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 centered around different kinds of file descriptors 11and ioctls that can be issued to these file descriptors. An initial 12open("/dev/kvm") obtains a handle to the kvm subsystem; this handle 13can be used to issue system ioctls. A KVM_CREATE_VM ioctl on this 14handle will create a VM file descriptor which can be used to issue VM 15ioctls. A KVM_CREATE_VCPU or KVM_CREATE_DEVICE ioctl on a VM fd will 16create a virtual cpu or device and return a file descriptor pointing to 17the new resource. 18 19In other words, the kvm API is a set of ioctls that are issued to 20different kinds of file descriptor in order to control various aspects of 21a virtual machine. Depending on the file descriptor that accepts them, 22ioctls belong to the following classes: 23 24 - System ioctls: These query and set global attributes which affect the 25 whole kvm subsystem. In addition a system ioctl is used to create 26 virtual machines. 27 28 - VM ioctls: These query and set attributes that affect an entire virtual 29 machine, for example memory layout. In addition a VM ioctl is used to 30 create virtual cpus (vcpus) and devices. 31 32 VM ioctls must be issued from the same process (address space) that was 33 used to create the VM. 34 35 - vcpu ioctls: These query and set attributes that control the operation 36 of a single virtual cpu. 37 38 vcpu ioctls should be issued from the same thread that was used to create 39 the vcpu, except for asynchronous vcpu ioctl that are marked as such in 40 the documentation. Otherwise, the first ioctl after switching threads 41 could see a performance impact. 42 43 - device ioctls: These query and set attributes that control the operation 44 of a single device. 45 46 device ioctls must be issued from the same process (address space) that 47 was used to create the VM. 48 49While most ioctls are specific to one kind of file descriptor, in some 50cases the same ioctl can belong to more than one class. 51 52The KVM API grew over time. For this reason, KVM defines many constants 53of the form ``KVM_CAP_*``, each corresponding to a set of functionality 54provided by one or more ioctls. Availability of these "capabilities" can 55be checked with :ref:`KVM_CHECK_EXTENSION <KVM_CHECK_EXTENSION>`. Some 56capabilities also need to be enabled for VMs or VCPUs where their 57functionality is desired (see :ref:`cap_enable` and :ref:`cap_enable_vm`). 58 59 602. Restrictions 61=============== 62 63In general file descriptors can be migrated among processes by means 64of fork() and the SCM_RIGHTS facility of unix domain socket. These 65kinds of tricks are explicitly not supported by kvm. While they will 66not cause harm to the host, their actual behavior is not guaranteed by 67the API. See "General description" for details on the ioctl usage 68model that is supported by KVM. 69 70It is important to note that although VM ioctls may only be issued from 71the process that created the VM, a VM's lifecycle is associated with its 72file descriptor, not its creator (process). In other words, the VM and 73its resources, *including the associated address space*, are not freed 74until the last reference to the VM's file descriptor has been released. 75For example, if fork() is issued after ioctl(KVM_CREATE_VM), the VM will 76not be freed until both the parent (original) process and its child have 77put their references to the VM's file descriptor. 78 79Because a VM's resources are not freed until the last reference to its 80file descriptor is released, creating additional references to a VM 81via fork(), dup(), etc... without careful consideration is strongly 82discouraged and may have unwanted side effects, e.g. memory allocated 83by and on behalf of the VM's process may not be freed/unaccounted when 84the VM is shut down. 85 86 873. Extensions 88============= 89 90As of Linux 2.6.22, the KVM ABI has been stabilized: no backward 91incompatible change are allowed. However, there is an extension 92facility that allows backward-compatible extensions to the API to be 93queried and used. 94 95The extension mechanism is not based on the Linux version number. 96Instead, kvm defines extension identifiers and a facility to query 97whether a particular extension identifier is available. If it is, a 98set of ioctls is available for application use. 99 100 1014. API description 102================== 103 104This section describes ioctls that can be used to control kvm guests. 105For each ioctl, the following information is provided along with a 106description: 107 108 Capability: 109 which KVM extension provides this ioctl. Can be 'basic', 110 which means that is will be provided by any kernel that supports 111 API version 12 (see :ref:`KVM_GET_API_VERSION <KVM_GET_API_VERSION>`), 112 or a KVM_CAP_xyz constant that can be checked with 113 :ref:`KVM_CHECK_EXTENSION <KVM_CHECK_EXTENSION>`. 114 115 Architectures: 116 which instruction set architectures provide this ioctl. 117 x86 includes both i386 and x86_64. 118 119 Type: 120 system, vm, or vcpu. 121 122 Parameters: 123 what parameters are accepted by the ioctl. 124 125 Returns: 126 the return value. General error numbers (EBADF, ENOMEM, EINVAL) 127 are not detailed, but errors with specific meanings are. 128 129 130.. _KVM_GET_API_VERSION: 131 1324.1 KVM_GET_API_VERSION 133----------------------- 134 135:Capability: basic 136:Architectures: all 137:Type: system ioctl 138:Parameters: none 139:Returns: the constant KVM_API_VERSION (=12) 140 141This identifies the API version as the stable kvm API. It is not 142expected that this number will change. However, Linux 2.6.20 and 1432.6.21 report earlier versions; these are not documented and not 144supported. Applications should refuse to run if KVM_GET_API_VERSION 145returns a value other than 12. If this check passes, all ioctls 146described as 'basic' will be available. 147 148 1494.2 KVM_CREATE_VM 150----------------- 151 152:Capability: basic 153:Architectures: all 154:Type: system ioctl 155:Parameters: machine type identifier (KVM_VM_*) 156:Returns: a VM fd that can be used to control the new virtual machine. 157 158The new VM has no virtual cpus and no memory. 159You probably want to use 0 as machine type. 160 161X86: 162^^^^ 163 164Supported X86 VM types can be queried via KVM_CAP_VM_TYPES. 165 166S390: 167^^^^^ 168 169In order to create user controlled virtual machines on S390, check 170KVM_CAP_S390_UCONTROL and use the flag KVM_VM_S390_UCONTROL as 171privileged user (CAP_SYS_ADMIN). 172 173MIPS: 174^^^^^ 175 176To use hardware assisted virtualization on MIPS (VZ ASE) rather than 177the default trap & emulate implementation (which changes the virtual 178memory layout to fit in user mode), check KVM_CAP_MIPS_VZ and use the 179flag KVM_VM_MIPS_VZ. 180 181ARM64: 182^^^^^^ 183 184On arm64, the physical address size for a VM (IPA Size limit) is limited 185to 40bits by default. The limit can be configured if the host supports the 186extension KVM_CAP_ARM_VM_IPA_SIZE. When supported, use 187KVM_VM_TYPE_ARM_IPA_SIZE(IPA_Bits) to set the size in the machine type 188identifier, where IPA_Bits is the maximum width of any physical 189address used by the VM. The IPA_Bits is encoded in bits[7-0] of the 190machine type identifier. 191 192e.g, to configure a guest to use 48bit physical address size:: 193 194 vm_fd = ioctl(dev_fd, KVM_CREATE_VM, KVM_VM_TYPE_ARM_IPA_SIZE(48)); 195 196The requested size (IPA_Bits) must be: 197 198 == ========================================================= 199 0 Implies default size, 40bits (for backward compatibility) 200 N Implies N bits, where N is a positive integer such that, 201 32 <= N <= Host_IPA_Limit 202 == ========================================================= 203 204Host_IPA_Limit is the maximum possible value for IPA_Bits on the host and 205is dependent on the CPU capability and the kernel configuration. The limit can 206be retrieved using KVM_CAP_ARM_VM_IPA_SIZE of the KVM_CHECK_EXTENSION 207ioctl() at run-time. 208 209Creation of the VM will fail if the requested IPA size (whether it is 210implicit or explicit) is unsupported on the host. 211 212Please note that configuring the IPA size does not affect the capability 213exposed by the guest CPUs in ID_AA64MMFR0_EL1[PARange]. It only affects 214size of the address translated by the stage2 level (guest physical to 215host physical address translations). 216 217 2184.3 KVM_GET_MSR_INDEX_LIST, KVM_GET_MSR_FEATURE_INDEX_LIST 219---------------------------------------------------------- 220 221:Capability: basic, KVM_CAP_GET_MSR_FEATURES for KVM_GET_MSR_FEATURE_INDEX_LIST 222:Architectures: x86 223:Type: system ioctl 224:Parameters: struct kvm_msr_list (in/out) 225:Returns: 0 on success; -1 on error 226 227Errors: 228 229 ====== ============================================================ 230 EFAULT the msr index list cannot be read from or written to 231 E2BIG the msr index list is too big to fit in the array specified by 232 the user. 233 ====== ============================================================ 234 235:: 236 237 struct kvm_msr_list { 238 __u32 nmsrs; /* number of msrs in entries */ 239 __u32 indices[0]; 240 }; 241 242The user fills in the size of the indices array in nmsrs, and in return 243kvm adjusts nmsrs to reflect the actual number of msrs and fills in the 244indices array with their numbers. 245 246KVM_GET_MSR_INDEX_LIST returns the guest msrs that are supported. The list 247varies by kvm version and host processor, but does not change otherwise. 248 249Note: if kvm indicates supports MCE (KVM_CAP_MCE), then the MCE bank MSRs are 250not returned in the MSR list, as different vcpus can have a different number 251of banks, as set via the KVM_X86_SETUP_MCE ioctl. 252 253KVM_GET_MSR_FEATURE_INDEX_LIST returns the list of MSRs that can be passed 254to the KVM_GET_MSRS system ioctl. This lets userspace probe host capabilities 255and processor features that are exposed via MSRs (e.g., VMX capabilities). 256This list also varies by kvm version and host processor, but does not change 257otherwise. 258 259 260.. _KVM_CHECK_EXTENSION: 261 2624.4 KVM_CHECK_EXTENSION 263----------------------- 264 265:Capability: basic, KVM_CAP_CHECK_EXTENSION_VM for vm ioctl 266:Architectures: all 267:Type: system ioctl, vm ioctl 268:Parameters: extension identifier (KVM_CAP_*) 269:Returns: 0 if unsupported; 1 (or some other positive integer) if supported 270 271The API allows the application to query about extensions to the core 272kvm API. Userspace passes an extension identifier (an integer) and 273receives an integer that describes the extension availability. 274Generally 0 means no and 1 means yes, but some extensions may report 275additional information in the integer return value. 276 277Based on their initialization different VMs may have different capabilities. 278It is thus encouraged to use the vm ioctl to query for capabilities (available 279with KVM_CAP_CHECK_EXTENSION_VM on the vm fd) 280 2814.5 KVM_GET_VCPU_MMAP_SIZE 282-------------------------- 283 284:Capability: basic 285:Architectures: all 286:Type: system ioctl 287:Parameters: none 288:Returns: size of vcpu mmap area, in bytes 289 290The KVM_RUN ioctl (cf.) communicates with userspace via a shared 291memory region. This ioctl returns the size of that region. See the 292KVM_RUN documentation for details. 293 294Besides the size of the KVM_RUN communication region, other areas of 295the VCPU file descriptor can be mmap-ed, including: 296 297- if KVM_CAP_COALESCED_MMIO is available, a page at 298 KVM_COALESCED_MMIO_PAGE_OFFSET * PAGE_SIZE; for historical reasons, 299 this page is included in the result of KVM_GET_VCPU_MMAP_SIZE. 300 KVM_CAP_COALESCED_MMIO is not documented yet. 301 302- if KVM_CAP_DIRTY_LOG_RING is available, a number of pages at 303 KVM_DIRTY_LOG_PAGE_OFFSET * PAGE_SIZE. For more information on 304 KVM_CAP_DIRTY_LOG_RING, see :ref:`KVM_CAP_DIRTY_LOG_RING`. 305 306 3074.7 KVM_CREATE_VCPU 308------------------- 309 310:Capability: basic 311:Architectures: all 312:Type: vm ioctl 313:Parameters: vcpu id (apic id on x86) 314:Returns: vcpu fd on success, -1 on error 315 316This API adds a vcpu to a virtual machine. No more than max_vcpus may be added. 317The vcpu id is an integer in the range [0, max_vcpu_id). 318 319The recommended max_vcpus value can be retrieved using the KVM_CAP_NR_VCPUS of 320the KVM_CHECK_EXTENSION ioctl() at run-time. 321The maximum possible value for max_vcpus can be retrieved using the 322KVM_CAP_MAX_VCPUS of the KVM_CHECK_EXTENSION ioctl() at run-time. 323 324If the KVM_CAP_NR_VCPUS does not exist, you should assume that max_vcpus is 4 325cpus max. 326If the KVM_CAP_MAX_VCPUS does not exist, you should assume that max_vcpus is 327same as the value returned from KVM_CAP_NR_VCPUS. 328 329The maximum possible value for max_vcpu_id can be retrieved using the 330KVM_CAP_MAX_VCPU_ID of the KVM_CHECK_EXTENSION ioctl() at run-time. 331 332If the KVM_CAP_MAX_VCPU_ID does not exist, you should assume that max_vcpu_id 333is the same as the value returned from KVM_CAP_MAX_VCPUS. 334 335On powerpc using book3s_hv mode, the vcpus are mapped onto virtual 336threads in one or more virtual CPU cores. (This is because the 337hardware requires all the hardware threads in a CPU core to be in the 338same partition.) The KVM_CAP_PPC_SMT capability indicates the number 339of vcpus per virtual core (vcore). The vcore id is obtained by 340dividing the vcpu id by the number of vcpus per vcore. The vcpus in a 341given vcore will always be in the same physical core as each other 342(though that might be a different physical core from time to time). 343Userspace can control the threading (SMT) mode of the guest by its 344allocation of vcpu ids. For example, if userspace wants 345single-threaded guest vcpus, it should make all vcpu ids be a multiple 346of the number of vcpus per vcore. 347 348For virtual cpus that have been created with S390 user controlled virtual 349machines, the resulting vcpu fd can be memory mapped at page offset 350KVM_S390_SIE_PAGE_OFFSET in order to obtain a memory map of the virtual 351cpu's hardware control block. 352 353 3544.8 KVM_GET_DIRTY_LOG 355--------------------- 356 357:Capability: basic 358:Architectures: all 359:Type: vm ioctl 360:Parameters: struct kvm_dirty_log (in/out) 361:Returns: 0 on success, -1 on error 362 363:: 364 365 /* for KVM_GET_DIRTY_LOG */ 366 struct kvm_dirty_log { 367 __u32 slot; 368 __u32 padding; 369 union { 370 void __user *dirty_bitmap; /* one bit per page */ 371 __u64 padding; 372 }; 373 }; 374 375Given a memory slot, return a bitmap containing any pages dirtied 376since the last call to this ioctl. Bit 0 is the first page in the 377memory slot. Ensure the entire structure is cleared to avoid padding 378issues. 379 380If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies 381the address space for which you want to return the dirty bitmap. See 382KVM_SET_USER_MEMORY_REGION for details on the usage of slot field. 383 384The bits in the dirty bitmap are cleared before the ioctl returns, unless 385KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is enabled. For more information, 386see the description of the capability. 387 388Note that the Xen shared_info page, if configured, shall always be assumed 389to be dirty. KVM will not explicitly mark it such. 390 391 3924.10 KVM_RUN 393------------ 394 395:Capability: basic 396:Architectures: all 397:Type: vcpu ioctl 398:Parameters: none 399:Returns: 0 on success, -1 on error 400 401Errors: 402 403 ======= ============================================================== 404 EINTR an unmasked signal is pending 405 ENOEXEC the vcpu hasn't been initialized or the guest tried to execute 406 instructions from device memory (arm64) 407 ENOSYS data abort outside memslots with no syndrome info and 408 KVM_CAP_ARM_NISV_TO_USER not enabled (arm64) 409 EPERM SVE feature set but not finalized (arm64) 410 ======= ============================================================== 411 412This ioctl is used to run a guest virtual cpu. While there are no 413explicit parameters, there is an implicit parameter block that can be 414obtained by mmap()ing the vcpu fd at offset 0, with the size given by 415KVM_GET_VCPU_MMAP_SIZE. The parameter block is formatted as a 'struct 416kvm_run' (see below). 417 418 4194.11 KVM_GET_REGS 420----------------- 421 422:Capability: basic 423:Architectures: all except arm64 424:Type: vcpu ioctl 425:Parameters: struct kvm_regs (out) 426:Returns: 0 on success, -1 on error 427 428Reads the general purpose registers from the vcpu. 429 430:: 431 432 /* x86 */ 433 struct kvm_regs { 434 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */ 435 __u64 rax, rbx, rcx, rdx; 436 __u64 rsi, rdi, rsp, rbp; 437 __u64 r8, r9, r10, r11; 438 __u64 r12, r13, r14, r15; 439 __u64 rip, rflags; 440 }; 441 442 /* mips */ 443 struct kvm_regs { 444 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */ 445 __u64 gpr[32]; 446 __u64 hi; 447 __u64 lo; 448 __u64 pc; 449 }; 450 451 /* LoongArch */ 452 struct kvm_regs { 453 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */ 454 unsigned long gpr[32]; 455 unsigned long pc; 456 }; 457 458 4594.12 KVM_SET_REGS 460----------------- 461 462:Capability: basic 463:Architectures: all except arm64 464:Type: vcpu ioctl 465:Parameters: struct kvm_regs (in) 466:Returns: 0 on success, -1 on error 467 468Writes the general purpose registers into the vcpu. 469 470See KVM_GET_REGS for the data structure. 471 472 4734.13 KVM_GET_SREGS 474------------------ 475 476:Capability: basic 477:Architectures: x86, ppc 478:Type: vcpu ioctl 479:Parameters: struct kvm_sregs (out) 480:Returns: 0 on success, -1 on error 481 482Reads special registers from the vcpu. 483 484:: 485 486 /* x86 */ 487 struct kvm_sregs { 488 struct kvm_segment cs, ds, es, fs, gs, ss; 489 struct kvm_segment tr, ldt; 490 struct kvm_dtable gdt, idt; 491 __u64 cr0, cr2, cr3, cr4, cr8; 492 __u64 efer; 493 __u64 apic_base; 494 __u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64]; 495 }; 496 497 /* ppc -- see arch/powerpc/include/uapi/asm/kvm.h */ 498 499interrupt_bitmap is a bitmap of pending external interrupts. At most 500one bit may be set. This interrupt has been acknowledged by the APIC 501but not yet injected into the cpu core. 502 503 5044.14 KVM_SET_SREGS 505------------------ 506 507:Capability: basic 508:Architectures: x86, ppc 509:Type: vcpu ioctl 510:Parameters: struct kvm_sregs (in) 511:Returns: 0 on success, -1 on error 512 513Writes special registers into the vcpu. See KVM_GET_SREGS for the 514data structures. 515 516 5174.15 KVM_TRANSLATE 518------------------ 519 520:Capability: basic 521:Architectures: x86 522:Type: vcpu ioctl 523:Parameters: struct kvm_translation (in/out) 524:Returns: 0 on success, -1 on error 525 526Translates a virtual address according to the vcpu's current address 527translation mode. 528 529:: 530 531 struct kvm_translation { 532 /* in */ 533 __u64 linear_address; 534 535 /* out */ 536 __u64 physical_address; 537 __u8 valid; 538 __u8 writeable; 539 __u8 usermode; 540 __u8 pad[5]; 541 }; 542 543 5444.16 KVM_INTERRUPT 545------------------ 546 547:Capability: basic 548:Architectures: x86, ppc, mips, riscv, loongarch 549:Type: vcpu ioctl 550:Parameters: struct kvm_interrupt (in) 551:Returns: 0 on success, negative on failure. 552 553Queues a hardware interrupt vector to be injected. 554 555:: 556 557 /* for KVM_INTERRUPT */ 558 struct kvm_interrupt { 559 /* in */ 560 __u32 irq; 561 }; 562 563X86: 564^^^^ 565 566:Returns: 567 568 ========= =================================== 569 0 on success, 570 -EEXIST if an interrupt is already enqueued 571 -EINVAL the irq number is invalid 572 -ENXIO if the PIC is in the kernel 573 -EFAULT if the pointer is invalid 574 ========= =================================== 575 576Note 'irq' is an interrupt vector, not an interrupt pin or line. This 577ioctl is useful if the in-kernel PIC is not used. 578 579PPC: 580^^^^ 581 582Queues an external interrupt to be injected. This ioctl is overloaded 583with 3 different irq values: 584 585a) KVM_INTERRUPT_SET 586 587 This injects an edge type external interrupt into the guest once it's ready 588 to receive interrupts. When injected, the interrupt is done. 589 590b) KVM_INTERRUPT_UNSET 591 592 This unsets any pending interrupt. 593 594 Only available with KVM_CAP_PPC_UNSET_IRQ. 595 596c) KVM_INTERRUPT_SET_LEVEL 597 598 This injects a level type external interrupt into the guest context. The 599 interrupt stays pending until a specific ioctl with KVM_INTERRUPT_UNSET 600 is triggered. 601 602 Only available with KVM_CAP_PPC_IRQ_LEVEL. 603 604Note that any value for 'irq' other than the ones stated above is invalid 605and incurs unexpected behavior. 606 607This is an asynchronous vcpu ioctl and can be invoked from any thread. 608 609MIPS: 610^^^^^ 611 612Queues an external interrupt to be injected into the virtual CPU. A negative 613interrupt number dequeues the interrupt. 614 615This is an asynchronous vcpu ioctl and can be invoked from any thread. 616 617RISC-V: 618^^^^^^^ 619 620Queues an external interrupt to be injected into the virtual CPU. This ioctl 621is overloaded with 2 different irq values: 622 623a) KVM_INTERRUPT_SET 624 625 This sets external interrupt for a virtual CPU and it will receive 626 once it is ready. 627 628b) KVM_INTERRUPT_UNSET 629 630 This clears pending external interrupt for a virtual CPU. 631 632This is an asynchronous vcpu ioctl and can be invoked from any thread. 633 634LOONGARCH: 635^^^^^^^^^^ 636 637Queues an external interrupt to be injected into the virtual CPU. A negative 638interrupt number dequeues the interrupt. 639 640This is an asynchronous vcpu ioctl and can be invoked from any thread. 641 642 6434.18 KVM_GET_MSRS 644----------------- 645 646:Capability: basic (vcpu), KVM_CAP_GET_MSR_FEATURES (system) 647:Architectures: x86 648:Type: system ioctl, vcpu ioctl 649:Parameters: struct kvm_msrs (in/out) 650:Returns: number of msrs successfully returned; 651 -1 on error 652 653When used as a system ioctl: 654Reads the values of MSR-based features that are available for the VM. This 655is similar to KVM_GET_SUPPORTED_CPUID, but it returns MSR indices and values. 656The list of msr-based features can be obtained using KVM_GET_MSR_FEATURE_INDEX_LIST 657in a system ioctl. 658 659When used as a vcpu ioctl: 660Reads model-specific registers from the vcpu. Supported msr indices can 661be obtained using KVM_GET_MSR_INDEX_LIST in a system ioctl. 662 663:: 664 665 struct kvm_msrs { 666 __u32 nmsrs; /* number of msrs in entries */ 667 __u32 pad; 668 669 struct kvm_msr_entry entries[0]; 670 }; 671 672 struct kvm_msr_entry { 673 __u32 index; 674 __u32 reserved; 675 __u64 data; 676 }; 677 678Application code should set the 'nmsrs' member (which indicates the 679size of the entries array) and the 'index' member of each array entry. 680kvm will fill in the 'data' member. 681 682 6834.19 KVM_SET_MSRS 684----------------- 685 686:Capability: basic 687:Architectures: x86 688:Type: vcpu ioctl 689:Parameters: struct kvm_msrs (in) 690:Returns: number of msrs successfully set (see below), -1 on error 691 692Writes model-specific registers to the vcpu. See KVM_GET_MSRS for the 693data structures. 694 695Application code should set the 'nmsrs' member (which indicates the 696size of the entries array), and the 'index' and 'data' members of each 697array entry. 698 699It tries to set the MSRs in array entries[] one by one. If setting an MSR 700fails, e.g., due to setting reserved bits, the MSR isn't supported/emulated 701by KVM, etc..., it stops processing the MSR list and returns the number of 702MSRs that have been set successfully. 703 704 7054.20 KVM_SET_CPUID 706------------------ 707 708:Capability: basic 709:Architectures: x86 710:Type: vcpu ioctl 711:Parameters: struct kvm_cpuid (in) 712:Returns: 0 on success, -1 on error 713 714Defines the vcpu responses to the cpuid instruction. Applications 715should use the KVM_SET_CPUID2 ioctl if available. 716 717Caveat emptor: 718 - If this IOCTL fails, KVM gives no guarantees that previous valid CPUID 719 configuration (if there is) is not corrupted. Userspace can get a copy 720 of the resulting CPUID configuration through KVM_GET_CPUID2 in case. 721 - Using KVM_SET_CPUID{,2} after KVM_RUN, i.e. changing the guest vCPU model 722 after running the guest, may cause guest instability. 723 - Using heterogeneous CPUID configurations, modulo APIC IDs, topology, etc... 724 may cause guest instability. 725 726:: 727 728 struct kvm_cpuid_entry { 729 __u32 function; 730 __u32 eax; 731 __u32 ebx; 732 __u32 ecx; 733 __u32 edx; 734 __u32 padding; 735 }; 736 737 /* for KVM_SET_CPUID */ 738 struct kvm_cpuid { 739 __u32 nent; 740 __u32 padding; 741 struct kvm_cpuid_entry entries[0]; 742 }; 743 744 7454.21 KVM_SET_SIGNAL_MASK 746------------------------ 747 748:Capability: basic 749:Architectures: all 750:Type: vcpu ioctl 751:Parameters: struct kvm_signal_mask (in) 752:Returns: 0 on success, -1 on error 753 754Defines which signals are blocked during execution of KVM_RUN. This 755signal mask temporarily overrides the threads signal mask. Any 756unblocked signal received (except SIGKILL and SIGSTOP, which retain 757their traditional behaviour) will cause KVM_RUN to return with -EINTR. 