1The Definitive KVM (Kernel-based Virtual Machine) API Documentation 2=================================================================== 3 41. General description 5---------------------- 6 7The kvm API is a set of ioctls that are issued to control various aspects 8of a virtual machine. The ioctls belong to three classes 9 10 - System ioctls: These query and set global attributes which affect the 11 whole kvm subsystem. In addition a system ioctl is used to create 12 virtual machines 13 14 - VM ioctls: These query and set attributes that affect an entire virtual 15 machine, for example memory layout. In addition a VM ioctl is used to 16 create virtual cpus (vcpus). 17 18 Only run VM ioctls from the same process (address space) that was used 19 to create the VM. 20 21 - vcpu ioctls: These query and set attributes that control the operation 22 of a single virtual cpu. 23 24 Only run vcpu ioctls from the same thread that was used to create the 25 vcpu. 26 27 282. File descriptors 29------------------- 30 31The kvm API is centered around file descriptors. An initial 32open("/dev/kvm") obtains a handle to the kvm subsystem; this handle 33can be used to issue system ioctls. A KVM_CREATE_VM ioctl on this 34handle will create a VM file descriptor which can be used to issue VM 35ioctls. A KVM_CREATE_VCPU ioctl on a VM fd will create a virtual cpu 36and return a file descriptor pointing to it. Finally, ioctls on a vcpu 37fd can be used to control the vcpu, including the important task of 38actually running guest code. 39 40In general file descriptors can be migrated among processes by means 41of fork() and the SCM_RIGHTS facility of unix domain socket. These 42kinds of tricks are explicitly not supported by kvm. While they will 43not cause harm to the host, their actual behavior is not guaranteed by 44the API. The only supported use is one virtual machine per process, 45and one vcpu per thread. 46 47 483. Extensions 49------------- 50 51As of Linux 2.6.22, the KVM ABI has been stabilized: no backward 52incompatible change are allowed. However, there is an extension 53facility that allows backward-compatible extensions to the API to be 54queried and used. 55 56The extension mechanism is not based on the Linux version number. 57Instead, kvm defines extension identifiers and a facility to query 58whether a particular extension identifier is available. If it is, a 59set of ioctls is available for application use. 60 61 624. API description 63------------------ 64 65This section describes ioctls that can be used to control kvm guests. 66For each ioctl, the following information is provided along with a 67description: 68 69 Capability: which KVM extension provides this ioctl. Can be 'basic', 70 which means that is will be provided by any kernel that supports 71 API version 12 (see section 4.1), a KVM_CAP_xyz constant, which 72 means availability needs to be checked with KVM_CHECK_EXTENSION 73 (see section 4.4), or 'none' which means that while not all kernels 74 support this ioctl, there's no capability bit to check its 75 availability: for kernels that don't support the ioctl, 76 the ioctl returns -ENOTTY. 77 78 Architectures: which instruction set architectures provide this ioctl. 79 x86 includes both i386 and x86_64. 80 81 Type: system, vm, or vcpu. 82 83 Parameters: what parameters are accepted by the ioctl. 84 85 Returns: the return value. General error numbers (EBADF, ENOMEM, EINVAL) 86 are not detailed, but errors with specific meanings are. 87 88 894.1 KVM_GET_API_VERSION 90 91Capability: basic 92Architectures: all 93Type: system ioctl 94Parameters: none 95Returns: the constant KVM_API_VERSION (=12) 96 97This identifies the API version as the stable kvm API. It is not 98expected that this number will change. However, Linux 2.6.20 and 992.6.21 report earlier versions; these are not documented and not 100supported. Applications should refuse to run if KVM_GET_API_VERSION 101returns a value other than 12. If this check passes, all ioctls 102described as 'basic' will be available. 103 104 1054.2 KVM_CREATE_VM 106 107Capability: basic 108Architectures: all 109Type: system ioctl 110Parameters: machine type identifier (KVM_VM_*) 111Returns: a VM fd that can be used to control the new virtual machine. 112 113The new VM has no virtual cpus and no memory. 114You probably want to use 0 as machine type. 115 116In order to create user controlled virtual machines on S390, check 117KVM_CAP_S390_UCONTROL and use the flag KVM_VM_S390_UCONTROL as 118privileged user (CAP_SYS_ADMIN). 119 120To use hardware assisted virtualization on MIPS (VZ ASE) rather than 121the default trap & emulate implementation (which changes the virtual 122memory layout to fit in user mode), check KVM_CAP_MIPS_VZ and use the 123flag KVM_VM_MIPS_VZ. 124 125 1264.3 KVM_GET_MSR_INDEX_LIST, KVM_GET_MSR_FEATURE_INDEX_LIST 127 128Capability: basic, KVM_CAP_GET_MSR_FEATURES for KVM_GET_MSR_FEATURE_INDEX_LIST 129Architectures: x86 130Type: system ioctl 131Parameters: struct kvm_msr_list (in/out) 132Returns: 0 on success; -1 on error 133Errors: 134 EFAULT: the msr index list cannot be read from or written to 135 E2BIG: the msr index list is to be to fit in the array specified by 136 the user. 137 138struct kvm_msr_list { 139 __u32 nmsrs; /* number of msrs in entries */ 140 __u32 indices[0]; 141}; 142 143The user fills in the size of the indices array in nmsrs, and in return 144kvm adjusts nmsrs to reflect the actual number of msrs and fills in the 145indices array with their numbers. 146 147KVM_GET_MSR_INDEX_LIST returns the guest msrs that are supported. The list 148varies by kvm version and host processor, but does not change otherwise. 149 150Note: if kvm indicates supports MCE (KVM_CAP_MCE), then the MCE bank MSRs are 151not returned in the MSR list, as different vcpus can have a different number 152of banks, as set via the KVM_X86_SETUP_MCE ioctl. 153 154KVM_GET_MSR_FEATURE_INDEX_LIST returns the list of MSRs that can be passed 155to the KVM_GET_MSRS system ioctl. This lets userspace probe host capabilities 156and processor features that are exposed via MSRs (e.g., VMX capabilities). 157This list also varies by kvm version and host processor, but does not change 158otherwise. 159 160 1614.4 KVM_CHECK_EXTENSION 162 163Capability: basic, KVM_CAP_CHECK_EXTENSION_VM for vm ioctl 164Architectures: all 165Type: system ioctl, vm ioctl 166Parameters: extension identifier (KVM_CAP_*) 167Returns: 0 if unsupported; 1 (or some other positive integer) if supported 168 169The API allows the application to query about extensions to the core 170kvm API. Userspace passes an extension identifier (an integer) and 171receives an integer that describes the extension availability. 172Generally 0 means no and 1 means yes, but some extensions may report 173additional information in the integer return value. 174 175Based on their initialization different VMs may have different capabilities. 176It is thus encouraged to use the vm ioctl to query for capabilities (available 177with KVM_CAP_CHECK_EXTENSION_VM on the vm fd) 178 1794.5 KVM_GET_VCPU_MMAP_SIZE 180 181Capability: basic 182Architectures: all 183Type: system ioctl 184Parameters: none 185Returns: size of vcpu mmap area, in bytes 186 187The KVM_RUN ioctl (cf.) communicates with userspace via a shared 188memory region. This ioctl returns the size of that region. See the 189KVM_RUN documentation for details. 190 191 1924.6 KVM_SET_MEMORY_REGION 193 194Capability: basic 195Architectures: all 196Type: vm ioctl 197Parameters: struct kvm_memory_region (in) 198Returns: 0 on success, -1 on error 199 200This ioctl is obsolete and has been removed. 201 202 2034.7 KVM_CREATE_VCPU 204 205Capability: basic 206Architectures: all 207Type: vm ioctl 208Parameters: vcpu id (apic id on x86) 209Returns: vcpu fd on success, -1 on error 210 211This API adds a vcpu to a virtual machine. No more than max_vcpus may be added. 212The vcpu id is an integer in the range [0, max_vcpu_id). 213 214The recommended max_vcpus value can be retrieved using the KVM_CAP_NR_VCPUS of 215the KVM_CHECK_EXTENSION ioctl() at run-time. 216The maximum possible value for max_vcpus can be retrieved using the 217KVM_CAP_MAX_VCPUS of the KVM_CHECK_EXTENSION ioctl() at run-time. 218 219If the KVM_CAP_NR_VCPUS does not exist, you should assume that max_vcpus is 4 220cpus max. 221If the KVM_CAP_MAX_VCPUS does not exist, you should assume that max_vcpus is 222same as the value returned from KVM_CAP_NR_VCPUS. 223 224The maximum possible value for max_vcpu_id can be retrieved using the 225KVM_CAP_MAX_VCPU_ID of the KVM_CHECK_EXTENSION ioctl() at run-time. 226 227If the KVM_CAP_MAX_VCPU_ID does not exist, you should assume that max_vcpu_id 228is the same as the value returned from KVM_CAP_MAX_VCPUS. 229 230On powerpc using book3s_hv mode, the vcpus are mapped onto virtual 231threads in one or more virtual CPU cores. (This is because the 232hardware requires all the hardware threads in a CPU core to be in the 233same partition.) The KVM_CAP_PPC_SMT capability indicates the number 234of vcpus per virtual core (vcore). The vcore id is obtained by 235dividing the vcpu id by the number of vcpus per vcore. The vcpus in a 236given vcore will always be in the same physical core as each other 237(though that might be a different physical core from time to time). 238Userspace can control the threading (SMT) mode of the guest by its 239allocation of vcpu ids. For example, if userspace wants 240single-threaded guest vcpus, it should make all vcpu ids be a multiple 241of the number of vcpus per vcore. 242 243For virtual cpus that have been created with S390 user controlled virtual 244machines, the resulting vcpu fd can be memory mapped at page offset 245KVM_S390_SIE_PAGE_OFFSET in order to obtain a memory map of the virtual 246cpu's hardware control block. 247 248 2494.8 KVM_GET_DIRTY_LOG (vm ioctl) 250 251Capability: basic 252Architectures: x86 253Type: vm ioctl 254Parameters: struct kvm_dirty_log (in/out) 255Returns: 0 on success, -1 on error 256 257/* for KVM_GET_DIRTY_LOG */ 258struct kvm_dirty_log { 259 __u32 slot; 260 __u32 padding; 261 union { 262 void __user *dirty_bitmap; /* one bit per page */ 263 __u64 padding; 264 }; 265}; 266 267Given a memory slot, return a bitmap containing any pages dirtied 268since the last call to this ioctl. Bit 0 is the first page in the 269memory slot. Ensure the entire structure is cleared to avoid padding 270issues. 271 272If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 specifies 273the address space for which you want to return the dirty bitmap. 274They must be less than the value that KVM_CHECK_EXTENSION returns for 275the KVM_CAP_MULTI_ADDRESS_SPACE capability. 276 277 2784.9 KVM_SET_MEMORY_ALIAS 279 280Capability: basic 281Architectures: x86 282Type: vm ioctl 283Parameters: struct kvm_memory_alias (in) 284Returns: 0 (success), -1 (error) 285 286This ioctl is obsolete and has been removed. 287 288 2894.10 KVM_RUN 290 291Capability: basic 292Architectures: all 293Type: vcpu ioctl 294Parameters: none 295Returns: 0 on success, -1 on error 296Errors: 297 EINTR: an unmasked signal is pending 298 299This ioctl is used to run a guest virtual cpu. While there are no 300explicit parameters, there is an implicit parameter block that can be 301obtained by mmap()ing the vcpu fd at offset 0, with the size given by 302KVM_GET_VCPU_MMAP_SIZE. The parameter block is formatted as a 'struct 303kvm_run' (see below). 304 305 3064.11 KVM_GET_REGS 307 308Capability: basic 309Architectures: all except ARM, arm64 310Type: vcpu ioctl 311Parameters: struct kvm_regs (out) 312Returns: 0 on success, -1 on error 313 314Reads the general purpose registers from the vcpu. 315 316/* x86 */ 317struct kvm_regs { 318 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */ 319 __u64 rax, rbx, rcx, rdx; 320 __u64 rsi, rdi, rsp, rbp; 321 __u64 r8, r9, r10, r11; 322 __u64 r12, r13, r14, r15; 323 __u64 rip, rflags; 324}; 325 326/* mips */ 327struct kvm_regs { 328 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */ 329 __u64 gpr[32]; 330 __u64 hi; 331 __u64 lo; 332 __u64 pc; 333}; 334 335 3364.12 KVM_SET_REGS 337 338Capability: basic 339Architectures: all except ARM, arm64 340Type: vcpu ioctl 341Parameters: struct kvm_regs (in) 342Returns: 0 on success, -1 on error 343 344Writes the general purpose registers into the vcpu. 345 346See KVM_GET_REGS for the data structure. 347 348 3494.13 KVM_GET_SREGS 350 351Capability: basic 352Architectures: x86, ppc 353Type: vcpu ioctl 354Parameters: struct kvm_sregs (out) 355Returns: 0 on success, -1 on error 356 357Reads special registers from the vcpu. 358 359/* x86 */ 360struct kvm_sregs { 361 struct kvm_segment cs, ds, es, fs, gs, ss; 362 struct kvm_segment tr, ldt; 363 struct kvm_dtable gdt, idt; 364 __u64 cr0, cr2, cr3, cr4, cr8; 365 __u64 efer; 366 __u64 apic_base; 367 __u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64]; 368}; 369 370/* ppc -- see arch/powerpc/include/uapi/asm/kvm.h */ 371 372interrupt_bitmap is a bitmap of pending external interrupts. At most 373one bit may be set. This interrupt has been acknowledged by the APIC 374but not yet injected into the cpu core. 375 376 3774.14 KVM_SET_SREGS 378 379Capability: basic 380Architectures: x86, ppc 381Type: vcpu ioctl 382Parameters: struct kvm_sregs (in) 383Returns: 0 on success, -1 on error 384 385Writes special registers into the vcpu. See KVM_GET_SREGS for the 386data structures. 387 388 3894.15 KVM_TRANSLATE 390 391Capability: basic 392Architectures: x86 393Type: vcpu ioctl 394Parameters: struct kvm_translation (in/out) 395Returns: 0 on success, -1 on error 396 397Translates a virtual address according to the vcpu's current address 398translation mode. 399 400struct kvm_translation { 401 /* in */ 402 __u64 linear_address; 403 404 /* out */ 405 __u64 physical_address; 406 __u8 valid; 407 __u8 writeable; 408 __u8 usermode; 409 __u8 pad[5]; 410}; 411 412 4134.16 KVM_INTERRUPT 414 415Capability: basic 416Architectures: x86, ppc, mips 417Type: vcpu ioctl 418Parameters: struct kvm_interrupt (in) 419Returns: 0 on success, negative on failure. 420 421Queues a hardware interrupt vector to be injected. 422 423/* for KVM_INTERRUPT */ 424struct kvm_interrupt { 425 /* in */ 426 __u32 irq; 427}; 428 429X86: 430 431Returns: 0 on success, 432 -EEXIST if an interrupt is already enqueued 433 -EINVAL the the irq number is invalid 434 -ENXIO if the PIC is in the kernel 435 -EFAULT if the pointer is invalid 436 437Note 'irq' is an interrupt vector, not an interrupt pin or line. This 438ioctl is useful if the in-kernel PIC is not used. 439 440PPC: 441 442Queues an external interrupt to be injected. This ioctl is overleaded 443with 3 different irq values: 444 445a) KVM_INTERRUPT_SET 446 447 This injects an edge type external interrupt into the guest once it's ready 448 to receive interrupts. When injected, the interrupt is done. 449 450b) KVM_INTERRUPT_UNSET 451 452 This unsets any pending interrupt. 453 454 Only available with KVM_CAP_PPC_UNSET_IRQ. 455 456c) KVM_INTERRUPT_SET_LEVEL 457 458 This injects a level type external interrupt into the guest context. The 459 interrupt stays pending until a specific ioctl with KVM_INTERRUPT_UNSET 460 is triggered. 461 462 Only available with KVM_CAP_PPC_IRQ_LEVEL. 463 464Note that any value for 'irq' other than the ones stated above is invalid 465and incurs unexpected behavior. 466 467MIPS: 468 469Queues an external interrupt to be injected into the virtual CPU. A negative 470interrupt number dequeues the interrupt. 471 472 4734.17 KVM_DEBUG_GUEST 474 475Capability: basic 476Architectures: none 477Type: vcpu ioctl 478Parameters: none) 479Returns: -1 on error 480 481Support for this has been removed. Use KVM_SET_GUEST_DEBUG instead. 482 483 4844.18 KVM_GET_MSRS 485 486Capability: basic (vcpu), KVM_CAP_GET_MSR_FEATURES (system) 487Architectures: x86 488Type: system ioctl, vcpu ioctl 489Parameters: struct kvm_msrs (in/out) 490Returns: number of msrs successfully returned; 491 -1 on error 492 493When used as a system ioctl: 494Reads the values of MSR-based features that are available for the VM. This 495is similar to KVM_GET_SUPPORTED_CPUID, but it returns MSR indices and values. 496The list of msr-based features can be obtained using KVM_GET_MSR_FEATURE_INDEX_LIST 497in a system ioctl. 498 499When used as a vcpu ioctl: 500Reads model-specific registers from the vcpu. Supported msr indices can 501be obtained using KVM_GET_MSR_INDEX_LIST in a system ioctl. 502 503struct kvm_msrs { 504 __u32 nmsrs; /* number of msrs in entries */ 505 __u32 pad; 506 507 struct kvm_msr_entry entries[0]; 508}; 509 510struct kvm_msr_entry { 511 __u32 index; 512 __u32 reserved; 513 __u64 data; 514}; 515 516Application code should set the 'nmsrs' member (which indicates the 517size of the entries array) and the 'index' member of each array entry. 518kvm will fill in the 'data' member. 519 520 5214.19 KVM_SET_MSRS 522 523Capability: basic 524Architectures: x86 525Type: vcpu ioctl 526Parameters: struct kvm_msrs (in) 527Returns: 0 on success, -1 on error 528 529Writes model-specific registers to the vcpu. See KVM_GET_MSRS for the 530data structures. 531 532Application code should set the 'nmsrs' member (which indicates the 533size of the entries array), and the 'index' and 'data' members of each 534array entry. 535 536 5374.20 KVM_SET_CPUID 538 539Capability: basic 540Architectures: x86 541Type: vcpu ioctl 542Parameters: struct kvm_cpuid (in) 543Returns: 0 on success, -1 on error 544 545Defines the vcpu responses to the cpuid instruction. Applications 546should use the KVM_SET_CPUID2 ioctl if available. 547 548 549struct kvm_cpuid_entry { 550 __u32 function; 551 __u32 eax; 552 __u32 ebx; 553 __u32 ecx; 554 __u32 edx; 555 __u32 padding; 556}; 557 558/* for KVM_SET_CPUID */ 559struct kvm_cpuid { 560 __u32 nent; 561 __u32 padding; 562 struct kvm_cpuid_entry entries[0]; 563}; 564 565 5664.21 KVM_SET_SIGNAL_MASK 567 568Capability: basic 569Architectures: all 570Type: vcpu ioctl 571Parameters: struct kvm_signal_mask (in) 572Returns: 0 on success, -1 on error 573 574Defines which signals are blocked during execution of KVM_RUN. This 575signal mask temporarily overrides the threads signal mask. Any 576unblocked signal received (except SIGKILL and SIGSTOP, which retain 577their traditional behaviour) will cause KVM_RUN to return with -EINTR. 578 579Note the signal will only be delivered if not blocked by the original 580signal mask. 581 582/* for KVM_SET_SIGNAL_MASK */ 583struct kvm_signal_mask { 584 __u32 len; 585 __u8 sigset[0]; 586}; 587 588 5894.22 KVM_GET_FPU 590 591Capability: basic 592Architectures: x86 593Type: vcpu ioctl 594Parameters: struct kvm_fpu (out) 595Returns: 0 on success, -1 on error 596 597Reads the floating point state from the vcpu. 598 599/* for KVM_GET_FPU and KVM_SET_FPU */ 600struct kvm_fpu { 601 __u8 fpr[8][16]; 602 __u16 fcw; 603 __u16 fsw; 604 __u8 ftwx; /* in fxsave format */ 605 __u8 pad1; 606 __u16 last_opcode; 607 __u64 last_ip; 608 __u64 last_dp; 609 __u8 xmm[16][16]; 610 __u32 mxcsr; 611 __u32 pad2; 612}; 613 614 6154.23 KVM_SET_FPU 616 617Capability: basic 618Architectures: x86 619Type: vcpu ioctl 620Parameters: struct kvm_fpu (in) 621Returns: 0 on success, -1 on error 622 623Writes the floating point state to the vcpu. 624 625/* for KVM_GET_FPU and KVM_SET_FPU */ 626struct kvm_fpu { 627 __u8 fpr[8][16]; 628 __u16 fcw; 629 __u16 fsw; 630 __u8 ftwx; /* in fxsave format */ 631 __u8 pad1; 632 __u16 last_opcode; 633 __u64 last_ip; 634 __u64 last_dp; 635 __u8 xmm[16][16]; 636 __u32 mxcsr; 637 __u32 pad2; 638}; 639 640 6414.24 KVM_CREATE_IRQCHIP 642 643Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQCHIP (s390) 644Architectures: x86, ARM, arm64, s390 645Type: vm ioctl 646Parameters: none 647Returns: 0 on success, -1 on error 648 649Creates an interrupt controller model in the kernel. 650On x86, creates a virtual ioapic, a virtual PIC (two PICs, nested), and sets up 651future vcpus to have a local APIC. IRQ routing for GSIs 0-15 is set to both 652PIC and IOAPIC; GSI 16-23 only go to the IOAPIC. 653On ARM/arm64, a GICv2 is created. Any other GIC versions require the usage of 654KVM_CREATE_DEVICE, which also supports creating a GICv2. Using 655KVM_CREATE_DEVICE is preferred over KVM_CREATE_IRQCHIP for GICv2. 656On s390, a dummy irq routing table is created. 657 658Note that on s390 the KVM_CAP_S390_IRQCHIP vm capability needs to be enabled 659before KVM_CREATE_IRQCHIP can be used. 660 661 6624.25 KVM_IRQ_LINE 663 664Capability: KVM_CAP_IRQCHIP 665Architectures: x86, arm, arm64 666Type: vm ioctl 667Parameters: struct kvm_irq_level 668Returns: 0 on success, -1 on error 669 670Sets the level of a GSI input to the interrupt controller model in the kernel. 671On some architectures it is required that an interrupt controller model has 672been previously created with KVM_CREATE_IRQCHIP. Note that edge-triggered 673interrupts require the level to be set to 1 and then back to 0. 674 675On real hardware, interrupt pins can be active-low or active-high. This 676does not matter for the level field of struct kvm_irq_level: 1 always 677means active (asserted), 0 means inactive (deasserted). 678 679x86 allows the operating system to program the interrupt polarity 680(active-low/active-high) for level-triggered interrupts, and KVM used 681to consider the polarity. However, due to bitrot in the handling of 682active-low interrupts, the above convention is now valid on x86 too. 683This is signaled by KVM_CAP_X86_IOAPIC_POLARITY_IGNORED. Userspace 684should not present interrupts to the guest as active-low unless this 685capability is present (or unless it is not using the in-kernel irqchip, 686of course). 