. 4818 4819If an MSR access is not permitted through the filtering, it generates a 4820#GP inside the guest. When combined with KVM_CAP_X86_USER_SPACE_MSR, that 4821allows user space to deflect and potentially handle various MSR accesses 4822into user space. 4823 4824Note, invoking this ioctl with a vCPU is running is inherently racy. However, 4825KVM does guarantee that vCPUs will see either the previous filter or the new 4826filter, e.g. MSRs with identical settings in both the old and new filter will 4827have deterministic behavior. 4828 4829 48305. The kvm_run structure 4831======================== 4832 4833Application code obtains a pointer to the kvm_run structure by 4834mmap()ing a vcpu fd. From that point, application code can control 4835execution by changing fields in kvm_run prior to calling the KVM_RUN 4836ioctl, and obtain information about the reason KVM_RUN returned by 4837looking up structure members. 4838 4839:: 4840 4841 struct kvm_run { 4842 /* in */ 4843 __u8 request_interrupt_window; 4844 4845Request that KVM_RUN return when it becomes possible to inject external 4846interrupts into the guest. Useful in conjunction with KVM_INTERRUPT. 4847 4848:: 4849 4850 __u8 immediate_exit; 4851 4852This field is polled once when KVM_RUN starts; if non-zero, KVM_RUN 4853exits immediately, returning -EINTR. In the common scenario where a 4854signal is used to "kick" a VCPU out of KVM_RUN, this field can be used 4855to avoid usage of KVM_SET_SIGNAL_MASK, which has worse scalability. 4856Rather than blocking the signal outside KVM_RUN, userspace can set up 4857a signal handler that sets run->immediate_exit to a non-zero value. 4858 4859This field is ignored if KVM_CAP_IMMEDIATE_EXIT is not available. 4860 4861:: 4862 4863 __u8 padding1[6]; 4864 4865 /* out */ 4866 __u32 exit_reason; 4867 4868When KVM_RUN has returned successfully (return value 0), this informs 4869application code why KVM_RUN has returned. Allowable values for this 4870field are detailed below. 4871 4872:: 4873 4874 __u8 ready_for_interrupt_injection; 4875 4876If request_interrupt_window has been specified, this field indicates 4877an interrupt can be injected now with KVM_INTERRUPT. 4878 4879:: 4880 4881 __u8 if_flag; 4882 4883The value of the current interrupt flag. Only valid if in-kernel 4884local APIC is not used. 4885 4886:: 4887 4888 __u16 flags; 4889 4890More architecture-specific flags detailing state of the VCPU that may 4891affect the device's behavior. The only currently defined flag is 4892KVM_RUN_X86_SMM, which is valid on x86 machines and is set if the 4893VCPU is in system management mode. 4894 4895:: 4896 4897 /* in (pre_kvm_run), out (post_kvm_run) */ 4898 __u64 cr8; 4899 4900The value of the cr8 register. Only valid if in-kernel local APIC is 4901not used. Both input and output. 4902 4903:: 4904 4905 __u64 apic_base; 4906 4907The value of the APIC BASE msr. Only valid if in-kernel local 4908APIC is not used. Both input and output. 4909 4910:: 4911 4912 union { 4913 /* KVM_EXIT_UNKNOWN */ 4914 struct { 4915 __u64 hardware_exit_reason; 4916 } hw; 4917 4918If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown 4919reasons. Further architecture-specific information is available in 4920hardware_exit_reason. 4921 4922:: 4923 4924 /* KVM_EXIT_FAIL_ENTRY */ 4925 struct { 4926 __u64 hardware_entry_failure_reason; 4927 __u32 cpu; /* if KVM_LAST_CPU */ 4928 } fail_entry; 4929 4930If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due 4931to unknown reasons. Further architecture-specific information is 4932available in hardware_entry_failure_reason. 4933 4934:: 4935 4936 /* KVM_EXIT_EXCEPTION */ 4937 struct { 4938 __u32 exception; 4939 __u32 error_code; 4940 } ex; 4941 4942Unused. 4943 4944:: 4945 4946 /* KVM_EXIT_IO */ 4947 struct { 4948 #define KVM_EXIT_IO_IN 0 4949 #define KVM_EXIT_IO_OUT 1 4950 __u8 direction; 4951 __u8 size; /* bytes */ 4952 __u16 port; 4953 __u32 count; 4954 __u64 data_offset; /* relative to kvm_run start */ 4955 } io; 4956 4957If exit_reason is KVM_EXIT_IO, then the vcpu has 4958executed a port I/O instruction which could not be satisfied by kvm. 4959data_offset describes where the data is located (KVM_EXIT_IO_OUT) or 4960where kvm expects application code to place the data for the next 4961KVM_RUN invocation (KVM_EXIT_IO_IN). Data format is a packed array. 4962 4963:: 4964 4965 /* KVM_EXIT_DEBUG */ 4966 struct { 4967 struct kvm_debug_exit_arch arch; 4968 } debug; 4969 4970If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event 4971for which architecture specific information is returned. 4972 4973:: 4974 4975 /* KVM_EXIT_MMIO */ 4976 struct { 4977 __u64 phys_addr; 4978 __u8 data[8]; 4979 __u32 len; 4980 __u8 is_write; 4981 } mmio; 4982 4983If exit_reason is KVM_EXIT_MMIO, then the vcpu has 4984executed a memory-mapped I/O instruction which could not be satisfied 4985by kvm. The 'data' member contains the written data if 'is_write' is 4986true, and should be filled by application code otherwise. 4987 4988The 'data' member contains, in its first 'len' bytes, the value as it would 4989appear if the VCPU performed a load or store of the appropriate width directly 4990to the byte array. 4991 4992.. note:: 4993 4994 For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR, 4995 KVM_EXIT_EPR, KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR the corresponding 4996 operations are complete (and guest state is consistent) only after userspace 4997 has re-entered the kernel with KVM_RUN. The kernel side will first finish 4998 incomplete operations and then check for pending signals. Userspace 4999 can re-enter the guest with an unmasked signal pending to complete 5000 pending operations. 5001 5002:: 5003 5004 /* KVM_EXIT_HYPERCALL */ 5005 struct { 5006 __u64 nr; 5007 __u64 args[6]; 5008 __u64 ret; 5009 __u32 longmode; 5010 __u32 pad; 5011 } hypercall; 5012 5013Unused. This was once used for 'hypercall to userspace'. To implement 5014such functionality, use KVM_EXIT_IO (x86) or KVM_EXIT_MMIO (all except s390). 5015 5016.. note:: KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO. 5017 5018:: 5019 5020 /* KVM_EXIT_TPR_ACCESS */ 5021 struct { 5022 __u64 rip; 5023 __u32 is_write; 5024 __u32 pad; 5025 } tpr_access; 5026 5027To be documented (KVM_TPR_ACCESS_REPORTING). 5028 5029:: 5030 5031 /* KVM_EXIT_S390_SIEIC */ 5032 struct { 5033 __u8 icptcode; 5034 __u64 mask; /* psw upper half */ 5035 __u64 addr; /* psw lower half */ 5036 __u16 ipa; 5037 __u32 ipb; 5038 } s390_sieic; 5039 5040s390 specific. 5041 5042:: 5043 5044 /* KVM_EXIT_S390_RESET */ 5045 #define KVM_S390_RESET_POR 1 5046 #define KVM_S390_RESET_CLEAR 2 5047 #define KVM_S390_RESET_SUBSYSTEM 4 5048 #define KVM_S390_RESET_CPU_INIT 8 5049 #define KVM_S390_RESET_IPL 16 5050 __u64 s390_reset_flags; 5051 5052s390 specific. 5053 5054:: 5055 5056 /* KVM_EXIT_S390_UCONTROL */ 5057 struct { 5058 __u64 trans_exc_code; 5059 __u32 pgm_code; 5060 } s390_ucontrol; 5061 5062s390 specific. A page fault has occurred for a user controlled virtual 5063machine (KVM_VM_S390_UNCONTROL) on it's host page table that cannot be 5064resolved by the kernel. 5065The program code and the translation exception code that were placed 5066in the cpu's lowcore are presented here as defined by the z Architecture 5067Principles of Operation Book in the Chapter for Dynamic Address Translation 5068(DAT) 5069 5070:: 5071 5072 /* KVM_EXIT_DCR */ 5073 struct { 5074 __u32 dcrn; 5075 __u32 data; 5076 __u8 is_write; 5077 } dcr; 5078 5079Deprecated - was used for 440 KVM. 5080 5081:: 5082 5083 /* KVM_EXIT_OSI */ 5084 struct { 5085 __u64 gprs[32]; 5086 } osi; 5087 5088MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch 5089hypercalls and exit with this exit struct that contains all the guest gprs. 5090 5091If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall. 5092Userspace can now handle the hypercall and when it's done modify the gprs as 5093necessary. Upon guest entry all guest GPRs will then be replaced by the values 5094in this struct. 