758 759Note the signal will only be delivered if not blocked by the original 760signal mask. 761 762:: 763 764 /* for KVM_SET_SIGNAL_MASK */ 765 struct kvm_signal_mask { 766 __u32 len; 767 __u8 sigset[0]; 768 }; 769 770 7714.22 KVM_GET_FPU 772---------------- 773 774:Capability: basic 775:Architectures: x86, loongarch 776:Type: vcpu ioctl 777:Parameters: struct kvm_fpu (out) 778:Returns: 0 on success, -1 on error 779 780Reads the floating point state from the vcpu. 781 782:: 783 784 /* x86: for KVM_GET_FPU and KVM_SET_FPU */ 785 struct kvm_fpu { 786 __u8 fpr[8][16]; 787 __u16 fcw; 788 __u16 fsw; 789 __u8 ftwx; /* in fxsave format */ 790 __u8 pad1; 791 __u16 last_opcode; 792 __u64 last_ip; 793 __u64 last_dp; 794 __u8 xmm[16][16]; 795 __u32 mxcsr; 796 __u32 pad2; 797 }; 798 799 /* LoongArch: for KVM_GET_FPU and KVM_SET_FPU */ 800 struct kvm_fpu { 801 __u32 fcsr; 802 __u64 fcc; 803 struct kvm_fpureg { 804 __u64 val64[4]; 805 }fpr[32]; 806 }; 807 808 8094.23 KVM_SET_FPU 810---------------- 811 812:Capability: basic 813:Architectures: x86, loongarch 814:Type: vcpu ioctl 815:Parameters: struct kvm_fpu (in) 816:Returns: 0 on success, -1 on error 817 818Writes the floating point state to the vcpu. 819 820:: 821 822 /* x86: for KVM_GET_FPU and KVM_SET_FPU */ 823 struct kvm_fpu { 824 __u8 fpr[8][16]; 825 __u16 fcw; 826 __u16 fsw; 827 __u8 ftwx; /* in fxsave format */ 828 __u8 pad1; 829 __u16 last_opcode; 830 __u64 last_ip; 831 __u64 last_dp; 832 __u8 xmm[16][16]; 833 __u32 mxcsr; 834 __u32 pad2; 835 }; 836 837 /* LoongArch: for KVM_GET_FPU and KVM_SET_FPU */ 838 struct kvm_fpu { 839 __u32 fcsr; 840 __u64 fcc; 841 struct kvm_fpureg { 842 __u64 val64[4]; 843 }fpr[32]; 844 }; 845 846 8474.24 KVM_CREATE_IRQCHIP 848----------------------- 849 850:Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQCHIP (s390) 851:Architectures: x86, arm64, s390 852:Type: vm ioctl 853:Parameters: none 854:Returns: 0 on success, -1 on error 855 856Creates an interrupt controller model in the kernel. 857On x86, creates a virtual ioapic, a virtual PIC (two PICs, nested), and sets up 858future vcpus to have a local APIC. IRQ routing for GSIs 0-15 is set to both 859PIC and IOAPIC; GSI 16-23 only go to the IOAPIC. 860On arm64, a GICv2 is created. Any other GIC versions require the usage of 861KVM_CREATE_DEVICE, which also supports creating a GICv2. Using 862KVM_CREATE_DEVICE is preferred over KVM_CREATE_IRQCHIP for GICv2. 863On s390, a dummy irq routing table is created. 864 865Note that on s390 the KVM_CAP_S390_IRQCHIP vm capability needs to be enabled 866before KVM_CREATE_IRQCHIP can be used. 867 868 8694.25 KVM_IRQ_LINE 870----------------- 871 872:Capability: KVM_CAP_IRQCHIP 873:Architectures: x86, arm64 874:Type: vm ioctl 875:Parameters: struct kvm_irq_level 876:Returns: 0 on success, -1 on error 877 878Sets the level of a GSI input to the interrupt controller model in the kernel. 879On some architectures it is required that an interrupt controller model has 880been previously created with KVM_CREATE_IRQCHIP. Note that edge-triggered 881interrupts require the level to be set to 1 and then back to 0. 882 883On real hardware, interrupt pins can be active-low or active-high. This 884does not matter for the level field of struct kvm_irq_level: 1 always 885means active (asserted), 0 means inactive (deasserted). 886 887x86 allows the operating system to program the interrupt polarity 888(active-low/active-high) for level-triggered interrupts, and KVM used 889to consider the polarity. However, due to bitrot in the handling of 890active-low interrupts, the above convention is now valid on x86 too. 891This is signaled by KVM_CAP_X86_IOAPIC_POLARITY_IGNORED. Userspace 892should not present interrupts to the guest as active-low unless this 893capability is present (or unless it is not using the in-kernel irqchip, 894of course). 895 896 897arm64 can signal an interrupt either at the CPU level, or at the 898in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to 899use PPIs designated for specific cpus. The irq field is interpreted 900like this:: 901 902 bits: | 31 ... 28 | 27 ... 24 | 23 ... 16 | 15 ... 0 | 903 field: | vcpu2_index | irq_type | vcpu_index | irq_id | 904 905The irq_type field has the following values: 906 907- KVM_ARM_IRQ_TYPE_CPU: 908 out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ 909- KVM_ARM_IRQ_TYPE_SPI: 910 in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.) 911 (the vcpu_index field is ignored) 912- KVM_ARM_IRQ_TYPE_PPI: 913 in-kernel GIC: PPI, irq_id between 16 and 31 (incl.) 914 915(The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs) 916 917In both cases, level is used to assert/deassert the line. 918 919When KVM_CAP_ARM_IRQ_LINE_LAYOUT_2 is supported, the target vcpu is 920identified as (256 * vcpu2_index + vcpu_index). Otherwise, vcpu2_index 921must be zero. 922 923Note that on arm64, the KVM_CAP_IRQCHIP capability only conditions 924injection of interrupts for the in-kernel irqchip. KVM_IRQ_LINE can always 925be used for a userspace interrupt controller. 926 927:: 928 929 struct kvm_irq_level { 930 union { 931 __u32 irq; /* GSI */ 932 __s32 status; /* not used for KVM_IRQ_LEVEL */ 933 }; 934 __u32 level; /* 0 or 1 */ 935 }; 936 937 9384.26 KVM_GET_IRQCHIP 939-------------------- 940 941:Capability: KVM_CAP_IRQCHIP 942:Architectures: x86 943:Type: vm ioctl 944:Parameters: struct kvm_irqchip (in/out) 945:Returns: 0 on success, -1 on error 946 947Reads the state of a kernel interrupt controller created with 948KVM_CREATE_IRQCHIP into a buffer provided by the caller. 949 950:: 951 952 struct kvm_irqchip { 953 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */ 954 __u32 pad; 955 union { 956 char dummy[512]; /* reserving space */ 957 struct kvm_pic_state pic; 958 struct kvm_ioapic_state ioapic; 959 } chip; 960 }; 961 962 9634.27 KVM_SET_IRQCHIP 964-------------------- 965 966:Capability: KVM_CAP_IRQCHIP 967:Architectures: x86 968:Type: vm ioctl 969:Parameters: struct kvm_irqchip (in) 970:Returns: 0 on success, -1 on error 971 972Sets the state of a kernel interrupt controller created with 973KVM_CREATE_IRQCHIP from a buffer provided by the caller. 974 975:: 976 977 struct kvm_irqchip { 978 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */ 979 __u32 pad; 980 union { 981 char dummy[512]; /* reserving space */ 982 struct kvm_pic_state pic; 983 struct kvm_ioapic_state ioapic; 984 } chip; 985 }; 986 987 9884.28 KVM_XEN_HVM_CONFIG 989----------------------- 990 991:Capability: KVM_CAP_XEN_HVM 992:Architectures: x86 993:Type: vm ioctl 994:Parameters: struct kvm_xen_hvm_config (in) 995:Returns: 0 on success, -1 on error 996 997Sets the MSR that the Xen HVM guest uses to initialize its hypercall 998page, and provides the starting address and size of the hypercall 999blobs in userspace. When the guest writes the MSR, kvm copies one 1000page of a blob (32- or 64-bit, depending on the vcpu mode) to guest 1001memory. 1002 1003:: 1004 1005 struct kvm_xen_hvm_config { 1006 __u32 flags; 1007 __u32 msr; 1008 __u64 blob_addr_32; 1009 __u64 blob_addr_64; 1010 __u8 blob_size_32; 1011 __u8 blob_size_64; 1012 __u8 pad2[30]; 1013 }; 1014 1015If certain flags are returned from the KVM_CAP_XEN_HVM check, they may 1016be set in the flags field of this ioctl: 1017 1018The KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL flag requests KVM to generate 1019the contents of the hypercall page automatically; hypercalls will be 1020intercepted and passed to userspace through KVM_EXIT_XEN. In this 1021case, all of the blob size and address fields must be zero. 1022 1023The KVM_XEN_HVM_CONFIG_EVTCHN_SEND flag indicates to KVM that userspace 1024will always use the KVM_XEN_HVM_EVTCHN_SEND ioctl to deliver event 1025channel interrupts rather than manipulating the guest's shared_info 1026structures directly. This, in turn, may allow KVM to enable features 1027such as intercepting the SCHEDOP_poll hypercall to accelerate PV 1028spinlock operation for the guest. Userspace may still use the ioctl 1029to deliver events if it was advertised, even if userspace does not 1030send this indication that it will always do so 1031 1032No other flags are currently valid in the struct kvm_xen_hvm_config. 1033 10344.29 KVM_GET_CLOCK 1035------------------ 1036 1037:Capability: KVM_CAP_ADJUST_CLOCK 1038:Architectures: x86 1039:Type: vm ioctl 1040:Parameters: struct kvm_clock_data (out) 1041:Returns: 0 on success, -1 on error 1042 1043Gets the current timestamp of kvmclock as seen by the current guest. In 1044conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios 1045such as migration. 1046 1047When KVM_CAP_ADJUST_CLOCK is passed to KVM_CHECK_EXTENSION, it returns the 1048set of bits that KVM can return in struct kvm_clock_data's flag member. 1049 1050The following flags are defined: 1051 1052KVM_CLOCK_TSC_STABLE 1053 If set, the returned value is the exact kvmclock 1054 value seen by all VCPUs at the instant when KVM_GET_CLOCK was called. 1055 If clear, the returned value is simply CLOCK_MONOTONIC plus a constant 1056 offset; the offset can be modified with KVM_SET_CLOCK. KVM will try 1057 to make all VCPUs follow this clock, but the exact value read by each 1058 VCPU could differ, because the host TSC is not stable. 1059 1060KVM_CLOCK_REALTIME 1061 If set, the `realtime` field in the kvm_clock_data 1062 structure is populated with the value of the host's real time 1063 clocksource at the instant when KVM_GET_CLOCK was called. If clear, 1064 the `realtime` field does not contain a value. 1065 1066KVM_CLOCK_HOST_TSC 1067 If set, the `host_tsc` field in the kvm_clock_data 1068 structure is populated with the value of the host's timestamp counter (TSC) 1069 at the instant when KVM_GET_CLOCK was called. If clear, the `host_tsc` field 1070 does not contain a value. 1071 1072:: 1073 1074 struct kvm_clock_data { 1075 __u64 clock; /* kvmclock current value */ 1076 __u32 flags; 1077 __u32 pad0; 1078 __u64 realtime; 1079 __u64 host_tsc; 1080 __u32 pad[4]; 1081 }; 1082 1083 10844.30 KVM_SET_CLOCK 1085------------------ 1086 1087:Capability: KVM_CAP_ADJUST_CLOCK 1088:Architectures: x86 1089:Type: vm ioctl 1090:Parameters: struct kvm_clock_data (in) 1091:Returns: 0 on success, -1 on error 1092 1093Sets the current timestamp of kvmclock to the value specified in its parameter. 1094In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios 1095such as migration. 1096 1097The following flags can be passed: 1098 1099KVM_CLOCK_REALTIME 1100 If set, KVM will compare the value of the `realtime` field 1101 with the value of the host's real time clocksource at the instant when 1102 KVM_SET_CLOCK was called. The difference in elapsed time is added to the final 1103 kvmclock value that will be provided to guests. 1104 1105Other flags returned by ``KVM_GET_CLOCK`` are accepted but ignored. 1106 1107:: 1108 1109 struct kvm_clock_data { 1110 __u64 clock; /* kvmclock current value */ 1111 __u32 flags; 1112 __u32 pad0; 1113 __u64 realtime; 1114 __u64 host_tsc; 1115 __u32 pad[4]; 1116 }; 1117 1118 11194.31 KVM_GET_VCPU_EVENTS 1120------------------------ 1121 1122:Capability: KVM_CAP_VCPU_EVENTS 1123:Extended by: KVM_CAP_INTR_SHADOW 1124:Architectures: x86, arm64 1125:Type: vcpu ioctl 1126:Parameters: struct kvm_vcpu_events (out) 1127:Returns: 0 on success, -1 on error 1128 1129X86: 1130^^^^ 1131 1132Gets currently pending exceptions, interrupts, and NMIs as well as related 1133states of the vcpu. 1134 1135:: 1136 1137 struct kvm_vcpu_events { 1138 struct { 1139 __u8 injected; 1140 __u8 nr; 1141 __u8 has_error_code; 1142 __u8 pending; 1143 __u32 error_code; 1144 } exception; 1145 struct { 1146 __u8 injected; 1147 __u8 nr; 1148 __u8 soft; 1149 __u8 shadow; 1150 } interrupt; 1151 struct { 1152 __u8 injected; 1153 __u8 pending; 1154 __u8 masked; 1155 __u8 pad; 1156 } nmi; 1157 __u32 sipi_vector; 1158 __u32 flags; 1159 struct { 1160 __u8 smm; 1161 __u8 pending; 1162 __u8 smm_inside_nmi; 1163 __u8 latched_init; 1164 } smi; 1165 __u8 reserved[27]; 1166 __u8 exception_has_payload; 1167 __u64 exception_payload; 1168 }; 1169 1170The following bits are defined in the flags field: 1171 1172- KVM_VCPUEVENT_VALID_SHADOW may be set to signal that 1173 interrupt.shadow contains a valid state. 1174 1175- KVM_VCPUEVENT_VALID_SMM may be set to signal that smi contains a 1176 valid state. 1177 1178- KVM_VCPUEVENT_VALID_PAYLOAD may be set to signal that the 1179 exception_has_payload, exception_payload, and exception.pending 1180 fields contain a valid state. This bit will be set whenever 1181 KVM_CAP_EXCEPTION_PAYLOAD is enabled. 1182 1183- KVM_VCPUEVENT_VALID_TRIPLE_FAULT may be set to signal that the 1184 triple_fault_pending field contains a valid state. This bit will 1185 be set whenever KVM_CAP_X86_TRIPLE_FAULT_EVENT is enabled. 1186 1187ARM64: 1188^^^^^^ 1189 1190If the guest accesses a device that is being emulated by the host kernel in 1191such a way that a real device would generate a physical SError, KVM may make 1192a virtual SError pending for that VCPU. This system error interrupt remains 1193pending until the guest takes the exception by unmasking PSTATE.A. 1194 1195Running the VCPU may cause it to take a pending SError, or make an access that 1196causes an SError to become pending. The event's description is only valid while 1197the VPCU is not running. 1198 1199This API provides a way to read and write the pending 'event' state that is not 1200visible to the guest. To save, restore or migrate a VCPU the struct representing 1201the state can be read then written using this GET/SET API, along with the other 1202guest-visible registers. It is not possible to 'cancel' an SError that has been 1203made pending. 1204 1205A device being emulated in user-space may also wish to generate an SError. To do 1206this the events structure can be populated by user-space. The current state 1207should be read first, to ensure no existing SError is pending. If an existing 1208SError is pending, the architecture's 'Multiple SError interrupts' rules should 1209be followed. (2.5.3 of DDI0587.a "ARM Reliability, Availability, and 1210Serviceability (RAS) Specification"). 1211 1212SError exceptions always have an ESR value. Some CPUs have the ability to 1213specify what the virtual SError's ESR value should be. These systems will 1214advertise KVM_CAP_ARM_INJECT_SERROR_ESR. In this case exception.has_esr will 1215always have a non-zero value when read, and the agent making an SError pending 1216should specify the ISS field in the lower 24 bits of exception.serror_esr. If 1217the system supports KVM_CAP_ARM_INJECT_SERROR_ESR, but user-space sets the events 1218with exception.has_esr as zero, KVM will choose an ESR. 1219 1220Specifying exception.has_esr on a system that does not support it will return 1221-EINVAL. Setting anything other than the lower 24bits of exception.serror_esr 1222will return -EINVAL. 1223 1224It is not possible to read back a pending external abort (injected via 1225KVM_SET_VCPU_EVENTS or otherwise) because such an exception is always delivered 1226directly to the virtual CPU). 1227 1228:: 1229 1230 struct kvm_vcpu_events { 1231 struct { 1232 __u8 serror_pending; 1233 __u8 serror_has_esr; 1234 __u8 ext_dabt_pending; 1235 /* Align it to 8 bytes */ 1236 __u8 pad[5]; 1237 __u64 serror_esr; 1238 } exception; 1239 __u32 reserved[12]; 1240 }; 1241 12424.32 KVM_SET_VCPU_EVENTS 1243------------------------ 1244 1245:Capability: KVM_CAP_VCPU_EVENTS 1246:Extended by: KVM_CAP_INTR_SHADOW 1247:Architectures: x86, arm64 1248:Type: vcpu ioctl 1249:Parameters: struct kvm_vcpu_events (in) 1250:Returns: 0 on success, -1 on error 1251 1252X86: 1253^^^^ 1254 1255Set pending exceptions, interrupts, and NMIs as well as related states of the 1256vcpu. 1257 1258See KVM_GET_VCPU_EVENTS for the data structure. 1259 1260Fields that may be modified asynchronously by running VCPUs can be excluded 1261from the update. These fields are nmi.pending, sipi_vector, smi.smm, 1262smi.pending. Keep the corresponding bits in the flags field cleared to 1263suppress overwriting the current in-kernel state. The bits are: 1264 1265=============================== ================================== 1266KVM_VCPUEVENT_VALID_NMI_PENDING transfer nmi.pending to the kernel 1267KVM_VCPUEVENT_VALID_SIPI_VECTOR transfer sipi_vector 1268KVM_VCPUEVENT_VALID_SMM transfer the smi sub-struct. 1269=============================== ================================== 1270 1271If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in 1272the flags field to signal that interrupt.shadow contains a valid state and 1273shall be written into the VCPU. 1274 1275KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available. 1276 1277If KVM_CAP_EXCEPTION_PAYLOAD is enabled, KVM_VCPUEVENT_VALID_PAYLOAD 1278can be set in the flags field to signal that the 1279exception_has_payload, exception_payload, and exception.pending fields 1280contain a valid state and shall be written into the VCPU. 1281 1282If KVM_CAP_X86_TRIPLE_FAULT_EVENT is enabled, KVM_VCPUEVENT_VALID_TRIPLE_FAULT 1283can be set in flags field to signal that the triple_fault field contains 1284a valid state and shall be written into the VCPU. 1285 1286ARM64: 1287^^^^^^ 1288 1289User space may need to inject several types of events to the guest. 1290 1291Set the pending SError exception state for this VCPU. It is not possible to 1292'cancel' an Serror that has been made pending. 1293 1294If the guest performed an access to I/O memory which could not be handled by 1295userspace, for example because of missing instruction syndrome decode 1296information or because there is no device mapped at the accessed IPA, then 1297userspace can ask the kernel to inject an external abort using the address 1298from the exiting fault on the VCPU. It is a programming error to set 1299ext_dabt_pending after an exit which was not either KVM_EXIT_MMIO or 1300KVM_EXIT_ARM_NISV. This feature is only available if the system supports 1301KVM_CAP_ARM_INJECT_EXT_DABT. This is a helper which provides commonality in 1302how userspace reports accesses for the above cases to guests, across different 1303userspace implementations. Nevertheless, userspace can still emulate all Arm 1304exceptions by manipulating individual registers using the KVM_SET_ONE_REG API. 1305 1306See KVM_GET_VCPU_EVENTS for the data structure. 1307 1308 13094.33 KVM_GET_DEBUGREGS 1310---------------------- 1311 1312:Capability: KVM_CAP_DEBUGREGS 1313:Architectures: x86 1314:Type: vcpu ioctl 1315:Parameters: struct kvm_debugregs (out) 1316:Returns: 0 on success, -1 on error 1317 1318Reads debug registers from the vcpu. 1319 1320:: 1321 1322 struct kvm_debugregs { 1323 __u64 db[4]; 1324 __u64 dr6; 1325 __u64 dr7; 1326 __u64 flags; 1327 __u64 reserved[9]; 1328 }; 1329 1330 13314.34 KVM_SET_DEBUGREGS 1332---------------------- 1333 1334:Capability: KVM_CAP_DEBUGREGS 1335:Architectures: x86 1336:Type: vcpu ioctl 1337:Parameters: struct kvm_debugregs (in) 1338:Returns: 0 on success, -1 on error 1339 1340Writes debug registers into the vcpu. 1341 1342See KVM_GET_DEBUGREGS for the data structure. The flags field is unused 1343yet and must be cleared on entry. 1344 1345 13464.35 KVM_SET_USER_MEMORY_REGION 1347------------------------------- 1348 1349:Capability: KVM_CAP_USER_MEMORY 1350:Architectures: all 1351:Type: vm ioctl 1352:Parameters: struct kvm_userspace_memory_region (in) 1353:Returns: 0 on success, -1 on error 1354 1355:: 1356 1357 struct kvm_userspace_memory_region { 1358 __u32 slot; 1359 __u32 flags; 1360 __u64 guest_phys_addr; 1361 __u64 memory_size; /* bytes */ 1362 __u64 userspace_addr; /* start of the userspace allocated memory */ 1363 }; 1364 1365 /* for kvm_userspace_memory_region::flags */ 1366 #define KVM_MEM_LOG_DIRTY_PAGES (1UL << 0) 1367 #define KVM_MEM_READONLY (1UL << 1) 1368 1369This ioctl allows the user to create, modify or delete a guest physical 1370memory slot. Bits 0-15 of "slot" specify the slot id and this value 1371should be less than the maximum number of user memory slots supported per 1372VM. The maximum allowed slots can be queried using KVM_CAP_NR_MEMSLOTS. 1373Slots may not overlap in guest physical address space. 1374 1375If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot" 1376specifies the address space which is being modified. They must be 1377less than the value that KVM_CHECK_EXTENSION returns for the 1378KVM_CAP_MULTI_ADDRESS_SPACE capability. Slots in separate address spaces 1379are unrelated; the restriction on overlapping slots only applies within 1380each address space. 1381 1382Deleting a slot is done by passing zero for memory_size. When changing 1383an existing slot, it may be moved in the guest physical memory space, 1384or its flags may be modified, but it may not be resized. 1385 1386Memory for the region is taken starting at the address denoted by the 1387field userspace_addr, which must point at user addressable memory for 1388the entire memory slot size. Any object may back this memory, including 1389anonymous memory, ordinary files, and hugetlbfs. 1390 1391On architectures that support a form of address tagging, userspace_addr must 1392be an untagged address. 1393 1394It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr 1395be identical. This allows large pages in the guest to be backed by large 1396pages in the host. 1397 1398The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and 1399KVM_MEM_READONLY. The former can be set to instruct KVM to keep track of 1400writes to memory within the slot. See KVM_GET_DIRTY_LOG ioctl to know how to 1401use it. The latter can be set, if KVM_CAP_READONLY_MEM capability allows it, 1402to make a new slot read-only. In this case, writes to this memory will be 1403posted to userspace as KVM_EXIT_MMIO exits. 1404 1405When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of 1406the memory region are automatically reflected into the guest. For example, an 1407mmap() that affects the region will be made visible immediately. Another 1408example is madvise(MADV_DROP). 1409 1410Note: On arm64, a write generated by the page-table walker (to update 1411the Access and Dirty flags, for example) never results in a 1412KVM_EXIT_MMIO exit when the slot has the KVM_MEM_READONLY flag. This 1413is because KVM cannot provide the data that would be written by the 1414page-table walker, making it impossible to emulate the access. 1415Instead, an abort (data abort if the cause of the page-table update 1416was a load or a store, instruction abort if it was an instruction 1417fetch) is injected in the guest. 1418 1419S390: 1420^^^^^ 1421 1422Returns -EINVAL or -EEXIST if the VM has the KVM_VM_S390_UCONTROL flag set. 1423Returns -EINVAL if called on a protected VM. 1424 14254.36 KVM_SET_TSS_ADDR 1426--------------------- 1427 1428:Capability: KVM_CAP_SET_TSS_ADDR 1429:Architectures: x86 1430:Type: vm ioctl 1431:Parameters: unsigned long tss_address (in) 1432:Returns: 0 on success, -1 on error 1433 1434This ioctl defines the physical address of a three-page region in the guest 1435physical address space. The region must be within the first 4GB of the 1436guest physical address space and must not conflict with any memory slot 1437or any mmio address. The guest may malfunction if it accesses this memory 1438region. 1439 1440This ioctl is required on Intel-based hosts. This is needed on Intel hardware 1441because of a quirk in the virtualization implementation (see the internals 1442documentation when it pops into existence). 1443 1444 1445.. _KVM_ENABLE_CAP: 1446 14474.37 KVM_ENABLE_CAP 1448------------------- 1449 1450:Capability: KVM_CAP_ENABLE_CAP 1451:Architectures: mips, ppc, s390, x86, loongarch 1452:Type: vcpu ioctl 1453:Parameters: struct kvm_enable_cap (in) 1454:Returns: 0 on success; -1 on error 1455 1456:Capability: KVM_CAP_ENABLE_CAP_VM 1457:Architectures: all 1458:Type: vm ioctl 1459:Parameters: struct kvm_enable_cap (in) 1460:Returns: 0 on success; -1 on error 1461 1462.. note:: 1463 1464 Not all extensions are enabled by default. Using this ioctl the application 1465 can enable an extension, making it available to the guest. 1466 1467On systems that do not support this ioctl, it always fails. On systems that 1468do support it, it only works for extensions that are supported for enablement. 1469 1470To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should 1471be used. 1472 1473:: 1474 1475 struct kvm_enable_cap { 1476 /* in */ 1477 __u32 cap; 1478 1479The capability that is supposed to get enabled. 1480 1481:: 1482 1483 __u32 flags; 1484 1485A bitfield indicating future enhancements. Has to be 0 for now. 1486 1487:: 1488 1489 __u64 args[4]; 1490 1491Arguments for enabling a feature. If a feature needs initial values to 1492function properly, this is the place to put them. 1493 1494:: 1495 1496 __u8 pad[64]; 1497 }; 1498 1499The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl 1500for vm-wide capabilities. 1501 15024.38 KVM_GET_MP_STATE 1503--------------------- 1504 1505:Capability: KVM_CAP_MP_STATE 1506:Architectures: x86, s390, arm64, riscv, loongarch 1507:Type: vcpu ioctl 1508:Parameters: struct kvm_mp_state (out) 1509:Returns: 0 on success; -1 on error 1510 1511:: 1512 1513 struct kvm_mp_state { 1514 __u32 mp_state; 1515 }; 1516 1517Returns the vcpu's current "multiprocessing state" (though also valid on 1518uniprocessor guests). 1519 1520Possible values are: 1521 1522 ========================== =============================================== 1523 KVM_MP_STATE_RUNNABLE the vcpu is currently running 1524 [x86,arm64,riscv,loongarch] 1525 KVM_MP_STATE_UNINITIALIZED the vcpu is an application processor (AP) 1526 which has not yet received an INIT signal [x86] 1527 KVM_MP_STATE_INIT_RECEIVED the vcpu has received an INIT signal, and is 1528 now ready for a SIPI [x86] 1529 KVM_MP_STATE_HALTED the vcpu has executed a HLT instruction and 1530 is waiting for an interrupt [x86] 1531 KVM_MP_STATE_SIPI_RECEIVED the vcpu has just received a SIPI (vector 1532 accessible via KVM_GET_VCPU_EVENTS) [x86] 1533 KVM_MP_STATE_STOPPED the vcpu is stopped [s390,arm64,riscv] 1534 KVM_MP_STATE_CHECK_STOP the vcpu is in a special error state [s390] 1535 KVM_MP_STATE_OPERATING the vcpu is operating (running or halted) 1536 [s390] 1537 KVM_MP_STATE_LOAD the vcpu is in a special load/startup state 1538 [s390] 1539 KVM_MP_STATE_SUSPENDED the vcpu is in a suspend state and is waiting 1540 for a wakeup event [arm64] 1541 ========================== =============================================== 1542 1543On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an 1544in-kernel irqchip, the multiprocessing state must be maintained by userspace on 1545these architectures. 1546 1547For arm64: 1548^^^^^^^^^^ 1549 1550If a vCPU is in the KVM_MP_STATE_SUSPENDED state, KVM will emulate the 1551architectural execution of a WFI instruction. 