687 688 689ARM/arm64 can signal an interrupt either at the CPU level, or at the 690in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to 691use PPIs designated for specific cpus. The irq field is interpreted 692like this: 693 694 bits: | 31 ... 24 | 23 ... 16 | 15 ... 0 | 695 field: | irq_type | vcpu_index | irq_id | 696 697The irq_type field has the following values: 698- irq_type[0]: out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ 699- irq_type[1]: in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.) 700 (the vcpu_index field is ignored) 701- irq_type[2]: in-kernel GIC: PPI, irq_id between 16 and 31 (incl.) 702 703(The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs) 704 705In both cases, level is used to assert/deassert the line. 706 707struct kvm_irq_level { 708 union { 709 __u32 irq; /* GSI */ 710 __s32 status; /* not used for KVM_IRQ_LEVEL */ 711 }; 712 __u32 level; /* 0 or 1 */ 713}; 714 715 7164.26 KVM_GET_IRQCHIP 717 718Capability: KVM_CAP_IRQCHIP 719Architectures: x86 720Type: vm ioctl 721Parameters: struct kvm_irqchip (in/out) 722Returns: 0 on success, -1 on error 723 724Reads the state of a kernel interrupt controller created with 725KVM_CREATE_IRQCHIP into a buffer provided by the caller. 726 727struct kvm_irqchip { 728 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */ 729 __u32 pad; 730 union { 731 char dummy[512]; /* reserving space */ 732 struct kvm_pic_state pic; 733 struct kvm_ioapic_state ioapic; 734 } chip; 735}; 736 737 7384.27 KVM_SET_IRQCHIP 739 740Capability: KVM_CAP_IRQCHIP 741Architectures: x86 742Type: vm ioctl 743Parameters: struct kvm_irqchip (in) 744Returns: 0 on success, -1 on error 745 746Sets the state of a kernel interrupt controller created with 747KVM_CREATE_IRQCHIP from a buffer provided by the caller. 748 749struct kvm_irqchip { 750 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */ 751 __u32 pad; 752 union { 753 char dummy[512]; /* reserving space */ 754 struct kvm_pic_state pic; 755 struct kvm_ioapic_state ioapic; 756 } chip; 757}; 758 759 7604.28 KVM_XEN_HVM_CONFIG 761 762Capability: KVM_CAP_XEN_HVM 763Architectures: x86 764Type: vm ioctl 765Parameters: struct kvm_xen_hvm_config (in) 766Returns: 0 on success, -1 on error 767 768Sets the MSR that the Xen HVM guest uses to initialize its hypercall 769page, and provides the starting address and size of the hypercall 770blobs in userspace. When the guest writes the MSR, kvm copies one 771page of a blob (32- or 64-bit, depending on the vcpu mode) to guest 772memory. 773 774struct kvm_xen_hvm_config { 775 __u32 flags; 776 __u32 msr; 777 __u64 blob_addr_32; 778 __u64 blob_addr_64; 779 __u8 blob_size_32; 780 __u8 blob_size_64; 781 __u8 pad2[30]; 782}; 783 784 7854.29 KVM_GET_CLOCK 786 787Capability: KVM_CAP_ADJUST_CLOCK 788Architectures: x86 789Type: vm ioctl 790Parameters: struct kvm_clock_data (out) 791Returns: 0 on success, -1 on error 792 793Gets the current timestamp of kvmclock as seen by the current guest. In 794conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios 795such as migration. 796 797When KVM_CAP_ADJUST_CLOCK is passed to KVM_CHECK_EXTENSION, it returns the 798set of bits that KVM can return in struct kvm_clock_data's flag member. 799 800The only flag defined now is KVM_CLOCK_TSC_STABLE. If set, the returned 801value is the exact kvmclock value seen by all VCPUs at the instant 802when KVM_GET_CLOCK was called. If clear, the returned value is simply 803CLOCK_MONOTONIC plus a constant offset; the offset can be modified 804with KVM_SET_CLOCK. KVM will try to make all VCPUs follow this clock, 805but the exact value read by each VCPU could differ, because the host 806TSC is not stable. 807 808struct kvm_clock_data { 809 __u64 clock; /* kvmclock current value */ 810 __u32 flags; 811 __u32 pad[9]; 812}; 813 814 8154.30 KVM_SET_CLOCK 816 817Capability: KVM_CAP_ADJUST_CLOCK 818Architectures: x86 819Type: vm ioctl 820Parameters: struct kvm_clock_data (in) 821Returns: 0 on success, -1 on error 822 823Sets the current timestamp of kvmclock to the value specified in its parameter. 824In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios 825such as migration. 826 827struct kvm_clock_data { 828 __u64 clock; /* kvmclock current value */ 829 __u32 flags; 830 __u32 pad[9]; 831}; 832 833 8344.31 KVM_GET_VCPU_EVENTS 835 836Capability: KVM_CAP_VCPU_EVENTS 837Extended by: KVM_CAP_INTR_SHADOW 838Architectures: x86, arm, arm64 839Type: vcpu ioctl 840Parameters: struct kvm_vcpu_event (out) 841Returns: 0 on success, -1 on error 842 843X86: 844 845Gets currently pending exceptions, interrupts, and NMIs as well as related 846states of the vcpu. 847 848struct kvm_vcpu_events { 849 struct { 850 __u8 injected; 851 __u8 nr; 852 __u8 has_error_code; 853 __u8 pad; 854 __u32 error_code; 855 } exception; 856 struct { 857 __u8 injected; 858 __u8 nr; 859 __u8 soft; 860 __u8 shadow; 861 } interrupt; 862 struct { 863 __u8 injected; 864 __u8 pending; 865 __u8 masked; 866 __u8 pad; 867 } nmi; 868 __u32 sipi_vector; 869 __u32 flags; 870 struct { 871 __u8 smm; 872 __u8 pending; 873 __u8 smm_inside_nmi; 874 __u8 latched_init; 875 } smi; 876}; 877 878Only two fields are defined in the flags field: 879 880- KVM_VCPUEVENT_VALID_SHADOW may be set in the flags field to signal that 881 interrupt.shadow contains a valid state. 882 883- KVM_VCPUEVENT_VALID_SMM may be set in the flags field to signal that 884 smi contains a valid state. 885 886ARM/ARM64: 887 888If the guest accesses a device that is being emulated by the host kernel in 889such a way that a real device would generate a physical SError, KVM may make 890a virtual SError pending for that VCPU. This system error interrupt remains 891pending until the guest takes the exception by unmasking PSTATE.A. 892 893Running the VCPU may cause it to take a pending SError, or make an access that 894causes an SError to become pending. The event's description is only valid while 895the VPCU is not running. 896 897This API provides a way to read and write the pending 'event' state that is not 898visible to the guest. To save, restore or migrate a VCPU the struct representing 899the state can be read then written using this GET/SET API, along with the other 900guest-visible registers. It is not possible to 'cancel' an SError that has been 901made pending. 902 903A device being emulated in user-space may also wish to generate an SError. To do 904this the events structure can be populated by user-space. The current state 905should be read first, to ensure no existing SError is pending. If an existing 906SError is pending, the architecture's 'Multiple SError interrupts' rules should 907be followed. (2.5.3 of DDI0587.a "ARM Reliability, Availability, and 908Serviceability (RAS) Specification"). 909 910SError exceptions always have an ESR value. Some CPUs have the ability to 911specify what the virtual SError's ESR value should be. These systems will 912advertise KVM_CAP_ARM_INJECT_SERROR_ESR. In this case exception.has_esr will 913always have a non-zero value when read, and the agent making an SError pending 914should specify the ISS field in the lower 24 bits of exception.serror_esr. If 915the system supports KVM_CAP_ARM_INJECT_SERROR_ESR, but user-space sets the events 916with exception.has_esr as zero, KVM will choose an ESR. 917 918Specifying exception.has_esr on a system that does not support it will return 919-EINVAL. Setting anything other than the lower 24bits of exception.serror_esr 920will return -EINVAL. 921 922struct kvm_vcpu_events { 923 struct { 924 __u8 serror_pending; 925 __u8 serror_has_esr; 926 /* Align it to 8 bytes */ 927 __u8 pad[6]; 928 __u64 serror_esr; 929 } exception; 930 __u32 reserved[12]; 931}; 932 9334.32 KVM_SET_VCPU_EVENTS 934 935Capability: KVM_CAP_VCPU_EVENTS 936Extended by: KVM_CAP_INTR_SHADOW 937Architectures: x86, arm, arm64 938Type: vcpu ioctl 939Parameters: struct kvm_vcpu_event (in) 940Returns: 0 on success, -1 on error 941 942X86: 943 944Set pending exceptions, interrupts, and NMIs as well as related states of the 945vcpu. 946 947See KVM_GET_VCPU_EVENTS for the data structure. 948 949Fields that may be modified asynchronously by running VCPUs can be excluded 950from the update. These fields are nmi.pending, sipi_vector, smi.smm, 951smi.pending. Keep the corresponding bits in the flags field cleared to 952suppress overwriting the current in-kernel state. The bits are: 953 954KVM_VCPUEVENT_VALID_NMI_PENDING - transfer nmi.pending to the kernel 955KVM_VCPUEVENT_VALID_SIPI_VECTOR - transfer sipi_vector 956KVM_VCPUEVENT_VALID_SMM - transfer the smi sub-struct. 957 958If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in 959the flags field to signal that interrupt.shadow contains a valid state and 960shall be written into the VCPU. 961 962KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available. 963 964ARM/ARM64: 965 966Set the pending SError exception state for this VCPU. It is not possible to 967'cancel' an Serror that has been made pending. 968 969See KVM_GET_VCPU_EVENTS for the data structure. 970 971 9724.33 KVM_GET_DEBUGREGS 973 974Capability: KVM_CAP_DEBUGREGS 975Architectures: x86 976Type: vm ioctl 977Parameters: struct kvm_debugregs (out) 978Returns: 0 on success, -1 on error 979 980Reads debug registers from the vcpu. 981 982struct kvm_debugregs { 983 __u64 db[4]; 984 __u64 dr6; 985 __u64 dr7; 986 __u64 flags; 987 __u64 reserved[9]; 988}; 989 990 9914.34 KVM_SET_DEBUGREGS 992 993Capability: KVM_CAP_DEBUGREGS 994Architectures: x86 995Type: vm ioctl 996Parameters: struct kvm_debugregs (in) 997Returns: 0 on success, -1 on error 998 999Writes debug registers into the vcpu. 1000 1001See KVM_GET_DEBUGREGS for the data structure. The flags field is unused 1002yet and must be cleared on entry. 1003 1004 10054.35 KVM_SET_USER_MEMORY_REGION 1006 1007Capability: KVM_CAP_USER_MEM 1008Architectures: all 1009Type: vm ioctl 1010Parameters: struct kvm_userspace_memory_region (in) 1011Returns: 0 on success, -1 on error 1012 1013struct kvm_userspace_memory_region { 1014 __u32 slot; 1015 __u32 flags; 1016 __u64 guest_phys_addr; 1017 __u64 memory_size; /* bytes */ 1018 __u64 userspace_addr; /* start of the userspace allocated memory */ 1019}; 1020 1021/* for kvm_memory_region::flags */ 1022#define KVM_MEM_LOG_DIRTY_PAGES (1UL << 0) 1023#define KVM_MEM_READONLY (1UL << 1) 1024 1025This ioctl allows the user to create or modify a guest physical memory 1026slot. When changing an existing slot, it may be moved in the guest 1027physical memory space, or its flags may be modified. It may not be 1028resized. Slots may not overlap in guest physical address space. 1029Bits 0-15 of "slot" specifies the slot id and this value should be 1030less than the maximum number of user memory slots supported per VM. 1031The maximum allowed slots can be queried using KVM_CAP_NR_MEMSLOTS, 1032if this capability is supported by the architecture. 1033 1034If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot" 1035specifies the address space which is being modified. They must be 1036less than the value that KVM_CHECK_EXTENSION returns for the 1037KVM_CAP_MULTI_ADDRESS_SPACE capability. Slots in separate address spaces 1038are unrelated; the restriction on overlapping slots only applies within 1039each address space. 1040 1041Memory for the region is taken starting at the address denoted by the 1042field userspace_addr, which must point at user addressable memory for 1043the entire memory slot size. Any object may back this memory, including 1044anonymous memory, ordinary files, and hugetlbfs. 1045 1046It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr 1047be identical. This allows large pages in the guest to be backed by large 1048pages in the host. 1049 1050The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and 1051KVM_MEM_READONLY. The former can be set to instruct KVM to keep track of 1052writes to memory within the slot. See KVM_GET_DIRTY_LOG ioctl to know how to 1053use it. The latter can be set, if KVM_CAP_READONLY_MEM capability allows it, 1054to make a new slot read-only. In this case, writes to this memory will be 1055posted to userspace as KVM_EXIT_MMIO exits. 1056 1057When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of 1058the memory region are automatically reflected into the guest. For example, an 1059mmap() that affects the region will be made visible immediately. Another 1060example is madvise(MADV_DROP). 1061 1062It is recommended to use this API instead of the KVM_SET_MEMORY_REGION ioctl. 1063The KVM_SET_MEMORY_REGION does not allow fine grained control over memory 1064allocation and is deprecated. 1065 1066 10674.36 KVM_SET_TSS_ADDR 1068 1069Capability: KVM_CAP_SET_TSS_ADDR 1070Architectures: x86 1071Type: vm ioctl 1072Parameters: unsigned long tss_address (in) 1073Returns: 0 on success, -1 on error 1074 1075This ioctl defines the physical address of a three-page region in the guest 1076physical address space. The region must be within the first 4GB of the 1077guest physical address space and must not conflict with any memory slot 1078or any mmio address. The guest may malfunction if it accesses this memory 1079region. 1080 1081This ioctl is required on Intel-based hosts. This is needed on Intel hardware 1082because of a quirk in the virtualization implementation (see the internals 1083documentation when it pops into existence). 1084 1085 10864.37 KVM_ENABLE_CAP 1087 1088Capability: KVM_CAP_ENABLE_CAP, KVM_CAP_ENABLE_CAP_VM 1089Architectures: x86 (only KVM_CAP_ENABLE_CAP_VM), 1090 mips (only KVM_CAP_ENABLE_CAP), ppc, s390 1091Type: vcpu ioctl, vm ioctl (with KVM_CAP_ENABLE_CAP_VM) 1092Parameters: struct kvm_enable_cap (in) 1093Returns: 0 on success; -1 on error 1094 1095+Not all extensions are enabled by default. Using this ioctl the application 1096can enable an extension, making it available to the guest. 1097 1098On systems that do not support this ioctl, it always fails. On systems that 1099do support it, it only works for extensions that are supported for enablement. 1100 1101To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should 1102be used. 1103 1104struct kvm_enable_cap { 1105 /* in */ 1106 __u32 cap; 1107 1108The capability that is supposed to get enabled. 1109 1110 __u32 flags; 1111 1112A bitfield indicating future enhancements. Has to be 0 for now. 1113 1114 __u64 args[4]; 1115 1116Arguments for enabling a feature. If a feature needs initial values to 1117function properly, this is the place to put them. 1118 1119 __u8 pad[64]; 1120}; 1121 1122The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl 1123for vm-wide capabilities. 1124 11254.38 KVM_GET_MP_STATE 1126 1127Capability: KVM_CAP_MP_STATE 1128Architectures: x86, s390, arm, arm64 1129Type: vcpu ioctl 1130Parameters: struct kvm_mp_state (out) 1131Returns: 0 on success; -1 on error 1132 1133struct kvm_mp_state { 1134 __u32 mp_state; 1135}; 1136 1137Returns the vcpu's current "multiprocessing state" (though also valid on 1138uniprocessor guests). 1139 1140Possible values are: 1141 1142 - KVM_MP_STATE_RUNNABLE: the vcpu is currently running [x86,arm/arm64] 1143 - KVM_MP_STATE_UNINITIALIZED: the vcpu is an application processor (AP) 1144 which has not yet received an INIT signal [x86] 1145 - KVM_MP_STATE_INIT_RECEIVED: the vcpu has received an INIT signal, and is 1146 now ready for a SIPI [x86] 1147 - KVM_MP_STATE_HALTED: the vcpu has executed a HLT instruction and 1148 is waiting for an interrupt [x86] 1149 - KVM_MP_STATE_SIPI_RECEIVED: the vcpu has just received a SIPI (vector 1150 accessible via KVM_GET_VCPU_EVENTS) [x86] 1151 - KVM_MP_STATE_STOPPED: the vcpu is stopped [s390,arm/arm64] 1152 - KVM_MP_STATE_CHECK_STOP: the vcpu is in a special error state [s390] 1153 - KVM_MP_STATE_OPERATING: the vcpu is operating (running or halted) 1154 [s390] 1155 - KVM_MP_STATE_LOAD: the vcpu is in a special load/startup state 1156 [s390] 1157 1158On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an 1159in-kernel irqchip, the multiprocessing state must be maintained by userspace on 1160these architectures. 1161 1162For arm/arm64: 1163 1164The only states that are valid are KVM_MP_STATE_STOPPED and 1165KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not. 1166 11674.39 KVM_SET_MP_STATE 1168 1169Capability: KVM_CAP_MP_STATE 1170Architectures: x86, s390, arm, arm64 1171Type: vcpu ioctl 1172Parameters: struct kvm_mp_state (in) 1173Returns: 0 on success; -1 on error 1174 1175Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for 1176arguments. 1177 1178On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an 1179in-kernel irqchip, the multiprocessing state must be maintained by userspace on 1180these architectures. 1181 1182For arm/arm64: 1183 1184The only states that are valid are KVM_MP_STATE_STOPPED and 1185KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not. 1186 11874.40 KVM_SET_IDENTITY_MAP_ADDR 1188 1189Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR 1190Architectures: x86 1191Type: vm ioctl 1192Parameters: unsigned long identity (in) 1193Returns: 0 on success, -1 on error 1194 1195This ioctl defines the physical address of a one-page region in the guest 1196physical address space. The region must be within the first 4GB of the 1197guest physical address space and must not conflict with any memory slot 1198or any mmio address. The guest may malfunction if it accesses this memory 1199region. 1200 1201Setting the address to 0 will result in resetting the address to its default 1202(0xfffbc000). 1203 1204This ioctl is required on Intel-based hosts. This is needed on Intel hardware 1205because of a quirk in the virtualization implementation (see the internals 1206documentation when it pops into existence). 1207 1208Fails if any VCPU has already been created. 1209 12104.41 KVM_SET_BOOT_CPU_ID 1211 1212Capability: KVM_CAP_SET_BOOT_CPU_ID 1213Architectures: x86 1214Type: vm ioctl 1215Parameters: unsigned long vcpu_id 1216Returns: 0 on success, -1 on error 1217 1218Define which vcpu is the Bootstrap Processor (BSP). Values are the same 1219as the vcpu id in KVM_CREATE_VCPU. If this ioctl is not called, the default 1220is vcpu 0. 1221 1222 12234.42 KVM_GET_XSAVE 1224 1225Capability: KVM_CAP_XSAVE 1226Architectures: x86 1227Type: vcpu ioctl 1228Parameters: struct kvm_xsave (out) 1229Returns: 0 on success, -1 on error 1230 1231struct kvm_xsave { 1232 __u32 region[1024]; 1233}; 1234 1235This ioctl would copy current vcpu's xsave struct to the userspace. 1236 1237 12384.43 KVM_SET_XSAVE 1239 1240Capability: KVM_CAP_XSAVE 1241Architectures: x86 1242Type: vcpu ioctl 1243Parameters: struct kvm_xsave (in) 1244Returns: 0 on success, -1 on error 1245 1246struct kvm_xsave { 1247 __u32 region[1024]; 1248}; 1249 1250This ioctl would copy userspace's xsave struct to the kernel. 1251 1252 12534.44 KVM_GET_XCRS 1254 1255Capability: KVM_CAP_XCRS 1256Architectures: x86 1257Type: vcpu ioctl 1258Parameters: struct kvm_xcrs (out) 1259Returns: 0 on success, -1 on error 1260 1261struct kvm_xcr { 1262 __u32 xcr; 1263 __u32 reserved; 1264 __u64 value; 1265}; 1266 1267struct kvm_xcrs { 1268 __u32 nr_xcrs; 1269 __u32 flags; 1270 struct kvm_xcr xcrs[KVM_MAX_XCRS]; 1271 __u64 padding[16]; 1272}; 1273 1274This ioctl would copy current vcpu's xcrs to the userspace. 1275 1276 12774.45 KVM_SET_XCRS 1278 1279Capability: KVM_CAP_XCRS 1280Architectures: x86 1281Type: vcpu ioctl 1282Parameters: struct kvm_xcrs (in) 1283Returns: 0 on success, -1 on error 1284 1285struct kvm_xcr { 1286 __u32 xcr; 1287 __u32 reserved; 1288 __u64 value; 1289}; 1290 1291struct kvm_xcrs { 1292 __u32 nr_xcrs; 1293 __u32 flags; 1294 struct kvm_xcr xcrs[KVM_MAX_XCRS]; 1295 __u64 padding[16]; 1296}; 1297 1298This ioctl would set vcpu's xcr to the value userspace specified. 1299 1300 13014.46 KVM_GET_SUPPORTED_CPUID 1302 1303Capability: KVM_CAP_EXT_CPUID 1304Architectures: x86 1305Type: system ioctl 1306Parameters: struct kvm_cpuid2 (in/out) 1307Returns: 0 on success, -1 on error 1308 1309struct kvm_cpuid2 { 1310 __u32 nent; 1311 __u32 padding; 1312 struct kvm_cpuid_entry2 entries[0]; 1313}; 1314 1315#define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0) 1316#define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) 1317#define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) 1318 1319struct kvm_cpuid_entry2 { 1320 __u32 function; 1321 __u32 index; 1322 __u32 flags; 1323 __u32 eax; 1324 __u32 ebx; 1325 __u32 ecx; 1326 __u32 edx; 1327 __u32 padding[3]; 1328}; 1329 1330This ioctl returns x86 cpuid features which are supported by both the 1331hardware and kvm in its default configuration. Userspace can use the 1332information returned by this ioctl to construct cpuid information (for 1333KVM_SET_CPUID2) that is consistent with hardware, kernel, and 1334userspace capabilities, and with user requirements (for example, the 1335user may wish to constrain cpuid to emulate older hardware, or for 1336feature consistency across a cluster). 1337 1338Note that certain capabilities, such as KVM_CAP_X86_DISABLE_EXITS, may 1339expose cpuid features (e.g. MONITOR) which are not supported by kvm in 1340its default configuration. If userspace enables such capabilities, it 1341is responsible for modifying the results of this ioctl appropriately. 1342 1343Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure 1344with the 'nent' field indicating the number of entries in the variable-size 1345array 'entries'. If the number of entries is too low to describe the cpu 1346capabilities, an error (E2BIG) is returned. If the number is too high, 1347the 'nent' field is adjusted and an error (ENOMEM) is returned. If the 1348number is just right, the 'nent' field is adjusted to the number of valid 1349entries in the 'entries' array, which is then filled. 