5095 5096:: 5097 5098 /* KVM_EXIT_PAPR_HCALL */ 5099 struct { 5100 __u64 nr; 5101 __u64 ret; 5102 __u64 args[9]; 5103 } papr_hcall; 5104 5105This is used on 64-bit PowerPC when emulating a pSeries partition, 5106e.g. with the 'pseries' machine type in qemu. It occurs when the 5107guest does a hypercall using the 'sc 1' instruction. The 'nr' field 5108contains the hypercall number (from the guest R3), and 'args' contains 5109the arguments (from the guest R4 - R12). Userspace should put the 5110return code in 'ret' and any extra returned values in args[]. 5111The possible hypercalls are defined in the Power Architecture Platform 5112Requirements (PAPR) document available from www.power.org (free 5113developer registration required to access it). 5114 5115:: 5116 5117 /* KVM_EXIT_S390_TSCH */ 5118 struct { 5119 __u16 subchannel_id; 5120 __u16 subchannel_nr; 5121 __u32 io_int_parm; 5122 __u32 io_int_word; 5123 __u32 ipb; 5124 __u8 dequeued; 5125 } s390_tsch; 5126 5127s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled 5128and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O 5129interrupt for the target subchannel has been dequeued and subchannel_id, 5130subchannel_nr, io_int_parm and io_int_word contain the parameters for that 5131interrupt. ipb is needed for instruction parameter decoding. 5132 5133:: 5134 5135 /* KVM_EXIT_EPR */ 5136 struct { 5137 __u32 epr; 5138 } epr; 5139 5140On FSL BookE PowerPC chips, the interrupt controller has a fast patch 5141interrupt acknowledge path to the core. When the core successfully 5142delivers an interrupt, it automatically populates the EPR register with 5143the interrupt vector number and acknowledges the interrupt inside 5144the interrupt controller. 5145 5146In case the interrupt controller lives in user space, we need to do 5147the interrupt acknowledge cycle through it to fetch the next to be 5148delivered interrupt vector using this exit. 5149 5150It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an 5151external interrupt has just been delivered into the guest. User space 5152should put the acknowledged interrupt vector into the 'epr' field. 5153 5154:: 5155 5156 /* KVM_EXIT_SYSTEM_EVENT */ 5157 struct { 5158 #define KVM_SYSTEM_EVENT_SHUTDOWN 1 5159 #define KVM_SYSTEM_EVENT_RESET 2 5160 #define KVM_SYSTEM_EVENT_CRASH 3 5161 __u32 type; 5162 __u64 flags; 5163 } system_event; 5164 5165If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered 5166a system-level event using some architecture specific mechanism (hypercall 5167or some special instruction). In case of ARM/ARM64, this is triggered using 5168HVC instruction based PSCI call from the vcpu. The 'type' field describes 5169the system-level event type. The 'flags' field describes architecture 5170specific flags for the system-level event. 5171 5172Valid values for 'type' are: 5173 5174 - KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the 5175 VM. Userspace is not obliged to honour this, and if it does honour 5176 this does not need to destroy the VM synchronously (ie it may call 5177 KVM_RUN again before shutdown finally occurs). 5178 - KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM. 5179 As with SHUTDOWN, userspace can choose to ignore the request, or 5180 to schedule the reset to occur in the future and may call KVM_RUN again. 5181 - KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest 5182 has requested a crash condition maintenance. Userspace can choose 5183 to ignore the request, or to gather VM memory core dump and/or 5184 reset/shutdown of the VM. 5185 5186:: 5187 5188 /* KVM_EXIT_IOAPIC_EOI */ 5189 struct { 5190 __u8 vector; 5191 } eoi; 5192 5193Indicates that the VCPU's in-kernel local APIC received an EOI for a 5194level-triggered IOAPIC interrupt. This exit only triggers when the 5195IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled); 5196the userspace IOAPIC should process the EOI and retrigger the interrupt if 5197it is still asserted. Vector is the LAPIC interrupt vector for which the 5198EOI was received. 5199 5200:: 5201 5202 struct kvm_hyperv_exit { 5203 #define KVM_EXIT_HYPERV_SYNIC 1 5204 #define KVM_EXIT_HYPERV_HCALL 2 5205 #define KVM_EXIT_HYPERV_SYNDBG 3 5206 __u32 type; 5207 __u32 pad1; 5208 union { 5209 struct { 5210 __u32 msr; 5211 __u32 pad2; 5212 __u64 control; 5213 __u64 evt_page; 5214 __u64 msg_page; 5215 } synic; 5216 struct { 5217 __u64 input; 5218 __u64 result; 5219 __u64 params[2]; 5220 } hcall; 5221 struct { 5222 __u32 msr; 5223 __u32 pad2; 5224 __u64 control; 5225 __u64 status; 5226 __u64 send_page; 5227 __u64 recv_page; 5228 __u64 pending_page; 5229 } syndbg; 5230 } u; 5231 }; 5232 /* KVM_EXIT_HYPERV */ 5233 struct kvm_hyperv_exit hyperv; 5234 5235Indicates that the VCPU exits into userspace to process some tasks 5236related to Hyper-V emulation. 5237 5238Valid values for 'type' are: 5239 5240 - KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about 5241 5242Hyper-V SynIC state change. Notification is used to remap SynIC 5243event/message pages and to enable/disable SynIC messages/events processing 5244in userspace. 5245 5246 - KVM_EXIT_HYPERV_SYNDBG -- synchronously notify user-space about 5247 5248Hyper-V Synthetic debugger state change. Notification is used to either update 5249the pending_page location or to send a control command (send the buffer located 5250in send_page or recv a buffer to recv_page). 5251 5252:: 5253 5254 /* KVM_EXIT_ARM_NISV */ 5255 struct { 5256 __u64 esr_iss; 5257 __u64 fault_ipa; 5258 } arm_nisv; 5259 5260Used on arm and arm64 systems. If a guest accesses memory not in a memslot, 5261KVM will typically return to userspace and ask it to do MMIO emulation on its 5262behalf. However, for certain classes of instructions, no instruction decode 5263(direction, length of memory access) is provided, and fetching and decoding 5264the instruction from the VM is overly complicated to live in the kernel. 5265 5266Historically, when this situation occurred, KVM would print a warning and kill 5267the VM. KVM assumed that if the guest accessed non-memslot memory, it was 5268trying to do I/O, which just couldn't be emulated, and the warning message was 5269phrased accordingly. However, what happened more often was that a guest bug 5270caused access outside the guest memory areas which should lead to a more 5271meaningful warning message and an external abort in the guest, if the access 5272did not fall within an I/O window. 5273 5274Userspace implementations can query for KVM_CAP_ARM_NISV_TO_USER, and enable 5275this capability at VM creation. Once this is done, these types of errors will 5276instead return to userspace with KVM_EXIT_ARM_NISV, with the valid bits from 5277the HSR (arm) and ESR_EL2 (arm64) in the esr_iss field, and the faulting IPA 5278in the fault_ipa field. Userspace can either fix up the access if it's 5279actually an I/O access by decoding the instruction from guest memory (if it's 5280very brave) and continue executing the guest, or it can decide to suspend, 5281dump, or restart the guest. 5282 5283Note that KVM does not skip the faulting instruction as it does for 5284KVM_EXIT_MMIO, but userspace has to emulate any change to the processing state 5285if it decides to decode and emulate the instruction. 5286 5287:: 5288 5289 /* KVM_EXIT_X86_RDMSR / KVM_EXIT_X86_WRMSR */ 5290 struct { 5291 __u8 error; /* user -> kernel */ 5292 __u8 pad[7]; 5293 __u32 reason; /* kernel -> user */ 5294 __u32 index; /* kernel -> user */ 5295 __u64 data; /* kernel <-> user */ 5296 } msr; 5297 5298Used on x86 systems. When the VM capability KVM_CAP_X86_USER_SPACE_MSR is 5299enabled, MSR accesses to registers that would invoke a #GP by KVM kernel code 5300will instead trigger a KVM_EXIT_X86_RDMSR exit for reads and KVM_EXIT_X86_WRMSR 5301exit for writes. 