1552 1553If a wakeup event is recognized, KVM will exit to userspace with a 1554KVM_SYSTEM_EVENT exit, where the event type is KVM_SYSTEM_EVENT_WAKEUP. If 1555userspace wants to honor the wakeup, it must set the vCPU's MP state to 1556KVM_MP_STATE_RUNNABLE. If it does not, KVM will continue to await a wakeup 1557event in subsequent calls to KVM_RUN. 1558 1559.. warning:: 1560 1561 If userspace intends to keep the vCPU in a SUSPENDED state, it is 1562 strongly recommended that userspace take action to suppress the 1563 wakeup event (such as masking an interrupt). Otherwise, subsequent 1564 calls to KVM_RUN will immediately exit with a KVM_SYSTEM_EVENT_WAKEUP 1565 event and inadvertently waste CPU cycles. 1566 1567 Additionally, if userspace takes action to suppress a wakeup event, 1568 it is strongly recommended that it also restores the vCPU to its 1569 original state when the vCPU is made RUNNABLE again. For example, 1570 if userspace masked a pending interrupt to suppress the wakeup, 1571 the interrupt should be unmasked before returning control to the 1572 guest. 1573 1574For riscv: 1575^^^^^^^^^^ 1576 1577The only states that are valid are KVM_MP_STATE_STOPPED and 1578KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not. 1579 1580On LoongArch, only the KVM_MP_STATE_RUNNABLE state is used to reflect 1581whether the vcpu is runnable. 1582 15834.39 KVM_SET_MP_STATE 1584--------------------- 1585 1586:Capability: KVM_CAP_MP_STATE 1587:Architectures: x86, s390, arm64, riscv, loongarch 1588:Type: vcpu ioctl 1589:Parameters: struct kvm_mp_state (in) 1590:Returns: 0 on success; -1 on error 1591 1592Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for 1593arguments. 1594 1595On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an 1596in-kernel irqchip, the multiprocessing state must be maintained by userspace on 1597these architectures. 1598 1599For arm64/riscv: 1600^^^^^^^^^^^^^^^^ 1601 1602The only states that are valid are KVM_MP_STATE_STOPPED and 1603KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not. 1604 1605On LoongArch, only the KVM_MP_STATE_RUNNABLE state is used to reflect 1606whether the vcpu is runnable. 1607 16084.40 KVM_SET_IDENTITY_MAP_ADDR 1609------------------------------ 1610 1611:Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR 1612:Architectures: x86 1613:Type: vm ioctl 1614:Parameters: unsigned long identity (in) 1615:Returns: 0 on success, -1 on error 1616 1617This ioctl defines the physical address of a one-page region in the guest 1618physical address space. The region must be within the first 4GB of the 1619guest physical address space and must not conflict with any memory slot 1620or any mmio address. The guest may malfunction if it accesses this memory 1621region. 1622 1623Setting the address to 0 will result in resetting the address to its default 1624(0xfffbc000). 1625 1626This ioctl is required on Intel-based hosts. This is needed on Intel hardware 1627because of a quirk in the virtualization implementation (see the internals 1628documentation when it pops into existence). 1629 1630Fails if any VCPU has already been created. 1631 16324.41 KVM_SET_BOOT_CPU_ID 1633------------------------ 1634 1635:Capability: KVM_CAP_SET_BOOT_CPU_ID 1636:Architectures: x86 1637:Type: vm ioctl 1638:Parameters: unsigned long vcpu_id 1639:Returns: 0 on success, -1 on error 1640 1641Define which vcpu is the Bootstrap Processor (BSP). Values are the same 1642as the vcpu id in KVM_CREATE_VCPU. If this ioctl is not called, the default 1643is vcpu 0. This ioctl has to be called before vcpu creation, 1644otherwise it will return EBUSY error. 1645 1646 16474.42 KVM_GET_XSAVE 1648------------------ 1649 1650:Capability: KVM_CAP_XSAVE 1651:Architectures: x86 1652:Type: vcpu ioctl 1653:Parameters: struct kvm_xsave (out) 1654:Returns: 0 on success, -1 on error 1655 1656 1657:: 1658 1659 struct kvm_xsave { 1660 __u32 region[1024]; 1661 __u32 extra[0]; 1662 }; 1663 1664This ioctl would copy current vcpu's xsave struct to the userspace. 1665 1666 16674.43 KVM_SET_XSAVE 1668------------------ 1669 1670:Capability: KVM_CAP_XSAVE and KVM_CAP_XSAVE2 1671:Architectures: x86 1672:Type: vcpu ioctl 1673:Parameters: struct kvm_xsave (in) 1674:Returns: 0 on success, -1 on error 1675 1676:: 1677 1678 1679 struct kvm_xsave { 1680 __u32 region[1024]; 1681 __u32 extra[0]; 1682 }; 1683 1684This ioctl would copy userspace's xsave struct to the kernel. It copies 1685as many bytes as are returned by KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2), 1686when invoked on the vm file descriptor. The size value returned by 1687KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) will always be at least 4096. 1688Currently, it is only greater than 4096 if a dynamic feature has been 1689enabled with ``arch_prctl()``, but this may change in the future. 1690 1691The offsets of the state save areas in struct kvm_xsave follow the 1692contents of CPUID leaf 0xD on the host. 1693 1694 16954.44 KVM_GET_XCRS 1696----------------- 1697 1698:Capability: KVM_CAP_XCRS 1699:Architectures: x86 1700:Type: vcpu ioctl 1701:Parameters: struct kvm_xcrs (out) 1702:Returns: 0 on success, -1 on error 1703 1704:: 1705 1706 struct kvm_xcr { 1707 __u32 xcr; 1708 __u32 reserved; 1709 __u64 value; 1710 }; 1711 1712 struct kvm_xcrs { 1713 __u32 nr_xcrs; 1714 __u32 flags; 1715 struct kvm_xcr xcrs[KVM_MAX_XCRS]; 1716 __u64 padding[16]; 1717 }; 1718 1719This ioctl would copy current vcpu's xcrs to the userspace. 1720 1721 17224.45 KVM_SET_XCRS 1723----------------- 1724 1725:Capability: KVM_CAP_XCRS 1726:Architectures: x86 1727:Type: vcpu ioctl 1728:Parameters: struct kvm_xcrs (in) 1729:Returns: 0 on success, -1 on error 1730 1731:: 1732 1733 struct kvm_xcr { 1734 __u32 xcr; 1735 __u32 reserved; 1736 __u64 value; 1737 }; 1738 1739 struct kvm_xcrs { 1740 __u32 nr_xcrs; 1741 __u32 flags; 1742 struct kvm_xcr xcrs[KVM_MAX_XCRS]; 1743 __u64 padding[16]; 1744 }; 1745 1746This ioctl would set vcpu's xcr to the value userspace specified. 1747 1748 17494.46 KVM_GET_SUPPORTED_CPUID 1750---------------------------- 1751 1752:Capability: KVM_CAP_EXT_CPUID 1753:Architectures: x86 1754:Type: system ioctl 1755:Parameters: struct kvm_cpuid2 (in/out) 1756:Returns: 0 on success, -1 on error 1757 1758:: 1759 1760 struct kvm_cpuid2 { 1761 __u32 nent; 1762 __u32 padding; 1763 struct kvm_cpuid_entry2 entries[0]; 1764 }; 1765 1766 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0) 1767 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */ 1768 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */ 1769 1770 struct kvm_cpuid_entry2 { 1771 __u32 function; 1772 __u32 index; 1773 __u32 flags; 1774 __u32 eax; 1775 __u32 ebx; 1776 __u32 ecx; 1777 __u32 edx; 1778 __u32 padding[3]; 1779 }; 1780 1781This ioctl returns x86 cpuid features which are supported by both the 1782hardware and kvm in its default configuration. Userspace can use the 1783information returned by this ioctl to construct cpuid information (for 1784KVM_SET_CPUID2) that is consistent with hardware, kernel, and 1785userspace capabilities, and with user requirements (for example, the 1786user may wish to constrain cpuid to emulate older hardware, or for 1787feature consistency across a cluster). 1788 1789Dynamically-enabled feature bits need to be requested with 1790``arch_prctl()`` before calling this ioctl. Feature bits that have not 1791been requested are excluded from the result. 1792 1793Note that certain capabilities, such as KVM_CAP_X86_DISABLE_EXITS, may 1794expose cpuid features (e.g. MONITOR) which are not supported by kvm in 1795its default configuration. If userspace enables such capabilities, it 1796is responsible for modifying the results of this ioctl appropriately. 1797 1798Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure 1799with the 'nent' field indicating the number of entries in the variable-size 1800array 'entries'. If the number of entries is too low to describe the cpu 1801capabilities, an error (E2BIG) is returned. If the number is too high, 1802the 'nent' field is adjusted and an error (ENOMEM) is returned. If the 1803number is just right, the 'nent' field is adjusted to the number of valid 1804entries in the 'entries' array, which is then filled. 1805 1806The entries returned are the host cpuid as returned by the cpuid instruction, 1807with unknown or unsupported features masked out. Some features (for example, 1808x2apic), may not be present in the host cpu, but are exposed by kvm if it can 1809emulate them efficiently. The fields in each entry are defined as follows: 1810 1811 function: 1812 the eax value used to obtain the entry 1813 1814 index: 1815 the ecx value used to obtain the entry (for entries that are 1816 affected by ecx) 1817 1818 flags: 1819 an OR of zero or more of the following: 1820 1821 KVM_CPUID_FLAG_SIGNIFCANT_INDEX: 1822 if the index field is valid 1823 1824 eax, ebx, ecx, edx: 1825 the values returned by the cpuid instruction for 1826 this function/index combination 1827 1828x2APIC (CPUID leaf 1, ecx[21) and TSC deadline timer (CPUID leaf 1, ecx[24]) 1829may be returned as true, but they depend on KVM_CREATE_IRQCHIP for in-kernel 1830emulation of the local APIC. TSC deadline timer support is also reported via:: 1831 1832 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER) 1833 1834if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the 1835feature in userspace, then you can enable the feature for KVM_SET_CPUID2. 1836 1837Enabling x2APIC in KVM_SET_CPUID2 requires KVM_CREATE_IRQCHIP as KVM doesn't 1838support forwarding x2APIC MSR accesses to userspace, i.e. KVM does not support 1839emulating x2APIC in userspace. 1840 18414.47 KVM_PPC_GET_PVINFO 1842----------------------- 1843 1844:Capability: KVM_CAP_PPC_GET_PVINFO 1845:Architectures: ppc 1846:Type: vm ioctl 1847:Parameters: struct kvm_ppc_pvinfo (out) 1848:Returns: 0 on success, !0 on error 1849 1850:: 1851 1852 struct kvm_ppc_pvinfo { 1853 __u32 flags; 1854 __u32 hcall[4]; 1855 __u8 pad[108]; 1856 }; 1857 1858This ioctl fetches PV specific information that need to be passed to the guest 1859using the device tree or other means from vm context. 1860 1861The hcall array defines 4 instructions that make up a hypercall. 1862 1863If any additional field gets added to this structure later on, a bit for that 1864additional piece of information will be set in the flags bitmap. 1865 1866The flags bitmap is defined as:: 1867 1868 /* the host supports the ePAPR idle hcall 1869 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1<<0) 1870 18714.52 KVM_SET_GSI_ROUTING 1872------------------------ 1873 1874:Capability: KVM_CAP_IRQ_ROUTING 1875:Architectures: x86 s390 arm64 1876:Type: vm ioctl 1877:Parameters: struct kvm_irq_routing (in) 1878:Returns: 0 on success, -1 on error 1879 1880Sets the GSI routing table entries, overwriting any previously set entries. 1881 1882On arm64, GSI routing has the following limitation: 1883 1884- GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD. 1885 1886:: 1887 1888 struct kvm_irq_routing { 1889 __u32 nr; 1890 __u32 flags; 1891 struct kvm_irq_routing_entry entries[0]; 1892 }; 1893 1894No flags are specified so far, the corresponding field must be set to zero. 1895 1896:: 1897 1898 struct kvm_irq_routing_entry { 1899 __u32 gsi; 1900 __u32 type; 1901 __u32 flags; 1902 __u32 pad; 1903 union { 1904 struct kvm_irq_routing_irqchip irqchip; 1905 struct kvm_irq_routing_msi msi; 1906 struct kvm_irq_routing_s390_adapter adapter; 1907 struct kvm_irq_routing_hv_sint hv_sint; 1908 struct kvm_irq_routing_xen_evtchn xen_evtchn; 1909 __u32 pad[8]; 1910 } u; 1911 }; 1912 1913 /* gsi routing entry types */ 1914 #define KVM_IRQ_ROUTING_IRQCHIP 1 1915 #define KVM_IRQ_ROUTING_MSI 2 1916 #define KVM_IRQ_ROUTING_S390_ADAPTER 3 1917 #define KVM_IRQ_ROUTING_HV_SINT 4 1918 #define KVM_IRQ_ROUTING_XEN_EVTCHN 5 1919 1920On s390, adding a KVM_IRQ_ROUTING_S390_ADAPTER is rejected on ucontrol VMs with 1921error -EINVAL. 1922 1923flags: 1924 1925- KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry 1926 type, specifies that the devid field contains a valid value. The per-VM 1927 KVM_CAP_MSI_DEVID capability advertises the requirement to provide 1928 the device ID. If this capability is not available, userspace should 1929 never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail. 1930- zero otherwise 1931 1932:: 1933 1934 struct kvm_irq_routing_irqchip { 1935 __u32 irqchip; 1936 __u32 pin; 1937 }; 1938 1939 struct kvm_irq_routing_msi { 1940 __u32 address_lo; 1941 __u32 address_hi; 1942 __u32 data; 1943 union { 1944 __u32 pad; 1945 __u32 devid; 1946 }; 1947 }; 1948 1949If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier 1950for the device that wrote the MSI message. For PCI, this is usually a 1951BDF identifier in the lower 16 bits. 1952 1953On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS 1954feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled, 1955address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of 1956address_hi must be zero. 1957 1958:: 1959 1960 struct kvm_irq_routing_s390_adapter { 1961 __u64 ind_addr; 1962 __u64 summary_addr; 1963 __u64 ind_offset; 1964 __u32 summary_offset; 1965 __u32 adapter_id; 1966 }; 1967 1968 struct kvm_irq_routing_hv_sint { 1969 __u32 vcpu; 1970 __u32 sint; 1971 }; 1972 1973 struct kvm_irq_routing_xen_evtchn { 1974 __u32 port; 1975 __u32 vcpu; 1976 __u32 priority; 1977 }; 1978 1979 1980When KVM_CAP_XEN_HVM includes the KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL bit 1981in its indication of supported features, routing to Xen event channels 1982is supported. Although the priority field is present, only the value 1983KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL is supported, which means delivery by 19842 level event channels. FIFO event channel support may be added in 1985the future. 1986 1987 19884.55 KVM_SET_TSC_KHZ 1989-------------------- 1990 1991:Capability: KVM_CAP_TSC_CONTROL / KVM_CAP_VM_TSC_CONTROL 1992:Architectures: x86 1993:Type: vcpu ioctl / vm ioctl 1994:Parameters: virtual tsc_khz 1995:Returns: 0 on success, -1 on error 1996 1997Specifies the tsc frequency for the virtual machine. The unit of the 1998frequency is KHz. 1999 2000If the KVM_CAP_VM_TSC_CONTROL capability is advertised, this can also 2001be used as a vm ioctl to set the initial tsc frequency of subsequently 2002created vCPUs. 2003 20044.56 KVM_GET_TSC_KHZ 2005-------------------- 2006 2007:Capability: KVM_CAP_GET_TSC_KHZ / KVM_CAP_VM_TSC_CONTROL 2008:Architectures: x86 2009:Type: vcpu ioctl / vm ioctl 2010:Parameters: none 2011:Returns: virtual tsc-khz on success, negative value on error 2012 2013Returns the tsc frequency of the guest. The unit of the return value is 2014KHz. If the host has unstable tsc this ioctl returns -EIO instead as an 2015error. 2016 2017 20184.57 KVM_GET_LAPIC 2019------------------ 2020 2021:Capability: KVM_CAP_IRQCHIP 2022:Architectures: x86 2023:Type: vcpu ioctl 2024:Parameters: struct kvm_lapic_state (out) 2025:Returns: 0 on success, -1 on error 2026 2027:: 2028 2029 #define KVM_APIC_REG_SIZE 0x400 2030 struct kvm_lapic_state { 2031 char regs[KVM_APIC_REG_SIZE]; 2032 }; 2033 2034Reads the Local APIC registers and copies them into the input argument. The 2035data format and layout are the same as documented in the architecture manual. 2036 2037If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is 2038enabled, then the format of APIC_ID register depends on the APIC mode 2039(reported by MSR_IA32_APICBASE) of its VCPU. x2APIC stores APIC ID in 2040the APIC_ID register (bytes 32-35). xAPIC only allows an 8-bit APIC ID 2041which is stored in bits 31-24 of the APIC register, or equivalently in 2042byte 35 of struct kvm_lapic_state's regs field. KVM_GET_LAPIC must then 2043be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR. 2044 2045If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state 2046always uses xAPIC format. 2047 2048 20494.58 KVM_SET_LAPIC 2050------------------ 2051 2052:Capability: KVM_CAP_IRQCHIP 2053:Architectures: x86 2054:Type: vcpu ioctl 2055:Parameters: struct kvm_lapic_state (in) 2056:Returns: 0 on success, -1 on error 2057 2058:: 2059 2060 #define KVM_APIC_REG_SIZE 0x400 2061 struct kvm_lapic_state { 2062 char regs[KVM_APIC_REG_SIZE]; 2063 }; 2064 2065Copies the input argument into the Local APIC registers. The data format 2066and layout are the same as documented in the architecture manual. 2067 2068The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's 2069regs field) depends on the state of the KVM_CAP_X2APIC_API capability. 2070See the note in KVM_GET_LAPIC. 2071 2072 20734.59 KVM_IOEVENTFD 2074------------------ 2075 2076:Capability: KVM_CAP_IOEVENTFD 2077:Architectures: all 2078:Type: vm ioctl 2079:Parameters: struct kvm_ioeventfd (in) 2080:Returns: 0 on success, !0 on error 2081 2082This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address 2083within the guest. A guest write in the registered address will signal the 2084provided event instead of triggering an exit. 2085 2086:: 2087 2088 struct kvm_ioeventfd { 2089 __u64 datamatch; 2090 __u64 addr; /* legal pio/mmio address */ 2091 __u32 len; /* 0, 1, 2, 4, or 8 bytes */ 2092 __s32 fd; 2093 __u32 flags; 2094 __u8 pad[36]; 2095 }; 2096 2097For the special case of virtio-ccw devices on s390, the ioevent is matched 2098to a subchannel/virtqueue tuple instead. 2099 2100The following flags are defined:: 2101 2102 #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch) 2103 #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio) 2104 #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign) 2105 #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \ 2106 (1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify) 2107 2108If datamatch flag is set, the event will be signaled only if the written value 2109to the registered address is equal to datamatch in struct kvm_ioeventfd. 2110 2111For virtio-ccw devices, addr contains the subchannel id and datamatch the 2112virtqueue index. 2113 2114With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and 2115the kernel will ignore the length of guest write and may get a faster vmexit. 2116The speedup may only apply to specific architectures, but the ioeventfd will 2117work anyway. 2118 21194.60 KVM_DIRTY_TLB 2120------------------ 2121 2122:Capability: KVM_CAP_SW_TLB 2123:Architectures: ppc 2124:Type: vcpu ioctl 2125:Parameters: struct kvm_dirty_tlb (in) 2126:Returns: 0 on success, -1 on error 2127 2128:: 2129 2130 struct kvm_dirty_tlb { 2131 __u64 bitmap; 2132 __u32 num_dirty; 2133 }; 2134 2135This must be called whenever userspace has changed an entry in the shared 2136TLB, prior to calling KVM_RUN on the associated vcpu. 2137 2138The "bitmap" field is the userspace address of an array. This array 2139consists of a number of bits, equal to the total number of TLB entries as 2140determined by the last successful call to ``KVM_ENABLE_CAP(KVM_CAP_SW_TLB)``, 2141rounded up to the nearest multiple of 64. 2142 2143Each bit corresponds to one TLB entry, ordered the same as in the shared TLB 2144array. 2145 2146The array is little-endian: the bit 0 is the least significant bit of the 2147first byte, bit 8 is the least significant bit of the second byte, etc. 2148This avoids any complications with differing word sizes. 2149 2150The "num_dirty" field is a performance hint for KVM to determine whether it 2151should skip processing the bitmap and just invalidate everything. It must 2152be set to the number of set bits in the bitmap. 2153 2154 21554.62 KVM_CREATE_SPAPR_TCE 2156------------------------- 2157 2158:Capability: KVM_CAP_SPAPR_TCE 2159:Architectures: powerpc 2160:Type: vm ioctl 2161:Parameters: struct kvm_create_spapr_tce (in) 2162:Returns: file descriptor for manipulating the created TCE table 2163 2164This creates a virtual TCE (translation control entry) table, which 2165is an IOMMU for PAPR-style virtual I/O. It is used to translate 2166logical addresses used in virtual I/O into guest physical addresses, 2167and provides a scatter/gather capability for PAPR virtual I/O. 2168 2169:: 2170 2171 /* for KVM_CAP_SPAPR_TCE */ 2172 struct kvm_create_spapr_tce { 2173 __u64 liobn; 2174 __u32 window_size; 2175 }; 2176 2177The liobn field gives the logical IO bus number for which to create a 2178TCE table. The window_size field specifies the size of the DMA window 2179which this TCE table will translate - the table will contain one 64 2180bit TCE entry for every 4kiB of the DMA window. 2181 2182When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE 2183table has been created using this ioctl(), the kernel will handle it 2184in real mode, updating the TCE table. H_PUT_TCE calls for other 2185liobns will cause a vm exit and must be handled by userspace. 2186 2187The return value is a file descriptor which can be passed to mmap(2) 2188to map the created TCE table into userspace. This lets userspace read 2189the entries written by kernel-handled H_PUT_TCE calls, and also lets 2190userspace update the TCE table directly which is useful in some 2191circumstances. 2192 2193 21944.64 KVM_NMI 2195------------ 2196 2197:Capability: KVM_CAP_USER_NMI 2198:Architectures: x86 2199:Type: vcpu ioctl 2200:Parameters: none 2201:Returns: 0 on success, -1 on error 2202 2203Queues an NMI on the thread's vcpu. Note this is well defined only 2204when KVM_CREATE_IRQCHIP has not been called, since this is an interface 2205between the virtual cpu core and virtual local APIC. After KVM_CREATE_IRQCHIP 2206has been called, this interface is completely emulated within the kernel. 2207 2208To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the 2209following algorithm: 2210 2211 - pause the vcpu 2212 - read the local APIC's state (KVM_GET_LAPIC) 2213 - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1) 2214 - if so, issue KVM_NMI 2215 - resume the vcpu 2216 2217Some guests configure the LINT1 NMI input to cause a panic, aiding in 2218debugging. 2219 2220 22214.65 KVM_S390_UCAS_MAP 2222---------------------- 2223 2224:Capability: KVM_CAP_S390_UCONTROL 2225:Architectures: s390 2226:Type: vcpu ioctl 2227:Parameters: struct kvm_s390_ucas_mapping (in) 2228:Returns: 0 in case of success 2229 2230The parameter is defined like this:: 2231 2232 struct kvm_s390_ucas_mapping { 2233 __u64 user_addr; 2234 __u64 vcpu_addr; 2235 __u64 length; 2236 }; 2237 2238This ioctl maps the memory at "user_addr" with the length "length" to 2239the vcpu's address space starting at "vcpu_addr". All parameters need to 2240be aligned by 1 megabyte. 2241 2242 22434.66 KVM_S390_UCAS_UNMAP 2244------------------------ 2245 2246:Capability: KVM_CAP_S390_UCONTROL 2247:Architectures: s390 2248:Type: vcpu ioctl 2249:Parameters: struct kvm_s390_ucas_mapping (in) 2250:Returns: 0 in case of success 2251 2252The parameter is defined like this:: 2253 2254 struct kvm_s390_ucas_mapping { 2255 __u64 user_addr; 2256 __u64 vcpu_addr; 2257 __u64 length; 2258 }; 2259 2260This ioctl unmaps the memory in the vcpu's address space starting at 2261"vcpu_addr" with the length "length". The field "user_addr" is ignored. 2262All parameters need to be aligned by 1 megabyte. 2263 2264 22654.67 KVM_S390_VCPU_FAULT 2266------------------------ 2267 2268:Capability: KVM_CAP_S390_UCONTROL 2269:Architectures: s390 2270:Type: vcpu ioctl 2271:Parameters: vcpu absolute address (in) 2272:Returns: 0 in case of success 2273 2274This call creates a page table entry on the virtual cpu's address space 2275(for user controlled virtual machines) or the virtual machine's address 2276space (for regular virtual machines). This only works for minor faults, 2277thus it's recommended to access subject memory page via the user page 2278table upfront. This is useful to handle validity intercepts for user 2279controlled virtual machines to fault in the virtual cpu's lowcore pages 2280prior to calling the KVM_RUN ioctl. 2281 2282 22834.68 KVM_SET_ONE_REG 2284-------------------- 2285 2286:Capability: KVM_CAP_ONE_REG 2287:Architectures: all 2288:Type: vcpu ioctl 2289:Parameters: struct kvm_one_reg (in) 2290:Returns: 0 on success, negative value on failure 2291 2292Errors: 2293 2294 ====== ============================================================ 2295 ENOENT no such register 2296 EINVAL invalid register ID, or no such register or used with VMs in 2297 protected virtualization mode on s390 2298 EPERM (arm64) register access not allowed before vcpu finalization 2299 EBUSY (riscv) changing register value not allowed after the vcpu 2300 has run at least once 2301 ====== ============================================================ 2302 2303(These error codes are indicative only: do not rely on a specific error 2304code being returned in a specific situation.) 2305 2306:: 2307 2308 struct kvm_one_reg { 2309 __u64 id; 2310 __u64 addr; 2311 }; 2312 2313Using this ioctl, a single vcpu register can be set to a specific value 2314defined by user space with the passed in struct kvm_one_reg, where id 2315refers to the register identifier as described below and addr is a pointer 2316to a variable with the respective size. There can be architecture agnostic 2317and architecture specific registers. Each have their own range of operation 2318and their own constants and width. To keep track of the implemented 2319registers, find a list below: 2320 2321 ======= =============================== ============ 2322 Arch Register Width (bits) 2323 ======= =============================== ============ 2324 PPC KVM_REG_PPC_HIOR 64 2325 PPC KVM_REG_PPC_IAC1 64 2326 PPC KVM_REG_PPC_IAC2 64 2327 PPC KVM_REG_PPC_IAC3 64 2328 PPC KVM_REG_PPC_IAC4 64 2329 PPC KVM_REG_PPC_DAC1 64 2330 PPC KVM_REG_PPC_DAC2 64 2331 PPC KVM_REG_PPC_DABR 64 2332 PPC KVM_REG_PPC_DSCR 64 2333 PPC KVM_REG_PPC_PURR 64 2334 PPC KVM_REG_PPC_SPURR 64 2335 PPC KVM_REG_PPC_DAR 64 2336 PPC KVM_REG_PPC_DSISR 32 2337 PPC KVM_REG_PPC_AMR 64 2338 PPC KVM_REG_PPC_UAMOR 64 2339 PPC KVM_REG_PPC_MMCR0 64 2340 PPC KVM_REG_PPC_MMCR1 64 2341 PPC KVM_REG_PPC_MMCRA 64 2342 PPC KVM_REG_PPC_MMCR2 64 2343 PPC KVM_REG_PPC_MMCRS 64 2344 PPC KVM_REG_PPC_MMCR3 64 2345 PPC KVM_REG_PPC_SIAR 64 2346 PPC KVM_REG_PPC_SDAR 64 2347 PPC KVM_REG_PPC_SIER 64 2348 PPC KVM_REG_PPC_SIER2 64 2349 PPC KVM_REG_PPC_SIER3 64 2350 PPC KVM_REG_PPC_PMC1 32 2351 PPC KVM_REG_PPC_PMC2 32 2352 PPC KVM_REG_PPC_PMC3 32 2353 PPC KVM_REG_PPC_PMC4 32 2354 PPC KVM_REG_PPC_PMC5 32 2355 PPC KVM_REG_PPC_PMC6 32 2356 PPC KVM_REG_PPC_PMC7 32 2357 PPC KVM_REG_PPC_PMC8 32 2358 PPC KVM_REG_PPC_FPR0 64 2359 ... 