1350 1351The entries returned are the host cpuid as returned by the cpuid instruction, 1352with unknown or unsupported features masked out. Some features (for example, 1353x2apic), may not be present in the host cpu, but are exposed by kvm if it can 1354emulate them efficiently. The fields in each entry are defined as follows: 1355 1356 function: the eax value used to obtain the entry 1357 index: the ecx value used to obtain the entry (for entries that are 1358 affected by ecx) 1359 flags: an OR of zero or more of the following: 1360 KVM_CPUID_FLAG_SIGNIFCANT_INDEX: 1361 if the index field is valid 1362 KVM_CPUID_FLAG_STATEFUL_FUNC: 1363 if cpuid for this function returns different values for successive 1364 invocations; there will be several entries with the same function, 1365 all with this flag set 1366 KVM_CPUID_FLAG_STATE_READ_NEXT: 1367 for KVM_CPUID_FLAG_STATEFUL_FUNC entries, set if this entry is 1368 the first entry to be read by a cpu 1369 eax, ebx, ecx, edx: the values returned by the cpuid instruction for 1370 this function/index combination 1371 1372The TSC deadline timer feature (CPUID leaf 1, ecx[24]) is always returned 1373as false, since the feature depends on KVM_CREATE_IRQCHIP for local APIC 1374support. Instead it is reported via 1375 1376 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER) 1377 1378if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the 1379feature in userspace, then you can enable the feature for KVM_SET_CPUID2. 1380 1381 13824.47 KVM_PPC_GET_PVINFO 1383 1384Capability: KVM_CAP_PPC_GET_PVINFO 1385Architectures: ppc 1386Type: vm ioctl 1387Parameters: struct kvm_ppc_pvinfo (out) 1388Returns: 0 on success, !0 on error 1389 1390struct kvm_ppc_pvinfo { 1391 __u32 flags; 1392 __u32 hcall[4]; 1393 __u8 pad[108]; 1394}; 1395 1396This ioctl fetches PV specific information that need to be passed to the guest 1397using the device tree or other means from vm context. 1398 1399The hcall array defines 4 instructions that make up a hypercall. 1400 1401If any additional field gets added to this structure later on, a bit for that 1402additional piece of information will be set in the flags bitmap. 1403 1404The flags bitmap is defined as: 1405 1406 /* the host supports the ePAPR idle hcall 1407 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1<<0) 1408 14094.52 KVM_SET_GSI_ROUTING 1410 1411Capability: KVM_CAP_IRQ_ROUTING 1412Architectures: x86 s390 arm arm64 1413Type: vm ioctl 1414Parameters: struct kvm_irq_routing (in) 1415Returns: 0 on success, -1 on error 1416 1417Sets the GSI routing table entries, overwriting any previously set entries. 1418 1419On arm/arm64, GSI routing has the following limitation: 1420- GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD. 1421 1422struct kvm_irq_routing { 1423 __u32 nr; 1424 __u32 flags; 1425 struct kvm_irq_routing_entry entries[0]; 1426}; 1427 1428No flags are specified so far, the corresponding field must be set to zero. 1429 1430struct kvm_irq_routing_entry { 1431 __u32 gsi; 1432 __u32 type; 1433 __u32 flags; 1434 __u32 pad; 1435 union { 1436 struct kvm_irq_routing_irqchip irqchip; 1437 struct kvm_irq_routing_msi msi; 1438 struct kvm_irq_routing_s390_adapter adapter; 1439 struct kvm_irq_routing_hv_sint hv_sint; 1440 __u32 pad[8]; 1441 } u; 1442}; 1443 1444/* gsi routing entry types */ 1445#define KVM_IRQ_ROUTING_IRQCHIP 1 1446#define KVM_IRQ_ROUTING_MSI 2 1447#define KVM_IRQ_ROUTING_S390_ADAPTER 3 1448#define KVM_IRQ_ROUTING_HV_SINT 4 1449 1450flags: 1451- KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry 1452 type, specifies that the devid field contains a valid value. The per-VM 1453 KVM_CAP_MSI_DEVID capability advertises the requirement to provide 1454 the device ID. If this capability is not available, userspace should 1455 never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail. 1456- zero otherwise 1457 1458struct kvm_irq_routing_irqchip { 1459 __u32 irqchip; 1460 __u32 pin; 1461}; 1462 1463struct kvm_irq_routing_msi { 1464 __u32 address_lo; 1465 __u32 address_hi; 1466 __u32 data; 1467 union { 1468 __u32 pad; 1469 __u32 devid; 1470 }; 1471}; 1472 1473If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier 1474for the device that wrote the MSI message. For PCI, this is usually a 1475BFD identifier in the lower 16 bits. 1476 1477On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS 1478feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled, 1479address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of 1480address_hi must be zero. 1481 1482struct kvm_irq_routing_s390_adapter { 1483 __u64 ind_addr; 1484 __u64 summary_addr; 1485 __u64 ind_offset; 1486 __u32 summary_offset; 1487 __u32 adapter_id; 1488}; 1489 1490struct kvm_irq_routing_hv_sint { 1491 __u32 vcpu; 1492 __u32 sint; 1493}; 1494 1495 14964.55 KVM_SET_TSC_KHZ 1497 1498Capability: KVM_CAP_TSC_CONTROL 1499Architectures: x86 1500Type: vcpu ioctl 1501Parameters: virtual tsc_khz 1502Returns: 0 on success, -1 on error 1503 1504Specifies the tsc frequency for the virtual machine. The unit of the 1505frequency is KHz. 1506 1507 15084.56 KVM_GET_TSC_KHZ 1509 1510Capability: KVM_CAP_GET_TSC_KHZ 1511Architectures: x86 1512Type: vcpu ioctl 1513Parameters: none 1514Returns: virtual tsc-khz on success, negative value on error 1515 1516Returns the tsc frequency of the guest. The unit of the return value is 1517KHz. If the host has unstable tsc this ioctl returns -EIO instead as an 1518error. 1519 1520 15214.57 KVM_GET_LAPIC 1522 1523Capability: KVM_CAP_IRQCHIP 1524Architectures: x86 1525Type: vcpu ioctl 1526Parameters: struct kvm_lapic_state (out) 1527Returns: 0 on success, -1 on error 1528 1529#define KVM_APIC_REG_SIZE 0x400 1530struct kvm_lapic_state { 1531 char regs[KVM_APIC_REG_SIZE]; 1532}; 1533 1534Reads the Local APIC registers and copies them into the input argument. The 1535data format and layout are the same as documented in the architecture manual. 1536 1537If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is 1538enabled, then the format of APIC_ID register depends on the APIC mode 1539(reported by MSR_IA32_APICBASE) of its VCPU. x2APIC stores APIC ID in 1540the APIC_ID register (bytes 32-35). xAPIC only allows an 8-bit APIC ID 1541which is stored in bits 31-24 of the APIC register, or equivalently in 1542byte 35 of struct kvm_lapic_state's regs field. KVM_GET_LAPIC must then 1543be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR. 1544 1545If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state 1546always uses xAPIC format. 1547 1548 15494.58 KVM_SET_LAPIC 1550 1551Capability: KVM_CAP_IRQCHIP 1552Architectures: x86 1553Type: vcpu ioctl 1554Parameters: struct kvm_lapic_state (in) 1555Returns: 0 on success, -1 on error 1556 1557#define KVM_APIC_REG_SIZE 0x400 1558struct kvm_lapic_state { 1559 char regs[KVM_APIC_REG_SIZE]; 1560}; 1561 1562Copies the input argument into the Local APIC registers. The data format 1563and layout are the same as documented in the architecture manual. 1564 1565The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's 1566regs field) depends on the state of the KVM_CAP_X2APIC_API capability. 1567See the note in KVM_GET_LAPIC. 1568 1569 15704.59 KVM_IOEVENTFD 1571 1572Capability: KVM_CAP_IOEVENTFD 1573Architectures: all 1574Type: vm ioctl 1575Parameters: struct kvm_ioeventfd (in) 1576Returns: 0 on success, !0 on error 1577 1578This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address 1579within the guest. A guest write in the registered address will signal the 1580provided event instead of triggering an exit. 1581 1582struct kvm_ioeventfd { 1583 __u64 datamatch; 1584 __u64 addr; /* legal pio/mmio address */ 1585 __u32 len; /* 0, 1, 2, 4, or 8 bytes */ 1586 __s32 fd; 1587 __u32 flags; 1588 __u8 pad[36]; 1589}; 1590 1591For the special case of virtio-ccw devices on s390, the ioevent is matched 1592to a subchannel/virtqueue tuple instead. 1593 1594The following flags are defined: 1595 1596#define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch) 1597#define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio) 1598#define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign) 1599#define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \ 1600 (1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify) 1601 1602If datamatch flag is set, the event will be signaled only if the written value 1603to the registered address is equal to datamatch in struct kvm_ioeventfd. 1604 1605For virtio-ccw devices, addr contains the subchannel id and datamatch the 1606virtqueue index. 1607 1608With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and 1609the kernel will ignore the length of guest write and may get a faster vmexit. 1610The speedup may only apply to specific architectures, but the ioeventfd will 1611work anyway. 1612 16134.60 KVM_DIRTY_TLB 1614 1615Capability: KVM_CAP_SW_TLB 1616Architectures: ppc 1617Type: vcpu ioctl 1618Parameters: struct kvm_dirty_tlb (in) 1619Returns: 0 on success, -1 on error 1620 1621struct kvm_dirty_tlb { 1622 __u64 bitmap; 1623 __u32 num_dirty; 1624}; 1625 1626This must be called whenever userspace has changed an entry in the shared 1627TLB, prior to calling KVM_RUN on the associated vcpu. 1628 1629The "bitmap" field is the userspace address of an array. This array 1630consists of a number of bits, equal to the total number of TLB entries as 1631determined by the last successful call to KVM_CONFIG_TLB, rounded up to the 1632nearest multiple of 64. 1633 1634Each bit corresponds to one TLB entry, ordered the same as in the shared TLB 1635array. 1636 1637The array is little-endian: the bit 0 is the least significant bit of the 1638first byte, bit 8 is the least significant bit of the second byte, etc. 1639This avoids any complications with differing word sizes. 1640 1641The "num_dirty" field is a performance hint for KVM to determine whether it 1642should skip processing the bitmap and just invalidate everything. It must 1643be set to the number of set bits in the bitmap. 1644 1645 16464.62 KVM_CREATE_SPAPR_TCE 1647 1648Capability: KVM_CAP_SPAPR_TCE 1649Architectures: powerpc 1650Type: vm ioctl 1651Parameters: struct kvm_create_spapr_tce (in) 1652Returns: file descriptor for manipulating the created TCE table 1653 1654This creates a virtual TCE (translation control entry) table, which 1655is an IOMMU for PAPR-style virtual I/O. It is used to translate 1656logical addresses used in virtual I/O into guest physical addresses, 1657and provides a scatter/gather capability for PAPR virtual I/O. 1658 1659/* for KVM_CAP_SPAPR_TCE */ 1660struct kvm_create_spapr_tce { 1661 __u64 liobn; 1662 __u32 window_size; 1663}; 1664 1665The liobn field gives the logical IO bus number for which to create a 1666TCE table. The window_size field specifies the size of the DMA window 1667which this TCE table will translate - the table will contain one 64 1668bit TCE entry for every 4kiB of the DMA window. 1669 1670When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE 1671table has been created using this ioctl(), the kernel will handle it 1672in real mode, updating the TCE table. H_PUT_TCE calls for other 1673liobns will cause a vm exit and must be handled by userspace. 1674 1675The return value is a file descriptor which can be passed to mmap(2) 1676to map the created TCE table into userspace. This lets userspace read 1677the entries written by kernel-handled H_PUT_TCE calls, and also lets 1678userspace update the TCE table directly which is useful in some 1679circumstances. 1680 1681 16824.63 KVM_ALLOCATE_RMA 1683 1684Capability: KVM_CAP_PPC_RMA 1685Architectures: powerpc 1686Type: vm ioctl 1687Parameters: struct kvm_allocate_rma (out) 1688Returns: file descriptor for mapping the allocated RMA 1689 1690This allocates a Real Mode Area (RMA) from the pool allocated at boot 1691time by the kernel. An RMA is a physically-contiguous, aligned region 1692of memory used on older POWER processors to provide the memory which 1693will be accessed by real-mode (MMU off) accesses in a KVM guest. 1694POWER processors support a set of sizes for the RMA that usually 1695includes 64MB, 128MB, 256MB and some larger powers of two. 1696 1697/* for KVM_ALLOCATE_RMA */ 1698struct kvm_allocate_rma { 1699 __u64 rma_size; 1700}; 1701 1702The return value is a file descriptor which can be passed to mmap(2) 1703to map the allocated RMA into userspace. The mapped area can then be 1704passed to the KVM_SET_USER_MEMORY_REGION ioctl to establish it as the 1705RMA for a virtual machine. The size of the RMA in bytes (which is 1706fixed at host kernel boot time) is returned in the rma_size field of 1707the argument structure. 1708 1709The KVM_CAP_PPC_RMA capability is 1 or 2 if the KVM_ALLOCATE_RMA ioctl 1710is supported; 2 if the processor requires all virtual machines to have 1711an RMA, or 1 if the processor can use an RMA but doesn't require it, 1712because it supports the Virtual RMA (VRMA) facility. 1713 1714 17154.64 KVM_NMI 1716 1717Capability: KVM_CAP_USER_NMI 1718Architectures: x86 1719Type: vcpu ioctl 1720Parameters: none 1721Returns: 0 on success, -1 on error 1722 1723Queues an NMI on the thread's vcpu. Note this is well defined only 1724when KVM_CREATE_IRQCHIP has not been called, since this is an interface 1725between the virtual cpu core and virtual local APIC. After KVM_CREATE_IRQCHIP 1726has been called, this interface is completely emulated within the kernel. 1727 1728To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the 1729following algorithm: 1730 1731 - pause the vcpu 1732 - read the local APIC's state (KVM_GET_LAPIC) 1733 - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1) 1734 - if so, issue KVM_NMI 1735 - resume the vcpu 1736 1737Some guests configure the LINT1 NMI input to cause a panic, aiding in 1738debugging. 1739 1740 17414.65 KVM_S390_UCAS_MAP 1742 1743Capability: KVM_CAP_S390_UCONTROL 1744Architectures: s390 1745Type: vcpu ioctl 1746Parameters: struct kvm_s390_ucas_mapping (in) 1747Returns: 0 in case of success 1748 1749The parameter is defined like this: 1750 struct kvm_s390_ucas_mapping { 1751 __u64 user_addr; 1752 __u64 vcpu_addr; 1753 __u64 length; 1754 }; 1755 1756This ioctl maps the memory at "user_addr" with the length "length" to 1757the vcpu's address space starting at "vcpu_addr". All parameters need to 1758be aligned by 1 megabyte. 1759 1760 17614.66 KVM_S390_UCAS_UNMAP 1762 1763Capability: KVM_CAP_S390_UCONTROL 1764Architectures: s390 1765Type: vcpu ioctl 1766Parameters: struct kvm_s390_ucas_mapping (in) 1767Returns: 0 in case of success 1768 1769The parameter is defined like this: 1770 struct kvm_s390_ucas_mapping { 1771 __u64 user_addr; 1772 __u64 vcpu_addr; 1773 __u64 length; 1774 }; 1775 1776This ioctl unmaps the memory in the vcpu's address space starting at 1777"vcpu_addr" with the length "length". The field "user_addr" is ignored. 1778All parameters need to be aligned by 1 megabyte. 1779 1780 17814.67 KVM_S390_VCPU_FAULT 1782 1783Capability: KVM_CAP_S390_UCONTROL 1784Architectures: s390 1785Type: vcpu ioctl 1786Parameters: vcpu absolute address (in) 1787Returns: 0 in case of success 1788 1789This call creates a page table entry on the virtual cpu's address space 1790(for user controlled virtual machines) or the virtual machine's address 1791space (for regular virtual machines). This only works for minor faults, 1792thus it's recommended to access subject memory page via the user page 1793table upfront. This is useful to handle validity intercepts for user 1794controlled virtual machines to fault in the virtual cpu's lowcore pages 1795prior to calling the KVM_RUN ioctl. 1796 1797 17984.68 KVM_SET_ONE_REG 1799 1800Capability: KVM_CAP_ONE_REG 1801Architectures: all 1802Type: vcpu ioctl 1803Parameters: struct kvm_one_reg (in) 1804Returns: 0 on success, negative value on failure 1805 1806struct kvm_one_reg { 1807 __u64 id; 1808 __u64 addr; 1809}; 1810 1811Using this ioctl, a single vcpu register can be set to a specific value 1812defined by user space with the passed in struct kvm_one_reg, where id 1813refers to the register identifier as described below and addr is a pointer 1814to a variable with the respective size. There can be architecture agnostic 1815and architecture specific registers. Each have their own range of operation 1816and their own constants and width. To keep track of the implemented 1817registers, find a list below: 1818 1819 Arch | Register | Width (bits) 1820 | | 1821 PPC | KVM_REG_PPC_HIOR | 64 1822 PPC | KVM_REG_PPC_IAC1 | 64 1823 PPC | KVM_REG_PPC_IAC2 | 64 1824 PPC | KVM_REG_PPC_IAC3 | 64 1825 PPC | KVM_REG_PPC_IAC4 | 64 1826 PPC | KVM_REG_PPC_DAC1 | 64 1827 PPC | KVM_REG_PPC_DAC2 | 64 1828 PPC | KVM_REG_PPC_DABR | 64 1829 PPC | KVM_REG_PPC_DSCR | 64 1830 PPC | KVM_REG_PPC_PURR | 64 1831 PPC | KVM_REG_PPC_SPURR | 64 1832 PPC | KVM_REG_PPC_DAR | 64 1833 PPC | KVM_REG_PPC_DSISR | 32 1834 PPC | KVM_REG_PPC_AMR | 64 1835 PPC | KVM_REG_PPC_UAMOR | 64 1836 PPC | KVM_REG_PPC_MMCR0 | 64 1837 PPC | KVM_REG_PPC_MMCR1 | 64 1838 PPC | KVM_REG_PPC_MMCRA | 64 1839 PPC | KVM_REG_PPC_MMCR2 | 64 1840 PPC | KVM_REG_PPC_MMCRS | 64 1841 PPC | KVM_REG_PPC_SIAR | 64 1842 PPC | KVM_REG_PPC_SDAR | 64 1843 PPC | KVM_REG_PPC_SIER | 64 1844 PPC | KVM_REG_PPC_PMC1 | 32 1845 PPC | KVM_REG_PPC_PMC2 | 32 1846 PPC | KVM_REG_PPC_PMC3 | 32 1847 PPC | KVM_REG_PPC_PMC4 | 32 1848 PPC | KVM_REG_PPC_PMC5 | 32 1849 PPC | KVM_REG_PPC_PMC6 | 32 1850 PPC | KVM_REG_PPC_PMC7 | 32 1851 PPC | KVM_REG_PPC_PMC8 | 32 1852 PPC | KVM_REG_PPC_FPR0 | 64 1853 ... 1854 PPC | KVM_REG_PPC_FPR31 | 64 1855 PPC | KVM_REG_PPC_VR0 | 128 1856 ... 1857 PPC | KVM_REG_PPC_VR31 | 128 1858 PPC | KVM_REG_PPC_VSR0 | 128 1859 ... 1860 PPC | KVM_REG_PPC_VSR31 | 128 1861 PPC | KVM_REG_PPC_FPSCR | 64 1862 PPC | KVM_REG_PPC_VSCR | 32 1863 PPC | KVM_REG_PPC_VPA_ADDR | 64 1864 PPC | KVM_REG_PPC_VPA_SLB | 128 1865 PPC | KVM_REG_PPC_VPA_DTL | 128 1866 PPC | KVM_REG_PPC_EPCR | 32 1867 PPC | KVM_REG_PPC_EPR | 32 1868 PPC | KVM_REG_PPC_TCR | 32 1869 PPC | KVM_REG_PPC_TSR | 32 1870 PPC | KVM_REG_PPC_OR_TSR | 32 1871 PPC | KVM_REG_PPC_CLEAR_TSR | 32 1872 PPC | KVM_REG_PPC_MAS0 | 32 1873 PPC | KVM_REG_PPC_MAS1 | 32 1874 PPC | KVM_REG_PPC_MAS2 | 64 1875 PPC | KVM_REG_PPC_MAS7_3 | 64 1876 PPC | KVM_REG_PPC_MAS4 | 32 1877 PPC | KVM_REG_PPC_MAS6 | 32 1878 PPC | KVM_REG_PPC_MMUCFG | 32 1879 PPC | KVM_REG_PPC_TLB0CFG | 32 1880 PPC | KVM_REG_PPC_TLB1CFG | 32 1881 PPC | KVM_REG_PPC_TLB2CFG | 32 1882 PPC | KVM_REG_PPC_TLB3CFG | 32 1883 PPC | KVM_REG_PPC_TLB0PS | 32 1884 PPC | KVM_REG_PPC_TLB1PS | 32 1885 PPC | KVM_REG_PPC_TLB2PS | 32 1886 PPC | KVM_REG_PPC_TLB3PS | 32 1887 PPC | KVM_REG_PPC_EPTCFG | 32 1888 PPC | KVM_REG_PPC_ICP_STATE | 64 1889 PPC | KVM_REG_PPC_TB_OFFSET | 64 1890 PPC | KVM_REG_PPC_SPMC1 | 32 1891 PPC | KVM_REG_PPC_SPMC2 | 32 1892 PPC | KVM_REG_PPC_IAMR | 64 1893 PPC | KVM_REG_PPC_TFHAR | 64 1894 PPC | KVM_REG_PPC_TFIAR | 64 1895 PPC | KVM_REG_PPC_TEXASR | 64 1896 PPC | KVM_REG_PPC_FSCR | 64 1897 PPC | KVM_REG_PPC_PSPB | 32 1898 PPC | KVM_REG_PPC_EBBHR | 64 1899 PPC | KVM_REG_PPC_EBBRR | 64 1900 PPC | KVM_REG_PPC_BESCR | 64 1901 PPC | KVM_REG_PPC_TAR | 64 1902 PPC | KVM_REG_PPC_DPDES | 64 1903 PPC | KVM_REG_PPC_DAWR | 64 1904 PPC | KVM_REG_PPC_DAWRX | 64 1905 PPC | KVM_REG_PPC_CIABR | 64 1906 PPC | KVM_REG_PPC_IC | 64 1907 PPC | KVM_REG_PPC_VTB | 64 1908 PPC | KVM_REG_PPC_CSIGR | 64 1909 PPC | KVM_REG_PPC_TACR | 64 1910 PPC | KVM_REG_PPC_TCSCR | 64 1911 PPC | KVM_REG_PPC_PID | 64 1912 PPC | KVM_REG_PPC_ACOP | 64 1913 PPC | KVM_REG_PPC_VRSAVE | 32 1914 PPC | KVM_REG_PPC_LPCR | 32 1915 PPC | KVM_REG_PPC_LPCR_64 | 64 1916 PPC | KVM_REG_PPC_PPR | 64 1917 PPC | KVM_REG_PPC_ARCH_COMPAT | 32 1918 PPC | KVM_REG_PPC_DABRX | 32 1919 PPC | KVM_REG_PPC_WORT | 64 1920 PPC | KVM_REG_PPC_SPRG9 | 64 1921 PPC | KVM_REG_PPC_DBSR | 32 1922 PPC | KVM_REG_PPC_TIDR | 64 1923 PPC | KVM_REG_PPC_PSSCR | 64 1924 PPC | KVM_REG_PPC_DEC_EXPIRY | 64 1925 PPC | KVM_REG_PPC_TM_GPR0 | 64 1926 ... 1927 PPC | KVM_REG_PPC_TM_GPR31 | 64 1928 PPC | KVM_REG_PPC_TM_VSR0 | 128 1929 ... 1930 PPC | KVM_REG_PPC_TM_VSR63 | 128 1931 PPC | KVM_REG_PPC_TM_CR | 64 1932 PPC | KVM_REG_PPC_TM_LR | 64 1933 PPC | KVM_REG_PPC_TM_CTR | 64 1934 PPC | KVM_REG_PPC_TM_FPSCR | 64 1935 PPC | KVM_REG_PPC_TM_AMR | 64 1936 PPC | KVM_REG_PPC_TM_PPR | 64 1937 PPC | KVM_REG_PPC_TM_VRSAVE | 64 1938 PPC | KVM_REG_PPC_TM_VSCR | 32 1939 PPC | KVM_REG_PPC_TM_DSCR | 64 1940 PPC | KVM_REG_PPC_TM_TAR | 64 1941 PPC | KVM_REG_PPC_TM_XER | 64 1942 | | 1943 MIPS | KVM_REG_MIPS_R0 | 64 1944 ... 