5302 5303The "reason" field specifies why the MSR trap occurred. User space will only 5304receive MSR exit traps when a particular reason was requested during through 5305ENABLE_CAP. Currently valid exit reasons are: 5306 5307 KVM_MSR_EXIT_REASON_UNKNOWN - access to MSR that is unknown to KVM 5308 KVM_MSR_EXIT_REASON_INVAL - access to invalid MSRs or reserved bits 5309 KVM_MSR_EXIT_REASON_FILTER - access blocked by KVM_X86_SET_MSR_FILTER 5310 5311For KVM_EXIT_X86_RDMSR, the "index" field tells user space which MSR the guest 5312wants to read. To respond to this request with a successful read, user space 5313writes the respective data into the "data" field and must continue guest 5314execution to ensure the read data is transferred into guest register state. 5315 5316If the RDMSR request was unsuccessful, user space indicates that with a "1" in 5317the "error" field. This will inject a #GP into the guest when the VCPU is 5318executed again. 5319 5320For KVM_EXIT_X86_WRMSR, the "index" field tells user space which MSR the guest 5321wants to write. Once finished processing the event, user space must continue 5322vCPU execution. If the MSR write was unsuccessful, user space also sets the 5323"error" field to "1". 5324 5325:: 5326 5327 /* Fix the size of the union. */ 5328 char padding[256]; 5329 }; 5330 5331 /* 5332 * shared registers between kvm and userspace. 5333 * kvm_valid_regs specifies the register classes set by the host 5334 * kvm_dirty_regs specified the register classes dirtied by userspace 5335 * struct kvm_sync_regs is architecture specific, as well as the 5336 * bits for kvm_valid_regs and kvm_dirty_regs 5337 */ 5338 __u64 kvm_valid_regs; 5339 __u64 kvm_dirty_regs; 5340 union { 5341 struct kvm_sync_regs regs; 5342 char padding[SYNC_REGS_SIZE_BYTES]; 5343 } s; 5344 5345If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access 5346certain guest registers without having to call SET/GET_*REGS. Thus we can 5347avoid some system call overhead if userspace has to handle the exit. 5348Userspace can query the validity of the structure by checking 5349kvm_valid_regs for specific bits. These bits are architecture specific 5350and usually define the validity of a groups of registers. (e.g. one bit 5351for general purpose registers) 5352 5353Please note that the kernel is allowed to use the kvm_run structure as the 5354primary storage for certain register types. Therefore, the kernel may use the 5355values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set. 5356 5357:: 5358 5359 }; 5360 5361 5362 53636. Capabilities that can be enabled on vCPUs 5364============================================ 5365 5366There are certain capabilities that change the behavior of the virtual CPU or 5367the virtual machine when enabled. To enable them, please see section 4.37. 5368Below you can find a list of capabilities and what their effect on the vCPU or 5369the virtual machine is when enabling them. 5370 5371The following information is provided along with the description: 5372 5373 Architectures: 5374 which instruction set architectures provide this ioctl. 5375 x86 includes both i386 and x86_64. 5376 5377 Target: 5378 whether this is a per-vcpu or per-vm capability. 5379 5380 Parameters: 5381 what parameters are accepted by the capability. 5382 5383 Returns: 5384 the return value. General error numbers (EBADF, ENOMEM, EINVAL) 5385 are not detailed, but errors with specific meanings are. 5386 5387 53886.1 KVM_CAP_PPC_OSI 5389------------------- 5390 5391:Architectures: ppc 5392:Target: vcpu 5393:Parameters: none 5394:Returns: 0 on success; -1 on error 5395 5396This capability enables interception of OSI hypercalls that otherwise would 5397be treated as normal system calls to be injected into the guest. OSI hypercalls 5398were invented by Mac-on-Linux to have a standardized communication mechanism 5399between the guest and the host. 5400 5401When this capability is enabled, KVM_EXIT_OSI can occur. 5402 5403 54046.2 KVM_CAP_PPC_PAPR 5405-------------------- 5406 5407:Architectures: ppc 5408:Target: vcpu 5409:Parameters: none 5410:Returns: 0 on success; -1 on error 5411 5412This capability enables interception of PAPR hypercalls. PAPR hypercalls are 5413done using the hypercall instruction "sc 1". 5414 5415It also sets the guest privilege level to "supervisor" mode. Usually the guest 5416runs in "hypervisor" privilege mode with a few missing features. 5417 5418In addition to the above, it changes the semantics of SDR1. In this mode, the 5419HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the 5420HTAB invisible to the guest. 5421 5422When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur. 5423 5424 54256.3 KVM_CAP_SW_TLB 5426------------------ 5427 5428:Architectures: ppc 5429:Target: vcpu 5430:Parameters: args[0] is the address of a struct kvm_config_tlb 5431:Returns: 0 on success; -1 on error 5432 5433:: 5434 5435 struct kvm_config_tlb { 5436 __u64 params; 5437 __u64 array; 5438 __u32 mmu_type; 5439 __u32 array_len; 5440 }; 5441 5442Configures the virtual CPU's TLB array, establishing a shared memory area 5443between userspace and KVM. The "params" and "array" fields are userspace 5444addresses of mmu-type-specific data structures. The "array_len" field is an 5445safety mechanism, and should be set to the size in bytes of the memory that 5446userspace has reserved for the array. It must be at least the size dictated 5447by "mmu_type" and "params". 5448 5449While KVM_RUN is active, the shared region is under control of KVM. Its 5450contents are undefined, and any modification by userspace results in 5451boundedly undefined behavior. 5452 5453On return from KVM_RUN, the shared region will reflect the current state of 5454the guest's TLB. If userspace makes any changes, it must call KVM_DIRTY_TLB 5455to tell KVM which entries have been changed, prior to calling KVM_RUN again 5456on this vcpu. 5457 5458For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV: 5459 5460 - The "params" field is of type "struct kvm_book3e_206_tlb_params". 5461 - The "array" field points to an array of type "struct 5462 kvm_book3e_206_tlb_entry". 5463 - The array consists of all entries in the first TLB, followed by all 5464 entries in the second TLB. 5465 - Within a TLB, entries are ordered first by increasing set number. Within a 5466 set, entries are ordered by way (increasing ESEL). 5467 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1) 5468 where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value. 5469 - The tsize field of mas1 shall be set to 4K on TLB0, even though the 5470 hardware ignores this value for TLB0. 5471 54726.4 KVM_CAP_S390_CSS_SUPPORT 5473---------------------------- 5474 5475:Architectures: s390 5476:Target: vcpu 5477:Parameters: none 5478:Returns: 0 on success; -1 on error 5479 5480This capability enables support for handling of channel I/O instructions. 5481 5482TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are 5483handled in-kernel, while the other I/O instructions are passed to userspace. 5484 5485When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST 5486SUBCHANNEL intercepts. 5487 5488Note that even though this capability is enabled per-vcpu, the complete 5489virtual machine is affected. 5490 54916.5 KVM_CAP_PPC_EPR 5492------------------- 5493 5494:Architectures: ppc 5495:Target: vcpu 5496:Parameters: args[0] defines whether the proxy facility is active 5497:Returns: 0 on success; -1 on error 5498 5499This capability enables or disables the delivery of interrupts through the 5500external proxy facility. 5501 5502When enabled (args[0] != 0), every time the guest gets an external interrupt 5503delivered, it automatically exits into user space with a KVM_EXIT_EPR exit 5504to receive the topmost interrupt vector. 