1.. SPDX-License-Identifier: GPL-2.0 2 3=================================================================== 4The Definitive KVM (Kernel-based Virtual Machine) API Documentation 5=================================================================== 6 71. General description 8====================== 9 10The kvm API is a set of ioctls that are issued to control various aspects 11of a virtual machine. The ioctls belong to the following classes: 12 13 - System ioctls: These query and set global attributes which affect the 14 whole kvm subsystem. In addition a system ioctl is used to create 15 virtual machines. 16 17 - VM ioctls: These query and set attributes that affect an entire virtual 18 machine, for example memory layout. In addition a VM ioctl is used to 19 create virtual cpus (vcpus) and devices. 20 21 VM ioctls must be issued from the same process (address space) that was 22 used to create the VM. 23 24 - vcpu ioctls: These query and set attributes that control the operation 25 of a single virtual cpu. 26 27 vcpu ioctls should be issued from the same thread that was used to create 28 the vcpu, except for asynchronous vcpu ioctl that are marked as such in 29 the documentation. Otherwise, the first ioctl after switching threads 30 could see a performance impact. 31 32 - device ioctls: These query and set attributes that control the operation 33 of a single device. 34 35 device ioctls must be issued from the same process (address space) that 36 was used to create the VM. 37 382. File descriptors 39=================== 40 41The kvm API is centered around file descriptors. An initial 42open("/dev/kvm") obtains a handle to the kvm subsystem; this handle 43can be used to issue system ioctls. A KVM_CREATE_VM ioctl on this 44handle will create a VM file descriptor which can be used to issue VM 45ioctls. A KVM_CREATE_VCPU or KVM_CREATE_DEVICE ioctl on a VM fd will 46create a virtual cpu or device and return a file descriptor pointing to 47the new resource. Finally, ioctls on a vcpu or device fd can be used 48to control the vcpu or device. For vcpus, this includes the important 49task of actually running guest code. 50 51In general file descriptors can be migrated among processes by means 52of fork() and the SCM_RIGHTS facility of unix domain socket. These 53kinds of tricks are explicitly not supported by kvm. While they will 54not cause harm to the host, their actual behavior is not guaranteed by 55the API. See "General description" for details on the ioctl usage 56model that is supported by KVM. 57 58It is important to note that although VM ioctls may only be issued from 59the process that created the VM, a VM's lifecycle is associated with its 60file descriptor, not its creator (process). In other words, the VM and 61its resources, *including the associated address space*, are not freed 62until the last reference to the VM's file descriptor has been released. 63For example, if fork() is issued after ioctl(KVM_CREATE_VM), the VM will 64not be freed until both the parent (original) process and its child have 65put their references to the VM's file descriptor. 66 67Because a VM's resources are not freed until the last reference to its 68file descriptor is released, creating additional references to a VM 69via fork(), dup(), etc... without careful consideration is strongly 70discouraged and may have unwanted side effects, e.g. memory allocated 71by and on behalf of the VM's process may not be freed/unaccounted when 72the VM is shut down. 73 74 753. Extensions 76============= 77 78As of Linux 2.6.22, the KVM ABI has been stabilized: no backward 79incompatible change are allowed. However, there is an extension 80facility that allows backward-compatible extensions to the API to be 81queried and used. 82 83The extension mechanism is not based on the Linux version number. 84Instead, kvm defines extension identifiers and a facility to query 85whether a particular extension identifier is available. If it is, a 86set of ioctls is available for application use. 87 88 894. API description 90================== 91 92This section describes ioctls that can be used to control kvm guests. 93For each ioctl, the following information is provided along with a 94description: 95 96 Capability: 97 which KVM extension provides this ioctl. Can be 'basic', 98 which means that is will be provided by any kernel that supports 99 API version 12 (see section 4.1), a KVM_CAP_xyz constant, which 100 means availability needs to be checked with KVM_CHECK_EXTENSION 101 (see section 4.4), or 'none' which means that while not all kernels 102 support this ioctl, there's no capability bit to check its 103 availability: for kernels that don't support the ioctl, 104 the ioctl returns -ENOTTY. 105 106 Architectures: 107 which instruction set architectures provide this ioctl. 108 x86 includes both i386 and x86_64. 109 110 Type: 111 system, vm, or vcpu. 112 113 Parameters: 114 what parameters are accepted by the ioctl. 115 116 Returns: 117 the return value. General error numbers (EBADF, ENOMEM, EINVAL) 118 are not detailed, but errors with specific meanings are. 119 120 1214.1 KVM_GET_API_VERSION 122----------------------- 123 124:Capability: basic 125:Architectures: all 126:Type: system ioctl 127:Parameters: none 128:Returns: the constant KVM_API_VERSION (=12) 129 130This identifies the API version as the stable kvm API. It is not 131expected that this number will change. However, Linux 2.6.20 and 1322.6.21 report earlier versions; these are not documented and not 133supported. Applications should refuse to run if KVM_GET_API_VERSION 134returns a value other than 12. If this check passes, all ioctls 135described as 'basic' will be available. 136 137 1384.2 KVM_CREATE_VM 139----------------- 140 141:Capability: basic 142:Architectures: all 143:Type: system ioctl 144:Parameters: machine type identifier (KVM_VM_*) 145:Returns: a VM fd that can be used to control the new virtual machine. 146 147The new VM has no virtual cpus and no memory. 148You probably want to use 0 as machine type. 149 150In order to create user controlled virtual machines on S390, check 151KVM_CAP_S390_UCONTROL and use the flag KVM_VM_S390_UCONTROL as 152privileged user (CAP_SYS_ADMIN). 153 154To use hardware assisted virtualization on MIPS (VZ ASE) rather than 155the default trap & emulate implementation (which changes the virtual 156memory layout to fit in user mode), check KVM_CAP_MIPS_VZ and use the 157flag KVM_VM_MIPS_VZ. 158 159 160On arm64, the physical address size for a VM (IPA Size limit) is limited 161to 40bits by default. The limit can be configured if the host supports the 162extension KVM_CAP_ARM_VM_IPA_SIZE. When supported, use 163KVM_VM_TYPE_ARM_IPA_SIZE(IPA_Bits) to set the size in the machine type 164identifier, where IPA_Bits is the maximum width of any physical 165address used by the VM. The IPA_Bits is encoded in bits[7-0] of the 166machine type identifier. 167 168e.g, to configure a guest to use 48bit physical address size:: 169 170 vm_fd = ioctl(dev_fd, KVM_CREATE_VM, KVM_VM_TYPE_ARM_IPA_SIZE(48)); 171 172The requested size (IPA_Bits) must be: 173 174 == ========================================================= 175 0 Implies default size, 40bits (for backward compatibility) 176 N Implies N bits, where N is a positive integer such that, 177 32 <= N <= Host_IPA_Limit 178 == ========================================================= 179 180Host_IPA_Limit is the maximum possible value for IPA_Bits on the host and 181is dependent on the CPU capability and the kernel configuration. The limit can 182be retrieved using KVM_CAP_ARM_VM_IPA_SIZE of the KVM_CHECK_EXTENSION 183ioctl() at run-time. 184 185Creation of the VM will fail if the requested IPA size (whether it is 186implicit or explicit) is unsupported on the host. 187 188Please note that configuring the IPA size does not affect the capability 189exposed by the guest CPUs in ID_AA64MMFR0_EL1[PARange]. It only affects 190size of the address translated by the stage2 level (guest physical to 191host physical address translations). 192 193 1944.3 KVM_GET_MSR_INDEX_LIST, KVM_GET_MSR_FEATURE_INDEX_LIST 195---------------------------------------------------------- 196 197:Capability: basic, KVM_CAP_GET_MSR_FEATURES for KVM_GET_MSR_FEATURE_INDEX_LIST 198:Architectures: x86 199:Type: system ioctl 200:Parameters: struct kvm_msr_list (in/out) 201:Returns: 0 on success; -1 on error 202 203Errors: 204 205 ====== ============================================================ 206 EFAULT the msr index list cannot be read from or written to 207 E2BIG the msr index list is too big to fit in the array specified by 208 the user. 209 ====== ============================================================ 210 211:: 212 213 struct kvm_msr_list { 214 __u32 nmsrs; /* number of msrs in entries */ 215 __u32 indices[0]; 216 }; 217 218The user fills in the size of the indices array in nmsrs, and in return 219kvm adjusts nmsrs to reflect the actual number of msrs and fills in the 220indices array with their numbers. 221 222KVM_GET_MSR_INDEX_LIST returns the guest msrs that are supported. The list 223varies by kvm version and host processor, but does not change otherwise. 224 225Note: if kvm indicates supports MCE (KVM_CAP_MCE), then the MCE bank MSRs are 226not returned in the MSR list, as different vcpus can have a different number 227of banks, as set via the KVM_X86_SETUP_MCE ioctl. 228 229KVM_GET_MSR_FEATURE_INDEX_LIST returns the list of MSRs that can be passed 230to the KVM_GET_MSRS system ioctl. This lets userspace probe host capabilities 231and processor features that are exposed via MSRs (e.g., VMX capabilities). 232This list also varies by kvm version and host processor, but does not change 233otherwise. 234 235 2364.4 KVM_CHECK_EXTENSION 237----------------------- 238 239:Capability: basic, KVM_CAP_CHECK_EXTENSION_VM for vm ioctl 240:Architectures: all 241:Type: system ioctl, vm ioctl 242:Parameters: extension identifier (KVM_CAP_*) 243:Returns: 0 if unsupported; 1 (or some other positive integer) if supported 244 245The API allows the application to query about extensions to the core 246kvm API. Userspace passes an extension identifier (an integer) and 247receives an integer that describes the extension availability. 248Generally 0 means no and 1 means yes, but some extensions may report 249additional information in the integer return value. 250 251Based on their initialization different VMs may have different capabilities. 252It is thus encouraged to use the vm ioctl to query for capabilities (available 253with KVM_CAP_CHECK_EXTENSION_VM on the vm fd) 254 2554.5 KVM_GET_VCPU_MMAP_SIZE 256-------------------------- 257 258:Capability: basic 259:Architectures: all 260:Type: system ioctl 261:Parameters: none 262:Returns: size of vcpu mmap area, in bytes 263 264The KVM_RUN ioctl (cf.) communicates with userspace via a shared 265memory region. This ioctl returns the size of that region. See the 266KVM_RUN documentation for details. 267 268Besides the size of the KVM_RUN communication region, other areas of 269the VCPU file descriptor can be mmap-ed, including: 270 271- if KVM_CAP_COALESCED_MMIO is available, a page at 272 KVM_COALESCED_MMIO_PAGE_OFFSET * PAGE_SIZE; for historical reasons, 273 this page is included in the result of KVM_GET_VCPU_MMAP_SIZE. 274 KVM_CAP_COALESCED_MMIO is not documented yet. 275 276- if KVM_CAP_DIRTY_LOG_RING is available, a number of pages at 277 KVM_DIRTY_LOG_PAGE_OFFSET * PAGE_SIZE. For more information on 278 KVM_CAP_DIRTY_LOG_RING, see section 8.3. 279 280 2814.6 KVM_SET_MEMORY_REGION 282------------------------- 283 284:Capability: basic 285:Architectures: all 286:Type: vm ioctl 287:Parameters: struct kvm_memory_region (in) 288:Returns: 0 on success, -1 on error 289 290This ioctl is obsolete and has been removed. 291 292 2934.7 KVM_CREATE_VCPU 294------------------- 295 296:Capability: basic 297:Architectures: all 298:Type: vm ioctl 299:Parameters: vcpu id (apic id on x86) 300:Returns: vcpu fd on success, -1 on error 301 302This API adds a vcpu to a virtual machine. No more than max_vcpus may be added. 303The vcpu id is an integer in the range [0, max_vcpu_id). 304 305The recommended max_vcpus value can be retrieved using the KVM_CAP_NR_VCPUS of 306the KVM_CHECK_EXTENSION ioctl() at run-time. 307The maximum possible value for max_vcpus can be retrieved using the 308KVM_CAP_MAX_VCPUS of the KVM_CHECK_EXTENSION ioctl() at run-time. 309 310If the KVM_CAP_NR_VCPUS does not exist, you should assume that max_vcpus is 4 311cpus max. 312If the KVM_CAP_MAX_VCPUS does not exist, you should assume that max_vcpus is 313same as the value returned from KVM_CAP_NR_VCPUS. 314 315The maximum possible value for max_vcpu_id can be retrieved using the 316KVM_CAP_MAX_VCPU_ID of the KVM_CHECK_EXTENSION ioctl() at run-time. 317 318If the KVM_CAP_MAX_VCPU_ID does not exist, you should assume that max_vcpu_id 319is the same as the value returned from KVM_CAP_MAX_VCPUS. 320 321On powerpc using book3s_hv mode, the vcpus are mapped onto virtual 322threads in one or more virtual CPU cores. (This is because the 323hardware requires all the hardware threads in a CPU core to be in the 324same partition.) The KVM_CAP_PPC_SMT capability indicates the number 325of vcpus per virtual core (vcore). The vcore id is obtained by 326dividing the vcpu id by the number of vcpus per vcore. The vcpus in a 327given vcore will always be in the same physical core as each other 328(though that might be a different physical core from time to time). 329Userspace can control the threading (SMT) mode of the guest by its 330allocation of vcpu ids. For example, if userspace wants 331single-threaded guest vcpus, it should make all vcpu ids be a multiple 332of the number of vcpus per vcore. 333 334For virtual cpus that have been created with S390 user controlled virtual 335machines, the resulting vcpu fd can be memory mapped at page offset 336KVM_S390_SIE_PAGE_OFFSET in order to obtain a memory map of the virtual 337cpu's hardware control block. 338 339 3404.8 KVM_GET_DIRTY_LOG (vm ioctl) 341-------------------------------- 342 343:Capability: basic 344:Architectures: all 345:Type: vm ioctl 346:Parameters: struct kvm_dirty_log (in/out) 347:Returns: 0 on success, -1 on error 348 349:: 350 351 /* for KVM_GET_DIRTY_LOG */ 352 struct kvm_dirty_log { 353 __u32 slot; 354 __u32 padding; 355 union { 356 void __user *dirty_bitmap; /* one bit per page */ 357 __u64 padding; 358 }; 359 }; 360 361Given a memory slot, return a bitmap containing any pages dirtied 362since the last call to this ioctl. Bit 0 is the first page in the 363memory slot. Ensure the entire structure is cleared to avoid padding 364issues. 365 366If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies 367the address space for which you want to return the dirty bitmap. See 368KVM_SET_USER_MEMORY_REGION for details on the usage of slot field. 369 370The bits in the dirty bitmap are cleared before the ioctl returns, unless 371KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is enabled. For more information, 372see the description of the capability. 373 3744.9 KVM_SET_MEMORY_ALIAS 375------------------------ 376 377:Capability: basic 378:Architectures: x86 379:Type: vm ioctl 380:Parameters: struct kvm_memory_alias (in) 381:Returns: 0 (success), -1 (error) 382 383This ioctl is obsolete and has been removed. 384 385 3864.10 KVM_RUN 387------------ 388 389:Capability: basic 390:Architectures: all 391:Type: vcpu ioctl 392:Parameters: none 393:Returns: 0 on success, -1 on error 394 395Errors: 396 397 ======= ============================================================== 398 EINTR an unmasked signal is pending 399 ENOEXEC the vcpu hasn't been initialized or the guest tried to execute 400 instructions from device memory (arm64) 401 ENOSYS data abort outside memslots with no syndrome info and 402 KVM_CAP_ARM_NISV_TO_USER not enabled (arm64) 403 EPERM SVE feature set but not finalized (arm64) 404 ======= ============================================================== 405 406This ioctl is used to run a guest virtual cpu. While there are no 407explicit parameters, there is an implicit parameter block that can be 408obtained by mmap()ing the vcpu fd at offset 0, with the size given by 409KVM_GET_VCPU_MMAP_SIZE. The parameter block is formatted as a 'struct 410kvm_run' (see below). 411 412 4134.11 KVM_GET_REGS 414----------------- 415 416:Capability: basic 417:Architectures: all except ARM, arm64 418:Type: vcpu ioctl 419:Parameters: struct kvm_regs (out) 420:Returns: 0 on success, -1 on error 421 422Reads the general purpose registers from the vcpu. 423 424:: 425 426 /* x86 */ 427 struct kvm_regs { 428 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */ 429 __u64 rax, rbx, rcx, rdx; 430 __u64 rsi, rdi, rsp, rbp; 431 __u64 r8, r9, r10, r11; 432 __u64 r12, r13, r14, r15; 433 __u64 rip, rflags; 434 }; 435 436 /* mips */ 437 struct kvm_regs { 438 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */ 439 __u64 gpr[32]; 440 __u64 hi; 441 __u64 lo; 442 __u64 pc; 443 }; 444 445 4464.12 KVM_SET_REGS 447----------------- 448 449:Capability: basic 450:Architectures: all except ARM, arm64 451:Type: vcpu ioctl 452:Parameters: struct kvm_regs (in) 453:Returns: 0 on success, -1 on error 454 455Writes the general purpose registers into the vcpu. 456 457See KVM_GET_REGS for the data structure. 458 459 4604.13 KVM_GET_SREGS 461------------------ 462 463:Capability: basic 464:Architectures: x86, ppc 465:Type: vcpu ioctl 466:Parameters: struct kvm_sregs (out) 467:Returns: 0 on success, -1 on error 468 469Reads special registers from the vcpu. 470 471:: 472 473 /* x86 */ 474 struct kvm_sregs { 475 struct kvm_segment cs, ds, es, fs, gs, ss; 476 struct kvm_segment tr, ldt; 477 struct kvm_dtable gdt, idt; 478 __u64 cr0, cr2, cr3, cr4, cr8; 479 __u64 efer; 480 __u64 apic_base; 481 __u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64]; 482 }; 483 484 /* ppc -- see arch/powerpc/include/uapi/asm/kvm.h */ 485 486interrupt_bitmap is a bitmap of pending external interrupts. At most 487one bit may be set. This interrupt has been acknowledged by the APIC 488but not yet injected into the cpu core. 489 490 4914.14 KVM_SET_SREGS 492------------------ 493 494:Capability: basic 495:Architectures: x86, ppc 496:Type: vcpu ioctl 497:Parameters: struct kvm_sregs (in) 498:Returns: 0 on success, -1 on error 499 500Writes special registers into the vcpu. See KVM_GET_SREGS for the 501data structures. 502 503 5044.15 KVM_TRANSLATE 505------------------ 506 507:Capability: basic 508:Architectures: x86 509:Type: vcpu ioctl 510:Parameters: struct kvm_translation (in/out) 511:Returns: 0 on success, -1 on error 512 513Translates a virtual address according to the vcpu's current address 514translation mode. 515 516:: 517 518 struct kvm_translation { 519 /* in */ 520 __u64 linear_address; 521 522 /* out */ 523 __u64 physical_address; 524 __u8 valid; 525 __u8 writeable; 526 __u8 usermode; 527 __u8 pad[5]; 528 }; 529 530 5314.16 KVM_INTERRUPT 532------------------ 533 534:Capability: basic 535:Architectures: x86, ppc, mips, riscv 536:Type: vcpu ioctl 537:Parameters: struct kvm_interrupt (in) 538:Returns: 0 on success, negative on failure. 539 540Queues a hardware interrupt vector to be injected. 541 542:: 543 544 /* for KVM_INTERRUPT */ 545 struct kvm_interrupt { 546 /* in */ 547 __u32 irq; 548 }; 549 550X86: 551^^^^ 552 553:Returns: 554 555 ========= =================================== 556 0 on success, 557 -EEXIST if an interrupt is already enqueued 558 -EINVAL the irq number is invalid 559 -ENXIO if the PIC is in the kernel 560 -EFAULT if the pointer is invalid 561 ========= =================================== 562 563Note 'irq' is an interrupt vector, not an interrupt pin or line. This 564ioctl is useful if the in-kernel PIC is not used. 565 566PPC: 567^^^^ 568 569Queues an external interrupt to be injected. This ioctl is overleaded 570with 3 different irq values: 571 572a) KVM_INTERRUPT_SET 573 574 This injects an edge type external interrupt into the guest once it's ready 575 to receive interrupts. When injected, the interrupt is done. 576 577b) KVM_INTERRUPT_UNSET 578 579 This unsets any pending interrupt. 580 581 Only available with KVM_CAP_PPC_UNSET_IRQ. 582 583c) KVM_INTERRUPT_SET_LEVEL 584 585 This injects a level type external interrupt into the guest context. The 586 interrupt stays pending until a specific ioctl with KVM_INTERRUPT_UNSET 587 is triggered. 588 589 Only available with KVM_CAP_PPC_IRQ_LEVEL. 590 591Note that any value for 'irq' other than the ones stated above is invalid 592and incurs unexpected behavior. 593 594This is an asynchronous vcpu ioctl and can be invoked from any thread. 595 596MIPS: 597^^^^^ 598 599Queues an external interrupt to be injected into the virtual CPU. A negative 600interrupt number dequeues the interrupt. 601 602This is an asynchronous vcpu ioctl and can be invoked from any thread. 603 604RISC-V: 605^^^^^^^ 606 607Queues an external interrupt to be injected into the virutal CPU. This ioctl 608is overloaded with 2 different irq values: 609 610a) KVM_INTERRUPT_SET 611 612 This sets external interrupt for a virtual CPU and it will receive 613 once it is ready. 614 615b) KVM_INTERRUPT_UNSET 616 617 This clears pending external interrupt for a virtual CPU. 618 619This is an asynchronous vcpu ioctl and can be invoked from any thread. 620 621 6224.17 KVM_DEBUG_GUEST 623-------------------- 624 625:Capability: basic 626:Architectures: none 627:Type: vcpu ioctl 628:Parameters: none) 629:Returns: -1 on error 630 631Support for this has been removed. Use KVM_SET_GUEST_DEBUG instead. 632 633 6344.18 KVM_GET_MSRS 635----------------- 636 637:Capability: basic (vcpu), KVM_CAP_GET_MSR_FEATURES (system) 638:Architectures: x86 639:Type: system ioctl, vcpu ioctl 640:Parameters: struct kvm_msrs (in/out) 641:Returns: number of msrs successfully returned; 642 -1 on error 643 644When used as a system ioctl: 645Reads the values of MSR-based features that are available for the VM. This 646is similar to KVM_GET_SUPPORTED_CPUID, but it returns MSR indices and values. 647The list of msr-based features can be obtained using KVM_GET_MSR_FEATURE_INDEX_LIST 648in a system ioctl. 649 650When used as a vcpu ioctl: 651Reads model-specific registers from the vcpu. Supported msr indices can 652be obtained using KVM_GET_MSR_INDEX_LIST in a system ioctl. 653 654:: 655 656 struct kvm_msrs { 657 __u32 nmsrs; /* number of msrs in entries */ 658 __u32 pad; 659 660 struct kvm_msr_entry entries[0]; 661 }; 662 663 struct kvm_msr_entry { 664 __u32 index; 665 __u32 reserved; 666 __u64 data; 667 }; 668 669Application code should set the 'nmsrs' member (which indicates the 670size of the entries array) and the 'index' member of each array entry. 671kvm will fill in the 'data' member. 672 673 6744.19 KVM_SET_MSRS 675----------------- 676 677:Capability: basic 678:Architectures: x86 679:Type: vcpu ioctl 680:Parameters: struct kvm_msrs (in) 681:Returns: number of msrs successfully set (see below), -1 on error 682 683Writes model-specific registers to the vcpu. See KVM_GET_MSRS for the 684data structures. 685 686Application code should set the 'nmsrs' member (which indicates the 687size of the entries array), and the 'index' and 'data' members of each 688array entry. 689 690It tries to set the MSRs in array entries[] one by one. If setting an MSR 691fails, e.g., due to setting reserved bits, the MSR isn't supported/emulated 692by KVM, etc..., it stops processing the MSR list and returns the number of 693MSRs that have been set successfully. 694 695 6964.20 KVM_SET_CPUID 697------------------ 698 699:Capability: basic 700:Architectures: x86 701:Type: vcpu ioctl 702:Parameters: struct kvm_cpuid (in) 703:Returns: 0 on success, -1 on error 704 705Defines the vcpu responses to the cpuid instruction. Applications 706should use the KVM_SET_CPUID2 ioctl if available. 707 708Caveat emptor: 709 - If this IOCTL fails, KVM gives no guarantees that previous valid CPUID 710 configuration (if there is) is not corrupted. Userspace can get a copy 711 of the resulting CPUID configuration through KVM_GET_CPUID2 in case. 712 - Using KVM_SET_CPUID{,2} after KVM_RUN, i.e. changing the guest vCPU model 713 after running the guest, may cause guest instability. 714 - Using heterogeneous CPUID configurations, modulo APIC IDs, topology, etc... 715 may cause guest instability. 716 717:: 718 719 struct kvm_cpuid_entry { 720 __u32 function; 721 __u32 eax; 722 __u32 ebx; 723 __u32 ecx; 724 __u32 edx; 725 __u32 padding; 726 }; 727 728 /* for KVM_SET_CPUID */ 729 struct kvm_cpuid { 730 __u32 nent; 731 __u32 padding; 732 struct kvm_cpuid_entry entries[0]; 733 }; 734 735 7364.21 KVM_SET_SIGNAL_MASK 737------------------------ 738 739:Capability: basic 740:Architectures: all 741:Type: vcpu ioctl 742:Parameters: struct kvm_signal_mask (in) 743:Returns: 0 on success, -1 on error 744 745Defines which signals are blocked during execution of KVM_RUN. This 746signal mask temporarily overrides the threads signal mask. Any 747unblocked signal received (except SIGKILL and SIGSTOP, which retain 748their traditional behaviour) will cause KVM_RUN to return with -EINTR. 749 750Note the signal will only be delivered if not blocked by the original 751signal mask. 752 753:: 754 755 /* for KVM_SET_SIGNAL_MASK */ 756 struct kvm_signal_mask { 757 __u32 len; 758 __u8 sigset[0]; 759 }; 760 761 7624.22 KVM_GET_FPU 763---------------- 764 765:Capability: basic 766:Architectures: x86 767:Type: vcpu ioctl 768:Parameters: struct kvm_fpu (out) 769:Returns: 0 on success, -1 on error 770 771Reads the floating point state from the vcpu. 772 773:: 774 775 /* for KVM_GET_FPU and KVM_SET_FPU */ 776 struct kvm_fpu { 777 __u8 fpr[8][16]; 778 __u16 fcw; 779 __u16 fsw; 780 __u8 ftwx; /* in fxsave format */ 781 __u8 pad1; 782 __u16 last_opcode; 783 __u64 last_ip; 784 __u64 last_dp; 785 __u8 xmm[16][16]; 786 __u32 mxcsr; 787 __u32 pad2; 788 }; 789 790 7914.23 KVM_SET_FPU 792---------------- 793 794:Capability: basic 795:Architectures: x86 796:Type: vcpu ioctl 797:Parameters: struct kvm_fpu (in) 798:Returns: 0 on success, -1 on error 799 800Writes the floating point state to the vcpu. 801 802:: 803 804 /* for KVM_GET_FPU and KVM_SET_FPU */ 805 struct kvm_fpu { 806 __u8 fpr[8][16]; 807 __u16 fcw; 808 __u16 fsw; 809 __u8 ftwx; /* in fxsave format */ 810 __u8 pad1; 811 __u16 last_opcode; 812 __u64 last_ip; 813 __u64 last_dp; 814 __u8 xmm[16][16]; 815 __u32 mxcsr; 816 __u32 pad2; 817 }; 818 819 8204.24 KVM_CREATE_IRQCHIP 821----------------------- 822 823:Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQCHIP (s390) 824:Architectures: x86, ARM, arm64, s390 825:Type: vm ioctl 826:Parameters: none 827:Returns: 0 on success, -1 on error 828 829Creates an interrupt controller model in the kernel. 830On x86, creates a virtual ioapic, a virtual PIC (two PICs, nested), and sets up 831future vcpus to have a local APIC. IRQ routing for GSIs 0-15 is set to both 832PIC and IOAPIC; GSI 16-23 only go to the IOAPIC. 833On ARM/arm64, a GICv2 is created. Any other GIC versions require the usage of 834KVM_CREATE_DEVICE, which also supports creating a GICv2. Using 835KVM_CREATE_DEVICE is preferred over KVM_CREATE_IRQCHIP for GICv2. 836On s390, a dummy irq routing table is created. 837 838Note that on s390 the KVM_CAP_S390_IRQCHIP vm capability needs to be enabled 839before KVM_CREATE_IRQCHIP can be used. 840 841 8424.25 KVM_IRQ_LINE 843----------------- 844 845:Capability: KVM_CAP_IRQCHIP 846:Architectures: x86, arm, arm64 847:Type: vm ioctl 848:Parameters: struct kvm_irq_level 849:Returns: 0 on success, -1 on error 850 851Sets the level of a GSI input to the interrupt controller model in the kernel. 852On some architectures it is required that an interrupt controller model has 853been previously created with KVM_CREATE_IRQCHIP. Note that edge-triggered 854interrupts require the level to be set to 1 and then back to 0. 855 856On real hardware, interrupt pins can be active-low or active-high. This 857does not matter for the level field of struct kvm_irq_level: 1 always 858means active (asserted), 0 means inactive (deasserted). 859 860x86 allows the operating system to program the interrupt polarity 861(active-low/active-high) for level-triggered interrupts, and KVM used 862to consider the polarity. However, due to bitrot in the handling of 863active-low interrupts, the above convention is now valid on x86 too. 864This is signaled by KVM_CAP_X86_IOAPIC_POLARITY_IGNORED. Userspace 865should not present interrupts to the guest as active-low unless this 866capability is present (or unless it is not using the in-kernel irqchip, 867of course). 868 869 870ARM/arm64 can signal an interrupt either at the CPU level, or at the 871in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to 872use PPIs designated for specific cpus. The irq field is interpreted 873like this:: 874 875 bits: | 31 ... 28 | 27 ... 24 | 23 ... 16 | 15 ... 0 | 876 field: | vcpu2_index | irq_type | vcpu_index | irq_id | 877 878The irq_type field has the following values: 879 880- irq_type[0]: 881 out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ 882- irq_type[1]: 883 in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.) 884 (the vcpu_index field is ignored) 885- irq_type[2]: 886 in-kernel GIC: PPI, irq_id between 16 and 31 (incl.) 887 888(The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs) 889 890In both cases, level is used to assert/deassert the line. 891 892When KVM_CAP_ARM_IRQ_LINE_LAYOUT_2 is supported, the target vcpu is 893identified as (256 * vcpu2_index + vcpu_index). Otherwise, vcpu2_index 894must be zero. 895 896Note that on arm/arm64, the KVM_CAP_IRQCHIP capability only conditions 897injection of interrupts for the in-kernel irqchip. KVM_IRQ_LINE can always 898be used for a userspace interrupt controller. 899 900:: 901 902 struct kvm_irq_level { 903 union { 904 __u32 irq; /* GSI */ 905 __s32 status; /* not used for KVM_IRQ_LEVEL */ 906 }; 907 __u32 level; /* 0 or 1 */ 908 }; 909 910 9114.26 KVM_GET_IRQCHIP 912-------------------- 913 914:Capability: KVM_CAP_IRQCHIP 915:Architectures: x86 916:Type: vm ioctl 917:Parameters: struct kvm_irqchip (in/out) 918:Returns: 0 on success, -1 on error 919 920Reads the state of a kernel interrupt controller created with 921KVM_CREATE_IRQCHIP into a buffer provided by the caller. 922 923:: 924 925 struct kvm_irqchip { 926 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */ 927 __u32 pad; 928 union { 929 char dummy[512]; /* reserving space */ 930 struct kvm_pic_state pic; 931 struct kvm_ioapic_state ioapic; 932 } chip; 933 }; 934 935 9364.27 KVM_SET_IRQCHIP 937-------------------- 938 939:Capability: KVM_CAP_IRQCHIP 940:Architectures: x86 941:Type: vm ioctl 942:Parameters: struct kvm_irqchip (in) 943:Returns: 0 on success, -1 on error 944 945Sets the state of a kernel interrupt controller created with 946KVM_CREATE_IRQCHIP from a buffer provided by the caller. 947 948:: 949 950 struct kvm_irqchip { 951 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */ 952 __u32 pad; 953 union { 954 char dummy[512]; /* reserving space */ 955 struct kvm_pic_state pic; 956 struct kvm_ioapic_state ioapic; 957 } chip; 958 }; 959 960 9614.28 KVM_XEN_HVM_CONFIG 962----------------------- 963 964:Capability: KVM_CAP_XEN_HVM 965:Architectures: x86 966:Type: vm ioctl 967:Parameters: struct kvm_xen_hvm_config (in) 968:Returns: 0 on success, -1 on error 969 970Sets the MSR that the Xen HVM guest uses to initialize its hypercall 971page, and provides the starting address and size of the hypercall 972blobs in userspace. When the guest writes the MSR, kvm copies one 973page of a blob (32- or 64-bit, depending on the vcpu mode) to guest 974memory. 975 976:: 977 978 struct kvm_xen_hvm_config { 979 __u32 flags; 980 __u32 msr; 981 __u64 blob_addr_32; 982 __u64 blob_addr_64; 983 __u8 blob_size_32; 984 __u8 blob_size_64; 985 __u8 pad2[30]; 986 }; 987 988If the KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL flag is returned from the 989KVM_CAP_XEN_HVM check, it may be set in the flags field of this ioctl. 990This requests KVM to generate the contents of the hypercall page 991automatically; hypercalls will be intercepted and passed to userspace 992through KVM_EXIT_XEN. In this case, all of the blob size and address 993fields must be zero. 994 995No other flags are currently valid in the struct kvm_xen_hvm_config. 996 9974.29 KVM_GET_CLOCK 998------------------ 999 1000:Capability: KVM_CAP_ADJUST_CLOCK 1001:Architectures: x86 1002:Type: vm ioctl 1003:Parameters: struct kvm_clock_data (out) 1004:Returns: 0 on success, -1 on error 1005 1006Gets the current timestamp of kvmclock as seen by the current guest. In 1007conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios 1008such as migration. 1009 1010When KVM_CAP_ADJUST_CLOCK is passed to KVM_CHECK_EXTENSION, it returns the 1011set of bits that KVM can return in struct kvm_clock_data's flag member. 1012 1013The following flags are defined: 1014 1015KVM_CLOCK_TSC_STABLE 1016 If set, the returned value is the exact kvmclock 1017 value seen by all VCPUs at the instant when KVM_GET_CLOCK was called. 1018 If clear, the returned value is simply CLOCK_MONOTONIC plus a constant 1019 offset; the offset can be modified with KVM_SET_CLOCK. KVM will try 1020 to make all VCPUs follow this clock, but the exact value read by each 1021 VCPU could differ, because the host TSC is not stable. 1022 1023KVM_CLOCK_REALTIME 1024 If set, the `realtime` field in the kvm_clock_data 1025 structure is populated with the value of the host's real time 1026 clocksource at the instant when KVM_GET_CLOCK was called. If clear, 1027 the `realtime` field does not contain a value. 1028 1029KVM_CLOCK_HOST_TSC 1030 If set, the `host_tsc` field in the kvm_clock_data 1031 structure is populated with the value of the host's timestamp counter (TSC) 1032 at the instant when KVM_GET_CLOCK was called. If clear, the `host_tsc` field 1033 does not contain a value. 1034 1035:: 1036 1037 struct kvm_clock_data { 1038 __u64 clock; /* kvmclock current value */ 1039 __u32 flags; 1040 __u32 pad0; 1041 __u64 realtime; 1042 __u64 host_tsc; 1043 __u32 pad[4]; 1044 }; 1045 1046 10474.30 KVM_SET_CLOCK 1048------------------ 1049 1050:Capability: KVM_CAP_ADJUST_CLOCK 1051:Architectures: x86 1052:Type: vm ioctl 1053:Parameters: struct kvm_clock_data (in) 1054:Returns: 0 on success, -1 on error 1055 1056Sets the current timestamp of kvmclock to the value specified in its parameter. 1057In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios 1058such as migration. 