1945 MIPS | KVM_REG_MIPS_R31 | 64 1946 MIPS | KVM_REG_MIPS_HI | 64 1947 MIPS | KVM_REG_MIPS_LO | 64 1948 MIPS | KVM_REG_MIPS_PC | 64 1949 MIPS | KVM_REG_MIPS_CP0_INDEX | 32 1950 MIPS | KVM_REG_MIPS_CP0_ENTRYLO0 | 64 1951 MIPS | KVM_REG_MIPS_CP0_ENTRYLO1 | 64 1952 MIPS | KVM_REG_MIPS_CP0_CONTEXT | 64 1953 MIPS | KVM_REG_MIPS_CP0_CONTEXTCONFIG| 32 1954 MIPS | KVM_REG_MIPS_CP0_USERLOCAL | 64 1955 MIPS | KVM_REG_MIPS_CP0_XCONTEXTCONFIG| 64 1956 MIPS | KVM_REG_MIPS_CP0_PAGEMASK | 32 1957 MIPS | KVM_REG_MIPS_CP0_PAGEGRAIN | 32 1958 MIPS | KVM_REG_MIPS_CP0_SEGCTL0 | 64 1959 MIPS | KVM_REG_MIPS_CP0_SEGCTL1 | 64 1960 MIPS | KVM_REG_MIPS_CP0_SEGCTL2 | 64 1961 MIPS | KVM_REG_MIPS_CP0_PWBASE | 64 1962 MIPS | KVM_REG_MIPS_CP0_PWFIELD | 64 1963 MIPS | KVM_REG_MIPS_CP0_PWSIZE | 64 1964 MIPS | KVM_REG_MIPS_CP0_WIRED | 32 1965 MIPS | KVM_REG_MIPS_CP0_PWCTL | 32 1966 MIPS | KVM_REG_MIPS_CP0_HWRENA | 32 1967 MIPS | KVM_REG_MIPS_CP0_BADVADDR | 64 1968 MIPS | KVM_REG_MIPS_CP0_BADINSTR | 32 1969 MIPS | KVM_REG_MIPS_CP0_BADINSTRP | 32 1970 MIPS | KVM_REG_MIPS_CP0_COUNT | 32 1971 MIPS | KVM_REG_MIPS_CP0_ENTRYHI | 64 1972 MIPS | KVM_REG_MIPS_CP0_COMPARE | 32 1973 MIPS | KVM_REG_MIPS_CP0_STATUS | 32 1974 MIPS | KVM_REG_MIPS_CP0_INTCTL | 32 1975 MIPS | KVM_REG_MIPS_CP0_CAUSE | 32 1976 MIPS | KVM_REG_MIPS_CP0_EPC | 64 1977 MIPS | KVM_REG_MIPS_CP0_PRID | 32 1978 MIPS | KVM_REG_MIPS_CP0_EBASE | 64 1979 MIPS | KVM_REG_MIPS_CP0_CONFIG | 32 1980 MIPS | KVM_REG_MIPS_CP0_CONFIG1 | 32 1981 MIPS | KVM_REG_MIPS_CP0_CONFIG2 | 32 1982 MIPS | KVM_REG_MIPS_CP0_CONFIG3 | 32 1983 MIPS | KVM_REG_MIPS_CP0_CONFIG4 | 32 1984 MIPS | KVM_REG_MIPS_CP0_CONFIG5 | 32 1985 MIPS | KVM_REG_MIPS_CP0_CONFIG7 | 32 1986 MIPS | KVM_REG_MIPS_CP0_XCONTEXT | 64 1987 MIPS | KVM_REG_MIPS_CP0_ERROREPC | 64 1988 MIPS | KVM_REG_MIPS_CP0_KSCRATCH1 | 64 1989 MIPS | KVM_REG_MIPS_CP0_KSCRATCH2 | 64 1990 MIPS | KVM_REG_MIPS_CP0_KSCRATCH3 | 64 1991 MIPS | KVM_REG_MIPS_CP0_KSCRATCH4 | 64 1992 MIPS | KVM_REG_MIPS_CP0_KSCRATCH5 | 64 1993 MIPS | KVM_REG_MIPS_CP0_KSCRATCH6 | 64 1994 MIPS | KVM_REG_MIPS_CP0_MAAR(0..63) | 64 1995 MIPS | KVM_REG_MIPS_COUNT_CTL | 64 1996 MIPS | KVM_REG_MIPS_COUNT_RESUME | 64 1997 MIPS | KVM_REG_MIPS_COUNT_HZ | 64 1998 MIPS | KVM_REG_MIPS_FPR_32(0..31) | 32 1999 MIPS | KVM_REG_MIPS_FPR_64(0..31) | 64 2000 MIPS | KVM_REG_MIPS_VEC_128(0..31) | 128 2001 MIPS | KVM_REG_MIPS_FCR_IR | 32 2002 MIPS | KVM_REG_MIPS_FCR_CSR | 32 2003 MIPS | KVM_REG_MIPS_MSA_IR | 32 2004 MIPS | KVM_REG_MIPS_MSA_CSR | 32 2005 2006ARM registers are mapped using the lower 32 bits. The upper 16 of that 2007is the register group type, or coprocessor number: 2008 2009ARM core registers have the following id bit patterns: 2010 0x4020 0000 0010 <index into the kvm_regs struct:16> 2011 2012ARM 32-bit CP15 registers have the following id bit patterns: 2013 0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3> 2014 2015ARM 64-bit CP15 registers have the following id bit patterns: 2016 0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3> 2017 2018ARM CCSIDR registers are demultiplexed by CSSELR value: 2019 0x4020 0000 0011 00 <csselr:8> 2020 2021ARM 32-bit VFP control registers have the following id bit patterns: 2022 0x4020 0000 0012 1 <regno:12> 2023 2024ARM 64-bit FP registers have the following id bit patterns: 2025 0x4030 0000 0012 0 <regno:12> 2026 2027ARM firmware pseudo-registers have the following bit pattern: 2028 0x4030 0000 0014 <regno:16> 2029 2030 2031arm64 registers are mapped using the lower 32 bits. The upper 16 of 2032that is the register group type, or coprocessor number: 2033 2034arm64 core/FP-SIMD registers have the following id bit patterns. Note 2035that the size of the access is variable, as the kvm_regs structure 2036contains elements ranging from 32 to 128 bits. The index is a 32bit 2037value in the kvm_regs structure seen as a 32bit array. 2038 0x60x0 0000 0010 <index into the kvm_regs struct:16> 2039 2040arm64 CCSIDR registers are demultiplexed by CSSELR value: 2041 0x6020 0000 0011 00 <csselr:8> 2042 2043arm64 system registers have the following id bit patterns: 2044 0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3> 2045 2046arm64 firmware pseudo-registers have the following bit pattern: 2047 0x6030 0000 0014 <regno:16> 2048 2049 2050MIPS registers are mapped using the lower 32 bits. The upper 16 of that is 2051the register group type: 2052 2053MIPS core registers (see above) have the following id bit patterns: 2054 0x7030 0000 0000 <reg:16> 2055 2056MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit 2057patterns depending on whether they're 32-bit or 64-bit registers: 2058 0x7020 0000 0001 00 <reg:5> <sel:3> (32-bit) 2059 0x7030 0000 0001 00 <reg:5> <sel:3> (64-bit) 2060 2061Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_MIPS_CP0_ENTRYLO1 are the MIPS64 2062versions of the EntryLo registers regardless of the word size of the host 2063hardware, host kernel, guest, and whether XPA is present in the guest, i.e. 2064with the RI and XI bits (if they exist) in bits 63 and 62 respectively, and 2065the PFNX field starting at bit 30. 2066 2067MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) above) have the following id bit 2068patterns: 2069 0x7030 0000 0001 01 <reg:8> 2070 2071MIPS KVM control registers (see above) have the following id bit patterns: 2072 0x7030 0000 0002 <reg:16> 2073 2074MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following 2075id bit patterns depending on the size of the register being accessed. They are 2076always accessed according to the current guest FPU mode (Status.FR and 2077Config5.FRE), i.e. as the guest would see them, and they become unpredictable 2078if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector 2079registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they 2080overlap the FPU registers: 2081 0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers) 2082 0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers) 2083 0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers) 2084 2085MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the 2086following id bit patterns: 2087 0x7020 0000 0003 01 <0:3> <reg:5> 2088 2089MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the 2090following id bit patterns: 2091 0x7020 0000 0003 02 <0:3> <reg:5> 2092 2093 20944.69 KVM_GET_ONE_REG 2095 2096Capability: KVM_CAP_ONE_REG 2097Architectures: all 2098Type: vcpu ioctl 2099Parameters: struct kvm_one_reg (in and out) 2100Returns: 0 on success, negative value on failure 2101 2102This ioctl allows to receive the value of a single register implemented 2103in a vcpu. The register to read is indicated by the "id" field of the 2104kvm_one_reg struct passed in. On success, the register value can be found 2105at the memory location pointed to by "addr". 2106 2107The list of registers accessible using this interface is identical to the 2108list in 4.68. 2109 2110 21114.70 KVM_KVMCLOCK_CTRL 2112 2113Capability: KVM_CAP_KVMCLOCK_CTRL 2114Architectures: Any that implement pvclocks (currently x86 only) 2115Type: vcpu ioctl 2116Parameters: None 2117Returns: 0 on success, -1 on error 2118 2119This signals to the host kernel that the specified guest is being paused by 2120userspace. The host will set a flag in the pvclock structure that is checked 2121from the soft lockup watchdog. The flag is part of the pvclock structure that 2122is shared between guest and host, specifically the second bit of the flags 2123field of the pvclock_vcpu_time_info structure. It will be set exclusively by 2124the host and read/cleared exclusively by the guest. The guest operation of 2125checking and clearing the flag must an atomic operation so 2126load-link/store-conditional, or equivalent must be used. There are two cases 2127where the guest will clear the flag: when the soft lockup watchdog timer resets 2128itself or when a soft lockup is detected. This ioctl can be called any time 2129after pausing the vcpu, but before it is resumed. 2130 2131 21324.71 KVM_SIGNAL_MSI 2133 2134Capability: KVM_CAP_SIGNAL_MSI 2135Architectures: x86 arm arm64 2136Type: vm ioctl 2137Parameters: struct kvm_msi (in) 2138Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error 2139 2140Directly inject a MSI message. Only valid with in-kernel irqchip that handles 2141MSI messages. 2142 2143struct kvm_msi { 2144 __u32 address_lo; 2145 __u32 address_hi; 2146 __u32 data; 2147 __u32 flags; 2148 __u32 devid; 2149 __u8 pad[12]; 2150}; 2151 2152flags: KVM_MSI_VALID_DEVID: devid contains a valid value. The per-VM 2153 KVM_CAP_MSI_DEVID capability advertises the requirement to provide 2154 the device ID. If this capability is not available, userspace 2155 should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail. 2156 2157If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier 2158for the device that wrote the MSI message. For PCI, this is usually a 2159BFD identifier in the lower 16 bits. 2160 2161On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS 2162feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled, 2163address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of 2164address_hi must be zero. 2165 2166 21674.71 KVM_CREATE_PIT2 2168 2169Capability: KVM_CAP_PIT2 2170Architectures: x86 2171Type: vm ioctl 2172Parameters: struct kvm_pit_config (in) 2173Returns: 0 on success, -1 on error 2174 2175Creates an in-kernel device model for the i8254 PIT. This call is only valid 2176after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following 2177parameters have to be passed: 2178 2179struct kvm_pit_config { 2180 __u32 flags; 2181 __u32 pad[15]; 2182}; 2183 2184Valid flags are: 2185 2186#define KVM_PIT_SPEAKER_DUMMY 1 /* emulate speaker port stub */ 2187 2188PIT timer interrupts may use a per-VM kernel thread for injection. If it 2189exists, this thread will have a name of the following pattern: 2190 2191kvm-pit/<owner-process-pid> 2192 2193When running a guest with elevated priorities, the scheduling parameters of 2194this thread may have to be adjusted accordingly. 2195 2196This IOCTL replaces the obsolete KVM_CREATE_PIT. 2197 2198 21994.72 KVM_GET_PIT2 2200 2201Capability: KVM_CAP_PIT_STATE2 2202Architectures: x86 2203Type: vm ioctl 2204Parameters: struct kvm_pit_state2 (out) 2205Returns: 0 on success, -1 on error 2206 2207Retrieves the state of the in-kernel PIT model. Only valid after 2208KVM_CREATE_PIT2. The state is returned in the following structure: 2209 2210struct kvm_pit_state2 { 2211 struct kvm_pit_channel_state channels[3]; 2212 __u32 flags; 2213 __u32 reserved[9]; 2214}; 2215 2216Valid flags are: 2217 2218/* disable PIT in HPET legacy mode */ 2219#define KVM_PIT_FLAGS_HPET_LEGACY 0x00000001 2220 2221This IOCTL replaces the obsolete KVM_GET_PIT. 2222 2223 22244.73 KVM_SET_PIT2 2225 2226Capability: KVM_CAP_PIT_STATE2 2227Architectures: x86 2228Type: vm ioctl 2229Parameters: struct kvm_pit_state2 (in) 2230Returns: 0 on success, -1 on error 2231 2232Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2. 2233See KVM_GET_PIT2 for details on struct kvm_pit_state2. 2234 2235This IOCTL replaces the obsolete KVM_SET_PIT. 2236 2237 22384.74 KVM_PPC_GET_SMMU_INFO 2239 2240Capability: KVM_CAP_PPC_GET_SMMU_INFO 2241Architectures: powerpc 2242Type: vm ioctl 2243Parameters: None 2244Returns: 0 on success, -1 on error 2245 2246This populates and returns a structure describing the features of 2247the "Server" class MMU emulation supported by KVM. 2248This can in turn be used by userspace to generate the appropriate 2249device-tree properties for the guest operating system. 2250 2251The structure contains some global information, followed by an 2252array of supported segment page sizes: 2253 2254 struct kvm_ppc_smmu_info { 2255 __u64 flags; 2256 __u32 slb_size; 2257 __u32 pad; 2258 struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ]; 2259 }; 2260 2261The supported flags are: 2262 2263 - KVM_PPC_PAGE_SIZES_REAL: 2264 When that flag is set, guest page sizes must "fit" the backing 2265 store page sizes. When not set, any page size in the list can 2266 be used regardless of how they are backed by userspace. 2267 2268 - KVM_PPC_1T_SEGMENTS 2269 The emulated MMU supports 1T segments in addition to the 2270 standard 256M ones. 2271 2272The "slb_size" field indicates how many SLB entries are supported 2273 2274The "sps" array contains 8 entries indicating the supported base 2275page sizes for a segment in increasing order. Each entry is defined 2276as follow: 2277 2278 struct kvm_ppc_one_seg_page_size { 2279 __u32 page_shift; /* Base page shift of segment (or 0) */ 2280 __u32 slb_enc; /* SLB encoding for BookS */ 2281 struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ]; 2282 }; 2283 2284An entry with a "page_shift" of 0 is unused. Because the array is 2285organized in increasing order, a lookup can stop when encoutering 2286such an entry. 2287 2288The "slb_enc" field provides the encoding to use in the SLB for the 2289page size. The bits are in positions such as the value can directly 2290be OR'ed into the "vsid" argument of the slbmte instruction. 2291 2292The "enc" array is a list which for each of those segment base page 2293size provides the list of supported actual page sizes (which can be 2294only larger or equal to the base page size), along with the 2295corresponding encoding in the hash PTE. Similarly, the array is 22968 entries sorted by increasing sizes and an entry with a "0" shift 2297is an empty entry and a terminator: 2298 2299 struct kvm_ppc_one_page_size { 2300 __u32 page_shift; /* Page shift (or 0) */ 2301 __u32 pte_enc; /* Encoding in the HPTE (>>12) */ 2302 }; 2303 2304The "pte_enc" field provides a value that can OR'ed into the hash 2305PTE's RPN field (ie, it needs to be shifted left by 12 to OR it 2306into the hash PTE second double word). 2307 23084.75 KVM_IRQFD 2309 2310Capability: KVM_CAP_IRQFD 2311Architectures: x86 s390 arm arm64 2312Type: vm ioctl 2313Parameters: struct kvm_irqfd (in) 2314Returns: 0 on success, -1 on error 2315 2316Allows setting an eventfd to directly trigger a guest interrupt. 2317kvm_irqfd.fd specifies the file descriptor to use as the eventfd and 2318kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When 2319an event is triggered on the eventfd, an interrupt is injected into 2320the guest using the specified gsi pin. The irqfd is removed using 2321the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd 2322and kvm_irqfd.gsi. 2323 2324With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify 2325mechanism allowing emulation of level-triggered, irqfd-based 2326interrupts. When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an 2327additional eventfd in the kvm_irqfd.resamplefd field. When operating 2328in resample mode, posting of an interrupt through kvm_irq.fd asserts 2329the specified gsi in the irqchip. When the irqchip is resampled, such 2330as from an EOI, the gsi is de-asserted and the user is notified via 2331kvm_irqfd.resamplefd. It is the user's responsibility to re-queue 2332the interrupt if the device making use of it still requires service. 2333Note that closing the resamplefd is not sufficient to disable the 2334irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment 2335and need not be specified with KVM_IRQFD_FLAG_DEASSIGN. 2336 2337On arm/arm64, gsi routing being supported, the following can happen: 2338- in case no routing entry is associated to this gsi, injection fails 2339- in case the gsi is associated to an irqchip routing entry, 2340 irqchip.pin + 32 corresponds to the injected SPI ID. 2341- in case the gsi is associated to an MSI routing entry, the MSI 2342 message and device ID are translated into an LPI (support restricted 2343 to GICv3 ITS in-kernel emulation). 2344 23454.76 KVM_PPC_ALLOCATE_HTAB 2346 2347Capability: KVM_CAP_PPC_ALLOC_HTAB 2348Architectures: powerpc 2349Type: vm ioctl 2350Parameters: Pointer to u32 containing hash table order (in/out) 2351Returns: 0 on success, -1 on error 2352 2353This requests the host kernel to allocate an MMU hash table for a 2354guest using the PAPR paravirtualization interface. This only does 2355anything if the kernel is configured to use the Book 3S HV style of 2356virtualization. Otherwise the capability doesn't exist and the ioctl 2357returns an ENOTTY error. The rest of this description assumes Book 3S 2358HV. 2359 2360There must be no vcpus running when this ioctl is called; if there 2361are, it will do nothing and return an EBUSY error. 2362 2363The parameter is a pointer to a 32-bit unsigned integer variable 2364containing the order (log base 2) of the desired size of the hash 2365table, which must be between 18 and 46. On successful return from the 2366ioctl, the value will not be changed by the kernel. 2367 2368If no hash table has been allocated when any vcpu is asked to run 2369(with the KVM_RUN ioctl), the host kernel will allocate a 2370default-sized hash table (16 MB). 2371 2372If this ioctl is called when a hash table has already been allocated, 2373with a different order from the existing hash table, the existing hash 2374table will be freed and a new one allocated. If this is ioctl is 2375called when a hash table has already been allocated of the same order 2376as specified, the kernel will clear out the existing hash table (zero 2377all HPTEs). In either case, if the guest is using the virtualized 2378real-mode area (VRMA) facility, the kernel will re-create the VMRA 2379HPTEs on the next KVM_RUN of any vcpu. 2380 23814.77 KVM_S390_INTERRUPT 2382 2383Capability: basic 2384Architectures: s390 2385Type: vm ioctl, vcpu ioctl 2386Parameters: struct kvm_s390_interrupt (in) 2387Returns: 0 on success, -1 on error 2388 2389Allows to inject an interrupt to the guest. Interrupts can be floating 2390(vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type. 2391 2392Interrupt parameters are passed via kvm_s390_interrupt: 2393 2394struct kvm_s390_interrupt { 2395 __u32 type; 2396 __u32 parm; 2397 __u64 parm64; 2398}; 2399 2400type can be one of the following: 2401 2402KVM_S390_SIGP_STOP (vcpu) - sigp stop; optional flags in parm 2403KVM_S390_PROGRAM_INT (vcpu) - program check; code in parm 2404KVM_S390_SIGP_SET_PREFIX (vcpu) - sigp set prefix; prefix address in parm 2405KVM_S390_RESTART (vcpu) - restart 2406KVM_S390_INT_CLOCK_COMP (vcpu) - clock comparator interrupt 2407KVM_S390_INT_CPU_TIMER (vcpu) - CPU timer interrupt 2408KVM_S390_INT_VIRTIO (vm) - virtio external interrupt; external interrupt 2409 parameters in parm and parm64 2410KVM_S390_INT_SERVICE (vm) - sclp external interrupt; sclp parameter in parm 2411KVM_S390_INT_EMERGENCY (vcpu) - sigp emergency; source cpu in parm 2412KVM_S390_INT_EXTERNAL_CALL (vcpu) - sigp external call; source cpu in parm 2413KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm) - compound value to indicate an 2414 I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel); 2415 I/O interruption parameters in parm (subchannel) and parm64 (intparm, 2416 interruption subclass) 2417KVM_S390_MCHK (vm, vcpu) - machine check interrupt; cr 14 bits in parm, 2418 machine check interrupt code in parm64 (note that 2419 machine checks needing further payload are not 2420 supported by this ioctl) 2421 2422Note that the vcpu ioctl is asynchronous to vcpu execution. 2423 24244.78 KVM_PPC_GET_HTAB_FD 2425 2426Capability: KVM_CAP_PPC_HTAB_FD 2427Architectures: powerpc 2428Type: vm ioctl 2429Parameters: Pointer to struct kvm_get_htab_fd (in) 2430Returns: file descriptor number (>= 0) on success, -1 on error 2431 2432This returns a file descriptor that can be used either to read out the 2433entries in the guest's hashed page table (HPT), or to write entries to 2434initialize the HPT. The returned fd can only be written to if the 2435KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and 2436can only be read if that bit is clear. The argument struct looks like 2437this: 2438 2439/* For KVM_PPC_GET_HTAB_FD */ 2440struct kvm_get_htab_fd { 2441 __u64 flags; 2442 __u64 start_index; 2443 __u64 reserved[2]; 2444}; 2445 2446/* Values for kvm_get_htab_fd.flags */ 2447#define KVM_GET_HTAB_BOLTED_ONLY ((__u64)0x1) 2448#define KVM_GET_HTAB_WRITE ((__u64)0x2) 2449 2450The `start_index' field gives the index in the HPT of the entry at 2451which to start reading. It is ignored when writing. 2452 2453Reads on the fd will initially supply information about all 2454"interesting" HPT entries. Interesting entries are those with the 2455bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise 2456all entries. When the end of the HPT is reached, the read() will 2457return. If read() is called again on the fd, it will start again from 2458the beginning of the HPT, but will only return HPT entries that have 2459changed since they were last read. 2460 2461Data read or written is structured as a header (8 bytes) followed by a 2462series of valid HPT entries (16 bytes) each. The header indicates how 2463many valid HPT entries there are and how many invalid entries follow 2464the valid entries. The invalid entries are not represented explicitly 2465in the stream. The header format is: 2466 2467struct kvm_get_htab_header { 2468 __u32 index; 2469 __u16 n_valid; 2470 __u16 n_invalid; 2471}; 2472 2473Writes to the fd create HPT entries starting at the index given in the 2474header; first `n_valid' valid entries with contents from the data 2475written, then `n_invalid' invalid entries, invalidating any previously 2476valid entries found. 2477 24784.79 KVM_CREATE_DEVICE 2479 2480Capability: KVM_CAP_DEVICE_CTRL 2481Type: vm ioctl 2482Parameters: struct kvm_create_device (in/out) 2483Returns: 0 on success, -1 on error 2484Errors: 2485 ENODEV: The device type is unknown or unsupported 2486 EEXIST: Device already created, and this type of device may not 2487 be instantiated multiple times 2488 2489 Other error conditions may be defined by individual device types or 2490 have their standard meanings. 2491 2492Creates an emulated device in the kernel. The file descriptor returned 2493in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR. 2494 2495If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the 2496device type is supported (not necessarily whether it can be created 2497in the current vm). 2498 2499Individual devices should not define flags. Attributes should be used 2500for specifying any behavior that is not implied by the device type 2501number. 2502 2503struct kvm_create_device { 2504 __u32 type; /* in: KVM_DEV_TYPE_xxx */ 2505 __u32 fd; /* out: device handle */ 2506 __u32 flags; /* in: KVM_CREATE_DEVICE_xxx */ 2507}; 2508 25094.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR 2510 2511Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device, 2512 KVM_CAP_VCPU_ATTRIBUTES for vcpu device 2513Type: device ioctl, vm ioctl, vcpu ioctl 2514Parameters: struct kvm_device_attr 2515Returns: 0 on success, -1 on error 2516Errors: 2517 ENXIO: The group or attribute is unknown/unsupported for this device 2518 or hardware support is missing. 