5505 5506When disabled (args[0] == 0), behavior is as if this facility is unsupported. 5507 5508When this capability is enabled, KVM_EXIT_EPR can occur. 5509 55106.6 KVM_CAP_IRQ_MPIC 5511-------------------- 5512 5513:Architectures: ppc 5514:Parameters: args[0] is the MPIC device fd; 5515 args[1] is the MPIC CPU number for this vcpu 5516 5517This capability connects the vcpu to an in-kernel MPIC device. 5518 55196.7 KVM_CAP_IRQ_XICS 5520-------------------- 5521 5522:Architectures: ppc 5523:Target: vcpu 5524:Parameters: args[0] is the XICS device fd; 5525 args[1] is the XICS CPU number (server ID) for this vcpu 5526 5527This capability connects the vcpu to an in-kernel XICS device. 5528 55296.8 KVM_CAP_S390_IRQCHIP 5530------------------------ 5531 5532:Architectures: s390 5533:Target: vm 5534:Parameters: none 5535 5536This capability enables the in-kernel irqchip for s390. Please refer to 5537"4.24 KVM_CREATE_IRQCHIP" for details. 5538 55396.9 KVM_CAP_MIPS_FPU 5540-------------------- 5541 5542:Architectures: mips 5543:Target: vcpu 5544:Parameters: args[0] is reserved for future use (should be 0). 5545 5546This capability allows the use of the host Floating Point Unit by the guest. It 5547allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is 5548done the ``KVM_REG_MIPS_FPR_*`` and ``KVM_REG_MIPS_FCR_*`` registers can be 5549accessed (depending on the current guest FPU register mode), and the Status.FR, 5550Config5.FRE bits are accessible via the KVM API and also from the guest, 5551depending on them being supported by the FPU. 5552 55536.10 KVM_CAP_MIPS_MSA 5554--------------------- 5555 5556:Architectures: mips 5557:Target: vcpu 5558:Parameters: args[0] is reserved for future use (should be 0). 5559 5560This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest. 5561It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest. 5562Once this is done the ``KVM_REG_MIPS_VEC_*`` and ``KVM_REG_MIPS_MSA_*`` 5563registers can be accessed, and the Config5.MSAEn bit is accessible via the 5564KVM API and also from the guest. 5565 55666.74 KVM_CAP_SYNC_REGS 5567---------------------- 5568 5569:Architectures: s390, x86 5570:Target: s390: always enabled, x86: vcpu 5571:Parameters: none 5572:Returns: x86: KVM_CHECK_EXTENSION returns a bit-array indicating which register 5573 sets are supported 5574 (bitfields defined in arch/x86/include/uapi/asm/kvm.h). 5575 5576As described above in the kvm_sync_regs struct info in section 5 (kvm_run): 5577KVM_CAP_SYNC_REGS "allow[s] userspace to access certain guest registers 5578without having to call SET/GET_*REGS". This reduces overhead by eliminating 5579repeated ioctl calls for setting and/or getting register values. This is 5580particularly important when userspace is making synchronous guest state 5581modifications, e.g. when emulating and/or intercepting instructions in 5582userspace. 5583 5584For s390 specifics, please refer to the source code. 5585 5586For x86: 5587 5588- the register sets to be copied out to kvm_run are selectable 5589 by userspace (rather that all sets being copied out for every exit). 5590- vcpu_events are available in addition to regs and sregs. 5591 5592For x86, the 'kvm_valid_regs' field of struct kvm_run is overloaded to 5593function as an input bit-array field set by userspace to indicate the 5594specific register sets to be copied out on the next exit. 5595 5596To indicate when userspace has modified values that should be copied into 5597the vCPU, the all architecture bitarray field, 'kvm_dirty_regs' must be set. 5598This is done using the same bitflags as for the 'kvm_valid_regs' field. 5599If the dirty bit is not set, then the register set values will not be copied 5600into the vCPU even if they've been modified. 5601 5602Unused bitfields in the bitarrays must be set to zero. 5603 5604:: 5605 5606 struct kvm_sync_regs { 5607 struct kvm_regs regs; 5608 struct kvm_sregs sregs; 5609 struct kvm_vcpu_events events; 5610 }; 5611 56126.75 KVM_CAP_PPC_IRQ_XIVE 5613------------------------- 5614 5615:Architectures: ppc 5616:Target: vcpu 5617:Parameters: args[0] is the XIVE device fd; 5618 args[1] is the XIVE CPU number (server ID) for this vcpu 5619 5620This capability connects the vcpu to an in-kernel XIVE device. 5621 56227. Capabilities that can be enabled on VMs 5623========================================== 5624 5625There are certain capabilities that change the behavior of the virtual 5626machine when enabled. To enable them, please see section 4.37. Below 5627you can find a list of capabilities and what their effect on the VM 5628is when enabling them. 5629 5630The following information is provided along with the description: 5631 5632 Architectures: 5633 which instruction set architectures provide this ioctl. 5634 x86 includes both i386 and x86_64. 5635 5636 Parameters: 5637 what parameters are accepted by the capability. 5638 5639 Returns: 5640 the return value. General error numbers (EBADF, ENOMEM, EINVAL) 5641 are not detailed, but errors with specific meanings are. 5642 5643 56447.1 KVM_CAP_PPC_ENABLE_HCALL 5645---------------------------- 5646 5647:Architectures: ppc 5648:Parameters: args[0] is the sPAPR hcall number; 5649 args[1] is 0 to disable, 1 to enable in-kernel handling 5650 5651This capability controls whether individual sPAPR hypercalls (hcalls) 5652get handled by the kernel or not. Enabling or disabling in-kernel 5653handling of an hcall is effective across the VM. On creation, an 5654initial set of hcalls are enabled for in-kernel handling, which 5655consists of those hcalls for which in-kernel handlers were implemented 5656before this capability was implemented. If disabled, the kernel will 5657not to attempt to handle the hcall, but will always exit to userspace 5658to handle it. Note that it may not make sense to enable some and 5659disable others of a group of related hcalls, but KVM does not prevent 5660userspace from doing that. 5661 5662If the hcall number specified is not one that has an in-kernel 5663implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL 5664error. 5665 56667.2 KVM_CAP_S390_USER_SIGP 5667-------------------------- 5668 5669:Architectures: s390 5670:Parameters: none 5671 5672This capability controls which SIGP orders will be handled completely in user 5673space. With this capability enabled, all fast orders will be handled completely 5674in the kernel: 5675 5676- SENSE 5677- SENSE RUNNING 5678- EXTERNAL CALL 5679- EMERGENCY SIGNAL 5680- CONDITIONAL EMERGENCY SIGNAL 5681 5682All other orders will be handled completely in user space. 5683 5684Only privileged operation exceptions will be checked for in the kernel (or even 5685in the hardware prior to interception). If this capability is not enabled, the 5686old way of handling SIGP orders is used (partially in kernel and user space). 5687 56887.3 KVM_CAP_S390_VECTOR_REGISTERS 5689--------------------------------- 5690 5691:Architectures: s390 5692:Parameters: none 5693:Returns: 0 on success, negative value on error 5694 5695Allows use of the vector registers introduced with z13 processor, and 5696provides for the synchronization between host and user space. Will 5697return -EINVAL if the machine does not support vectors. 5698 56997.4 KVM_CAP_S390_USER_STSI 5700-------------------------- 5701 5702:Architectures: s390 5703:Parameters: none 5704 5705This capability allows post-handlers for the STSI instruction. After 5706initial handling in the kernel, KVM exits to user space with 5707KVM_EXIT_S390_STSI to allow user space to insert further data. 5708 5709Before exiting to userspace, kvm handlers should fill in s390_stsi field of 5710vcpu->run:: 5711 5712 struct { 5713 __u64 addr; 5714 __u8 ar; 5715 __u8 reserved; 5716 __u8 fc; 5717 __u8 sel1; 5718 __u16 sel2; 5719 } s390_stsi; 5720 5721 @addr - guest address of STSI SYSIB 5722 @fc - function code 5723 @sel1 - selector 1 5724 @sel2 - selector 2 5725 @ar - access register number 5726 5727KVM handlers should exit to userspace with rc = -EREMOTE. 