2360 PPC KVM_REG_PPC_FPR31 64 2361 PPC KVM_REG_PPC_VR0 128 2362 ... 2363 PPC KVM_REG_PPC_VR31 128 2364 PPC KVM_REG_PPC_VSR0 128 2365 ... 2366 PPC KVM_REG_PPC_VSR31 128 2367 PPC KVM_REG_PPC_FPSCR 64 2368 PPC KVM_REG_PPC_VSCR 32 2369 PPC KVM_REG_PPC_VPA_ADDR 64 2370 PPC KVM_REG_PPC_VPA_SLB 128 2371 PPC KVM_REG_PPC_VPA_DTL 128 2372 PPC KVM_REG_PPC_EPCR 32 2373 PPC KVM_REG_PPC_EPR 32 2374 PPC KVM_REG_PPC_TCR 32 2375 PPC KVM_REG_PPC_TSR 32 2376 PPC KVM_REG_PPC_OR_TSR 32 2377 PPC KVM_REG_PPC_CLEAR_TSR 32 2378 PPC KVM_REG_PPC_MAS0 32 2379 PPC KVM_REG_PPC_MAS1 32 2380 PPC KVM_REG_PPC_MAS2 64 2381 PPC KVM_REG_PPC_MAS7_3 64 2382 PPC KVM_REG_PPC_MAS4 32 2383 PPC KVM_REG_PPC_MAS6 32 2384 PPC KVM_REG_PPC_MMUCFG 32 2385 PPC KVM_REG_PPC_TLB0CFG 32 2386 PPC KVM_REG_PPC_TLB1CFG 32 2387 PPC KVM_REG_PPC_TLB2CFG 32 2388 PPC KVM_REG_PPC_TLB3CFG 32 2389 PPC KVM_REG_PPC_TLB0PS 32 2390 PPC KVM_REG_PPC_TLB1PS 32 2391 PPC KVM_REG_PPC_TLB2PS 32 2392 PPC KVM_REG_PPC_TLB3PS 32 2393 PPC KVM_REG_PPC_EPTCFG 32 2394 PPC KVM_REG_PPC_ICP_STATE 64 2395 PPC KVM_REG_PPC_VP_STATE 128 2396 PPC KVM_REG_PPC_TB_OFFSET 64 2397 PPC KVM_REG_PPC_SPMC1 32 2398 PPC KVM_REG_PPC_SPMC2 32 2399 PPC KVM_REG_PPC_IAMR 64 2400 PPC KVM_REG_PPC_TFHAR 64 2401 PPC KVM_REG_PPC_TFIAR 64 2402 PPC KVM_REG_PPC_TEXASR 64 2403 PPC KVM_REG_PPC_FSCR 64 2404 PPC KVM_REG_PPC_PSPB 32 2405 PPC KVM_REG_PPC_EBBHR 64 2406 PPC KVM_REG_PPC_EBBRR 64 2407 PPC KVM_REG_PPC_BESCR 64 2408 PPC KVM_REG_PPC_TAR 64 2409 PPC KVM_REG_PPC_DPDES 64 2410 PPC KVM_REG_PPC_DAWR 64 2411 PPC KVM_REG_PPC_DAWRX 64 2412 PPC KVM_REG_PPC_CIABR 64 2413 PPC KVM_REG_PPC_IC 64 2414 PPC KVM_REG_PPC_VTB 64 2415 PPC KVM_REG_PPC_CSIGR 64 2416 PPC KVM_REG_PPC_TACR 64 2417 PPC KVM_REG_PPC_TCSCR 64 2418 PPC KVM_REG_PPC_PID 64 2419 PPC KVM_REG_PPC_ACOP 64 2420 PPC KVM_REG_PPC_VRSAVE 32 2421 PPC KVM_REG_PPC_LPCR 32 2422 PPC KVM_REG_PPC_LPCR_64 64 2423 PPC KVM_REG_PPC_PPR 64 2424 PPC KVM_REG_PPC_ARCH_COMPAT 32 2425 PPC KVM_REG_PPC_DABRX 32 2426 PPC KVM_REG_PPC_WORT 64 2427 PPC KVM_REG_PPC_SPRG9 64 2428 PPC KVM_REG_PPC_DBSR 32 2429 PPC KVM_REG_PPC_TIDR 64 2430 PPC KVM_REG_PPC_PSSCR 64 2431 PPC KVM_REG_PPC_DEC_EXPIRY 64 2432 PPC KVM_REG_PPC_PTCR 64 2433 PPC KVM_REG_PPC_HASHKEYR 64 2434 PPC KVM_REG_PPC_HASHPKEYR 64 2435 PPC KVM_REG_PPC_DAWR1 64 2436 PPC KVM_REG_PPC_DAWRX1 64 2437 PPC KVM_REG_PPC_DEXCR 64 2438 PPC KVM_REG_PPC_TM_GPR0 64 2439 ... 2440 PPC KVM_REG_PPC_TM_GPR31 64 2441 PPC KVM_REG_PPC_TM_VSR0 128 2442 ... 2443 PPC KVM_REG_PPC_TM_VSR63 128 2444 PPC KVM_REG_PPC_TM_CR 64 2445 PPC KVM_REG_PPC_TM_LR 64 2446 PPC KVM_REG_PPC_TM_CTR 64 2447 PPC KVM_REG_PPC_TM_FPSCR 64 2448 PPC KVM_REG_PPC_TM_AMR 64 2449 PPC KVM_REG_PPC_TM_PPR 64 2450 PPC KVM_REG_PPC_TM_VRSAVE 64 2451 PPC KVM_REG_PPC_TM_VSCR 32 2452 PPC KVM_REG_PPC_TM_DSCR 64 2453 PPC KVM_REG_PPC_TM_TAR 64 2454 PPC KVM_REG_PPC_TM_XER 64 2455 2456 MIPS KVM_REG_MIPS_R0 64 2457 ... 2458 MIPS KVM_REG_MIPS_R31 64 2459 MIPS KVM_REG_MIPS_HI 64 2460 MIPS KVM_REG_MIPS_LO 64 2461 MIPS KVM_REG_MIPS_PC 64 2462 MIPS KVM_REG_MIPS_CP0_INDEX 32 2463 MIPS KVM_REG_MIPS_CP0_ENTRYLO0 64 2464 MIPS KVM_REG_MIPS_CP0_ENTRYLO1 64 2465 MIPS KVM_REG_MIPS_CP0_CONTEXT 64 2466 MIPS KVM_REG_MIPS_CP0_CONTEXTCONFIG 32 2467 MIPS KVM_REG_MIPS_CP0_USERLOCAL 64 2468 MIPS KVM_REG_MIPS_CP0_XCONTEXTCONFIG 64 2469 MIPS KVM_REG_MIPS_CP0_PAGEMASK 32 2470 MIPS KVM_REG_MIPS_CP0_PAGEGRAIN 32 2471 MIPS KVM_REG_MIPS_CP0_SEGCTL0 64 2472 MIPS KVM_REG_MIPS_CP0_SEGCTL1 64 2473 MIPS KVM_REG_MIPS_CP0_SEGCTL2 64 2474 MIPS KVM_REG_MIPS_CP0_PWBASE 64 2475 MIPS KVM_REG_MIPS_CP0_PWFIELD 64 2476 MIPS KVM_REG_MIPS_CP0_PWSIZE 64 2477 MIPS KVM_REG_MIPS_CP0_WIRED 32 2478 MIPS KVM_REG_MIPS_CP0_PWCTL 32 2479 MIPS KVM_REG_MIPS_CP0_HWRENA 32 2480 MIPS KVM_REG_MIPS_CP0_BADVADDR 64 2481 MIPS KVM_REG_MIPS_CP0_BADINSTR 32 2482 MIPS KVM_REG_MIPS_CP0_BADINSTRP 32 2483 MIPS KVM_REG_MIPS_CP0_COUNT 32 2484 MIPS KVM_REG_MIPS_CP0_ENTRYHI 64 2485 MIPS KVM_REG_MIPS_CP0_COMPARE 32 2486 MIPS KVM_REG_MIPS_CP0_STATUS 32 2487 MIPS KVM_REG_MIPS_CP0_INTCTL 32 2488 MIPS KVM_REG_MIPS_CP0_CAUSE 32 2489 MIPS KVM_REG_MIPS_CP0_EPC 64 2490 MIPS KVM_REG_MIPS_CP0_PRID 32 2491 MIPS KVM_REG_MIPS_CP0_EBASE 64 2492 MIPS KVM_REG_MIPS_CP0_CONFIG 32 2493 MIPS KVM_REG_MIPS_CP0_CONFIG1 32 2494 MIPS KVM_REG_MIPS_CP0_CONFIG2 32 2495 MIPS KVM_REG_MIPS_CP0_CONFIG3 32 2496 MIPS KVM_REG_MIPS_CP0_CONFIG4 32 2497 MIPS KVM_REG_MIPS_CP0_CONFIG5 32 2498 MIPS KVM_REG_MIPS_CP0_CONFIG7 32 2499 MIPS KVM_REG_MIPS_CP0_XCONTEXT 64 2500 MIPS KVM_REG_MIPS_CP0_ERROREPC 64 2501 MIPS KVM_REG_MIPS_CP0_KSCRATCH1 64 2502 MIPS KVM_REG_MIPS_CP0_KSCRATCH2 64 2503 MIPS KVM_REG_MIPS_CP0_KSCRATCH3 64 2504 MIPS KVM_REG_MIPS_CP0_KSCRATCH4 64 2505 MIPS KVM_REG_MIPS_CP0_KSCRATCH5 64 2506 MIPS KVM_REG_MIPS_CP0_KSCRATCH6 64 2507 MIPS KVM_REG_MIPS_CP0_MAAR(0..63) 64 2508 MIPS KVM_REG_MIPS_COUNT_CTL 64 2509 MIPS KVM_REG_MIPS_COUNT_RESUME 64 2510 MIPS KVM_REG_MIPS_COUNT_HZ 64 2511 MIPS KVM_REG_MIPS_FPR_32(0..31) 32 2512 MIPS KVM_REG_MIPS_FPR_64(0..31) 64 2513 MIPS KVM_REG_MIPS_VEC_128(0..31) 128 2514 MIPS KVM_REG_MIPS_FCR_IR 32 2515 MIPS KVM_REG_MIPS_FCR_CSR 32 2516 MIPS KVM_REG_MIPS_MSA_IR 32 2517 MIPS KVM_REG_MIPS_MSA_CSR 32 2518 ======= =============================== ============ 2519 2520ARM registers are mapped using the lower 32 bits. The upper 16 of that 2521is the register group type, or coprocessor number: 2522 2523ARM core registers have the following id bit patterns:: 2524 2525 0x4020 0000 0010 <index into the kvm_regs struct:16> 2526 2527ARM 32-bit CP15 registers have the following id bit patterns:: 2528 2529 0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3> 2530 2531ARM 64-bit CP15 registers have the following id bit patterns:: 2532 2533 0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3> 2534 2535ARM CCSIDR registers are demultiplexed by CSSELR value:: 2536 2537 0x4020 0000 0011 00 <csselr:8> 2538 2539ARM 32-bit VFP control registers have the following id bit patterns:: 2540 2541 0x4020 0000 0012 1 <regno:12> 2542 2543ARM 64-bit FP registers have the following id bit patterns:: 2544 2545 0x4030 0000 0012 0 <regno:12> 2546 2547ARM firmware pseudo-registers have the following bit pattern:: 2548 2549 0x4030 0000 0014 <regno:16> 2550 2551 2552arm64 registers are mapped using the lower 32 bits. The upper 16 of 2553that is the register group type, or coprocessor number: 2554 2555arm64 core/FP-SIMD registers have the following id bit patterns. Note 2556that the size of the access is variable, as the kvm_regs structure 2557contains elements ranging from 32 to 128 bits. The index is a 32bit 2558value in the kvm_regs structure seen as a 32bit array:: 2559 2560 0x60x0 0000 0010 <index into the kvm_regs struct:16> 2561 2562Specifically: 2563 2564======================= ========= ===== ======================================= 2565 Encoding Register Bits kvm_regs member 2566======================= ========= ===== ======================================= 2567 0x6030 0000 0010 0000 X0 64 regs.regs[0] 2568 0x6030 0000 0010 0002 X1 64 regs.regs[1] 2569 ... 2570 0x6030 0000 0010 003c X30 64 regs.regs[30] 2571 0x6030 0000 0010 003e SP 64 regs.sp 2572 0x6030 0000 0010 0040 PC 64 regs.pc 2573 0x6030 0000 0010 0042 PSTATE 64 regs.pstate 2574 0x6030 0000 0010 0044 SP_EL1 64 sp_el1 2575 0x6030 0000 0010 0046 ELR_EL1 64 elr_el1 2576 0x6030 0000 0010 0048 SPSR_EL1 64 spsr[KVM_SPSR_EL1] (alias SPSR_SVC) 2577 0x6030 0000 0010 004a SPSR_ABT 64 spsr[KVM_SPSR_ABT] 2578 0x6030 0000 0010 004c SPSR_UND 64 spsr[KVM_SPSR_UND] 2579 0x6030 0000 0010 004e SPSR_IRQ 64 spsr[KVM_SPSR_IRQ] 2580 0x6030 0000 0010 0050 SPSR_FIQ 64 spsr[KVM_SPSR_FIQ] 2581 0x6040 0000 0010 0054 V0 128 fp_regs.vregs[0] [1]_ 2582 0x6040 0000 0010 0058 V1 128 fp_regs.vregs[1] [1]_ 2583 ... 2584 0x6040 0000 0010 00d0 V31 128 fp_regs.vregs[31] [1]_ 2585 0x6020 0000 0010 00d4 FPSR 32 fp_regs.fpsr 2586 0x6020 0000 0010 00d5 FPCR 32 fp_regs.fpcr 2587======================= ========= ===== ======================================= 2588 2589.. [1] These encodings are not accepted for SVE-enabled vcpus. See 2590 :ref:`KVM_ARM_VCPU_INIT`. 2591 2592 The equivalent register content can be accessed via bits [127:0] of 2593 the corresponding SVE Zn registers instead for vcpus that have SVE 2594 enabled (see below). 2595 2596arm64 CCSIDR registers are demultiplexed by CSSELR value:: 2597 2598 0x6020 0000 0011 00 <csselr:8> 2599 2600arm64 system registers have the following id bit patterns:: 2601 2602 0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3> 2603 2604.. warning:: 2605 2606 Two system register IDs do not follow the specified pattern. These 2607 are KVM_REG_ARM_TIMER_CVAL and KVM_REG_ARM_TIMER_CNT, which map to 2608 system registers CNTV_CVAL_EL0 and CNTVCT_EL0 respectively. These 2609 two had their values accidentally swapped, which means TIMER_CVAL is 2610 derived from the register encoding for CNTVCT_EL0 and TIMER_CNT is 2611 derived from the register encoding for CNTV_CVAL_EL0. As this is 2612 API, it must remain this way. 2613 2614arm64 firmware pseudo-registers have the following bit pattern:: 2615 2616 0x6030 0000 0014 <regno:16> 2617 2618arm64 SVE registers have the following bit patterns:: 2619 2620 0x6080 0000 0015 00 <n:5> <slice:5> Zn bits[2048*slice + 2047 : 2048*slice] 2621 0x6050 0000 0015 04 <n:4> <slice:5> Pn bits[256*slice + 255 : 256*slice] 2622 0x6050 0000 0015 060 <slice:5> FFR bits[256*slice + 255 : 256*slice] 2623 0x6060 0000 0015 ffff KVM_REG_ARM64_SVE_VLS pseudo-register 2624 2625Access to register IDs where 2048 * slice >= 128 * max_vq will fail with 2626ENOENT. max_vq is the vcpu's maximum supported vector length in 128-bit 2627quadwords: see [2]_ below. 2628 2629These registers are only accessible on vcpus for which SVE is enabled. 2630See KVM_ARM_VCPU_INIT for details. 2631 2632In addition, except for KVM_REG_ARM64_SVE_VLS, these registers are not 2633accessible until the vcpu's SVE configuration has been finalized 2634using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). See KVM_ARM_VCPU_INIT 2635and KVM_ARM_VCPU_FINALIZE for more information about this procedure. 2636 2637KVM_REG_ARM64_SVE_VLS is a pseudo-register that allows the set of vector 2638lengths supported by the vcpu to be discovered and configured by 2639userspace. When transferred to or from user memory via KVM_GET_ONE_REG 2640or KVM_SET_ONE_REG, the value of this register is of type 2641__u64[KVM_ARM64_SVE_VLS_WORDS], and encodes the set of vector lengths as 2642follows:: 2643 2644 __u64 vector_lengths[KVM_ARM64_SVE_VLS_WORDS]; 2645 2646 if (vq >= SVE_VQ_MIN && vq <= SVE_VQ_MAX && 2647 ((vector_lengths[(vq - KVM_ARM64_SVE_VQ_MIN) / 64] >> 2648 ((vq - KVM_ARM64_SVE_VQ_MIN) % 64)) & 1)) 2649 /* Vector length vq * 16 bytes supported */ 2650 else 2651 /* Vector length vq * 16 bytes not supported */ 2652 2653.. [2] The maximum value vq for which the above condition is true is 2654 max_vq. This is the maximum vector length available to the guest on 2655 this vcpu, and determines which register slices are visible through 2656 this ioctl interface. 2657 2658(See Documentation/arch/arm64/sve.rst for an explanation of the "vq" 2659nomenclature.) 2660 2661KVM_REG_ARM64_SVE_VLS is only accessible after KVM_ARM_VCPU_INIT. 2662KVM_ARM_VCPU_INIT initialises it to the best set of vector lengths that 2663the host supports. 2664 2665Userspace may subsequently modify it if desired until the vcpu's SVE 2666configuration is finalized using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). 2667 2668Apart from simply removing all vector lengths from the host set that 2669exceed some value, support for arbitrarily chosen sets of vector lengths 2670is hardware-dependent and may not be available. Attempting to configure 2671an invalid set of vector lengths via KVM_SET_ONE_REG will fail with 2672EINVAL. 2673 2674After the vcpu's SVE configuration is finalized, further attempts to 2675write this register will fail with EPERM. 2676 2677arm64 bitmap feature firmware pseudo-registers have the following bit pattern:: 2678 2679 0x6030 0000 0016 <regno:16> 2680 2681The bitmap feature firmware registers exposes the hypercall services that 2682are available for userspace to configure. The set bits corresponds to the 2683services that are available for the guests to access. By default, KVM 2684sets all the supported bits during VM initialization. The userspace can 2685discover the available services via KVM_GET_ONE_REG, and write back the 2686bitmap corresponding to the features that it wishes guests to see via 2687KVM_SET_ONE_REG. 2688 2689Note: These registers are immutable once any of the vCPUs of the VM has 2690run at least once. A KVM_SET_ONE_REG in such a scenario will return 2691a -EBUSY to userspace. 2692 2693(See Documentation/virt/kvm/arm/hypercalls.rst for more details.) 2694 2695 2696MIPS registers are mapped using the lower 32 bits. The upper 16 of that is 2697the register group type: 2698 2699MIPS core registers (see above) have the following id bit patterns:: 2700 2701 0x7030 0000 0000 <reg:16> 2702 2703MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit 2704patterns depending on whether they're 32-bit or 64-bit registers:: 2705 2706 0x7020 0000 0001 00 <reg:5> <sel:3> (32-bit) 2707 0x7030 0000 0001 00 <reg:5> <sel:3> (64-bit) 2708 2709Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_MIPS_CP0_ENTRYLO1 are the MIPS64 2710versions of the EntryLo registers regardless of the word size of the host 2711hardware, host kernel, guest, and whether XPA is present in the guest, i.e. 2712with the RI and XI bits (if they exist) in bits 63 and 62 respectively, and 2713the PFNX field starting at bit 30. 2714 2715MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) above) have the following id bit 2716patterns:: 2717 2718 0x7030 0000 0001 01 <reg:8> 2719 2720MIPS KVM control registers (see above) have the following id bit patterns:: 2721 2722 0x7030 0000 0002 <reg:16> 2723 2724MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following 2725id bit patterns depending on the size of the register being accessed. They are 2726always accessed according to the current guest FPU mode (Status.FR and 2727Config5.FRE), i.e. as the guest would see them, and they become unpredictable 2728if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector 2729registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they 2730overlap the FPU registers:: 2731 2732 0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers) 2733 0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers) 2734 0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers) 2735 2736MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the 2737following id bit patterns:: 2738 2739 0x7020 0000 0003 01 <0:3> <reg:5> 2740 2741MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the 2742following id bit patterns:: 2743 2744 0x7020 0000 0003 02 <0:3> <reg:5> 2745 2746RISC-V registers are mapped using the lower 32 bits. The upper 8 bits of 2747that is the register group type. 2748 2749RISC-V config registers are meant for configuring a Guest VCPU and it has 2750the following id bit patterns:: 2751 2752 0x8020 0000 01 <index into the kvm_riscv_config struct:24> (32bit Host) 2753 0x8030 0000 01 <index into the kvm_riscv_config struct:24> (64bit Host) 2754 2755Following are the RISC-V config registers: 2756 2757======================= ========= ============================================= 2758 Encoding Register Description 2759======================= ========= ============================================= 2760 0x80x0 0000 0100 0000 isa ISA feature bitmap of Guest VCPU 2761======================= ========= ============================================= 2762 2763The isa config register can be read anytime but can only be written before 2764a Guest VCPU runs. It will have ISA feature bits matching underlying host 2765set by default. 2766 2767RISC-V core registers represent the general execution state of a Guest VCPU 2768and it has the following id bit patterns:: 2769 2770 0x8020 0000 02 <index into the kvm_riscv_core struct:24> (32bit Host) 2771 0x8030 0000 02 <index into the kvm_riscv_core struct:24> (64bit Host) 2772 2773Following are the RISC-V core registers: 2774 2775======================= ========= ============================================= 2776 Encoding Register Description 2777======================= ========= ============================================= 2778 0x80x0 0000 0200 0000 regs.pc Program counter 2779 0x80x0 0000 0200 0001 regs.ra Return address 2780 0x80x0 0000 0200 0002 regs.sp Stack pointer 2781 0x80x0 0000 0200 0003 regs.gp Global pointer 2782 0x80x0 0000 0200 0004 regs.tp Task pointer 2783 0x80x0 0000 0200 0005 regs.t0 Caller saved register 0 2784 0x80x0 0000 0200 0006 regs.t1 Caller saved register 1 2785 0x80x0 0000 0200 0007 regs.t2 Caller saved register 2 2786 0x80x0 0000 0200 0008 regs.s0 Callee saved register 0 2787 0x80x0 0000 0200 0009 regs.s1 Callee saved register 1 2788 0x80x0 0000 0200 000a regs.a0 Function argument (or return value) 0 2789 0x80x0 0000 0200 000b regs.a1 Function argument (or return value) 1 2790 0x80x0 0000 0200 000c regs.a2 Function argument 2 2791 0x80x0 0000 0200 000d regs.a3 Function argument 3 2792 0x80x0 0000 0200 000e regs.a4 Function argument 4 2793 0x80x0 0000 0200 000f regs.a5 Function argument 5 2794 0x80x0 0000 0200 0010 regs.a6 Function argument 6 2795 0x80x0 0000 0200 0011 regs.a7 Function argument 7 2796 0x80x0 0000 0200 0012 regs.s2 Callee saved register 2 2797 0x80x0 0000 0200 0013 regs.s3 Callee saved register 3 2798 0x80x0 0000 0200 0014 regs.s4 Callee saved register 4 2799 0x80x0 0000 0200 0015 regs.s5 Callee saved register 5 2800 0x80x0 0000 0200 0016 regs.s6 Callee saved register 6 2801 0x80x0 0000 0200 0017 regs.s7 Callee saved register 7 2802 0x80x0 0000 0200 0018 regs.s8 Callee saved register 8 2803 0x80x0 0000 0200 0019 regs.s9 Callee saved register 9 2804 0x80x0 0000 0200 001a regs.s10 Callee saved register 10 2805 0x80x0 0000 0200 001b regs.s11 Callee saved register 11 2806 0x80x0 0000 0200 001c regs.t3 Caller saved register 3 2807 0x80x0 0000 0200 001d regs.t4 Caller saved register 4 2808 0x80x0 0000 0200 001e regs.t5 Caller saved register 5 2809 0x80x0 0000 0200 001f regs.t6 Caller saved register 6 2810 0x80x0 0000 0200 0020 mode Privilege mode (1 = S-mode or 0 = U-mode) 2811======================= ========= ============================================= 2812 2813RISC-V csr registers represent the supervisor mode control/status registers 2814of a Guest VCPU and it has the following id bit patterns:: 2815 2816 0x8020 0000 03 <index into the kvm_riscv_csr struct:24> (32bit Host) 2817 0x8030 0000 03 <index into the kvm_riscv_csr struct:24> (64bit Host) 2818 2819Following are the RISC-V csr registers: 2820 2821======================= ========= ============================================= 2822 Encoding Register Description 2823======================= ========= ============================================= 2824 0x80x0 0000 0300 0000 sstatus Supervisor status 2825 0x80x0 0000 0300 0001 sie Supervisor interrupt enable 2826 0x80x0 0000 0300 0002 stvec Supervisor trap vector base 2827 0x80x0 0000 0300 0003 sscratch Supervisor scratch register 2828 0x80x0 0000 0300 0004 sepc Supervisor exception program counter 2829 0x80x0 0000 0300 0005 scause Supervisor trap cause 2830 0x80x0 0000 0300 0006 stval Supervisor bad address or instruction 2831 0x80x0 0000 0300 0007 sip Supervisor interrupt pending 2832 0x80x0 0000 0300 0008 satp Supervisor address translation and protection 2833======================= ========= ============================================= 2834 2835RISC-V timer registers represent the timer state of a Guest VCPU and it has 2836the following id bit patterns:: 2837 2838 0x8030 0000 04 <index into the kvm_riscv_timer struct:24> 2839 2840Following are the RISC-V timer registers: 2841 2842======================= ========= ============================================= 2843 Encoding Register Description 2844======================= ========= ============================================= 2845 0x8030 0000 0400 0000 frequency Time base frequency (read-only) 2846 0x8030 0000 0400 0001 time Time value visible to Guest 2847 0x8030 0000 0400 0002 compare Time compare programmed by Guest 2848 0x8030 0000 0400 0003 state Time compare state (1 = ON or 0 = OFF) 2849======================= ========= ============================================= 2850 2851RISC-V F-extension registers represent the single precision floating point 2852state of a Guest VCPU and it has the following id bit patterns:: 2853 2854 0x8020 0000 05 <index into the __riscv_f_ext_state struct:24> 2855 2856Following are the RISC-V F-extension registers: 2857 2858======================= ========= ============================================= 2859 Encoding Register Description 2860======================= ========= ============================================= 2861 0x8020 0000 0500 0000 f[0] Floating point register 0 2862 ... 2863 0x8020 0000 0500 001f f[31] Floating point register 31 2864 0x8020 0000 0500 0020 fcsr Floating point control and status register 2865======================= ========= ============================================= 2866 2867RISC-V D-extension registers represent the double precision floating point 2868state of a Guest VCPU and it has the following id bit patterns:: 2869 2870 0x8020 0000 06 <index into the __riscv_d_ext_state struct:24> (fcsr) 2871 0x8030 0000 06 <index into the __riscv_d_ext_state struct:24> (non-fcsr) 2872 2873Following are the RISC-V D-extension registers: 2874 2875======================= ========= ============================================= 2876 Encoding Register Description 2877======================= ========= ============================================= 2878 0x8030 0000 0600 0000 f[0] Floating point register 0 2879 ... 2880 0x8030 0000 0600 001f f[31] Floating point register 31 2881 0x8020 0000 0600 0020 fcsr Floating point control and status register 2882======================= ========= ============================================= 2883 2884LoongArch registers are mapped using the lower 32 bits. The upper 16 bits of 2885that is the register group type. 2886 2887LoongArch csr registers are used to control guest cpu or get status of guest 2888cpu, and they have the following id bit patterns:: 2889 2890 0x9030 0000 0001 00 <reg:5> <sel:3> (64-bit) 2891 2892LoongArch KVM control registers are used to implement some new defined functions 2893such as set vcpu counter or reset vcpu, and they have the following id bit patterns:: 2894 2895 0x9030 0000 0002 <reg:16> 2896 2897 28984.69 KVM_GET_ONE_REG 2899-------------------- 2900 2901:Capability: KVM_CAP_ONE_REG 2902:Architectures: all 2903:Type: vcpu ioctl 2904:Parameters: struct kvm_one_reg (in and out) 2905:Returns: 0 on success, negative value on failure 2906 2907Errors include: 2908 2909 ======== ============================================================ 2910 ENOENT no such register 2911 EINVAL invalid register ID, or no such register or used with VMs in 2912 protected virtualization mode on s390 2913 EPERM (arm64) register access not allowed before vcpu finalization 2914 ======== ============================================================ 2915 2916(These error codes are indicative only: do not rely on a specific error 2917code being returned in a specific situation.) 2918 2919This ioctl allows to receive the value of a single register implemented 2920in a vcpu. The register to read is indicated by the "id" field of the 2921kvm_one_reg struct passed in. On success, the register value can be found 2922at the memory location pointed to by "addr". 2923 2924The list of registers accessible using this interface is identical to the 2925list in 4.68. 2926 2927 29284.70 KVM_KVMCLOCK_CTRL 2929---------------------- 2930 2931:Capability: KVM_CAP_KVMCLOCK_CTRL 2932:Architectures: Any that implement pvclocks (currently x86 only) 2933:Type: vcpu ioctl 2934:Parameters: None 2935:Returns: 0 on success, -1 on error 2936 2937This ioctl sets a flag accessible to the guest indicating that the specified 2938vCPU has been paused by the host userspace. 2939 2940The host will set a flag in the pvclock structure that is checked from the 2941soft lockup watchdog. The flag is part of the pvclock structure that is 2942shared between guest and host, specifically the second bit of the flags 2943field of the pvclock_vcpu_time_info structure. It will be set exclusively by 2944the host and read/cleared exclusively by the guest. The guest operation of 2945checking and clearing the flag must be an atomic operation so 2946load-link/store-conditional, or equivalent must be used. There are two cases 2947where the guest will clear the flag: when the soft lockup watchdog timer resets 2948itself or when a soft lockup is detected. This ioctl can be called any time 2949after pausing the vcpu, but before it is resumed. 2950 2951 29524.71 KVM_SIGNAL_MSI 2953------------------- 2954 2955:Capability: KVM_CAP_SIGNAL_MSI 2956:Architectures: x86 arm64 2957:Type: vm ioctl 2958:Parameters: struct kvm_msi (in) 2959:Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error 2960 2961Directly inject a MSI message. Only valid with in-kernel irqchip that handles 2962MSI messages. 2963 2964:: 2965 2966 struct kvm_msi { 2967 __u32 address_lo; 2968 __u32 address_hi; 2969 __u32 data; 2970 __u32 flags; 2971 __u32 devid; 2972 __u8 pad[12]; 2973 }; 2974 2975flags: 2976 KVM_MSI_VALID_DEVID: devid contains a valid value. The per-VM 2977 KVM_CAP_MSI_DEVID capability advertises the requirement to provide 2978 the device ID. If this capability is not available, userspace 2979 should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail. 2980 2981If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier 2982for the device that wrote the MSI message. For PCI, this is usually a 2983BDF identifier in the lower 16 bits. 2984 2985On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS 2986feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled, 2987address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of 2988address_hi must be zero. 2989 2990 29914.71 KVM_CREATE_PIT2 2992-------------------- 2993 2994:Capability: KVM_CAP_PIT2 2995:Architectures: x86 2996:Type: vm ioctl 2997:Parameters: struct kvm_pit_config (in) 2998:Returns: 0 on success, -1 on error 2999 3000Creates an in-kernel device model for the i8254 PIT. This call is only valid 3001after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following 3002parameters have to be passed:: 3003 3004 struct kvm_pit_config { 3005 __u32 flags; 3006 __u32 pad[15]; 3007 }; 3008 3009Valid flags are:: 3010 3011 #define KVM_PIT_SPEAKER_DUMMY 1 /* emulate speaker port stub */ 3012 3013PIT timer interrupts may use a per-VM kernel thread for injection. If it 3014exists, this thread will have a name of the following pattern:: 3015 3016 kvm-pit/<owner-process-pid> 3017 3018When running a guest with elevated priorities, the scheduling parameters of 3019this thread may have to be adjusted accordingly. 