1059 1060The following flags can be passed: 1061 1062KVM_CLOCK_REALTIME 1063 If set, KVM will compare the value of the `realtime` field 1064 with the value of the host's real time clocksource at the instant when 1065 KVM_SET_CLOCK was called. The difference in elapsed time is added to the final 1066 kvmclock value that will be provided to guests. 1067 1068Other flags returned by ``KVM_GET_CLOCK`` are accepted but ignored. 1069 1070:: 1071 1072 struct kvm_clock_data { 1073 __u64 clock; /* kvmclock current value */ 1074 __u32 flags; 1075 __u32 pad0; 1076 __u64 realtime; 1077 __u64 host_tsc; 1078 __u32 pad[4]; 1079 }; 1080 1081 10824.31 KVM_GET_VCPU_EVENTS 1083------------------------ 1084 1085:Capability: KVM_CAP_VCPU_EVENTS 1086:Extended by: KVM_CAP_INTR_SHADOW 1087:Architectures: x86, arm, arm64 1088:Type: vcpu ioctl 1089:Parameters: struct kvm_vcpu_event (out) 1090:Returns: 0 on success, -1 on error 1091 1092X86: 1093^^^^ 1094 1095Gets currently pending exceptions, interrupts, and NMIs as well as related 1096states of the vcpu. 1097 1098:: 1099 1100 struct kvm_vcpu_events { 1101 struct { 1102 __u8 injected; 1103 __u8 nr; 1104 __u8 has_error_code; 1105 __u8 pending; 1106 __u32 error_code; 1107 } exception; 1108 struct { 1109 __u8 injected; 1110 __u8 nr; 1111 __u8 soft; 1112 __u8 shadow; 1113 } interrupt; 1114 struct { 1115 __u8 injected; 1116 __u8 pending; 1117 __u8 masked; 1118 __u8 pad; 1119 } nmi; 1120 __u32 sipi_vector; 1121 __u32 flags; 1122 struct { 1123 __u8 smm; 1124 __u8 pending; 1125 __u8 smm_inside_nmi; 1126 __u8 latched_init; 1127 } smi; 1128 __u8 reserved[27]; 1129 __u8 exception_has_payload; 1130 __u64 exception_payload; 1131 }; 1132 1133The following bits are defined in the flags field: 1134 1135- KVM_VCPUEVENT_VALID_SHADOW may be set to signal that 1136 interrupt.shadow contains a valid state. 1137 1138- KVM_VCPUEVENT_VALID_SMM may be set to signal that smi contains a 1139 valid state. 1140 1141- KVM_VCPUEVENT_VALID_PAYLOAD may be set to signal that the 1142 exception_has_payload, exception_payload, and exception.pending 1143 fields contain a valid state. This bit will be set whenever 1144 KVM_CAP_EXCEPTION_PAYLOAD is enabled. 1145 1146ARM/ARM64: 1147^^^^^^^^^^ 1148 1149If the guest accesses a device that is being emulated by the host kernel in 1150such a way that a real device would generate a physical SError, KVM may make 1151a virtual SError pending for that VCPU. This system error interrupt remains 1152pending until the guest takes the exception by unmasking PSTATE.A. 1153 1154Running the VCPU may cause it to take a pending SError, or make an access that 1155causes an SError to become pending. The event's description is only valid while 1156the VPCU is not running. 1157 1158This API provides a way to read and write the pending 'event' state that is not 1159visible to the guest. To save, restore or migrate a VCPU the struct representing 1160the state can be read then written using this GET/SET API, along with the other 1161guest-visible registers. It is not possible to 'cancel' an SError that has been 1162made pending. 1163 1164A device being emulated in user-space may also wish to generate an SError. To do 1165this the events structure can be populated by user-space. The current state 1166should be read first, to ensure no existing SError is pending. If an existing 1167SError is pending, the architecture's 'Multiple SError interrupts' rules should 1168be followed. (2.5.3 of DDI0587.a "ARM Reliability, Availability, and 1169Serviceability (RAS) Specification"). 1170 1171SError exceptions always have an ESR value. Some CPUs have the ability to 1172specify what the virtual SError's ESR value should be. These systems will 1173advertise KVM_CAP_ARM_INJECT_SERROR_ESR. In this case exception.has_esr will 1174always have a non-zero value when read, and the agent making an SError pending 1175should specify the ISS field in the lower 24 bits of exception.serror_esr. If 1176the system supports KVM_CAP_ARM_INJECT_SERROR_ESR, but user-space sets the events 1177with exception.has_esr as zero, KVM will choose an ESR. 1178 1179Specifying exception.has_esr on a system that does not support it will return 1180-EINVAL. Setting anything other than the lower 24bits of exception.serror_esr 1181will return -EINVAL. 1182 1183It is not possible to read back a pending external abort (injected via 1184KVM_SET_VCPU_EVENTS or otherwise) because such an exception is always delivered 1185directly to the virtual CPU). 1186 1187:: 1188 1189 struct kvm_vcpu_events { 1190 struct { 1191 __u8 serror_pending; 1192 __u8 serror_has_esr; 1193 __u8 ext_dabt_pending; 1194 /* Align it to 8 bytes */ 1195 __u8 pad[5]; 1196 __u64 serror_esr; 1197 } exception; 1198 __u32 reserved[12]; 1199 }; 1200 12014.32 KVM_SET_VCPU_EVENTS 1202------------------------ 1203 1204:Capability: KVM_CAP_VCPU_EVENTS 1205:Extended by: KVM_CAP_INTR_SHADOW 1206:Architectures: x86, arm, arm64 1207:Type: vcpu ioctl 1208:Parameters: struct kvm_vcpu_event (in) 1209:Returns: 0 on success, -1 on error 1210 1211X86: 1212^^^^ 1213 1214Set pending exceptions, interrupts, and NMIs as well as related states of the 1215vcpu. 1216 1217See KVM_GET_VCPU_EVENTS for the data structure. 1218 1219Fields that may be modified asynchronously by running VCPUs can be excluded 1220from the update. These fields are nmi.pending, sipi_vector, smi.smm, 1221smi.pending. Keep the corresponding bits in the flags field cleared to 1222suppress overwriting the current in-kernel state. The bits are: 1223 1224=============================== ================================== 1225KVM_VCPUEVENT_VALID_NMI_PENDING transfer nmi.pending to the kernel 1226KVM_VCPUEVENT_VALID_SIPI_VECTOR transfer sipi_vector 1227KVM_VCPUEVENT_VALID_SMM transfer the smi sub-struct. 1228=============================== ================================== 1229 1230If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in 1231the flags field to signal that interrupt.shadow contains a valid state and 1232shall be written into the VCPU. 1233 1234KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available. 1235 1236If KVM_CAP_EXCEPTION_PAYLOAD is enabled, KVM_VCPUEVENT_VALID_PAYLOAD 1237can be set in the flags field to signal that the 1238exception_has_payload, exception_payload, and exception.pending fields 1239contain a valid state and shall be written into the VCPU. 1240 1241ARM/ARM64: 1242^^^^^^^^^^ 1243 1244User space may need to inject several types of events to the guest. 1245 1246Set the pending SError exception state for this VCPU. It is not possible to 1247'cancel' an Serror that has been made pending. 1248 1249If the guest performed an access to I/O memory which could not be handled by 1250userspace, for example because of missing instruction syndrome decode 1251information or because there is no device mapped at the accessed IPA, then 1252userspace can ask the kernel to inject an external abort using the address 1253from the exiting fault on the VCPU. It is a programming error to set 1254ext_dabt_pending after an exit which was not either KVM_EXIT_MMIO or 1255KVM_EXIT_ARM_NISV. This feature is only available if the system supports 1256KVM_CAP_ARM_INJECT_EXT_DABT. This is a helper which provides commonality in 1257how userspace reports accesses for the above cases to guests, across different 1258userspace implementations. Nevertheless, userspace can still emulate all Arm 1259exceptions by manipulating individual registers using the KVM_SET_ONE_REG API. 1260 1261See KVM_GET_VCPU_EVENTS for the data structure. 1262 1263 12644.33 KVM_GET_DEBUGREGS 1265---------------------- 1266 1267:Capability: KVM_CAP_DEBUGREGS 1268:Architectures: x86 1269:Type: vm ioctl 1270:Parameters: struct kvm_debugregs (out) 1271:Returns: 0 on success, -1 on error 1272 1273Reads debug registers from the vcpu. 1274 1275:: 1276 1277 struct kvm_debugregs { 1278 __u64 db[4]; 1279 __u64 dr6; 1280 __u64 dr7; 1281 __u64 flags; 1282 __u64 reserved[9]; 1283 }; 1284 1285 12864.34 KVM_SET_DEBUGREGS 1287---------------------- 1288 1289:Capability: KVM_CAP_DEBUGREGS 1290:Architectures: x86 1291:Type: vm ioctl 1292:Parameters: struct kvm_debugregs (in) 1293:Returns: 0 on success, -1 on error 1294 1295Writes debug registers into the vcpu. 1296 1297See KVM_GET_DEBUGREGS for the data structure. The flags field is unused 1298yet and must be cleared on entry. 1299 1300 13014.35 KVM_SET_USER_MEMORY_REGION 1302------------------------------- 1303 1304:Capability: KVM_CAP_USER_MEMORY 1305:Architectures: all 1306:Type: vm ioctl 1307:Parameters: struct kvm_userspace_memory_region (in) 1308:Returns: 0 on success, -1 on error 1309 1310:: 1311 1312 struct kvm_userspace_memory_region { 1313 __u32 slot; 1314 __u32 flags; 1315 __u64 guest_phys_addr; 1316 __u64 memory_size; /* bytes */ 1317 __u64 userspace_addr; /* start of the userspace allocated memory */ 1318 }; 1319 1320 /* for kvm_memory_region::flags */ 1321 #define KVM_MEM_LOG_DIRTY_PAGES (1UL << 0) 1322 #define KVM_MEM_READONLY (1UL << 1) 1323 1324This ioctl allows the user to create, modify or delete a guest physical 1325memory slot. Bits 0-15 of "slot" specify the slot id and this value 1326should be less than the maximum number of user memory slots supported per 1327VM. The maximum allowed slots can be queried using KVM_CAP_NR_MEMSLOTS. 1328Slots may not overlap in guest physical address space. 1329 1330If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot" 1331specifies the address space which is being modified. They must be 1332less than the value that KVM_CHECK_EXTENSION returns for the 1333KVM_CAP_MULTI_ADDRESS_SPACE capability. Slots in separate address spaces 1334are unrelated; the restriction on overlapping slots only applies within 1335each address space. 1336 1337Deleting a slot is done by passing zero for memory_size. When changing 1338an existing slot, it may be moved in the guest physical memory space, 1339or its flags may be modified, but it may not be resized. 1340 1341Memory for the region is taken starting at the address denoted by the 1342field userspace_addr, which must point at user addressable memory for 1343the entire memory slot size. Any object may back this memory, including 1344anonymous memory, ordinary files, and hugetlbfs. 1345 1346On architectures that support a form of address tagging, userspace_addr must 1347be an untagged address. 1348 1349It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr 1350be identical. This allows large pages in the guest to be backed by large 1351pages in the host. 1352 1353The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and 1354KVM_MEM_READONLY. The former can be set to instruct KVM to keep track of 1355writes to memory within the slot. See KVM_GET_DIRTY_LOG ioctl to know how to 1356use it. The latter can be set, if KVM_CAP_READONLY_MEM capability allows it, 1357to make a new slot read-only. In this case, writes to this memory will be 1358posted to userspace as KVM_EXIT_MMIO exits. 1359 1360When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of 1361the memory region are automatically reflected into the guest. For example, an 1362mmap() that affects the region will be made visible immediately. Another 1363example is madvise(MADV_DROP). 1364 1365It is recommended to use this API instead of the KVM_SET_MEMORY_REGION ioctl. 1366The KVM_SET_MEMORY_REGION does not allow fine grained control over memory 1367allocation and is deprecated. 1368 1369 13704.36 KVM_SET_TSS_ADDR 1371--------------------- 1372 1373:Capability: KVM_CAP_SET_TSS_ADDR 1374:Architectures: x86 1375:Type: vm ioctl 1376:Parameters: unsigned long tss_address (in) 1377:Returns: 0 on success, -1 on error 1378 1379This ioctl defines the physical address of a three-page region in the guest 1380physical address space. The region must be within the first 4GB of the 1381guest physical address space and must not conflict with any memory slot 1382or any mmio address. The guest may malfunction if it accesses this memory 1383region. 1384 1385This ioctl is required on Intel-based hosts. This is needed on Intel hardware 1386because of a quirk in the virtualization implementation (see the internals 1387documentation when it pops into existence). 1388 1389 13904.37 KVM_ENABLE_CAP 1391------------------- 1392 1393:Capability: KVM_CAP_ENABLE_CAP 1394:Architectures: mips, ppc, s390 1395:Type: vcpu ioctl 1396:Parameters: struct kvm_enable_cap (in) 1397:Returns: 0 on success; -1 on error 1398 1399:Capability: KVM_CAP_ENABLE_CAP_VM 1400:Architectures: all 1401:Type: vm ioctl 1402:Parameters: struct kvm_enable_cap (in) 1403:Returns: 0 on success; -1 on error 1404 1405.. note:: 1406 1407 Not all extensions are enabled by default. Using this ioctl the application 1408 can enable an extension, making it available to the guest. 1409 1410On systems that do not support this ioctl, it always fails. On systems that 1411do support it, it only works for extensions that are supported for enablement. 1412 1413To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should 1414be used. 1415 1416:: 1417 1418 struct kvm_enable_cap { 1419 /* in */ 1420 __u32 cap; 1421 1422The capability that is supposed to get enabled. 1423 1424:: 1425 1426 __u32 flags; 1427 1428A bitfield indicating future enhancements. Has to be 0 for now. 1429 1430:: 1431 1432 __u64 args[4]; 1433 1434Arguments for enabling a feature. If a feature needs initial values to 1435function properly, this is the place to put them. 1436 1437:: 1438 1439 __u8 pad[64]; 1440 }; 1441 1442The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl 1443for vm-wide capabilities. 1444 14454.38 KVM_GET_MP_STATE 1446--------------------- 1447 1448:Capability: KVM_CAP_MP_STATE 1449:Architectures: x86, s390, arm, arm64, riscv 1450:Type: vcpu ioctl 1451:Parameters: struct kvm_mp_state (out) 1452:Returns: 0 on success; -1 on error 1453 1454:: 1455 1456 struct kvm_mp_state { 1457 __u32 mp_state; 1458 }; 1459 1460Returns the vcpu's current "multiprocessing state" (though also valid on 1461uniprocessor guests). 1462 1463Possible values are: 1464 1465 ========================== =============================================== 1466 KVM_MP_STATE_RUNNABLE the vcpu is currently running 1467 [x86,arm/arm64,riscv] 1468 KVM_MP_STATE_UNINITIALIZED the vcpu is an application processor (AP) 1469 which has not yet received an INIT signal [x86] 1470 KVM_MP_STATE_INIT_RECEIVED the vcpu has received an INIT signal, and is 1471 now ready for a SIPI [x86] 1472 KVM_MP_STATE_HALTED the vcpu has executed a HLT instruction and 1473 is waiting for an interrupt [x86] 1474 KVM_MP_STATE_SIPI_RECEIVED the vcpu has just received a SIPI (vector 1475 accessible via KVM_GET_VCPU_EVENTS) [x86] 1476 KVM_MP_STATE_STOPPED the vcpu is stopped [s390,arm/arm64,riscv] 1477 KVM_MP_STATE_CHECK_STOP the vcpu is in a special error state [s390] 1478 KVM_MP_STATE_OPERATING the vcpu is operating (running or halted) 1479 [s390] 1480 KVM_MP_STATE_LOAD the vcpu is in a special load/startup state 1481 [s390] 1482 ========================== =============================================== 1483 1484On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an 1485in-kernel irqchip, the multiprocessing state must be maintained by userspace on 1486these architectures. 1487 1488For arm/arm64/riscv: 1489^^^^^^^^^^^^^^^^^^^^ 1490 1491The only states that are valid are KVM_MP_STATE_STOPPED and 1492KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not. 1493 14944.39 KVM_SET_MP_STATE 1495--------------------- 1496 1497:Capability: KVM_CAP_MP_STATE 1498:Architectures: x86, s390, arm, arm64, riscv 1499:Type: vcpu ioctl 1500:Parameters: struct kvm_mp_state (in) 1501:Returns: 0 on success; -1 on error 1502 1503Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for 1504arguments. 1505 1506On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an 1507in-kernel irqchip, the multiprocessing state must be maintained by userspace on 1508these architectures. 1509 1510For arm/arm64/riscv: 1511^^^^^^^^^^^^^^^^^^^^ 1512 1513The only states that are valid are KVM_MP_STATE_STOPPED and 1514KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not. 1515 15164.40 KVM_SET_IDENTITY_MAP_ADDR 1517------------------------------ 1518 1519:Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR 1520:Architectures: x86 1521:Type: vm ioctl 1522:Parameters: unsigned long identity (in) 1523:Returns: 0 on success, -1 on error 1524 1525This ioctl defines the physical address of a one-page region in the guest 1526physical address space. The region must be within the first 4GB of the 1527guest physical address space and must not conflict with any memory slot 1528or any mmio address. The guest may malfunction if it accesses this memory 1529region. 1530 1531Setting the address to 0 will result in resetting the address to its default 1532(0xfffbc000). 1533 1534This ioctl is required on Intel-based hosts. This is needed on Intel hardware 1535because of a quirk in the virtualization implementation (see the internals 1536documentation when it pops into existence). 1537 1538Fails if any VCPU has already been created. 1539 15404.41 KVM_SET_BOOT_CPU_ID 1541------------------------ 1542 1543:Capability: KVM_CAP_SET_BOOT_CPU_ID 1544:Architectures: x86 1545:Type: vm ioctl 1546:Parameters: unsigned long vcpu_id 1547:Returns: 0 on success, -1 on error 1548 1549Define which vcpu is the Bootstrap Processor (BSP). Values are the same 1550as the vcpu id in KVM_CREATE_VCPU. If this ioctl is not called, the default 1551is vcpu 0. This ioctl has to be called before vcpu creation, 1552otherwise it will return EBUSY error. 1553 1554 15554.42 KVM_GET_XSAVE 1556------------------ 1557 1558:Capability: KVM_CAP_XSAVE 1559:Architectures: x86 1560:Type: vcpu ioctl 1561:Parameters: struct kvm_xsave (out) 1562:Returns: 0 on success, -1 on error 1563 1564 1565:: 1566 1567 struct kvm_xsave { 1568 __u32 region[1024]; 1569 }; 1570 1571This ioctl would copy current vcpu's xsave struct to the userspace. 1572 1573 15744.43 KVM_SET_XSAVE 1575------------------ 1576 1577:Capability: KVM_CAP_XSAVE 1578:Architectures: x86 1579:Type: vcpu ioctl 1580:Parameters: struct kvm_xsave (in) 1581:Returns: 0 on success, -1 on error 1582 1583:: 1584 1585 1586 struct kvm_xsave { 1587 __u32 region[1024]; 1588 }; 1589 1590This ioctl would copy userspace's xsave struct to the kernel. 1591 1592 15934.44 KVM_GET_XCRS 1594----------------- 1595 1596:Capability: KVM_CAP_XCRS 1597:Architectures: x86 1598:Type: vcpu ioctl 1599:Parameters: struct kvm_xcrs (out) 1600:Returns: 0 on success, -1 on error 1601 1602:: 1603 1604 struct kvm_xcr { 1605 __u32 xcr; 1606 __u32 reserved; 1607 __u64 value; 1608 }; 1609 1610 struct kvm_xcrs { 1611 __u32 nr_xcrs; 1612 __u32 flags; 1613 struct kvm_xcr xcrs[KVM_MAX_XCRS]; 1614 __u64 padding[16]; 1615 }; 1616 1617This ioctl would copy current vcpu's xcrs to the userspace. 1618 1619 16204.45 KVM_SET_XCRS 1621----------------- 1622 1623:Capability: KVM_CAP_XCRS 1624:Architectures: x86 1625:Type: vcpu ioctl 1626:Parameters: struct kvm_xcrs (in) 1627:Returns: 0 on success, -1 on error 1628 1629:: 1630 1631 struct kvm_xcr { 1632 __u32 xcr; 1633 __u32 reserved; 1634 __u64 value; 1635 }; 1636 1637 struct kvm_xcrs { 1638 __u32 nr_xcrs; 1639 __u32 flags; 1640 struct kvm_xcr xcrs[KVM_MAX_XCRS]; 1641 __u64 padding[16]; 1642 }; 1643 1644This ioctl would set vcpu's xcr to the value userspace specified. 1645 1646 16474.46 KVM_GET_SUPPORTED_CPUID 1648---------------------------- 1649 1650:Capability: KVM_CAP_EXT_CPUID 1651:Architectures: x86 1652:Type: system ioctl 1653:Parameters: struct kvm_cpuid2 (in/out) 1654:Returns: 0 on success, -1 on error 1655 1656:: 1657 1658 struct kvm_cpuid2 { 1659 __u32 nent; 1660 __u32 padding; 1661 struct kvm_cpuid_entry2 entries[0]; 1662 }; 1663 1664 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0) 1665 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */ 1666 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */ 1667 1668 struct kvm_cpuid_entry2 { 1669 __u32 function; 1670 __u32 index; 1671 __u32 flags; 1672 __u32 eax; 1673 __u32 ebx; 1674 __u32 ecx; 1675 __u32 edx; 1676 __u32 padding[3]; 1677 }; 1678 1679This ioctl returns x86 cpuid features which are supported by both the 1680hardware and kvm in its default configuration. Userspace can use the 1681information returned by this ioctl to construct cpuid information (for 1682KVM_SET_CPUID2) that is consistent with hardware, kernel, and 1683userspace capabilities, and with user requirements (for example, the 1684user may wish to constrain cpuid to emulate older hardware, or for 1685feature consistency across a cluster). 1686 1687Note that certain capabilities, such as KVM_CAP_X86_DISABLE_EXITS, may 1688expose cpuid features (e.g. MONITOR) which are not supported by kvm in 1689its default configuration. If userspace enables such capabilities, it 1690is responsible for modifying the results of this ioctl appropriately. 1691 1692Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure 1693with the 'nent' field indicating the number of entries in the variable-size 1694array 'entries'. If the number of entries is too low to describe the cpu 1695capabilities, an error (E2BIG) is returned. If the number is too high, 1696the 'nent' field is adjusted and an error (ENOMEM) is returned. If the 1697number is just right, the 'nent' field is adjusted to the number of valid 1698entries in the 'entries' array, which is then filled. 1699 1700The entries returned are the host cpuid as returned by the cpuid instruction, 1701with unknown or unsupported features masked out. Some features (for example, 1702x2apic), may not be present in the host cpu, but are exposed by kvm if it can 1703emulate them efficiently. The fields in each entry are defined as follows: 1704 1705 function: 1706 the eax value used to obtain the entry 1707 1708 index: 1709 the ecx value used to obtain the entry (for entries that are 1710 affected by ecx) 1711 1712 flags: 1713 an OR of zero or more of the following: 1714 1715 KVM_CPUID_FLAG_SIGNIFCANT_INDEX: 1716 if the index field is valid 1717 1718 eax, ebx, ecx, edx: 1719 the values returned by the cpuid instruction for 1720 this function/index combination 1721 1722The TSC deadline timer feature (CPUID leaf 1, ecx[24]) is always returned 1723as false, since the feature depends on KVM_CREATE_IRQCHIP for local APIC 1724support. Instead it is reported via:: 1725 1726 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER) 1727 1728if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the 1729feature in userspace, then you can enable the feature for KVM_SET_CPUID2. 1730 1731 17324.47 KVM_PPC_GET_PVINFO 1733----------------------- 1734 1735:Capability: KVM_CAP_PPC_GET_PVINFO 1736:Architectures: ppc 1737:Type: vm ioctl 1738:Parameters: struct kvm_ppc_pvinfo (out) 1739:Returns: 0 on success, !0 on error 1740 1741:: 1742 1743 struct kvm_ppc_pvinfo { 1744 __u32 flags; 1745 __u32 hcall[4]; 1746 __u8 pad[108]; 1747 }; 1748 1749This ioctl fetches PV specific information that need to be passed to the guest 1750using the device tree or other means from vm context. 1751 1752The hcall array defines 4 instructions that make up a hypercall. 1753 1754If any additional field gets added to this structure later on, a bit for that 1755additional piece of information will be set in the flags bitmap. 1756 1757The flags bitmap is defined as:: 1758 1759 /* the host supports the ePAPR idle hcall 1760 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1<<0) 1761 17624.52 KVM_SET_GSI_ROUTING 1763------------------------ 1764 1765:Capability: KVM_CAP_IRQ_ROUTING 1766:Architectures: x86 s390 arm arm64 1767:Type: vm ioctl 1768:Parameters: struct kvm_irq_routing (in) 1769:Returns: 0 on success, -1 on error 1770 1771Sets the GSI routing table entries, overwriting any previously set entries. 1772 1773On arm/arm64, GSI routing has the following limitation: 1774 1775- GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD. 1776 1777:: 1778 1779 struct kvm_irq_routing { 1780 __u32 nr; 1781 __u32 flags; 1782 struct kvm_irq_routing_entry entries[0]; 1783 }; 1784 1785No flags are specified so far, the corresponding field must be set to zero. 1786 1787:: 1788 1789 struct kvm_irq_routing_entry { 1790 __u32 gsi; 1791 __u32 type; 1792 __u32 flags; 1793 __u32 pad; 1794 union { 1795 struct kvm_irq_routing_irqchip irqchip; 1796 struct kvm_irq_routing_msi msi; 1797 struct kvm_irq_routing_s390_adapter adapter; 1798 struct kvm_irq_routing_hv_sint hv_sint; 1799 __u32 pad[8]; 1800 } u; 1801 }; 1802 1803 /* gsi routing entry types */ 1804 #define KVM_IRQ_ROUTING_IRQCHIP 1 1805 #define KVM_IRQ_ROUTING_MSI 2 1806 #define KVM_IRQ_ROUTING_S390_ADAPTER 3 1807 #define KVM_IRQ_ROUTING_HV_SINT 4 1808 1809flags: 1810 1811- KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry 1812 type, specifies that the devid field contains a valid value. The per-VM 1813 KVM_CAP_MSI_DEVID capability advertises the requirement to provide 1814 the device ID. If this capability is not available, userspace should 1815 never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail. 1816- zero otherwise 1817 1818:: 1819 1820 struct kvm_irq_routing_irqchip { 1821 __u32 irqchip; 1822 __u32 pin; 1823 }; 1824 1825 struct kvm_irq_routing_msi { 1826 __u32 address_lo; 1827 __u32 address_hi; 1828 __u32 data; 1829 union { 1830 __u32 pad; 1831 __u32 devid; 1832 }; 1833 }; 1834 1835If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier 1836for the device that wrote the MSI message. For PCI, this is usually a 1837BFD identifier in the lower 16 bits. 1838 1839On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS 1840feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled, 1841address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of 1842address_hi must be zero. 1843 1844:: 1845 1846 struct kvm_irq_routing_s390_adapter { 1847 __u64 ind_addr; 1848 __u64 summary_addr; 1849 __u64 ind_offset; 1850 __u32 summary_offset; 1851 __u32 adapter_id; 1852 }; 1853 1854 struct kvm_irq_routing_hv_sint { 1855 __u32 vcpu; 1856 __u32 sint; 1857 }; 1858 1859 18604.55 KVM_SET_TSC_KHZ 1861-------------------- 1862 1863:Capability: KVM_CAP_TSC_CONTROL 1864:Architectures: x86 1865:Type: vcpu ioctl 1866:Parameters: virtual tsc_khz 1867:Returns: 0 on success, -1 on error 1868 1869Specifies the tsc frequency for the virtual machine. The unit of the 1870frequency is KHz. 1871 1872 18734.56 KVM_GET_TSC_KHZ 1874-------------------- 1875 1876:Capability: KVM_CAP_GET_TSC_KHZ 1877:Architectures: x86 1878:Type: vcpu ioctl 1879:Parameters: none 1880:Returns: virtual tsc-khz on success, negative value on error 1881 1882Returns the tsc frequency of the guest. The unit of the return value is 1883KHz. If the host has unstable tsc this ioctl returns -EIO instead as an 1884error. 1885 1886 18874.57 KVM_GET_LAPIC 1888------------------ 1889 1890:Capability: KVM_CAP_IRQCHIP 1891:Architectures: x86 1892:Type: vcpu ioctl 1893:Parameters: struct kvm_lapic_state (out) 1894:Returns: 0 on success, -1 on error 1895 1896:: 1897 1898 #define KVM_APIC_REG_SIZE 0x400 1899 struct kvm_lapic_state { 1900 char regs[KVM_APIC_REG_SIZE]; 1901 }; 1902 1903Reads the Local APIC registers and copies them into the input argument. The 1904data format and layout are the same as documented in the architecture manual. 1905 1906If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is 1907enabled, then the format of APIC_ID register depends on the APIC mode 1908(reported by MSR_IA32_APICBASE) of its VCPU. x2APIC stores APIC ID in 1909the APIC_ID register (bytes 32-35). xAPIC only allows an 8-bit APIC ID 1910which is stored in bits 31-24 of the APIC register, or equivalently in 1911byte 35 of struct kvm_lapic_state's regs field. KVM_GET_LAPIC must then 1912be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR. 1913 1914If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state 1915always uses xAPIC format. 1916 1917 19184.58 KVM_SET_LAPIC 1919------------------ 1920 1921:Capability: KVM_CAP_IRQCHIP 1922:Architectures: x86 1923:Type: vcpu ioctl 1924:Parameters: struct kvm_lapic_state (in) 1925:Returns: 0 on success, -1 on error 1926 1927:: 1928 1929 #define KVM_APIC_REG_SIZE 0x400 1930 struct kvm_lapic_state { 1931 char regs[KVM_APIC_REG_SIZE]; 1932 }; 1933 1934Copies the input argument into the Local APIC registers. The data format 1935and layout are the same as documented in the architecture manual. 1936 1937The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's 1938regs field) depends on the state of the KVM_CAP_X2APIC_API capability. 1939See the note in KVM_GET_LAPIC. 1940 1941 19424.59 KVM_IOEVENTFD 1943------------------ 1944 1945:Capability: KVM_CAP_IOEVENTFD 1946:Architectures: all 1947:Type: vm ioctl 1948:Parameters: struct kvm_ioeventfd (in) 1949:Returns: 0 on success, !0 on error 1950 1951This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address 1952within the guest. A guest write in the registered address will signal the 1953provided event instead of triggering an exit. 1954 1955:: 1956 1957 struct kvm_ioeventfd { 1958 __u64 datamatch; 1959 __u64 addr; /* legal pio/mmio address */ 1960 __u32 len; /* 0, 1, 2, 4, or 8 bytes */ 1961 __s32 fd; 1962 __u32 flags; 1963 __u8 pad[36]; 1964 }; 1965 1966For the special case of virtio-ccw devices on s390, the ioevent is matched 1967to a subchannel/virtqueue tuple instead. 1968 1969The following flags are defined:: 1970 1971 #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch) 1972 #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio) 1973 #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign) 1974 #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \ 1975 (1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify) 1976 1977If datamatch flag is set, the event will be signaled only if the written value 1978to the registered address is equal to datamatch in struct kvm_ioeventfd. 1979 1980For virtio-ccw devices, addr contains the subchannel id and datamatch the 1981virtqueue index. 1982 1983With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and 1984the kernel will ignore the length of guest write and may get a faster vmexit. 1985The speedup may only apply to specific architectures, but the ioeventfd will 1986work anyway. 1987 19884.60 KVM_DIRTY_TLB 1989------------------ 1990 1991:Capability: KVM_CAP_SW_TLB 1992:Architectures: ppc 1993:Type: vcpu ioctl 1994:Parameters: struct kvm_dirty_tlb (in) 1995:Returns: 0 on success, -1 on error 1996 1997:: 1998 1999 struct kvm_dirty_tlb { 2000 __u64 bitmap; 2001 __u32 num_dirty; 2002 }; 2003 2004This must be called whenever userspace has changed an entry in the shared 2005TLB, prior to calling KVM_RUN on the associated vcpu. 2006 2007The "bitmap" field is the userspace address of an array. This array 2008consists of a number of bits, equal to the total number of TLB entries as 2009determined by the last successful call to KVM_CONFIG_TLB, rounded up to the 2010nearest multiple of 64. 2011 2012Each bit corresponds to one TLB entry, ordered the same as in the shared TLB 2013array. 2014 2015The array is little-endian: the bit 0 is the least significant bit of the 2016first byte, bit 8 is the least significant bit of the second byte, etc. 2017This avoids any complications with differing word sizes. 2018 2019The "num_dirty" field is a performance hint for KVM to determine whether it 2020should skip processing the bitmap and just invalidate everything. It must 2021be set to the number of set bits in the bitmap. 2022 2023 20244.62 KVM_CREATE_SPAPR_TCE 2025------------------------- 2026 2027:Capability: KVM_CAP_SPAPR_TCE 2028:Architectures: powerpc 2029:Type: vm ioctl 2030:Parameters: struct kvm_create_spapr_tce (in) 2031:Returns: file descriptor for manipulating the created TCE table 2032 2033This creates a virtual TCE (translation control entry) table, which 2034is an IOMMU for PAPR-style virtual I/O. It is used to translate 2035logical addresses used in virtual I/O into guest physical addresses, 2036and provides a scatter/gather capability for PAPR virtual I/O. 2037 2038:: 2039 2040 /* for KVM_CAP_SPAPR_TCE */ 2041 struct kvm_create_spapr_tce { 2042 __u64 liobn; 2043 __u32 window_size; 2044 }; 2045 2046The liobn field gives the logical IO bus number for which to create a 2047TCE table. The window_size field specifies the size of the DMA window 2048which this TCE table will translate - the table will contain one 64 2049bit TCE entry for every 4kiB of the DMA window. 2050 2051When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE 2052table has been created using this ioctl(), the kernel will handle it 2053in real mode, updating the TCE table. H_PUT_TCE calls for other 2054liobns will cause a vm exit and must be handled by userspace. 2055 2056The return value is a file descriptor which can be passed to mmap(2) 2057to map the created TCE table into userspace. This lets userspace read 2058the entries written by kernel-handled H_PUT_TCE calls, and also lets 2059userspace update the TCE table directly which is useful in some 2060circumstances. 2061 2062 20634.63 KVM_ALLOCATE_RMA 2064--------------------- 2065 2066:Capability: KVM_CAP_PPC_RMA 2067:Architectures: powerpc 2068:Type: vm ioctl 2069:Parameters: struct kvm_allocate_rma (out) 2070:Returns: file descriptor for mapping the allocated RMA 2071 2072This allocates a Real Mode Area (RMA) from the pool allocated at boot 2073time by the kernel. An RMA is a physically-contiguous, aligned region 2074of memory used on older POWER processors to provide the memory which 2075will be accessed by real-mode (MMU off) accesses in a KVM guest. 2076POWER processors support a set of sizes for the RMA that usually 2077includes 64MB, 128MB, 256MB and some larger powers of two. 2078 2079:: 2080 2081 /* for KVM_ALLOCATE_RMA */ 2082 struct kvm_allocate_rma { 2083 __u64 rma_size; 2084 }; 2085 2086The return value is a file descriptor which can be passed to mmap(2) 2087to map the allocated RMA into userspace. The mapped area can then be 2088passed to the KVM_SET_USER_MEMORY_REGION ioctl to establish it as the 2089RMA for a virtual machine. The size of the RMA in bytes (which is 2090fixed at host kernel boot time) is returned in the rma_size field of 2091the argument structure. 2092 2093The KVM_CAP_PPC_RMA capability is 1 or 2 if the KVM_ALLOCATE_RMA ioctl 2094is supported; 2 if the processor requires all virtual machines to have 2095an RMA, or 1 if the processor can use an RMA but doesn't require it, 2096because it supports the Virtual RMA (VRMA) facility. 2097 2098 20994.64 KVM_NMI 2100------------ 2101 2102:Capability: KVM_CAP_USER_NMI 2103:Architectures: x86 2104:Type: vcpu ioctl 2105:Parameters: none 2106:Returns: 0 on success, -1 on error 2107 2108Queues an NMI on the thread's vcpu. Note this is well defined only 2109when KVM_CREATE_IRQCHIP has not been called, since this is an interface 2110between the virtual cpu core and virtual local APIC. After KVM_CREATE_IRQCHIP 2111has been called, this interface is completely emulated within the kernel. 