5728 57297.5 KVM_CAP_SPLIT_IRQCHIP 5730------------------------- 5731 5732:Architectures: x86 5733:Parameters: args[0] - number of routes reserved for userspace IOAPICs 5734:Returns: 0 on success, -1 on error 5735 5736Create a local apic for each processor in the kernel. This can be used 5737instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the 5738IOAPIC and PIC (and also the PIT, even though this has to be enabled 5739separately). 5740 5741This capability also enables in kernel routing of interrupt requests; 5742when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are 5743used in the IRQ routing table. The first args[0] MSI routes are reserved 5744for the IOAPIC pins. Whenever the LAPIC receives an EOI for these routes, 5745a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace. 5746 5747Fails if VCPU has already been created, or if the irqchip is already in the 5748kernel (i.e. KVM_CREATE_IRQCHIP has already been called). 5749 57507.6 KVM_CAP_S390_RI 5751------------------- 5752 5753:Architectures: s390 5754:Parameters: none 5755 5756Allows use of runtime-instrumentation introduced with zEC12 processor. 5757Will return -EINVAL if the machine does not support runtime-instrumentation. 5758Will return -EBUSY if a VCPU has already been created. 5759 57607.7 KVM_CAP_X2APIC_API 5761---------------------- 5762 5763:Architectures: x86 5764:Parameters: args[0] - features that should be enabled 5765:Returns: 0 on success, -EINVAL when args[0] contains invalid features 5766 5767Valid feature flags in args[0] are:: 5768 5769 #define KVM_X2APIC_API_USE_32BIT_IDS (1ULL << 0) 5770 #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK (1ULL << 1) 5771 5772Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of 5773KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC, 5774allowing the use of 32-bit APIC IDs. See KVM_CAP_X2APIC_API in their 5775respective sections. 5776 5777KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work 5778in logical mode or with more than 255 VCPUs. Otherwise, KVM treats 0xff 5779as a broadcast even in x2APIC mode in order to support physical x2APIC 5780without interrupt remapping. This is undesirable in logical mode, 5781where 0xff represents CPUs 0-7 in cluster 0. 5782 57837.8 KVM_CAP_S390_USER_INSTR0 5784---------------------------- 5785 5786:Architectures: s390 5787:Parameters: none 5788 5789With this capability enabled, all illegal instructions 0x0000 (2 bytes) will 5790be intercepted and forwarded to user space. User space can use this 5791mechanism e.g. to realize 2-byte software breakpoints. The kernel will 5792not inject an operating exception for these instructions, user space has 5793to take care of that. 5794 5795This capability can be enabled dynamically even if VCPUs were already 5796created and are running. 5797 57987.9 KVM_CAP_S390_GS 5799------------------- 5800 5801:Architectures: s390 5802:Parameters: none 5803:Returns: 0 on success; -EINVAL if the machine does not support 5804 guarded storage; -EBUSY if a VCPU has already been created. 5805 5806Allows use of guarded storage for the KVM guest. 5807 58087.10 KVM_CAP_S390_AIS 5809--------------------- 5810 5811:Architectures: s390 5812:Parameters: none 5813 5814Allow use of adapter-interruption suppression. 5815:Returns: 0 on success; -EBUSY if a VCPU has already been created. 5816 58177.11 KVM_CAP_PPC_SMT 5818-------------------- 5819 5820:Architectures: ppc 5821:Parameters: vsmt_mode, flags 5822 5823Enabling this capability on a VM provides userspace with a way to set 5824the desired virtual SMT mode (i.e. the number of virtual CPUs per 5825virtual core). The virtual SMT mode, vsmt_mode, must be a power of 2 5826between 1 and 8. On POWER8, vsmt_mode must also be no greater than 5827the number of threads per subcore for the host. Currently flags must 5828be 0. A successful call to enable this capability will result in 5829vsmt_mode being returned when the KVM_CAP_PPC_SMT capability is 5830subsequently queried for the VM. This capability is only supported by 5831HV KVM, and can only be set before any VCPUs have been created. 5832The KVM_CAP_PPC_SMT_POSSIBLE capability indicates which virtual SMT 5833modes are available. 5834 58357.12 KVM_CAP_PPC_FWNMI 5836---------------------- 5837 5838:Architectures: ppc 5839:Parameters: none 5840 5841With this capability a machine check exception in the guest address 5842space will cause KVM to exit the guest with NMI exit reason. This 5843enables QEMU to build error log and branch to guest kernel registered 5844machine check handling routine. Without this capability KVM will 5845branch to guests' 0x200 interrupt vector. 5846 58477.13 KVM_CAP_X86_DISABLE_EXITS 5848------------------------------ 5849 5850:Architectures: x86 5851:Parameters: args[0] defines which exits are disabled 5852:Returns: 0 on success, -EINVAL when args[0] contains invalid exits 5853 5854Valid bits in args[0] are:: 5855 5856 #define KVM_X86_DISABLE_EXITS_MWAIT (1 << 0) 5857 #define KVM_X86_DISABLE_EXITS_HLT (1 << 1) 5858 #define KVM_X86_DISABLE_EXITS_PAUSE (1 << 2) 5859 #define KVM_X86_DISABLE_EXITS_CSTATE (1 << 3) 5860 5861Enabling this capability on a VM provides userspace with a way to no 5862longer intercept some instructions for improved latency in some 5863workloads, and is suggested when vCPUs are associated to dedicated 5864physical CPUs. More bits can be added in the future; userspace can 5865just pass the KVM_CHECK_EXTENSION result to KVM_ENABLE_CAP to disable 5866all such vmexits. 5867 5868Do not enable KVM_FEATURE_PV_UNHALT if you disable HLT exits. 5869 58707.14 KVM_CAP_S390_HPAGE_1M 5871-------------------------- 5872 5873:Architectures: s390 5874:Parameters: none 5875:Returns: 0 on success, -EINVAL if hpage module parameter was not set 5876 or cmma is enabled, or the VM has the KVM_VM_S390_UCONTROL 5877 flag set 5878 5879With this capability the KVM support for memory backing with 1m pages 5880through hugetlbfs can be enabled for a VM. After the capability is 5881enabled, cmma can't be enabled anymore and pfmfi and the storage key 5882interpretation are disabled. If cmma has already been enabled or the 5883hpage module parameter is not set to 1, -EINVAL is returned. 5884 5885While it is generally possible to create a huge page backed VM without 5886this capability, the VM will not be able to run. 5887 58887.15 KVM_CAP_MSR_PLATFORM_INFO 5889------------------------------ 5890 5891:Architectures: x86 5892:Parameters: args[0] whether feature should be enabled or not 5893 5894With this capability, a guest may read the MSR_PLATFORM_INFO MSR. Otherwise, 5895a #GP would be raised when the guest tries to access. Currently, this 5896capability does not enable write permissions of this MSR for the guest. 5897 58987.16 KVM_CAP_PPC_NESTED_HV 5899-------------------------- 5900 5901:Architectures: ppc 5902:Parameters: none 5903:Returns: 0 on success, -EINVAL when the implementation doesn't support 5904 nested-HV virtualization. 5905 5906HV-KVM on POWER9 and later systems allows for "nested-HV" 5907virtualization, which provides a way for a guest VM to run guests that 5908can run using the CPU's supervisor mode (privileged non-hypervisor 5909state). Enabling this capability on a VM depends on the CPU having 5910the necessary functionality and on the facility being enabled with a 5911kvm-hv module parameter. 