2519 EPERM: The attribute cannot (currently) be accessed this way 2520 (e.g. read-only attribute, or attribute that only makes 2521 sense when the device is in a different state) 2522 2523 Other error conditions may be defined by individual device types. 2524 2525Gets/sets a specified piece of device configuration and/or state. The 2526semantics are device-specific. See individual device documentation in 2527the "devices" directory. As with ONE_REG, the size of the data 2528transferred is defined by the particular attribute. 2529 2530struct kvm_device_attr { 2531 __u32 flags; /* no flags currently defined */ 2532 __u32 group; /* device-defined */ 2533 __u64 attr; /* group-defined */ 2534 __u64 addr; /* userspace address of attr data */ 2535}; 2536 25374.81 KVM_HAS_DEVICE_ATTR 2538 2539Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device, 2540 KVM_CAP_VCPU_ATTRIBUTES for vcpu device 2541Type: device ioctl, vm ioctl, vcpu ioctl 2542Parameters: struct kvm_device_attr 2543Returns: 0 on success, -1 on error 2544Errors: 2545 ENXIO: The group or attribute is unknown/unsupported for this device 2546 or hardware support is missing. 2547 2548Tests whether a device supports a particular attribute. A successful 2549return indicates the attribute is implemented. It does not necessarily 2550indicate that the attribute can be read or written in the device's 2551current state. "addr" is ignored. 2552 25534.82 KVM_ARM_VCPU_INIT 2554 2555Capability: basic 2556Architectures: arm, arm64 2557Type: vcpu ioctl 2558Parameters: struct kvm_vcpu_init (in) 2559Returns: 0 on success; -1 on error 2560Errors: 2561 EINVAL: the target is unknown, or the combination of features is invalid. 2562 ENOENT: a features bit specified is unknown. 2563 2564This tells KVM what type of CPU to present to the guest, and what 2565optional features it should have. This will cause a reset of the cpu 2566registers to their initial values. If this is not called, KVM_RUN will 2567return ENOEXEC for that vcpu. 2568 2569Note that because some registers reflect machine topology, all vcpus 2570should be created before this ioctl is invoked. 2571 2572Userspace can call this function multiple times for a given vcpu, including 2573after the vcpu has been run. This will reset the vcpu to its initial 2574state. All calls to this function after the initial call must use the same 2575target and same set of feature flags, otherwise EINVAL will be returned. 2576 2577Possible features: 2578 - KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state. 2579 Depends on KVM_CAP_ARM_PSCI. If not set, the CPU will be powered on 2580 and execute guest code when KVM_RUN is called. 2581 - KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode. 2582 Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only). 2583 - KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 (or a future revision 2584 backward compatible with v0.2) for the CPU. 2585 Depends on KVM_CAP_ARM_PSCI_0_2. 2586 - KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU. 2587 Depends on KVM_CAP_ARM_PMU_V3. 2588 2589 25904.83 KVM_ARM_PREFERRED_TARGET 2591 2592Capability: basic 2593Architectures: arm, arm64 2594Type: vm ioctl 2595Parameters: struct struct kvm_vcpu_init (out) 2596Returns: 0 on success; -1 on error 2597Errors: 2598 ENODEV: no preferred target available for the host 2599 2600This queries KVM for preferred CPU target type which can be emulated 2601by KVM on underlying host. 2602 2603The ioctl returns struct kvm_vcpu_init instance containing information 2604about preferred CPU target type and recommended features for it. The 2605kvm_vcpu_init->features bitmap returned will have feature bits set if 2606the preferred target recommends setting these features, but this is 2607not mandatory. 2608 2609The information returned by this ioctl can be used to prepare an instance 2610of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in 2611in VCPU matching underlying host. 2612 2613 26144.84 KVM_GET_REG_LIST 2615 2616Capability: basic 2617Architectures: arm, arm64, mips 2618Type: vcpu ioctl 2619Parameters: struct kvm_reg_list (in/out) 2620Returns: 0 on success; -1 on error 2621Errors: 2622 E2BIG: the reg index list is too big to fit in the array specified by 2623 the user (the number required will be written into n). 2624 2625struct kvm_reg_list { 2626 __u64 n; /* number of registers in reg[] */ 2627 __u64 reg[0]; 2628}; 2629 2630This ioctl returns the guest registers that are supported for the 2631KVM_GET_ONE_REG/KVM_SET_ONE_REG calls. 2632 2633 26344.85 KVM_ARM_SET_DEVICE_ADDR (deprecated) 2635 2636Capability: KVM_CAP_ARM_SET_DEVICE_ADDR 2637Architectures: arm, arm64 2638Type: vm ioctl 2639Parameters: struct kvm_arm_device_address (in) 2640Returns: 0 on success, -1 on error 2641Errors: 2642 ENODEV: The device id is unknown 2643 ENXIO: Device not supported on current system 2644 EEXIST: Address already set 2645 E2BIG: Address outside guest physical address space 2646 EBUSY: Address overlaps with other device range 2647 2648struct kvm_arm_device_addr { 2649 __u64 id; 2650 __u64 addr; 2651}; 2652 2653Specify a device address in the guest's physical address space where guests 2654can access emulated or directly exposed devices, which the host kernel needs 2655to know about. The id field is an architecture specific identifier for a 2656specific device. 2657 2658ARM/arm64 divides the id field into two parts, a device id and an 2659address type id specific to the individual device. 2660 2661 bits: | 63 ... 32 | 31 ... 16 | 15 ... 0 | 2662 field: | 0x00000000 | device id | addr type id | 2663 2664ARM/arm64 currently only require this when using the in-kernel GIC 2665support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2 2666as the device id. When setting the base address for the guest's 2667mapping of the VGIC virtual CPU and distributor interface, the ioctl 2668must be called after calling KVM_CREATE_IRQCHIP, but before calling 2669KVM_RUN on any of the VCPUs. Calling this ioctl twice for any of the 2670base addresses will return -EEXIST. 2671 2672Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API 2673should be used instead. 2674 2675 26764.86 KVM_PPC_RTAS_DEFINE_TOKEN 2677 2678Capability: KVM_CAP_PPC_RTAS 2679Architectures: ppc 2680Type: vm ioctl 2681Parameters: struct kvm_rtas_token_args 2682Returns: 0 on success, -1 on error 2683 2684Defines a token value for a RTAS (Run Time Abstraction Services) 2685service in order to allow it to be handled in the kernel. The 2686argument struct gives the name of the service, which must be the name 2687of a service that has a kernel-side implementation. If the token 2688value is non-zero, it will be associated with that service, and 2689subsequent RTAS calls by the guest specifying that token will be 2690handled by the kernel. If the token value is 0, then any token 2691associated with the service will be forgotten, and subsequent RTAS 2692calls by the guest for that service will be passed to userspace to be 2693handled. 2694 26954.87 KVM_SET_GUEST_DEBUG 2696 2697Capability: KVM_CAP_SET_GUEST_DEBUG 2698Architectures: x86, s390, ppc, arm64 2699Type: vcpu ioctl 2700Parameters: struct kvm_guest_debug (in) 2701Returns: 0 on success; -1 on error 2702 2703struct kvm_guest_debug { 2704 __u32 control; 2705 __u32 pad; 2706 struct kvm_guest_debug_arch arch; 2707}; 2708 2709Set up the processor specific debug registers and configure vcpu for 2710handling guest debug events. There are two parts to the structure, the 2711first a control bitfield indicates the type of debug events to handle 2712when running. Common control bits are: 2713 2714 - KVM_GUESTDBG_ENABLE: guest debugging is enabled 2715 - KVM_GUESTDBG_SINGLESTEP: the next run should single-step 2716 2717The top 16 bits of the control field are architecture specific control 2718flags which can include the following: 2719 2720 - KVM_GUESTDBG_USE_SW_BP: using software breakpoints [x86, arm64] 2721 - KVM_GUESTDBG_USE_HW_BP: using hardware breakpoints [x86, s390, arm64] 2722 - KVM_GUESTDBG_INJECT_DB: inject DB type exception [x86] 2723 - KVM_GUESTDBG_INJECT_BP: inject BP type exception [x86] 2724 - KVM_GUESTDBG_EXIT_PENDING: trigger an immediate guest exit [s390] 2725 2726For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints 2727are enabled in memory so we need to ensure breakpoint exceptions are 2728correctly trapped and the KVM run loop exits at the breakpoint and not 2729running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP 2730we need to ensure the guest vCPUs architecture specific registers are 2731updated to the correct (supplied) values. 2732 2733The second part of the structure is architecture specific and 2734typically contains a set of debug registers. 2735 2736For arm64 the number of debug registers is implementation defined and 2737can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and 2738KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number 2739indicating the number of supported registers. 2740 2741When debug events exit the main run loop with the reason 2742KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run 2743structure containing architecture specific debug information. 2744 27454.88 KVM_GET_EMULATED_CPUID 2746 2747Capability: KVM_CAP_EXT_EMUL_CPUID 2748Architectures: x86 2749Type: system ioctl 2750Parameters: struct kvm_cpuid2 (in/out) 2751Returns: 0 on success, -1 on error 2752 2753struct kvm_cpuid2 { 2754 __u32 nent; 2755 __u32 flags; 2756 struct kvm_cpuid_entry2 entries[0]; 2757}; 2758 2759The member 'flags' is used for passing flags from userspace. 2760 2761#define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0) 2762#define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) 2763#define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) 2764 2765struct kvm_cpuid_entry2 { 2766 __u32 function; 2767 __u32 index; 2768 __u32 flags; 2769 __u32 eax; 2770 __u32 ebx; 2771 __u32 ecx; 2772 __u32 edx; 2773 __u32 padding[3]; 2774}; 2775 2776This ioctl returns x86 cpuid features which are emulated by 2777kvm.Userspace can use the information returned by this ioctl to query 2778which features are emulated by kvm instead of being present natively. 2779 2780Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2 2781structure with the 'nent' field indicating the number of entries in 2782the variable-size array 'entries'. If the number of entries is too low 2783to describe the cpu capabilities, an error (E2BIG) is returned. If the 2784number is too high, the 'nent' field is adjusted and an error (ENOMEM) 2785is returned. If the number is just right, the 'nent' field is adjusted 2786to the number of valid entries in the 'entries' array, which is then 2787filled. 2788 2789The entries returned are the set CPUID bits of the respective features 2790which kvm emulates, as returned by the CPUID instruction, with unknown 2791or unsupported feature bits cleared. 2792 2793Features like x2apic, for example, may not be present in the host cpu 2794but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be 2795emulated efficiently and thus not included here. 2796 2797The fields in each entry are defined as follows: 2798 2799 function: the eax value used to obtain the entry 2800 index: the ecx value used to obtain the entry (for entries that are 2801 affected by ecx) 2802 flags: an OR of zero or more of the following: 2803 KVM_CPUID_FLAG_SIGNIFCANT_INDEX: 2804 if the index field is valid 2805 KVM_CPUID_FLAG_STATEFUL_FUNC: 2806 if cpuid for this function returns different values for successive 2807 invocations; there will be several entries with the same function, 2808 all with this flag set 2809 KVM_CPUID_FLAG_STATE_READ_NEXT: 2810 for KVM_CPUID_FLAG_STATEFUL_FUNC entries, set if this entry is 2811 the first entry to be read by a cpu 2812 eax, ebx, ecx, edx: the values returned by the cpuid instruction for 2813 this function/index combination 2814 28154.89 KVM_S390_MEM_OP 2816 2817Capability: KVM_CAP_S390_MEM_OP 2818Architectures: s390 2819Type: vcpu ioctl 2820Parameters: struct kvm_s390_mem_op (in) 2821Returns: = 0 on success, 2822 < 0 on generic error (e.g. -EFAULT or -ENOMEM), 2823 > 0 if an exception occurred while walking the page tables 2824 2825Read or write data from/to the logical (virtual) memory of a VCPU. 2826 2827Parameters are specified via the following structure: 2828 2829struct kvm_s390_mem_op { 2830 __u64 gaddr; /* the guest address */ 2831 __u64 flags; /* flags */ 2832 __u32 size; /* amount of bytes */ 2833 __u32 op; /* type of operation */ 2834 __u64 buf; /* buffer in userspace */ 2835 __u8 ar; /* the access register number */ 2836 __u8 reserved[31]; /* should be set to 0 */ 2837}; 2838 2839The type of operation is specified in the "op" field. It is either 2840KVM_S390_MEMOP_LOGICAL_READ for reading from logical memory space or 2841KVM_S390_MEMOP_LOGICAL_WRITE for writing to logical memory space. The 2842KVM_S390_MEMOP_F_CHECK_ONLY flag can be set in the "flags" field to check 2843whether the corresponding memory access would create an access exception 2844(without touching the data in the memory at the destination). In case an 2845access exception occurred while walking the MMU tables of the guest, the 2846ioctl returns a positive error number to indicate the type of exception. 2847This exception is also raised directly at the corresponding VCPU if the 2848flag KVM_S390_MEMOP_F_INJECT_EXCEPTION is set in the "flags" field. 2849 2850The start address of the memory region has to be specified in the "gaddr" 2851field, and the length of the region in the "size" field. "buf" is the buffer 2852supplied by the userspace application where the read data should be written 2853to for KVM_S390_MEMOP_LOGICAL_READ, or where the data that should be written 2854is stored for a KVM_S390_MEMOP_LOGICAL_WRITE. "buf" is unused and can be NULL 2855when KVM_S390_MEMOP_F_CHECK_ONLY is specified. "ar" designates the access 2856register number to be used. 2857 2858The "reserved" field is meant for future extensions. It is not used by 2859KVM with the currently defined set of flags. 2860 28614.90 KVM_S390_GET_SKEYS 2862 2863Capability: KVM_CAP_S390_SKEYS 2864Architectures: s390 2865Type: vm ioctl 2866Parameters: struct kvm_s390_skeys 2867Returns: 0 on success, KVM_S390_GET_KEYS_NONE if guest is not using storage 2868 keys, negative value on error 2869 2870This ioctl is used to get guest storage key values on the s390 2871architecture. The ioctl takes parameters via the kvm_s390_skeys struct. 2872 2873struct kvm_s390_skeys { 2874 __u64 start_gfn; 2875 __u64 count; 2876 __u64 skeydata_addr; 2877 __u32 flags; 2878 __u32 reserved[9]; 2879}; 2880 2881The start_gfn field is the number of the first guest frame whose storage keys 2882you want to get. 2883 2884The count field is the number of consecutive frames (starting from start_gfn) 2885whose storage keys to get. The count field must be at least 1 and the maximum 2886allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range 2887will cause the ioctl to return -EINVAL. 2888 2889The skeydata_addr field is the address to a buffer large enough to hold count 2890bytes. This buffer will be filled with storage key data by the ioctl. 2891 28924.91 KVM_S390_SET_SKEYS 2893 2894Capability: KVM_CAP_S390_SKEYS 2895Architectures: s390 2896Type: vm ioctl 2897Parameters: struct kvm_s390_skeys 2898Returns: 0 on success, negative value on error 2899 2900This ioctl is used to set guest storage key values on the s390 2901architecture. The ioctl takes parameters via the kvm_s390_skeys struct. 2902See section on KVM_S390_GET_SKEYS for struct definition. 2903 2904The start_gfn field is the number of the first guest frame whose storage keys 2905you want to set. 2906 2907The count field is the number of consecutive frames (starting from start_gfn) 2908whose storage keys to get. The count field must be at least 1 and the maximum 2909allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range 2910will cause the ioctl to return -EINVAL. 2911 2912The skeydata_addr field is the address to a buffer containing count bytes of 2913storage keys. Each byte in the buffer will be set as the storage key for a 2914single frame starting at start_gfn for count frames. 2915 2916Note: If any architecturally invalid key value is found in the given data then 2917the ioctl will return -EINVAL. 2918 29194.92 KVM_S390_IRQ 2920 2921Capability: KVM_CAP_S390_INJECT_IRQ 2922Architectures: s390 2923Type: vcpu ioctl 2924Parameters: struct kvm_s390_irq (in) 2925Returns: 0 on success, -1 on error 2926Errors: 2927 EINVAL: interrupt type is invalid 2928 type is KVM_S390_SIGP_STOP and flag parameter is invalid value 2929 type is KVM_S390_INT_EXTERNAL_CALL and code is bigger 2930 than the maximum of VCPUs 2931 EBUSY: type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped 2932 type is KVM_S390_SIGP_STOP and a stop irq is already pending 2933 type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt 2934 is already pending 2935 2936Allows to inject an interrupt to the guest. 2937 2938Using struct kvm_s390_irq as a parameter allows 2939to inject additional payload which is not 2940possible via KVM_S390_INTERRUPT. 2941 2942Interrupt parameters are passed via kvm_s390_irq: 2943 2944struct kvm_s390_irq { 2945 __u64 type; 2946 union { 2947 struct kvm_s390_io_info io; 2948 struct kvm_s390_ext_info ext; 2949 struct kvm_s390_pgm_info pgm; 2950 struct kvm_s390_emerg_info emerg; 2951 struct kvm_s390_extcall_info extcall; 2952 struct kvm_s390_prefix_info prefix; 2953 struct kvm_s390_stop_info stop; 2954 struct kvm_s390_mchk_info mchk; 2955 char reserved[64]; 2956 } u; 2957}; 2958 2959type can be one of the following: 2960 2961KVM_S390_SIGP_STOP - sigp stop; parameter in .stop 2962KVM_S390_PROGRAM_INT - program check; parameters in .pgm 2963KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix 2964KVM_S390_RESTART - restart; no parameters 2965KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters 2966KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters 2967KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg 2968KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall 2969KVM_S390_MCHK - machine check interrupt; parameters in .mchk 2970 2971 2972Note that the vcpu ioctl is asynchronous to vcpu execution. 2973 29744.94 KVM_S390_GET_IRQ_STATE 2975 2976Capability: KVM_CAP_S390_IRQ_STATE 2977Architectures: s390 2978Type: vcpu ioctl 2979Parameters: struct kvm_s390_irq_state (out) 2980Returns: >= number of bytes copied into buffer, 2981 -EINVAL if buffer size is 0, 2982 -ENOBUFS if buffer size is too small to fit all pending interrupts, 2983 -EFAULT if the buffer address was invalid 2984 2985This ioctl allows userspace to retrieve the complete state of all currently 2986pending interrupts in a single buffer. Use cases include migration 2987and introspection. The parameter structure contains the address of a 2988userspace buffer and its length: 2989 2990struct kvm_s390_irq_state { 2991 __u64 buf; 2992 __u32 flags; /* will stay unused for compatibility reasons */ 2993 __u32 len; 2994 __u32 reserved[4]; /* will stay unused for compatibility reasons */ 2995}; 2996 2997Userspace passes in the above struct and for each pending interrupt a 2998struct kvm_s390_irq is copied to the provided buffer. 2999 3000The structure contains a flags and a reserved field for future extensions. As 3001the kernel never checked for flags == 0 and QEMU never pre-zeroed flags and 3002reserved, these fields can not be used in the future without breaking 3003compatibility. 3004 3005If -ENOBUFS is returned the buffer provided was too small and userspace 3006may retry with a bigger buffer. 3007 30084.95 KVM_S390_SET_IRQ_STATE 3009 3010Capability: KVM_CAP_S390_IRQ_STATE 3011Architectures: s390 3012Type: vcpu ioctl 3013Parameters: struct kvm_s390_irq_state (in) 3014Returns: 0 on success, 3015 -EFAULT if the buffer address was invalid, 3016 -EINVAL for an invalid buffer length (see below), 3017 -EBUSY if there were already interrupts pending, 3018 errors occurring when actually injecting the 3019 interrupt. See KVM_S390_IRQ. 3020 3021This ioctl allows userspace to set the complete state of all cpu-local 3022interrupts currently pending for the vcpu. It is intended for restoring 3023interrupt state after a migration. The input parameter is a userspace buffer 3024containing a struct kvm_s390_irq_state: 3025 3026struct kvm_s390_irq_state { 3027 __u64 buf; 3028 __u32 flags; /* will stay unused for compatibility reasons */ 3029 __u32 len; 3030 __u32 reserved[4]; /* will stay unused for compatibility reasons */ 3031}; 3032 3033The restrictions for flags and reserved apply as well. 3034(see KVM_S390_GET_IRQ_STATE) 3035 3036The userspace memory referenced by buf contains a struct kvm_s390_irq 3037for each interrupt to be injected into the guest. 3038If one of the interrupts could not be injected for some reason the 3039ioctl aborts. 3040 3041len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0 3042and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq), 3043which is the maximum number of possibly pending cpu-local interrupts. 3044 30454.96 KVM_SMI 3046 3047Capability: KVM_CAP_X86_SMM 3048Architectures: x86 3049Type: vcpu ioctl 3050Parameters: none 3051Returns: 0 on success, -1 on error 3052 3053Queues an SMI on the thread's vcpu. 3054 30554.97 KVM_CAP_PPC_MULTITCE 3056 3057Capability: KVM_CAP_PPC_MULTITCE 3058Architectures: ppc 3059Type: vm 3060 3061This capability means the kernel is capable of handling hypercalls 3062H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user 3063space. This significantly accelerates DMA operations for PPC KVM guests. 3064User space should expect that its handlers for these hypercalls 3065are not going to be called if user space previously registered LIOBN 3066in KVM (via KVM_CREATE_SPAPR_TCE or similar calls). 