2112 2113To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the 2114following algorithm: 2115 2116 - pause the vcpu 2117 - read the local APIC's state (KVM_GET_LAPIC) 2118 - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1) 2119 - if so, issue KVM_NMI 2120 - resume the vcpu 2121 2122Some guests configure the LINT1 NMI input to cause a panic, aiding in 2123debugging. 2124 2125 21264.65 KVM_S390_UCAS_MAP 2127---------------------- 2128 2129:Capability: KVM_CAP_S390_UCONTROL 2130:Architectures: s390 2131:Type: vcpu ioctl 2132:Parameters: struct kvm_s390_ucas_mapping (in) 2133:Returns: 0 in case of success 2134 2135The parameter is defined like this:: 2136 2137 struct kvm_s390_ucas_mapping { 2138 __u64 user_addr; 2139 __u64 vcpu_addr; 2140 __u64 length; 2141 }; 2142 2143This ioctl maps the memory at "user_addr" with the length "length" to 2144the vcpu's address space starting at "vcpu_addr". All parameters need to 2145be aligned by 1 megabyte. 2146 2147 21484.66 KVM_S390_UCAS_UNMAP 2149------------------------ 2150 2151:Capability: KVM_CAP_S390_UCONTROL 2152:Architectures: s390 2153:Type: vcpu ioctl 2154:Parameters: struct kvm_s390_ucas_mapping (in) 2155:Returns: 0 in case of success 2156 2157The parameter is defined like this:: 2158 2159 struct kvm_s390_ucas_mapping { 2160 __u64 user_addr; 2161 __u64 vcpu_addr; 2162 __u64 length; 2163 }; 2164 2165This ioctl unmaps the memory in the vcpu's address space starting at 2166"vcpu_addr" with the length "length". The field "user_addr" is ignored. 2167All parameters need to be aligned by 1 megabyte. 2168 2169 21704.67 KVM_S390_VCPU_FAULT 2171------------------------ 2172 2173:Capability: KVM_CAP_S390_UCONTROL 2174:Architectures: s390 2175:Type: vcpu ioctl 2176:Parameters: vcpu absolute address (in) 2177:Returns: 0 in case of success 2178 2179This call creates a page table entry on the virtual cpu's address space 2180(for user controlled virtual machines) or the virtual machine's address 2181space (for regular virtual machines). This only works for minor faults, 2182thus it's recommended to access subject memory page via the user page 2183table upfront. This is useful to handle validity intercepts for user 2184controlled virtual machines to fault in the virtual cpu's lowcore pages 2185prior to calling the KVM_RUN ioctl. 2186 2187 21884.68 KVM_SET_ONE_REG 2189-------------------- 2190 2191:Capability: KVM_CAP_ONE_REG 2192:Architectures: all 2193:Type: vcpu ioctl 2194:Parameters: struct kvm_one_reg (in) 2195:Returns: 0 on success, negative value on failure 2196 2197Errors: 2198 2199 ====== ============================================================ 2200 ENOENT no such register 2201 EINVAL invalid register ID, or no such register or used with VMs in 2202 protected virtualization mode on s390 2203 EPERM (arm64) register access not allowed before vcpu finalization 2204 ====== ============================================================ 2205 2206(These error codes are indicative only: do not rely on a specific error 2207code being returned in a specific situation.) 2208 2209:: 2210 2211 struct kvm_one_reg { 2212 __u64 id; 2213 __u64 addr; 2214 }; 2215 2216Using this ioctl, a single vcpu register can be set to a specific value 2217defined by user space with the passed in struct kvm_one_reg, where id 2218refers to the register identifier as described below and addr is a pointer 2219to a variable with the respective size. There can be architecture agnostic 2220and architecture specific registers. Each have their own range of operation 2221and their own constants and width. To keep track of the implemented 2222registers, find a list below: 2223 2224 ======= =============================== ============ 2225 Arch Register Width (bits) 2226 ======= =============================== ============ 2227 PPC KVM_REG_PPC_HIOR 64 2228 PPC KVM_REG_PPC_IAC1 64 2229 PPC KVM_REG_PPC_IAC2 64 2230 PPC KVM_REG_PPC_IAC3 64 2231 PPC KVM_REG_PPC_IAC4 64 2232 PPC KVM_REG_PPC_DAC1 64 2233 PPC KVM_REG_PPC_DAC2 64 2234 PPC KVM_REG_PPC_DABR 64 2235 PPC KVM_REG_PPC_DSCR 64 2236 PPC KVM_REG_PPC_PURR 64 2237 PPC KVM_REG_PPC_SPURR 64 2238 PPC KVM_REG_PPC_DAR 64 2239 PPC KVM_REG_PPC_DSISR 32 2240 PPC KVM_REG_PPC_AMR 64 2241 PPC KVM_REG_PPC_UAMOR 64 2242 PPC KVM_REG_PPC_MMCR0 64 2243 PPC KVM_REG_PPC_MMCR1 64 2244 PPC KVM_REG_PPC_MMCRA 64 2245 PPC KVM_REG_PPC_MMCR2 64 2246 PPC KVM_REG_PPC_MMCRS 64 2247 PPC KVM_REG_PPC_MMCR3 64 2248 PPC KVM_REG_PPC_SIAR 64 2249 PPC KVM_REG_PPC_SDAR 64 2250 PPC KVM_REG_PPC_SIER 64 2251 PPC KVM_REG_PPC_SIER2 64 2252 PPC KVM_REG_PPC_SIER3 64 2253 PPC KVM_REG_PPC_PMC1 32 2254 PPC KVM_REG_PPC_PMC2 32 2255 PPC KVM_REG_PPC_PMC3 32 2256 PPC KVM_REG_PPC_PMC4 32 2257 PPC KVM_REG_PPC_PMC5 32 2258 PPC KVM_REG_PPC_PMC6 32 2259 PPC KVM_REG_PPC_PMC7 32 2260 PPC KVM_REG_PPC_PMC8 32 2261 PPC KVM_REG_PPC_FPR0 64 2262 ... 2263 PPC KVM_REG_PPC_FPR31 64 2264 PPC KVM_REG_PPC_VR0 128 2265 ... 2266 PPC KVM_REG_PPC_VR31 128 2267 PPC KVM_REG_PPC_VSR0 128 2268 ... 2269 PPC KVM_REG_PPC_VSR31 128 2270 PPC KVM_REG_PPC_FPSCR 64 2271 PPC KVM_REG_PPC_VSCR 32 2272 PPC KVM_REG_PPC_VPA_ADDR 64 2273 PPC KVM_REG_PPC_VPA_SLB 128 2274 PPC KVM_REG_PPC_VPA_DTL 128 2275 PPC KVM_REG_PPC_EPCR 32 2276 PPC KVM_REG_PPC_EPR 32 2277 PPC KVM_REG_PPC_TCR 32 2278 PPC KVM_REG_PPC_TSR 32 2279 PPC KVM_REG_PPC_OR_TSR 32 2280 PPC KVM_REG_PPC_CLEAR_TSR 32 2281 PPC KVM_REG_PPC_MAS0 32 2282 PPC KVM_REG_PPC_MAS1 32 2283 PPC KVM_REG_PPC_MAS2 64 2284 PPC KVM_REG_PPC_MAS7_3 64 2285 PPC KVM_REG_PPC_MAS4 32 2286 PPC KVM_REG_PPC_MAS6 32 2287 PPC KVM_REG_PPC_MMUCFG 32 2288 PPC KVM_REG_PPC_TLB0CFG 32 2289 PPC KVM_REG_PPC_TLB1CFG 32 2290 PPC KVM_REG_PPC_TLB2CFG 32 2291 PPC KVM_REG_PPC_TLB3CFG 32 2292 PPC KVM_REG_PPC_TLB0PS 32 2293 PPC KVM_REG_PPC_TLB1PS 32 2294 PPC KVM_REG_PPC_TLB2PS 32 2295 PPC KVM_REG_PPC_TLB3PS 32 2296 PPC KVM_REG_PPC_EPTCFG 32 2297 PPC KVM_REG_PPC_ICP_STATE 64 2298 PPC KVM_REG_PPC_VP_STATE 128 2299 PPC KVM_REG_PPC_TB_OFFSET 64 2300 PPC KVM_REG_PPC_SPMC1 32 2301 PPC KVM_REG_PPC_SPMC2 32 2302 PPC KVM_REG_PPC_IAMR 64 2303 PPC KVM_REG_PPC_TFHAR 64 2304 PPC KVM_REG_PPC_TFIAR 64 2305 PPC KVM_REG_PPC_TEXASR 64 2306 PPC KVM_REG_PPC_FSCR 64 2307 PPC KVM_REG_PPC_PSPB 32 2308 PPC KVM_REG_PPC_EBBHR 64 2309 PPC KVM_REG_PPC_EBBRR 64 2310 PPC KVM_REG_PPC_BESCR 64 2311 PPC KVM_REG_PPC_TAR 64 2312 PPC KVM_REG_PPC_DPDES 64 2313 PPC KVM_REG_PPC_DAWR 64 2314 PPC KVM_REG_PPC_DAWRX 64 2315 PPC KVM_REG_PPC_CIABR 64 2316 PPC KVM_REG_PPC_IC 64 2317 PPC KVM_REG_PPC_VTB 64 2318 PPC KVM_REG_PPC_CSIGR 64 2319 PPC KVM_REG_PPC_TACR 64 2320 PPC KVM_REG_PPC_TCSCR 64 2321 PPC KVM_REG_PPC_PID 64 2322 PPC KVM_REG_PPC_ACOP 64 2323 PPC KVM_REG_PPC_VRSAVE 32 2324 PPC KVM_REG_PPC_LPCR 32 2325 PPC KVM_REG_PPC_LPCR_64 64 2326 PPC KVM_REG_PPC_PPR 64 2327 PPC KVM_REG_PPC_ARCH_COMPAT 32 2328 PPC KVM_REG_PPC_DABRX 32 2329 PPC KVM_REG_PPC_WORT 64 2330 PPC KVM_REG_PPC_SPRG9 64 2331 PPC KVM_REG_PPC_DBSR 32 2332 PPC KVM_REG_PPC_TIDR 64 2333 PPC KVM_REG_PPC_PSSCR 64 2334 PPC KVM_REG_PPC_DEC_EXPIRY 64 2335 PPC KVM_REG_PPC_PTCR 64 2336 PPC KVM_REG_PPC_DAWR1 64 2337 PPC KVM_REG_PPC_DAWRX1 64 2338 PPC KVM_REG_PPC_TM_GPR0 64 2339 ... 2340 PPC KVM_REG_PPC_TM_GPR31 64 2341 PPC KVM_REG_PPC_TM_VSR0 128 2342 ... 2343 PPC KVM_REG_PPC_TM_VSR63 128 2344 PPC KVM_REG_PPC_TM_CR 64 2345 PPC KVM_REG_PPC_TM_LR 64 2346 PPC KVM_REG_PPC_TM_CTR 64 2347 PPC KVM_REG_PPC_TM_FPSCR 64 2348 PPC KVM_REG_PPC_TM_AMR 64 2349 PPC KVM_REG_PPC_TM_PPR 64 2350 PPC KVM_REG_PPC_TM_VRSAVE 64 2351 PPC KVM_REG_PPC_TM_VSCR 32 2352 PPC KVM_REG_PPC_TM_DSCR 64 2353 PPC KVM_REG_PPC_TM_TAR 64 2354 PPC KVM_REG_PPC_TM_XER 64 2355 2356 MIPS KVM_REG_MIPS_R0 64 2357 ... 2358 MIPS KVM_REG_MIPS_R31 64 2359 MIPS KVM_REG_MIPS_HI 64 2360 MIPS KVM_REG_MIPS_LO 64 2361 MIPS KVM_REG_MIPS_PC 64 2362 MIPS KVM_REG_MIPS_CP0_INDEX 32 2363 MIPS KVM_REG_MIPS_CP0_ENTRYLO0 64 2364 MIPS KVM_REG_MIPS_CP0_ENTRYLO1 64 2365 MIPS KVM_REG_MIPS_CP0_CONTEXT 64 2366 MIPS KVM_REG_MIPS_CP0_CONTEXTCONFIG 32 2367 MIPS KVM_REG_MIPS_CP0_USERLOCAL 64 2368 MIPS KVM_REG_MIPS_CP0_XCONTEXTCONFIG 64 2369 MIPS KVM_REG_MIPS_CP0_PAGEMASK 32 2370 MIPS KVM_REG_MIPS_CP0_PAGEGRAIN 32 2371 MIPS KVM_REG_MIPS_CP0_SEGCTL0 64 2372 MIPS KVM_REG_MIPS_CP0_SEGCTL1 64 2373 MIPS KVM_REG_MIPS_CP0_SEGCTL2 64 2374 MIPS KVM_REG_MIPS_CP0_PWBASE 64 2375 MIPS KVM_REG_MIPS_CP0_PWFIELD 64 2376 MIPS KVM_REG_MIPS_CP0_PWSIZE 64 2377 MIPS KVM_REG_MIPS_CP0_WIRED 32 2378 MIPS KVM_REG_MIPS_CP0_PWCTL 32 2379 MIPS KVM_REG_MIPS_CP0_HWRENA 32 2380 MIPS KVM_REG_MIPS_CP0_BADVADDR 64 2381 MIPS KVM_REG_MIPS_CP0_BADINSTR 32 2382 MIPS KVM_REG_MIPS_CP0_BADINSTRP 32 2383 MIPS KVM_REG_MIPS_CP0_COUNT 32 2384 MIPS KVM_REG_MIPS_CP0_ENTRYHI 64 2385 MIPS KVM_REG_MIPS_CP0_COMPARE 32 2386 MIPS KVM_REG_MIPS_CP0_STATUS 32 2387 MIPS KVM_REG_MIPS_CP0_INTCTL 32 2388 MIPS KVM_REG_MIPS_CP0_CAUSE 32 2389 MIPS KVM_REG_MIPS_CP0_EPC 64 2390 MIPS KVM_REG_MIPS_CP0_PRID 32 2391 MIPS KVM_REG_MIPS_CP0_EBASE 64 2392 MIPS KVM_REG_MIPS_CP0_CONFIG 32 2393 MIPS KVM_REG_MIPS_CP0_CONFIG1 32 2394 MIPS KVM_REG_MIPS_CP0_CONFIG2 32 2395 MIPS KVM_REG_MIPS_CP0_CONFIG3 32 2396 MIPS KVM_REG_MIPS_CP0_CONFIG4 32 2397 MIPS KVM_REG_MIPS_CP0_CONFIG5 32 2398 MIPS KVM_REG_MIPS_CP0_CONFIG7 32 2399 MIPS KVM_REG_MIPS_CP0_XCONTEXT 64 2400 MIPS KVM_REG_MIPS_CP0_ERROREPC 64 2401 MIPS KVM_REG_MIPS_CP0_KSCRATCH1 64 2402 MIPS KVM_REG_MIPS_CP0_KSCRATCH2 64 2403 MIPS KVM_REG_MIPS_CP0_KSCRATCH3 64 2404 MIPS KVM_REG_MIPS_CP0_KSCRATCH4 64 2405 MIPS KVM_REG_MIPS_CP0_KSCRATCH5 64 2406 MIPS KVM_REG_MIPS_CP0_KSCRATCH6 64 2407 MIPS KVM_REG_MIPS_CP0_MAAR(0..63) 64 2408 MIPS KVM_REG_MIPS_COUNT_CTL 64 2409 MIPS KVM_REG_MIPS_COUNT_RESUME 64 2410 MIPS KVM_REG_MIPS_COUNT_HZ 64 2411 MIPS KVM_REG_MIPS_FPR_32(0..31) 32 2412 MIPS KVM_REG_MIPS_FPR_64(0..31) 64 2413 MIPS KVM_REG_MIPS_VEC_128(0..31) 128 2414 MIPS KVM_REG_MIPS_FCR_IR 32 2415 MIPS KVM_REG_MIPS_FCR_CSR 32 2416 MIPS KVM_REG_MIPS_MSA_IR 32 2417 MIPS KVM_REG_MIPS_MSA_CSR 32 2418 ======= =============================== ============ 2419 2420ARM registers are mapped using the lower 32 bits. The upper 16 of that 2421is the register group type, or coprocessor number: 2422 2423ARM core registers have the following id bit patterns:: 2424 2425 0x4020 0000 0010 <index into the kvm_regs struct:16> 2426 2427ARM 32-bit CP15 registers have the following id bit patterns:: 2428 2429 0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3> 2430 2431ARM 64-bit CP15 registers have the following id bit patterns:: 2432 2433 0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3> 2434 2435ARM CCSIDR registers are demultiplexed by CSSELR value:: 2436 2437 0x4020 0000 0011 00 <csselr:8> 2438 2439ARM 32-bit VFP control registers have the following id bit patterns:: 2440 2441 0x4020 0000 0012 1 <regno:12> 2442 2443ARM 64-bit FP registers have the following id bit patterns:: 2444 2445 0x4030 0000 0012 0 <regno:12> 2446 2447ARM firmware pseudo-registers have the following bit pattern:: 2448 2449 0x4030 0000 0014 <regno:16> 2450 2451 2452arm64 registers are mapped using the lower 32 bits. The upper 16 of 2453that is the register group type, or coprocessor number: 2454 2455arm64 core/FP-SIMD registers have the following id bit patterns. Note 2456that the size of the access is variable, as the kvm_regs structure 2457contains elements ranging from 32 to 128 bits. The index is a 32bit 2458value in the kvm_regs structure seen as a 32bit array:: 2459 2460 0x60x0 0000 0010 <index into the kvm_regs struct:16> 2461 2462Specifically: 2463 2464======================= ========= ===== ======================================= 2465 Encoding Register Bits kvm_regs member 2466======================= ========= ===== ======================================= 2467 0x6030 0000 0010 0000 X0 64 regs.regs[0] 2468 0x6030 0000 0010 0002 X1 64 regs.regs[1] 2469 ... 2470 0x6030 0000 0010 003c X30 64 regs.regs[30] 2471 0x6030 0000 0010 003e SP 64 regs.sp 2472 0x6030 0000 0010 0040 PC 64 regs.pc 2473 0x6030 0000 0010 0042 PSTATE 64 regs.pstate 2474 0x6030 0000 0010 0044 SP_EL1 64 sp_el1 2475 0x6030 0000 0010 0046 ELR_EL1 64 elr_el1 2476 0x6030 0000 0010 0048 SPSR_EL1 64 spsr[KVM_SPSR_EL1] (alias SPSR_SVC) 2477 0x6030 0000 0010 004a SPSR_ABT 64 spsr[KVM_SPSR_ABT] 2478 0x6030 0000 0010 004c SPSR_UND 64 spsr[KVM_SPSR_UND] 2479 0x6030 0000 0010 004e SPSR_IRQ 64 spsr[KVM_SPSR_IRQ] 2480 0x6060 0000 0010 0050 SPSR_FIQ 64 spsr[KVM_SPSR_FIQ] 2481 0x6040 0000 0010 0054 V0 128 fp_regs.vregs[0] [1]_ 2482 0x6040 0000 0010 0058 V1 128 fp_regs.vregs[1] [1]_ 2483 ... 2484 0x6040 0000 0010 00d0 V31 128 fp_regs.vregs[31] [1]_ 2485 0x6020 0000 0010 00d4 FPSR 32 fp_regs.fpsr 2486 0x6020 0000 0010 00d5 FPCR 32 fp_regs.fpcr 2487======================= ========= ===== ======================================= 2488 2489.. [1] These encodings are not accepted for SVE-enabled vcpus. See 2490 KVM_ARM_VCPU_INIT. 2491 2492 The equivalent register content can be accessed via bits [127:0] of 2493 the corresponding SVE Zn registers instead for vcpus that have SVE 2494 enabled (see below). 2495 2496arm64 CCSIDR registers are demultiplexed by CSSELR value:: 2497 2498 0x6020 0000 0011 00 <csselr:8> 2499 2500arm64 system registers have the following id bit patterns:: 2501 2502 0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3> 2503 2504.. warning:: 2505 2506 Two system register IDs do not follow the specified pattern. These 2507 are KVM_REG_ARM_TIMER_CVAL and KVM_REG_ARM_TIMER_CNT, which map to 2508 system registers CNTV_CVAL_EL0 and CNTVCT_EL0 respectively. These 2509 two had their values accidentally swapped, which means TIMER_CVAL is 2510 derived from the register encoding for CNTVCT_EL0 and TIMER_CNT is 2511 derived from the register encoding for CNTV_CVAL_EL0. As this is 2512 API, it must remain this way. 2513 2514arm64 firmware pseudo-registers have the following bit pattern:: 2515 2516 0x6030 0000 0014 <regno:16> 2517 2518arm64 SVE registers have the following bit patterns:: 2519 2520 0x6080 0000 0015 00 <n:5> <slice:5> Zn bits[2048*slice + 2047 : 2048*slice] 2521 0x6050 0000 0015 04 <n:4> <slice:5> Pn bits[256*slice + 255 : 256*slice] 2522 0x6050 0000 0015 060 <slice:5> FFR bits[256*slice + 255 : 256*slice] 2523 0x6060 0000 0015 ffff KVM_REG_ARM64_SVE_VLS pseudo-register 2524 2525Access to register IDs where 2048 * slice >= 128 * max_vq will fail with 2526ENOENT. max_vq is the vcpu's maximum supported vector length in 128-bit 2527quadwords: see [2]_ below. 2528 2529These registers are only accessible on vcpus for which SVE is enabled. 2530See KVM_ARM_VCPU_INIT for details. 2531 2532In addition, except for KVM_REG_ARM64_SVE_VLS, these registers are not 2533accessible until the vcpu's SVE configuration has been finalized 2534using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). See KVM_ARM_VCPU_INIT 2535and KVM_ARM_VCPU_FINALIZE for more information about this procedure. 2536 2537KVM_REG_ARM64_SVE_VLS is a pseudo-register that allows the set of vector 2538lengths supported by the vcpu to be discovered and configured by 2539userspace. When transferred to or from user memory via KVM_GET_ONE_REG 2540or KVM_SET_ONE_REG, the value of this register is of type 2541__u64[KVM_ARM64_SVE_VLS_WORDS], and encodes the set of vector lengths as 2542follows:: 2543 2544 __u64 vector_lengths[KVM_ARM64_SVE_VLS_WORDS]; 2545 2546 if (vq >= SVE_VQ_MIN && vq <= SVE_VQ_MAX && 2547 ((vector_lengths[(vq - KVM_ARM64_SVE_VQ_MIN) / 64] >> 2548 ((vq - KVM_ARM64_SVE_VQ_MIN) % 64)) & 1)) 2549 /* Vector length vq * 16 bytes supported */ 2550 else 2551 /* Vector length vq * 16 bytes not supported */ 2552 2553.. [2] The maximum value vq for which the above condition is true is 2554 max_vq. This is the maximum vector length available to the guest on 2555 this vcpu, and determines which register slices are visible through 2556 this ioctl interface. 2557 2558(See Documentation/arm64/sve.rst for an explanation of the "vq" 2559nomenclature.) 2560 2561KVM_REG_ARM64_SVE_VLS is only accessible after KVM_ARM_VCPU_INIT. 2562KVM_ARM_VCPU_INIT initialises it to the best set of vector lengths that 2563the host supports. 2564 2565Userspace may subsequently modify it if desired until the vcpu's SVE 2566configuration is finalized using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). 2567 2568Apart from simply removing all vector lengths from the host set that 2569exceed some value, support for arbitrarily chosen sets of vector lengths 2570is hardware-dependent and may not be available. Attempting to configure 2571an invalid set of vector lengths via KVM_SET_ONE_REG will fail with 2572EINVAL. 2573 2574After the vcpu's SVE configuration is finalized, further attempts to 2575write this register will fail with EPERM. 2576 2577 2578MIPS registers are mapped using the lower 32 bits. The upper 16 of that is 2579the register group type: 2580 2581MIPS core registers (see above) have the following id bit patterns:: 2582 2583 0x7030 0000 0000 <reg:16> 2584 2585MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit 2586patterns depending on whether they're 32-bit or 64-bit registers:: 2587 2588 0x7020 0000 0001 00 <reg:5> <sel:3> (32-bit) 2589 0x7030 0000 0001 00 <reg:5> <sel:3> (64-bit) 2590 2591Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_MIPS_CP0_ENTRYLO1 are the MIPS64 2592versions of the EntryLo registers regardless of the word size of the host 2593hardware, host kernel, guest, and whether XPA is present in the guest, i.e. 2594with the RI and XI bits (if they exist) in bits 63 and 62 respectively, and 2595the PFNX field starting at bit 30. 2596 2597MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) above) have the following id bit 2598patterns:: 2599 2600 0x7030 0000 0001 01 <reg:8> 2601 2602MIPS KVM control registers (see above) have the following id bit patterns:: 2603 2604 0x7030 0000 0002 <reg:16> 2605 2606MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following 2607id bit patterns depending on the size of the register being accessed. They are 2608always accessed according to the current guest FPU mode (Status.FR and 2609Config5.FRE), i.e. as the guest would see them, and they become unpredictable 2610if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector 2611registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they 2612overlap the FPU registers:: 2613 2614 0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers) 2615 0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers) 2616 0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers) 2617 2618MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the 2619following id bit patterns:: 2620 2621 0x7020 0000 0003 01 <0:3> <reg:5> 2622 2623MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the 2624following id bit patterns:: 2625 2626 0x7020 0000 0003 02 <0:3> <reg:5> 2627 2628RISC-V registers are mapped using the lower 32 bits. The upper 8 bits of 2629that is the register group type. 2630 2631RISC-V config registers are meant for configuring a Guest VCPU and it has 2632the following id bit patterns:: 2633 2634 0x8020 0000 01 <index into the kvm_riscv_config struct:24> (32bit Host) 2635 0x8030 0000 01 <index into the kvm_riscv_config struct:24> (64bit Host) 2636 2637Following are the RISC-V config registers: 2638 2639======================= ========= ============================================= 2640 Encoding Register Description 2641======================= ========= ============================================= 2642 0x80x0 0000 0100 0000 isa ISA feature bitmap of Guest VCPU 2643======================= ========= ============================================= 2644 2645The isa config register can be read anytime but can only be written before 2646a Guest VCPU runs. It will have ISA feature bits matching underlying host 2647set by default. 2648 2649RISC-V core registers represent the general excution state of a Guest VCPU 2650and it has the following id bit patterns:: 2651 2652 0x8020 0000 02 <index into the kvm_riscv_core struct:24> (32bit Host) 2653 0x8030 0000 02 <index into the kvm_riscv_core struct:24> (64bit Host) 2654 2655Following are the RISC-V core registers: 2656 2657======================= ========= ============================================= 2658 Encoding Register Description 2659======================= ========= ============================================= 2660 0x80x0 0000 0200 0000 regs.pc Program counter 2661 0x80x0 0000 0200 0001 regs.ra Return address 2662 0x80x0 0000 0200 0002 regs.sp Stack pointer 2663 0x80x0 0000 0200 0003 regs.gp Global pointer 2664 0x80x0 0000 0200 0004 regs.tp Task pointer 2665 0x80x0 0000 0200 0005 regs.t0 Caller saved register 0 2666 0x80x0 0000 0200 0006 regs.t1 Caller saved register 1 2667 0x80x0 0000 0200 0007 regs.t2 Caller saved register 2 2668 0x80x0 0000 0200 0008 regs.s0 Callee saved register 0 2669 0x80x0 0000 0200 0009 regs.s1 Callee saved register 1 2670 0x80x0 0000 0200 000a regs.a0 Function argument (or return value) 0 2671 0x80x0 0000 0200 000b regs.a1 Function argument (or return value) 1 2672 0x80x0 0000 0200 000c regs.a2 Function argument 2 2673 0x80x0 0000 0200 000d regs.a3 Function argument 3 2674 0x80x0 0000 0200 000e regs.a4 Function argument 4 2675 0x80x0 0000 0200 000f regs.a5 Function argument 5 2676 0x80x0 0000 0200 0010 regs.a6 Function argument 6 2677 0x80x0 0000 0200 0011 regs.a7 Function argument 7 2678 0x80x0 0000 0200 0012 regs.s2 Callee saved register 2 2679 0x80x0 0000 0200 0013 regs.s3 Callee saved register 3 2680 0x80x0 0000 0200 0014 regs.s4 Callee saved register 4 2681 0x80x0 0000 0200 0015 regs.s5 Callee saved register 5 2682 0x80x0 0000 0200 0016 regs.s6 Callee saved register 6 2683 0x80x0 0000 0200 0017 regs.s7 Callee saved register 7 2684 0x80x0 0000 0200 0018 regs.s8 Callee saved register 8 2685 0x80x0 0000 0200 0019 regs.s9 Callee saved register 9 2686 0x80x0 0000 0200 001a regs.s10 Callee saved register 10 2687 0x80x0 0000 0200 001b regs.s11 Callee saved register 11 2688 0x80x0 0000 0200 001c regs.t3 Caller saved register 3 2689 0x80x0 0000 0200 001d regs.t4 Caller saved register 4 2690 0x80x0 0000 0200 001e regs.t5 Caller saved register 5 2691 0x80x0 0000 0200 001f regs.t6 Caller saved register 6 2692 0x80x0 0000 0200 0020 mode Privilege mode (1 = S-mode or 0 = U-mode) 2693======================= ========= ============================================= 2694 2695RISC-V csr registers represent the supervisor mode control/status registers 2696of a Guest VCPU and it has the following id bit patterns:: 2697 2698 0x8020 0000 03 <index into the kvm_riscv_csr struct:24> (32bit Host) 2699 0x8030 0000 03 <index into the kvm_riscv_csr struct:24> (64bit Host) 2700 2701Following are the RISC-V csr registers: 2702 2703======================= ========= ============================================= 2704 Encoding Register Description 2705======================= ========= ============================================= 2706 0x80x0 0000 0300 0000 sstatus Supervisor status 2707 0x80x0 0000 0300 0001 sie Supervisor interrupt enable 2708 0x80x0 0000 0300 0002 stvec Supervisor trap vector base 2709 0x80x0 0000 0300 0003 sscratch Supervisor scratch register 2710 0x80x0 0000 0300 0004 sepc Supervisor exception program counter 2711 0x80x0 0000 0300 0005 scause Supervisor trap cause 2712 0x80x0 0000 0300 0006 stval Supervisor bad address or instruction 2713 0x80x0 0000 0300 0007 sip Supervisor interrupt pending 2714 0x80x0 0000 0300 0008 satp Supervisor address translation and protection 2715======================= ========= ============================================= 2716 2717RISC-V timer registers represent the timer state of a Guest VCPU and it has 2718the following id bit patterns:: 2719 2720 0x8030 0000 04 <index into the kvm_riscv_timer struct:24> 2721 2722Following are the RISC-V timer registers: 2723 2724======================= ========= ============================================= 2725 Encoding Register Description 2726======================= ========= ============================================= 2727 0x8030 0000 0400 0000 frequency Time base frequency (read-only) 2728 0x8030 0000 0400 0001 time Time value visible to Guest 2729 0x8030 0000 0400 0002 compare Time compare programmed by Guest 2730 0x8030 0000 0400 0003 state Time compare state (1 = ON or 0 = OFF) 2731======================= ========= ============================================= 2732 2733RISC-V F-extension registers represent the single precision floating point 2734state of a Guest VCPU and it has the following id bit patterns:: 2735 2736 0x8020 0000 05 <index into the __riscv_f_ext_state struct:24> 2737 2738Following are the RISC-V F-extension registers: 2739 2740======================= ========= ============================================= 2741 Encoding Register Description 2742======================= ========= ============================================= 2743 0x8020 0000 0500 0000 f[0] Floating point register 0 2744 ... 2745 0x8020 0000 0500 001f f[31] Floating point register 31 2746 0x8020 0000 0500 0020 fcsr Floating point control and status register 2747======================= ========= ============================================= 2748 2749RISC-V D-extension registers represent the double precision floating point 2750state of a Guest VCPU and it has the following id bit patterns:: 2751 2752 0x8020 0000 06 <index into the __riscv_d_ext_state struct:24> (fcsr) 2753 0x8030 0000 06 <index into the __riscv_d_ext_state struct:24> (non-fcsr) 2754 2755Following are the RISC-V D-extension registers: 2756 2757======================= ========= ============================================= 2758 Encoding Register Description 2759======================= ========= ============================================= 2760 0x8030 0000 0600 0000 f[0] Floating point register 0 2761 ... 2762 0x8030 0000 0600 001f f[31] Floating point register 31 2763 0x8020 0000 0600 0020 fcsr Floating point control and status register 2764======================= ========= ============================================= 2765 2766 27674.69 KVM_GET_ONE_REG 2768-------------------- 2769 2770:Capability: KVM_CAP_ONE_REG 2771:Architectures: all 2772:Type: vcpu ioctl 2773:Parameters: struct kvm_one_reg (in and out) 2774:Returns: 0 on success, negative value on failure 2775 2776Errors include: 2777 2778 ======== ============================================================ 2779 ENOENT no such register 2780 EINVAL invalid register ID, or no such register or used with VMs in 2781 protected virtualization mode on s390 2782 EPERM (arm64) register access not allowed before vcpu finalization 2783 ======== ============================================================ 2784 2785(These error codes are indicative only: do not rely on a specific error 2786code being returned in a specific situation.) 2787 2788This ioctl allows to receive the value of a single register implemented 2789in a vcpu. The register to read is indicated by the "id" field of the 2790kvm_one_reg struct passed in. On success, the register value can be found 2791at the memory location pointed to by "addr". 2792 2793The list of registers accessible using this interface is identical to the 2794list in 4.68. 2795 2796 27974.70 KVM_KVMCLOCK_CTRL 2798---------------------- 2799 2800:Capability: KVM_CAP_KVMCLOCK_CTRL 2801:Architectures: Any that implement pvclocks (currently x86 only) 2802:Type: vcpu ioctl 2803:Parameters: None 2804:Returns: 0 on success, -1 on error 2805 2806This ioctl sets a flag accessible to the guest indicating that the specified 2807vCPU has been paused by the host userspace. 2808 2809The host will set a flag in the pvclock structure that is checked from the 2810soft lockup watchdog. The flag is part of the pvclock structure that is 2811shared between guest and host, specifically the second bit of the flags 2812field of the pvclock_vcpu_time_info structure. It will be set exclusively by 2813the host and read/cleared exclusively by the guest. The guest operation of 2814checking and clearing the flag must be an atomic operation so 2815load-link/store-conditional, or equivalent must be used. There are two cases 2816where the guest will clear the flag: when the soft lockup watchdog timer resets 2817itself or when a soft lockup is detected. This ioctl can be called any time 2818after pausing the vcpu, but before it is resumed. 2819 2820 28214.71 KVM_SIGNAL_MSI 2822------------------- 2823 2824:Capability: KVM_CAP_SIGNAL_MSI 2825:Architectures: x86 arm arm64 2826:Type: vm ioctl 2827:Parameters: struct kvm_msi (in) 2828:Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error 2829 2830Directly inject a MSI message. Only valid with in-kernel irqchip that handles 2831MSI messages. 2832 2833:: 2834 2835 struct kvm_msi { 2836 __u32 address_lo; 2837 __u32 address_hi; 2838 __u32 data; 2839 __u32 flags; 2840 __u32 devid; 2841 __u8 pad[12]; 2842 }; 2843 2844flags: 2845 KVM_MSI_VALID_DEVID: devid contains a valid value. The per-VM 2846 KVM_CAP_MSI_DEVID capability advertises the requirement to provide 2847 the device ID. If this capability is not available, userspace 2848 should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail. 2849 2850If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier 2851for the device that wrote the MSI message. For PCI, this is usually a 2852BFD identifier in the lower 16 bits. 2853 2854On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS 2855feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled, 2856address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of 2857address_hi must be zero. 2858 2859 28604.71 KVM_CREATE_PIT2 2861-------------------- 2862 2863:Capability: KVM_CAP_PIT2 2864:Architectures: x86 2865:Type: vm ioctl 2866:Parameters: struct kvm_pit_config (in) 2867:Returns: 0 on success, -1 on error 2868 2869Creates an in-kernel device model for the i8254 PIT. This call is only valid 2870after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following 2871parameters have to be passed:: 2872 2873 struct kvm_pit_config { 2874 __u32 flags; 2875 __u32 pad[15]; 2876 }; 2877 2878Valid flags are:: 2879 2880 #define KVM_PIT_SPEAKER_DUMMY 1 /* emulate speaker port stub */ 2881 2882PIT timer interrupts may use a per-VM kernel thread for injection. If it 2883exists, this thread will have a name of the following pattern:: 2884 2885 kvm-pit/<owner-process-pid> 2886 2887When running a guest with elevated priorities, the scheduling parameters of 2888this thread may have to be adjusted accordingly. 2889 2890This IOCTL replaces the obsolete KVM_CREATE_PIT. 2891 2892 28934.72 KVM_GET_PIT2 2894----------------- 2895 2896:Capability: KVM_CAP_PIT_STATE2 2897:Architectures: x86 2898:Type: vm ioctl 2899:Parameters: struct kvm_pit_state2 (out) 2900:Returns: 0 on success, -1 on error 2901 2902Retrieves the state of the in-kernel PIT model. Only valid after 2903KVM_CREATE_PIT2. The state is returned in the following structure:: 2904 2905 struct kvm_pit_state2 { 2906 struct kvm_pit_channel_state channels[3]; 2907 __u32 flags; 2908 __u32 reserved[9]; 2909 }; 2910 2911Valid flags are:: 2912 2913 /* disable PIT in HPET legacy mode */ 2914 #define KVM_PIT_FLAGS_HPET_LEGACY 0x00000001 2915 2916This IOCTL replaces the obsolete KVM_GET_PIT. 2917 2918 29194.73 KVM_SET_PIT2 2920----------------- 2921 2922:Capability: KVM_CAP_PIT_STATE2 2923:Architectures: x86 2924:Type: vm ioctl 2925:Parameters: struct kvm_pit_state2 (in) 2926:Returns: 0 on success, -1 on error 2927 2928Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2. 2929See KVM_GET_PIT2 for details on struct kvm_pit_state2. 2930 2931This IOCTL replaces the obsolete KVM_SET_PIT. 2932 2933 29344.74 KVM_PPC_GET_SMMU_INFO 2935-------------------------- 2936 2937:Capability: KVM_CAP_PPC_GET_SMMU_INFO 2938:Architectures: powerpc 2939:Type: vm ioctl 2940:Parameters: None 2941:Returns: 0 on success, -1 on error 2942 2943This populates and returns a structure describing the features of 2944the "Server" class MMU emulation supported by KVM. 2945This can in turn be used by userspace to generate the appropriate 2946device-tree properties for the guest operating system. 2947 2948The structure contains some global information, followed by an 2949array of supported segment page sizes:: 2950 2951 struct kvm_ppc_smmu_info { 2952 __u64 flags; 2953 __u32 slb_size; 2954 __u32 pad; 2955 struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ]; 2956 }; 2957 2958The supported flags are: 2959 2960 - KVM_PPC_PAGE_SIZES_REAL: 2961 When that flag is set, guest page sizes must "fit" the backing 2962 store page sizes. When not set, any page size in the list can 2963 be used regardless of how they are backed by userspace. 2964 2965 - KVM_PPC_1T_SEGMENTS 2966 The emulated MMU supports 1T segments in addition to the 2967 standard 256M ones. 2968 2969 - KVM_PPC_NO_HASH 2970 This flag indicates that HPT guests are not supported by KVM, 2971 thus all guests must use radix MMU mode. 2972 2973The "slb_size" field indicates how many SLB entries are supported 2974 2975The "sps" array contains 8 entries indicating the supported base 2976page sizes for a segment in increasing order. Each entry is defined 2977as follow:: 2978 2979 struct kvm_ppc_one_seg_page_size { 2980 __u32 page_shift; /* Base page shift of segment (or 0) */ 2981 __u32 slb_enc; /* SLB encoding for BookS */ 2982 struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ]; 2983 }; 2984 2985An entry with a "page_shift" of 0 is unused. Because the array is 2986organized in increasing order, a lookup can stop when encoutering 2987such an entry. 2988 2989The "slb_enc" field provides the encoding to use in the SLB for the 2990page size. The bits are in positions such as the value can directly 2991be OR'ed into the "vsid" argument of the slbmte instruction. 2992 2993The "enc" array is a list which for each of those segment base page 2994size provides the list of supported actual page sizes (which can be 2995only larger or equal to the base page size), along with the 2996corresponding encoding in the hash PTE. Similarly, the array is 29978 entries sorted by increasing sizes and an entry with a "0" shift 2998is an empty entry and a terminator:: 2999 3000 struct kvm_ppc_one_page_size { 3001 __u32 page_shift; /* Page shift (or 0) */ 3002 __u32 pte_enc; /* Encoding in the HPTE (>>12) */ 3003 }; 3004 3005The "pte_enc" field provides a value that can OR'ed into the hash 3006PTE's RPN field (ie, it needs to be shifted left by 12 to OR it 3007into the hash PTE second double word). 3008 30094.75 KVM_IRQFD 3010-------------- 3011 3012:Capability: KVM_CAP_IRQFD 3013:Architectures: x86 s390 arm arm64 3014:Type: vm ioctl 3015:Parameters: struct kvm_irqfd (in) 3016:Returns: 0 on success, -1 on error 3017 3018Allows setting an eventfd to directly trigger a guest interrupt. 