5912 59137.17 KVM_CAP_EXCEPTION_PAYLOAD 5914------------------------------ 5915 5916:Architectures: x86 5917:Parameters: args[0] whether feature should be enabled or not 5918 5919With this capability enabled, CR2 will not be modified prior to the 5920emulated VM-exit when L1 intercepts a #PF exception that occurs in 5921L2. Similarly, for kvm-intel only, DR6 will not be modified prior to 5922the emulated VM-exit when L1 intercepts a #DB exception that occurs in 5923L2. As a result, when KVM_GET_VCPU_EVENTS reports a pending #PF (or 5924#DB) exception for L2, exception.has_payload will be set and the 5925faulting address (or the new DR6 bits*) will be reported in the 5926exception_payload field. Similarly, when userspace injects a #PF (or 5927#DB) into L2 using KVM_SET_VCPU_EVENTS, it is expected to set 5928exception.has_payload and to put the faulting address - or the new DR6 5929bits\ [#]_ - in the exception_payload field. 5930 5931This capability also enables exception.pending in struct 5932kvm_vcpu_events, which allows userspace to distinguish between pending 5933and injected exceptions. 5934 5935 5936.. [#] For the new DR6 bits, note that bit 16 is set iff the #DB exception 5937 will clear DR6.RTM. 5938 59397.18 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 5940 5941:Architectures: x86, arm, arm64, mips 5942:Parameters: args[0] whether feature should be enabled or not 5943 5944Valid flags are:: 5945 5946 #define KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (1 << 0) 5947 #define KVM_DIRTY_LOG_INITIALLY_SET (1 << 1) 5948 5949With KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE is set, KVM_GET_DIRTY_LOG will not 5950automatically clear and write-protect all pages that are returned as dirty. 5951Rather, userspace will have to do this operation separately using 5952KVM_CLEAR_DIRTY_LOG. 5953 5954At the cost of a slightly more complicated operation, this provides better 5955scalability and responsiveness for two reasons. First, 5956KVM_CLEAR_DIRTY_LOG ioctl can operate on a 64-page granularity rather 5957than requiring to sync a full memslot; this ensures that KVM does not 5958take spinlocks for an extended period of time. Second, in some cases a 5959large amount of time can pass between a call to KVM_GET_DIRTY_LOG and 5960userspace actually using the data in the page. Pages can be modified 5961during this time, which is inefficient for both the guest and userspace: 5962the guest will incur a higher penalty due to write protection faults, 5963while userspace can see false reports of dirty pages. Manual reprotection 5964helps reducing this time, improving guest performance and reducing the 5965number of dirty log false positives. 5966 5967With KVM_DIRTY_LOG_INITIALLY_SET set, all the bits of the dirty bitmap 5968will be initialized to 1 when created. This also improves performance because 5969dirty logging can be enabled gradually in small chunks on the first call 5970to KVM_CLEAR_DIRTY_LOG. KVM_DIRTY_LOG_INITIALLY_SET depends on 5971KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (it is also only available on 5972x86 and arm64 for now). 5973 5974KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 was previously available under the name 5975KVM_CAP_MANUAL_DIRTY_LOG_PROTECT, but the implementation had bugs that make 5976it hard or impossible to use it correctly. The availability of 5977KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 signals that those bugs are fixed. 5978Userspace should not try to use KVM_CAP_MANUAL_DIRTY_LOG_PROTECT. 5979 59807.19 KVM_CAP_PPC_SECURE_GUEST 5981------------------------------ 5982 5983:Architectures: ppc 5984 5985This capability indicates that KVM is running on a host that has 5986ultravisor firmware and thus can support a secure guest. On such a 5987system, a guest can ask the ultravisor to make it a secure guest, 5988one whose memory is inaccessible to the host except for pages which 5989are explicitly requested to be shared with the host. The ultravisor 5990notifies KVM when a guest requests to become a secure guest, and KVM 5991has the opportunity to veto the transition. 5992 5993If present, this capability can be enabled for a VM, meaning that KVM 5994will allow the transition to secure guest mode. Otherwise KVM will 5995veto the transition. 5996 59977.20 KVM_CAP_HALT_POLL 5998---------------------- 5999 6000:Architectures: all 6001:Target: VM 6002:Parameters: args[0] is the maximum poll time in nanoseconds 6003:Returns: 0 on success; -1 on error 6004 6005This capability overrides the kvm module parameter halt_poll_ns for the 6006target VM. 6007 6008VCPU polling allows a VCPU to poll for wakeup events instead of immediately 6009scheduling during guest halts. The maximum time a VCPU can spend polling is 6010controlled by the kvm module parameter halt_poll_ns. This capability allows 6011the maximum halt time to specified on a per-VM basis, effectively overriding 6012the module parameter for the target VM. 6013 60147.21 KVM_CAP_X86_USER_SPACE_MSR 6015------------------------------- 6016 6017:Architectures: x86 6018:Target: VM 6019:Parameters: args[0] contains the mask of KVM_MSR_EXIT_REASON_* events to report 6020:Returns: 0 on success; -1 on error 6021 6022This capability enables trapping of #GP invoking RDMSR and WRMSR instructions 6023into user space. 6024 6025When a guest requests to read or write an MSR, KVM may not implement all MSRs 6026that are relevant to a respective system. It also does not differentiate by 6027CPU type. 6028 6029To allow more fine grained control over MSR handling, user space may enable 6030this capability. With it enabled, MSR accesses that match the mask specified in 6031args[0] and trigger a #GP event inside the guest by KVM will instead trigger 6032KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR exit notifications which user space 6033can then handle to implement model specific MSR handling and/or user notifications 6034to inform a user that an MSR was not handled. 6035 60368. Other capabilities. 6037====================== 6038 6039This section lists capabilities that give information about other 6040features of the KVM implementation. 6041 60428.1 KVM_CAP_PPC_HWRNG 6043--------------------- 6044 6045:Architectures: ppc 6046 6047This capability, if KVM_CHECK_EXTENSION indicates that it is 6048available, means that the kernel has an implementation of the 6049H_RANDOM hypercall backed by a hardware random-number generator. 6050If present, the kernel H_RANDOM handler can be enabled for guest use 6051with the KVM_CAP_PPC_ENABLE_HCALL capability. 6052 60538.2 KVM_CAP_HYPERV_SYNIC 6054------------------------ 6055 6056:Architectures: x86 6057 6058This capability, if KVM_CHECK_EXTENSION indicates that it is 6059available, means that the kernel has an implementation of the 6060Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is 6061used to support Windows Hyper-V based guest paravirt drivers(VMBus). 6062 6063In order to use SynIC, it has to be activated by setting this 6064capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this 6065will disable the use of APIC hardware virtualization even if supported 6066by the CPU, as it's incompatible with SynIC auto-EOI behavior. 6067 60688.3 KVM_CAP_PPC_RADIX_MMU 6069------------------------- 6070 6071:Architectures: ppc 6072 6073This capability, if KVM_CHECK_EXTENSION indicates that it is 6074available, means that the kernel can support guests using the 6075radix MMU defined in Power ISA V3.00 (as implemented in the POWER9 6076processor). 6077 60788.4 KVM_CAP_PPC_HASH_MMU_V3 6079--------------------------- 6080 6081:Architectures: ppc 6082 6083This capability, if KVM_CHECK_EXTENSION indicates that it is 6084available, means that the kernel can support guests using the 6085hashed page table MMU defined in Power ISA V3.00 (as implemented in 6086the POWER9 processor), including in-memory segment tables. 6087 60888.5 KVM_CAP_MIPS_VZ 6089------------------- 6090 6091:Architectures: mips 6092 6093This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that 6094it is available, means that full hardware assisted virtualization capabilities 6095of the hardware are available for use through KVM. An appropriate 6096KVM_VM_MIPS_* type must be passed to KVM_CREATE_VM to create a VM which 6097utilises it. 6098 6099If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is 6100available, it means that the VM is using full hardware assisted virtualization 6101capabilities of the hardware. This is useful to check after creating a VM with 6102KVM_VM_MIPS_DEFAULT. 