3067 3068In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest, 3069user space might have to advertise it for the guest. For example, 3070IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is 3071present in the "ibm,hypertas-functions" device-tree property. 3072 3073The hypercalls mentioned above may or may not be processed successfully 3074in the kernel based fast path. If they can not be handled by the kernel, 3075they will get passed on to user space. So user space still has to have 3076an implementation for these despite the in kernel acceleration. 3077 3078This capability is always enabled. 3079 30804.98 KVM_CREATE_SPAPR_TCE_64 3081 3082Capability: KVM_CAP_SPAPR_TCE_64 3083Architectures: powerpc 3084Type: vm ioctl 3085Parameters: struct kvm_create_spapr_tce_64 (in) 3086Returns: file descriptor for manipulating the created TCE table 3087 3088This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit 3089windows, described in 4.62 KVM_CREATE_SPAPR_TCE 3090 3091This capability uses extended struct in ioctl interface: 3092 3093/* for KVM_CAP_SPAPR_TCE_64 */ 3094struct kvm_create_spapr_tce_64 { 3095 __u64 liobn; 3096 __u32 page_shift; 3097 __u32 flags; 3098 __u64 offset; /* in pages */ 3099 __u64 size; /* in pages */ 3100}; 3101 3102The aim of extension is to support an additional bigger DMA window with 3103a variable page size. 3104KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and 3105a bus offset of the corresponding DMA window, @size and @offset are numbers 3106of IOMMU pages. 3107 3108@flags are not used at the moment. 3109 3110The rest of functionality is identical to KVM_CREATE_SPAPR_TCE. 3111 31124.99 KVM_REINJECT_CONTROL 3113 3114Capability: KVM_CAP_REINJECT_CONTROL 3115Architectures: x86 3116Type: vm ioctl 3117Parameters: struct kvm_reinject_control (in) 3118Returns: 0 on success, 3119 -EFAULT if struct kvm_reinject_control cannot be read, 3120 -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier. 3121 3122i8254 (PIT) has two modes, reinject and !reinject. The default is reinject, 3123where KVM queues elapsed i8254 ticks and monitors completion of interrupt from 3124vector(s) that i8254 injects. Reinject mode dequeues a tick and injects its 3125interrupt whenever there isn't a pending interrupt from i8254. 3126!reinject mode injects an interrupt as soon as a tick arrives. 3127 3128struct kvm_reinject_control { 3129 __u8 pit_reinject; 3130 __u8 reserved[31]; 3131}; 3132 3133pit_reinject = 0 (!reinject mode) is recommended, unless running an old 3134operating system that uses the PIT for timing (e.g. Linux 2.4.x). 3135 31364.100 KVM_PPC_CONFIGURE_V3_MMU 3137 3138Capability: KVM_CAP_PPC_RADIX_MMU or KVM_CAP_PPC_HASH_MMU_V3 3139Architectures: ppc 3140Type: vm ioctl 3141Parameters: struct kvm_ppc_mmuv3_cfg (in) 3142Returns: 0 on success, 3143 -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read, 3144 -EINVAL if the configuration is invalid 3145 3146This ioctl controls whether the guest will use radix or HPT (hashed 3147page table) translation, and sets the pointer to the process table for 3148the guest. 3149 3150struct kvm_ppc_mmuv3_cfg { 3151 __u64 flags; 3152 __u64 process_table; 3153}; 3154 3155There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and 3156KVM_PPC_MMUV3_GTSE. KVM_PPC_MMUV3_RADIX, if set, configures the guest 3157to use radix tree translation, and if clear, to use HPT translation. 3158KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest 3159to be able to use the global TLB and SLB invalidation instructions; 3160if clear, the guest may not use these instructions. 3161 3162The process_table field specifies the address and size of the guest 3163process table, which is in the guest's space. This field is formatted 3164as the second doubleword of the partition table entry, as defined in 3165the Power ISA V3.00, Book III section 5.7.6.1. 3166 31674.101 KVM_PPC_GET_RMMU_INFO 3168 3169Capability: KVM_CAP_PPC_RADIX_MMU 3170Architectures: ppc 3171Type: vm ioctl 3172Parameters: struct kvm_ppc_rmmu_info (out) 3173Returns: 0 on success, 3174 -EFAULT if struct kvm_ppc_rmmu_info cannot be written, 3175 -EINVAL if no useful information can be returned 3176 3177This ioctl returns a structure containing two things: (a) a list 3178containing supported radix tree geometries, and (b) a list that maps 3179page sizes to put in the "AP" (actual page size) field for the tlbie 3180(TLB invalidate entry) instruction. 3181 3182struct kvm_ppc_rmmu_info { 3183 struct kvm_ppc_radix_geom { 3184 __u8 page_shift; 3185 __u8 level_bits[4]; 3186 __u8 pad[3]; 3187 } geometries[8]; 3188 __u32 ap_encodings[8]; 3189}; 3190 3191The geometries[] field gives up to 8 supported geometries for the 3192radix page table, in terms of the log base 2 of the smallest page 3193size, and the number of bits indexed at each level of the tree, from 3194the PTE level up to the PGD level in that order. Any unused entries 3195will have 0 in the page_shift field. 3196 3197The ap_encodings gives the supported page sizes and their AP field 3198encodings, encoded with the AP value in the top 3 bits and the log 3199base 2 of the page size in the bottom 6 bits. 3200 32014.102 KVM_PPC_RESIZE_HPT_PREPARE 3202 3203Capability: KVM_CAP_SPAPR_RESIZE_HPT 3204Architectures: powerpc 3205Type: vm ioctl 3206Parameters: struct kvm_ppc_resize_hpt (in) 3207Returns: 0 on successful completion, 3208 >0 if a new HPT is being prepared, the value is an estimated 3209 number of milliseconds until preparation is complete 3210 -EFAULT if struct kvm_reinject_control cannot be read, 3211 -EINVAL if the supplied shift or flags are invalid 3212 -ENOMEM if unable to allocate the new HPT 3213 -ENOSPC if there was a hash collision when moving existing 3214 HPT entries to the new HPT 3215 -EIO on other error conditions 3216 3217Used to implement the PAPR extension for runtime resizing of a guest's 3218Hashed Page Table (HPT). Specifically this starts, stops or monitors 3219the preparation of a new potential HPT for the guest, essentially 3220implementing the H_RESIZE_HPT_PREPARE hypercall. 3221 3222If called with shift > 0 when there is no pending HPT for the guest, 3223this begins preparation of a new pending HPT of size 2^(shift) bytes. 3224It then returns a positive integer with the estimated number of 3225milliseconds until preparation is complete. 3226 3227If called when there is a pending HPT whose size does not match that 3228requested in the parameters, discards the existing pending HPT and 3229creates a new one as above. 3230 3231If called when there is a pending HPT of the size requested, will: 3232 * If preparation of the pending HPT is already complete, return 0 3233 * If preparation of the pending HPT has failed, return an error 3234 code, then discard the pending HPT. 3235 * If preparation of the pending HPT is still in progress, return an 3236 estimated number of milliseconds until preparation is complete. 3237 3238If called with shift == 0, discards any currently pending HPT and 3239returns 0 (i.e. cancels any in-progress preparation). 3240 3241flags is reserved for future expansion, currently setting any bits in 3242flags will result in an -EINVAL. 3243 3244Normally this will be called repeatedly with the same parameters until 3245it returns <= 0. The first call will initiate preparation, subsequent 3246ones will monitor preparation until it completes or fails. 3247 3248struct kvm_ppc_resize_hpt { 3249 __u64 flags; 3250 __u32 shift; 3251 __u32 pad; 3252}; 3253 32544.103 KVM_PPC_RESIZE_HPT_COMMIT 3255 3256Capability: KVM_CAP_SPAPR_RESIZE_HPT 3257Architectures: powerpc 3258Type: vm ioctl 3259Parameters: struct kvm_ppc_resize_hpt (in) 3260Returns: 0 on successful completion, 3261 -EFAULT if struct kvm_reinject_control cannot be read, 3262 -EINVAL if the supplied shift or flags are invalid 3263 -ENXIO is there is no pending HPT, or the pending HPT doesn't 3264 have the requested size 3265 -EBUSY if the pending HPT is not fully prepared 3266 -ENOSPC if there was a hash collision when moving existing 3267 HPT entries to the new HPT 3268 -EIO on other error conditions 3269 3270Used to implement the PAPR extension for runtime resizing of a guest's 3271Hashed Page Table (HPT). Specifically this requests that the guest be 3272transferred to working with the new HPT, essentially implementing the 3273H_RESIZE_HPT_COMMIT hypercall. 3274 3275This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has 3276returned 0 with the same parameters. In other cases 3277KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or 3278-EBUSY, though others may be possible if the preparation was started, 3279but failed). 3280 3281This will have undefined effects on the guest if it has not already 3282placed itself in a quiescent state where no vcpu will make MMU enabled 3283memory accesses. 3284 3285On succsful completion, the pending HPT will become the guest's active 3286HPT and the previous HPT will be discarded. 3287 3288On failure, the guest will still be operating on its previous HPT. 3289 3290struct kvm_ppc_resize_hpt { 3291 __u64 flags; 3292 __u32 shift; 3293 __u32 pad; 3294}; 3295 32964.104 KVM_X86_GET_MCE_CAP_SUPPORTED 3297 3298Capability: KVM_CAP_MCE 3299Architectures: x86 3300Type: system ioctl 3301Parameters: u64 mce_cap (out) 3302Returns: 0 on success, -1 on error 3303 3304Returns supported MCE capabilities. The u64 mce_cap parameter 3305has the same format as the MSR_IA32_MCG_CAP register. Supported 3306capabilities will have the corresponding bits set. 3307 33084.105 KVM_X86_SETUP_MCE 3309 3310Capability: KVM_CAP_MCE 3311Architectures: x86 3312Type: vcpu ioctl 3313Parameters: u64 mcg_cap (in) 3314Returns: 0 on success, 3315 -EFAULT if u64 mcg_cap cannot be read, 3316 -EINVAL if the requested number of banks is invalid, 3317 -EINVAL if requested MCE capability is not supported. 3318 3319Initializes MCE support for use. The u64 mcg_cap parameter 3320has the same format as the MSR_IA32_MCG_CAP register and 3321specifies which capabilities should be enabled. The maximum 3322supported number of error-reporting banks can be retrieved when 3323checking for KVM_CAP_MCE. The supported capabilities can be 3324retrieved with KVM_X86_GET_MCE_CAP_SUPPORTED. 3325 33264.106 KVM_X86_SET_MCE 3327 3328Capability: KVM_CAP_MCE 3329Architectures: x86 3330Type: vcpu ioctl 3331Parameters: struct kvm_x86_mce (in) 3332Returns: 0 on success, 3333 -EFAULT if struct kvm_x86_mce cannot be read, 3334 -EINVAL if the bank number is invalid, 3335 -EINVAL if VAL bit is not set in status field. 3336 3337Inject a machine check error (MCE) into the guest. The input 3338parameter is: 3339 3340struct kvm_x86_mce { 3341 __u64 status; 3342 __u64 addr; 3343 __u64 misc; 3344 __u64 mcg_status; 3345 __u8 bank; 3346 __u8 pad1[7]; 3347 __u64 pad2[3]; 3348}; 3349 3350If the MCE being reported is an uncorrected error, KVM will 3351inject it as an MCE exception into the guest. If the guest 3352MCG_STATUS register reports that an MCE is in progress, KVM 3353causes an KVM_EXIT_SHUTDOWN vmexit. 3354 3355Otherwise, if the MCE is a corrected error, KVM will just 3356store it in the corresponding bank (provided this bank is 3357not holding a previously reported uncorrected error). 3358 33594.107 KVM_S390_GET_CMMA_BITS 3360 3361Capability: KVM_CAP_S390_CMMA_MIGRATION 3362Architectures: s390 3363Type: vm ioctl 3364Parameters: struct kvm_s390_cmma_log (in, out) 3365Returns: 0 on success, a negative value on error 3366 3367This ioctl is used to get the values of the CMMA bits on the s390 3368architecture. It is meant to be used in two scenarios: 3369- During live migration to save the CMMA values. Live migration needs 3370 to be enabled via the KVM_REQ_START_MIGRATION VM property. 3371- To non-destructively peek at the CMMA values, with the flag 3372 KVM_S390_CMMA_PEEK set. 3373 3374The ioctl takes parameters via the kvm_s390_cmma_log struct. The desired 3375values are written to a buffer whose location is indicated via the "values" 3376member in the kvm_s390_cmma_log struct. The values in the input struct are 3377also updated as needed. 3378Each CMMA value takes up one byte. 3379 3380struct kvm_s390_cmma_log { 3381 __u64 start_gfn; 3382 __u32 count; 3383 __u32 flags; 3384 union { 3385 __u64 remaining; 3386 __u64 mask; 3387 }; 3388 __u64 values; 3389}; 3390 3391start_gfn is the number of the first guest frame whose CMMA values are 3392to be retrieved, 3393 3394count is the length of the buffer in bytes, 3395 3396values points to the buffer where the result will be written to. 3397 3398If count is greater than KVM_S390_SKEYS_MAX, then it is considered to be 3399KVM_S390_SKEYS_MAX. KVM_S390_SKEYS_MAX is re-used for consistency with 3400other ioctls. 3401 3402The result is written in the buffer pointed to by the field values, and 3403the values of the input parameter are updated as follows. 3404 3405Depending on the flags, different actions are performed. The only 3406supported flag so far is KVM_S390_CMMA_PEEK. 3407 3408The default behaviour if KVM_S390_CMMA_PEEK is not set is: 3409start_gfn will indicate the first page frame whose CMMA bits were dirty. 3410It is not necessarily the same as the one passed as input, as clean pages 3411are skipped. 3412 3413count will indicate the number of bytes actually written in the buffer. 3414It can (and very often will) be smaller than the input value, since the 3415buffer is only filled until 16 bytes of clean values are found (which 3416are then not copied in the buffer). Since a CMMA migration block needs 3417the base address and the length, for a total of 16 bytes, we will send 3418back some clean data if there is some dirty data afterwards, as long as 3419the size of the clean data does not exceed the size of the header. This 3420allows to minimize the amount of data to be saved or transferred over 3421the network at the expense of more roundtrips to userspace. The next 3422invocation of the ioctl will skip over all the clean values, saving 3423potentially more than just the 16 bytes we found. 3424 3425If KVM_S390_CMMA_PEEK is set: 3426the existing storage attributes are read even when not in migration 3427mode, and no other action is performed; 3428 3429the output start_gfn will be equal to the input start_gfn, 3430 3431the output count will be equal to the input count, except if the end of 3432memory has been reached. 3433 3434In both cases: 3435the field "remaining" will indicate the total number of dirty CMMA values 3436still remaining, or 0 if KVM_S390_CMMA_PEEK is set and migration mode is 3437not enabled. 3438 3439mask is unused. 3440 3441values points to the userspace buffer where the result will be stored. 3442 3443This ioctl can fail with -ENOMEM if not enough memory can be allocated to 3444complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if 3445KVM_S390_CMMA_PEEK is not set but migration mode was not enabled, with 3446-EFAULT if the userspace address is invalid or if no page table is 3447present for the addresses (e.g. when using hugepages). 3448 34494.108 KVM_S390_SET_CMMA_BITS 3450 3451Capability: KVM_CAP_S390_CMMA_MIGRATION 3452Architectures: s390 3453Type: vm ioctl 3454Parameters: struct kvm_s390_cmma_log (in) 3455Returns: 0 on success, a negative value on error 3456 3457This ioctl is used to set the values of the CMMA bits on the s390 3458architecture. It is meant to be used during live migration to restore 3459the CMMA values, but there are no restrictions on its use. 3460The ioctl takes parameters via the kvm_s390_cmma_values struct. 3461Each CMMA value takes up one byte. 3462 3463struct kvm_s390_cmma_log { 3464 __u64 start_gfn; 3465 __u32 count; 3466 __u32 flags; 3467 union { 3468 __u64 remaining; 3469 __u64 mask; 3470 }; 3471 __u64 values; 3472}; 3473 3474start_gfn indicates the starting guest frame number, 3475 3476count indicates how many values are to be considered in the buffer, 3477 3478flags is not used and must be 0. 3479 3480mask indicates which PGSTE bits are to be considered. 3481 3482remaining is not used. 3483 3484values points to the buffer in userspace where to store the values. 3485 3486This ioctl can fail with -ENOMEM if not enough memory can be allocated to 3487complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if 3488the count field is too large (e.g. more than KVM_S390_CMMA_SIZE_MAX) or 3489if the flags field was not 0, with -EFAULT if the userspace address is 3490invalid, if invalid pages are written to (e.g. after the end of memory) 3491or if no page table is present for the addresses (e.g. when using 3492hugepages). 3493 34944.109 KVM_PPC_GET_CPU_CHAR 3495 3496Capability: KVM_CAP_PPC_GET_CPU_CHAR 3497Architectures: powerpc 3498Type: vm ioctl 3499Parameters: struct kvm_ppc_cpu_char (out) 3500Returns: 0 on successful completion 3501 -EFAULT if struct kvm_ppc_cpu_char cannot be written 3502 3503This ioctl gives userspace information about certain characteristics 3504of the CPU relating to speculative execution of instructions and 3505possible information leakage resulting from speculative execution (see 3506CVE-2017-5715, CVE-2017-5753 and CVE-2017-5754). The information is 3507returned in struct kvm_ppc_cpu_char, which looks like this: 3508 3509struct kvm_ppc_cpu_char { 3510 __u64 character; /* characteristics of the CPU */ 3511 __u64 behaviour; /* recommended software behaviour */ 3512 __u64 character_mask; /* valid bits in character */ 3513 __u64 behaviour_mask; /* valid bits in behaviour */ 3514}; 3515 3516For extensibility, the character_mask and behaviour_mask fields 3517indicate which bits of character and behaviour have been filled in by 3518the kernel. If the set of defined bits is extended in future then 3519userspace will be able to tell whether it is running on a kernel that 3520knows about the new bits. 3521 3522The character field describes attributes of the CPU which can help 3523with preventing inadvertent information disclosure - specifically, 3524whether there is an instruction to flash-invalidate the L1 data cache 3525(ori 30,30,0 or mtspr SPRN_TRIG2,rN), whether the L1 data cache is set 3526to a mode where entries can only be used by the thread that created 3527them, whether the bcctr[l] instruction prevents speculation, and 3528whether a speculation barrier instruction (ori 31,31,0) is provided. 3529 3530The behaviour field describes actions that software should take to 3531prevent inadvertent information disclosure, and thus describes which 3532vulnerabilities the hardware is subject to; specifically whether the 3533L1 data cache should be flushed when returning to user mode from the 3534kernel, and whether a speculation barrier should be placed between an 3535array bounds check and the array access. 3536 3537These fields use the same bit definitions as the new 3538H_GET_CPU_CHARACTERISTICS hypercall. 3539 35404.110 KVM_MEMORY_ENCRYPT_OP 3541 3542Capability: basic 3543Architectures: x86 3544Type: system 3545Parameters: an opaque platform specific structure (in/out) 3546Returns: 0 on success; -1 on error 3547 3548If the platform supports creating encrypted VMs then this ioctl can be used 3549for issuing platform-specific memory encryption commands to manage those 3550encrypted VMs. 3551 3552Currently, this ioctl is used for issuing Secure Encrypted Virtualization 3553(SEV) commands on AMD Processors. The SEV commands are defined in 3554Documentation/virtual/kvm/amd-memory-encryption.rst. 3555 35564.111 KVM_MEMORY_ENCRYPT_REG_REGION 3557 3558Capability: basic 3559Architectures: x86 3560Type: system 3561Parameters: struct kvm_enc_region (in) 3562Returns: 0 on success; -1 on error 3563 3564This ioctl can be used to register a guest memory region which may 3565contain encrypted data (e.g. guest RAM, SMRAM etc). 3566 3567It is used in the SEV-enabled guest. When encryption is enabled, a guest 3568memory region may contain encrypted data. The SEV memory encryption 3569engine uses a tweak such that two identical plaintext pages, each at 3570different locations will have differing ciphertexts. So swapping or 3571moving ciphertext of those pages will not result in plaintext being 3572swapped. So relocating (or migrating) physical backing pages for the SEV 3573guest will require some additional steps. 3574 3575Note: The current SEV key management spec does not provide commands to 3576swap or migrate (move) ciphertext pages. Hence, for now we pin the guest 3577memory region registered with the ioctl. 3578 35794.112 KVM_MEMORY_ENCRYPT_UNREG_REGION 3580 3581Capability: basic 3582Architectures: x86 3583Type: system 3584Parameters: struct kvm_enc_region (in) 3585Returns: 0 on success; -1 on error 3586 3587This ioctl can be used to unregister the guest memory region registered 3588with KVM_MEMORY_ENCRYPT_REG_REGION ioctl above. 