3020 3021This IOCTL replaces the obsolete KVM_CREATE_PIT. 3022 3023 30244.72 KVM_GET_PIT2 3025----------------- 3026 3027:Capability: KVM_CAP_PIT_STATE2 3028:Architectures: x86 3029:Type: vm ioctl 3030:Parameters: struct kvm_pit_state2 (out) 3031:Returns: 0 on success, -1 on error 3032 3033Retrieves the state of the in-kernel PIT model. Only valid after 3034KVM_CREATE_PIT2. The state is returned in the following structure:: 3035 3036 struct kvm_pit_state2 { 3037 struct kvm_pit_channel_state channels[3]; 3038 __u32 flags; 3039 __u32 reserved[9]; 3040 }; 3041 3042Valid flags are:: 3043 3044 /* disable PIT in HPET legacy mode */ 3045 #define KVM_PIT_FLAGS_HPET_LEGACY 0x00000001 3046 /* speaker port data bit enabled */ 3047 #define KVM_PIT_FLAGS_SPEAKER_DATA_ON 0x00000002 3048 3049This IOCTL replaces the obsolete KVM_GET_PIT. 3050 3051 30524.73 KVM_SET_PIT2 3053----------------- 3054 3055:Capability: KVM_CAP_PIT_STATE2 3056:Architectures: x86 3057:Type: vm ioctl 3058:Parameters: struct kvm_pit_state2 (in) 3059:Returns: 0 on success, -1 on error 3060 3061Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2. 3062See KVM_GET_PIT2 for details on struct kvm_pit_state2. 3063 3064This IOCTL replaces the obsolete KVM_SET_PIT. 3065 3066 30674.74 KVM_PPC_GET_SMMU_INFO 3068-------------------------- 3069 3070:Capability: KVM_CAP_PPC_GET_SMMU_INFO 3071:Architectures: powerpc 3072:Type: vm ioctl 3073:Parameters: None 3074:Returns: 0 on success, -1 on error 3075 3076This populates and returns a structure describing the features of 3077the "Server" class MMU emulation supported by KVM. 3078This can in turn be used by userspace to generate the appropriate 3079device-tree properties for the guest operating system. 3080 3081The structure contains some global information, followed by an 3082array of supported segment page sizes:: 3083 3084 struct kvm_ppc_smmu_info { 3085 __u64 flags; 3086 __u32 slb_size; 3087 __u32 pad; 3088 struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ]; 3089 }; 3090 3091The supported flags are: 3092 3093 - KVM_PPC_PAGE_SIZES_REAL: 3094 When that flag is set, guest page sizes must "fit" the backing 3095 store page sizes. When not set, any page size in the list can 3096 be used regardless of how they are backed by userspace. 3097 3098 - KVM_PPC_1T_SEGMENTS 3099 The emulated MMU supports 1T segments in addition to the 3100 standard 256M ones. 3101 3102 - KVM_PPC_NO_HASH 3103 This flag indicates that HPT guests are not supported by KVM, 3104 thus all guests must use radix MMU mode. 3105 3106The "slb_size" field indicates how many SLB entries are supported 3107 3108The "sps" array contains 8 entries indicating the supported base 3109page sizes for a segment in increasing order. Each entry is defined 3110as follow:: 3111 3112 struct kvm_ppc_one_seg_page_size { 3113 __u32 page_shift; /* Base page shift of segment (or 0) */ 3114 __u32 slb_enc; /* SLB encoding for BookS */ 3115 struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ]; 3116 }; 3117 3118An entry with a "page_shift" of 0 is unused. Because the array is 3119organized in increasing order, a lookup can stop when encountering 3120such an entry. 3121 3122The "slb_enc" field provides the encoding to use in the SLB for the 3123page size. The bits are in positions such as the value can directly 3124be OR'ed into the "vsid" argument of the slbmte instruction. 3125 3126The "enc" array is a list which for each of those segment base page 3127size provides the list of supported actual page sizes (which can be 3128only larger or equal to the base page size), along with the 3129corresponding encoding in the hash PTE. Similarly, the array is 31308 entries sorted by increasing sizes and an entry with a "0" shift 3131is an empty entry and a terminator:: 3132 3133 struct kvm_ppc_one_page_size { 3134 __u32 page_shift; /* Page shift (or 0) */ 3135 __u32 pte_enc; /* Encoding in the HPTE (>>12) */ 3136 }; 3137 3138The "pte_enc" field provides a value that can OR'ed into the hash 3139PTE's RPN field (ie, it needs to be shifted left by 12 to OR it 3140into the hash PTE second double word). 3141 31424.75 KVM_IRQFD 3143-------------- 3144 3145:Capability: KVM_CAP_IRQFD 3146:Architectures: x86 s390 arm64 3147:Type: vm ioctl 3148:Parameters: struct kvm_irqfd (in) 3149:Returns: 0 on success, -1 on error 3150 3151Allows setting an eventfd to directly trigger a guest interrupt. 3152kvm_irqfd.fd specifies the file descriptor to use as the eventfd and 3153kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When 3154an event is triggered on the eventfd, an interrupt is injected into 3155the guest using the specified gsi pin. The irqfd is removed using 3156the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd 3157and kvm_irqfd.gsi. 3158 3159With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify 3160mechanism allowing emulation of level-triggered, irqfd-based 3161interrupts. When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an 3162additional eventfd in the kvm_irqfd.resamplefd field. When operating 3163in resample mode, posting of an interrupt through kvm_irq.fd asserts 3164the specified gsi in the irqchip. When the irqchip is resampled, such 3165as from an EOI, the gsi is de-asserted and the user is notified via 3166kvm_irqfd.resamplefd. It is the user's responsibility to re-queue 3167the interrupt if the device making use of it still requires service. 3168Note that closing the resamplefd is not sufficient to disable the 3169irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment 3170and need not be specified with KVM_IRQFD_FLAG_DEASSIGN. 3171 3172On arm64, gsi routing being supported, the following can happen: 3173 3174- in case no routing entry is associated to this gsi, injection fails 3175- in case the gsi is associated to an irqchip routing entry, 3176 irqchip.pin + 32 corresponds to the injected SPI ID. 3177- in case the gsi is associated to an MSI routing entry, the MSI 3178 message and device ID are translated into an LPI (support restricted 3179 to GICv3 ITS in-kernel emulation). 3180 31814.76 KVM_PPC_ALLOCATE_HTAB 3182-------------------------- 3183 3184:Capability: KVM_CAP_PPC_ALLOC_HTAB 3185:Architectures: powerpc 3186:Type: vm ioctl 3187:Parameters: Pointer to u32 containing hash table order (in/out) 3188:Returns: 0 on success, -1 on error 3189 3190This requests the host kernel to allocate an MMU hash table for a 3191guest using the PAPR paravirtualization interface. This only does 3192anything if the kernel is configured to use the Book 3S HV style of 3193virtualization. Otherwise the capability doesn't exist and the ioctl 3194returns an ENOTTY error. The rest of this description assumes Book 3S 3195HV. 3196 3197There must be no vcpus running when this ioctl is called; if there 3198are, it will do nothing and return an EBUSY error. 3199 3200The parameter is a pointer to a 32-bit unsigned integer variable 3201containing the order (log base 2) of the desired size of the hash 3202table, which must be between 18 and 46. On successful return from the 3203ioctl, the value will not be changed by the kernel. 3204 3205If no hash table has been allocated when any vcpu is asked to run 3206(with the KVM_RUN ioctl), the host kernel will allocate a 3207default-sized hash table (16 MB). 3208 3209If this ioctl is called when a hash table has already been allocated, 3210with a different order from the existing hash table, the existing hash 3211table will be freed and a new one allocated. If this is ioctl is 3212called when a hash table has already been allocated of the same order 3213as specified, the kernel will clear out the existing hash table (zero 3214all HPTEs). In either case, if the guest is using the virtualized 3215real-mode area (VRMA) facility, the kernel will re-create the VMRA 3216HPTEs on the next KVM_RUN of any vcpu. 3217 32184.77 KVM_S390_INTERRUPT 3219----------------------- 3220 3221:Capability: basic 3222:Architectures: s390 3223:Type: vm ioctl, vcpu ioctl 3224:Parameters: struct kvm_s390_interrupt (in) 3225:Returns: 0 on success, -1 on error 3226 3227Allows to inject an interrupt to the guest. Interrupts can be floating 3228(vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type. 3229 3230Interrupt parameters are passed via kvm_s390_interrupt:: 3231 3232 struct kvm_s390_interrupt { 3233 __u32 type; 3234 __u32 parm; 3235 __u64 parm64; 3236 }; 3237 3238type can be one of the following: 3239 3240KVM_S390_SIGP_STOP (vcpu) 3241 - sigp stop; optional flags in parm 3242KVM_S390_PROGRAM_INT (vcpu) 3243 - program check; code in parm 3244KVM_S390_SIGP_SET_PREFIX (vcpu) 3245 - sigp set prefix; prefix address in parm 3246KVM_S390_RESTART (vcpu) 3247 - restart 3248KVM_S390_INT_CLOCK_COMP (vcpu) 3249 - clock comparator interrupt 3250KVM_S390_INT_CPU_TIMER (vcpu) 3251 - CPU timer interrupt 3252KVM_S390_INT_VIRTIO (vm) 3253 - virtio external interrupt; external interrupt 3254 parameters in parm and parm64 3255KVM_S390_INT_SERVICE (vm) 3256 - sclp external interrupt; sclp parameter in parm 3257KVM_S390_INT_EMERGENCY (vcpu) 3258 - sigp emergency; source cpu in parm 3259KVM_S390_INT_EXTERNAL_CALL (vcpu) 3260 - sigp external call; source cpu in parm 3261KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm) 3262 - compound value to indicate an 3263 I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel); 3264 I/O interruption parameters in parm (subchannel) and parm64 (intparm, 3265 interruption subclass) 3266KVM_S390_MCHK (vm, vcpu) 3267 - machine check interrupt; cr 14 bits in parm, machine check interrupt 3268 code in parm64 (note that machine checks needing further payload are not 3269 supported by this ioctl) 3270 3271This is an asynchronous vcpu ioctl and can be invoked from any thread. 3272 32734.78 KVM_PPC_GET_HTAB_FD 3274------------------------ 3275 3276:Capability: KVM_CAP_PPC_HTAB_FD 3277:Architectures: powerpc 3278:Type: vm ioctl 3279:Parameters: Pointer to struct kvm_get_htab_fd (in) 3280:Returns: file descriptor number (>= 0) on success, -1 on error 3281 3282This returns a file descriptor that can be used either to read out the 3283entries in the guest's hashed page table (HPT), or to write entries to 3284initialize the HPT. The returned fd can only be written to if the 3285KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and 3286can only be read if that bit is clear. The argument struct looks like 3287this:: 3288 3289 /* For KVM_PPC_GET_HTAB_FD */ 3290 struct kvm_get_htab_fd { 3291 __u64 flags; 3292 __u64 start_index; 3293 __u64 reserved[2]; 3294 }; 3295 3296 /* Values for kvm_get_htab_fd.flags */ 3297 #define KVM_GET_HTAB_BOLTED_ONLY ((__u64)0x1) 3298 #define KVM_GET_HTAB_WRITE ((__u64)0x2) 3299 3300The 'start_index' field gives the index in the HPT of the entry at 3301which to start reading. It is ignored when writing. 3302 3303Reads on the fd will initially supply information about all 3304"interesting" HPT entries. Interesting entries are those with the 3305bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise 3306all entries. When the end of the HPT is reached, the read() will 3307return. If read() is called again on the fd, it will start again from 3308the beginning of the HPT, but will only return HPT entries that have 3309changed since they were last read. 3310 3311Data read or written is structured as a header (8 bytes) followed by a 3312series of valid HPT entries (16 bytes) each. The header indicates how 3313many valid HPT entries there are and how many invalid entries follow 3314the valid entries. The invalid entries are not represented explicitly 3315in the stream. The header format is:: 3316 3317 struct kvm_get_htab_header { 3318 __u32 index; 3319 __u16 n_valid; 3320 __u16 n_invalid; 3321 }; 3322 3323Writes to the fd create HPT entries starting at the index given in the 3324header; first 'n_valid' valid entries with contents from the data 3325written, then 'n_invalid' invalid entries, invalidating any previously 3326valid entries found. 3327 33284.79 KVM_CREATE_DEVICE 3329---------------------- 3330 3331:Capability: KVM_CAP_DEVICE_CTRL 3332:Architectures: all 3333:Type: vm ioctl 3334:Parameters: struct kvm_create_device (in/out) 3335:Returns: 0 on success, -1 on error 3336 3337Errors: 3338 3339 ====== ======================================================= 3340 ENODEV The device type is unknown or unsupported 3341 EEXIST Device already created, and this type of device may not 3342 be instantiated multiple times 3343 ====== ======================================================= 3344 3345 Other error conditions may be defined by individual device types or 3346 have their standard meanings. 3347 3348Creates an emulated device in the kernel. The file descriptor returned 3349in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR. 3350 3351If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the 3352device type is supported (not necessarily whether it can be created 3353in the current vm). 3354 3355Individual devices should not define flags. Attributes should be used 3356for specifying any behavior that is not implied by the device type 3357number. 3358 3359:: 3360 3361 struct kvm_create_device { 3362 __u32 type; /* in: KVM_DEV_TYPE_xxx */ 3363 __u32 fd; /* out: device handle */ 3364 __u32 flags; /* in: KVM_CREATE_DEVICE_xxx */ 3365 }; 3366 33674.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR 3368-------------------------------------------- 3369 3370:Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device, 3371 KVM_CAP_VCPU_ATTRIBUTES for vcpu device 3372 KVM_CAP_SYS_ATTRIBUTES for system (/dev/kvm) device (no set) 3373:Architectures: x86, arm64, s390 3374:Type: device ioctl, vm ioctl, vcpu ioctl 3375:Parameters: struct kvm_device_attr 3376:Returns: 0 on success, -1 on error 3377 3378Errors: 3379 3380 ===== ============================================================= 3381 ENXIO The group or attribute is unknown/unsupported for this device 3382 or hardware support is missing. 3383 EPERM The attribute cannot (currently) be accessed this way 3384 (e.g. read-only attribute, or attribute that only makes 3385 sense when the device is in a different state) 3386 ===== ============================================================= 3387 3388 Other error conditions may be defined by individual device types. 3389 3390Gets/sets a specified piece of device configuration and/or state. The 3391semantics are device-specific. See individual device documentation in 3392the "devices" directory. As with ONE_REG, the size of the data 3393transferred is defined by the particular attribute. 3394 3395:: 3396 3397 struct kvm_device_attr { 3398 __u32 flags; /* no flags currently defined */ 3399 __u32 group; /* device-defined */ 3400 __u64 attr; /* group-defined */ 3401 __u64 addr; /* userspace address of attr data */ 3402 }; 3403 34044.81 KVM_HAS_DEVICE_ATTR 3405------------------------ 3406 3407:Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device, 3408 KVM_CAP_VCPU_ATTRIBUTES for vcpu device 3409 KVM_CAP_SYS_ATTRIBUTES for system (/dev/kvm) device 3410:Type: device ioctl, vm ioctl, vcpu ioctl 3411:Parameters: struct kvm_device_attr 3412:Returns: 0 on success, -1 on error 3413 3414Errors: 3415 3416 ===== ============================================================= 3417 ENXIO The group or attribute is unknown/unsupported for this device 3418 or hardware support is missing. 3419 ===== ============================================================= 3420 3421Tests whether a device supports a particular attribute. A successful 3422return indicates the attribute is implemented. It does not necessarily 3423indicate that the attribute can be read or written in the device's 3424current state. "addr" is ignored. 3425 3426.. _KVM_ARM_VCPU_INIT: 3427 34284.82 KVM_ARM_VCPU_INIT 3429---------------------- 3430 3431:Capability: basic 3432:Architectures: arm64 3433:Type: vcpu ioctl 3434:Parameters: struct kvm_vcpu_init (in) 3435:Returns: 0 on success; -1 on error 3436 3437Errors: 3438 3439 ====== ================================================================= 3440 EINVAL the target is unknown, or the combination of features is invalid. 3441 ENOENT a features bit specified is unknown. 3442 ====== ================================================================= 3443 3444This tells KVM what type of CPU to present to the guest, and what 3445optional features it should have. This will cause a reset of the cpu 3446registers to their initial values. If this is not called, KVM_RUN will 3447return ENOEXEC for that vcpu. 3448 3449The initial values are defined as: 3450 - Processor state: 3451 * AArch64: EL1h, D, A, I and F bits set. All other bits 3452 are cleared. 3453 * AArch32: SVC, A, I and F bits set. All other bits are 3454 cleared. 3455 - General Purpose registers, including PC and SP: set to 0 3456 - FPSIMD/NEON registers: set to 0 3457 - SVE registers: set to 0 3458 - System registers: Reset to their architecturally defined 3459 values as for a warm reset to EL1 (resp. SVC) 3460 3461Note that because some registers reflect machine topology, all vcpus 3462should be created before this ioctl is invoked. 3463 3464Userspace can call this function multiple times for a given vcpu, including 3465after the vcpu has been run. This will reset the vcpu to its initial 3466state. All calls to this function after the initial call must use the same 3467target and same set of feature flags, otherwise EINVAL will be returned. 3468 3469Possible features: 3470 3471 - KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state. 3472 Depends on KVM_CAP_ARM_PSCI. If not set, the CPU will be powered on 3473 and execute guest code when KVM_RUN is called. 3474 - KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode. 3475 Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only). 3476 - KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 (or a future revision 3477 backward compatible with v0.2) for the CPU. 3478 Depends on KVM_CAP_ARM_PSCI_0_2. 3479 - KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU. 3480 Depends on KVM_CAP_ARM_PMU_V3. 3481 3482 - KVM_ARM_VCPU_PTRAUTH_ADDRESS: Enables Address Pointer authentication 3483 for arm64 only. 3484 Depends on KVM_CAP_ARM_PTRAUTH_ADDRESS. 3485 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are 3486 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and 3487 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be 3488 requested. 3489 3490 - KVM_ARM_VCPU_PTRAUTH_GENERIC: Enables Generic Pointer authentication 3491 for arm64 only. 3492 Depends on KVM_CAP_ARM_PTRAUTH_GENERIC. 3493 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are 3494 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and 3495 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be 3496 requested. 3497 3498 - KVM_ARM_VCPU_SVE: Enables SVE for the CPU (arm64 only). 3499 Depends on KVM_CAP_ARM_SVE. 3500 Requires KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3501 3502 * After KVM_ARM_VCPU_INIT: 3503 3504 - KVM_REG_ARM64_SVE_VLS may be read using KVM_GET_ONE_REG: the 3505 initial value of this pseudo-register indicates the best set of 3506 vector lengths possible for a vcpu on this host. 3507 3508 * Before KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3509 3510 - KVM_RUN and KVM_GET_REG_LIST are not available; 3511 3512 - KVM_GET_ONE_REG and KVM_SET_ONE_REG cannot be used to access 3513 the scalable architectural SVE registers 3514 KVM_REG_ARM64_SVE_ZREG(), KVM_REG_ARM64_SVE_PREG() or 3515 KVM_REG_ARM64_SVE_FFR; 3516 3517 - KVM_REG_ARM64_SVE_VLS may optionally be written using 3518 KVM_SET_ONE_REG, to modify the set of vector lengths available 3519 for the vcpu. 3520 3521 * After KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3522 3523 - the KVM_REG_ARM64_SVE_VLS pseudo-register is immutable, and can 3524 no longer be written using KVM_SET_ONE_REG. 3525 35264.83 KVM_ARM_PREFERRED_TARGET 3527----------------------------- 3528 3529:Capability: basic 3530:Architectures: arm64 3531:Type: vm ioctl 3532:Parameters: struct kvm_vcpu_init (out) 3533:Returns: 0 on success; -1 on error 3534 3535Errors: 3536 3537 ====== ========================================== 3538 ENODEV no preferred target available for the host 3539 ====== ========================================== 3540 3541This queries KVM for preferred CPU target type which can be emulated 3542by KVM on underlying host. 3543 3544The ioctl returns struct kvm_vcpu_init instance containing information 3545about preferred CPU target type and recommended features for it. The 3546kvm_vcpu_init->features bitmap returned will have feature bits set if 3547the preferred target recommends setting these features, but this is 3548not mandatory. 3549 3550The information returned by this ioctl can be used to prepare an instance 3551of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in 3552VCPU matching underlying host. 3553 3554 35554.84 KVM_GET_REG_LIST 3556--------------------- 3557 3558:Capability: basic 3559:Architectures: arm64, mips, riscv 3560:Type: vcpu ioctl 3561:Parameters: struct kvm_reg_list (in/out) 3562:Returns: 0 on success; -1 on error 3563 3564Errors: 3565 3566 ===== ============================================================== 3567 E2BIG the reg index list is too big to fit in the array specified by 3568 the user (the number required will be written into n). 3569 ===== ============================================================== 3570 3571:: 3572 3573 struct kvm_reg_list { 3574 __u64 n; /* number of registers in reg[] */ 3575 __u64 reg[0]; 3576 }; 3577 3578This ioctl returns the guest registers that are supported for the 3579KVM_GET_ONE_REG/KVM_SET_ONE_REG calls. 3580 3581Note that s390 does not support KVM_GET_REG_LIST for historical reasons 3582(read: nobody cared). The set of registers in kernels 4.x and newer is: 3583 3584- KVM_REG_S390_TODPR 3585 3586- KVM_REG_S390_EPOCHDIFF 3587 3588- KVM_REG_S390_CPU_TIMER 3589 3590- KVM_REG_S390_CLOCK_COMP 3591 3592- KVM_REG_S390_PFTOKEN 3593 3594- KVM_REG_S390_PFCOMPARE 3595 3596- KVM_REG_S390_PFSELECT 3597 3598- KVM_REG_S390_PP 3599 3600- KVM_REG_S390_GBEA 3601 3602 36034.85 KVM_ARM_SET_DEVICE_ADDR (deprecated) 3604----------------------------------------- 3605 3606:Capability: KVM_CAP_ARM_SET_DEVICE_ADDR 3607:Architectures: arm64 3608:Type: vm ioctl 3609:Parameters: struct kvm_arm_device_address (in) 3610:Returns: 0 on success, -1 on error 3611 3612Errors: 3613 3614 ====== ============================================ 3615 ENODEV The device id is unknown 3616 ENXIO Device not supported on current system 3617 EEXIST Address already set 3618 E2BIG Address outside guest physical address space 3619 EBUSY Address overlaps with other device range 3620 ====== ============================================ 3621 3622:: 3623 3624 struct kvm_arm_device_addr { 3625 __u64 id; 3626 __u64 addr; 3627 }; 3628 3629Specify a device address in the guest's physical address space where guests 3630can access emulated or directly exposed devices, which the host kernel needs 3631to know about. The id field is an architecture specific identifier for a 3632specific device. 3633 3634arm64 divides the id field into two parts, a device id and an 3635address type id specific to the individual device:: 3636 3637 bits: | 63 ... 32 | 31 ... 16 | 15 ... 0 | 3638 field: | 0x00000000 | device id | addr type id | 3639 3640arm64 currently only require this when using the in-kernel GIC 3641support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2 3642as the device id. When setting the base address for the guest's 3643mapping of the VGIC virtual CPU and distributor interface, the ioctl 3644must be called after calling KVM_CREATE_IRQCHIP, but before calling 3645KVM_RUN on any of the VCPUs. Calling this ioctl twice for any of the 3646base addresses will return -EEXIST. 3647 3648Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API 3649should be used instead. 3650 3651 36524.86 KVM_PPC_RTAS_DEFINE_TOKEN 3653------------------------------ 3654 3655:Capability: KVM_CAP_PPC_RTAS 3656:Architectures: ppc 3657:Type: vm ioctl 3658:Parameters: struct kvm_rtas_token_args 3659:Returns: 0 on success, -1 on error 3660 3661Defines a token value for a RTAS (Run Time Abstraction Services) 3662service in order to allow it to be handled in the kernel. The 3663argument struct gives the name of the service, which must be the name 3664of a service that has a kernel-side implementation. If the token 3665value is non-zero, it will be associated with that service, and 3666subsequent RTAS calls by the guest specifying that token will be 3667handled by the kernel. If the token value is 0, then any token 3668associated with the service will be forgotten, and subsequent RTAS 3669calls by the guest for that service will be passed to userspace to be 3670handled. 