3019kvm_irqfd.fd specifies the file descriptor to use as the eventfd and 3020kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When 3021an event is triggered on the eventfd, an interrupt is injected into 3022the guest using the specified gsi pin. The irqfd is removed using 3023the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd 3024and kvm_irqfd.gsi. 3025 3026With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify 3027mechanism allowing emulation of level-triggered, irqfd-based 3028interrupts. When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an 3029additional eventfd in the kvm_irqfd.resamplefd field. When operating 3030in resample mode, posting of an interrupt through kvm_irq.fd asserts 3031the specified gsi in the irqchip. When the irqchip is resampled, such 3032as from an EOI, the gsi is de-asserted and the user is notified via 3033kvm_irqfd.resamplefd. It is the user's responsibility to re-queue 3034the interrupt if the device making use of it still requires service. 3035Note that closing the resamplefd is not sufficient to disable the 3036irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment 3037and need not be specified with KVM_IRQFD_FLAG_DEASSIGN. 3038 3039On arm/arm64, gsi routing being supported, the following can happen: 3040 3041- in case no routing entry is associated to this gsi, injection fails 3042- in case the gsi is associated to an irqchip routing entry, 3043 irqchip.pin + 32 corresponds to the injected SPI ID. 3044- in case the gsi is associated to an MSI routing entry, the MSI 3045 message and device ID are translated into an LPI (support restricted 3046 to GICv3 ITS in-kernel emulation). 3047 30484.76 KVM_PPC_ALLOCATE_HTAB 3049-------------------------- 3050 3051:Capability: KVM_CAP_PPC_ALLOC_HTAB 3052:Architectures: powerpc 3053:Type: vm ioctl 3054:Parameters: Pointer to u32 containing hash table order (in/out) 3055:Returns: 0 on success, -1 on error 3056 3057This requests the host kernel to allocate an MMU hash table for a 3058guest using the PAPR paravirtualization interface. This only does 3059anything if the kernel is configured to use the Book 3S HV style of 3060virtualization. Otherwise the capability doesn't exist and the ioctl 3061returns an ENOTTY error. The rest of this description assumes Book 3S 3062HV. 3063 3064There must be no vcpus running when this ioctl is called; if there 3065are, it will do nothing and return an EBUSY error. 3066 3067The parameter is a pointer to a 32-bit unsigned integer variable 3068containing the order (log base 2) of the desired size of the hash 3069table, which must be between 18 and 46. On successful return from the 3070ioctl, the value will not be changed by the kernel. 3071 3072If no hash table has been allocated when any vcpu is asked to run 3073(with the KVM_RUN ioctl), the host kernel will allocate a 3074default-sized hash table (16 MB). 3075 3076If this ioctl is called when a hash table has already been allocated, 3077with a different order from the existing hash table, the existing hash 3078table will be freed and a new one allocated. If this is ioctl is 3079called when a hash table has already been allocated of the same order 3080as specified, the kernel will clear out the existing hash table (zero 3081all HPTEs). In either case, if the guest is using the virtualized 3082real-mode area (VRMA) facility, the kernel will re-create the VMRA 3083HPTEs on the next KVM_RUN of any vcpu. 3084 30854.77 KVM_S390_INTERRUPT 3086----------------------- 3087 3088:Capability: basic 3089:Architectures: s390 3090:Type: vm ioctl, vcpu ioctl 3091:Parameters: struct kvm_s390_interrupt (in) 3092:Returns: 0 on success, -1 on error 3093 3094Allows to inject an interrupt to the guest. Interrupts can be floating 3095(vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type. 3096 3097Interrupt parameters are passed via kvm_s390_interrupt:: 3098 3099 struct kvm_s390_interrupt { 3100 __u32 type; 3101 __u32 parm; 3102 __u64 parm64; 3103 }; 3104 3105type can be one of the following: 3106 3107KVM_S390_SIGP_STOP (vcpu) 3108 - sigp stop; optional flags in parm 3109KVM_S390_PROGRAM_INT (vcpu) 3110 - program check; code in parm 3111KVM_S390_SIGP_SET_PREFIX (vcpu) 3112 - sigp set prefix; prefix address in parm 3113KVM_S390_RESTART (vcpu) 3114 - restart 3115KVM_S390_INT_CLOCK_COMP (vcpu) 3116 - clock comparator interrupt 3117KVM_S390_INT_CPU_TIMER (vcpu) 3118 - CPU timer interrupt 3119KVM_S390_INT_VIRTIO (vm) 3120 - virtio external interrupt; external interrupt 3121 parameters in parm and parm64 3122KVM_S390_INT_SERVICE (vm) 3123 - sclp external interrupt; sclp parameter in parm 3124KVM_S390_INT_EMERGENCY (vcpu) 3125 - sigp emergency; source cpu in parm 3126KVM_S390_INT_EXTERNAL_CALL (vcpu) 3127 - sigp external call; source cpu in parm 3128KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm) 3129 - compound value to indicate an 3130 I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel); 3131 I/O interruption parameters in parm (subchannel) and parm64 (intparm, 3132 interruption subclass) 3133KVM_S390_MCHK (vm, vcpu) 3134 - machine check interrupt; cr 14 bits in parm, machine check interrupt 3135 code in parm64 (note that machine checks needing further payload are not 3136 supported by this ioctl) 3137 3138This is an asynchronous vcpu ioctl and can be invoked from any thread. 3139 31404.78 KVM_PPC_GET_HTAB_FD 3141------------------------ 3142 3143:Capability: KVM_CAP_PPC_HTAB_FD 3144:Architectures: powerpc 3145:Type: vm ioctl 3146:Parameters: Pointer to struct kvm_get_htab_fd (in) 3147:Returns: file descriptor number (>= 0) on success, -1 on error 3148 3149This returns a file descriptor that can be used either to read out the 3150entries in the guest's hashed page table (HPT), or to write entries to 3151initialize the HPT. The returned fd can only be written to if the 3152KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and 3153can only be read if that bit is clear. The argument struct looks like 3154this:: 3155 3156 /* For KVM_PPC_GET_HTAB_FD */ 3157 struct kvm_get_htab_fd { 3158 __u64 flags; 3159 __u64 start_index; 3160 __u64 reserved[2]; 3161 }; 3162 3163 /* Values for kvm_get_htab_fd.flags */ 3164 #define KVM_GET_HTAB_BOLTED_ONLY ((__u64)0x1) 3165 #define KVM_GET_HTAB_WRITE ((__u64)0x2) 3166 3167The 'start_index' field gives the index in the HPT of the entry at 3168which to start reading. It is ignored when writing. 3169 3170Reads on the fd will initially supply information about all 3171"interesting" HPT entries. Interesting entries are those with the 3172bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise 3173all entries. When the end of the HPT is reached, the read() will 3174return. If read() is called again on the fd, it will start again from 3175the beginning of the HPT, but will only return HPT entries that have 3176changed since they were last read. 3177 3178Data read or written is structured as a header (8 bytes) followed by a 3179series of valid HPT entries (16 bytes) each. The header indicates how 3180many valid HPT entries there are and how many invalid entries follow 3181the valid entries. The invalid entries are not represented explicitly 3182in the stream. The header format is:: 3183 3184 struct kvm_get_htab_header { 3185 __u32 index; 3186 __u16 n_valid; 3187 __u16 n_invalid; 3188 }; 3189 3190Writes to the fd create HPT entries starting at the index given in the 3191header; first 'n_valid' valid entries with contents from the data 3192written, then 'n_invalid' invalid entries, invalidating any previously 3193valid entries found. 3194 31954.79 KVM_CREATE_DEVICE 3196---------------------- 3197 3198:Capability: KVM_CAP_DEVICE_CTRL 3199:Type: vm ioctl 3200:Parameters: struct kvm_create_device (in/out) 3201:Returns: 0 on success, -1 on error 3202 3203Errors: 3204 3205 ====== ======================================================= 3206 ENODEV The device type is unknown or unsupported 3207 EEXIST Device already created, and this type of device may not 3208 be instantiated multiple times 3209 ====== ======================================================= 3210 3211 Other error conditions may be defined by individual device types or 3212 have their standard meanings. 3213 3214Creates an emulated device in the kernel. The file descriptor returned 3215in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR. 3216 3217If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the 3218device type is supported (not necessarily whether it can be created 3219in the current vm). 3220 3221Individual devices should not define flags. Attributes should be used 3222for specifying any behavior that is not implied by the device type 3223number. 3224 3225:: 3226 3227 struct kvm_create_device { 3228 __u32 type; /* in: KVM_DEV_TYPE_xxx */ 3229 __u32 fd; /* out: device handle */ 3230 __u32 flags; /* in: KVM_CREATE_DEVICE_xxx */ 3231 }; 3232 32334.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR 3234-------------------------------------------- 3235 3236:Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device, 3237 KVM_CAP_VCPU_ATTRIBUTES for vcpu device 3238:Type: device ioctl, vm ioctl, vcpu ioctl 3239:Parameters: struct kvm_device_attr 3240:Returns: 0 on success, -1 on error 3241 3242Errors: 3243 3244 ===== ============================================================= 3245 ENXIO The group or attribute is unknown/unsupported for this device 3246 or hardware support is missing. 3247 EPERM The attribute cannot (currently) be accessed this way 3248 (e.g. read-only attribute, or attribute that only makes 3249 sense when the device is in a different state) 3250 ===== ============================================================= 3251 3252 Other error conditions may be defined by individual device types. 3253 3254Gets/sets a specified piece of device configuration and/or state. The 3255semantics are device-specific. See individual device documentation in 3256the "devices" directory. As with ONE_REG, the size of the data 3257transferred is defined by the particular attribute. 3258 3259:: 3260 3261 struct kvm_device_attr { 3262 __u32 flags; /* no flags currently defined */ 3263 __u32 group; /* device-defined */ 3264 __u64 attr; /* group-defined */ 3265 __u64 addr; /* userspace address of attr data */ 3266 }; 3267 32684.81 KVM_HAS_DEVICE_ATTR 3269------------------------ 3270 3271:Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device, 3272 KVM_CAP_VCPU_ATTRIBUTES for vcpu device 3273:Type: device ioctl, vm ioctl, vcpu ioctl 3274:Parameters: struct kvm_device_attr 3275:Returns: 0 on success, -1 on error 3276 3277Errors: 3278 3279 ===== ============================================================= 3280 ENXIO The group or attribute is unknown/unsupported for this device 3281 or hardware support is missing. 3282 ===== ============================================================= 3283 3284Tests whether a device supports a particular attribute. A successful 3285return indicates the attribute is implemented. It does not necessarily 3286indicate that the attribute can be read or written in the device's 3287current state. "addr" is ignored. 3288 32894.82 KVM_ARM_VCPU_INIT 3290---------------------- 3291 3292:Capability: basic 3293:Architectures: arm, arm64 3294:Type: vcpu ioctl 3295:Parameters: struct kvm_vcpu_init (in) 3296:Returns: 0 on success; -1 on error 3297 3298Errors: 3299 3300 ====== ================================================================= 3301 EINVAL the target is unknown, or the combination of features is invalid. 3302 ENOENT a features bit specified is unknown. 3303 ====== ================================================================= 3304 3305This tells KVM what type of CPU to present to the guest, and what 3306optional features it should have. This will cause a reset of the cpu 3307registers to their initial values. If this is not called, KVM_RUN will 3308return ENOEXEC for that vcpu. 3309 3310The initial values are defined as: 3311 - Processor state: 3312 * AArch64: EL1h, D, A, I and F bits set. All other bits 3313 are cleared. 3314 * AArch32: SVC, A, I and F bits set. All other bits are 3315 cleared. 3316 - General Purpose registers, including PC and SP: set to 0 3317 - FPSIMD/NEON registers: set to 0 3318 - SVE registers: set to 0 3319 - System registers: Reset to their architecturally defined 3320 values as for a warm reset to EL1 (resp. SVC) 3321 3322Note that because some registers reflect machine topology, all vcpus 3323should be created before this ioctl is invoked. 3324 3325Userspace can call this function multiple times for a given vcpu, including 3326after the vcpu has been run. This will reset the vcpu to its initial 3327state. All calls to this function after the initial call must use the same 3328target and same set of feature flags, otherwise EINVAL will be returned. 3329 3330Possible features: 3331 3332 - KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state. 3333 Depends on KVM_CAP_ARM_PSCI. If not set, the CPU will be powered on 3334 and execute guest code when KVM_RUN is called. 3335 - KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode. 3336 Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only). 3337 - KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 (or a future revision 3338 backward compatible with v0.2) for the CPU. 3339 Depends on KVM_CAP_ARM_PSCI_0_2. 3340 - KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU. 3341 Depends on KVM_CAP_ARM_PMU_V3. 3342 3343 - KVM_ARM_VCPU_PTRAUTH_ADDRESS: Enables Address Pointer authentication 3344 for arm64 only. 3345 Depends on KVM_CAP_ARM_PTRAUTH_ADDRESS. 3346 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are 3347 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and 3348 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be 3349 requested. 3350 3351 - KVM_ARM_VCPU_PTRAUTH_GENERIC: Enables Generic Pointer authentication 3352 for arm64 only. 3353 Depends on KVM_CAP_ARM_PTRAUTH_GENERIC. 3354 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are 3355 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and 3356 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be 3357 requested. 3358 3359 - KVM_ARM_VCPU_SVE: Enables SVE for the CPU (arm64 only). 3360 Depends on KVM_CAP_ARM_SVE. 3361 Requires KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3362 3363 * After KVM_ARM_VCPU_INIT: 3364 3365 - KVM_REG_ARM64_SVE_VLS may be read using KVM_GET_ONE_REG: the 3366 initial value of this pseudo-register indicates the best set of 3367 vector lengths possible for a vcpu on this host. 3368 3369 * Before KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3370 3371 - KVM_RUN and KVM_GET_REG_LIST are not available; 3372 3373 - KVM_GET_ONE_REG and KVM_SET_ONE_REG cannot be used to access 3374 the scalable archietctural SVE registers 3375 KVM_REG_ARM64_SVE_ZREG(), KVM_REG_ARM64_SVE_PREG() or 3376 KVM_REG_ARM64_SVE_FFR; 3377 3378 - KVM_REG_ARM64_SVE_VLS may optionally be written using 3379 KVM_SET_ONE_REG, to modify the set of vector lengths available 3380 for the vcpu. 3381 3382 * After KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3383 3384 - the KVM_REG_ARM64_SVE_VLS pseudo-register is immutable, and can 3385 no longer be written using KVM_SET_ONE_REG. 3386 33874.83 KVM_ARM_PREFERRED_TARGET 3388----------------------------- 3389 3390:Capability: basic 3391:Architectures: arm, arm64 3392:Type: vm ioctl 3393:Parameters: struct kvm_vcpu_init (out) 3394:Returns: 0 on success; -1 on error 3395 3396Errors: 3397 3398 ====== ========================================== 3399 ENODEV no preferred target available for the host 3400 ====== ========================================== 3401 3402This queries KVM for preferred CPU target type which can be emulated 3403by KVM on underlying host. 3404 3405The ioctl returns struct kvm_vcpu_init instance containing information 3406about preferred CPU target type and recommended features for it. The 3407kvm_vcpu_init->features bitmap returned will have feature bits set if 3408the preferred target recommends setting these features, but this is 3409not mandatory. 3410 3411The information returned by this ioctl can be used to prepare an instance 3412of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in 3413VCPU matching underlying host. 3414 3415 34164.84 KVM_GET_REG_LIST 3417--------------------- 3418 3419:Capability: basic 3420:Architectures: arm, arm64, mips 3421:Type: vcpu ioctl 3422:Parameters: struct kvm_reg_list (in/out) 3423:Returns: 0 on success; -1 on error 3424 3425Errors: 3426 3427 ===== ============================================================== 3428 E2BIG the reg index list is too big to fit in the array specified by 3429 the user (the number required will be written into n). 3430 ===== ============================================================== 3431 3432:: 3433 3434 struct kvm_reg_list { 3435 __u64 n; /* number of registers in reg[] */ 3436 __u64 reg[0]; 3437 }; 3438 3439This ioctl returns the guest registers that are supported for the 3440KVM_GET_ONE_REG/KVM_SET_ONE_REG calls. 3441 3442 34434.85 KVM_ARM_SET_DEVICE_ADDR (deprecated) 3444----------------------------------------- 3445 3446:Capability: KVM_CAP_ARM_SET_DEVICE_ADDR 3447:Architectures: arm, arm64 3448:Type: vm ioctl 3449:Parameters: struct kvm_arm_device_address (in) 3450:Returns: 0 on success, -1 on error 3451 3452Errors: 3453 3454 ====== ============================================ 3455 ENODEV The device id is unknown 3456 ENXIO Device not supported on current system 3457 EEXIST Address already set 3458 E2BIG Address outside guest physical address space 3459 EBUSY Address overlaps with other device range 3460 ====== ============================================ 3461 3462:: 3463 3464 struct kvm_arm_device_addr { 3465 __u64 id; 3466 __u64 addr; 3467 }; 3468 3469Specify a device address in the guest's physical address space where guests 3470can access emulated or directly exposed devices, which the host kernel needs 3471to know about. The id field is an architecture specific identifier for a 3472specific device. 3473 3474ARM/arm64 divides the id field into two parts, a device id and an 3475address type id specific to the individual device:: 3476 3477 bits: | 63 ... 32 | 31 ... 16 | 15 ... 0 | 3478 field: | 0x00000000 | device id | addr type id | 3479 3480ARM/arm64 currently only require this when using the in-kernel GIC 3481support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2 3482as the device id. When setting the base address for the guest's 3483mapping of the VGIC virtual CPU and distributor interface, the ioctl 3484must be called after calling KVM_CREATE_IRQCHIP, but before calling 3485KVM_RUN on any of the VCPUs. Calling this ioctl twice for any of the 3486base addresses will return -EEXIST. 3487 3488Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API 3489should be used instead. 3490 3491 34924.86 KVM_PPC_RTAS_DEFINE_TOKEN 3493------------------------------ 3494 3495:Capability: KVM_CAP_PPC_RTAS 3496:Architectures: ppc 3497:Type: vm ioctl 3498:Parameters: struct kvm_rtas_token_args 3499:Returns: 0 on success, -1 on error 3500 3501Defines a token value for a RTAS (Run Time Abstraction Services) 3502service in order to allow it to be handled in the kernel. The 3503argument struct gives the name of the service, which must be the name 3504of a service that has a kernel-side implementation. If the token 3505value is non-zero, it will be associated with that service, and 3506subsequent RTAS calls by the guest specifying that token will be 3507handled by the kernel. If the token value is 0, then any token 3508associated with the service will be forgotten, and subsequent RTAS 3509calls by the guest for that service will be passed to userspace to be 3510handled. 3511 35124.87 KVM_SET_GUEST_DEBUG 3513------------------------ 3514 3515:Capability: KVM_CAP_SET_GUEST_DEBUG 3516:Architectures: x86, s390, ppc, arm64 3517:Type: vcpu ioctl 3518:Parameters: struct kvm_guest_debug (in) 3519:Returns: 0 on success; -1 on error 3520 3521:: 3522 3523 struct kvm_guest_debug { 3524 __u32 control; 3525 __u32 pad; 3526 struct kvm_guest_debug_arch arch; 3527 }; 3528 3529Set up the processor specific debug registers and configure vcpu for 3530handling guest debug events. There are two parts to the structure, the 3531first a control bitfield indicates the type of debug events to handle 3532when running. Common control bits are: 3533 3534 - KVM_GUESTDBG_ENABLE: guest debugging is enabled 3535 - KVM_GUESTDBG_SINGLESTEP: the next run should single-step 3536 3537The top 16 bits of the control field are architecture specific control 3538flags which can include the following: 3539 3540 - KVM_GUESTDBG_USE_SW_BP: using software breakpoints [x86, arm64] 3541 - KVM_GUESTDBG_USE_HW_BP: using hardware breakpoints [x86, s390] 3542 - KVM_GUESTDBG_USE_HW: using hardware debug events [arm64] 3543 - KVM_GUESTDBG_INJECT_DB: inject DB type exception [x86] 3544 - KVM_GUESTDBG_INJECT_BP: inject BP type exception [x86] 3545 - KVM_GUESTDBG_EXIT_PENDING: trigger an immediate guest exit [s390] 3546 - KVM_GUESTDBG_BLOCKIRQ: avoid injecting interrupts/NMI/SMI [x86] 3547 3548For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints 3549are enabled in memory so we need to ensure breakpoint exceptions are 3550correctly trapped and the KVM run loop exits at the breakpoint and not 3551running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP 3552we need to ensure the guest vCPUs architecture specific registers are 3553updated to the correct (supplied) values. 3554 3555The second part of the structure is architecture specific and 3556typically contains a set of debug registers. 3557 3558For arm64 the number of debug registers is implementation defined and 3559can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and 3560KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number 3561indicating the number of supported registers. 3562 3563For ppc, the KVM_CAP_PPC_GUEST_DEBUG_SSTEP capability indicates whether 3564the single-step debug event (KVM_GUESTDBG_SINGLESTEP) is supported. 3565 3566Also when supported, KVM_CAP_SET_GUEST_DEBUG2 capability indicates the 3567supported KVM_GUESTDBG_* bits in the control field. 3568 3569When debug events exit the main run loop with the reason 3570KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run 3571structure containing architecture specific debug information. 3572 35734.88 KVM_GET_EMULATED_CPUID 3574--------------------------- 3575 3576:Capability: KVM_CAP_EXT_EMUL_CPUID 3577:Architectures: x86 3578:Type: system ioctl 3579:Parameters: struct kvm_cpuid2 (in/out) 3580:Returns: 0 on success, -1 on error 3581 3582:: 3583 3584 struct kvm_cpuid2 { 3585 __u32 nent; 3586 __u32 flags; 3587 struct kvm_cpuid_entry2 entries[0]; 3588 }; 3589 3590The member 'flags' is used for passing flags from userspace. 3591 3592:: 3593 3594 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0) 3595 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */ 3596 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */ 3597 3598 struct kvm_cpuid_entry2 { 3599 __u32 function; 3600 __u32 index; 3601 __u32 flags; 3602 __u32 eax; 3603 __u32 ebx; 3604 __u32 ecx; 3605 __u32 edx; 3606 __u32 padding[3]; 3607 }; 3608 3609This ioctl returns x86 cpuid features which are emulated by 3610kvm.Userspace can use the information returned by this ioctl to query 3611which features are emulated by kvm instead of being present natively. 3612 3613Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2 3614structure with the 'nent' field indicating the number of entries in 3615the variable-size array 'entries'. If the number of entries is too low 3616to describe the cpu capabilities, an error (E2BIG) is returned. If the 3617number is too high, the 'nent' field is adjusted and an error (ENOMEM) 3618is returned. If the number is just right, the 'nent' field is adjusted 3619to the number of valid entries in the 'entries' array, which is then 3620filled. 3621 3622The entries returned are the set CPUID bits of the respective features 3623which kvm emulates, as returned by the CPUID instruction, with unknown 3624or unsupported feature bits cleared. 3625 3626Features like x2apic, for example, may not be present in the host cpu 3627but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be 3628emulated efficiently and thus not included here. 3629 3630The fields in each entry are defined as follows: 3631 3632 function: 3633 the eax value used to obtain the entry 3634 index: 3635 the ecx value used to obtain the entry (for entries that are 3636 affected by ecx) 3637 flags: 3638 an OR of zero or more of the following: 3639 3640 KVM_CPUID_FLAG_SIGNIFCANT_INDEX: 3641 if the index field is valid 3642 3643 eax, ebx, ecx, edx: 3644 3645 the values returned by the cpuid instruction for 3646 this function/index combination 3647 36484.89 KVM_S390_MEM_OP 3649-------------------- 3650 3651:Capability: KVM_CAP_S390_MEM_OP 3652:Architectures: s390 3653:Type: vcpu ioctl 3654:Parameters: struct kvm_s390_mem_op (in) 3655:Returns: = 0 on success, 3656 < 0 on generic error (e.g. -EFAULT or -ENOMEM), 3657 > 0 if an exception occurred while walking the page tables 3658 3659Read or write data from/to the logical (virtual) memory of a VCPU. 3660 3661Parameters are specified via the following structure:: 3662 3663 struct kvm_s390_mem_op { 3664 __u64 gaddr; /* the guest address */ 3665 __u64 flags; /* flags */ 3666 __u32 size; /* amount of bytes */ 3667 __u32 op; /* type of operation */ 3668 __u64 buf; /* buffer in userspace */ 3669 __u8 ar; /* the access register number */ 3670 __u8 reserved[31]; /* should be set to 0 */ 3671 }; 3672 3673The type of operation is specified in the "op" field. It is either 3674KVM_S390_MEMOP_LOGICAL_READ for reading from logical memory space or 3675KVM_S390_MEMOP_LOGICAL_WRITE for writing to logical memory space. The 3676KVM_S390_MEMOP_F_CHECK_ONLY flag can be set in the "flags" field to check 3677whether the corresponding memory access would create an access exception 3678(without touching the data in the memory at the destination). In case an 3679access exception occurred while walking the MMU tables of the guest, the 3680ioctl returns a positive error number to indicate the type of exception. 3681This exception is also raised directly at the corresponding VCPU if the 3682flag KVM_S390_MEMOP_F_INJECT_EXCEPTION is set in the "flags" field. 3683 3684The start address of the memory region has to be specified in the "gaddr" 3685field, and the length of the region in the "size" field (which must not 3686be 0). The maximum value for "size" can be obtained by checking the 3687KVM_CAP_S390_MEM_OP capability. "buf" is the buffer supplied by the 3688userspace application where the read data should be written to for 3689KVM_S390_MEMOP_LOGICAL_READ, or where the data that should be written is 3690stored for a KVM_S390_MEMOP_LOGICAL_WRITE. When KVM_S390_MEMOP_F_CHECK_ONLY 3691is specified, "buf" is unused and can be NULL. "ar" designates the access 3692register number to be used; the valid range is 0..15. 3693 3694The "reserved" field is meant for future extensions. It is not used by 3695KVM with the currently defined set of flags. 3696 36974.90 KVM_S390_GET_SKEYS 3698----------------------- 3699 3700:Capability: KVM_CAP_S390_SKEYS 3701:Architectures: s390 3702:Type: vm ioctl 3703:Parameters: struct kvm_s390_skeys 3704:Returns: 0 on success, KVM_S390_GET_KEYS_NONE if guest is not using storage 3705 keys, negative value on error 3706 3707This ioctl is used to get guest storage key values on the s390 3708architecture. The ioctl takes parameters via the kvm_s390_skeys struct:: 3709 3710 struct kvm_s390_skeys { 3711 __u64 start_gfn; 3712 __u64 count; 3713 __u64 skeydata_addr; 3714 __u32 flags; 3715 __u32 reserved[9]; 3716 }; 3717 3718The start_gfn field is the number of the first guest frame whose storage keys 3719you want to get. 3720 3721The count field is the number of consecutive frames (starting from start_gfn) 3722whose storage keys to get. The count field must be at least 1 and the maximum 3723allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range 3724will cause the ioctl to return -EINVAL. 3725 3726The skeydata_addr field is the address to a buffer large enough to hold count 3727bytes. This buffer will be filled with storage key data by the ioctl. 3728 37294.91 KVM_S390_SET_SKEYS 3730----------------------- 3731 3732:Capability: KVM_CAP_S390_SKEYS 3733:Architectures: s390 3734:Type: vm ioctl 3735:Parameters: struct kvm_s390_skeys 3736:Returns: 0 on success, negative value on error 3737 3738This ioctl is used to set guest storage key values on the s390 3739architecture. The ioctl takes parameters via the kvm_s390_skeys struct. 3740See section on KVM_S390_GET_SKEYS for struct definition. 3741 3742The start_gfn field is the number of the first guest frame whose storage keys 3743you want to set. 3744 3745The count field is the number of consecutive frames (starting from start_gfn) 3746whose storage keys to get. The count field must be at least 1 and the maximum 3747allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range 3748will cause the ioctl to return -EINVAL. 3749 3750The skeydata_addr field is the address to a buffer containing count bytes of 3751storage keys. Each byte in the buffer will be set as the storage key for a 3752single frame starting at start_gfn for count frames. 3753 3754Note: If any architecturally invalid key value is found in the given data then 3755the ioctl will return -EINVAL. 3756 37574.92 KVM_S390_IRQ 3758----------------- 3759 3760:Capability: KVM_CAP_S390_INJECT_IRQ 3761:Architectures: s390 3762:Type: vcpu ioctl 3763:Parameters: struct kvm_s390_irq (in) 3764:Returns: 0 on success, -1 on error 3765 3766Errors: 3767 3768 3769 ====== ================================================================= 3770 EINVAL interrupt type is invalid 3771 type is KVM_S390_SIGP_STOP and flag parameter is invalid value, 3772 type is KVM_S390_INT_EXTERNAL_CALL and code is bigger 3773 than the maximum of VCPUs 3774 EBUSY type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped, 3775 type is KVM_S390_SIGP_STOP and a stop irq is already pending, 3776 type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt 3777 is already pending 3778 ====== ================================================================= 3779 3780Allows to inject an interrupt to the guest. 3781 3782Using struct kvm_s390_irq as a parameter allows 3783to inject additional payload which is not 3784possible via KVM_S390_INTERRUPT. 3785 3786Interrupt parameters are passed via kvm_s390_irq:: 3787 3788 struct kvm_s390_irq { 3789 __u64 type; 3790 union { 3791 struct kvm_s390_io_info io; 3792 struct kvm_s390_ext_info ext; 3793 struct kvm_s390_pgm_info pgm; 3794 struct kvm_s390_emerg_info emerg; 3795 struct kvm_s390_extcall_info extcall; 3796 struct kvm_s390_prefix_info prefix; 3797 struct kvm_s390_stop_info stop; 3798 struct kvm_s390_mchk_info mchk; 3799 char reserved[64]; 3800 } u; 3801 }; 3802 3803type can be one of the following: 3804 3805- KVM_S390_SIGP_STOP - sigp stop; parameter in .stop 3806- KVM_S390_PROGRAM_INT - program check; parameters in .pgm 3807- KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix 3808- KVM_S390_RESTART - restart; no parameters 3809- KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters 3810- KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters 3811- KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg 3812- KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall 3813- KVM_S390_MCHK - machine check interrupt; parameters in .mchk 3814 3815This is an asynchronous vcpu ioctl and can be invoked from any thread. 3816 38174.94 KVM_S390_GET_IRQ_STATE 3818--------------------------- 3819 3820:Capability: KVM_CAP_S390_IRQ_STATE 3821:Architectures: s390 3822:Type: vcpu ioctl 3823:Parameters: struct kvm_s390_irq_state (out) 3824:Returns: >= number of bytes copied into buffer, 3825 -EINVAL if buffer size is 0, 3826 -ENOBUFS if buffer size is too small to fit all pending interrupts, 3827 -EFAULT if the buffer address was invalid 3828 3829This ioctl allows userspace to retrieve the complete state of all currently 3830pending interrupts in a single buffer. Use cases include migration 3831and introspection. The parameter structure contains the address of a 3832userspace buffer and its length:: 3833 3834 struct kvm_s390_irq_state { 3835 __u64 buf; 3836 __u32 flags; /* will stay unused for compatibility reasons */ 3837 __u32 len; 3838 __u32 reserved[4]; /* will stay unused for compatibility reasons */ 3839 }; 3840 3841Userspace passes in the above struct and for each pending interrupt a 3842struct kvm_s390_irq is copied to the provided buffer. 3843 3844The structure contains a flags and a reserved field for future extensions. As 3845the kernel never checked for flags == 0 and QEMU never pre-zeroed flags and 3846reserved, these fields can not be used in the future without breaking 3847compatibility. 3848 3849If -ENOBUFS is returned the buffer provided was too small and userspace 3850may retry with a bigger buffer. 3851 38524.95 KVM_S390_SET_IRQ_STATE 3853--------------------------- 3854 3855:Capability: KVM_CAP_S390_IRQ_STATE 3856:Architectures: s390 3857:Type: vcpu ioctl 3858:Parameters: struct kvm_s390_irq_state (in) 3859:Returns: 0 on success, 3860 -EFAULT if the buffer address was invalid, 3861 -EINVAL for an invalid buffer length (see below), 3862 -EBUSY if there were already interrupts pending, 3863 errors occurring when actually injecting the 3864 interrupt. See KVM_S390_IRQ. 3865 3866This ioctl allows userspace to set the complete state of all cpu-local 3867interrupts currently pending for the vcpu. It is intended for restoring 3868interrupt state after a migration. The input parameter is a userspace buffer 3869containing a struct kvm_s390_irq_state:: 3870 3871 struct kvm_s390_irq_state { 3872 __u64 buf; 3873 __u32 flags; /* will stay unused for compatibility reasons */ 3874 __u32 len; 3875 __u32 reserved[4]; /* will stay unused for compatibility reasons */ 3876 }; 3877 3878The restrictions for flags and reserved apply as well. 3879(see KVM_S390_GET_IRQ_STATE) 3880 3881The userspace memory referenced by buf contains a struct kvm_s390_irq 3882for each interrupt to be injected into the guest. 3883If one of the interrupts could not be injected for some reason the 3884ioctl aborts. 3885 3886len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0 3887and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq), 3888which is the maximum number of possibly pending cpu-local interrupts. 