3589 35904.113 KVM_HYPERV_EVENTFD 3591 3592Capability: KVM_CAP_HYPERV_EVENTFD 3593Architectures: x86 3594Type: vm ioctl 3595Parameters: struct kvm_hyperv_eventfd (in) 3596 3597This ioctl (un)registers an eventfd to receive notifications from the guest on 3598the specified Hyper-V connection id through the SIGNAL_EVENT hypercall, without 3599causing a user exit. SIGNAL_EVENT hypercall with non-zero event flag number 3600(bits 24-31) still triggers a KVM_EXIT_HYPERV_HCALL user exit. 3601 3602struct kvm_hyperv_eventfd { 3603 __u32 conn_id; 3604 __s32 fd; 3605 __u32 flags; 3606 __u32 padding[3]; 3607}; 3608 3609The conn_id field should fit within 24 bits: 3610 3611#define KVM_HYPERV_CONN_ID_MASK 0x00ffffff 3612 3613The acceptable values for the flags field are: 3614 3615#define KVM_HYPERV_EVENTFD_DEASSIGN (1 << 0) 3616 3617Returns: 0 on success, 3618 -EINVAL if conn_id or flags is outside the allowed range 3619 -ENOENT on deassign if the conn_id isn't registered 3620 -EEXIST on assign if the conn_id is already registered 3621 36224.114 KVM_GET_NESTED_STATE 3623 3624Capability: KVM_CAP_NESTED_STATE 3625Architectures: x86 3626Type: vcpu ioctl 3627Parameters: struct kvm_nested_state (in/out) 3628Returns: 0 on success, -1 on error 3629Errors: 3630 E2BIG: the total state size (including the fixed-size part of struct 3631 kvm_nested_state) exceeds the value of 'size' specified by 3632 the user; the size required will be written into size. 3633 3634struct kvm_nested_state { 3635 __u16 flags; 3636 __u16 format; 3637 __u32 size; 3638 union { 3639 struct kvm_vmx_nested_state vmx; 3640 struct kvm_svm_nested_state svm; 3641 __u8 pad[120]; 3642 }; 3643 __u8 data[0]; 3644}; 3645 3646#define KVM_STATE_NESTED_GUEST_MODE 0x00000001 3647#define KVM_STATE_NESTED_RUN_PENDING 0x00000002 3648 3649#define KVM_STATE_NESTED_SMM_GUEST_MODE 0x00000001 3650#define KVM_STATE_NESTED_SMM_VMXON 0x00000002 3651 3652struct kvm_vmx_nested_state { 3653 __u64 vmxon_pa; 3654 __u64 vmcs_pa; 3655 3656 struct { 3657 __u16 flags; 3658 } smm; 3659}; 3660 3661This ioctl copies the vcpu's nested virtualization state from the kernel to 3662userspace. 3663 3664The maximum size of the state, including the fixed-size part of struct 3665kvm_nested_state, can be retrieved by passing KVM_CAP_NESTED_STATE to 3666the KVM_CHECK_EXTENSION ioctl(). 3667 36684.115 KVM_SET_NESTED_STATE 3669 3670Capability: KVM_CAP_NESTED_STATE 3671Architectures: x86 3672Type: vcpu ioctl 3673Parameters: struct kvm_nested_state (in) 3674Returns: 0 on success, -1 on error 3675 3676This copies the vcpu's kvm_nested_state struct from userspace to the kernel. For 3677the definition of struct kvm_nested_state, see KVM_GET_NESTED_STATE. 3678 36795. The kvm_run structure 3680------------------------ 3681 3682Application code obtains a pointer to the kvm_run structure by 3683mmap()ing a vcpu fd. From that point, application code can control 3684execution by changing fields in kvm_run prior to calling the KVM_RUN 3685ioctl, and obtain information about the reason KVM_RUN returned by 3686looking up structure members. 3687 3688struct kvm_run { 3689 /* in */ 3690 __u8 request_interrupt_window; 3691 3692Request that KVM_RUN return when it becomes possible to inject external 3693interrupts into the guest. Useful in conjunction with KVM_INTERRUPT. 3694 3695 __u8 immediate_exit; 3696 3697This field is polled once when KVM_RUN starts; if non-zero, KVM_RUN 3698exits immediately, returning -EINTR. In the common scenario where a 3699signal is used to "kick" a VCPU out of KVM_RUN, this field can be used 3700to avoid usage of KVM_SET_SIGNAL_MASK, which has worse scalability. 3701Rather than blocking the signal outside KVM_RUN, userspace can set up 3702a signal handler that sets run->immediate_exit to a non-zero value. 3703 3704This field is ignored if KVM_CAP_IMMEDIATE_EXIT is not available. 3705 3706 __u8 padding1[6]; 3707 3708 /* out */ 3709 __u32 exit_reason; 3710 3711When KVM_RUN has returned successfully (return value 0), this informs 3712application code why KVM_RUN has returned. Allowable values for this 3713field are detailed below. 3714 3715 __u8 ready_for_interrupt_injection; 3716 3717If request_interrupt_window has been specified, this field indicates 3718an interrupt can be injected now with KVM_INTERRUPT. 3719 3720 __u8 if_flag; 3721 3722The value of the current interrupt flag. Only valid if in-kernel 3723local APIC is not used. 3724 3725 __u16 flags; 3726 3727More architecture-specific flags detailing state of the VCPU that may 3728affect the device's behavior. The only currently defined flag is 3729KVM_RUN_X86_SMM, which is valid on x86 machines and is set if the 3730VCPU is in system management mode. 3731 3732 /* in (pre_kvm_run), out (post_kvm_run) */ 3733 __u64 cr8; 3734 3735The value of the cr8 register. Only valid if in-kernel local APIC is 3736not used. Both input and output. 3737 3738 __u64 apic_base; 3739 3740The value of the APIC BASE msr. Only valid if in-kernel local 3741APIC is not used. Both input and output. 3742 3743 union { 3744 /* KVM_EXIT_UNKNOWN */ 3745 struct { 3746 __u64 hardware_exit_reason; 3747 } hw; 3748 3749If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown 3750reasons. Further architecture-specific information is available in 3751hardware_exit_reason. 3752 3753 /* KVM_EXIT_FAIL_ENTRY */ 3754 struct { 3755 __u64 hardware_entry_failure_reason; 3756 } fail_entry; 3757 3758If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due 3759to unknown reasons. Further architecture-specific information is 3760available in hardware_entry_failure_reason. 3761 3762 /* KVM_EXIT_EXCEPTION */ 3763 struct { 3764 __u32 exception; 3765 __u32 error_code; 3766 } ex; 3767 3768Unused. 3769 3770 /* KVM_EXIT_IO */ 3771 struct { 3772#define KVM_EXIT_IO_IN 0 3773#define KVM_EXIT_IO_OUT 1 3774 __u8 direction; 3775 __u8 size; /* bytes */ 3776 __u16 port; 3777 __u32 count; 3778 __u64 data_offset; /* relative to kvm_run start */ 3779 } io; 3780 3781If exit_reason is KVM_EXIT_IO, then the vcpu has 3782executed a port I/O instruction which could not be satisfied by kvm. 3783data_offset describes where the data is located (KVM_EXIT_IO_OUT) or 3784where kvm expects application code to place the data for the next 3785KVM_RUN invocation (KVM_EXIT_IO_IN). Data format is a packed array. 3786 3787 /* KVM_EXIT_DEBUG */ 3788 struct { 3789 struct kvm_debug_exit_arch arch; 3790 } debug; 3791 3792If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event 3793for which architecture specific information is returned. 3794 3795 /* KVM_EXIT_MMIO */ 3796 struct { 3797 __u64 phys_addr; 3798 __u8 data[8]; 3799 __u32 len; 3800 __u8 is_write; 3801 } mmio; 3802 3803If exit_reason is KVM_EXIT_MMIO, then the vcpu has 3804executed a memory-mapped I/O instruction which could not be satisfied 3805by kvm. The 'data' member contains the written data if 'is_write' is 3806true, and should be filled by application code otherwise. 3807 3808The 'data' member contains, in its first 'len' bytes, the value as it would 3809appear if the VCPU performed a load or store of the appropriate width directly 3810to the byte array. 3811 3812NOTE: For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR and 3813 KVM_EXIT_EPR the corresponding 3814operations are complete (and guest state is consistent) only after userspace 3815has re-entered the kernel with KVM_RUN. The kernel side will first finish 3816incomplete operations and then check for pending signals. Userspace 3817can re-enter the guest with an unmasked signal pending to complete 3818pending operations. 3819 3820 /* KVM_EXIT_HYPERCALL */ 3821 struct { 3822 __u64 nr; 3823 __u64 args[6]; 3824 __u64 ret; 3825 __u32 longmode; 3826 __u32 pad; 3827 } hypercall; 3828 3829Unused. This was once used for 'hypercall to userspace'. To implement 3830such functionality, use KVM_EXIT_IO (x86) or KVM_EXIT_MMIO (all except s390). 3831Note KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO. 3832 3833 /* KVM_EXIT_TPR_ACCESS */ 3834 struct { 3835 __u64 rip; 3836 __u32 is_write; 3837 __u32 pad; 3838 } tpr_access; 3839 3840To be documented (KVM_TPR_ACCESS_REPORTING). 3841 3842 /* KVM_EXIT_S390_SIEIC */ 3843 struct { 3844 __u8 icptcode; 3845 __u64 mask; /* psw upper half */ 3846 __u64 addr; /* psw lower half */ 3847 __u16 ipa; 3848 __u32 ipb; 3849 } s390_sieic; 3850 3851s390 specific. 3852 3853 /* KVM_EXIT_S390_RESET */ 3854#define KVM_S390_RESET_POR 1 3855#define KVM_S390_RESET_CLEAR 2 3856#define KVM_S390_RESET_SUBSYSTEM 4 3857#define KVM_S390_RESET_CPU_INIT 8 3858#define KVM_S390_RESET_IPL 16 3859 __u64 s390_reset_flags; 3860 3861s390 specific. 3862 3863 /* KVM_EXIT_S390_UCONTROL */ 3864 struct { 3865 __u64 trans_exc_code; 3866 __u32 pgm_code; 3867 } s390_ucontrol; 3868 3869s390 specific. A page fault has occurred for a user controlled virtual 3870machine (KVM_VM_S390_UNCONTROL) on it's host page table that cannot be 3871resolved by the kernel. 3872The program code and the translation exception code that were placed 3873in the cpu's lowcore are presented here as defined by the z Architecture 3874Principles of Operation Book in the Chapter for Dynamic Address Translation 3875(DAT) 3876 3877 /* KVM_EXIT_DCR */ 3878 struct { 3879 __u32 dcrn; 3880 __u32 data; 3881 __u8 is_write; 3882 } dcr; 3883 3884Deprecated - was used for 440 KVM. 3885 3886 /* KVM_EXIT_OSI */ 3887 struct { 3888 __u64 gprs[32]; 3889 } osi; 3890 3891MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch 3892hypercalls and exit with this exit struct that contains all the guest gprs. 3893 3894If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall. 3895Userspace can now handle the hypercall and when it's done modify the gprs as 3896necessary. Upon guest entry all guest GPRs will then be replaced by the values 3897in this struct. 3898 3899 /* KVM_EXIT_PAPR_HCALL */ 3900 struct { 3901 __u64 nr; 3902 __u64 ret; 3903 __u64 args[9]; 3904 } papr_hcall; 3905 3906This is used on 64-bit PowerPC when emulating a pSeries partition, 3907e.g. with the 'pseries' machine type in qemu. It occurs when the 3908guest does a hypercall using the 'sc 1' instruction. The 'nr' field 3909contains the hypercall number (from the guest R3), and 'args' contains 3910the arguments (from the guest R4 - R12). Userspace should put the 3911return code in 'ret' and any extra returned values in args[]. 3912The possible hypercalls are defined in the Power Architecture Platform 3913Requirements (PAPR) document available from www.power.org (free 3914developer registration required to access it). 3915 3916 /* KVM_EXIT_S390_TSCH */ 3917 struct { 3918 __u16 subchannel_id; 3919 __u16 subchannel_nr; 3920 __u32 io_int_parm; 3921 __u32 io_int_word; 3922 __u32 ipb; 3923 __u8 dequeued; 3924 } s390_tsch; 3925 3926s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled 3927and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O 3928interrupt for the target subchannel has been dequeued and subchannel_id, 3929subchannel_nr, io_int_parm and io_int_word contain the parameters for that 3930interrupt. ipb is needed for instruction parameter decoding. 3931 3932 /* KVM_EXIT_EPR */ 3933 struct { 3934 __u32 epr; 3935 } epr; 3936 3937On FSL BookE PowerPC chips, the interrupt controller has a fast patch 3938interrupt acknowledge path to the core. When the core successfully 3939delivers an interrupt, it automatically populates the EPR register with 3940the interrupt vector number and acknowledges the interrupt inside 3941the interrupt controller. 3942 3943In case the interrupt controller lives in user space, we need to do 3944the interrupt acknowledge cycle through it to fetch the next to be 3945delivered interrupt vector using this exit. 3946 3947It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an 3948external interrupt has just been delivered into the guest. User space 3949should put the acknowledged interrupt vector into the 'epr' field. 3950 3951 /* KVM_EXIT_SYSTEM_EVENT */ 3952 struct { 3953#define KVM_SYSTEM_EVENT_SHUTDOWN 1 3954#define KVM_SYSTEM_EVENT_RESET 2 3955#define KVM_SYSTEM_EVENT_CRASH 3 3956 __u32 type; 3957 __u64 flags; 3958 } system_event; 3959 3960If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered 3961a system-level event using some architecture specific mechanism (hypercall 3962or some special instruction). In case of ARM/ARM64, this is triggered using 3963HVC instruction based PSCI call from the vcpu. The 'type' field describes 3964the system-level event type. The 'flags' field describes architecture 3965specific flags for the system-level event. 3966 3967Valid values for 'type' are: 3968 KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the 3969 VM. Userspace is not obliged to honour this, and if it does honour 3970 this does not need to destroy the VM synchronously (ie it may call 3971 KVM_RUN again before shutdown finally occurs). 3972 KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM. 3973 As with SHUTDOWN, userspace can choose to ignore the request, or 3974 to schedule the reset to occur in the future and may call KVM_RUN again. 3975 KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest 3976 has requested a crash condition maintenance. Userspace can choose 3977 to ignore the request, or to gather VM memory core dump and/or 3978 reset/shutdown of the VM. 3979 3980 /* KVM_EXIT_IOAPIC_EOI */ 3981 struct { 3982 __u8 vector; 3983 } eoi; 3984 3985Indicates that the VCPU's in-kernel local APIC received an EOI for a 3986level-triggered IOAPIC interrupt. This exit only triggers when the 3987IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled); 3988the userspace IOAPIC should process the EOI and retrigger the interrupt if 3989it is still asserted. Vector is the LAPIC interrupt vector for which the 3990EOI was received. 3991 3992 struct kvm_hyperv_exit { 3993#define KVM_EXIT_HYPERV_SYNIC 1 3994#define KVM_EXIT_HYPERV_HCALL 2 3995 __u32 type; 3996 union { 3997 struct { 3998 __u32 msr; 3999 __u64 control; 4000 __u64 evt_page; 4001 __u64 msg_page; 4002 } synic; 4003 struct { 4004 __u64 input; 4005 __u64 result; 4006 __u64 params[2]; 4007 } hcall; 4008 } u; 4009 }; 4010 /* KVM_EXIT_HYPERV */ 4011 struct kvm_hyperv_exit hyperv; 4012Indicates that the VCPU exits into userspace to process some tasks 4013related to Hyper-V emulation. 4014Valid values for 'type' are: 4015 KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about 4016Hyper-V SynIC state change. Notification is used to remap SynIC 4017event/message pages and to enable/disable SynIC messages/events processing 4018in userspace. 4019 4020 /* Fix the size of the union. */ 4021 char padding[256]; 4022 }; 4023 4024 /* 4025 * shared registers between kvm and userspace. 4026 * kvm_valid_regs specifies the register classes set by the host 4027 * kvm_dirty_regs specified the register classes dirtied by userspace 4028 * struct kvm_sync_regs is architecture specific, as well as the 4029 * bits for kvm_valid_regs and kvm_dirty_regs 4030 */ 4031 __u64 kvm_valid_regs; 4032 __u64 kvm_dirty_regs; 4033 union { 4034 struct kvm_sync_regs regs; 4035 char padding[SYNC_REGS_SIZE_BYTES]; 4036 } s; 4037 4038If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access 4039certain guest registers without having to call SET/GET_*REGS. Thus we can 4040avoid some system call overhead if userspace has to handle the exit. 4041Userspace can query the validity of the structure by checking 4042kvm_valid_regs for specific bits. These bits are architecture specific 4043and usually define the validity of a groups of registers. (e.g. one bit 4044 for general purpose registers) 4045 4046Please note that the kernel is allowed to use the kvm_run structure as the 4047primary storage for certain register types. Therefore, the kernel may use the 4048values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set. 4049 4050}; 4051 4052 4053 40546. Capabilities that can be enabled on vCPUs 4055-------------------------------------------- 4056 4057There are certain capabilities that change the behavior of the virtual CPU or 4058the virtual machine when enabled. To enable them, please see section 4.37. 4059Below you can find a list of capabilities and what their effect on the vCPU or 4060the virtual machine is when enabling them. 4061 4062The following information is provided along with the description: 4063 4064 Architectures: which instruction set architectures provide this ioctl. 4065 x86 includes both i386 and x86_64. 4066 4067 Target: whether this is a per-vcpu or per-vm capability. 4068 4069 Parameters: what parameters are accepted by the capability. 4070 4071 Returns: the return value. General error numbers (EBADF, ENOMEM, EINVAL) 4072 are not detailed, but errors with specific meanings are. 4073 4074 40756.1 KVM_CAP_PPC_OSI 4076 4077Architectures: ppc 4078Target: vcpu 4079Parameters: none 4080Returns: 0 on success; -1 on error 4081 4082This capability enables interception of OSI hypercalls that otherwise would 4083be treated as normal system calls to be injected into the guest. OSI hypercalls 4084were invented by Mac-on-Linux to have a standardized communication mechanism 4085between the guest and the host. 4086 4087When this capability is enabled, KVM_EXIT_OSI can occur. 4088 4089 40906.2 KVM_CAP_PPC_PAPR 4091 4092Architectures: ppc 4093Target: vcpu 4094Parameters: none 4095Returns: 0 on success; -1 on error 4096 4097This capability enables interception of PAPR hypercalls. PAPR hypercalls are 4098done using the hypercall instruction "sc 1". 4099 4100It also sets the guest privilege level to "supervisor" mode. Usually the guest 4101runs in "hypervisor" privilege mode with a few missing features. 4102 4103In addition to the above, it changes the semantics of SDR1. In this mode, the 4104HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the 4105HTAB invisible to the guest. 4106 4107When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur. 4108 4109 41106.3 KVM_CAP_SW_TLB 4111 4112Architectures: ppc 4113Target: vcpu 4114Parameters: args[0] is the address of a struct kvm_config_tlb 4115Returns: 0 on success; -1 on error 4116 4117struct kvm_config_tlb { 4118 __u64 params; 4119 __u64 array; 4120 __u32 mmu_type; 4121 __u32 array_len; 4122}; 4123 4124Configures the virtual CPU's TLB array, establishing a shared memory area 4125between userspace and KVM. The "params" and "array" fields are userspace 4126addresses of mmu-type-specific data structures. The "array_len" field is an 4127safety mechanism, and should be set to the size in bytes of the memory that 4128userspace has reserved for the array. It must be at least the size dictated 4129by "mmu_type" and "params". 4130 4131While KVM_RUN is active, the shared region is under control of KVM. Its 4132contents are undefined, and any modification by userspace results in 4133boundedly undefined behavior. 4134 4135On return from KVM_RUN, the shared region will reflect the current state of 4136the guest's TLB. If userspace makes any changes, it must call KVM_DIRTY_TLB 4137to tell KVM which entries have been changed, prior to calling KVM_RUN again 4138on this vcpu. 4139 4140For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV: 4141 - The "params" field is of type "struct kvm_book3e_206_tlb_params". 4142 - The "array" field points to an array of type "struct 4143 kvm_book3e_206_tlb_entry". 4144 - The array consists of all entries in the first TLB, followed by all 4145 entries in the second TLB. 4146 - Within a TLB, entries are ordered first by increasing set number. Within a 4147 set, entries are ordered by way (increasing ESEL). 4148 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1) 4149 where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value. 4150 - The tsize field of mas1 shall be set to 4K on TLB0, even though the 4151 hardware ignores this value for TLB0. 4152 41536.4 KVM_CAP_S390_CSS_SUPPORT 4154 4155Architectures: s390 4156Target: vcpu 4157Parameters: none 4158Returns: 0 on success; -1 on error 4159 4160This capability enables support for handling of channel I/O instructions. 4161 4162TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are 4163handled in-kernel, while the other I/O instructions are passed to userspace. 4164 4165When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST 4166SUBCHANNEL intercepts. 4167 4168Note that even though this capability is enabled per-vcpu, the complete 4169virtual machine is affected. 4170 41716.5 KVM_CAP_PPC_EPR 4172 4173Architectures: ppc 4174Target: vcpu 4175Parameters: args[0] defines whether the proxy facility is active 4176Returns: 0 on success; -1 on error 4177 4178This capability enables or disables the delivery of interrupts through the 4179external proxy facility. 4180 4181When enabled (args[0] != 0), every time the guest gets an external interrupt 4182delivered, it automatically exits into user space with a KVM_EXIT_EPR exit 4183to receive the topmost interrupt vector. 4184 4185When disabled (args[0] == 0), behavior is as if this facility is unsupported. 4186 4187When this capability is enabled, KVM_EXIT_EPR can occur. 4188 41896.6 KVM_CAP_IRQ_MPIC 4190 4191Architectures: ppc 4192Parameters: args[0] is the MPIC device fd 4193 args[1] is the MPIC CPU number for this vcpu 4194 4195This capability connects the vcpu to an in-kernel MPIC device. 4196 41976.7 KVM_CAP_IRQ_XICS 4198 4199Architectures: ppc 4200Target: vcpu 4201Parameters: args[0] is the XICS device fd 4202 args[1] is the XICS CPU number (server ID) for this vcpu 4203 4204This capability connects the vcpu to an in-kernel XICS device. 4205 42066.8 KVM_CAP_S390_IRQCHIP 4207 4208Architectures: s390 4209Target: vm 4210Parameters: none 4211 4212This capability enables the in-kernel irqchip for s390. Please refer to 4213"4.24 KVM_CREATE_IRQCHIP" for details. 4214 42156.9 KVM_CAP_MIPS_FPU 4216 4217Architectures: mips 4218Target: vcpu 4219Parameters: args[0] is reserved for future use (should be 0). 4220 4221This capability allows the use of the host Floating Point Unit by the guest. It 4222allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is 4223done the KVM_REG_MIPS_FPR_* and KVM_REG_MIPS_FCR_* registers can be accessed 4224(depending on the current guest FPU register mode), and the Status.FR, 4225Config5.FRE bits are accessible via the KVM API and also from the guest, 4226depending on them being supported by the FPU. 4227 42286.10 KVM_CAP_MIPS_MSA 4229 4230Architectures: mips 4231Target: vcpu 4232Parameters: args[0] is reserved for future use (should be 0). 4233 4234This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest. 4235It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest. 4236Once this is done the KVM_REG_MIPS_VEC_* and KVM_REG_MIPS_MSA_* registers can be 4237accessed, and the Config5.MSAEn bit is accessible via the KVM API and also from 4238the guest. 