6103 6104The value returned by KVM_CHECK_EXTENSION should be compared against known 6105values (see below). All other values are reserved. This is to allow for the 6106possibility of other hardware assisted virtualization implementations which 6107may be incompatible with the MIPS VZ ASE. 6108 6109== ========================================================================== 6110 0 The trap & emulate implementation is in use to run guest code in user 6111 mode. Guest virtual memory segments are rearranged to fit the guest in the 6112 user mode address space. 6113 6114 1 The MIPS VZ ASE is in use, providing full hardware assisted 6115 virtualization, including standard guest virtual memory segments. 6116== ========================================================================== 6117 61188.6 KVM_CAP_MIPS_TE 6119------------------- 6120 6121:Architectures: mips 6122 6123This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that 6124it is available, means that the trap & emulate implementation is available to 6125run guest code in user mode, even if KVM_CAP_MIPS_VZ indicates that hardware 6126assisted virtualisation is also available. KVM_VM_MIPS_TE (0) must be passed 6127to KVM_CREATE_VM to create a VM which utilises it. 6128 6129If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is 6130available, it means that the VM is using trap & emulate. 6131 61328.7 KVM_CAP_MIPS_64BIT 6133---------------------- 6134 6135:Architectures: mips 6136 6137This capability indicates the supported architecture type of the guest, i.e. the 6138supported register and address width. 6139 6140The values returned when this capability is checked by KVM_CHECK_EXTENSION on a 6141kvm VM handle correspond roughly to the CP0_Config.AT register field, and should 6142be checked specifically against known values (see below). All other values are 6143reserved. 6144 6145== ======================================================================== 6146 0 MIPS32 or microMIPS32. 6147 Both registers and addresses are 32-bits wide. 6148 It will only be possible to run 32-bit guest code. 6149 6150 1 MIPS64 or microMIPS64 with access only to 32-bit compatibility segments. 6151 Registers are 64-bits wide, but addresses are 32-bits wide. 6152 64-bit guest code may run but cannot access MIPS64 memory segments. 6153 It will also be possible to run 32-bit guest code. 6154 6155 2 MIPS64 or microMIPS64 with access to all address segments. 6156 Both registers and addresses are 64-bits wide. 6157 It will be possible to run 64-bit or 32-bit guest code. 6158== ======================================================================== 6159 61608.9 KVM_CAP_ARM_USER_IRQ 6161------------------------ 6162 6163:Architectures: arm, arm64 6164 6165This capability, if KVM_CHECK_EXTENSION indicates that it is available, means 6166that if userspace creates a VM without an in-kernel interrupt controller, it 6167will be notified of changes to the output level of in-kernel emulated devices, 6168which can generate virtual interrupts, presented to the VM. 6169For such VMs, on every return to userspace, the kernel 6170updates the vcpu's run->s.regs.device_irq_level field to represent the actual 6171output level of the device. 6172 6173Whenever kvm detects a change in the device output level, kvm guarantees at 6174least one return to userspace before running the VM. This exit could either 6175be a KVM_EXIT_INTR or any other exit event, like KVM_EXIT_MMIO. This way, 6176userspace can always sample the device output level and re-compute the state of 6177the userspace interrupt controller. Userspace should always check the state 6178of run->s.regs.device_irq_level on every kvm exit. 6179The value in run->s.regs.device_irq_level can represent both level and edge 6180triggered interrupt signals, depending on the device. Edge triggered interrupt 6181signals will exit to userspace with the bit in run->s.regs.device_irq_level 6182set exactly once per edge signal. 6183 6184The field run->s.regs.device_irq_level is available independent of 6185run->kvm_valid_regs or run->kvm_dirty_regs bits. 6186 6187If KVM_CAP_ARM_USER_IRQ is supported, the KVM_CHECK_EXTENSION ioctl returns a 6188number larger than 0 indicating the version of this capability is implemented 6189and thereby which bits in run->s.regs.device_irq_level can signal values. 6190 6191Currently the following bits are defined for the device_irq_level bitmap:: 6192 6193 KVM_CAP_ARM_USER_IRQ >= 1: 6194 6195 KVM_ARM_DEV_EL1_VTIMER - EL1 virtual timer 6196 KVM_ARM_DEV_EL1_PTIMER - EL1 physical timer 6197 KVM_ARM_DEV_PMU - ARM PMU overflow interrupt signal 6198 6199Future versions of kvm may implement additional events. These will get 6200indicated by returning a higher number from KVM_CHECK_EXTENSION and will be 6201listed above. 6202 62038.10 KVM_CAP_PPC_SMT_POSSIBLE 6204----------------------------- 6205 6206:Architectures: ppc 6207 6208Querying this capability returns a bitmap indicating the possible 6209virtual SMT modes that can be set using KVM_CAP_PPC_SMT. If bit N 6210(counting from the right) is set, then a virtual SMT mode of 2^N is 6211available. 6212 62138.11 KVM_CAP_HYPERV_SYNIC2 6214-------------------------- 6215 6216:Architectures: x86 6217 6218This capability enables a newer version of Hyper-V Synthetic interrupt 6219controller (SynIC). The only difference with KVM_CAP_HYPERV_SYNIC is that KVM 6220doesn't clear SynIC message and event flags pages when they are enabled by 6221writing to the respective MSRs. 6222 62238.12 KVM_CAP_HYPERV_VP_INDEX 6224---------------------------- 6225 6226:Architectures: x86 6227 6228This capability indicates that userspace can load HV_X64_MSR_VP_INDEX msr. Its 6229value is used to denote the target vcpu for a SynIC interrupt. For 6230compatibilty, KVM initializes this msr to KVM's internal vcpu index. When this 6231capability is absent, userspace can still query this msr's value. 6232 62338.13 KVM_CAP_S390_AIS_MIGRATION 6234------------------------------- 6235 6236:Architectures: s390 6237:Parameters: none 6238 6239This capability indicates if the flic device will be able to get/set the 6240AIS states for migration via the KVM_DEV_FLIC_AISM_ALL attribute and allows 6241to discover this without having to create a flic device. 6242 62438.14 KVM_CAP_S390_PSW 6244--------------------- 6245 6246:Architectures: s390 6247 6248This capability indicates that the PSW is exposed via the kvm_run structure. 6249 62508.15 KVM_CAP_S390_GMAP 6251---------------------- 6252 6253:Architectures: s390 6254 6255This capability indicates that the user space memory used as guest mapping can 6256be anywhere in the user memory address space, as long as the memory slots are 6257aligned and sized to a segment (1MB) boundary. 6258 62598.16 KVM_CAP_S390_COW 6260--------------------- 6261 6262:Architectures: s390 6263 6264This capability indicates that the user space memory used as guest mapping can 6265use copy-on-write semantics as well as dirty pages tracking via read-only page 6266tables. 6267 62688.17 KVM_CAP_S390_BPB 6269--------------------- 6270 6271:Architectures: s390 6272 6273This capability indicates that kvm will implement the interfaces to handle 6274reset, migration and nested KVM for branch prediction blocking. The stfle 6275facility 82 should not be provided to the guest without this capability. 6276 62778.18 KVM_CAP_HYPERV_TLBFLUSH 6278---------------------------- 6279 6280:Architectures: x86 6281 6282This capability indicates that KVM supports paravirtualized Hyper-V TLB Flush 6283hypercalls: 6284HvFlushVirtualAddressSpace, HvFlushVirtualAddressSpaceEx, 6285HvFlushVirtualAddressList, HvFlushVirtualAddressListEx. 