3889 38904.96 KVM_SMI 3891------------ 3892 3893:Capability: KVM_CAP_X86_SMM 3894:Architectures: x86 3895:Type: vcpu ioctl 3896:Parameters: none 3897:Returns: 0 on success, -1 on error 3898 3899Queues an SMI on the thread's vcpu. 3900 39014.97 KVM_X86_SET_MSR_FILTER 3902---------------------------- 3903 3904:Capability: KVM_X86_SET_MSR_FILTER 3905:Architectures: x86 3906:Type: vm ioctl 3907:Parameters: struct kvm_msr_filter 3908:Returns: 0 on success, < 0 on error 3909 3910:: 3911 3912 struct kvm_msr_filter_range { 3913 #define KVM_MSR_FILTER_READ (1 << 0) 3914 #define KVM_MSR_FILTER_WRITE (1 << 1) 3915 __u32 flags; 3916 __u32 nmsrs; /* number of msrs in bitmap */ 3917 __u32 base; /* MSR index the bitmap starts at */ 3918 __u8 *bitmap; /* a 1 bit allows the operations in flags, 0 denies */ 3919 }; 3920 3921 #define KVM_MSR_FILTER_MAX_RANGES 16 3922 struct kvm_msr_filter { 3923 #define KVM_MSR_FILTER_DEFAULT_ALLOW (0 << 0) 3924 #define KVM_MSR_FILTER_DEFAULT_DENY (1 << 0) 3925 __u32 flags; 3926 struct kvm_msr_filter_range ranges[KVM_MSR_FILTER_MAX_RANGES]; 3927 }; 3928 3929flags values for ``struct kvm_msr_filter_range``: 3930 3931``KVM_MSR_FILTER_READ`` 3932 3933 Filter read accesses to MSRs using the given bitmap. A 0 in the bitmap 3934 indicates that a read should immediately fail, while a 1 indicates that 3935 a read for a particular MSR should be handled regardless of the default 3936 filter action. 3937 3938``KVM_MSR_FILTER_WRITE`` 3939 3940 Filter write accesses to MSRs using the given bitmap. A 0 in the bitmap 3941 indicates that a write should immediately fail, while a 1 indicates that 3942 a write for a particular MSR should be handled regardless of the default 3943 filter action. 3944 3945``KVM_MSR_FILTER_READ | KVM_MSR_FILTER_WRITE`` 3946 3947 Filter both read and write accesses to MSRs using the given bitmap. A 0 3948 in the bitmap indicates that both reads and writes should immediately fail, 3949 while a 1 indicates that reads and writes for a particular MSR are not 3950 filtered by this range. 3951 3952flags values for ``struct kvm_msr_filter``: 3953 3954``KVM_MSR_FILTER_DEFAULT_ALLOW`` 3955 3956 If no filter range matches an MSR index that is getting accessed, KVM will 3957 fall back to allowing access to the MSR. 3958 3959``KVM_MSR_FILTER_DEFAULT_DENY`` 3960 3961 If no filter range matches an MSR index that is getting accessed, KVM will 3962 fall back to rejecting access to the MSR. In this mode, all MSRs that should 3963 be processed by KVM need to explicitly be marked as allowed in the bitmaps. 3964 3965This ioctl allows user space to define up to 16 bitmaps of MSR ranges to 3966specify whether a certain MSR access should be explicitly filtered for or not. 3967 3968If this ioctl has never been invoked, MSR accesses are not guarded and the 3969default KVM in-kernel emulation behavior is fully preserved. 3970 3971Calling this ioctl with an empty set of ranges (all nmsrs == 0) disables MSR 3972filtering. In that mode, ``KVM_MSR_FILTER_DEFAULT_DENY`` is invalid and causes 3973an error. 3974 3975As soon as the filtering is in place, every MSR access is processed through 3976the filtering except for accesses to the x2APIC MSRs (from 0x800 to 0x8ff); 3977x2APIC MSRs are always allowed, independent of the ``default_allow`` setting, 3978and their behavior depends on the ``X2APIC_ENABLE`` bit of the APIC base 3979register. 3980 3981If a bit is within one of the defined ranges, read and write accesses are 3982guarded by the bitmap's value for the MSR index if the kind of access 3983is included in the ``struct kvm_msr_filter_range`` flags. If no range 3984cover this particular access, the behavior is determined by the flags 3985field in the kvm_msr_filter struct: ``KVM_MSR_FILTER_DEFAULT_ALLOW`` 3986and ``KVM_MSR_FILTER_DEFAULT_DENY``. 3987 3988Each bitmap range specifies a range of MSRs to potentially allow access on. 3989The range goes from MSR index [base .. base+nmsrs]. The flags field 3990indicates whether reads, writes or both reads and writes are filtered 3991by setting a 1 bit in the bitmap for the corresponding MSR index. 3992 3993If an MSR access is not permitted through the filtering, it generates a 3994#GP inside the guest. When combined with KVM_CAP_X86_USER_SPACE_MSR, that 3995allows user space to deflect and potentially handle various MSR accesses 3996into user space. 3997 3998If a vCPU is in running state while this ioctl is invoked, the vCPU may 3999experience inconsistent filtering behavior on MSR accesses. 4000 40014.98 KVM_CREATE_SPAPR_TCE_64 4002---------------------------- 4003 4004:Capability: KVM_CAP_SPAPR_TCE_64 4005:Architectures: powerpc 4006:Type: vm ioctl 4007:Parameters: struct kvm_create_spapr_tce_64 (in) 4008:Returns: file descriptor for manipulating the created TCE table 4009 4010This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit 4011windows, described in 4.62 KVM_CREATE_SPAPR_TCE 4012 4013This capability uses extended struct in ioctl interface:: 4014 4015 /* for KVM_CAP_SPAPR_TCE_64 */ 4016 struct kvm_create_spapr_tce_64 { 4017 __u64 liobn; 4018 __u32 page_shift; 4019 __u32 flags; 4020 __u64 offset; /* in pages */ 4021 __u64 size; /* in pages */ 4022 }; 4023 4024The aim of extension is to support an additional bigger DMA window with 4025a variable page size. 4026KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and 4027a bus offset of the corresponding DMA window, @size and @offset are numbers 4028of IOMMU pages. 4029 4030@flags are not used at the moment. 4031 4032The rest of functionality is identical to KVM_CREATE_SPAPR_TCE. 4033 40344.99 KVM_REINJECT_CONTROL 4035------------------------- 4036 4037:Capability: KVM_CAP_REINJECT_CONTROL 4038:Architectures: x86 4039:Type: vm ioctl 4040:Parameters: struct kvm_reinject_control (in) 4041:Returns: 0 on success, 4042 -EFAULT if struct kvm_reinject_control cannot be read, 4043 -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier. 4044 4045i8254 (PIT) has two modes, reinject and !reinject. The default is reinject, 4046where KVM queues elapsed i8254 ticks and monitors completion of interrupt from 4047vector(s) that i8254 injects. Reinject mode dequeues a tick and injects its 4048interrupt whenever there isn't a pending interrupt from i8254. 4049!reinject mode injects an interrupt as soon as a tick arrives. 4050 4051:: 4052 4053 struct kvm_reinject_control { 4054 __u8 pit_reinject; 4055 __u8 reserved[31]; 4056 }; 4057 4058pit_reinject = 0 (!reinject mode) is recommended, unless running an old 4059operating system that uses the PIT for timing (e.g. Linux 2.4.x). 4060 40614.100 KVM_PPC_CONFIGURE_V3_MMU 4062------------------------------ 4063 4064:Capability: KVM_CAP_PPC_RADIX_MMU or KVM_CAP_PPC_HASH_MMU_V3 4065:Architectures: ppc 4066:Type: vm ioctl 4067:Parameters: struct kvm_ppc_mmuv3_cfg (in) 4068:Returns: 0 on success, 4069 -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read, 4070 -EINVAL if the configuration is invalid 4071 4072This ioctl controls whether the guest will use radix or HPT (hashed 4073page table) translation, and sets the pointer to the process table for 4074the guest. 4075 4076:: 4077 4078 struct kvm_ppc_mmuv3_cfg { 4079 __u64 flags; 4080 __u64 process_table; 4081 }; 4082 4083There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and 4084KVM_PPC_MMUV3_GTSE. KVM_PPC_MMUV3_RADIX, if set, configures the guest 4085to use radix tree translation, and if clear, to use HPT translation. 4086KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest 4087to be able to use the global TLB and SLB invalidation instructions; 4088if clear, the guest may not use these instructions. 4089 4090The process_table field specifies the address and size of the guest 4091process table, which is in the guest's space. This field is formatted 4092as the second doubleword of the partition table entry, as defined in 4093the Power ISA V3.00, Book III section 5.7.6.1. 4094 40954.101 KVM_PPC_GET_RMMU_INFO 4096--------------------------- 4097 4098:Capability: KVM_CAP_PPC_RADIX_MMU 4099:Architectures: ppc 4100:Type: vm ioctl 4101:Parameters: struct kvm_ppc_rmmu_info (out) 4102:Returns: 0 on success, 4103 -EFAULT if struct kvm_ppc_rmmu_info cannot be written, 4104 -EINVAL if no useful information can be returned 4105 4106This ioctl returns a structure containing two things: (a) a list 4107containing supported radix tree geometries, and (b) a list that maps 4108page sizes to put in the "AP" (actual page size) field for the tlbie 4109(TLB invalidate entry) instruction. 4110 4111:: 4112 4113 struct kvm_ppc_rmmu_info { 4114 struct kvm_ppc_radix_geom { 4115 __u8 page_shift; 4116 __u8 level_bits[4]; 4117 __u8 pad[3]; 4118 } geometries[8]; 4119 __u32 ap_encodings[8]; 4120 }; 4121 4122The geometries[] field gives up to 8 supported geometries for the 4123radix page table, in terms of the log base 2 of the smallest page 4124size, and the number of bits indexed at each level of the tree, from 4125the PTE level up to the PGD level in that order. Any unused entries 4126will have 0 in the page_shift field. 4127 4128The ap_encodings gives the supported page sizes and their AP field 4129encodings, encoded with the AP value in the top 3 bits and the log 4130base 2 of the page size in the bottom 6 bits. 4131 41324.102 KVM_PPC_RESIZE_HPT_PREPARE 4133-------------------------------- 4134 4135:Capability: KVM_CAP_SPAPR_RESIZE_HPT 4136:Architectures: powerpc 4137:Type: vm ioctl 4138:Parameters: struct kvm_ppc_resize_hpt (in) 4139:Returns: 0 on successful completion, 4140 >0 if a new HPT is being prepared, the value is an estimated 4141 number of milliseconds until preparation is complete, 4142 -EFAULT if struct kvm_reinject_control cannot be read, 4143 -EINVAL if the supplied shift or flags are invalid, 4144 -ENOMEM if unable to allocate the new HPT, 4145 4146Used to implement the PAPR extension for runtime resizing of a guest's 4147Hashed Page Table (HPT). Specifically this starts, stops or monitors 4148the preparation of a new potential HPT for the guest, essentially 4149implementing the H_RESIZE_HPT_PREPARE hypercall. 4150 4151:: 4152 4153 struct kvm_ppc_resize_hpt { 4154 __u64 flags; 4155 __u32 shift; 4156 __u32 pad; 4157 }; 4158 4159If called with shift > 0 when there is no pending HPT for the guest, 4160this begins preparation of a new pending HPT of size 2^(shift) bytes. 4161It then returns a positive integer with the estimated number of 4162milliseconds until preparation is complete. 4163 4164If called when there is a pending HPT whose size does not match that 4165requested in the parameters, discards the existing pending HPT and 4166creates a new one as above. 4167 4168If called when there is a pending HPT of the size requested, will: 4169 4170 * If preparation of the pending HPT is already complete, return 0 4171 * If preparation of the pending HPT has failed, return an error 4172 code, then discard the pending HPT. 4173 * If preparation of the pending HPT is still in progress, return an 4174 estimated number of milliseconds until preparation is complete. 4175 4176If called with shift == 0, discards any currently pending HPT and 4177returns 0 (i.e. cancels any in-progress preparation). 4178 4179flags is reserved for future expansion, currently setting any bits in 4180flags will result in an -EINVAL. 4181 4182Normally this will be called repeatedly with the same parameters until 4183it returns <= 0. The first call will initiate preparation, subsequent 4184ones will monitor preparation until it completes or fails. 4185 41864.103 KVM_PPC_RESIZE_HPT_COMMIT 4187------------------------------- 4188 4189:Capability: KVM_CAP_SPAPR_RESIZE_HPT 4190:Architectures: powerpc 4191:Type: vm ioctl 4192:Parameters: struct kvm_ppc_resize_hpt (in) 4193:Returns: 0 on successful completion, 4194 -EFAULT if struct kvm_reinject_control cannot be read, 4195 -EINVAL if the supplied shift or flags are invalid, 4196 -ENXIO is there is no pending HPT, or the pending HPT doesn't 4197 have the requested size, 4198 -EBUSY if the pending HPT is not fully prepared, 4199 -ENOSPC if there was a hash collision when moving existing 4200 HPT entries to the new HPT, 4201 -EIO on other error conditions 4202 4203Used to implement the PAPR extension for runtime resizing of a guest's 4204Hashed Page Table (HPT). Specifically this requests that the guest be 4205transferred to working with the new HPT, essentially implementing the 4206H_RESIZE_HPT_COMMIT hypercall. 4207 4208:: 4209 4210 struct kvm_ppc_resize_hpt { 4211 __u64 flags; 4212 __u32 shift; 4213 __u32 pad; 4214 }; 4215 4216This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has 4217returned 0 with the same parameters. In other cases 4218KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or 4219-EBUSY, though others may be possible if the preparation was started, 4220but failed). 4221 4222This will have undefined effects on the guest if it has not already 4223placed itself in a quiescent state where no vcpu will make MMU enabled 4224memory accesses. 4225 4226On succsful completion, the pending HPT will become the guest's active 4227HPT and the previous HPT will be discarded. 4228 4229On failure, the guest will still be operating on its previous HPT. 4230 42314.104 KVM_X86_GET_MCE_CAP_SUPPORTED 4232----------------------------------- 4233 4234:Capability: KVM_CAP_MCE 4235:Architectures: x86 4236:Type: system ioctl 4237:Parameters: u64 mce_cap (out) 4238:Returns: 0 on success, -1 on error 4239 4240Returns supported MCE capabilities. The u64 mce_cap parameter 4241has the same format as the MSR_IA32_MCG_CAP register. Supported 4242capabilities will have the corresponding bits set. 4243 42444.105 KVM_X86_SETUP_MCE 4245----------------------- 4246 4247:Capability: KVM_CAP_MCE 4248:Architectures: x86 4249:Type: vcpu ioctl 4250:Parameters: u64 mcg_cap (in) 4251:Returns: 0 on success, 4252 -EFAULT if u64 mcg_cap cannot be read, 4253 -EINVAL if the requested number of banks is invalid, 4254 -EINVAL if requested MCE capability is not supported. 4255 4256Initializes MCE support for use. The u64 mcg_cap parameter 4257has the same format as the MSR_IA32_MCG_CAP register and 4258specifies which capabilities should be enabled. The maximum 4259supported number of error-reporting banks can be retrieved when 4260checking for KVM_CAP_MCE. The supported capabilities can be 4261retrieved with KVM_X86_GET_MCE_CAP_SUPPORTED. 4262 42634.106 KVM_X86_SET_MCE 4264--------------------- 4265 4266:Capability: KVM_CAP_MCE 4267:Architectures: x86 4268:Type: vcpu ioctl 4269:Parameters: struct kvm_x86_mce (in) 4270:Returns: 0 on success, 4271 -EFAULT if struct kvm_x86_mce cannot be read, 4272 -EINVAL if the bank number is invalid, 4273 -EINVAL if VAL bit is not set in status field. 4274 4275Inject a machine check error (MCE) into the guest. The input 4276parameter is:: 4277 4278 struct kvm_x86_mce { 4279 __u64 status; 4280 __u64 addr; 4281 __u64 misc; 4282 __u64 mcg_status; 4283 __u8 bank; 4284 __u8 pad1[7]; 4285 __u64 pad2[3]; 4286 }; 4287 4288If the MCE being reported is an uncorrected error, KVM will 4289inject it as an MCE exception into the guest. If the guest 4290MCG_STATUS register reports that an MCE is in progress, KVM 4291causes an KVM_EXIT_SHUTDOWN vmexit. 4292 4293Otherwise, if the MCE is a corrected error, KVM will just 4294store it in the corresponding bank (provided this bank is 4295not holding a previously reported uncorrected error). 4296 42974.107 KVM_S390_GET_CMMA_BITS 4298---------------------------- 4299 4300:Capability: KVM_CAP_S390_CMMA_MIGRATION 4301:Architectures: s390 4302:Type: vm ioctl 4303:Parameters: struct kvm_s390_cmma_log (in, out) 4304:Returns: 0 on success, a negative value on error 4305 4306This ioctl is used to get the values of the CMMA bits on the s390 4307architecture. It is meant to be used in two scenarios: 4308 4309- During live migration to save the CMMA values. Live migration needs 4310 to be enabled via the KVM_REQ_START_MIGRATION VM property. 4311- To non-destructively peek at the CMMA values, with the flag 4312 KVM_S390_CMMA_PEEK set. 4313 4314The ioctl takes parameters via the kvm_s390_cmma_log struct. The desired 4315values are written to a buffer whose location is indicated via the "values" 4316member in the kvm_s390_cmma_log struct. The values in the input struct are 4317also updated as needed. 4318 4319Each CMMA value takes up one byte. 4320 4321:: 4322 4323 struct kvm_s390_cmma_log { 4324 __u64 start_gfn; 4325 __u32 count; 4326 __u32 flags; 4327 union { 4328 __u64 remaining; 4329 __u64 mask; 4330 }; 4331 __u64 values; 4332 }; 4333 4334start_gfn is the number of the first guest frame whose CMMA values are 4335to be retrieved, 4336 4337count is the length of the buffer in bytes, 4338 4339values points to the buffer where the result will be written to. 4340 4341If count is greater than KVM_S390_SKEYS_MAX, then it is considered to be 4342KVM_S390_SKEYS_MAX. KVM_S390_SKEYS_MAX is re-used for consistency with 4343other ioctls. 4344 4345The result is written in the buffer pointed to by the field values, and 4346the values of the input parameter are updated as follows. 4347 4348Depending on the flags, different actions are performed. The only 4349supported flag so far is KVM_S390_CMMA_PEEK. 4350 4351The default behaviour if KVM_S390_CMMA_PEEK is not set is: 4352start_gfn will indicate the first page frame whose CMMA bits were dirty. 4353It is not necessarily the same as the one passed as input, as clean pages 4354are skipped. 4355 4356count will indicate the number of bytes actually written in the buffer. 4357It can (and very often will) be smaller than the input value, since the 4358buffer is only filled until 16 bytes of clean values are found (which 4359are then not copied in the buffer). Since a CMMA migration block needs 4360the base address and the length, for a total of 16 bytes, we will send 4361back some clean data if there is some dirty data afterwards, as long as 4362the size of the clean data does not exceed the size of the header. This 4363allows to minimize the amount of data to be saved or transferred over 4364the network at the expense of more roundtrips to userspace. The next 4365invocation of the ioctl will skip over all the clean values, saving 4366potentially more than just the 16 bytes we found. 4367 4368If KVM_S390_CMMA_PEEK is set: 4369the existing storage attributes are read even when not in migration 4370mode, and no other action is performed; 4371 4372the output start_gfn will be equal to the input start_gfn, 4373 4374the output count will be equal to the input count, except if the end of 4375memory has been reached. 4376 4377In both cases: 4378the field "remaining" will indicate the total number of dirty CMMA values 4379still remaining, or 0 if KVM_S390_CMMA_PEEK is set and migration mode is 4380not enabled. 4381 4382mask is unused. 4383 4384values points to the userspace buffer where the result will be stored. 4385 4386This ioctl can fail with -ENOMEM if not enough memory can be allocated to 4387complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if 4388KVM_S390_CMMA_PEEK is not set but migration mode was not enabled, with 4389-EFAULT if the userspace address is invalid or if no page table is 4390present for the addresses (e.g. when using hugepages). 4391 43924.108 KVM_S390_SET_CMMA_BITS 4393---------------------------- 4394 4395:Capability: KVM_CAP_S390_CMMA_MIGRATION 4396:Architectures: s390 4397:Type: vm ioctl 4398:Parameters: struct kvm_s390_cmma_log (in) 4399:Returns: 0 on success, a negative value on error 4400 4401This ioctl is used to set the values of the CMMA bits on the s390 4402architecture. It is meant to be used during live migration to restore 4403the CMMA values, but there are no restrictions on its use. 4404The ioctl takes parameters via the kvm_s390_cmma_values struct. 4405Each CMMA value takes up one byte. 4406 4407:: 4408 4409 struct kvm_s390_cmma_log { 4410 __u64 start_gfn; 4411 __u32 count; 4412 __u32 flags; 4413 union { 4414 __u64 remaining; 4415 __u64 mask; 4416 }; 4417 __u64 values; 4418 }; 4419 4420start_gfn indicates the starting guest frame number, 4421 4422count indicates how many values are to be considered in the buffer, 4423 4424flags is not used and must be 0. 4425 4426mask indicates which PGSTE bits are to be considered. 4427 4428remaining is not used. 4429 4430values points to the buffer in userspace where to store the values. 4431 4432This ioctl can fail with -ENOMEM if not enough memory can be allocated to 4433complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if 4434the count field is too large (e.g. more than KVM_S390_CMMA_SIZE_MAX) or 4435if the flags field was not 0, with -EFAULT if the userspace address is 4436invalid, if invalid pages are written to (e.g. after the end of memory) 4437or if no page table is present for the addresses (e.g. when using 4438hugepages). 4439 44404.109 KVM_PPC_GET_CPU_CHAR 4441-------------------------- 4442 4443:Capability: KVM_CAP_PPC_GET_CPU_CHAR 4444:Architectures: powerpc 4445:Type: vm ioctl 4446:Parameters: struct kvm_ppc_cpu_char (out) 4447:Returns: 0 on successful completion, 4448 -EFAULT if struct kvm_ppc_cpu_char cannot be written 4449 4450This ioctl gives userspace information about certain characteristics 4451of the CPU relating to speculative execution of instructions and 4452possible information leakage resulting from speculative execution (see 4453CVE-2017-5715, CVE-2017-5753 and CVE-2017-5754). The information is 4454returned in struct kvm_ppc_cpu_char, which looks like this:: 4455 4456 struct kvm_ppc_cpu_char { 4457 __u64 character; /* characteristics of the CPU */ 4458 __u64 behaviour; /* recommended software behaviour */ 4459 __u64 character_mask; /* valid bits in character */ 4460 __u64 behaviour_mask; /* valid bits in behaviour */ 4461 }; 4462 4463For extensibility, the character_mask and behaviour_mask fields 4464indicate which bits of character and behaviour have been filled in by 4465the kernel. If the set of defined bits is extended in future then 4466userspace will be able to tell whether it is running on a kernel that 4467knows about the new bits. 4468 4469The character field describes attributes of the CPU which can help 4470with preventing inadvertent information disclosure - specifically, 4471whether there is an instruction to flash-invalidate the L1 data cache 4472(ori 30,30,0 or mtspr SPRN_TRIG2,rN), whether the L1 data cache is set 4473to a mode where entries can only be used by the thread that created 4474them, whether the bcctr[l] instruction prevents speculation, and 4475whether a speculation barrier instruction (ori 31,31,0) is provided. 4476 4477The behaviour field describes actions that software should take to 4478prevent inadvertent information disclosure, and thus describes which 4479vulnerabilities the hardware is subject to; specifically whether the 4480L1 data cache should be flushed when returning to user mode from the 4481kernel, and whether a speculation barrier should be placed between an 4482array bounds check and the array access. 4483 4484These fields use the same bit definitions as the new 4485H_GET_CPU_CHARACTERISTICS hypercall. 4486 44874.110 KVM_MEMORY_ENCRYPT_OP 4488--------------------------- 4489 4490:Capability: basic 4491:Architectures: x86 4492:Type: vm 4493:Parameters: an opaque platform specific structure (in/out) 4494:Returns: 0 on success; -1 on error 4495 4496If the platform supports creating encrypted VMs then this ioctl can be used 4497for issuing platform-specific memory encryption commands to manage those 4498encrypted VMs. 4499 4500Currently, this ioctl is used for issuing Secure Encrypted Virtualization 4501(SEV) commands on AMD Processors. The SEV commands are defined in 4502Documentation/virt/kvm/amd-memory-encryption.rst. 4503 45044.111 KVM_MEMORY_ENCRYPT_REG_REGION 4505----------------------------------- 4506 4507:Capability: basic 4508:Architectures: x86 4509:Type: system 4510:Parameters: struct kvm_enc_region (in) 4511:Returns: 0 on success; -1 on error 4512 4513This ioctl can be used to register a guest memory region which may 4514contain encrypted data (e.g. guest RAM, SMRAM etc). 4515 4516It is used in the SEV-enabled guest. When encryption is enabled, a guest 4517memory region may contain encrypted data. The SEV memory encryption 4518engine uses a tweak such that two identical plaintext pages, each at 4519different locations will have differing ciphertexts. So swapping or 4520moving ciphertext of those pages will not result in plaintext being 4521swapped. So relocating (or migrating) physical backing pages for the SEV 4522guest will require some additional steps. 4523 4524Note: The current SEV key management spec does not provide commands to 4525swap or migrate (move) ciphertext pages. Hence, for now we pin the guest 4526memory region registered with the ioctl. 4527 45284.112 KVM_MEMORY_ENCRYPT_UNREG_REGION 4529------------------------------------- 4530 4531:Capability: basic 4532:Architectures: x86 4533:Type: system 4534:Parameters: struct kvm_enc_region (in) 4535:Returns: 0 on success; -1 on error 4536 4537This ioctl can be used to unregister the guest memory region registered 4538with KVM_MEMORY_ENCRYPT_REG_REGION ioctl above. 4539 45404.113 KVM_HYPERV_EVENTFD 4541------------------------ 4542 4543:Capability: KVM_CAP_HYPERV_EVENTFD 4544:Architectures: x86 4545:Type: vm ioctl 4546:Parameters: struct kvm_hyperv_eventfd (in) 4547 4548This ioctl (un)registers an eventfd to receive notifications from the guest on 4549the specified Hyper-V connection id through the SIGNAL_EVENT hypercall, without 4550causing a user exit. SIGNAL_EVENT hypercall with non-zero event flag number 4551(bits 24-31) still triggers a KVM_EXIT_HYPERV_HCALL user exit. 4552 4553:: 4554 4555 struct kvm_hyperv_eventfd { 4556 __u32 conn_id; 4557 __s32 fd; 4558 __u32 flags; 4559 __u32 padding[3]; 4560 }; 4561 4562The conn_id field should fit within 24 bits:: 4563 4564 #define KVM_HYPERV_CONN_ID_MASK 0x00ffffff 4565 4566The acceptable values for the flags field are:: 4567 4568 #define KVM_HYPERV_EVENTFD_DEASSIGN (1 << 0) 4569 4570:Returns: 0 on success, 4571 -EINVAL if conn_id or flags is outside the allowed range, 4572 -ENOENT on deassign if the conn_id isn't registered, 4573 -EEXIST on assign if the conn_id is already registered 4574 45754.114 KVM_GET_NESTED_STATE 4576-------------------------- 4577 4578:Capability: KVM_CAP_NESTED_STATE 4579:Architectures: x86 4580:Type: vcpu ioctl 4581:Parameters: struct kvm_nested_state (in/out) 4582:Returns: 0 on success, -1 on error 4583 4584Errors: 4585 4586 ===== ============================================================= 4587 E2BIG the total state size exceeds the value of 'size' specified by 4588 the user; the size required will be written into size. 4589 ===== ============================================================= 4590 4591:: 4592 4593 struct kvm_nested_state { 4594 __u16 flags; 4595 __u16 format; 4596 __u32 size; 4597 4598 union { 4599 struct kvm_vmx_nested_state_hdr vmx; 4600 struct kvm_svm_nested_state_hdr svm; 4601 4602 /* Pad the header to 128 bytes. */ 4603 __u8 pad[120]; 4604 } hdr; 4605 4606 union { 4607 struct kvm_vmx_nested_state_data vmx[0]; 4608 struct kvm_svm_nested_state_data svm[0]; 4609 } data; 4610 }; 4611 4612 #define KVM_STATE_NESTED_GUEST_MODE 0x00000001 4613 #define KVM_STATE_NESTED_RUN_PENDING 0x00000002 4614 #define KVM_STATE_NESTED_EVMCS 0x00000004 4615 4616 #define KVM_STATE_NESTED_FORMAT_VMX 0 4617 #define KVM_STATE_NESTED_FORMAT_SVM 1 4618 4619 #define KVM_STATE_NESTED_VMX_VMCS_SIZE 0x1000 4620 4621 #define KVM_STATE_NESTED_VMX_SMM_GUEST_MODE 0x00000001 4622 #define KVM_STATE_NESTED_VMX_SMM_VMXON 0x00000002 4623 4624 #define KVM_STATE_VMX_PREEMPTION_TIMER_DEADLINE 0x00000001 4625 4626 struct kvm_vmx_nested_state_hdr { 4627 __u64 vmxon_pa; 4628 __u64 vmcs12_pa; 4629 4630 struct { 4631 __u16 flags; 4632 } smm; 4633 4634 __u32 flags; 4635 __u64 preemption_timer_deadline; 4636 }; 4637 4638 struct kvm_vmx_nested_state_data { 4639 __u8 vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE]; 4640 __u8 shadow_vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE]; 4641 }; 4642 4643This ioctl copies the vcpu's nested virtualization state from the kernel to 4644userspace. 4645 4646The maximum size of the state can be retrieved by passing KVM_CAP_NESTED_STATE 4647to the KVM_CHECK_EXTENSION ioctl(). 4648 46494.115 KVM_SET_NESTED_STATE 4650-------------------------- 4651 4652:Capability: KVM_CAP_NESTED_STATE 4653:Architectures: x86 4654:Type: vcpu ioctl 4655:Parameters: struct kvm_nested_state (in) 4656:Returns: 0 on success, -1 on error 4657 4658This copies the vcpu's kvm_nested_state struct from userspace to the kernel. 4659For the definition of struct kvm_nested_state, see KVM_GET_NESTED_STATE. 4660 46614.116 KVM_(UN)REGISTER_COALESCED_MMIO 4662------------------------------------- 4663 4664:Capability: KVM_CAP_COALESCED_MMIO (for coalesced mmio) 4665 KVM_CAP_COALESCED_PIO (for coalesced pio) 4666:Architectures: all 4667:Type: vm ioctl 4668:Parameters: struct kvm_coalesced_mmio_zone 4669:Returns: 0 on success, < 0 on error 4670 4671Coalesced I/O is a performance optimization that defers hardware 4672register write emulation so that userspace exits are avoided. It is 4673typically used to reduce the overhead of emulating frequently accessed 4674hardware registers. 4675 4676When a hardware register is configured for coalesced I/O, write accesses 4677do not exit to userspace and their value is recorded in a ring buffer 4678that is shared between kernel and userspace. 4679 4680Coalesced I/O is used if one or more write accesses to a hardware 4681register can be deferred until a read or a write to another hardware 4682register on the same device. This last access will cause a vmexit and 4683userspace will process accesses from the ring buffer before emulating 4684it. That will avoid exiting to userspace on repeated writes. 4685 4686Coalesced pio is based on coalesced mmio. There is little difference 4687between coalesced mmio and pio except that coalesced pio records accesses 4688to I/O ports. 4689 46904.117 KVM_CLEAR_DIRTY_LOG (vm ioctl) 4691------------------------------------ 4692 4693:Capability: KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 4694:Architectures: x86, arm, arm64, mips 4695:Type: vm ioctl 4696:Parameters: struct kvm_clear_dirty_log (in) 4697:Returns: 0 on success, -1 on error 4698 4699:: 4700 4701 /* for KVM_CLEAR_DIRTY_LOG */ 4702 struct kvm_clear_dirty_log { 4703 __u32 slot; 4704 __u32 num_pages; 4705 __u64 first_page; 4706 union { 4707 void __user *dirty_bitmap; /* one bit per page */ 4708 __u64 padding; 4709 }; 4710 }; 4711 4712The ioctl clears the dirty status of pages in a memory slot, according to 4713the bitmap that is passed in struct kvm_clear_dirty_log's dirty_bitmap 4714field. Bit 0 of the bitmap corresponds to page "first_page" in the 4715memory slot, and num_pages is the size in bits of the input bitmap. 4716first_page must be a multiple of 64; num_pages must also be a multiple of 471764 unless first_page + num_pages is the size of the memory slot. For each 4718bit that is set in the input bitmap, the corresponding page is marked "clean" 4719in KVM's dirty bitmap, and dirty tracking is re-enabled for that page 4720(for example via write-protection, or by clearing the dirty bit in 4721a page table entry). 4722 4723If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies 4724the address space for which you want to clear the dirty status. See 4725KVM_SET_USER_MEMORY_REGION for details on the usage of slot field. 4726 4727This ioctl is mostly useful when KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 4728is enabled; for more information, see the description of the capability. 4729However, it can always be used as long as KVM_CHECK_EXTENSION confirms 4730that KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is present. 4731 47324.118 KVM_GET_SUPPORTED_HV_CPUID 4733-------------------------------- 4734 4735:Capability: KVM_CAP_HYPERV_CPUID (vcpu), KVM_CAP_SYS_HYPERV_CPUID (system) 4736:Architectures: x86 4737:Type: system ioctl, vcpu ioctl 4738:Parameters: struct kvm_cpuid2 (in/out) 4739:Returns: 0 on success, -1 on error 4740 4741:: 4742 4743 struct kvm_cpuid2 { 4744 __u32 nent; 4745 __u32 padding; 4746 struct kvm_cpuid_entry2 entries[0]; 4747 }; 4748 4749 struct kvm_cpuid_entry2 { 4750 __u32 function; 4751 __u32 index; 4752 __u32 flags; 4753 __u32 eax; 4754 __u32 ebx; 4755 __u32 ecx; 4756 __u32 edx; 4757 __u32 padding[3]; 4758 }; 4759 4760This ioctl returns x86 cpuid features leaves related to Hyper-V emulation in 4761KVM. Userspace can use the information returned by this ioctl to construct 4762cpuid information presented to guests consuming Hyper-V enlightenments (e.g. 4763Windows or Hyper-V guests). 4764 4765CPUID feature leaves returned by this ioctl are defined by Hyper-V Top Level 4766Functional Specification (TLFS). These leaves can't be obtained with 4767KVM_GET_SUPPORTED_CPUID ioctl because some of them intersect with KVM feature 4768leaves (0x40000000, 0x40000001). 4769 4770Currently, the following list of CPUID leaves are returned: 4771 4772 - HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS 4773 - HYPERV_CPUID_INTERFACE 4774 - HYPERV_CPUID_VERSION 4775 - HYPERV_CPUID_FEATURES 4776 - HYPERV_CPUID_ENLIGHTMENT_INFO 4777 - HYPERV_CPUID_IMPLEMENT_LIMITS 4778 - HYPERV_CPUID_NESTED_FEATURES 4779 - HYPERV_CPUID_SYNDBG_VENDOR_AND_MAX_FUNCTIONS 4780 - HYPERV_CPUID_SYNDBG_INTERFACE 4781 - HYPERV_CPUID_SYNDBG_PLATFORM_CAPABILITIES 4782 4783Userspace invokes KVM_GET_SUPPORTED_HV_CPUID by passing a kvm_cpuid2 structure 4784with the 'nent' field indicating the number of entries in the variable-size 4785array 'entries'. If the number of entries is too low to describe all Hyper-V 4786feature leaves, an error (E2BIG) is returned. If the number is more or equal 4787to the number of Hyper-V feature leaves, the 'nent' field is adjusted to the 4788number of valid entries in the 'entries' array, which is then filled. 4789 4790'index' and 'flags' fields in 'struct kvm_cpuid_entry2' are currently reserved, 4791userspace should not expect to get any particular value there. 4792 4793Note, vcpu version of KVM_GET_SUPPORTED_HV_CPUID is currently deprecated. Unlike 4794system ioctl which exposes all supported feature bits unconditionally, vcpu 4795version has the following quirks: 4796 4797- HYPERV_CPUID_NESTED_FEATURES leaf and HV_X64_ENLIGHTENED_VMCS_RECOMMENDED 4798 feature bit are only exposed when Enlightened VMCS was previously enabled 4799 on the corresponding vCPU (KVM_CAP_HYPERV_ENLIGHTENED_VMCS). 4800- HV_STIMER_DIRECT_MODE_AVAILABLE bit is only exposed with in-kernel LAPIC. 4801 (presumes KVM_CREATE_IRQCHIP has already been called). 