3671 36724.87 KVM_SET_GUEST_DEBUG 3673------------------------ 3674 3675:Capability: KVM_CAP_SET_GUEST_DEBUG 3676:Architectures: x86, s390, ppc, arm64 3677:Type: vcpu ioctl 3678:Parameters: struct kvm_guest_debug (in) 3679:Returns: 0 on success; -1 on error 3680 3681:: 3682 3683 struct kvm_guest_debug { 3684 __u32 control; 3685 __u32 pad; 3686 struct kvm_guest_debug_arch arch; 3687 }; 3688 3689Set up the processor specific debug registers and configure vcpu for 3690handling guest debug events. There are two parts to the structure, the 3691first a control bitfield indicates the type of debug events to handle 3692when running. Common control bits are: 3693 3694 - KVM_GUESTDBG_ENABLE: guest debugging is enabled 3695 - KVM_GUESTDBG_SINGLESTEP: the next run should single-step 3696 3697The top 16 bits of the control field are architecture specific control 3698flags which can include the following: 3699 3700 - KVM_GUESTDBG_USE_SW_BP: using software breakpoints [x86, arm64] 3701 - KVM_GUESTDBG_USE_HW_BP: using hardware breakpoints [x86, s390] 3702 - KVM_GUESTDBG_USE_HW: using hardware debug events [arm64] 3703 - KVM_GUESTDBG_INJECT_DB: inject DB type exception [x86] 3704 - KVM_GUESTDBG_INJECT_BP: inject BP type exception [x86] 3705 - KVM_GUESTDBG_EXIT_PENDING: trigger an immediate guest exit [s390] 3706 - KVM_GUESTDBG_BLOCKIRQ: avoid injecting interrupts/NMI/SMI [x86] 3707 3708For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints 3709are enabled in memory so we need to ensure breakpoint exceptions are 3710correctly trapped and the KVM run loop exits at the breakpoint and not 3711running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP 3712we need to ensure the guest vCPUs architecture specific registers are 3713updated to the correct (supplied) values. 3714 3715The second part of the structure is architecture specific and 3716typically contains a set of debug registers. 3717 3718For arm64 the number of debug registers is implementation defined and 3719can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and 3720KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number 3721indicating the number of supported registers. 3722 3723For ppc, the KVM_CAP_PPC_GUEST_DEBUG_SSTEP capability indicates whether 3724the single-step debug event (KVM_GUESTDBG_SINGLESTEP) is supported. 3725 3726Also when supported, KVM_CAP_SET_GUEST_DEBUG2 capability indicates the 3727supported KVM_GUESTDBG_* bits in the control field. 3728 3729When debug events exit the main run loop with the reason 3730KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run 3731structure containing architecture specific debug information. 3732 37334.88 KVM_GET_EMULATED_CPUID 3734--------------------------- 3735 3736:Capability: KVM_CAP_EXT_EMUL_CPUID 3737:Architectures: x86 3738:Type: system ioctl 3739:Parameters: struct kvm_cpuid2 (in/out) 3740:Returns: 0 on success, -1 on error 3741 3742:: 3743 3744 struct kvm_cpuid2 { 3745 __u32 nent; 3746 __u32 flags; 3747 struct kvm_cpuid_entry2 entries[0]; 3748 }; 3749 3750The member 'flags' is used for passing flags from userspace. 3751 3752:: 3753 3754 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0) 3755 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */ 3756 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */ 3757 3758 struct kvm_cpuid_entry2 { 3759 __u32 function; 3760 __u32 index; 3761 __u32 flags; 3762 __u32 eax; 3763 __u32 ebx; 3764 __u32 ecx; 3765 __u32 edx; 3766 __u32 padding[3]; 3767 }; 3768 3769This ioctl returns x86 cpuid features which are emulated by 3770kvm.Userspace can use the information returned by this ioctl to query 3771which features are emulated by kvm instead of being present natively. 3772 3773Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2 3774structure with the 'nent' field indicating the number of entries in 3775the variable-size array 'entries'. If the number of entries is too low 3776to describe the cpu capabilities, an error (E2BIG) is returned. If the 3777number is too high, the 'nent' field is adjusted and an error (ENOMEM) 3778is returned. If the number is just right, the 'nent' field is adjusted 3779to the number of valid entries in the 'entries' array, which is then 3780filled. 3781 3782The entries returned are the set CPUID bits of the respective features 3783which kvm emulates, as returned by the CPUID instruction, with unknown 3784or unsupported feature bits cleared. 3785 3786Features like x2apic, for example, may not be present in the host cpu 3787but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be 3788emulated efficiently and thus not included here. 3789 3790The fields in each entry are defined as follows: 3791 3792 function: 3793 the eax value used to obtain the entry 3794 index: 3795 the ecx value used to obtain the entry (for entries that are 3796 affected by ecx) 3797 flags: 3798 an OR of zero or more of the following: 3799 3800 KVM_CPUID_FLAG_SIGNIFCANT_INDEX: 3801 if the index field is valid 3802 3803 eax, ebx, ecx, edx: 3804 3805 the values returned by the cpuid instruction for 3806 this function/index combination 3807 38084.89 KVM_S390_MEM_OP 3809-------------------- 3810 3811:Capability: KVM_CAP_S390_MEM_OP, KVM_CAP_S390_PROTECTED, KVM_CAP_S390_MEM_OP_EXTENSION 3812:Architectures: s390 3813:Type: vm ioctl, vcpu ioctl 3814:Parameters: struct kvm_s390_mem_op (in) 3815:Returns: = 0 on success, 3816 < 0 on generic error (e.g. -EFAULT or -ENOMEM), 3817 16 bit program exception code if the access causes such an exception 3818 3819Read or write data from/to the VM's memory. 3820The KVM_CAP_S390_MEM_OP_EXTENSION capability specifies what functionality is 3821supported. 3822 3823Parameters are specified via the following structure:: 3824 3825 struct kvm_s390_mem_op { 3826 __u64 gaddr; /* the guest address */ 3827 __u64 flags; /* flags */ 3828 __u32 size; /* amount of bytes */ 3829 __u32 op; /* type of operation */ 3830 __u64 buf; /* buffer in userspace */ 3831 union { 3832 struct { 3833 __u8 ar; /* the access register number */ 3834 __u8 key; /* access key, ignored if flag unset */ 3835 __u8 pad1[6]; /* ignored */ 3836 __u64 old_addr; /* ignored if flag unset */ 3837 }; 3838 __u32 sida_offset; /* offset into the sida */ 3839 __u8 reserved[32]; /* ignored */ 3840 }; 3841 }; 3842 3843The start address of the memory region has to be specified in the "gaddr" 3844field, and the length of the region in the "size" field (which must not 3845be 0). The maximum value for "size" can be obtained by checking the 3846KVM_CAP_S390_MEM_OP capability. "buf" is the buffer supplied by the 3847userspace application where the read data should be written to for 3848a read access, or where the data that should be written is stored for 3849a write access. The "reserved" field is meant for future extensions. 3850Reserved and unused values are ignored. Future extension that add members must 3851introduce new flags. 3852 3853The type of operation is specified in the "op" field. Flags modifying 3854their behavior can be set in the "flags" field. Undefined flag bits must 3855be set to 0. 3856 3857Possible operations are: 3858 * ``KVM_S390_MEMOP_LOGICAL_READ`` 3859 * ``KVM_S390_MEMOP_LOGICAL_WRITE`` 3860 * ``KVM_S390_MEMOP_ABSOLUTE_READ`` 3861 * ``KVM_S390_MEMOP_ABSOLUTE_WRITE`` 3862 * ``KVM_S390_MEMOP_SIDA_READ`` 3863 * ``KVM_S390_MEMOP_SIDA_WRITE`` 3864 * ``KVM_S390_MEMOP_ABSOLUTE_CMPXCHG`` 3865 3866Logical read/write: 3867^^^^^^^^^^^^^^^^^^^ 3868 3869Access logical memory, i.e. translate the given guest address to an absolute 3870address given the state of the VCPU and use the absolute address as target of 3871the access. "ar" designates the access register number to be used; the valid 3872range is 0..15. 3873Logical accesses are permitted for the VCPU ioctl only. 3874Logical accesses are permitted for non-protected guests only. 3875 3876Supported flags: 3877 * ``KVM_S390_MEMOP_F_CHECK_ONLY`` 3878 * ``KVM_S390_MEMOP_F_INJECT_EXCEPTION`` 3879 * ``KVM_S390_MEMOP_F_SKEY_PROTECTION`` 3880 3881The KVM_S390_MEMOP_F_CHECK_ONLY flag can be set to check whether the 3882corresponding memory access would cause an access exception; however, 3883no actual access to the data in memory at the destination is performed. 3884In this case, "buf" is unused and can be NULL. 3885 3886In case an access exception occurred during the access (or would occur 3887in case of KVM_S390_MEMOP_F_CHECK_ONLY), the ioctl returns a positive 3888error number indicating the type of exception. This exception is also 3889raised directly at the corresponding VCPU if the flag 3890KVM_S390_MEMOP_F_INJECT_EXCEPTION is set. 3891On protection exceptions, unless specified otherwise, the injected 3892translation-exception identifier (TEID) indicates suppression. 3893 3894If the KVM_S390_MEMOP_F_SKEY_PROTECTION flag is set, storage key 3895protection is also in effect and may cause exceptions if accesses are 3896prohibited given the access key designated by "key"; the valid range is 0..15. 3897KVM_S390_MEMOP_F_SKEY_PROTECTION is available if KVM_CAP_S390_MEM_OP_EXTENSION 3898is > 0. 3899Since the accessed memory may span multiple pages and those pages might have 3900different storage keys, it is possible that a protection exception occurs 3901after memory has been modified. In this case, if the exception is injected, 3902the TEID does not indicate suppression. 3903 3904Absolute read/write: 3905^^^^^^^^^^^^^^^^^^^^ 3906 3907Access absolute memory. This operation is intended to be used with the 3908KVM_S390_MEMOP_F_SKEY_PROTECTION flag, to allow accessing memory and performing 3909the checks required for storage key protection as one operation (as opposed to 3910user space getting the storage keys, performing the checks, and accessing 3911memory thereafter, which could lead to a delay between check and access). 3912Absolute accesses are permitted for the VM ioctl if KVM_CAP_S390_MEM_OP_EXTENSION 3913has the KVM_S390_MEMOP_EXTENSION_CAP_BASE bit set. 3914Currently absolute accesses are not permitted for VCPU ioctls. 3915Absolute accesses are permitted for non-protected guests only. 3916 3917Supported flags: 3918 * ``KVM_S390_MEMOP_F_CHECK_ONLY`` 3919 * ``KVM_S390_MEMOP_F_SKEY_PROTECTION`` 3920 3921The semantics of the flags common with logical accesses are as for logical 3922accesses. 3923 3924Absolute cmpxchg: 3925^^^^^^^^^^^^^^^^^ 3926 3927Perform cmpxchg on absolute guest memory. Intended for use with the 3928KVM_S390_MEMOP_F_SKEY_PROTECTION flag. 3929Instead of doing an unconditional write, the access occurs only if the target 3930location contains the value pointed to by "old_addr". 3931This is performed as an atomic cmpxchg with the length specified by the "size" 3932parameter. "size" must be a power of two up to and including 16. 3933If the exchange did not take place because the target value doesn't match the 3934old value, the value "old_addr" points to is replaced by the target value. 3935User space can tell if an exchange took place by checking if this replacement 3936occurred. The cmpxchg op is permitted for the VM ioctl if 3937KVM_CAP_S390_MEM_OP_EXTENSION has flag KVM_S390_MEMOP_EXTENSION_CAP_CMPXCHG set. 3938 3939Supported flags: 3940 * ``KVM_S390_MEMOP_F_SKEY_PROTECTION`` 3941 3942SIDA read/write: 3943^^^^^^^^^^^^^^^^ 3944 3945Access the secure instruction data area which contains memory operands necessary 3946for instruction emulation for protected guests. 3947SIDA accesses are available if the KVM_CAP_S390_PROTECTED capability is available. 3948SIDA accesses are permitted for the VCPU ioctl only. 3949SIDA accesses are permitted for protected guests only. 3950 3951No flags are supported. 3952 39534.90 KVM_S390_GET_SKEYS 3954----------------------- 3955 3956:Capability: KVM_CAP_S390_SKEYS 3957:Architectures: s390 3958:Type: vm ioctl 3959:Parameters: struct kvm_s390_skeys 3960:Returns: 0 on success, KVM_S390_GET_SKEYS_NONE if guest is not using storage 3961 keys, negative value on error 3962 3963This ioctl is used to get guest storage key values on the s390 3964architecture. The ioctl takes parameters via the kvm_s390_skeys struct:: 3965 3966 struct kvm_s390_skeys { 3967 __u64 start_gfn; 3968 __u64 count; 3969 __u64 skeydata_addr; 3970 __u32 flags; 3971 __u32 reserved[9]; 3972 }; 3973 3974The start_gfn field is the number of the first guest frame whose storage keys 3975you want to get. 3976 3977The count field is the number of consecutive frames (starting from start_gfn) 3978whose storage keys to get. The count field must be at least 1 and the maximum 3979allowed value is defined as KVM_S390_SKEYS_MAX. Values outside this range 3980will cause the ioctl to return -EINVAL. 3981 3982The skeydata_addr field is the address to a buffer large enough to hold count 3983bytes. This buffer will be filled with storage key data by the ioctl. 3984 39854.91 KVM_S390_SET_SKEYS 3986----------------------- 3987 3988:Capability: KVM_CAP_S390_SKEYS 3989:Architectures: s390 3990:Type: vm ioctl 3991:Parameters: struct kvm_s390_skeys 3992:Returns: 0 on success, negative value on error 3993 3994This ioctl is used to set guest storage key values on the s390 3995architecture. The ioctl takes parameters via the kvm_s390_skeys struct. 3996See section on KVM_S390_GET_SKEYS for struct definition. 3997 3998The start_gfn field is the number of the first guest frame whose storage keys 3999you want to set. 4000 4001The count field is the number of consecutive frames (starting from start_gfn) 4002whose storage keys to get. The count field must be at least 1 and the maximum 4003allowed value is defined as KVM_S390_SKEYS_MAX. Values outside this range 4004will cause the ioctl to return -EINVAL. 4005 4006The skeydata_addr field is the address to a buffer containing count bytes of 4007storage keys. Each byte in the buffer will be set as the storage key for a 4008single frame starting at start_gfn for count frames. 4009 4010Note: If any architecturally invalid key value is found in the given data then 4011the ioctl will return -EINVAL. 4012 40134.92 KVM_S390_IRQ 4014----------------- 4015 4016:Capability: KVM_CAP_S390_INJECT_IRQ 4017:Architectures: s390 4018:Type: vcpu ioctl 4019:Parameters: struct kvm_s390_irq (in) 4020:Returns: 0 on success, -1 on error 4021 4022Errors: 4023 4024 4025 ====== ================================================================= 4026 EINVAL interrupt type is invalid 4027 type is KVM_S390_SIGP_STOP and flag parameter is invalid value, 4028 type is KVM_S390_INT_EXTERNAL_CALL and code is bigger 4029 than the maximum of VCPUs 4030 EBUSY type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped, 4031 type is KVM_S390_SIGP_STOP and a stop irq is already pending, 4032 type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt 4033 is already pending 4034 ====== ================================================================= 4035 4036Allows to inject an interrupt to the guest. 4037 4038Using struct kvm_s390_irq as a parameter allows 4039to inject additional payload which is not 4040possible via KVM_S390_INTERRUPT. 4041 4042Interrupt parameters are passed via kvm_s390_irq:: 4043 4044 struct kvm_s390_irq { 4045 __u64 type; 4046 union { 4047 struct kvm_s390_io_info io; 4048 struct kvm_s390_ext_info ext; 4049 struct kvm_s390_pgm_info pgm; 4050 struct kvm_s390_emerg_info emerg; 4051 struct kvm_s390_extcall_info extcall; 4052 struct kvm_s390_prefix_info prefix; 4053 struct kvm_s390_stop_info stop; 4054 struct kvm_s390_mchk_info mchk; 4055 char reserved[64]; 4056 } u; 4057 }; 4058 4059type can be one of the following: 4060 4061- KVM_S390_SIGP_STOP - sigp stop; parameter in .stop 4062- KVM_S390_PROGRAM_INT - program check; parameters in .pgm 4063- KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix 4064- KVM_S390_RESTART - restart; no parameters 4065- KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters 4066- KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters 4067- KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg 4068- KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall 4069- KVM_S390_MCHK - machine check interrupt; parameters in .mchk 4070 4071This is an asynchronous vcpu ioctl and can be invoked from any thread. 4072 40734.94 KVM_S390_GET_IRQ_STATE 4074--------------------------- 4075 4076:Capability: KVM_CAP_S390_IRQ_STATE 4077:Architectures: s390 4078:Type: vcpu ioctl 4079:Parameters: struct kvm_s390_irq_state (out) 4080:Returns: >= number of bytes copied into buffer, 4081 -EINVAL if buffer size is 0, 4082 -ENOBUFS if buffer size is too small to fit all pending interrupts, 4083 -EFAULT if the buffer address was invalid 4084 4085This ioctl allows userspace to retrieve the complete state of all currently 4086pending interrupts in a single buffer. Use cases include migration 4087and introspection. The parameter structure contains the address of a 4088userspace buffer and its length:: 4089 4090 struct kvm_s390_irq_state { 4091 __u64 buf; 4092 __u32 flags; /* will stay unused for compatibility reasons */ 4093 __u32 len; 4094 __u32 reserved[4]; /* will stay unused for compatibility reasons */ 4095 }; 4096 4097Userspace passes in the above struct and for each pending interrupt a 4098struct kvm_s390_irq is copied to the provided buffer. 4099 4100The structure contains a flags and a reserved field for future extensions. As 4101the kernel never checked for flags == 0 and QEMU never pre-zeroed flags and 4102reserved, these fields can not be used in the future without breaking 4103compatibility. 4104 4105If -ENOBUFS is returned the buffer provided was too small and userspace 4106may retry with a bigger buffer. 4107 41084.95 KVM_S390_SET_IRQ_STATE 4109--------------------------- 4110 4111:Capability: KVM_CAP_S390_IRQ_STATE 4112:Architectures: s390 4113:Type: vcpu ioctl 4114:Parameters: struct kvm_s390_irq_state (in) 4115:Returns: 0 on success, 4116 -EFAULT if the buffer address was invalid, 4117 -EINVAL for an invalid buffer length (see below), 4118 -EBUSY if there were already interrupts pending, 4119 errors occurring when actually injecting the 4120 interrupt. See KVM_S390_IRQ. 4121 4122This ioctl allows userspace to set the complete state of all cpu-local 4123interrupts currently pending for the vcpu. It is intended for restoring 4124interrupt state after a migration. The input parameter is a userspace buffer 4125containing a struct kvm_s390_irq_state:: 4126 4127 struct kvm_s390_irq_state { 4128 __u64 buf; 4129 __u32 flags; /* will stay unused for compatibility reasons */ 4130 __u32 len; 4131 __u32 reserved[4]; /* will stay unused for compatibility reasons */ 4132 }; 4133 4134The restrictions for flags and reserved apply as well. 4135(see KVM_S390_GET_IRQ_STATE) 4136 4137The userspace memory referenced by buf contains a struct kvm_s390_irq 4138for each interrupt to be injected into the guest. 4139If one of the interrupts could not be injected for some reason the 4140ioctl aborts. 4141 4142len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0 4143and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq), 4144which is the maximum number of possibly pending cpu-local interrupts. 4145 41464.96 KVM_SMI 4147------------ 4148 4149:Capability: KVM_CAP_X86_SMM 4150:Architectures: x86 4151:Type: vcpu ioctl 4152:Parameters: none 4153:Returns: 0 on success, -1 on error 4154 4155Queues an SMI on the thread's vcpu. 4156 41574.97 KVM_X86_SET_MSR_FILTER 4158---------------------------- 4159 4160:Capability: KVM_CAP_X86_MSR_FILTER 4161:Architectures: x86 4162:Type: vm ioctl 4163:Parameters: struct kvm_msr_filter 4164:Returns: 0 on success, < 0 on error 4165 4166:: 4167 4168 struct kvm_msr_filter_range { 4169 #define KVM_MSR_FILTER_READ (1 << 0) 4170 #define KVM_MSR_FILTER_WRITE (1 << 1) 4171 __u32 flags; 4172 __u32 nmsrs; /* number of msrs in bitmap */ 4173 __u32 base; /* MSR index the bitmap starts at */ 4174 __u8 *bitmap; /* a 1 bit allows the operations in flags, 0 denies */ 4175 }; 4176 4177 #define KVM_MSR_FILTER_MAX_RANGES 16 4178 struct kvm_msr_filter { 4179 #define KVM_MSR_FILTER_DEFAULT_ALLOW (0 << 0) 4180 #define KVM_MSR_FILTER_DEFAULT_DENY (1 << 0) 4181 __u32 flags; 4182 struct kvm_msr_filter_range ranges[KVM_MSR_FILTER_MAX_RANGES]; 4183 }; 4184 4185flags values for ``struct kvm_msr_filter_range``: 4186 4187``KVM_MSR_FILTER_READ`` 4188 4189 Filter read accesses to MSRs using the given bitmap. A 0 in the bitmap 4190 indicates that read accesses should be denied, while a 1 indicates that 4191 a read for a particular MSR should be allowed regardless of the default 4192 filter action. 4193 4194``KVM_MSR_FILTER_WRITE`` 4195 4196 Filter write accesses to MSRs using the given bitmap. A 0 in the bitmap 4197 indicates that write accesses should be denied, while a 1 indicates that 4198 a write for a particular MSR should be allowed regardless of the default 4199 filter action. 4200 4201flags values for ``struct kvm_msr_filter``: 4202 4203``KVM_MSR_FILTER_DEFAULT_ALLOW`` 4204 4205 If no filter range matches an MSR index that is getting accessed, KVM will 4206 allow accesses to all MSRs by default. 4207 4208``KVM_MSR_FILTER_DEFAULT_DENY`` 4209 4210 If no filter range matches an MSR index that is getting accessed, KVM will 4211 deny accesses to all MSRs by default. 4212 4213This ioctl allows userspace to define up to 16 bitmaps of MSR ranges to deny 4214guest MSR accesses that would normally be allowed by KVM. If an MSR is not 4215covered by a specific range, the "default" filtering behavior applies. Each 4216bitmap range covers MSRs from [base .. base+nmsrs). 4217 4218If an MSR access is denied by userspace, the resulting KVM behavior depends on 4219whether or not KVM_CAP_X86_USER_SPACE_MSR's KVM_MSR_EXIT_REASON_FILTER is 4220enabled. If KVM_MSR_EXIT_REASON_FILTER is enabled, KVM will exit to userspace 4221on denied accesses, i.e. userspace effectively intercepts the MSR access. If 4222KVM_MSR_EXIT_REASON_FILTER is not enabled, KVM will inject a #GP into the guest 4223on denied accesses. Note, if an MSR access is denied during emulation of MSR 4224load/stores during VMX transitions, KVM ignores KVM_MSR_EXIT_REASON_FILTER. 4225See the below warning for full details. 4226 4227If an MSR access is allowed by userspace, KVM will emulate and/or virtualize 4228the access in accordance with the vCPU model. Note, KVM may still ultimately 4229inject a #GP if an access is allowed by userspace, e.g. if KVM doesn't support 4230the MSR, or to follow architectural behavior for the MSR. 4231 4232By default, KVM operates in KVM_MSR_FILTER_DEFAULT_ALLOW mode with no MSR range 4233filters. 4234 4235Calling this ioctl with an empty set of ranges (all nmsrs == 0) disables MSR 4236filtering. In that mode, ``KVM_MSR_FILTER_DEFAULT_DENY`` is invalid and causes 4237an error. 4238 4239.. warning:: 4240 MSR accesses that are side effects of instruction execution (emulated or 4241 native) are not filtered as hardware does not honor MSR bitmaps outside of 4242 RDMSR and WRMSR, and KVM mimics that behavior when emulating instructions 4243 to avoid pointless divergence from hardware. E.g. RDPID reads MSR_TSC_AUX, 4244 SYSENTER reads the SYSENTER MSRs, etc. 4245 4246 MSRs that are loaded/stored via dedicated VMCS fields are not filtered as 4247 part of VM-Enter/VM-Exit emulation. 4248 4249 MSRs that are loaded/store via VMX's load/store lists _are_ filtered as part 4250 of VM-Enter/VM-Exit emulation. If an MSR access is denied on VM-Enter, KVM 4251 synthesizes a consistency check VM-Exit(EXIT_REASON_MSR_LOAD_FAIL). If an 4252 MSR access is denied on VM-Exit, KVM synthesizes a VM-Abort. In short, KVM 4253 extends Intel's architectural list of MSRs that cannot be loaded/saved via 4254 the VM-Enter/VM-Exit MSR list. It is platform owner's responsibility to 4255 to communicate any such restrictions to their end users. 4256 4257 x2APIC MSR accesses cannot be filtered (KVM silently ignores filters that 4258 cover any x2APIC MSRs). 4259 4260Note, invoking this ioctl while a vCPU is running is inherently racy. However, 4261KVM does guarantee that vCPUs will see either the previous filter or the new 4262filter, e.g. MSRs with identical settings in both the old and new filter will 4263have deterministic behavior. 4264 4265Similarly, if userspace wishes to intercept on denied accesses, 4266KVM_MSR_EXIT_REASON_FILTER must be enabled before activating any filters, and 4267left enabled until after all filters are deactivated. Failure to do so may 4268result in KVM injecting a #GP instead of exiting to userspace. 4269 42704.98 KVM_CREATE_SPAPR_TCE_64 4271---------------------------- 4272 4273:Capability: KVM_CAP_SPAPR_TCE_64 4274:Architectures: powerpc 4275:Type: vm ioctl 4276:Parameters: struct kvm_create_spapr_tce_64 (in) 4277:Returns: file descriptor for manipulating the created TCE table 4278 4279This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit 4280windows, described in 4.62 KVM_CREATE_SPAPR_TCE 4281 4282This capability uses extended struct in ioctl interface:: 4283 4284 /* for KVM_CAP_SPAPR_TCE_64 */ 4285 struct kvm_create_spapr_tce_64 { 4286 __u64 liobn; 4287 __u32 page_shift; 4288 __u32 flags; 4289 __u64 offset; /* in pages */ 4290 __u64 size; /* in pages */ 4291 }; 4292 4293The aim of extension is to support an additional bigger DMA window with 4294a variable page size. 4295KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and 4296a bus offset of the corresponding DMA window, @size and @offset are numbers 4297of IOMMU pages. 4298 4299@flags are not used at the moment. 4300 4301The rest of functionality is identical to KVM_CREATE_SPAPR_TCE. 4302 43034.99 KVM_REINJECT_CONTROL 4304------------------------- 4305 4306:Capability: KVM_CAP_REINJECT_CONTROL 4307:Architectures: x86 4308:Type: vm ioctl 4309:Parameters: struct kvm_reinject_control (in) 4310:Returns: 0 on success, 4311 -EFAULT if struct kvm_reinject_control cannot be read, 4312 -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier. 4313 4314i8254 (PIT) has two modes, reinject and !reinject. The default is reinject, 4315where KVM queues elapsed i8254 ticks and monitors completion of interrupt from 4316vector(s) that i8254 injects. Reinject mode dequeues a tick and injects its 4317interrupt whenever there isn't a pending interrupt from i8254. 4318!reinject mode injects an interrupt as soon as a tick arrives. 4319 4320:: 4321 4322 struct kvm_reinject_control { 4323 __u8 pit_reinject; 4324 __u8 reserved[31]; 4325 }; 4326 4327pit_reinject = 0 (!reinject mode) is recommended, unless running an old 4328operating system that uses the PIT for timing (e.g. Linux 2.4.x). 4329 43304.100 KVM_PPC_CONFIGURE_V3_MMU 4331------------------------------ 4332 4333:Capability: KVM_CAP_PPC_MMU_RADIX or KVM_CAP_PPC_MMU_HASH_V3 4334:Architectures: ppc 4335:Type: vm ioctl 4336:Parameters: struct kvm_ppc_mmuv3_cfg (in) 4337:Returns: 0 on success, 4338 -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read, 4339 -EINVAL if the configuration is invalid 4340 4341This ioctl controls whether the guest will use radix or HPT (hashed 4342page table) translation, and sets the pointer to the process table for 4343the guest. 