4239 42406.74 KVM_CAP_SYNC_REGS 4241Architectures: s390, x86 4242Target: s390: always enabled, x86: vcpu 4243Parameters: none 4244Returns: x86: KVM_CHECK_EXTENSION returns a bit-array indicating which register 4245sets are supported (bitfields defined in arch/x86/include/uapi/asm/kvm.h). 4246 4247As described above in the kvm_sync_regs struct info in section 5 (kvm_run): 4248KVM_CAP_SYNC_REGS "allow[s] userspace to access certain guest registers 4249without having to call SET/GET_*REGS". This reduces overhead by eliminating 4250repeated ioctl calls for setting and/or getting register values. This is 4251particularly important when userspace is making synchronous guest state 4252modifications, e.g. when emulating and/or intercepting instructions in 4253userspace. 4254 4255For s390 specifics, please refer to the source code. 4256 4257For x86: 4258- the register sets to be copied out to kvm_run are selectable 4259 by userspace (rather that all sets being copied out for every exit). 4260- vcpu_events are available in addition to regs and sregs. 4261 4262For x86, the 'kvm_valid_regs' field of struct kvm_run is overloaded to 4263function as an input bit-array field set by userspace to indicate the 4264specific register sets to be copied out on the next exit. 4265 4266To indicate when userspace has modified values that should be copied into 4267the vCPU, the all architecture bitarray field, 'kvm_dirty_regs' must be set. 4268This is done using the same bitflags as for the 'kvm_valid_regs' field. 4269If the dirty bit is not set, then the register set values will not be copied 4270into the vCPU even if they've been modified. 4271 4272Unused bitfields in the bitarrays must be set to zero. 4273 4274struct kvm_sync_regs { 4275 struct kvm_regs regs; 4276 struct kvm_sregs sregs; 4277 struct kvm_vcpu_events events; 4278}; 4279 42807. Capabilities that can be enabled on VMs 4281------------------------------------------ 4282 4283There are certain capabilities that change the behavior of the virtual 4284machine when enabled. To enable them, please see section 4.37. Below 4285you can find a list of capabilities and what their effect on the VM 4286is when enabling them. 4287 4288The following information is provided along with the description: 4289 4290 Architectures: which instruction set architectures provide this ioctl. 4291 x86 includes both i386 and x86_64. 4292 4293 Parameters: what parameters are accepted by the capability. 4294 4295 Returns: the return value. General error numbers (EBADF, ENOMEM, EINVAL) 4296 are not detailed, but errors with specific meanings are. 4297 4298 42997.1 KVM_CAP_PPC_ENABLE_HCALL 4300 4301Architectures: ppc 4302Parameters: args[0] is the sPAPR hcall number 4303 args[1] is 0 to disable, 1 to enable in-kernel handling 4304 4305This capability controls whether individual sPAPR hypercalls (hcalls) 4306get handled by the kernel or not. Enabling or disabling in-kernel 4307handling of an hcall is effective across the VM. On creation, an 4308initial set of hcalls are enabled for in-kernel handling, which 4309consists of those hcalls for which in-kernel handlers were implemented 4310before this capability was implemented. If disabled, the kernel will 4311not to attempt to handle the hcall, but will always exit to userspace 4312to handle it. Note that it may not make sense to enable some and 4313disable others of a group of related hcalls, but KVM does not prevent 4314userspace from doing that. 4315 4316If the hcall number specified is not one that has an in-kernel 4317implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL 4318error. 4319 43207.2 KVM_CAP_S390_USER_SIGP 4321 4322Architectures: s390 4323Parameters: none 4324 4325This capability controls which SIGP orders will be handled completely in user 4326space. With this capability enabled, all fast orders will be handled completely 4327in the kernel: 4328- SENSE 4329- SENSE RUNNING 4330- EXTERNAL CALL 4331- EMERGENCY SIGNAL 4332- CONDITIONAL EMERGENCY SIGNAL 4333 4334All other orders will be handled completely in user space. 4335 4336Only privileged operation exceptions will be checked for in the kernel (or even 4337in the hardware prior to interception). If this capability is not enabled, the 4338old way of handling SIGP orders is used (partially in kernel and user space). 4339 43407.3 KVM_CAP_S390_VECTOR_REGISTERS 4341 4342Architectures: s390 4343Parameters: none 4344Returns: 0 on success, negative value on error 4345 4346Allows use of the vector registers introduced with z13 processor, and 4347provides for the synchronization between host and user space. Will 4348return -EINVAL if the machine does not support vectors. 4349 43507.4 KVM_CAP_S390_USER_STSI 4351 4352Architectures: s390 4353Parameters: none 4354 4355This capability allows post-handlers for the STSI instruction. After 4356initial handling in the kernel, KVM exits to user space with 4357KVM_EXIT_S390_STSI to allow user space to insert further data. 4358 4359Before exiting to userspace, kvm handlers should fill in s390_stsi field of 4360vcpu->run: 4361struct { 4362 __u64 addr; 4363 __u8 ar; 4364 __u8 reserved; 4365 __u8 fc; 4366 __u8 sel1; 4367 __u16 sel2; 4368} s390_stsi; 4369 4370@addr - guest address of STSI SYSIB 4371@fc - function code 4372@sel1 - selector 1 4373@sel2 - selector 2 4374@ar - access register number 4375 4376KVM handlers should exit to userspace with rc = -EREMOTE. 4377 43787.5 KVM_CAP_SPLIT_IRQCHIP 4379 4380Architectures: x86 4381Parameters: args[0] - number of routes reserved for userspace IOAPICs 4382Returns: 0 on success, -1 on error 4383 4384Create a local apic for each processor in the kernel. This can be used 4385instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the 4386IOAPIC and PIC (and also the PIT, even though this has to be enabled 4387separately). 4388 4389This capability also enables in kernel routing of interrupt requests; 4390when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are 4391used in the IRQ routing table. The first args[0] MSI routes are reserved 4392for the IOAPIC pins. Whenever the LAPIC receives an EOI for these routes, 4393a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace. 4394 4395Fails if VCPU has already been created, or if the irqchip is already in the 4396kernel (i.e. KVM_CREATE_IRQCHIP has already been called). 4397 43987.6 KVM_CAP_S390_RI 4399 4400Architectures: s390 4401Parameters: none 4402 4403Allows use of runtime-instrumentation introduced with zEC12 processor. 4404Will return -EINVAL if the machine does not support runtime-instrumentation. 4405Will return -EBUSY if a VCPU has already been created. 4406 44077.7 KVM_CAP_X2APIC_API 4408 4409Architectures: x86 4410Parameters: args[0] - features that should be enabled 4411Returns: 0 on success, -EINVAL when args[0] contains invalid features 4412 4413Valid feature flags in args[0] are 4414 4415#define KVM_X2APIC_API_USE_32BIT_IDS (1ULL << 0) 4416#define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK (1ULL << 1) 4417 4418Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of 4419KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC, 4420allowing the use of 32-bit APIC IDs. See KVM_CAP_X2APIC_API in their 4421respective sections. 4422 4423KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work 4424in logical mode or with more than 255 VCPUs. Otherwise, KVM treats 0xff 4425as a broadcast even in x2APIC mode in order to support physical x2APIC 4426without interrupt remapping. This is undesirable in logical mode, 4427where 0xff represents CPUs 0-7 in cluster 0. 4428 44297.8 KVM_CAP_S390_USER_INSTR0 4430 4431Architectures: s390 4432Parameters: none 4433 4434With this capability enabled, all illegal instructions 0x0000 (2 bytes) will 4435be intercepted and forwarded to user space. User space can use this 4436mechanism e.g. to realize 2-byte software breakpoints. The kernel will 4437not inject an operating exception for these instructions, user space has 4438to take care of that. 4439 4440This capability can be enabled dynamically even if VCPUs were already 4441created and are running. 4442 44437.9 KVM_CAP_S390_GS 4444 4445Architectures: s390 4446Parameters: none 4447Returns: 0 on success; -EINVAL if the machine does not support 4448 guarded storage; -EBUSY if a VCPU has already been created. 4449 4450Allows use of guarded storage for the KVM guest. 4451 44527.10 KVM_CAP_S390_AIS 4453 4454Architectures: s390 4455Parameters: none 4456 4457Allow use of adapter-interruption suppression. 4458Returns: 0 on success; -EBUSY if a VCPU has already been created. 4459 44607.11 KVM_CAP_PPC_SMT 4461 4462Architectures: ppc 4463Parameters: vsmt_mode, flags 4464 4465Enabling this capability on a VM provides userspace with a way to set 4466the desired virtual SMT mode (i.e. the number of virtual CPUs per 4467virtual core). The virtual SMT mode, vsmt_mode, must be a power of 2 4468between 1 and 8. On POWER8, vsmt_mode must also be no greater than 4469the number of threads per subcore for the host. Currently flags must 4470be 0. A successful call to enable this capability will result in 4471vsmt_mode being returned when the KVM_CAP_PPC_SMT capability is 4472subsequently queried for the VM. This capability is only supported by 4473HV KVM, and can only be set before any VCPUs have been created. 4474The KVM_CAP_PPC_SMT_POSSIBLE capability indicates which virtual SMT 4475modes are available. 4476 44777.12 KVM_CAP_PPC_FWNMI 4478 4479Architectures: ppc 4480Parameters: none 4481 4482With this capability a machine check exception in the guest address 4483space will cause KVM to exit the guest with NMI exit reason. This 4484enables QEMU to build error log and branch to guest kernel registered 4485machine check handling routine. Without this capability KVM will 4486branch to guests' 0x200 interrupt vector. 4487 44887.13 KVM_CAP_X86_DISABLE_EXITS 4489 4490Architectures: x86 4491Parameters: args[0] defines which exits are disabled 4492Returns: 0 on success, -EINVAL when args[0] contains invalid exits 4493 4494Valid bits in args[0] are 4495 4496#define KVM_X86_DISABLE_EXITS_MWAIT (1 << 0) 4497#define KVM_X86_DISABLE_EXITS_HLT (1 << 1) 4498 4499Enabling this capability on a VM provides userspace with a way to no 4500longer intercept some instructions for improved latency in some 4501workloads, and is suggested when vCPUs are associated to dedicated 4502physical CPUs. More bits can be added in the future; userspace can 4503just pass the KVM_CHECK_EXTENSION result to KVM_ENABLE_CAP to disable 4504all such vmexits. 4505 4506Do not enable KVM_FEATURE_PV_UNHALT if you disable HLT exits. 4507 45087.14 KVM_CAP_S390_HPAGE_1M 4509 4510Architectures: s390 4511Parameters: none 4512Returns: 0 on success, -EINVAL if hpage module parameter was not set 4513 or cmma is enabled, or the VM has the KVM_VM_S390_UCONTROL 4514 flag set 4515 4516With this capability the KVM support for memory backing with 1m pages 4517through hugetlbfs can be enabled for a VM. After the capability is 4518enabled, cmma can't be enabled anymore and pfmfi and the storage key 4519interpretation are disabled. If cmma has already been enabled or the 4520hpage module parameter is not set to 1, -EINVAL is returned. 4521 4522While it is generally possible to create a huge page backed VM without 4523this capability, the VM will not be able to run. 4524 45257.14 KVM_CAP_MSR_PLATFORM_INFO 4526 4527Architectures: x86 4528Parameters: args[0] whether feature should be enabled or not 4529 4530With this capability, a guest may read the MSR_PLATFORM_INFO MSR. Otherwise, 4531a #GP would be raised when the guest tries to access. Currently, this 4532capability does not enable write permissions of this MSR for the guest. 4533 45348. Other capabilities. 4535---------------------- 4536 4537This section lists capabilities that give information about other 4538features of the KVM implementation. 4539 45408.1 KVM_CAP_PPC_HWRNG 4541 4542Architectures: ppc 4543 4544This capability, if KVM_CHECK_EXTENSION indicates that it is 4545available, means that that the kernel has an implementation of the 4546H_RANDOM hypercall backed by a hardware random-number generator. 4547If present, the kernel H_RANDOM handler can be enabled for guest use 4548with the KVM_CAP_PPC_ENABLE_HCALL capability. 4549 45508.2 KVM_CAP_HYPERV_SYNIC 4551 4552Architectures: x86 4553This capability, if KVM_CHECK_EXTENSION indicates that it is 4554available, means that that the kernel has an implementation of the 4555Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is 4556used to support Windows Hyper-V based guest paravirt drivers(VMBus). 4557 4558In order to use SynIC, it has to be activated by setting this 4559capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this 4560will disable the use of APIC hardware virtualization even if supported 4561by the CPU, as it's incompatible with SynIC auto-EOI behavior. 4562 45638.3 KVM_CAP_PPC_RADIX_MMU 4564 4565Architectures: ppc 4566 4567This capability, if KVM_CHECK_EXTENSION indicates that it is 4568available, means that that the kernel can support guests using the 4569radix MMU defined in Power ISA V3.00 (as implemented in the POWER9 4570processor). 4571 45728.4 KVM_CAP_PPC_HASH_MMU_V3 4573 4574Architectures: ppc 4575 4576This capability, if KVM_CHECK_EXTENSION indicates that it is 4577available, means that that the kernel can support guests using the 4578hashed page table MMU defined in Power ISA V3.00 (as implemented in 4579the POWER9 processor), including in-memory segment tables. 4580 45818.5 KVM_CAP_MIPS_VZ 4582 4583Architectures: mips 4584 4585This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that 4586it is available, means that full hardware assisted virtualization capabilities 4587of the hardware are available for use through KVM. An appropriate 4588KVM_VM_MIPS_* type must be passed to KVM_CREATE_VM to create a VM which 4589utilises it. 4590 4591If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is 4592available, it means that the VM is using full hardware assisted virtualization 4593capabilities of the hardware. This is useful to check after creating a VM with 4594KVM_VM_MIPS_DEFAULT. 4595 4596The value returned by KVM_CHECK_EXTENSION should be compared against known 4597values (see below). All other values are reserved. This is to allow for the 4598possibility of other hardware assisted virtualization implementations which 4599may be incompatible with the MIPS VZ ASE. 4600 4601 0: The trap & emulate implementation is in use to run guest code in user 4602 mode. Guest virtual memory segments are rearranged to fit the guest in the 4603 user mode address space. 4604 4605 1: The MIPS VZ ASE is in use, providing full hardware assisted 4606 virtualization, including standard guest virtual memory segments. 4607 46088.6 KVM_CAP_MIPS_TE 4609 4610Architectures: mips 4611 4612This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that 4613it is available, means that the trap & emulate implementation is available to 4614run guest code in user mode, even if KVM_CAP_MIPS_VZ indicates that hardware 4615assisted virtualisation is also available. KVM_VM_MIPS_TE (0) must be passed 4616to KVM_CREATE_VM to create a VM which utilises it. 4617 4618If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is 4619available, it means that the VM is using trap & emulate. 4620 46218.7 KVM_CAP_MIPS_64BIT 4622 4623Architectures: mips 4624 4625This capability indicates the supported architecture type of the guest, i.e. the 4626supported register and address width. 4627 4628The values returned when this capability is checked by KVM_CHECK_EXTENSION on a 4629kvm VM handle correspond roughly to the CP0_Config.AT register field, and should 4630be checked specifically against known values (see below). All other values are 4631reserved. 4632 4633 0: MIPS32 or microMIPS32. 4634 Both registers and addresses are 32-bits wide. 4635 It will only be possible to run 32-bit guest code. 4636 4637 1: MIPS64 or microMIPS64 with access only to 32-bit compatibility segments. 4638 Registers are 64-bits wide, but addresses are 32-bits wide. 4639 64-bit guest code may run but cannot access MIPS64 memory segments. 4640 It will also be possible to run 32-bit guest code. 4641 4642 2: MIPS64 or microMIPS64 with access to all address segments. 4643 Both registers and addresses are 64-bits wide. 4644 It will be possible to run 64-bit or 32-bit guest code. 4645 46468.9 KVM_CAP_ARM_USER_IRQ 4647 4648Architectures: arm, arm64 4649This capability, if KVM_CHECK_EXTENSION indicates that it is available, means 4650that if userspace creates a VM without an in-kernel interrupt controller, it 4651will be notified of changes to the output level of in-kernel emulated devices, 4652which can generate virtual interrupts, presented to the VM. 4653For such VMs, on every return to userspace, the kernel 4654updates the vcpu's run->s.regs.device_irq_level field to represent the actual 4655output level of the device. 4656 4657Whenever kvm detects a change in the device output level, kvm guarantees at 4658least one return to userspace before running the VM. This exit could either 4659be a KVM_EXIT_INTR or any other exit event, like KVM_EXIT_MMIO. This way, 4660userspace can always sample the device output level and re-compute the state of 4661the userspace interrupt controller. Userspace should always check the state 4662of run->s.regs.device_irq_level on every kvm exit. 4663The value in run->s.regs.device_irq_level can represent both level and edge 4664triggered interrupt signals, depending on the device. Edge triggered interrupt 4665signals will exit to userspace with the bit in run->s.regs.device_irq_level 4666set exactly once per edge signal. 4667 4668The field run->s.regs.device_irq_level is available independent of 4669run->kvm_valid_regs or run->kvm_dirty_regs bits. 4670 4671If KVM_CAP_ARM_USER_IRQ is supported, the KVM_CHECK_EXTENSION ioctl returns a 4672number larger than 0 indicating the version of this capability is implemented 4673and thereby which bits in in run->s.regs.device_irq_level can signal values. 4674 4675Currently the following bits are defined for the device_irq_level bitmap: 4676 4677 KVM_CAP_ARM_USER_IRQ >= 1: 4678 4679 KVM_ARM_DEV_EL1_VTIMER - EL1 virtual timer 4680 KVM_ARM_DEV_EL1_PTIMER - EL1 physical timer 4681 KVM_ARM_DEV_PMU - ARM PMU overflow interrupt signal 4682 4683Future versions of kvm may implement additional events. These will get 4684indicated by returning a higher number from KVM_CHECK_EXTENSION and will be 4685listed above. 4686 46878.10 KVM_CAP_PPC_SMT_POSSIBLE 4688 4689Architectures: ppc 4690 4691Querying this capability returns a bitmap indicating the possible 4692virtual SMT modes that can be set using KVM_CAP_PPC_SMT. If bit N 4693(counting from the right) is set, then a virtual SMT mode of 2^N is 4694available. 4695 46968.11 KVM_CAP_HYPERV_SYNIC2 4697 4698Architectures: x86 4699 4700This capability enables a newer version of Hyper-V Synthetic interrupt 4701controller (SynIC). The only difference with KVM_CAP_HYPERV_SYNIC is that KVM 4702doesn't clear SynIC message and event flags pages when they are enabled by 4703writing to the respective MSRs. 4704 47058.12 KVM_CAP_HYPERV_VP_INDEX 4706 4707Architectures: x86 4708 4709This capability indicates that userspace can load HV_X64_MSR_VP_INDEX msr. Its 4710value is used to denote the target vcpu for a SynIC interrupt. For 4711compatibilty, KVM initializes this msr to KVM's internal vcpu index. When this 4712capability is absent, userspace can still query this msr's value. 4713 47148.13 KVM_CAP_S390_AIS_MIGRATION 4715 4716Architectures: s390 4717Parameters: none 4718 4719This capability indicates if the flic device will be able to get/set the 4720AIS states for migration via the KVM_DEV_FLIC_AISM_ALL attribute and allows 4721to discover this without having to create a flic device. 4722 47238.14 KVM_CAP_S390_PSW 4724 4725Architectures: s390 4726 4727This capability indicates that the PSW is exposed via the kvm_run structure. 4728 47298.15 KVM_CAP_S390_GMAP 4730 4731Architectures: s390 4732 4733This capability indicates that the user space memory used as guest mapping can 4734be anywhere in the user memory address space, as long as the memory slots are 4735aligned and sized to a segment (1MB) boundary. 4736 47378.16 KVM_CAP_S390_COW 4738 4739Architectures: s390 4740 4741This capability indicates that the user space memory used as guest mapping can 4742use copy-on-write semantics as well as dirty pages tracking via read-only page 4743tables. 4744 47458.17 KVM_CAP_S390_BPB 4746 4747Architectures: s390 4748 4749This capability indicates that kvm will implement the interfaces to handle 4750reset, migration and nested KVM for branch prediction blocking. The stfle 4751facility 82 should not be provided to the guest without this capability. 4752 47538.18 KVM_CAP_HYPERV_TLBFLUSH 4754 4755Architectures: x86 4756 4757This capability indicates that KVM supports paravirtualized Hyper-V TLB Flush 4758hypercalls: 4759HvFlushVirtualAddressSpace, HvFlushVirtualAddressSpaceEx, 4760HvFlushVirtualAddressList, HvFlushVirtualAddressListEx. 4761 47628.19 KVM_CAP_ARM_INJECT_SERROR_ESR 4763 4764Architectures: arm, arm64 4765 4766This capability indicates that userspace can specify (via the 4767KVM_SET_VCPU_EVENTS ioctl) the syndrome value reported to the guest when it 4768takes a virtual SError interrupt exception. 4769If KVM advertises this capability, userspace can only specify the ISS field for 4770the ESR syndrome. Other parts of the ESR, such as the EC are generated by the 4771CPU when the exception is taken. If this virtual SError is taken to EL1 using 4772AArch64, this value will be reported in the ISS field of ESR_ELx. 4773 4774See KVM_CAP_VCPU_EVENTS for more details. 4775