4802 48034.119 KVM_ARM_VCPU_FINALIZE 4804--------------------------- 4805 4806:Architectures: arm, arm64 4807:Type: vcpu ioctl 4808:Parameters: int feature (in) 4809:Returns: 0 on success, -1 on error 4810 4811Errors: 4812 4813 ====== ============================================================== 4814 EPERM feature not enabled, needs configuration, or already finalized 4815 EINVAL feature unknown or not present 4816 ====== ============================================================== 4817 4818Recognised values for feature: 4819 4820 ===== =========================================== 4821 arm64 KVM_ARM_VCPU_SVE (requires KVM_CAP_ARM_SVE) 4822 ===== =========================================== 4823 4824Finalizes the configuration of the specified vcpu feature. 4825 4826The vcpu must already have been initialised, enabling the affected feature, by 4827means of a successful KVM_ARM_VCPU_INIT call with the appropriate flag set in 4828features[]. 4829 4830For affected vcpu features, this is a mandatory step that must be performed 4831before the vcpu is fully usable. 4832 4833Between KVM_ARM_VCPU_INIT and KVM_ARM_VCPU_FINALIZE, the feature may be 4834configured by use of ioctls such as KVM_SET_ONE_REG. The exact configuration 4835that should be performaned and how to do it are feature-dependent. 4836 4837Other calls that depend on a particular feature being finalized, such as 4838KVM_RUN, KVM_GET_REG_LIST, KVM_GET_ONE_REG and KVM_SET_ONE_REG, will fail with 4839-EPERM unless the feature has already been finalized by means of a 4840KVM_ARM_VCPU_FINALIZE call. 4841 4842See KVM_ARM_VCPU_INIT for details of vcpu features that require finalization 4843using this ioctl. 4844 48454.120 KVM_SET_PMU_EVENT_FILTER 4846------------------------------ 4847 4848:Capability: KVM_CAP_PMU_EVENT_FILTER 4849:Architectures: x86 4850:Type: vm ioctl 4851:Parameters: struct kvm_pmu_event_filter (in) 4852:Returns: 0 on success, -1 on error 4853 4854:: 4855 4856 struct kvm_pmu_event_filter { 4857 __u32 action; 4858 __u32 nevents; 4859 __u32 fixed_counter_bitmap; 4860 __u32 flags; 4861 __u32 pad[4]; 4862 __u64 events[0]; 4863 }; 4864 4865This ioctl restricts the set of PMU events that the guest can program. 4866The argument holds a list of events which will be allowed or denied. 4867The eventsel+umask of each event the guest attempts to program is compared 4868against the events field to determine whether the guest should have access. 4869The events field only controls general purpose counters; fixed purpose 4870counters are controlled by the fixed_counter_bitmap. 4871 4872No flags are defined yet, the field must be zero. 4873 4874Valid values for 'action':: 4875 4876 #define KVM_PMU_EVENT_ALLOW 0 4877 #define KVM_PMU_EVENT_DENY 1 4878 48794.121 KVM_PPC_SVM_OFF 4880--------------------- 4881 4882:Capability: basic 4883:Architectures: powerpc 4884:Type: vm ioctl 4885:Parameters: none 4886:Returns: 0 on successful completion, 4887 4888Errors: 4889 4890 ====== ================================================================ 4891 EINVAL if ultravisor failed to terminate the secure guest 4892 ENOMEM if hypervisor failed to allocate new radix page tables for guest 4893 ====== ================================================================ 4894 4895This ioctl is used to turn off the secure mode of the guest or transition 4896the guest from secure mode to normal mode. This is invoked when the guest 4897is reset. This has no effect if called for a normal guest. 4898 4899This ioctl issues an ultravisor call to terminate the secure guest, 4900unpins the VPA pages and releases all the device pages that are used to 4901track the secure pages by hypervisor. 4902 49034.122 KVM_S390_NORMAL_RESET 4904--------------------------- 4905 4906:Capability: KVM_CAP_S390_VCPU_RESETS 4907:Architectures: s390 4908:Type: vcpu ioctl 4909:Parameters: none 4910:Returns: 0 4911 4912This ioctl resets VCPU registers and control structures according to 4913the cpu reset definition in the POP (Principles Of Operation). 4914 49154.123 KVM_S390_INITIAL_RESET 4916---------------------------- 4917 4918:Capability: none 4919:Architectures: s390 4920:Type: vcpu ioctl 4921:Parameters: none 4922:Returns: 0 4923 4924This ioctl resets VCPU registers and control structures according to 4925the initial cpu reset definition in the POP. However, the cpu is not 4926put into ESA mode. This reset is a superset of the normal reset. 4927 49284.124 KVM_S390_CLEAR_RESET 4929-------------------------- 4930 4931:Capability: KVM_CAP_S390_VCPU_RESETS 4932:Architectures: s390 4933:Type: vcpu ioctl 4934:Parameters: none 4935:Returns: 0 4936 4937This ioctl resets VCPU registers and control structures according to 4938the clear cpu reset definition in the POP. However, the cpu is not put 4939into ESA mode. This reset is a superset of the initial reset. 4940 4941 49424.125 KVM_S390_PV_COMMAND 4943------------------------- 4944 4945:Capability: KVM_CAP_S390_PROTECTED 4946:Architectures: s390 4947:Type: vm ioctl 4948:Parameters: struct kvm_pv_cmd 4949:Returns: 0 on success, < 0 on error 4950 4951:: 4952 4953 struct kvm_pv_cmd { 4954 __u32 cmd; /* Command to be executed */ 4955 __u16 rc; /* Ultravisor return code */ 4956 __u16 rrc; /* Ultravisor return reason code */ 4957 __u64 data; /* Data or address */ 4958 __u32 flags; /* flags for future extensions. Must be 0 for now */ 4959 __u32 reserved[3]; 4960 }; 4961 4962cmd values: 4963 4964KVM_PV_ENABLE 4965 Allocate memory and register the VM with the Ultravisor, thereby 4966 donating memory to the Ultravisor that will become inaccessible to 4967 KVM. All existing CPUs are converted to protected ones. After this 4968 command has succeeded, any CPU added via hotplug will become 4969 protected during its creation as well. 4970 4971 Errors: 4972 4973 ===== ============================= 4974 EINTR an unmasked signal is pending 4975 ===== ============================= 4976 4977KVM_PV_DISABLE 4978 4979 Deregister the VM from the Ultravisor and reclaim the memory that 4980 had been donated to the Ultravisor, making it usable by the kernel 4981 again. All registered VCPUs are converted back to non-protected 4982 ones. 4983 4984KVM_PV_VM_SET_SEC_PARMS 4985 Pass the image header from VM memory to the Ultravisor in 4986 preparation of image unpacking and verification. 4987 4988KVM_PV_VM_UNPACK 4989 Unpack (protect and decrypt) a page of the encrypted boot image. 4990 4991KVM_PV_VM_VERIFY 4992 Verify the integrity of the unpacked image. Only if this succeeds, 4993 KVM is allowed to start protected VCPUs. 4994 49954.126 KVM_X86_SET_MSR_FILTER 4996---------------------------- 4997 4998:Capability: KVM_CAP_X86_MSR_FILTER 4999:Architectures: x86 5000:Type: vm ioctl 5001:Parameters: struct kvm_msr_filter 5002:Returns: 0 on success, < 0 on error 5003 5004:: 5005 5006 struct kvm_msr_filter_range { 5007 #define KVM_MSR_FILTER_READ (1 << 0) 5008 #define KVM_MSR_FILTER_WRITE (1 << 1) 5009 __u32 flags; 5010 __u32 nmsrs; /* number of msrs in bitmap */ 5011 __u32 base; /* MSR index the bitmap starts at */ 5012 __u8 *bitmap; /* a 1 bit allows the operations in flags, 0 denies */ 5013 }; 5014 5015 #define KVM_MSR_FILTER_MAX_RANGES 16 5016 struct kvm_msr_filter { 5017 #define KVM_MSR_FILTER_DEFAULT_ALLOW (0 << 0) 5018 #define KVM_MSR_FILTER_DEFAULT_DENY (1 << 0) 5019 __u32 flags; 5020 struct kvm_msr_filter_range ranges[KVM_MSR_FILTER_MAX_RANGES]; 5021 }; 5022 5023flags values for ``struct kvm_msr_filter_range``: 5024 5025``KVM_MSR_FILTER_READ`` 5026 5027 Filter read accesses to MSRs using the given bitmap. A 0 in the bitmap 5028 indicates that a read should immediately fail, while a 1 indicates that 5029 a read for a particular MSR should be handled regardless of the default 5030 filter action. 5031 5032``KVM_MSR_FILTER_WRITE`` 5033 5034 Filter write accesses to MSRs using the given bitmap. A 0 in the bitmap 5035 indicates that a write should immediately fail, while a 1 indicates that 5036 a write for a particular MSR should be handled regardless of the default 5037 filter action. 5038 5039``KVM_MSR_FILTER_READ | KVM_MSR_FILTER_WRITE`` 5040 5041 Filter both read and write accesses to MSRs using the given bitmap. A 0 5042 in the bitmap indicates that both reads and writes should immediately fail, 5043 while a 1 indicates that reads and writes for a particular MSR are not 5044 filtered by this range. 5045 5046flags values for ``struct kvm_msr_filter``: 5047 5048``KVM_MSR_FILTER_DEFAULT_ALLOW`` 5049 5050 If no filter range matches an MSR index that is getting accessed, KVM will 5051 fall back to allowing access to the MSR. 5052 5053``KVM_MSR_FILTER_DEFAULT_DENY`` 5054 5055 If no filter range matches an MSR index that is getting accessed, KVM will 5056 fall back to rejecting access to the MSR. In this mode, all MSRs that should 5057 be processed by KVM need to explicitly be marked as allowed in the bitmaps. 5058 5059This ioctl allows user space to define up to 16 bitmaps of MSR ranges to 5060specify whether a certain MSR access should be explicitly filtered for or not. 5061 5062If this ioctl has never been invoked, MSR accesses are not guarded and the 5063default KVM in-kernel emulation behavior is fully preserved. 5064 5065Calling this ioctl with an empty set of ranges (all nmsrs == 0) disables MSR 5066filtering. In that mode, ``KVM_MSR_FILTER_DEFAULT_DENY`` is invalid and causes 5067an error. 5068 5069As soon as the filtering is in place, every MSR access is processed through 5070the filtering except for accesses to the x2APIC MSRs (from 0x800 to 0x8ff); 5071x2APIC MSRs are always allowed, independent of the ``default_allow`` setting, 5072and their behavior depends on the ``X2APIC_ENABLE`` bit of the APIC base 5073register. 5074 5075If a bit is within one of the defined ranges, read and write accesses are 5076guarded by the bitmap's value for the MSR index if the kind of access 5077is included in the ``struct kvm_msr_filter_range`` flags. If no range 5078cover this particular access, the behavior is determined by the flags 5079field in the kvm_msr_filter struct: ``KVM_MSR_FILTER_DEFAULT_ALLOW`` 5080and ``KVM_MSR_FILTER_DEFAULT_DENY``. 5081 5082Each bitmap range specifies a range of MSRs to potentially allow access on. 5083The range goes from MSR index [base .. base+nmsrs]. The flags field 5084indicates whether reads, writes or both reads and writes are filtered 5085by setting a 1 bit in the bitmap for the corresponding MSR index. 5086 5087If an MSR access is not permitted through the filtering, it generates a 5088#GP inside the guest. When combined with KVM_CAP_X86_USER_SPACE_MSR, that 5089allows user space to deflect and potentially handle various MSR accesses 5090into user space. 5091 5092Note, invoking this ioctl with a vCPU is running is inherently racy. However, 5093KVM does guarantee that vCPUs will see either the previous filter or the new 5094filter, e.g. MSRs with identical settings in both the old and new filter will 5095have deterministic behavior. 5096 50974.127 KVM_XEN_HVM_SET_ATTR 5098-------------------------- 5099 5100:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 5101:Architectures: x86 5102:Type: vm ioctl 5103:Parameters: struct kvm_xen_hvm_attr 5104:Returns: 0 on success, < 0 on error 5105 5106:: 5107 5108 struct kvm_xen_hvm_attr { 5109 __u16 type; 5110 __u16 pad[3]; 5111 union { 5112 __u8 long_mode; 5113 __u8 vector; 5114 struct { 5115 __u64 gfn; 5116 } shared_info; 5117 __u64 pad[4]; 5118 } u; 5119 }; 5120 5121type values: 5122 5123KVM_XEN_ATTR_TYPE_LONG_MODE 5124 Sets the ABI mode of the VM to 32-bit or 64-bit (long mode). This 5125 determines the layout of the shared info pages exposed to the VM. 5126 5127KVM_XEN_ATTR_TYPE_SHARED_INFO 5128 Sets the guest physical frame number at which the Xen "shared info" 5129 page resides. Note that although Xen places vcpu_info for the first 5130 32 vCPUs in the shared_info page, KVM does not automatically do so 5131 and instead requires that KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO be used 5132 explicitly even when the vcpu_info for a given vCPU resides at the 5133 "default" location in the shared_info page. This is because KVM is 5134 not aware of the Xen CPU id which is used as the index into the 5135 vcpu_info[] array, so cannot know the correct default location. 5136 5137KVM_XEN_ATTR_TYPE_UPCALL_VECTOR 5138 Sets the exception vector used to deliver Xen event channel upcalls. 5139 51404.127 KVM_XEN_HVM_GET_ATTR 5141-------------------------- 5142 5143:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 5144:Architectures: x86 5145:Type: vm ioctl 5146:Parameters: struct kvm_xen_hvm_attr 5147:Returns: 0 on success, < 0 on error 5148 5149Allows Xen VM attributes to be read. For the structure and types, 5150see KVM_XEN_HVM_SET_ATTR above. 5151 51524.128 KVM_XEN_VCPU_SET_ATTR 5153--------------------------- 5154 5155:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 5156:Architectures: x86 5157:Type: vcpu ioctl 5158:Parameters: struct kvm_xen_vcpu_attr 5159:Returns: 0 on success, < 0 on error 5160 5161:: 5162 5163 struct kvm_xen_vcpu_attr { 5164 __u16 type; 5165 __u16 pad[3]; 5166 union { 5167 __u64 gpa; 5168 __u64 pad[4]; 5169 struct { 5170 __u64 state; 5171 __u64 state_entry_time; 5172 __u64 time_running; 5173 __u64 time_runnable; 5174 __u64 time_blocked; 5175 __u64 time_offline; 5176 } runstate; 5177 } u; 5178 }; 5179 5180type values: 5181 5182KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO 5183 Sets the guest physical address of the vcpu_info for a given vCPU. 5184 5185KVM_XEN_VCPU_ATTR_TYPE_VCPU_TIME_INFO 5186 Sets the guest physical address of an additional pvclock structure 5187 for a given vCPU. This is typically used for guest vsyscall support. 5188 5189KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR 5190 Sets the guest physical address of the vcpu_runstate_info for a given 5191 vCPU. This is how a Xen guest tracks CPU state such as steal time. 5192 5193KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_CURRENT 5194 Sets the runstate (RUNSTATE_running/_runnable/_blocked/_offline) of 5195 the given vCPU from the .u.runstate.state member of the structure. 5196 KVM automatically accounts running and runnable time but blocked 5197 and offline states are only entered explicitly. 5198 5199KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_DATA 5200 Sets all fields of the vCPU runstate data from the .u.runstate member 5201 of the structure, including the current runstate. The state_entry_time 5202 must equal the sum of the other four times. 5203 5204KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST 5205 This *adds* the contents of the .u.runstate members of the structure 5206 to the corresponding members of the given vCPU's runstate data, thus 5207 permitting atomic adjustments to the runstate times. The adjustment 5208 to the state_entry_time must equal the sum of the adjustments to the 5209 other four times. The state field must be set to -1, or to a valid 5210 runstate value (RUNSTATE_running, RUNSTATE_runnable, RUNSTATE_blocked 5211 or RUNSTATE_offline) to set the current accounted state as of the 5212 adjusted state_entry_time. 5213 52144.129 KVM_XEN_VCPU_GET_ATTR 5215--------------------------- 5216 5217:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 5218:Architectures: x86 5219:Type: vcpu ioctl 5220:Parameters: struct kvm_xen_vcpu_attr 5221:Returns: 0 on success, < 0 on error 5222 5223Allows Xen vCPU attributes to be read. For the structure and types, 5224see KVM_XEN_VCPU_SET_ATTR above. 5225 5226The KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST type may not be used 5227with the KVM_XEN_VCPU_GET_ATTR ioctl. 5228 52294.130 KVM_ARM_MTE_COPY_TAGS 5230--------------------------- 5231 5232:Capability: KVM_CAP_ARM_MTE 5233:Architectures: arm64 5234:Type: vm ioctl 5235:Parameters: struct kvm_arm_copy_mte_tags 5236:Returns: number of bytes copied, < 0 on error (-EINVAL for incorrect 5237 arguments, -EFAULT if memory cannot be accessed). 5238 5239:: 5240 5241 struct kvm_arm_copy_mte_tags { 5242 __u64 guest_ipa; 5243 __u64 length; 5244 void __user *addr; 5245 __u64 flags; 5246 __u64 reserved[2]; 5247 }; 5248 5249Copies Memory Tagging Extension (MTE) tags to/from guest tag memory. The 5250``guest_ipa`` and ``length`` fields must be ``PAGE_SIZE`` aligned. The ``addr`` 5251field must point to a buffer which the tags will be copied to or from. 5252 5253``flags`` specifies the direction of copy, either ``KVM_ARM_TAGS_TO_GUEST`` or 5254``KVM_ARM_TAGS_FROM_GUEST``. 5255 5256The size of the buffer to store the tags is ``(length / 16)`` bytes 5257(granules in MTE are 16 bytes long). Each byte contains a single tag 5258value. This matches the format of ``PTRACE_PEEKMTETAGS`` and 5259``PTRACE_POKEMTETAGS``. 5260 5261If an error occurs before any data is copied then a negative error code is 5262returned. If some tags have been copied before an error occurs then the number 5263of bytes successfully copied is returned. If the call completes successfully 5264then ``length`` is returned. 5265 52664.131 KVM_GET_SREGS2 5267-------------------- 5268 5269:Capability: KVM_CAP_SREGS2 5270:Architectures: x86 5271:Type: vcpu ioctl 5272:Parameters: struct kvm_sregs2 (out) 5273:Returns: 0 on success, -1 on error 5274 5275Reads special registers from the vcpu. 5276This ioctl (when supported) replaces the KVM_GET_SREGS. 5277 5278:: 5279 5280 struct kvm_sregs2 { 5281 /* out (KVM_GET_SREGS2) / in (KVM_SET_SREGS2) */ 5282 struct kvm_segment cs, ds, es, fs, gs, ss; 5283 struct kvm_segment tr, ldt; 5284 struct kvm_dtable gdt, idt; 5285 __u64 cr0, cr2, cr3, cr4, cr8; 5286 __u64 efer; 5287 __u64 apic_base; 5288 __u64 flags; 5289 __u64 pdptrs[4]; 5290 }; 5291 5292flags values for ``kvm_sregs2``: 5293 5294``KVM_SREGS2_FLAGS_PDPTRS_VALID`` 5295 5296 Indicates thats the struct contain valid PDPTR values. 5297 5298 52994.132 KVM_SET_SREGS2 5300-------------------- 5301 5302:Capability: KVM_CAP_SREGS2 5303:Architectures: x86 5304:Type: vcpu ioctl 5305:Parameters: struct kvm_sregs2 (in) 5306:Returns: 0 on success, -1 on error 5307 5308Writes special registers into the vcpu. 5309See KVM_GET_SREGS2 for the data structures. 5310This ioctl (when supported) replaces the KVM_SET_SREGS. 5311 53124.133 KVM_GET_STATS_FD 5313---------------------- 5314 5315:Capability: KVM_CAP_STATS_BINARY_FD 5316:Architectures: all 5317:Type: vm ioctl, vcpu ioctl 5318:Parameters: none 5319:Returns: statistics file descriptor on success, < 0 on error 5320 5321Errors: 5322 5323 ====== ====================================================== 5324 ENOMEM if the fd could not be created due to lack of memory 5325 EMFILE if the number of opened files exceeds the limit 5326 ====== ====================================================== 5327 5328The returned file descriptor can be used to read VM/vCPU statistics data in 5329binary format. The data in the file descriptor consists of four blocks 5330organized as follows: 5331 5332+-------------+ 5333| Header | 5334+-------------+ 5335| id string | 5336+-------------+ 5337| Descriptors | 5338+-------------+ 5339| Stats Data | 5340+-------------+ 5341 5342Apart from the header starting at offset 0, please be aware that it is 5343not guaranteed that the four blocks are adjacent or in the above order; 5344the offsets of the id, descriptors and data blocks are found in the 5345header. However, all four blocks are aligned to 64 bit offsets in the 5346file and they do not overlap. 5347 5348All blocks except the data block are immutable. Userspace can read them 5349only one time after retrieving the file descriptor, and then use ``pread`` or 5350``lseek`` to read the statistics repeatedly. 5351 5352All data is in system endianness. 5353 5354The format of the header is as follows:: 5355 5356 struct kvm_stats_header { 5357 __u32 flags; 5358 __u32 name_size; 5359 __u32 num_desc; 5360 __u32 id_offset; 5361 __u32 desc_offset; 5362 __u32 data_offset; 5363 }; 5364 5365The ``flags`` field is not used at the moment. It is always read as 0. 5366 5367The ``name_size`` field is the size (in byte) of the statistics name string 5368(including trailing '\0') which is contained in the "id string" block and 5369appended at the end of every descriptor. 5370 5371The ``num_desc`` field is the number of descriptors that are included in the 5372descriptor block. (The actual number of values in the data block may be 5373larger, since each descriptor may comprise more than one value). 5374 5375The ``id_offset`` field is the offset of the id string from the start of the 5376file indicated by the file descriptor. It is a multiple of 8. 5377 5378The ``desc_offset`` field is the offset of the Descriptors block from the start 5379of the file indicated by the file descriptor. It is a multiple of 8. 5380 5381The ``data_offset`` field is the offset of the Stats Data block from the start 5382of the file indicated by the file descriptor. It is a multiple of 8. 5383 5384The id string block contains a string which identifies the file descriptor on 5385which KVM_GET_STATS_FD was invoked. The size of the block, including the 5386trailing ``'\0'``, is indicated by the ``name_size`` field in the header. 5387 5388The descriptors block is only needed to be read once for the lifetime of the 5389file descriptor contains a sequence of ``struct kvm_stats_desc``, each followed 5390by a string of size ``name_size``. 5391:: 5392 5393 #define KVM_STATS_TYPE_SHIFT 0 5394 #define KVM_STATS_TYPE_MASK (0xF << KVM_STATS_TYPE_SHIFT) 5395 #define KVM_STATS_TYPE_CUMULATIVE (0x0 << KVM_STATS_TYPE_SHIFT) 5396 #define KVM_STATS_TYPE_INSTANT (0x1 << KVM_STATS_TYPE_SHIFT) 5397 #define KVM_STATS_TYPE_PEAK (0x2 << KVM_STATS_TYPE_SHIFT) 5398 #define KVM_STATS_TYPE_LINEAR_HIST (0x3 << KVM_STATS_TYPE_SHIFT) 5399 #define KVM_STATS_TYPE_LOG_HIST (0x4 << KVM_STATS_TYPE_SHIFT) 5400 #define KVM_STATS_TYPE_MAX KVM_STATS_TYPE_LOG_HIST 5401 5402 #define KVM_STATS_UNIT_SHIFT 4 5403 #define KVM_STATS_UNIT_MASK (0xF << KVM_STATS_UNIT_SHIFT) 5404 #define KVM_STATS_UNIT_NONE (0x0 << KVM_STATS_UNIT_SHIFT) 5405 #define KVM_STATS_UNIT_BYTES (0x1 << KVM_STATS_UNIT_SHIFT) 5406 #define KVM_STATS_UNIT_SECONDS (0x2 << KVM_STATS_UNIT_SHIFT) 5407 #define KVM_STATS_UNIT_CYCLES (0x3 << KVM_STATS_UNIT_SHIFT) 5408 #define KVM_STATS_UNIT_MAX KVM_STATS_UNIT_CYCLES 5409 5410 #define KVM_STATS_BASE_SHIFT 8 5411 #define KVM_STATS_BASE_MASK (0xF << KVM_STATS_BASE_SHIFT) 5412 #define KVM_STATS_BASE_POW10 (0x0 << KVM_STATS_BASE_SHIFT) 5413 #define KVM_STATS_BASE_POW2 (0x1 << KVM_STATS_BASE_SHIFT) 5414 #define KVM_STATS_BASE_MAX KVM_STATS_BASE_POW2 5415 5416 struct kvm_stats_desc { 5417 __u32 flags; 5418 __s16 exponent; 5419 __u16 size; 5420 __u32 offset; 5421 __u32 bucket_size; 5422 char name[]; 5423 }; 5424 5425The ``flags`` field contains the type and unit of the statistics data described 5426by this descriptor. Its endianness is CPU native. 5427The following flags are supported: 5428 5429Bits 0-3 of ``flags`` encode the type: 5430 5431 * ``KVM_STATS_TYPE_CUMULATIVE`` 5432 The statistics reports a cumulative count. The value of data can only be increased. 5433 Most of the counters used in KVM are of this type. 5434 The corresponding ``size`` field for this type is always 1. 5435 All cumulative statistics data are read/write. 5436 * ``KVM_STATS_TYPE_INSTANT`` 5437 The statistics reports an instantaneous value. Its value can be increased or 5438 decreased. This type is usually used as a measurement of some resources, 5439 like the number of dirty pages, the number of large pages, etc. 5440 All instant statistics are read only. 5441 The corresponding ``size`` field for this type is always 1. 5442 * ``KVM_STATS_TYPE_PEAK`` 5443 The statistics data reports a peak value, for example the maximum number 5444 of items in a hash table bucket, the longest time waited and so on. 5445 The value of data can only be increased. 5446 The corresponding ``size`` field for this type is always 1. 5447 * ``KVM_STATS_TYPE_LINEAR_HIST`` 5448 The statistic is reported as a linear histogram. The number of 5449 buckets is specified by the ``size`` field. The size of buckets is specified 5450 by the ``hist_param`` field. The range of the Nth bucket (1 <= N < ``size``) 5451 is [``hist_param``*(N-1), ``hist_param``*N), while the range of the last 5452 bucket is [``hist_param``*(``size``-1), +INF). (+INF means positive infinity 5453 value.) The bucket value indicates how many samples fell in the bucket's range. 5454 * ``KVM_STATS_TYPE_LOG_HIST`` 5455 The statistic is reported as a logarithmic histogram. The number of 5456 buckets is specified by the ``size`` field. The range of the first bucket is 5457 [0, 1), while the range of the last bucket is [pow(2, ``size``-2), +INF). 5458 Otherwise, The Nth bucket (1 < N < ``size``) covers 5459 [pow(2, N-2), pow(2, N-1)). The bucket value indicates how many samples fell 5460 in the bucket's range. 5461 5462Bits 4-7 of ``flags`` encode the unit: 5463 5464 * ``KVM_STATS_UNIT_NONE`` 5465 There is no unit for the value of statistics data. This usually means that 5466 the value is a simple counter of an event. 5467 * ``KVM_STATS_UNIT_BYTES`` 5468 It indicates that the statistics data is used to measure memory size, in the 5469 unit of Byte, KiByte, MiByte, GiByte, etc. The unit of the data is 5470 determined by the ``exponent`` field in the descriptor. 5471 * ``KVM_STATS_UNIT_SECONDS`` 5472 It indicates that the statistics data is used to measure time or latency. 5473 * ``KVM_STATS_UNIT_CYCLES`` 5474 It indicates that the statistics data is used to measure CPU clock cycles. 5475 5476Bits 8-11 of ``flags``, together with ``exponent``, encode the scale of the 5477unit: 5478 5479 * ``KVM_STATS_BASE_POW10`` 5480 The scale is based on power of 10. It is used for measurement of time and 5481 CPU clock cycles. For example, an exponent of -9 can be used with 5482 ``KVM_STATS_UNIT_SECONDS`` to express that the unit is nanoseconds. 5483 * ``KVM_STATS_BASE_POW2`` 5484 The scale is based on power of 2. It is used for measurement of memory size. 5485 For example, an exponent of 20 can be used with ``KVM_STATS_UNIT_BYTES`` to 5486 express that the unit is MiB. 5487 5488The ``size`` field is the number of values of this statistics data. Its 5489value is usually 1 for most of simple statistics. 1 means it contains an 5490unsigned 64bit data. 5491 5492The ``offset`` field is the offset from the start of Data Block to the start of 5493the corresponding statistics data. 5494 5495The ``bucket_size`` field is used as a parameter for histogram statistics data. 5496It is only used by linear histogram statistics data, specifying the size of a 5497bucket. 5498 5499The ``name`` field is the name string of the statistics data. The name string 5500starts at the end of ``struct kvm_stats_desc``. The maximum length including 5501the trailing ``'\0'``, is indicated by ``name_size`` in the header. 5502 5503The Stats Data block contains an array of 64-bit values in the same order 5504as the descriptors in Descriptors block. 5505 55065. The kvm_run structure 5507======================== 5508 5509Application code obtains a pointer to the kvm_run structure by 5510mmap()ing a vcpu fd. From that point, application code can control 5511execution by changing fields in kvm_run prior to calling the KVM_RUN 5512ioctl, and obtain information about the reason KVM_RUN returned by 5513looking up structure members. 5514 5515:: 5516 5517 struct kvm_run { 5518 /* in */ 5519 __u8 request_interrupt_window; 5520 5521Request that KVM_RUN return when it becomes possible to inject external 5522interrupts into the guest. Useful in conjunction with KVM_INTERRUPT. 5523 5524:: 5525 5526 __u8 immediate_exit; 5527 5528This field is polled once when KVM_RUN starts; if non-zero, KVM_RUN 5529exits immediately, returning -EINTR. In the common scenario where a 5530signal is used to "kick" a VCPU out of KVM_RUN, this field can be used 5531to avoid usage of KVM_SET_SIGNAL_MASK, which has worse scalability. 5532Rather than blocking the signal outside KVM_RUN, userspace can set up 5533a signal handler that sets run->immediate_exit to a non-zero value. 5534 5535This field is ignored if KVM_CAP_IMMEDIATE_EXIT is not available. 5536 5537:: 5538 5539 __u8 padding1[6]; 5540 5541 /* out */ 5542 __u32 exit_reason; 5543 5544When KVM_RUN has returned successfully (return value 0), this informs 5545application code why KVM_RUN has returned. Allowable values for this 5546field are detailed below. 5547 5548:: 5549 5550 __u8 ready_for_interrupt_injection; 5551 5552If request_interrupt_window has been specified, this field indicates 5553an interrupt can be injected now with KVM_INTERRUPT. 5554 5555:: 5556 5557 __u8 if_flag; 5558 5559The value of the current interrupt flag. Only valid if in-kernel 5560local APIC is not used. 5561 5562:: 5563 5564 __u16 flags; 5565 5566More architecture-specific flags detailing state of the VCPU that may 5567affect the device's behavior. Current defined flags:: 5568 5569 /* x86, set if the VCPU is in system management mode */ 5570 #define KVM_RUN_X86_SMM (1 << 0) 5571 /* x86, set if bus lock detected in VM */ 5572 #define KVM_RUN_BUS_LOCK (1 << 1) 5573 5574:: 5575 5576 /* in (pre_kvm_run), out (post_kvm_run) */ 5577 __u64 cr8; 5578 5579The value of the cr8 register. Only valid if in-kernel local APIC is 5580not used. Both input and output. 5581 5582:: 5583 5584 __u64 apic_base; 5585 5586The value of the APIC BASE msr. Only valid if in-kernel local 5587APIC is not used. Both input and output. 5588 5589:: 5590 5591 union { 5592 /* KVM_EXIT_UNKNOWN */ 5593 struct { 5594 __u64 hardware_exit_reason; 5595 } hw; 5596 5597If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown 5598reasons. Further architecture-specific information is available in 5599hardware_exit_reason. 5600 5601:: 5602 5603 /* KVM_EXIT_FAIL_ENTRY */ 5604 struct { 5605 __u64 hardware_entry_failure_reason; 5606 __u32 cpu; /* if KVM_LAST_CPU */ 5607 } fail_entry; 5608 5609If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due 5610to unknown reasons. Further architecture-specific information is 5611available in hardware_entry_failure_reason. 5612 5613:: 5614 5615 /* KVM_EXIT_EXCEPTION */ 5616 struct { 5617 __u32 exception; 5618 __u32 error_code; 5619 } ex; 5620 5621Unused. 5622 5623:: 5624 5625 /* KVM_EXIT_IO */ 5626 struct { 5627 #define KVM_EXIT_IO_IN 0 5628 #define KVM_EXIT_IO_OUT 1 5629 __u8 direction; 5630 __u8 size; /* bytes */ 5631 __u16 port; 5632 __u32 count; 5633 __u64 data_offset; /* relative to kvm_run start */ 5634 } io; 5635 5636If exit_reason is KVM_EXIT_IO, then the vcpu has 5637executed a port I/O instruction which could not be satisfied by kvm. 5638data_offset describes where the data is located (KVM_EXIT_IO_OUT) or 5639where kvm expects application code to place the data for the next 5640KVM_RUN invocation (KVM_EXIT_IO_IN). Data format is a packed array. 5641 5642:: 5643 5644 /* KVM_EXIT_DEBUG */ 5645 struct { 5646 struct kvm_debug_exit_arch arch; 5647 } debug; 5648 5649If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event 5650for which architecture specific information is returned. 5651 5652:: 5653 5654 /* KVM_EXIT_MMIO */ 5655 struct { 5656 __u64 phys_addr; 5657 __u8 data[8]; 5658 __u32 len; 5659 __u8 is_write; 5660 } mmio; 5661 5662If exit_reason is KVM_EXIT_MMIO, then the vcpu has 5663executed a memory-mapped I/O instruction which could not be satisfied 5664by kvm. The 'data' member contains the written data if 'is_write' is 5665true, and should be filled by application code otherwise. 5666 5667The 'data' member contains, in its first 'len' bytes, the value as it would 5668appear if the VCPU performed a load or store of the appropriate width directly 5669to the byte array. 5670 5671.. note:: 5672 5673 For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR, KVM_EXIT_XEN, 5674 KVM_EXIT_EPR, KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR the corresponding 5675 operations are complete (and guest state is consistent) only after userspace 5676 has re-entered the kernel with KVM_RUN. The kernel side will first finish 5677 incomplete operations and then check for pending signals. 5678 5679 The pending state of the operation is not preserved in state which is 5680 visible to userspace, thus userspace should ensure that the operation is 5681 completed before performing a live migration. Userspace can re-enter the 5682 guest with an unmasked signal pending or with the immediate_exit field set 5683 to complete pending operations without allowing any further instructions 5684 to be executed. 5685 5686:: 5687 5688 /* KVM_EXIT_HYPERCALL */ 5689 struct { 5690 __u64 nr; 5691 __u64 args[6]; 5692 __u64 ret; 5693 __u32 longmode; 5694 __u32 pad; 5695 } hypercall; 5696 5697Unused. This was once used for 'hypercall to userspace'. To implement 5698such functionality, use KVM_EXIT_IO (x86) or KVM_EXIT_MMIO (all except s390). 5699 5700.. note:: KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO. 5701 5702:: 5703 5704 /* KVM_EXIT_TPR_ACCESS */ 5705 struct { 5706 __u64 rip; 5707 __u32 is_write; 5708 __u32 pad; 5709 } tpr_access; 5710 5711To be documented (KVM_TPR_ACCESS_REPORTING). 5712 5713:: 5714 5715 /* KVM_EXIT_S390_SIEIC */ 5716 struct { 5717 __u8 icptcode; 5718 __u64 mask; /* psw upper half */ 5719 __u64 addr; /* psw lower half */ 5720 __u16 ipa; 5721 __u32 ipb; 5722 } s390_sieic; 5723 5724s390 specific. 5725 5726:: 5727 5728 /* KVM_EXIT_S390_RESET */ 5729 #define KVM_S390_RESET_POR 1 5730 #define KVM_S390_RESET_CLEAR 2 5731 #define KVM_S390_RESET_SUBSYSTEM 4 5732 #define KVM_S390_RESET_CPU_INIT 8 5733 #define KVM_S390_RESET_IPL 16 5734 __u64 s390_reset_flags; 5735 5736s390 specific. 5737 5738:: 5739 5740 /* KVM_EXIT_S390_UCONTROL */ 5741 struct { 5742 __u64 trans_exc_code; 5743 __u32 pgm_code; 5744 } s390_ucontrol; 5745 5746s390 specific. A page fault has occurred for a user controlled virtual 5747machine (KVM_VM_S390_UNCONTROL) on it's host page table that cannot be 5748resolved by the kernel. 5749The program code and the translation exception code that were placed 5750in the cpu's lowcore are presented here as defined by the z Architecture 5751Principles of Operation Book in the Chapter for Dynamic Address Translation 5752(DAT) 5753 5754:: 5755 5756 /* KVM_EXIT_DCR */ 5757 struct { 5758 __u32 dcrn; 5759 __u32 data; 5760 __u8 is_write; 5761 } dcr; 5762 5763Deprecated - was used for 440 KVM. 5764 5765:: 5766 5767 /* KVM_EXIT_OSI */ 5768 struct { 5769 __u64 gprs[32]; 5770 } osi; 5771 5772MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch 5773hypercalls and exit with this exit struct that contains all the guest gprs. 5774 5775If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall. 5776Userspace can now handle the hypercall and when it's done modify the gprs as 5777necessary. Upon guest entry all guest GPRs will then be replaced by the values 5778in this struct. 