4344 4345:: 4346 4347 struct kvm_ppc_mmuv3_cfg { 4348 __u64 flags; 4349 __u64 process_table; 4350 }; 4351 4352There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and 4353KVM_PPC_MMUV3_GTSE. KVM_PPC_MMUV3_RADIX, if set, configures the guest 4354to use radix tree translation, and if clear, to use HPT translation. 4355KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest 4356to be able to use the global TLB and SLB invalidation instructions; 4357if clear, the guest may not use these instructions. 4358 4359The process_table field specifies the address and size of the guest 4360process table, which is in the guest's space. This field is formatted 4361as the second doubleword of the partition table entry, as defined in 4362the Power ISA V3.00, Book III section 5.7.6.1. 4363 43644.101 KVM_PPC_GET_RMMU_INFO 4365--------------------------- 4366 4367:Capability: KVM_CAP_PPC_MMU_RADIX 4368:Architectures: ppc 4369:Type: vm ioctl 4370:Parameters: struct kvm_ppc_rmmu_info (out) 4371:Returns: 0 on success, 4372 -EFAULT if struct kvm_ppc_rmmu_info cannot be written, 4373 -EINVAL if no useful information can be returned 4374 4375This ioctl returns a structure containing two things: (a) a list 4376containing supported radix tree geometries, and (b) a list that maps 4377page sizes to put in the "AP" (actual page size) field for the tlbie 4378(TLB invalidate entry) instruction. 4379 4380:: 4381 4382 struct kvm_ppc_rmmu_info { 4383 struct kvm_ppc_radix_geom { 4384 __u8 page_shift; 4385 __u8 level_bits[4]; 4386 __u8 pad[3]; 4387 } geometries[8]; 4388 __u32 ap_encodings[8]; 4389 }; 4390 4391The geometries[] field gives up to 8 supported geometries for the 4392radix page table, in terms of the log base 2 of the smallest page 4393size, and the number of bits indexed at each level of the tree, from 4394the PTE level up to the PGD level in that order. Any unused entries 4395will have 0 in the page_shift field. 4396 4397The ap_encodings gives the supported page sizes and their AP field 4398encodings, encoded with the AP value in the top 3 bits and the log 4399base 2 of the page size in the bottom 6 bits. 4400 44014.102 KVM_PPC_RESIZE_HPT_PREPARE 4402-------------------------------- 4403 4404:Capability: KVM_CAP_SPAPR_RESIZE_HPT 4405:Architectures: powerpc 4406:Type: vm ioctl 4407:Parameters: struct kvm_ppc_resize_hpt (in) 4408:Returns: 0 on successful completion, 4409 >0 if a new HPT is being prepared, the value is an estimated 4410 number of milliseconds until preparation is complete, 4411 -EFAULT if struct kvm_reinject_control cannot be read, 4412 -EINVAL if the supplied shift or flags are invalid, 4413 -ENOMEM if unable to allocate the new HPT, 4414 4415Used to implement the PAPR extension for runtime resizing of a guest's 4416Hashed Page Table (HPT). Specifically this starts, stops or monitors 4417the preparation of a new potential HPT for the guest, essentially 4418implementing the H_RESIZE_HPT_PREPARE hypercall. 4419 4420:: 4421 4422 struct kvm_ppc_resize_hpt { 4423 __u64 flags; 4424 __u32 shift; 4425 __u32 pad; 4426 }; 4427 4428If called with shift > 0 when there is no pending HPT for the guest, 4429this begins preparation of a new pending HPT of size 2^(shift) bytes. 4430It then returns a positive integer with the estimated number of 4431milliseconds until preparation is complete. 4432 4433If called when there is a pending HPT whose size does not match that 4434requested in the parameters, discards the existing pending HPT and 4435creates a new one as above. 4436 4437If called when there is a pending HPT of the size requested, will: 4438 4439 * If preparation of the pending HPT is already complete, return 0 4440 * If preparation of the pending HPT has failed, return an error 4441 code, then discard the pending HPT. 4442 * If preparation of the pending HPT is still in progress, return an 4443 estimated number of milliseconds until preparation is complete. 4444 4445If called with shift == 0, discards any currently pending HPT and 4446returns 0 (i.e. cancels any in-progress preparation). 4447 4448flags is reserved for future expansion, currently setting any bits in 4449flags will result in an -EINVAL. 4450 4451Normally this will be called repeatedly with the same parameters until 4452it returns <= 0. The first call will initiate preparation, subsequent 4453ones will monitor preparation until it completes or fails. 4454 44554.103 KVM_PPC_RESIZE_HPT_COMMIT 4456------------------------------- 4457 4458:Capability: KVM_CAP_SPAPR_RESIZE_HPT 4459:Architectures: powerpc 4460:Type: vm ioctl 4461:Parameters: struct kvm_ppc_resize_hpt (in) 4462:Returns: 0 on successful completion, 4463 -EFAULT if struct kvm_reinject_control cannot be read, 4464 -EINVAL if the supplied shift or flags are invalid, 4465 -ENXIO is there is no pending HPT, or the pending HPT doesn't 4466 have the requested size, 4467 -EBUSY if the pending HPT is not fully prepared, 4468 -ENOSPC if there was a hash collision when moving existing 4469 HPT entries to the new HPT, 4470 -EIO on other error conditions 4471 4472Used to implement the PAPR extension for runtime resizing of a guest's 4473Hashed Page Table (HPT). Specifically this requests that the guest be 4474transferred to working with the new HPT, essentially implementing the 4475H_RESIZE_HPT_COMMIT hypercall. 4476 4477:: 4478 4479 struct kvm_ppc_resize_hpt { 4480 __u64 flags; 4481 __u32 shift; 4482 __u32 pad; 4483 }; 4484 4485This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has 4486returned 0 with the same parameters. In other cases 4487KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or 4488-EBUSY, though others may be possible if the preparation was started, 4489but failed). 4490 4491This will have undefined effects on the guest if it has not already 4492placed itself in a quiescent state where no vcpu will make MMU enabled 4493memory accesses. 4494 4495On successful completion, the pending HPT will become the guest's active 4496HPT and the previous HPT will be discarded. 4497 4498On failure, the guest will still be operating on its previous HPT. 4499 45004.104 KVM_X86_GET_MCE_CAP_SUPPORTED 4501----------------------------------- 4502 4503:Capability: KVM_CAP_MCE 4504:Architectures: x86 4505:Type: system ioctl 4506:Parameters: u64 mce_cap (out) 4507:Returns: 0 on success, -1 on error 4508 4509Returns supported MCE capabilities. The u64 mce_cap parameter 4510has the same format as the MSR_IA32_MCG_CAP register. Supported 4511capabilities will have the corresponding bits set. 4512 45134.105 KVM_X86_SETUP_MCE 4514----------------------- 4515 4516:Capability: KVM_CAP_MCE 4517:Architectures: x86 4518:Type: vcpu ioctl 4519:Parameters: u64 mcg_cap (in) 4520:Returns: 0 on success, 4521 -EFAULT if u64 mcg_cap cannot be read, 4522 -EINVAL if the requested number of banks is invalid, 4523 -EINVAL if requested MCE capability is not supported. 4524 4525Initializes MCE support for use. The u64 mcg_cap parameter 4526has the same format as the MSR_IA32_MCG_CAP register and 4527specifies which capabilities should be enabled. The maximum 4528supported number of error-reporting banks can be retrieved when 4529checking for KVM_CAP_MCE. The supported capabilities can be 4530retrieved with KVM_X86_GET_MCE_CAP_SUPPORTED. 4531 45324.106 KVM_X86_SET_MCE 4533--------------------- 4534 4535:Capability: KVM_CAP_MCE 4536:Architectures: x86 4537:Type: vcpu ioctl 4538:Parameters: struct kvm_x86_mce (in) 4539:Returns: 0 on success, 4540 -EFAULT if struct kvm_x86_mce cannot be read, 4541 -EINVAL if the bank number is invalid, 4542 -EINVAL if VAL bit is not set in status field. 4543 4544Inject a machine check error (MCE) into the guest. The input 4545parameter is:: 4546 4547 struct kvm_x86_mce { 4548 __u64 status; 4549 __u64 addr; 4550 __u64 misc; 4551 __u64 mcg_status; 4552 __u8 bank; 4553 __u8 pad1[7]; 4554 __u64 pad2[3]; 4555 }; 4556 4557If the MCE being reported is an uncorrected error, KVM will 4558inject it as an MCE exception into the guest. If the guest 4559MCG_STATUS register reports that an MCE is in progress, KVM 4560causes an KVM_EXIT_SHUTDOWN vmexit. 4561 4562Otherwise, if the MCE is a corrected error, KVM will just 4563store it in the corresponding bank (provided this bank is 4564not holding a previously reported uncorrected error). 4565 45664.107 KVM_S390_GET_CMMA_BITS 4567---------------------------- 4568 4569:Capability: KVM_CAP_S390_CMMA_MIGRATION 4570:Architectures: s390 4571:Type: vm ioctl 4572:Parameters: struct kvm_s390_cmma_log (in, out) 4573:Returns: 0 on success, a negative value on error 4574 4575Errors: 4576 4577 ====== ============================================================= 4578 ENOMEM not enough memory can be allocated to complete the task 4579 ENXIO if CMMA is not enabled 4580 EINVAL if KVM_S390_CMMA_PEEK is not set but migration mode was not enabled 4581 EINVAL if KVM_S390_CMMA_PEEK is not set but dirty tracking has been 4582 disabled (and thus migration mode was automatically disabled) 4583 EFAULT if the userspace address is invalid or if no page table is 4584 present for the addresses (e.g. when using hugepages). 4585 ====== ============================================================= 4586 4587This ioctl is used to get the values of the CMMA bits on the s390 4588architecture. It is meant to be used in two scenarios: 4589 4590- During live migration to save the CMMA values. Live migration needs 4591 to be enabled via the KVM_REQ_START_MIGRATION VM property. 4592- To non-destructively peek at the CMMA values, with the flag 4593 KVM_S390_CMMA_PEEK set. 4594 4595The ioctl takes parameters via the kvm_s390_cmma_log struct. The desired 4596values are written to a buffer whose location is indicated via the "values" 4597member in the kvm_s390_cmma_log struct. The values in the input struct are 4598also updated as needed. 4599 4600Each CMMA value takes up one byte. 4601 4602:: 4603 4604 struct kvm_s390_cmma_log { 4605 __u64 start_gfn; 4606 __u32 count; 4607 __u32 flags; 4608 union { 4609 __u64 remaining; 4610 __u64 mask; 4611 }; 4612 __u64 values; 4613 }; 4614 4615start_gfn is the number of the first guest frame whose CMMA values are 4616to be retrieved, 4617 4618count is the length of the buffer in bytes, 4619 4620values points to the buffer where the result will be written to. 4621 4622If count is greater than KVM_S390_SKEYS_MAX, then it is considered to be 4623KVM_S390_SKEYS_MAX. KVM_S390_SKEYS_MAX is re-used for consistency with 4624other ioctls. 4625 4626The result is written in the buffer pointed to by the field values, and 4627the values of the input parameter are updated as follows. 4628 4629Depending on the flags, different actions are performed. The only 4630supported flag so far is KVM_S390_CMMA_PEEK. 4631 4632The default behaviour if KVM_S390_CMMA_PEEK is not set is: 4633start_gfn will indicate the first page frame whose CMMA bits were dirty. 4634It is not necessarily the same as the one passed as input, as clean pages 4635are skipped. 4636 4637count will indicate the number of bytes actually written in the buffer. 4638It can (and very often will) be smaller than the input value, since the 4639buffer is only filled until 16 bytes of clean values are found (which 4640are then not copied in the buffer). Since a CMMA migration block needs 4641the base address and the length, for a total of 16 bytes, we will send 4642back some clean data if there is some dirty data afterwards, as long as 4643the size of the clean data does not exceed the size of the header. This 4644allows to minimize the amount of data to be saved or transferred over 4645the network at the expense of more roundtrips to userspace. The next 4646invocation of the ioctl will skip over all the clean values, saving 4647potentially more than just the 16 bytes we found. 4648 4649If KVM_S390_CMMA_PEEK is set: 4650the existing storage attributes are read even when not in migration 4651mode, and no other action is performed; 4652 4653the output start_gfn will be equal to the input start_gfn, 4654 4655the output count will be equal to the input count, except if the end of 4656memory has been reached. 4657 4658In both cases: 4659the field "remaining" will indicate the total number of dirty CMMA values 4660still remaining, or 0 if KVM_S390_CMMA_PEEK is set and migration mode is 4661not enabled. 4662 4663mask is unused. 4664 4665values points to the userspace buffer where the result will be stored. 4666 46674.108 KVM_S390_SET_CMMA_BITS 4668---------------------------- 4669 4670:Capability: KVM_CAP_S390_CMMA_MIGRATION 4671:Architectures: s390 4672:Type: vm ioctl 4673:Parameters: struct kvm_s390_cmma_log (in) 4674:Returns: 0 on success, a negative value on error 4675 4676This ioctl is used to set the values of the CMMA bits on the s390 4677architecture. It is meant to be used during live migration to restore 4678the CMMA values, but there are no restrictions on its use. 4679The ioctl takes parameters via the kvm_s390_cmma_values struct. 4680Each CMMA value takes up one byte. 4681 4682:: 4683 4684 struct kvm_s390_cmma_log { 4685 __u64 start_gfn; 4686 __u32 count; 4687 __u32 flags; 4688 union { 4689 __u64 remaining; 4690 __u64 mask; 4691 }; 4692 __u64 values; 4693 }; 4694 4695start_gfn indicates the starting guest frame number, 4696 4697count indicates how many values are to be considered in the buffer, 4698 4699flags is not used and must be 0. 4700 4701mask indicates which PGSTE bits are to be considered. 4702 4703remaining is not used. 4704 4705values points to the buffer in userspace where to store the values. 4706 4707This ioctl can fail with -ENOMEM if not enough memory can be allocated to 4708complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if 4709the count field is too large (e.g. more than KVM_S390_CMMA_SIZE_MAX) or 4710if the flags field was not 0, with -EFAULT if the userspace address is 4711invalid, if invalid pages are written to (e.g. after the end of memory) 4712or if no page table is present for the addresses (e.g. when using 4713hugepages). 4714 47154.109 KVM_PPC_GET_CPU_CHAR 4716-------------------------- 4717 4718:Capability: KVM_CAP_PPC_GET_CPU_CHAR 4719:Architectures: powerpc 4720:Type: vm ioctl 4721:Parameters: struct kvm_ppc_cpu_char (out) 4722:Returns: 0 on successful completion, 4723 -EFAULT if struct kvm_ppc_cpu_char cannot be written 4724 4725This ioctl gives userspace information about certain characteristics 4726of the CPU relating to speculative execution of instructions and 4727possible information leakage resulting from speculative execution (see 4728CVE-2017-5715, CVE-2017-5753 and CVE-2017-5754). The information is 4729returned in struct kvm_ppc_cpu_char, which looks like this:: 4730 4731 struct kvm_ppc_cpu_char { 4732 __u64 character; /* characteristics of the CPU */ 4733 __u64 behaviour; /* recommended software behaviour */ 4734 __u64 character_mask; /* valid bits in character */ 4735 __u64 behaviour_mask; /* valid bits in behaviour */ 4736 }; 4737 4738For extensibility, the character_mask and behaviour_mask fields 4739indicate which bits of character and behaviour have been filled in by 4740the kernel. If the set of defined bits is extended in future then 4741userspace will be able to tell whether it is running on a kernel that 4742knows about the new bits. 4743 4744The character field describes attributes of the CPU which can help 4745with preventing inadvertent information disclosure - specifically, 4746whether there is an instruction to flash-invalidate the L1 data cache 4747(ori 30,30,0 or mtspr SPRN_TRIG2,rN), whether the L1 data cache is set 4748to a mode where entries can only be used by the thread that created 4749them, whether the bcctr[l] instruction prevents speculation, and 4750whether a speculation barrier instruction (ori 31,31,0) is provided. 4751 4752The behaviour field describes actions that software should take to 4753prevent inadvertent information disclosure, and thus describes which 4754vulnerabilities the hardware is subject to; specifically whether the 4755L1 data cache should be flushed when returning to user mode from the 4756kernel, and whether a speculation barrier should be placed between an 4757array bounds check and the array access. 4758 4759These fields use the same bit definitions as the new 4760H_GET_CPU_CHARACTERISTICS hypercall. 4761 47624.110 KVM_MEMORY_ENCRYPT_OP 4763--------------------------- 4764 4765:Capability: basic 4766:Architectures: x86 4767:Type: vm 4768:Parameters: an opaque platform specific structure (in/out) 4769:Returns: 0 on success; -1 on error 4770 4771If the platform supports creating encrypted VMs then this ioctl can be used 4772for issuing platform-specific memory encryption commands to manage those 4773encrypted VMs. 4774 4775Currently, this ioctl is used for issuing Secure Encrypted Virtualization 4776(SEV) commands on AMD Processors. The SEV commands are defined in 4777Documentation/virt/kvm/x86/amd-memory-encryption.rst. 4778 47794.111 KVM_MEMORY_ENCRYPT_REG_REGION 4780----------------------------------- 4781 4782:Capability: basic 4783:Architectures: x86 4784:Type: system 4785:Parameters: struct kvm_enc_region (in) 4786:Returns: 0 on success; -1 on error 4787 4788This ioctl can be used to register a guest memory region which may 4789contain encrypted data (e.g. guest RAM, SMRAM etc). 4790 4791It is used in the SEV-enabled guest. When encryption is enabled, a guest 4792memory region may contain encrypted data. The SEV memory encryption 4793engine uses a tweak such that two identical plaintext pages, each at 4794different locations will have differing ciphertexts. So swapping or 4795moving ciphertext of those pages will not result in plaintext being 4796swapped. So relocating (or migrating) physical backing pages for the SEV 4797guest will require some additional steps. 4798 4799Note: The current SEV key management spec does not provide commands to 4800swap or migrate (move) ciphertext pages. Hence, for now we pin the guest 4801memory region registered with the ioctl. 4802 48034.112 KVM_MEMORY_ENCRYPT_UNREG_REGION 4804------------------------------------- 4805 4806:Capability: basic 4807:Architectures: x86 4808:Type: system 4809:Parameters: struct kvm_enc_region (in) 4810:Returns: 0 on success; -1 on error 4811 4812This ioctl can be used to unregister the guest memory region registered 4813with KVM_MEMORY_ENCRYPT_REG_REGION ioctl above. 4814 48154.113 KVM_HYPERV_EVENTFD 4816------------------------ 4817 4818:Capability: KVM_CAP_HYPERV_EVENTFD 4819:Architectures: x86 4820:Type: vm ioctl 4821:Parameters: struct kvm_hyperv_eventfd (in) 4822 4823This ioctl (un)registers an eventfd to receive notifications from the guest on 4824the specified Hyper-V connection id through the SIGNAL_EVENT hypercall, without 4825causing a user exit. SIGNAL_EVENT hypercall with non-zero event flag number 4826(bits 24-31) still triggers a KVM_EXIT_HYPERV_HCALL user exit. 4827 4828:: 4829 4830 struct kvm_hyperv_eventfd { 4831 __u32 conn_id; 4832 __s32 fd; 4833 __u32 flags; 4834 __u32 padding[3]; 4835 }; 4836 4837The conn_id field should fit within 24 bits:: 4838 4839 #define KVM_HYPERV_CONN_ID_MASK 0x00ffffff 4840 4841The acceptable values for the flags field are:: 4842 4843 #define KVM_HYPERV_EVENTFD_DEASSIGN (1 << 0) 4844 4845:Returns: 0 on success, 4846 -EINVAL if conn_id or flags is outside the allowed range, 4847 -ENOENT on deassign if the conn_id isn't registered, 4848 -EEXIST on assign if the conn_id is already registered 4849 48504.114 KVM_GET_NESTED_STATE 4851-------------------------- 4852 4853:Capability: KVM_CAP_NESTED_STATE 4854:Architectures: x86 4855:Type: vcpu ioctl 4856:Parameters: struct kvm_nested_state (in/out) 4857:Returns: 0 on success, -1 on error 4858 4859Errors: 4860 4861 ===== ============================================================= 4862 E2BIG the total state size exceeds the value of 'size' specified by 4863 the user; the size required will be written into size. 4864 ===== ============================================================= 4865 4866:: 4867 4868 struct kvm_nested_state { 4869 __u16 flags; 4870 __u16 format; 4871 __u32 size; 4872 4873 union { 4874 struct kvm_vmx_nested_state_hdr vmx; 4875 struct kvm_svm_nested_state_hdr svm; 4876 4877 /* Pad the header to 128 bytes. */ 4878 __u8 pad[120]; 4879 } hdr; 4880 4881 union { 4882 struct kvm_vmx_nested_state_data vmx[0]; 4883 struct kvm_svm_nested_state_data svm[0]; 4884 } data; 4885 }; 4886 4887 #define KVM_STATE_NESTED_GUEST_MODE 0x00000001 4888 #define KVM_STATE_NESTED_RUN_PENDING 0x00000002 4889 #define KVM_STATE_NESTED_EVMCS 0x00000004 4890 4891 #define KVM_STATE_NESTED_FORMAT_VMX 0 4892 #define KVM_STATE_NESTED_FORMAT_SVM 1 4893 4894 #define KVM_STATE_NESTED_VMX_VMCS_SIZE 0x1000 4895 4896 #define KVM_STATE_NESTED_VMX_SMM_GUEST_MODE 0x00000001 4897 #define KVM_STATE_NESTED_VMX_SMM_VMXON 0x00000002 4898 4899 #define KVM_STATE_VMX_PREEMPTION_TIMER_DEADLINE 0x00000001 4900 4901 struct kvm_vmx_nested_state_hdr { 4902 __u64 vmxon_pa; 4903 __u64 vmcs12_pa; 4904 4905 struct { 4906 __u16 flags; 4907 } smm; 4908 4909 __u32 flags; 4910 __u64 preemption_timer_deadline; 4911 }; 4912 4913 struct kvm_vmx_nested_state_data { 4914 __u8 vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE]; 4915 __u8 shadow_vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE]; 4916 }; 4917 4918This ioctl copies the vcpu's nested virtualization state from the kernel to 4919userspace. 4920 4921The maximum size of the state can be retrieved by passing KVM_CAP_NESTED_STATE 4922to the KVM_CHECK_EXTENSION ioctl(). 4923 49244.115 KVM_SET_NESTED_STATE 4925-------------------------- 4926 4927:Capability: KVM_CAP_NESTED_STATE 4928:Architectures: x86 4929:Type: vcpu ioctl 4930:Parameters: struct kvm_nested_state (in) 4931:Returns: 0 on success, -1 on error 4932 4933This copies the vcpu's kvm_nested_state struct from userspace to the kernel. 4934For the definition of struct kvm_nested_state, see KVM_GET_NESTED_STATE. 4935 49364.116 KVM_(UN)REGISTER_COALESCED_MMIO 4937------------------------------------- 4938 4939:Capability: KVM_CAP_COALESCED_MMIO (for coalesced mmio) 4940 KVM_CAP_COALESCED_PIO (for coalesced pio) 4941:Architectures: all 4942:Type: vm ioctl 4943:Parameters: struct kvm_coalesced_mmio_zone 4944:Returns: 0 on success, < 0 on error 4945 4946Coalesced I/O is a performance optimization that defers hardware 4947register write emulation so that userspace exits are avoided. It is 4948typically used to reduce the overhead of emulating frequently accessed 4949hardware registers. 4950 4951When a hardware register is configured for coalesced I/O, write accesses 4952do not exit to userspace and their value is recorded in a ring buffer 4953that is shared between kernel and userspace. 4954 4955Coalesced I/O is used if one or more write accesses to a hardware 4956register can be deferred until a read or a write to another hardware 4957register on the same device. This last access will cause a vmexit and 4958userspace will process accesses from the ring buffer before emulating 4959it. That will avoid exiting to userspace on repeated writes. 4960 4961Coalesced pio is based on coalesced mmio. There is little difference 4962between coalesced mmio and pio except that coalesced pio records accesses 4963to I/O ports. 4964 49654.117 KVM_CLEAR_DIRTY_LOG 4966------------------------- 4967 4968:Capability: KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 4969:Architectures: x86, arm64, mips 4970:Type: vm ioctl 4971:Parameters: struct kvm_clear_dirty_log (in) 4972:Returns: 0 on success, -1 on error 4973 4974:: 4975 4976 /* for KVM_CLEAR_DIRTY_LOG */ 4977 struct kvm_clear_dirty_log { 4978 __u32 slot; 4979 __u32 num_pages; 4980 __u64 first_page; 4981 union { 4982 void __user *dirty_bitmap; /* one bit per page */ 4983 __u64 padding; 4984 }; 4985 }; 4986 4987The ioctl clears the dirty status of pages in a memory slot, according to 4988the bitmap that is passed in struct kvm_clear_dirty_log's dirty_bitmap 4989field. Bit 0 of the bitmap corresponds to page "first_page" in the 4990memory slot, and num_pages is the size in bits of the input bitmap. 4991first_page must be a multiple of 64; num_pages must also be a multiple of 499264 unless first_page + num_pages is the size of the memory slot. For each 4993bit that is set in the input bitmap, the corresponding page is marked "clean" 4994in KVM's dirty bitmap, and dirty tracking is re-enabled for that page 4995(for example via write-protection, or by clearing the dirty bit in 4996a page table entry). 4997 4998If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies 4999the address space for which you want to clear the dirty status. See 5000KVM_SET_USER_MEMORY_REGION for details on the usage of slot field. 5001 5002This ioctl is mostly useful when KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 5003is enabled; for more information, see the description of the capability. 5004However, it can always be used as long as KVM_CHECK_EXTENSION confirms 5005that KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is present. 5006 50074.118 KVM_GET_SUPPORTED_HV_CPUID 5008-------------------------------- 5009 5010:Capability: KVM_CAP_HYPERV_CPUID (vcpu), KVM_CAP_SYS_HYPERV_CPUID (system) 5011:Architectures: x86 5012:Type: system ioctl, vcpu ioctl 5013:Parameters: struct kvm_cpuid2 (in/out) 5014:Returns: 0 on success, -1 on error 5015 5016:: 5017 5018 struct kvm_cpuid2 { 5019 __u32 nent; 5020 __u32 padding; 5021 struct kvm_cpuid_entry2 entries[0]; 5022 }; 5023 5024 struct kvm_cpuid_entry2 { 5025 __u32 function; 5026 __u32 index; 5027 __u32 flags; 5028 __u32 eax; 5029 __u32 ebx; 5030 __u32 ecx; 5031 __u32 edx; 5032 __u32 padding[3]; 5033 }; 5034 5035This ioctl returns x86 cpuid features leaves related to Hyper-V emulation in 5036KVM. Userspace can use the information returned by this ioctl to construct 5037cpuid information presented to guests consuming Hyper-V enlightenments (e.g. 5038Windows or Hyper-V guests). 5039 5040CPUID feature leaves returned by this ioctl are defined by Hyper-V Top Level 5041Functional Specification (TLFS). These leaves can't be obtained with 5042KVM_GET_SUPPORTED_CPUID ioctl because some of them intersect with KVM feature 5043leaves (0x40000000, 0x40000001). 