6286 62878.19 KVM_CAP_ARM_INJECT_SERROR_ESR 6288---------------------------------- 6289 6290:Architectures: arm, arm64 6291 6292This capability indicates that userspace can specify (via the 6293KVM_SET_VCPU_EVENTS ioctl) the syndrome value reported to the guest when it 6294takes a virtual SError interrupt exception. 6295If KVM advertises this capability, userspace can only specify the ISS field for 6296the ESR syndrome. Other parts of the ESR, such as the EC are generated by the 6297CPU when the exception is taken. If this virtual SError is taken to EL1 using 6298AArch64, this value will be reported in the ISS field of ESR_ELx. 6299 6300See KVM_CAP_VCPU_EVENTS for more details. 6301 63028.20 KVM_CAP_HYPERV_SEND_IPI 6303---------------------------- 6304 6305:Architectures: x86 6306 6307This capability indicates that KVM supports paravirtualized Hyper-V IPI send 6308hypercalls: 6309HvCallSendSyntheticClusterIpi, HvCallSendSyntheticClusterIpiEx. 6310 63118.21 KVM_CAP_HYPERV_DIRECT_TLBFLUSH 6312----------------------------------- 6313 6314:Architectures: x86 6315 6316This capability indicates that KVM running on top of Hyper-V hypervisor 6317enables Direct TLB flush for its guests meaning that TLB flush 6318hypercalls are handled by Level 0 hypervisor (Hyper-V) bypassing KVM. 6319Due to the different ABI for hypercall parameters between Hyper-V and 6320KVM, enabling this capability effectively disables all hypercall 6321handling by KVM (as some KVM hypercall may be mistakenly treated as TLB 6322flush hypercalls by Hyper-V) so userspace should disable KVM identification 6323in CPUID and only exposes Hyper-V identification. In this case, guest 6324thinks it's running on Hyper-V and only use Hyper-V hypercalls. 6325 63268.22 KVM_CAP_S390_VCPU_RESETS 6327----------------------------- 6328 6329:Architectures: s390 6330 6331This capability indicates that the KVM_S390_NORMAL_RESET and 6332KVM_S390_CLEAR_RESET ioctls are available. 6333 63348.23 KVM_CAP_S390_PROTECTED 6335--------------------------- 6336 6337:Architectures: s390 6338 6339This capability indicates that the Ultravisor has been initialized and 6340KVM can therefore start protected VMs. 6341This capability governs the KVM_S390_PV_COMMAND ioctl and the 6342KVM_MP_STATE_LOAD MP_STATE. KVM_SET_MP_STATE can fail for protected 6343guests when the state change is invalid. 6344 63458.24 KVM_CAP_STEAL_TIME 6346----------------------- 6347 6348:Architectures: arm64, x86 6349 6350This capability indicates that KVM supports steal time accounting. 6351When steal time accounting is supported it may be enabled with 6352architecture-specific interfaces. This capability and the architecture- 6353specific interfaces must be consistent, i.e. if one says the feature 6354is supported, than the other should as well and vice versa. For arm64 6355see Documentation/virt/kvm/devices/vcpu.rst "KVM_ARM_VCPU_PVTIME_CTRL". 6356For x86 see Documentation/virt/kvm/msr.rst "MSR_KVM_STEAL_TIME". 6357 63588.25 KVM_CAP_S390_DIAG318 6359------------------------- 6360 6361:Architectures: s390 6362 6363This capability enables a guest to set information about its control program 6364(i.e. guest kernel type and version). The information is helpful during 6365system/firmware service events, providing additional data about the guest 6366environments running on the machine. 6367 6368The information is associated with the DIAGNOSE 0x318 instruction, which sets 6369an 8-byte value consisting of a one-byte Control Program Name Code (CPNC) and 6370a 7-byte Control Program Version Code (CPVC). The CPNC determines what 6371environment the control program is running in (e.g. Linux, z/VM...), and the 6372CPVC is used for information specific to OS (e.g. Linux version, Linux 6373distribution...) 6374 6375If this capability is available, then the CPNC and CPVC can be synchronized 6376between KVM and userspace via the sync regs mechanism (KVM_SYNC_DIAG318). 6377 63788.26 KVM_CAP_X86_USER_SPACE_MSR 6379------------------------------- 6380 6381:Architectures: x86 6382 6383This capability indicates that KVM supports deflection of MSR reads and 6384writes to user space. It can be enabled on a VM level. If enabled, MSR 6385accesses that would usually trigger a #GP by KVM into the guest will 6386instead get bounced to user space through the KVM_EXIT_X86_RDMSR and 6387KVM_EXIT_X86_WRMSR exit notifications. 6388 63898.27 KVM_X86_SET_MSR_FILTER 6390--------------------------- 6391 6392:Architectures: x86 6393 6394This capability indicates that KVM supports that accesses to user defined MSRs 6395may be rejected. With this capability exposed, KVM exports new VM ioctl 6396KVM_X86_SET_MSR_FILTER which user space can call to specify bitmaps of MSR 6397ranges that KVM should reject access to. 6398 6399In combination with KVM_CAP_X86_USER_SPACE_MSR, this allows user space to 6400trap and emulate MSRs that are outside of the scope of KVM as well as 6401limit the attack surface on KVM's MSR emulation code. 6402 64038.28 KVM_CAP_ENFORCE_PV_CPUID 6404----------------------------- 6405 6406Architectures: x86 6407 6408When enabled, KVM will disable paravirtual features provided to the 6409guest according to the bits in the KVM_CPUID_FEATURES CPUID leaf 6410(0x40000001). Otherwise, a guest may use the paravirtual features 6411regardless of what has actually been exposed through the CPUID leaf. 6412 64139. Known KVM API problems 6414========================= 6415 6416In some cases, KVM's API has some inconsistencies or common pitfalls 6417that userspace need to be aware of. This section details some of 6418these issues. 6419 6420Most of them are architecture specific, so the section is split by 6421architecture. 6422 64239.1. x86 6424-------- 6425 6426``KVM_GET_SUPPORTED_CPUID`` issues 6427^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6428 6429In general, ``KVM_GET_SUPPORTED_CPUID`` is designed so that it is possible 6430to take its result and pass it directly to ``KVM_SET_CPUID2``. This section 6431documents some cases in which that requires some care. 6432 6433Local APIC features 6434~~~~~~~~~~~~~~~~~~~ 6435 6436CPU[EAX=1]:ECX[21] (X2APIC) is reported by ``KVM_GET_SUPPORTED_CPUID``, 6437but it can only be enabled if ``KVM_CREATE_IRQCHIP`` or 6438``KVM_ENABLE_CAP(KVM_CAP_IRQCHIP_SPLIT)`` are used to enable in-kernel emulation of 6439the local APIC. 6440 6441The same is true for the ``KVM_FEATURE_PV_UNHALT`` paravirtualized feature. 6442 6443CPU[EAX=1]:ECX[24] (TSC_DEADLINE) is not reported by ``KVM_GET_SUPPORTED_CPUID``. 6444It can be enabled if ``KVM_CAP_TSC_DEADLINE_TIMER`` is present and the kernel 6445has enabled in-kernel emulation of the local APIC. 6446 6447CPU topology 6448~~~~~~~~~~~~ 6449 6450Several CPUID values include topology information for the host CPU: 64510x0b and 0x1f for Intel systems, 0x8000001e for AMD systems. Different 6452versions of KVM return different values for this information and userspace 6453should not rely on it. Currently they return all zeroes. 6454 6455If userspace wishes to set up a guest topology, it should be careful that 6456the values of these three leaves differ for each CPU. In particular, 6457the APIC ID is found in EDX for all subleaves of 0x0b and 0x1f, and in EAX 6458for 0x8000001e; the latter also encodes the core id and node id in bits 64597:0 of EBX and ECX respectively. 6460 6461Obsolete ioctls and capabilities 6462^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6463 6464KVM_CAP_DISABLE_QUIRKS does not let userspace know which quirks are actually 6465available. Use ``KVM_CHECK_EXTENSION(KVM_CAP_DISABLE_QUIRKS2)`` instead if 6466available. 6467 6468Ordering of KVM_GET_*/KVM_SET_* ioctls 6469^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6470 6471TBD 6472