5779 5780:: 5781 5782 /* KVM_EXIT_PAPR_HCALL */ 5783 struct { 5784 __u64 nr; 5785 __u64 ret; 5786 __u64 args[9]; 5787 } papr_hcall; 5788 5789This is used on 64-bit PowerPC when emulating a pSeries partition, 5790e.g. with the 'pseries' machine type in qemu. It occurs when the 5791guest does a hypercall using the 'sc 1' instruction. The 'nr' field 5792contains the hypercall number (from the guest R3), and 'args' contains 5793the arguments (from the guest R4 - R12). Userspace should put the 5794return code in 'ret' and any extra returned values in args[]. 5795The possible hypercalls are defined in the Power Architecture Platform 5796Requirements (PAPR) document available from www.power.org (free 5797developer registration required to access it). 5798 5799:: 5800 5801 /* KVM_EXIT_S390_TSCH */ 5802 struct { 5803 __u16 subchannel_id; 5804 __u16 subchannel_nr; 5805 __u32 io_int_parm; 5806 __u32 io_int_word; 5807 __u32 ipb; 5808 __u8 dequeued; 5809 } s390_tsch; 5810 5811s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled 5812and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O 5813interrupt for the target subchannel has been dequeued and subchannel_id, 5814subchannel_nr, io_int_parm and io_int_word contain the parameters for that 5815interrupt. ipb is needed for instruction parameter decoding. 5816 5817:: 5818 5819 /* KVM_EXIT_EPR */ 5820 struct { 5821 __u32 epr; 5822 } epr; 5823 5824On FSL BookE PowerPC chips, the interrupt controller has a fast patch 5825interrupt acknowledge path to the core. When the core successfully 5826delivers an interrupt, it automatically populates the EPR register with 5827the interrupt vector number and acknowledges the interrupt inside 5828the interrupt controller. 5829 5830In case the interrupt controller lives in user space, we need to do 5831the interrupt acknowledge cycle through it to fetch the next to be 5832delivered interrupt vector using this exit. 5833 5834It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an 5835external interrupt has just been delivered into the guest. User space 5836should put the acknowledged interrupt vector into the 'epr' field. 5837 5838:: 5839 5840 /* KVM_EXIT_SYSTEM_EVENT */ 5841 struct { 5842 #define KVM_SYSTEM_EVENT_SHUTDOWN 1 5843 #define KVM_SYSTEM_EVENT_RESET 2 5844 #define KVM_SYSTEM_EVENT_CRASH 3 5845 __u32 type; 5846 __u64 flags; 5847 } system_event; 5848 5849If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered 5850a system-level event using some architecture specific mechanism (hypercall 5851or some special instruction). In case of ARM/ARM64, this is triggered using 5852HVC instruction based PSCI call from the vcpu. The 'type' field describes 5853the system-level event type. The 'flags' field describes architecture 5854specific flags for the system-level event. 5855 5856Valid values for 'type' are: 5857 5858 - KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the 5859 VM. Userspace is not obliged to honour this, and if it does honour 5860 this does not need to destroy the VM synchronously (ie it may call 5861 KVM_RUN again before shutdown finally occurs). 5862 - KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM. 5863 As with SHUTDOWN, userspace can choose to ignore the request, or 5864 to schedule the reset to occur in the future and may call KVM_RUN again. 5865 - KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest 5866 has requested a crash condition maintenance. Userspace can choose 5867 to ignore the request, or to gather VM memory core dump and/or 5868 reset/shutdown of the VM. 5869 5870:: 5871 5872 /* KVM_EXIT_IOAPIC_EOI */ 5873 struct { 5874 __u8 vector; 5875 } eoi; 5876 5877Indicates that the VCPU's in-kernel local APIC received an EOI for a 5878level-triggered IOAPIC interrupt. This exit only triggers when the 5879IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled); 5880the userspace IOAPIC should process the EOI and retrigger the interrupt if 5881it is still asserted. Vector is the LAPIC interrupt vector for which the 5882EOI was received. 5883 5884:: 5885 5886 struct kvm_hyperv_exit { 5887 #define KVM_EXIT_HYPERV_SYNIC 1 5888 #define KVM_EXIT_HYPERV_HCALL 2 5889 #define KVM_EXIT_HYPERV_SYNDBG 3 5890 __u32 type; 5891 __u32 pad1; 5892 union { 5893 struct { 5894 __u32 msr; 5895 __u32 pad2; 5896 __u64 control; 5897 __u64 evt_page; 5898 __u64 msg_page; 5899 } synic; 5900 struct { 5901 __u64 input; 5902 __u64 result; 5903 __u64 params[2]; 5904 } hcall; 5905 struct { 5906 __u32 msr; 5907 __u32 pad2; 5908 __u64 control; 5909 __u64 status; 5910 __u64 send_page; 5911 __u64 recv_page; 5912 __u64 pending_page; 5913 } syndbg; 5914 } u; 5915 }; 5916 /* KVM_EXIT_HYPERV */ 5917 struct kvm_hyperv_exit hyperv; 5918 5919Indicates that the VCPU exits into userspace to process some tasks 5920related to Hyper-V emulation. 5921 5922Valid values for 'type' are: 5923 5924 - KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about 5925 5926Hyper-V SynIC state change. Notification is used to remap SynIC 5927event/message pages and to enable/disable SynIC messages/events processing 5928in userspace. 5929 5930 - KVM_EXIT_HYPERV_SYNDBG -- synchronously notify user-space about 5931 5932Hyper-V Synthetic debugger state change. Notification is used to either update 5933the pending_page location or to send a control command (send the buffer located 5934in send_page or recv a buffer to recv_page). 5935 5936:: 5937 5938 /* KVM_EXIT_ARM_NISV */ 5939 struct { 5940 __u64 esr_iss; 5941 __u64 fault_ipa; 5942 } arm_nisv; 5943 5944Used on arm and arm64 systems. If a guest accesses memory not in a memslot, 5945KVM will typically return to userspace and ask it to do MMIO emulation on its 5946behalf. However, for certain classes of instructions, no instruction decode 5947(direction, length of memory access) is provided, and fetching and decoding 5948the instruction from the VM is overly complicated to live in the kernel. 5949 5950Historically, when this situation occurred, KVM would print a warning and kill 5951the VM. KVM assumed that if the guest accessed non-memslot memory, it was 5952trying to do I/O, which just couldn't be emulated, and the warning message was 5953phrased accordingly. However, what happened more often was that a guest bug 5954caused access outside the guest memory areas which should lead to a more 5955meaningful warning message and an external abort in the guest, if the access 5956did not fall within an I/O window. 5957 5958Userspace implementations can query for KVM_CAP_ARM_NISV_TO_USER, and enable 5959this capability at VM creation. Once this is done, these types of errors will 5960instead return to userspace with KVM_EXIT_ARM_NISV, with the valid bits from 5961the HSR (arm) and ESR_EL2 (arm64) in the esr_iss field, and the faulting IPA 5962in the fault_ipa field. Userspace can either fix up the access if it's 5963actually an I/O access by decoding the instruction from guest memory (if it's 5964very brave) and continue executing the guest, or it can decide to suspend, 5965dump, or restart the guest. 5966 5967Note that KVM does not skip the faulting instruction as it does for 5968KVM_EXIT_MMIO, but userspace has to emulate any change to the processing state 5969if it decides to decode and emulate the instruction. 5970 5971:: 5972 5973 /* KVM_EXIT_X86_RDMSR / KVM_EXIT_X86_WRMSR */ 5974 struct { 5975 __u8 error; /* user -> kernel */ 5976 __u8 pad[7]; 5977 __u32 reason; /* kernel -> user */ 5978 __u32 index; /* kernel -> user */ 5979 __u64 data; /* kernel <-> user */ 5980 } msr; 5981 5982Used on x86 systems. When the VM capability KVM_CAP_X86_USER_SPACE_MSR is 5983enabled, MSR accesses to registers that would invoke a #GP by KVM kernel code 5984will instead trigger a KVM_EXIT_X86_RDMSR exit for reads and KVM_EXIT_X86_WRMSR 5985exit for writes. 5986 5987The "reason" field specifies why the MSR trap occurred. User space will only 5988receive MSR exit traps when a particular reason was requested during through 5989ENABLE_CAP. Currently valid exit reasons are: 5990 5991 KVM_MSR_EXIT_REASON_UNKNOWN - access to MSR that is unknown to KVM 5992 KVM_MSR_EXIT_REASON_INVAL - access to invalid MSRs or reserved bits 5993 KVM_MSR_EXIT_REASON_FILTER - access blocked by KVM_X86_SET_MSR_FILTER 5994 5995For KVM_EXIT_X86_RDMSR, the "index" field tells user space which MSR the guest 5996wants to read. To respond to this request with a successful read, user space 5997writes the respective data into the "data" field and must continue guest 5998execution to ensure the read data is transferred into guest register state. 5999 6000If the RDMSR request was unsuccessful, user space indicates that with a "1" in 6001the "error" field. This will inject a #GP into the guest when the VCPU is 6002executed again. 6003 6004For KVM_EXIT_X86_WRMSR, the "index" field tells user space which MSR the guest 6005wants to write. Once finished processing the event, user space must continue 6006vCPU execution. If the MSR write was unsuccessful, user space also sets the 6007"error" field to "1". 6008 6009:: 6010 6011 6012 struct kvm_xen_exit { 6013 #define KVM_EXIT_XEN_HCALL 1 6014 __u32 type; 6015 union { 6016 struct { 6017 __u32 longmode; 6018 __u32 cpl; 6019 __u64 input; 6020 __u64 result; 6021 __u64 params[6]; 6022 } hcall; 6023 } u; 6024 }; 6025 /* KVM_EXIT_XEN */ 6026 struct kvm_hyperv_exit xen; 6027 6028Indicates that the VCPU exits into userspace to process some tasks 6029related to Xen emulation. 6030 6031Valid values for 'type' are: 6032 6033 - KVM_EXIT_XEN_HCALL -- synchronously notify user-space about Xen hypercall. 6034 Userspace is expected to place the hypercall result into the appropriate 6035 field before invoking KVM_RUN again. 6036 6037:: 6038 6039 /* KVM_EXIT_RISCV_SBI */ 6040 struct { 6041 unsigned long extension_id; 6042 unsigned long function_id; 6043 unsigned long args[6]; 6044 unsigned long ret[2]; 6045 } riscv_sbi; 6046If exit reason is KVM_EXIT_RISCV_SBI then it indicates that the VCPU has 6047done a SBI call which is not handled by KVM RISC-V kernel module. The details 6048of the SBI call are available in 'riscv_sbi' member of kvm_run structure. The 6049'extension_id' field of 'riscv_sbi' represents SBI extension ID whereas the 6050'function_id' field represents function ID of given SBI extension. The 'args' 6051array field of 'riscv_sbi' represents parameters for the SBI call and 'ret' 6052array field represents return values. The userspace should update the return 6053values of SBI call before resuming the VCPU. For more details on RISC-V SBI 6054spec refer, https://github.com/riscv/riscv-sbi-doc. 6055 6056:: 6057 6058 /* Fix the size of the union. */ 6059 char padding[256]; 6060 }; 6061 6062 /* 6063 * shared registers between kvm and userspace. 6064 * kvm_valid_regs specifies the register classes set by the host 6065 * kvm_dirty_regs specified the register classes dirtied by userspace 6066 * struct kvm_sync_regs is architecture specific, as well as the 6067 * bits for kvm_valid_regs and kvm_dirty_regs 6068 */ 6069 __u64 kvm_valid_regs; 6070 __u64 kvm_dirty_regs; 6071 union { 6072 struct kvm_sync_regs regs; 6073 char padding[SYNC_REGS_SIZE_BYTES]; 6074 } s; 6075 6076If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access 6077certain guest registers without having to call SET/GET_*REGS. Thus we can 6078avoid some system call overhead if userspace has to handle the exit. 6079Userspace can query the validity of the structure by checking 6080kvm_valid_regs for specific bits. These bits are architecture specific 6081and usually define the validity of a groups of registers. (e.g. one bit 6082for general purpose registers) 6083 6084Please note that the kernel is allowed to use the kvm_run structure as the 6085primary storage for certain register types. Therefore, the kernel may use the 6086values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set. 6087 6088:: 6089 6090 }; 6091 6092 6093 60946. Capabilities that can be enabled on vCPUs 6095============================================ 6096 6097There are certain capabilities that change the behavior of the virtual CPU or 6098the virtual machine when enabled. To enable them, please see section 4.37. 6099Below you can find a list of capabilities and what their effect on the vCPU or 6100the virtual machine is when enabling them. 6101 6102The following information is provided along with the description: 6103 6104 Architectures: 6105 which instruction set architectures provide this ioctl. 6106 x86 includes both i386 and x86_64. 6107 6108 Target: 6109 whether this is a per-vcpu or per-vm capability. 6110 6111 Parameters: 6112 what parameters are accepted by the capability. 6113 6114 Returns: 6115 the return value. General error numbers (EBADF, ENOMEM, EINVAL) 6116 are not detailed, but errors with specific meanings are. 6117 6118 61196.1 KVM_CAP_PPC_OSI 6120------------------- 6121 6122:Architectures: ppc 6123:Target: vcpu 6124:Parameters: none 6125:Returns: 0 on success; -1 on error 6126 6127This capability enables interception of OSI hypercalls that otherwise would 6128be treated as normal system calls to be injected into the guest. OSI hypercalls 6129were invented by Mac-on-Linux to have a standardized communication mechanism 6130between the guest and the host. 6131 6132When this capability is enabled, KVM_EXIT_OSI can occur. 6133 6134 61356.2 KVM_CAP_PPC_PAPR 6136-------------------- 6137 6138:Architectures: ppc 6139:Target: vcpu 6140:Parameters: none 6141:Returns: 0 on success; -1 on error 6142 6143This capability enables interception of PAPR hypercalls. PAPR hypercalls are 6144done using the hypercall instruction "sc 1". 6145 6146It also sets the guest privilege level to "supervisor" mode. Usually the guest 6147runs in "hypervisor" privilege mode with a few missing features. 6148 6149In addition to the above, it changes the semantics of SDR1. In this mode, the 6150HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the 6151HTAB invisible to the guest. 6152 6153When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur. 6154 6155 61566.3 KVM_CAP_SW_TLB 6157------------------ 6158 6159:Architectures: ppc 6160:Target: vcpu 6161:Parameters: args[0] is the address of a struct kvm_config_tlb 6162:Returns: 0 on success; -1 on error 6163 6164:: 6165 6166 struct kvm_config_tlb { 6167 __u64 params; 6168 __u64 array; 6169 __u32 mmu_type; 6170 __u32 array_len; 6171 }; 6172 6173Configures the virtual CPU's TLB array, establishing a shared memory area 6174between userspace and KVM. The "params" and "array" fields are userspace 6175addresses of mmu-type-specific data structures. The "array_len" field is an 6176safety mechanism, and should be set to the size in bytes of the memory that 6177userspace has reserved for the array. It must be at least the size dictated 6178by "mmu_type" and "params". 6179 6180While KVM_RUN is active, the shared region is under control of KVM. Its 6181contents are undefined, and any modification by userspace results in 6182boundedly undefined behavior. 6183 6184On return from KVM_RUN, the shared region will reflect the current state of 6185the guest's TLB. If userspace makes any changes, it must call KVM_DIRTY_TLB 6186to tell KVM which entries have been changed, prior to calling KVM_RUN again 6187on this vcpu. 6188 6189For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV: 6190 6191 - The "params" field is of type "struct kvm_book3e_206_tlb_params". 6192 - The "array" field points to an array of type "struct 6193 kvm_book3e_206_tlb_entry". 6194 - The array consists of all entries in the first TLB, followed by all 6195 entries in the second TLB. 6196 - Within a TLB, entries are ordered first by increasing set number. Within a 6197 set, entries are ordered by way (increasing ESEL). 6198 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1) 6199 where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value. 6200 - The tsize field of mas1 shall be set to 4K on TLB0, even though the 6201 hardware ignores this value for TLB0. 6202 62036.4 KVM_CAP_S390_CSS_SUPPORT 6204---------------------------- 6205 6206:Architectures: s390 6207:Target: vcpu 6208:Parameters: none 6209:Returns: 0 on success; -1 on error 6210 6211This capability enables support for handling of channel I/O instructions. 6212 6213TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are 6214handled in-kernel, while the other I/O instructions are passed to userspace. 6215 6216When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST 6217SUBCHANNEL intercepts. 6218 6219Note that even though this capability is enabled per-vcpu, the complete 6220virtual machine is affected. 6221 62226.5 KVM_CAP_PPC_EPR 6223------------------- 6224 6225:Architectures: ppc 6226:Target: vcpu 6227:Parameters: args[0] defines whether the proxy facility is active 6228:Returns: 0 on success; -1 on error 6229 6230This capability enables or disables the delivery of interrupts through the 6231external proxy facility. 6232 6233When enabled (args[0] != 0), every time the guest gets an external interrupt 6234delivered, it automatically exits into user space with a KVM_EXIT_EPR exit 6235to receive the topmost interrupt vector. 6236 6237When disabled (args[0] == 0), behavior is as if this facility is unsupported. 6238 6239When this capability is enabled, KVM_EXIT_EPR can occur. 6240 62416.6 KVM_CAP_IRQ_MPIC 6242-------------------- 6243 6244:Architectures: ppc 6245:Parameters: args[0] is the MPIC device fd; 6246 args[1] is the MPIC CPU number for this vcpu 6247 6248This capability connects the vcpu to an in-kernel MPIC device. 6249 62506.7 KVM_CAP_IRQ_XICS 6251-------------------- 6252 6253:Architectures: ppc 6254:Target: vcpu 6255:Parameters: args[0] is the XICS device fd; 6256 args[1] is the XICS CPU number (server ID) for this vcpu 6257 6258This capability connects the vcpu to an in-kernel XICS device. 6259 62606.8 KVM_CAP_S390_IRQCHIP 6261------------------------ 6262 6263:Architectures: s390 6264:Target: vm 6265:Parameters: none 6266 6267This capability enables the in-kernel irqchip for s390. Please refer to 6268"4.24 KVM_CREATE_IRQCHIP" for details. 6269 62706.9 KVM_CAP_MIPS_FPU 6271-------------------- 6272 6273:Architectures: mips 6274:Target: vcpu 6275:Parameters: args[0] is reserved for future use (should be 0). 6276 6277This capability allows the use of the host Floating Point Unit by the guest. It 6278allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is 6279done the ``KVM_REG_MIPS_FPR_*`` and ``KVM_REG_MIPS_FCR_*`` registers can be 6280accessed (depending on the current guest FPU register mode), and the Status.FR, 6281Config5.FRE bits are accessible via the KVM API and also from the guest, 6282depending on them being supported by the FPU. 6283 62846.10 KVM_CAP_MIPS_MSA 6285--------------------- 6286 6287:Architectures: mips 6288:Target: vcpu 6289:Parameters: args[0] is reserved for future use (should be 0). 6290 6291This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest. 6292It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest. 6293Once this is done the ``KVM_REG_MIPS_VEC_*`` and ``KVM_REG_MIPS_MSA_*`` 6294registers can be accessed, and the Config5.MSAEn bit is accessible via the 6295KVM API and also from the guest. 6296 62976.74 KVM_CAP_SYNC_REGS 6298---------------------- 6299 6300:Architectures: s390, x86 6301:Target: s390: always enabled, x86: vcpu 6302:Parameters: none 6303:Returns: x86: KVM_CHECK_EXTENSION returns a bit-array indicating which register 6304 sets are supported 6305 (bitfields defined in arch/x86/include/uapi/asm/kvm.h). 6306 6307As described above in the kvm_sync_regs struct info in section 5 (kvm_run): 6308KVM_CAP_SYNC_REGS "allow[s] userspace to access certain guest registers 6309without having to call SET/GET_*REGS". This reduces overhead by eliminating 6310repeated ioctl calls for setting and/or getting register values. This is 6311particularly important when userspace is making synchronous guest state 6312modifications, e.g. when emulating and/or intercepting instructions in 6313userspace. 6314 6315For s390 specifics, please refer to the source code. 6316 6317For x86: 6318 6319- the register sets to be copied out to kvm_run are selectable 6320 by userspace (rather that all sets being copied out for every exit). 6321- vcpu_events are available in addition to regs and sregs. 6322 6323For x86, the 'kvm_valid_regs' field of struct kvm_run is overloaded to 6324function as an input bit-array field set by userspace to indicate the 6325specific register sets to be copied out on the next exit. 6326 6327To indicate when userspace has modified values that should be copied into 6328the vCPU, the all architecture bitarray field, 'kvm_dirty_regs' must be set. 6329This is done using the same bitflags as for the 'kvm_valid_regs' field. 6330If the dirty bit is not set, then the register set values will not be copied 6331into the vCPU even if they've been modified. 6332 6333Unused bitfields in the bitarrays must be set to zero. 6334 6335:: 6336 6337 struct kvm_sync_regs { 6338 struct kvm_regs regs; 6339 struct kvm_sregs sregs; 6340 struct kvm_vcpu_events events; 6341 }; 6342 63436.75 KVM_CAP_PPC_IRQ_XIVE 6344------------------------- 6345 6346:Architectures: ppc 6347:Target: vcpu 6348:Parameters: args[0] is the XIVE device fd; 6349 args[1] is the XIVE CPU number (server ID) for this vcpu 6350 6351This capability connects the vcpu to an in-kernel XIVE device. 6352 63537. Capabilities that can be enabled on VMs 6354========================================== 6355 6356There are certain capabilities that change the behavior of the virtual 6357machine when enabled. To enable them, please see section 4.37. Below 6358you can find a list of capabilities and what their effect on the VM 6359is when enabling them. 6360 6361The following information is provided along with the description: 6362 6363 Architectures: 6364 which instruction set architectures provide this ioctl. 6365 x86 includes both i386 and x86_64. 6366 6367 Parameters: 6368 what parameters are accepted by the capability. 6369 6370 Returns: 6371 the return value. General error numbers (EBADF, ENOMEM, EINVAL) 6372 are not detailed, but errors with specific meanings are. 6373 6374 63757.1 KVM_CAP_PPC_ENABLE_HCALL 6376---------------------------- 6377 6378:Architectures: ppc 6379:Parameters: args[0] is the sPAPR hcall number; 6380 args[1] is 0 to disable, 1 to enable in-kernel handling 6381 6382This capability controls whether individual sPAPR hypercalls (hcalls) 6383get handled by the kernel or not. Enabling or disabling in-kernel 6384handling of an hcall is effective across the VM. On creation, an 6385initial set of hcalls are enabled for in-kernel handling, which 6386consists of those hcalls for which in-kernel handlers were implemented 6387before this capability was implemented. If disabled, the kernel will 6388not to attempt to handle the hcall, but will always exit to userspace 6389to handle it. Note that it may not make sense to enable some and 6390disable others of a group of related hcalls, but KVM does not prevent 6391userspace from doing that. 6392 6393If the hcall number specified is not one that has an in-kernel 6394implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL 6395error. 6396 63977.2 KVM_CAP_S390_USER_SIGP 6398-------------------------- 6399 6400:Architectures: s390 6401:Parameters: none 6402 6403This capability controls which SIGP orders will be handled completely in user 6404space. With this capability enabled, all fast orders will be handled completely 6405in the kernel: 6406 6407- SENSE 6408- SENSE RUNNING 6409- EXTERNAL CALL 6410- EMERGENCY SIGNAL 6411- CONDITIONAL EMERGENCY SIGNAL 6412 6413All other orders will be handled completely in user space. 6414 6415Only privileged operation exceptions will be checked for in the kernel (or even 6416in the hardware prior to interception). If this capability is not enabled, the 6417old way of handling SIGP orders is used (partially in kernel and user space). 6418 64197.3 KVM_CAP_S390_VECTOR_REGISTERS 6420--------------------------------- 6421 6422:Architectures: s390 6423:Parameters: none 6424:Returns: 0 on success, negative value on error 6425 6426Allows use of the vector registers introduced with z13 processor, and 6427provides for the synchronization between host and user space. Will 6428return -EINVAL if the machine does not support vectors. 6429 64307.4 KVM_CAP_S390_USER_STSI 6431-------------------------- 6432 6433:Architectures: s390 6434:Parameters: none 6435 6436This capability allows post-handlers for the STSI instruction. After 6437initial handling in the kernel, KVM exits to user space with 6438KVM_EXIT_S390_STSI to allow user space to insert further data. 6439 6440Before exiting to userspace, kvm handlers should fill in s390_stsi field of 6441vcpu->run:: 6442 6443 struct { 6444 __u64 addr; 6445 __u8 ar; 6446 __u8 reserved; 6447 __u8 fc; 6448 __u8 sel1; 6449 __u16 sel2; 6450 } s390_stsi; 6451 6452 @addr - guest address of STSI SYSIB 6453 @fc - function code 6454 @sel1 - selector 1 6455 @sel2 - selector 2 6456 @ar - access register number 6457 6458KVM handlers should exit to userspace with rc = -EREMOTE. 6459 64607.5 KVM_CAP_SPLIT_IRQCHIP 6461------------------------- 6462 6463:Architectures: x86 6464:Parameters: args[0] - number of routes reserved for userspace IOAPICs 6465:Returns: 0 on success, -1 on error 6466 6467Create a local apic for each processor in the kernel. This can be used 6468instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the 6469IOAPIC and PIC (and also the PIT, even though this has to be enabled 6470separately). 6471 6472This capability also enables in kernel routing of interrupt requests; 6473when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are 6474used in the IRQ routing table. The first args[0] MSI routes are reserved 6475for the IOAPIC pins. Whenever the LAPIC receives an EOI for these routes, 6476a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace. 6477 6478Fails if VCPU has already been created, or if the irqchip is already in the 6479kernel (i.e. KVM_CREATE_IRQCHIP has already been called). 6480 64817.6 KVM_CAP_S390_RI 6482------------------- 6483 6484:Architectures: s390 6485:Parameters: none 6486 6487Allows use of runtime-instrumentation introduced with zEC12 processor. 6488Will return -EINVAL if the machine does not support runtime-instrumentation. 6489Will return -EBUSY if a VCPU has already been created. 6490 64917.7 KVM_CAP_X2APIC_API 6492---------------------- 6493 6494:Architectures: x86 6495:Parameters: args[0] - features that should be enabled 6496:Returns: 0 on success, -EINVAL when args[0] contains invalid features 6497 6498Valid feature flags in args[0] are:: 6499 6500 #define KVM_X2APIC_API_USE_32BIT_IDS (1ULL << 0) 6501 #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK (1ULL << 1) 6502 6503Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of 6504KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC, 6505allowing the use of 32-bit APIC IDs. See KVM_CAP_X2APIC_API in their 6506respective sections. 6507 6508KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work 6509in logical mode or with more than 255 VCPUs. Otherwise, KVM treats 0xff 6510as a broadcast even in x2APIC mode in order to support physical x2APIC 6511without interrupt remapping. This is undesirable in logical mode, 6512where 0xff represents CPUs 0-7 in cluster 0. 6513 65147.8 KVM_CAP_S390_USER_INSTR0 6515---------------------------- 6516 6517:Architectures: s390 6518:Parameters: none 6519 6520With this capability enabled, all illegal instructions 0x0000 (2 bytes) will 6521be intercepted and forwarded to user space. User space can use this 6522mechanism e.g. to realize 2-byte software breakpoints. The kernel will 6523not inject an operating exception for these instructions, user space has 6524to take care of that. 6525 6526This capability can be enabled dynamically even if VCPUs were already 6527created and are running. 6528 65297.9 KVM_CAP_S390_GS 6530------------------- 6531 6532:Architectures: s390 6533:Parameters: none 6534:Returns: 0 on success; -EINVAL if the machine does not support 6535 guarded storage; -EBUSY if a VCPU has already been created. 6536 6537Allows use of guarded storage for the KVM guest. 6538 65397.10 KVM_CAP_S390_AIS 6540--------------------- 6541 6542:Architectures: s390 6543:Parameters: none 6544 6545Allow use of adapter-interruption suppression. 6546:Returns: 0 on success; -EBUSY if a VCPU has already been created. 6547 65487.11 KVM_CAP_PPC_SMT 6549-------------------- 6550 6551:Architectures: ppc 6552:Parameters: vsmt_mode, flags 6553 6554Enabling this capability on a VM provides userspace with a way to set 6555the desired virtual SMT mode (i.e. the number of virtual CPUs per 6556virtual core). The virtual SMT mode, vsmt_mode, must be a power of 2 6557between 1 and 8. On POWER8, vsmt_mode must also be no greater than 6558the number of threads per subcore for the host. Currently flags must 6559be 0. A successful call to enable this capability will result in 6560vsmt_mode being returned when the KVM_CAP_PPC_SMT capability is 6561subsequently queried for the VM. This capability is only supported by 6562HV KVM, and can only be set before any VCPUs have been created. 6563The KVM_CAP_PPC_SMT_POSSIBLE capability indicates which virtual SMT 6564modes are available. 6565 65667.12 KVM_CAP_PPC_FWNMI 6567---------------------- 6568 6569:Architectures: ppc 6570:Parameters: none 6571 6572With this capability a machine check exception in the guest address 6573space will cause KVM to exit the guest with NMI exit reason. This 6574enables QEMU to build error log and branch to guest kernel registered 6575machine check handling routine. Without this capability KVM will 6576branch to guests' 0x200 interrupt vector. 6577 65787.13 KVM_CAP_X86_DISABLE_EXITS 6579------------------------------ 6580 6581:Architectures: x86 6582:Parameters: args[0] defines which exits are disabled 6583:Returns: 0 on success, -EINVAL when args[0] contains invalid exits 6584 6585Valid bits in args[0] are:: 6586 6587 #define KVM_X86_DISABLE_EXITS_MWAIT (1 << 0) 6588 #define KVM_X86_DISABLE_EXITS_HLT (1 << 1) 6589 #define KVM_X86_DISABLE_EXITS_PAUSE (1 << 2) 6590 #define KVM_X86_DISABLE_EXITS_CSTATE (1 << 3) 6591 6592Enabling this capability on a VM provides userspace with a way to no 6593longer intercept some instructions for improved latency in some 6594workloads, and is suggested when vCPUs are associated to dedicated 6595physical CPUs. More bits can be added in the future; userspace can 6596just pass the KVM_CHECK_EXTENSION result to KVM_ENABLE_CAP to disable 6597all such vmexits. 6598 6599Do not enable KVM_FEATURE_PV_UNHALT if you disable HLT exits. 6600 66017.14 KVM_CAP_S390_HPAGE_1M 6602-------------------------- 6603 6604:Architectures: s390 6605:Parameters: none 6606:Returns: 0 on success, -EINVAL if hpage module parameter was not set 6607 or cmma is enabled, or the VM has the KVM_VM_S390_UCONTROL 6608 flag set 6609 6610With this capability the KVM support for memory backing with 1m pages 6611through hugetlbfs can be enabled for a VM. After the capability is 6612enabled, cmma can't be enabled anymore and pfmfi and the storage key 6613interpretation are disabled. If cmma has already been enabled or the 6614hpage module parameter is not set to 1, -EINVAL is returned. 6615 6616While it is generally possible to create a huge page backed VM without 6617this capability, the VM will not be able to run. 6618 66197.15 KVM_CAP_MSR_PLATFORM_INFO 6620------------------------------ 6621 6622:Architectures: x86 6623:Parameters: args[0] whether feature should be enabled or not 6624 6625With this capability, a guest may read the MSR_PLATFORM_INFO MSR. Otherwise, 6626a #GP would be raised when the guest tries to access. Currently, this 6627capability does not enable write permissions of this MSR for the guest. 6628 66297.16 KVM_CAP_PPC_NESTED_HV 6630-------------------------- 6631 6632:Architectures: ppc 6633:Parameters: none 6634:Returns: 0 on success, -EINVAL when the implementation doesn't support 6635 nested-HV virtualization. 6636 6637HV-KVM on POWER9 and later systems allows for "nested-HV" 6638virtualization, which provides a way for a guest VM to run guests that 6639can run using the CPU's supervisor mode (privileged non-hypervisor 6640state). Enabling this capability on a VM depends on the CPU having 6641the necessary functionality and on the facility being enabled with a 6642kvm-hv module parameter. 6643 66447.17 KVM_CAP_EXCEPTION_PAYLOAD 6645------------------------------ 6646 6647:Architectures: x86 6648:Parameters: args[0] whether feature should be enabled or not 6649 6650With this capability enabled, CR2 will not be modified prior to the 6651emulated VM-exit when L1 intercepts a #PF exception that occurs in 6652L2. Similarly, for kvm-intel only, DR6 will not be modified prior to 6653the emulated VM-exit when L1 intercepts a #DB exception that occurs in 6654L2. As a result, when KVM_GET_VCPU_EVENTS reports a pending #PF (or 6655#DB) exception for L2, exception.has_payload will be set and the 6656faulting address (or the new DR6 bits*) will be reported in the 6657exception_payload field. Similarly, when userspace injects a #PF (or 6658#DB) into L2 using KVM_SET_VCPU_EVENTS, it is expected to set 6659exception.has_payload and to put the faulting address - or the new DR6 6660bits\ [#]_ - in the exception_payload field. 6661 6662This capability also enables exception.pending in struct 6663kvm_vcpu_events, which allows userspace to distinguish between pending 6664and injected exceptions. 6665 6666 6667.. [#] For the new DR6 bits, note that bit 16 is set iff the #DB exception 6668 will clear DR6.RTM. 6669 66707.18 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 6671 6672:Architectures: x86, arm, arm64, mips 6673:Parameters: args[0] whether feature should be enabled or not 6674 6675Valid flags are:: 6676 6677 #define KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (1 << 0) 6678 #define KVM_DIRTY_LOG_INITIALLY_SET (1 << 1) 6679 6680With KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE is set, KVM_GET_DIRTY_LOG will not 6681automatically clear and write-protect all pages that are returned as dirty. 6682Rather, userspace will have to do this operation separately using 6683KVM_CLEAR_DIRTY_LOG. 6684 6685At the cost of a slightly more complicated operation, this provides better 6686scalability and responsiveness for two reasons. First, 6687KVM_CLEAR_DIRTY_LOG ioctl can operate on a 64-page granularity rather 6688than requiring to sync a full memslot; this ensures that KVM does not 6689take spinlocks for an extended period of time. Second, in some cases a 6690large amount of time can pass between a call to KVM_GET_DIRTY_LOG and 6691userspace actually using the data in the page. Pages can be modified 6692during this time, which is inefficient for both the guest and userspace: 6693the guest will incur a higher penalty due to write protection faults, 6694while userspace can see false reports of dirty pages. Manual reprotection 6695helps reducing this time, improving guest performance and reducing the 6696number of dirty log false positives. 6697 6698With KVM_DIRTY_LOG_INITIALLY_SET set, all the bits of the dirty bitmap 6699will be initialized to 1 when created. This also improves performance because 6700dirty logging can be enabled gradually in small chunks on the first call 6701to KVM_CLEAR_DIRTY_LOG. KVM_DIRTY_LOG_INITIALLY_SET depends on 6702KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (it is also only available on 6703x86 and arm64 for now). 6704 6705KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 was previously available under the name 6706KVM_CAP_MANUAL_DIRTY_LOG_PROTECT, but the implementation had bugs that make 6707it hard or impossible to use it correctly. The availability of 6708KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 signals that those bugs are fixed. 6709Userspace should not try to use KVM_CAP_MANUAL_DIRTY_LOG_PROTECT. 6710 67117.19 KVM_CAP_PPC_SECURE_GUEST 6712------------------------------ 6713 6714:Architectures: ppc 6715 6716This capability indicates that KVM is running on a host that has 6717ultravisor firmware and thus can support a secure guest. On such a 6718system, a guest can ask the ultravisor to make it a secure guest, 6719one whose memory is inaccessible to the host except for pages which 6720are explicitly requested to be shared with the host. The ultravisor 6721notifies KVM when a guest requests to become a secure guest, and KVM 6722has the opportunity to veto the transition. 6723 6724If present, this capability can be enabled for a VM, meaning that KVM 6725will allow the transition to secure guest mode. Otherwise KVM will 6726veto the transition. 6727 67287.20 KVM_CAP_HALT_POLL 6729---------------------- 6730 6731:Architectures: all 6732:Target: VM 6733:Parameters: args[0] is the maximum poll time in nanoseconds 6734:Returns: 0 on success; -1 on error 6735 6736This capability overrides the kvm module parameter halt_poll_ns for the 6737target VM. 6738 6739VCPU polling allows a VCPU to poll for wakeup events instead of immediately 6740scheduling during guest halts. The maximum time a VCPU can spend polling is 6741controlled by the kvm module parameter halt_poll_ns. This capability allows 6742the maximum halt time to specified on a per-VM basis, effectively overriding 6743the module parameter for the target VM. 6744 67457.21 KVM_CAP_X86_USER_SPACE_MSR 6746------------------------------- 6747 6748:Architectures: x86 6749:Target: VM 6750:Parameters: args[0] contains the mask of KVM_MSR_EXIT_REASON_* events to report 6751:Returns: 0 on success; -1 on error 6752 6753This capability enables trapping of #GP invoking RDMSR and WRMSR instructions 6754into user space. 6755 6756When a guest requests to read or write an MSR, KVM may not implement all MSRs 6757that are relevant to a respective system. It also does not differentiate by 6758CPU type. 6759 6760To allow more fine grained control over MSR handling, user space may enable 6761this capability. With it enabled, MSR accesses that match the mask specified in 6762args[0] and trigger a #GP event inside the guest by KVM will instead trigger 6763KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR exit notifications which user space 6764can then handle to implement model specific MSR handling and/or user notifications 6765to inform a user that an MSR was not handled. 6766 67677.22 KVM_CAP_X86_BUS_LOCK_EXIT 6768------------------------------- 6769 6770:Architectures: x86 6771:Target: VM 6772:Parameters: args[0] defines the policy used when bus locks detected in guest 6773:Returns: 0 on success, -EINVAL when args[0] contains invalid bits 6774 6775Valid bits in args[0] are:: 6776 6777 #define KVM_BUS_LOCK_DETECTION_OFF (1 << 0) 6778 #define KVM_BUS_LOCK_DETECTION_EXIT (1 << 1) 6779 6780Enabling this capability on a VM provides userspace with a way to select 6781a policy to handle the bus locks detected in guest. Userspace can obtain 6782the supported modes from the result of KVM_CHECK_EXTENSION and define it 6783through the KVM_ENABLE_CAP. 6784 6785KVM_BUS_LOCK_DETECTION_OFF and KVM_BUS_LOCK_DETECTION_EXIT are supported 6786currently and mutually exclusive with each other. More bits can be added in 6787the future. 6788 6789With KVM_BUS_LOCK_DETECTION_OFF set, bus locks in guest will not cause vm exits 6790so that no additional actions are needed. This is the default mode. 6791 6792With KVM_BUS_LOCK_DETECTION_EXIT set, vm exits happen when bus lock detected 6793in VM. KVM just exits to userspace when handling them. Userspace can enforce 6794its own throttling or other policy based mitigations. 6795 6796This capability is aimed to address the thread that VM can exploit bus locks to 6797degree the performance of the whole system. Once the userspace enable this 6798capability and select the KVM_BUS_LOCK_DETECTION_EXIT mode, KVM will set the 6799KVM_RUN_BUS_LOCK flag in vcpu-run->flags field and exit to userspace. Concerning 6800the bus lock vm exit can be preempted by a higher priority VM exit, the exit 6801notifications to userspace can be KVM_EXIT_BUS_LOCK or other reasons. 6802KVM_RUN_BUS_LOCK flag is used to distinguish between them. 6803 68047.23 KVM_CAP_PPC_DAWR1 6805---------------------- 6806 6807:Architectures: ppc 6808:Parameters: none 6809:Returns: 0 on success, -EINVAL when CPU doesn't support 2nd DAWR 6810 6811This capability can be used to check / enable 2nd DAWR feature provided 6812by POWER10 processor. 6813 6814 68157.24 KVM_CAP_VM_COPY_ENC_CONTEXT_FROM 6816------------------------------------- 6817 6818Architectures: x86 SEV enabled 6819Type: vm 6820Parameters: args[0] is the fd of the source vm 6821Returns: 0 on success; ENOTTY on error 6822 6823This capability enables userspace to copy encryption context from the vm 6824indicated by the fd to the vm this is called on. 6825 6826This is intended to support in-guest workloads scheduled by the host. This 6827allows the in-guest workload to maintain its own NPTs and keeps the two vms 6828from accidentally clobbering each other with interrupts and the like (separate 6829APIC/MSRs/etc). 6830 68317.25 KVM_CAP_SGX_ATTRIBUTE 6832-------------------------- 6833 6834:Architectures: x86 6835:Target: VM 6836:Parameters: args[0] is a file handle of a SGX attribute file in securityfs 6837:Returns: 0 on success, -EINVAL if the file handle is invalid or if a requested 6838 attribute is not supported by KVM. 6839 6840KVM_CAP_SGX_ATTRIBUTE enables a userspace VMM to grant a VM access to one or 6841more priveleged enclave attributes. args[0] must hold a file handle to a valid 6842SGX attribute file corresponding to an attribute that is supported/restricted 6843by KVM (currently only PROVISIONKEY). 6844 6845The SGX subsystem restricts access to a subset of enclave attributes to provide 6846additional security for an uncompromised kernel, e.g. use of the PROVISIONKEY 6847is restricted to deter malware from using the PROVISIONKEY to obtain a stable 6848system fingerprint. To prevent userspace from circumventing such restrictions 6849by running an enclave in a VM, KVM prevents access to privileged attributes by 6850default. 6851 6852See Documentation/x86/sgx.rst for more details. 6853 68547.26 KVM_CAP_PPC_RPT_INVALIDATE 6855------------------------------- 6856 6857:Capability: KVM_CAP_PPC_RPT_INVALIDATE 6858:Architectures: ppc 6859:Type: vm 6860 6861This capability indicates that the kernel is capable of handling 6862H_RPT_INVALIDATE hcall. 6863 6864In order to enable the use of H_RPT_INVALIDATE in the guest, 6865user space might have to advertise it for the guest. For example, 6866IBM pSeries (sPAPR) guest starts using it if "hcall-rpt-invalidate" is 6867present in the "ibm,hypertas-functions" device-tree property. 6868 6869This capability is enabled for hypervisors on platforms like POWER9 6870that support radix MMU. 6871 68727.27 KVM_CAP_EXIT_ON_EMULATION_FAILURE 6873-------------------------------------- 6874 6875:Architectures: x86 6876:Parameters: args[0] whether the feature should be enabled or not 6877 6878When this capability is enabled, an emulation failure will result in an exit 6879to userspace with KVM_INTERNAL_ERROR (except when the emulator was invoked 6880to handle a VMware backdoor instruction). Furthermore, KVM will now provide up 6881to 15 instruction bytes for any exit to userspace resulting from an emulation 6882failure. When these exits to userspace occur use the emulation_failure struct 6883instead of the internal struct. They both have the same layout, but the 6884emulation_failure struct matches the content better. It also explicitly 6885defines the 'flags' field which is used to describe the fields in the struct 6886that are valid (ie: if KVM_INTERNAL_ERROR_EMULATION_FLAG_INSTRUCTION_BYTES is 6887set in the 'flags' field then both 'insn_size' and 'insn_bytes' have valid data 6888in them.) 6889 68907.28 KVM_CAP_ARM_MTE 6891-------------------- 6892 6893:Architectures: arm64 6894:Parameters: none 6895 6896This capability indicates that KVM (and the hardware) supports exposing the 6897Memory Tagging Extensions (MTE) to the guest. It must also be enabled by the 6898VMM before creating any VCPUs to allow the guest access. Note that MTE is only 6899available to a guest running in AArch64 mode and enabling this capability will 6900cause attempts to create AArch32 VCPUs to fail. 6901 6902When enabled the guest is able to access tags associated with any memory given 6903to the guest. KVM will ensure that the tags are maintained during swap or 6904hibernation of the host; however the VMM needs to manually save/restore the 6905tags as appropriate if the VM is migrated. 6906 6907When this capability is enabled all memory in memslots must be mapped as 6908not-shareable (no MAP_SHARED), attempts to create a memslot with a 6909MAP_SHARED mmap will result in an -EINVAL return. 6910 6911When enabled the VMM may make use of the ``KVM_ARM_MTE_COPY_TAGS`` ioctl to 6912perform a bulk copy of tags to/from the guest. 6913 69147.29 KVM_CAP_VM_MOVE_ENC_CONTEXT_FROM 6915------------------------------------- 6916 6917Architectures: x86 SEV enabled 6918Type: vm 6919Parameters: args[0] is the fd of the source vm 6920Returns: 0 on success 6921 6922This capability enables userspace to migrate the encryption context from the VM 6923indicated by the fd to the VM this is called on. 6924 6925This is intended to support intra-host migration of VMs between userspace VMMs, 6926upgrading the VMM process without interrupting the guest. 6927 69288. Other capabilities. 6929====================== 6930 6931This section lists capabilities that give information about other 6932features of the KVM implementation. 6933 69348.1 KVM_CAP_PPC_HWRNG 6935--------------------- 6936 6937:Architectures: ppc 6938 6939This capability, if KVM_CHECK_EXTENSION indicates that it is 6940available, means that the kernel has an implementation of the 6941H_RANDOM hypercall backed by a hardware random-number generator. 6942If present, the kernel H_RANDOM handler can be enabled for guest use 6943with the KVM_CAP_PPC_ENABLE_HCALL capability. 6944 69458.2 KVM_CAP_HYPERV_SYNIC 6946------------------------ 6947 6948:Architectures: x86 6949 6950This capability, if KVM_CHECK_EXTENSION indicates that it is 6951available, means that the kernel has an implementation of the 6952Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is 6953used to support Windows Hyper-V based guest paravirt drivers(VMBus). 6954 6955In order to use SynIC, it has to be activated by setting this 6956capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this 6957will disable the use of APIC hardware virtualization even if supported 6958by the CPU, as it's incompatible with SynIC auto-EOI behavior. 6959 69608.3 KVM_CAP_PPC_RADIX_MMU 6961------------------------- 6962 6963:Architectures: ppc 6964 6965This capability, if KVM_CHECK_EXTENSION indicates that it is 6966available, means that the kernel can support guests using the 6967radix MMU defined in Power ISA V3.00 (as implemented in the POWER9 6968processor). 6969 69708.4 KVM_CAP_PPC_HASH_MMU_V3 6971--------------------------- 6972 6973:Architectures: ppc 6974 6975This capability, if KVM_CHECK_EXTENSION indicates that it is 6976available, means that the kernel can support guests using the 6977hashed page table MMU defined in Power ISA V3.00 (as implemented in 6978the POWER9 processor), including in-memory segment tables. 6979 69808.5 KVM_CAP_MIPS_VZ 6981------------------- 6982 6983:Architectures: mips 6984 6985This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that 6986it is available, means that full hardware assisted virtualization capabilities 6987of the hardware are available for use through KVM. An appropriate 6988KVM_VM_MIPS_* type must be passed to KVM_CREATE_VM to create a VM which 6989utilises it. 6990 6991If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is 6992available, it means that the VM is using full hardware assisted virtualization 6993capabilities of the hardware. This is useful to check after creating a VM with 6994KVM_VM_MIPS_DEFAULT. 6995 6996The value returned by KVM_CHECK_EXTENSION should be compared against known 6997values (see below). All other values are reserved. This is to allow for the 6998possibility of other hardware assisted virtualization implementations which 6999may be incompatible with the MIPS VZ ASE. 7000 7001== ========================================================================== 7002 0 The trap & emulate implementation is in use to run guest code in user 7003 mode. Guest virtual memory segments are rearranged to fit the guest in the 7004 user mode address space. 7005 7006 1 The MIPS VZ ASE is in use, providing full hardware assisted 7007 virtualization, including standard guest virtual memory segments. 7008== ========================================================================== 7009 70108.6 KVM_CAP_MIPS_TE 7011------------------- 7012 7013:Architectures: mips 7014 7015This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that 7016it is available, means that the trap & emulate implementation is available to 7017run guest code in user mode, even if KVM_CAP_MIPS_VZ indicates that hardware 7018assisted virtualisation is also available. KVM_VM_MIPS_TE (0) must be passed 7019to KVM_CREATE_VM to create a VM which utilises it. 7020 7021If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is 7022available, it means that the VM is using trap & emulate. 7023 70248.7 KVM_CAP_MIPS_64BIT 7025---------------------- 7026 7027:Architectures: mips 7028 7029This capability indicates the supported architecture type of the guest, i.e. the 7030supported register and address width. 7031 7032The values returned when this capability is checked by KVM_CHECK_EXTENSION on a 7033kvm VM handle correspond roughly to the CP0_Config.AT register field, and should 7034be checked specifically against known values (see below). All other values are 7035reserved. 7036 7037== ======================================================================== 7038 0 MIPS32 or microMIPS32. 7039 Both registers and addresses are 32-bits wide. 7040 It will only be possible to run 32-bit guest code. 7041 7042 1 MIPS64 or microMIPS64 with access only to 32-bit compatibility segments. 7043 Registers are 64-bits wide, but addresses are 32-bits wide. 7044 64-bit guest code may run but cannot access MIPS64 memory segments. 7045 It will also be possible to run 32-bit guest code. 7046 7047 2 MIPS64 or microMIPS64 with access to all address segments. 7048 Both registers and addresses are 64-bits wide. 7049 It will be possible to run 64-bit or 32-bit guest code. 7050== ======================================================================== 7051 70528.9 KVM_CAP_ARM_USER_IRQ 7053------------------------ 7054 7055:Architectures: arm, arm64 7056 7057This capability, if KVM_CHECK_EXTENSION indicates that it is available, means 7058that if userspace creates a VM without an in-kernel interrupt controller, it 7059will be notified of changes to the output level of in-kernel emulated devices, 7060which can generate virtual interrupts, presented to the VM. 7061For such VMs, on every return to userspace, the kernel 7062updates the vcpu's run->s.regs.device_irq_level field to represent the actual 7063output level of the device. 7064 7065Whenever kvm detects a change in the device output level, kvm guarantees at 7066least one return to userspace before running the VM. This exit could either 7067be a KVM_EXIT_INTR or any other exit event, like KVM_EXIT_MMIO. This way, 7068userspace can always sample the device output level and re-compute the state of 7069the userspace interrupt controller. Userspace should always check the state 7070of run->s.regs.device_irq_level on every kvm exit. 7071The value in run->s.regs.device_irq_level can represent both level and edge 7072triggered interrupt signals, depending on the device. Edge triggered interrupt 7073signals will exit to userspace with the bit in run->s.regs.device_irq_level 7074set exactly once per edge signal. 7075 7076The field run->s.regs.device_irq_level is available independent of 7077run->kvm_valid_regs or run->kvm_dirty_regs bits. 7078 7079If KVM_CAP_ARM_USER_IRQ is supported, the KVM_CHECK_EXTENSION ioctl returns a 7080number larger than 0 indicating the version of this capability is implemented 7081and thereby which bits in run->s.regs.device_irq_level can signal values. 7082 7083Currently the following bits are defined for the device_irq_level bitmap:: 7084 7085 KVM_CAP_ARM_USER_IRQ >= 1: 7086 7087 KVM_ARM_DEV_EL1_VTIMER - EL1 virtual timer 7088 KVM_ARM_DEV_EL1_PTIMER - EL1 physical timer 7089 KVM_ARM_DEV_PMU - ARM PMU overflow interrupt signal 7090 7091Future versions of kvm may implement additional events. These will get 7092indicated by returning a higher number from KVM_CHECK_EXTENSION and will be 7093listed above. 7094 70958.10 KVM_CAP_PPC_SMT_POSSIBLE 7096----------------------------- 7097 7098:Architectures: ppc 7099 7100Querying this capability returns a bitmap indicating the possible 7101virtual SMT modes that can be set using KVM_CAP_PPC_SMT. If bit N 7102(counting from the right) is set, then a virtual SMT mode of 2^N is 7103available. 7104 71058.11 KVM_CAP_HYPERV_SYNIC2 7106-------------------------- 7107 7108:Architectures: x86 7109 7110This capability enables a newer version of Hyper-V Synthetic interrupt 7111controller (SynIC). The only difference with KVM_CAP_HYPERV_SYNIC is that KVM 7112doesn't clear SynIC message and event flags pages when they are enabled by 7113writing to the respective MSRs. 7114 71158.12 KVM_CAP_HYPERV_VP_INDEX 7116---------------------------- 7117 7118:Architectures: x86 7119 7120This capability indicates that userspace can load HV_X64_MSR_VP_INDEX msr. Its 7121value is used to denote the target vcpu for a SynIC interrupt. For 7122compatibilty, KVM initializes this msr to KVM's internal vcpu index. When this 7123capability is absent, userspace can still query this msr's value. 7124 71258.13 KVM_CAP_S390_AIS_MIGRATION 7126------------------------------- 7127 7128:Architectures: s390 7129:Parameters: none 7130 7131This capability indicates if the flic device will be able to get/set the 7132AIS states for migration via the KVM_DEV_FLIC_AISM_ALL attribute and allows 7133to discover this without having to create a flic device. 7134 71358.14 KVM_CAP_S390_PSW 7136--------------------- 7137 7138:Architectures: s390 7139 7140This capability indicates that the PSW is exposed via the kvm_run structure. 7141 71428.15 KVM_CAP_S390_GMAP 7143---------------------- 7144 7145:Architectures: s390 7146 7147This capability indicates that the user space memory used as guest mapping can 7148be anywhere in the user memory address space, as long as the memory slots are 7149aligned and sized to a segment (1MB) boundary. 7150 71518.16 KVM_CAP_S390_COW 7152--------------------- 7153 7154:Architectures: s390 7155 7156This capability indicates that the user space memory used as guest mapping can 7157use copy-on-write semantics as well as dirty pages tracking via read-only page 7158tables. 7159 71608.17 KVM_CAP_S390_BPB 7161--------------------- 7162 7163:Architectures: s390 7164 7165This capability indicates that kvm will implement the interfaces to handle 7166reset, migration and nested KVM for branch prediction blocking. The stfle 7167facility 82 should not be provided to the guest without this capability. 7168 71698.18 KVM_CAP_HYPERV_TLBFLUSH 7170---------------------------- 7171 7172:Architectures: x86 7173 7174This capability indicates that KVM supports paravirtualized Hyper-V TLB Flush 7175hypercalls: 7176HvFlushVirtualAddressSpace, HvFlushVirtualAddressSpaceEx, 7177HvFlushVirtualAddressList, HvFlushVirtualAddressListEx. 7178 71798.19 KVM_CAP_ARM_INJECT_SERROR_ESR 7180---------------------------------- 7181 7182:Architectures: arm, arm64 7183 7184This capability indicates that userspace can specify (via the 7185KVM_SET_VCPU_EVENTS ioctl) the syndrome value reported to the guest when it 7186takes a virtual SError interrupt exception. 7187If KVM advertises this capability, userspace can only specify the ISS field for 7188the ESR syndrome. Other parts of the ESR, such as the EC are generated by the 7189CPU when the exception is taken. If this virtual SError is taken to EL1 using 7190AArch64, this value will be reported in the ISS field of ESR_ELx. 7191 7192See KVM_CAP_VCPU_EVENTS for more details. 7193 71948.20 KVM_CAP_HYPERV_SEND_IPI 7195---------------------------- 7196 7197:Architectures: x86 7198 7199This capability indicates that KVM supports paravirtualized Hyper-V IPI send 7200hypercalls: 7201HvCallSendSyntheticClusterIpi, HvCallSendSyntheticClusterIpiEx. 7202 72038.21 KVM_CAP_HYPERV_DIRECT_TLBFLUSH 7204----------------------------------- 7205 7206:Architectures: x86 7207 7208This capability indicates that KVM running on top of Hyper-V hypervisor 7209enables Direct TLB flush for its guests meaning that TLB flush 7210hypercalls are handled by Level 0 hypervisor (Hyper-V) bypassing KVM. 7211Due to the different ABI for hypercall parameters between Hyper-V and 7212KVM, enabling this capability effectively disables all hypercall 7213handling by KVM (as some KVM hypercall may be mistakenly treated as TLB 7214flush hypercalls by Hyper-V) so userspace should disable KVM identification 7215in CPUID and only exposes Hyper-V identification. In this case, guest 7216thinks it's running on Hyper-V and only use Hyper-V hypercalls. 7217 72188.22 KVM_CAP_S390_VCPU_RESETS 7219----------------------------- 7220 7221:Architectures: s390 7222 7223This capability indicates that the KVM_S390_NORMAL_RESET and 7224KVM_S390_CLEAR_RESET ioctls are available. 7225 72268.23 KVM_CAP_S390_PROTECTED 7227--------------------------- 7228 7229:Architectures: s390 7230 7231This capability indicates that the Ultravisor has been initialized and 7232KVM can therefore start protected VMs. 7233This capability governs the KVM_S390_PV_COMMAND ioctl and the 7234KVM_MP_STATE_LOAD MP_STATE. KVM_SET_MP_STATE can fail for protected 7235guests when the state change is invalid. 7236 72378.24 KVM_CAP_STEAL_TIME 7238----------------------- 7239 7240:Architectures: arm64, x86 7241 7242This capability indicates that KVM supports steal time accounting. 7243When steal time accounting is supported it may be enabled with 7244architecture-specific interfaces. This capability and the architecture- 7245specific interfaces must be consistent, i.e. if one says the feature 7246is supported, than the other should as well and vice versa. For arm64 7247see Documentation/virt/kvm/devices/vcpu.rst "KVM_ARM_VCPU_PVTIME_CTRL". 7248For x86 see Documentation/virt/kvm/msr.rst "MSR_KVM_STEAL_TIME". 7249 72508.25 KVM_CAP_S390_DIAG318 7251------------------------- 7252 7253:Architectures: s390 7254 7255This capability enables a guest to set information about its control program 7256(i.e. guest kernel type and version). The information is helpful during 7257system/firmware service events, providing additional data about the guest 7258environments running on the machine. 7259 7260The information is associated with the DIAGNOSE 0x318 instruction, which sets 7261an 8-byte value consisting of a one-byte Control Program Name Code (CPNC) and 7262a 7-byte Control Program Version Code (CPVC). The CPNC determines what 7263environment the control program is running in (e.g. Linux, z/VM...), and the 7264CPVC is used for information specific to OS (e.g. Linux version, Linux 7265distribution...) 7266 7267If this capability is available, then the CPNC and CPVC can be synchronized 7268between KVM and userspace via the sync regs mechanism (KVM_SYNC_DIAG318). 7269 72708.26 KVM_CAP_X86_USER_SPACE_MSR 7271------------------------------- 7272 7273:Architectures: x86 7274 7275This capability indicates that KVM supports deflection of MSR reads and 7276writes to user space. It can be enabled on a VM level. If enabled, MSR 7277accesses that would usually trigger a #GP by KVM into the guest will 7278instead get bounced to user space through the KVM_EXIT_X86_RDMSR and 7279KVM_EXIT_X86_WRMSR exit notifications. 7280 72818.27 KVM_CAP_X86_MSR_FILTER 7282--------------------------- 7283 7284:Architectures: x86 7285 7286This capability indicates that KVM supports that accesses to user defined MSRs 7287may be rejected. With this capability exposed, KVM exports new VM ioctl 7288KVM_X86_SET_MSR_FILTER which user space can call to specify bitmaps of MSR 7289ranges that KVM should reject access to. 7290 7291In combination with KVM_CAP_X86_USER_SPACE_MSR, this allows user space to 7292trap and emulate MSRs that are outside of the scope of KVM as well as 7293limit the attack surface on KVM's MSR emulation code. 7294 72958.28 KVM_CAP_ENFORCE_PV_FEATURE_CPUID 7296----------------------------- 7297 7298Architectures: x86 7299 7300When enabled, KVM will disable paravirtual features provided to the 7301guest according to the bits in the KVM_CPUID_FEATURES CPUID leaf 7302(0x40000001). Otherwise, a guest may use the paravirtual features 7303regardless of what has actually been exposed through the CPUID leaf. 7304 73058.29 KVM_CAP_DIRTY_LOG_RING 7306--------------------------- 7307 7308:Architectures: x86 7309:Parameters: args[0] - size of the dirty log ring 7310 7311KVM is capable of tracking dirty memory using ring buffers that are 7312mmaped into userspace; there is one dirty ring per vcpu. 7313 7314The dirty ring is available to userspace as an array of 7315``struct kvm_dirty_gfn``. Each dirty entry it's defined as:: 7316 7317 struct kvm_dirty_gfn { 7318 __u32 flags; 7319 __u32 slot; /* as_id | slot_id */ 7320 __u64 offset; 7321 }; 7322 7323The following values are defined for the flags field to define the 7324current state of the entry:: 7325 7326 #define KVM_DIRTY_GFN_F_DIRTY BIT(0) 7327 #define KVM_DIRTY_GFN_F_RESET BIT(1) 7328 #define KVM_DIRTY_GFN_F_MASK 0x3 7329 7330Userspace should call KVM_ENABLE_CAP ioctl right after KVM_CREATE_VM 7331ioctl to enable this capability for the new guest and set the size of 7332the rings. Enabling the capability is only allowed before creating any 7333vCPU, and the size of the ring must be a power of two. The larger the 7334ring buffer, the less likely the ring is full and the VM is forced to 7335exit to userspace. The optimal size depends on the workload, but it is 7336recommended that it be at least 64 KiB (4096 entries). 7337 7338Just like for dirty page bitmaps, the buffer tracks writes to 7339all user memory regions for which the KVM_MEM_LOG_DIRTY_PAGES flag was 7340set in KVM_SET_USER_MEMORY_REGION. Once a memory region is registered 7341with the flag set, userspace can start harvesting dirty pages from the 7342ring buffer. 7343 7344An entry in the ring buffer can be unused (flag bits ``00``), 7345dirty (flag bits ``01``) or harvested (flag bits ``1X``). The 7346state machine for the entry is as follows:: 7347 7348 dirtied harvested reset 7349 00 -----------> 01 -------------> 1X -------+ 7350 ^ | 7351 | | 7352 +------------------------------------------+ 7353 7354To harvest the dirty pages, userspace accesses the mmaped ring buffer 7355to read the dirty GFNs. If the flags has the DIRTY bit set (at this stage 7356the RESET bit must be cleared), then it means this GFN is a dirty GFN. 7357The userspace should harvest this GFN and mark the flags from state 7358``01b`` to ``1Xb`` (bit 0 will be ignored by KVM, but bit 1 must be set 7359to show that this GFN is harvested and waiting for a reset), and move 7360on to the next GFN. The userspace should continue to do this until the 7361flags of a GFN have the DIRTY bit cleared, meaning that it has harvested 7362all the dirty GFNs that were available. 7363 7364It's not necessary for userspace to harvest the all dirty GFNs at once. 7365However it must collect the dirty GFNs in sequence, i.e., the userspace 7366program cannot skip one dirty GFN to collect the one next to it. 7367 7368After processing one or more entries in the ring buffer, userspace 7369calls the VM ioctl KVM_RESET_DIRTY_RINGS to notify the kernel about 7370it, so that the kernel will reprotect those collected GFNs. 7371Therefore, the ioctl must be called *before* reading the content of 7372the dirty pages. 7373 7374The dirty ring can get full. When it happens, the KVM_RUN of the 7375vcpu will return with exit reason KVM_EXIT_DIRTY_LOG_FULL. 7376 7377The dirty ring interface has a major difference comparing to the 7378KVM_GET_DIRTY_LOG interface in that, when reading the dirty ring from 7379userspace, it's still possible that the kernel has not yet flushed the 7380processor's dirty page buffers into the kernel buffer (with dirty bitmaps, the 7381flushing is done by the KVM_GET_DIRTY_LOG ioctl). To achieve that, one 7382needs to kick the vcpu out of KVM_RUN using a signal. The resulting 7383vmexit ensures that all dirty GFNs are flushed to the dirty rings. 7384 7385NOTE: the capability KVM_CAP_DIRTY_LOG_RING and the corresponding 7386ioctl KVM_RESET_DIRTY_RINGS are mutual exclusive to the existing ioctls 7387KVM_GET_DIRTY_LOG and KVM_CLEAR_DIRTY_LOG. After enabling 7388KVM_CAP_DIRTY_LOG_RING with an acceptable dirty ring size, the virtual 7389machine will switch to ring-buffer dirty page tracking and further 7390KVM_GET_DIRTY_LOG or KVM_CLEAR_DIRTY_LOG ioctls will fail. 7391 73928.30 KVM_CAP_XEN_HVM 7393-------------------- 7394 7395:Architectures: x86 7396 7397This capability indicates the features that Xen supports for hosting Xen 7398PVHVM guests. Valid flags are:: 7399 7400 #define KVM_XEN_HVM_CONFIG_HYPERCALL_MSR (1 << 0) 7401 #define KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL (1 << 1) 7402 #define KVM_XEN_HVM_CONFIG_SHARED_INFO (1 << 2) 7403 #define KVM_XEN_HVM_CONFIG_RUNSTATE (1 << 2) 7404 7405The KVM_XEN_HVM_CONFIG_HYPERCALL_MSR flag indicates that the KVM_XEN_HVM_CONFIG 7406ioctl is available, for the guest to set its hypercall page. 7407 7408If KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL is also set, the same flag may also be 7409provided in the flags to KVM_XEN_HVM_CONFIG, without providing hypercall page 7410contents, to request that KVM generate hypercall page content automatically 7411and also enable interception of guest hypercalls with KVM_EXIT_XEN. 7412 7413The KVM_XEN_HVM_CONFIG_SHARED_INFO flag indicates the availability of the 7414KVM_XEN_HVM_SET_ATTR, KVM_XEN_HVM_GET_ATTR, KVM_XEN_VCPU_SET_ATTR and 7415KVM_XEN_VCPU_GET_ATTR ioctls, as well as the delivery of exception vectors 7416for event channel upcalls when the evtchn_upcall_pending field of a vcpu's 7417vcpu_info is set. 7418 7419The KVM_XEN_HVM_CONFIG_RUNSTATE flag indicates that the runstate-related 7420features KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR/_CURRENT/_DATA/_ADJUST are 7421supported by the KVM_XEN_VCPU_SET_ATTR/KVM_XEN_VCPU_GET_ATTR ioctls. 7422 74238.31 KVM_CAP_PPC_MULTITCE 7424------------------------- 7425 7426:Capability: KVM_CAP_PPC_MULTITCE 7427:Architectures: ppc 7428:Type: vm 7429 7430This capability means the kernel is capable of handling hypercalls 7431H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user 7432space. This significantly accelerates DMA operations for PPC KVM guests. 7433User space should expect that its handlers for these hypercalls 7434are not going to be called if user space previously registered LIOBN 7435in KVM (via KVM_CREATE_SPAPR_TCE or similar calls). 7436 7437In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest, 7438user space might have to advertise it for the guest. For example, 7439IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is 7440present in the "ibm,hypertas-functions" device-tree property. 7441 7442The hypercalls mentioned above may or may not be processed successfully 7443in the kernel based fast path. If they can not be handled by the kernel, 7444they will get passed on to user space. So user space still has to have 7445an implementation for these despite the in kernel acceleration. 7446 7447This capability is always enabled. 7448 74498.32 KVM_CAP_PTP_KVM 7450-------------------- 7451 7452:Architectures: arm64 7453 7454This capability indicates that the KVM virtual PTP service is 7455supported in the host. A VMM can check whether the service is 7456available to the guest on migration. 7457 74588.33 KVM_CAP_HYPERV_ENFORCE_CPUID 7459--------------------------------- 7460 7461Architectures: x86 7462 7463When enabled, KVM will disable emulated Hyper-V features provided to the 7464guest according to the bits Hyper-V CPUID feature leaves. Otherwise, all 7465currently implmented Hyper-V features are provided unconditionally when 7466Hyper-V identification is set in the HYPERV_CPUID_INTERFACE (0x40000001) 7467leaf. 7468 74698.34 KVM_CAP_EXIT_HYPERCALL 7470--------------------------- 7471 7472:Capability: KVM_CAP_EXIT_HYPERCALL 7473:Architectures: x86 7474:Type: vm 7475 7476This capability, if enabled, will cause KVM to exit to userspace 7477with KVM_EXIT_HYPERCALL exit reason to process some hypercalls. 7478 7479Calling KVM_CHECK_EXTENSION for this capability will return a bitmask 7480of hypercalls that can be configured to exit to userspace. 7481Right now, the only such hypercall is KVM_HC_MAP_GPA_RANGE. 7482 7483The argument to KVM_ENABLE_CAP is also a bitmask, and must be a subset 7484of the result of KVM_CHECK_EXTENSION. KVM will forward to userspace 7485the hypercalls whose corresponding bit is in the argument, and return 7486ENOSYS for the others. 7487