5044 5045Currently, the following list of CPUID leaves are returned: 5046 5047 - HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS 5048 - HYPERV_CPUID_INTERFACE 5049 - HYPERV_CPUID_VERSION 5050 - HYPERV_CPUID_FEATURES 5051 - HYPERV_CPUID_ENLIGHTMENT_INFO 5052 - HYPERV_CPUID_IMPLEMENT_LIMITS 5053 - HYPERV_CPUID_NESTED_FEATURES 5054 - HYPERV_CPUID_SYNDBG_VENDOR_AND_MAX_FUNCTIONS 5055 - HYPERV_CPUID_SYNDBG_INTERFACE 5056 - HYPERV_CPUID_SYNDBG_PLATFORM_CAPABILITIES 5057 5058Userspace invokes KVM_GET_SUPPORTED_HV_CPUID by passing a kvm_cpuid2 structure 5059with the 'nent' field indicating the number of entries in the variable-size 5060array 'entries'. If the number of entries is too low to describe all Hyper-V 5061feature leaves, an error (E2BIG) is returned. If the number is more or equal 5062to the number of Hyper-V feature leaves, the 'nent' field is adjusted to the 5063number of valid entries in the 'entries' array, which is then filled. 5064 5065'index' and 'flags' fields in 'struct kvm_cpuid_entry2' are currently reserved, 5066userspace should not expect to get any particular value there. 5067 5068Note, vcpu version of KVM_GET_SUPPORTED_HV_CPUID is currently deprecated. Unlike 5069system ioctl which exposes all supported feature bits unconditionally, vcpu 5070version has the following quirks: 5071 5072- HYPERV_CPUID_NESTED_FEATURES leaf and HV_X64_ENLIGHTENED_VMCS_RECOMMENDED 5073 feature bit are only exposed when Enlightened VMCS was previously enabled 5074 on the corresponding vCPU (KVM_CAP_HYPERV_ENLIGHTENED_VMCS). 5075- HV_STIMER_DIRECT_MODE_AVAILABLE bit is only exposed with in-kernel LAPIC. 5076 (presumes KVM_CREATE_IRQCHIP has already been called). 5077 50784.119 KVM_ARM_VCPU_FINALIZE 5079--------------------------- 5080 5081:Architectures: arm64 5082:Type: vcpu ioctl 5083:Parameters: int feature (in) 5084:Returns: 0 on success, -1 on error 5085 5086Errors: 5087 5088 ====== ============================================================== 5089 EPERM feature not enabled, needs configuration, or already finalized 5090 EINVAL feature unknown or not present 5091 ====== ============================================================== 5092 5093Recognised values for feature: 5094 5095 ===== =========================================== 5096 arm64 KVM_ARM_VCPU_SVE (requires KVM_CAP_ARM_SVE) 5097 ===== =========================================== 5098 5099Finalizes the configuration of the specified vcpu feature. 5100 5101The vcpu must already have been initialised, enabling the affected feature, by 5102means of a successful :ref:`KVM_ARM_VCPU_INIT <KVM_ARM_VCPU_INIT>` call with the 5103appropriate flag set in features[]. 5104 5105For affected vcpu features, this is a mandatory step that must be performed 5106before the vcpu is fully usable. 5107 5108Between KVM_ARM_VCPU_INIT and KVM_ARM_VCPU_FINALIZE, the feature may be 5109configured by use of ioctls such as KVM_SET_ONE_REG. The exact configuration 5110that should be performed and how to do it are feature-dependent. 5111 5112Other calls that depend on a particular feature being finalized, such as 5113KVM_RUN, KVM_GET_REG_LIST, KVM_GET_ONE_REG and KVM_SET_ONE_REG, will fail with 5114-EPERM unless the feature has already been finalized by means of a 5115KVM_ARM_VCPU_FINALIZE call. 5116 5117See KVM_ARM_VCPU_INIT for details of vcpu features that require finalization 5118using this ioctl. 5119 51204.120 KVM_SET_PMU_EVENT_FILTER 5121------------------------------ 5122 5123:Capability: KVM_CAP_PMU_EVENT_FILTER 5124:Architectures: x86 5125:Type: vm ioctl 5126:Parameters: struct kvm_pmu_event_filter (in) 5127:Returns: 0 on success, -1 on error 5128 5129Errors: 5130 5131 ====== ============================================================ 5132 EFAULT args[0] cannot be accessed 5133 EINVAL args[0] contains invalid data in the filter or filter events 5134 E2BIG nevents is too large 5135 EBUSY not enough memory to allocate the filter 5136 ====== ============================================================ 5137 5138:: 5139 5140 struct kvm_pmu_event_filter { 5141 __u32 action; 5142 __u32 nevents; 5143 __u32 fixed_counter_bitmap; 5144 __u32 flags; 5145 __u32 pad[4]; 5146 __u64 events[0]; 5147 }; 5148 5149This ioctl restricts the set of PMU events the guest can program by limiting 5150which event select and unit mask combinations are permitted. 5151 5152The argument holds a list of filter events which will be allowed or denied. 5153 5154Filter events only control general purpose counters; fixed purpose counters 5155are controlled by the fixed_counter_bitmap. 5156 5157Valid values for 'flags':: 5158 5159``0`` 5160 5161To use this mode, clear the 'flags' field. 5162 5163In this mode each event will contain an event select + unit mask. 5164 5165When the guest attempts to program the PMU the guest's event select + 5166unit mask is compared against the filter events to determine whether the 5167guest should have access. 5168 5169``KVM_PMU_EVENT_FLAG_MASKED_EVENTS`` 5170:Capability: KVM_CAP_PMU_EVENT_MASKED_EVENTS 5171 5172In this mode each filter event will contain an event select, mask, match, and 5173exclude value. To encode a masked event use:: 5174 5175 KVM_PMU_ENCODE_MASKED_ENTRY() 5176 5177An encoded event will follow this layout:: 5178 5179 Bits Description 5180 ---- ----------- 5181 7:0 event select (low bits) 5182 15:8 umask match 5183 31:16 unused 5184 35:32 event select (high bits) 5185 36:54 unused 5186 55 exclude bit 5187 63:56 umask mask 5188 5189When the guest attempts to program the PMU, these steps are followed in 5190determining if the guest should have access: 5191 5192 1. Match the event select from the guest against the filter events. 5193 2. If a match is found, match the guest's unit mask to the mask and match 5194 values of the included filter events. 5195 I.e. (unit mask & mask) == match && !exclude. 5196 3. If a match is found, match the guest's unit mask to the mask and match 5197 values of the excluded filter events. 5198 I.e. (unit mask & mask) == match && exclude. 5199 4. 5200 a. If an included match is found and an excluded match is not found, filter 5201 the event. 5202 b. For everything else, do not filter the event. 5203 5. 5204 a. If the event is filtered and it's an allow list, allow the guest to 5205 program the event. 5206 b. If the event is filtered and it's a deny list, do not allow the guest to 5207 program the event. 5208 5209When setting a new pmu event filter, -EINVAL will be returned if any of the 5210unused fields are set or if any of the high bits (35:32) in the event 5211select are set when called on Intel. 5212 5213Valid values for 'action':: 5214 5215 #define KVM_PMU_EVENT_ALLOW 0 5216 #define KVM_PMU_EVENT_DENY 1 5217 5218Via this API, KVM userspace can also control the behavior of the VM's fixed 5219counters (if any) by configuring the "action" and "fixed_counter_bitmap" fields. 5220 5221Specifically, KVM follows the following pseudo-code when determining whether to 5222allow the guest FixCtr[i] to count its pre-defined fixed event:: 5223 5224 FixCtr[i]_is_allowed = (action == ALLOW) && (bitmap & BIT(i)) || 5225 (action == DENY) && !(bitmap & BIT(i)); 5226 FixCtr[i]_is_denied = !FixCtr[i]_is_allowed; 5227 5228KVM always consumes fixed_counter_bitmap, it's userspace's responsibility to 5229ensure fixed_counter_bitmap is set correctly, e.g. if userspace wants to define 5230a filter that only affects general purpose counters. 5231 5232Note, the "events" field also applies to fixed counters' hardcoded event_select 5233and unit_mask values. "fixed_counter_bitmap" has higher priority than "events" 5234if there is a contradiction between the two. 5235 52364.121 KVM_PPC_SVM_OFF 5237--------------------- 5238 5239:Capability: basic 5240:Architectures: powerpc 5241:Type: vm ioctl 5242:Parameters: none 5243:Returns: 0 on successful completion, 5244 5245Errors: 5246 5247 ====== ================================================================ 5248 EINVAL if ultravisor failed to terminate the secure guest 5249 ENOMEM if hypervisor failed to allocate new radix page tables for guest 5250 ====== ================================================================ 5251 5252This ioctl is used to turn off the secure mode of the guest or transition 5253the guest from secure mode to normal mode. This is invoked when the guest 5254is reset. This has no effect if called for a normal guest. 5255 5256This ioctl issues an ultravisor call to terminate the secure guest, 5257unpins the VPA pages and releases all the device pages that are used to 5258track the secure pages by hypervisor. 5259 52604.122 KVM_S390_NORMAL_RESET 5261--------------------------- 5262 5263:Capability: KVM_CAP_S390_VCPU_RESETS 5264:Architectures: s390 5265:Type: vcpu ioctl 5266:Parameters: none 5267:Returns: 0 5268 5269This ioctl resets VCPU registers and control structures according to 5270the cpu reset definition in the POP (Principles Of Operation). 5271 52724.123 KVM_S390_INITIAL_RESET 5273---------------------------- 5274 5275:Capability: basic 5276:Architectures: s390 5277:Type: vcpu ioctl 5278:Parameters: none 5279:Returns: 0 5280 5281This ioctl resets VCPU registers and control structures according to 5282the initial cpu reset definition in the POP. However, the cpu is not 5283put into ESA mode. This reset is a superset of the normal reset. 5284 52854.124 KVM_S390_CLEAR_RESET 5286-------------------------- 5287 5288:Capability: KVM_CAP_S390_VCPU_RESETS 5289:Architectures: s390 5290:Type: vcpu ioctl 5291:Parameters: none 5292:Returns: 0 5293 5294This ioctl resets VCPU registers and control structures according to 5295the clear cpu reset definition in the POP. However, the cpu is not put 5296into ESA mode. This reset is a superset of the initial reset. 5297 5298 52994.125 KVM_S390_PV_COMMAND 5300------------------------- 5301 5302:Capability: KVM_CAP_S390_PROTECTED 5303:Architectures: s390 5304:Type: vm ioctl 5305:Parameters: struct kvm_pv_cmd 5306:Returns: 0 on success, < 0 on error 5307 5308:: 5309 5310 struct kvm_pv_cmd { 5311 __u32 cmd; /* Command to be executed */ 5312 __u16 rc; /* Ultravisor return code */ 5313 __u16 rrc; /* Ultravisor return reason code */ 5314 __u64 data; /* Data or address */ 5315 __u32 flags; /* flags for future extensions. Must be 0 for now */ 5316 __u32 reserved[3]; 5317 }; 5318 5319**Ultravisor return codes** 5320The Ultravisor return (reason) codes are provided by the kernel if a 5321Ultravisor call has been executed to achieve the results expected by 5322the command. Therefore they are independent of the IOCTL return 5323code. If KVM changes `rc`, its value will always be greater than 0 5324hence setting it to 0 before issuing a PV command is advised to be 5325able to detect a change of `rc`. 5326 5327**cmd values:** 5328 5329KVM_PV_ENABLE 5330 Allocate memory and register the VM with the Ultravisor, thereby 5331 donating memory to the Ultravisor that will become inaccessible to 5332 KVM. All existing CPUs are converted to protected ones. After this 5333 command has succeeded, any CPU added via hotplug will become 5334 protected during its creation as well. 5335 5336 Errors: 5337 5338 ===== ============================= 5339 EINTR an unmasked signal is pending 5340 ===== ============================= 5341 5342KVM_PV_DISABLE 5343 Deregister the VM from the Ultravisor and reclaim the memory that had 5344 been donated to the Ultravisor, making it usable by the kernel again. 5345 All registered VCPUs are converted back to non-protected ones. If a 5346 previous protected VM had been prepared for asynchronous teardown with 5347 KVM_PV_ASYNC_CLEANUP_PREPARE and not subsequently torn down with 5348 KVM_PV_ASYNC_CLEANUP_PERFORM, it will be torn down in this call 5349 together with the current protected VM. 5350 5351KVM_PV_VM_SET_SEC_PARMS 5352 Pass the image header from VM memory to the Ultravisor in 5353 preparation of image unpacking and verification. 5354 5355KVM_PV_VM_UNPACK 5356 Unpack (protect and decrypt) a page of the encrypted boot image. 5357 5358KVM_PV_VM_VERIFY 5359 Verify the integrity of the unpacked image. Only if this succeeds, 5360 KVM is allowed to start protected VCPUs. 5361 5362KVM_PV_INFO 5363 :Capability: KVM_CAP_S390_PROTECTED_DUMP 5364 5365 Presents an API that provides Ultravisor related data to userspace 5366 via subcommands. len_max is the size of the user space buffer, 5367 len_written is KVM's indication of how much bytes of that buffer 5368 were actually written to. len_written can be used to determine the 5369 valid fields if more response fields are added in the future. 5370 5371 :: 5372 5373 enum pv_cmd_info_id { 5374 KVM_PV_INFO_VM, 5375 KVM_PV_INFO_DUMP, 5376 }; 5377 5378 struct kvm_s390_pv_info_header { 5379 __u32 id; 5380 __u32 len_max; 5381 __u32 len_written; 5382 __u32 reserved; 5383 }; 5384 5385 struct kvm_s390_pv_info { 5386 struct kvm_s390_pv_info_header header; 5387 struct kvm_s390_pv_info_dump dump; 5388 struct kvm_s390_pv_info_vm vm; 5389 }; 5390 5391**subcommands:** 5392 5393 KVM_PV_INFO_VM 5394 This subcommand provides basic Ultravisor information for PV 5395 hosts. These values are likely also exported as files in the sysfs 5396 firmware UV query interface but they are more easily available to 5397 programs in this API. 5398 5399 The installed calls and feature_indication members provide the 5400 installed UV calls and the UV's other feature indications. 5401 5402 The max_* members provide information about the maximum number of PV 5403 vcpus, PV guests and PV guest memory size. 5404 5405 :: 5406 5407 struct kvm_s390_pv_info_vm { 5408 __u64 inst_calls_list[4]; 5409 __u64 max_cpus; 5410 __u64 max_guests; 5411 __u64 max_guest_addr; 5412 __u64 feature_indication; 5413 }; 5414 5415 5416 KVM_PV_INFO_DUMP 5417 This subcommand provides information related to dumping PV guests. 5418 5419 :: 5420 5421 struct kvm_s390_pv_info_dump { 5422 __u64 dump_cpu_buffer_len; 5423 __u64 dump_config_mem_buffer_per_1m; 5424 __u64 dump_config_finalize_len; 5425 }; 5426 5427KVM_PV_DUMP 5428 :Capability: KVM_CAP_S390_PROTECTED_DUMP 5429 5430 Presents an API that provides calls which facilitate dumping a 5431 protected VM. 5432 5433 :: 5434 5435 struct kvm_s390_pv_dmp { 5436 __u64 subcmd; 5437 __u64 buff_addr; 5438 __u64 buff_len; 5439 __u64 gaddr; /* For dump storage state */ 5440 }; 5441 5442 **subcommands:** 5443 5444 KVM_PV_DUMP_INIT 5445 Initializes the dump process of a protected VM. If this call does 5446 not succeed all other subcommands will fail with -EINVAL. This 5447 subcommand will return -EINVAL if a dump process has not yet been 5448 completed. 5449 5450 Not all PV vms can be dumped, the owner needs to set `dump 5451 allowed` PCF bit 34 in the SE header to allow dumping. 5452 5453 KVM_PV_DUMP_CONFIG_STOR_STATE 5454 Stores `buff_len` bytes of tweak component values starting with 5455 the 1MB block specified by the absolute guest address 5456 (`gaddr`). `buff_len` needs to be `conf_dump_storage_state_len` 5457 aligned and at least >= the `conf_dump_storage_state_len` value 5458 provided by the dump uv_info data. buff_user might be written to 5459 even if an error rc is returned. For instance if we encounter a 5460 fault after writing the first page of data. 5461 5462 KVM_PV_DUMP_COMPLETE 5463 If the subcommand succeeds it completes the dump process and lets 5464 KVM_PV_DUMP_INIT be called again. 5465 5466 On success `conf_dump_finalize_len` bytes of completion data will be 5467 stored to the `buff_addr`. The completion data contains a key 5468 derivation seed, IV, tweak nonce and encryption keys as well as an 5469 authentication tag all of which are needed to decrypt the dump at a 5470 later time. 5471 5472KVM_PV_ASYNC_CLEANUP_PREPARE 5473 :Capability: KVM_CAP_S390_PROTECTED_ASYNC_DISABLE 5474 5475 Prepare the current protected VM for asynchronous teardown. Most 5476 resources used by the current protected VM will be set aside for a 5477 subsequent asynchronous teardown. The current protected VM will then 5478 resume execution immediately as non-protected. There can be at most 5479 one protected VM prepared for asynchronous teardown at any time. If 5480 a protected VM had already been prepared for teardown without 5481 subsequently calling KVM_PV_ASYNC_CLEANUP_PERFORM, this call will 5482 fail. In that case, the userspace process should issue a normal 5483 KVM_PV_DISABLE. The resources set aside with this call will need to 5484 be cleaned up with a subsequent call to KVM_PV_ASYNC_CLEANUP_PERFORM 5485 or KVM_PV_DISABLE, otherwise they will be cleaned up when KVM 5486 terminates. KVM_PV_ASYNC_CLEANUP_PREPARE can be called again as soon 5487 as cleanup starts, i.e. before KVM_PV_ASYNC_CLEANUP_PERFORM finishes. 5488 5489KVM_PV_ASYNC_CLEANUP_PERFORM 5490 :Capability: KVM_CAP_S390_PROTECTED_ASYNC_DISABLE 5491 5492 Tear down the protected VM previously prepared for teardown with 5493 KVM_PV_ASYNC_CLEANUP_PREPARE. The resources that had been set aside 5494 will be freed during the execution of this command. This PV command 5495 should ideally be issued by userspace from a separate thread. If a 5496 fatal signal is received (or the process terminates naturally), the 5497 command will terminate immediately without completing, and the normal 5498 KVM shutdown procedure will take care of cleaning up all remaining 5499 protected VMs, including the ones whose teardown was interrupted by 5500 process termination. 5501 55024.126 KVM_XEN_HVM_SET_ATTR 5503-------------------------- 5504 5505:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 5506:Architectures: x86 5507:Type: vm ioctl 5508:Parameters: struct kvm_xen_hvm_attr 5509:Returns: 0 on success, < 0 on error 5510 5511:: 5512 5513 struct kvm_xen_hvm_attr { 5514 __u16 type; 5515 __u16 pad[3]; 5516 union { 5517 __u8 long_mode; 5518 __u8 vector; 5519 __u8 runstate_update_flag; 5520 union { 5521 __u64 gfn; 5522 __u64 hva; 5523 } shared_info; 5524 struct { 5525 __u32 send_port; 5526 __u32 type; /* EVTCHNSTAT_ipi / EVTCHNSTAT_interdomain */ 5527 __u32 flags; 5528 union { 5529 struct { 5530 __u32 port; 5531 __u32 vcpu; 5532 __u32 priority; 5533 } port; 5534 struct { 5535 __u32 port; /* Zero for eventfd */ 5536 __s32 fd; 5537 } eventfd; 5538 __u32 padding[4]; 5539 } deliver; 5540 } evtchn; 5541 __u32 xen_version; 5542 __u64 pad[8]; 5543 } u; 5544 }; 5545 5546type values: 5547 5548KVM_XEN_ATTR_TYPE_LONG_MODE 5549 Sets the ABI mode of the VM to 32-bit or 64-bit (long mode). This 5550 determines the layout of the shared_info page exposed to the VM. 5551 5552KVM_XEN_ATTR_TYPE_SHARED_INFO 5553 Sets the guest physical frame number at which the Xen shared_info 5554 page resides. Note that although Xen places vcpu_info for the first 5555 32 vCPUs in the shared_info page, KVM does not automatically do so 5556 and instead requires that KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO or 5557 KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO_HVA be used explicitly even when 5558 the vcpu_info for a given vCPU resides at the "default" location 5559 in the shared_info page. This is because KVM may not be aware of 5560 the Xen CPU id which is used as the index into the vcpu_info[] 5561 array, so may know the correct default location. 5562 5563 Note that the shared_info page may be constantly written to by KVM; 5564 it contains the event channel bitmap used to deliver interrupts to 5565 a Xen guest, amongst other things. It is exempt from dirty tracking 5566 mechanisms — KVM will not explicitly mark the page as dirty each 5567 time an event channel interrupt is delivered to the guest! Thus, 5568 userspace should always assume that the designated GFN is dirty if 5569 any vCPU has been running or any event channel interrupts can be 5570 routed to the guest. 5571 5572 Setting the gfn to KVM_XEN_INVALID_GFN will disable the shared_info 5573 page. 5574 5575KVM_XEN_ATTR_TYPE_SHARED_INFO_HVA 5576 If the KVM_XEN_HVM_CONFIG_SHARED_INFO_HVA flag is also set in the 5577 Xen capabilities, then this attribute may be used to set the 5578 userspace address at which the shared_info page resides, which 5579 will always be fixed in the VMM regardless of where it is mapped 5580 in guest physical address space. This attribute should be used in 5581 preference to KVM_XEN_ATTR_TYPE_SHARED_INFO as it avoids 5582 unnecessary invalidation of an internal cache when the page is 5583 re-mapped in guest physical address space. 5584 5585 Setting the hva to zero will disable the shared_info page. 5586 5587KVM_XEN_ATTR_TYPE_UPCALL_VECTOR 5588 Sets the exception vector used to deliver Xen event channel upcalls. 5589 This is the HVM-wide vector injected directly by the hypervisor 5590 (not through the local APIC), typically configured by a guest via 5591 HVM_PARAM_CALLBACK_IRQ. This can be disabled again (e.g. for guest 5592 SHUTDOWN_soft_reset) by setting it to zero. 5593 5594KVM_XEN_ATTR_TYPE_EVTCHN 5595 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5596 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It configures 5597 an outbound port number for interception of EVTCHNOP_send requests 5598 from the guest. A given sending port number may be directed back to 5599 a specified vCPU (by APIC ID) / port / priority on the guest, or to 5600 trigger events on an eventfd. The vCPU and priority can be changed 5601 by setting KVM_XEN_EVTCHN_UPDATE in a subsequent call, but other 5602 fields cannot change for a given sending port. A port mapping is 5603 removed by using KVM_XEN_EVTCHN_DEASSIGN in the flags field. Passing 5604 KVM_XEN_EVTCHN_RESET in the flags field removes all interception of 5605 outbound event channels. The values of the flags field are mutually 5606 exclusive and cannot be combined as a bitmask. 5607 5608KVM_XEN_ATTR_TYPE_XEN_VERSION 5609 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5610 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It configures 5611 the 32-bit version code returned to the guest when it invokes the 5612 XENVER_version call; typically (XEN_MAJOR << 16 | XEN_MINOR). PV 5613 Xen guests will often use this to as a dummy hypercall to trigger 5614 event channel delivery, so responding within the kernel without 5615 exiting to userspace is beneficial. 5616 5617KVM_XEN_ATTR_TYPE_RUNSTATE_UPDATE_FLAG 5618 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5619 support for KVM_XEN_HVM_CONFIG_RUNSTATE_UPDATE_FLAG. It enables the 5620 XEN_RUNSTATE_UPDATE flag which allows guest vCPUs to safely read 5621 other vCPUs' vcpu_runstate_info. Xen guests enable this feature via 5622 the VMASST_TYPE_runstate_update_flag of the HYPERVISOR_vm_assist 5623 hypercall. 5624 56254.127 KVM_XEN_HVM_GET_ATTR 5626-------------------------- 5627 5628:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 5629:Architectures: x86 5630:Type: vm ioctl 5631:Parameters: struct kvm_xen_hvm_attr 5632:Returns: 0 on success, < 0 on error 5633 5634Allows Xen VM attributes to be read. For the structure and types, 5635see KVM_XEN_HVM_SET_ATTR above. The KVM_XEN_ATTR_TYPE_EVTCHN 5636attribute cannot be read. 5637 56384.128 KVM_XEN_VCPU_SET_ATTR 5639--------------------------- 5640 5641:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 5642:Architectures: x86 5643:Type: vcpu ioctl 5644:Parameters: struct kvm_xen_vcpu_attr 5645:Returns: 0 on success, < 0 on error 5646 5647:: 5648 5649 struct kvm_xen_vcpu_attr { 5650 __u16 type; 5651 __u16 pad[3]; 5652 union { 5653 __u64 gpa; 5654 __u64 pad[4]; 5655 struct { 5656 __u64 state; 5657 __u64 state_entry_time; 5658 __u64 time_running; 5659 __u64 time_runnable; 5660 __u64 time_blocked; 5661 __u64 time_offline; 5662 } runstate; 5663 __u32 vcpu_id; 5664 struct { 5665 __u32 port; 5666 __u32 priority; 5667 __u64 expires_ns; 5668 } timer; 5669 __u8 vector; 5670 } u; 5671 }; 5672 5673type values: 5674 5675KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO 5676 Sets the guest physical address of the vcpu_info for a given vCPU. 5677 As with the shared_info page for the VM, the corresponding page may be 5678 dirtied at any time if event channel interrupt delivery is enabled, so 5679 userspace should always assume that the page is dirty without relying 5680 on dirty logging. Setting the gpa to KVM_XEN_INVALID_GPA will disable 5681 the vcpu_info. 5682 5683KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO_HVA 5684 If the KVM_XEN_HVM_CONFIG_SHARED_INFO_HVA flag is also set in the 5685 Xen capabilities, then this attribute may be used to set the 5686 userspace address of the vcpu_info for a given vCPU. It should 5687 only be used when the vcpu_info resides at the "default" location 5688 in the shared_info page. In this case it is safe to assume the 5689 userspace address will not change, because the shared_info page is 5690 an overlay on guest memory and remains at a fixed host address 5691 regardless of where it is mapped in guest physical address space 5692 and hence unnecessary invalidation of an internal cache may be 5693 avoided if the guest memory layout is modified. 5694 If the vcpu_info does not reside at the "default" location then 5695 it is not guaranteed to remain at the same host address and 5696 hence the aforementioned cache invalidation is required. 5697 5698KVM_XEN_VCPU_ATTR_TYPE_VCPU_TIME_INFO 5699 Sets the guest physical address of an additional pvclock structure 5700 for a given vCPU. This is typically used for guest vsyscall support. 5701 Setting the gpa to KVM_XEN_INVALID_GPA will disable the structure. 5702 5703KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR 5704 Sets the guest physical address of the vcpu_runstate_info for a given 5705 vCPU. This is how a Xen guest tracks CPU state such as steal time. 5706 Setting the gpa to KVM_XEN_INVALID_GPA will disable the runstate area. 5707 5708KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_CURRENT 5709 Sets the runstate (RUNSTATE_running/_runnable/_blocked/_offline) of 5710 the given vCPU from the .u.runstate.state member of the structure. 5711 KVM automatically accounts running and runnable time but blocked 5712 and offline states are only entered explicitly. 5713 5714KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_DATA 5715 Sets all fields of the vCPU runstate data from the .u.runstate member 5716 of the structure, including the current runstate. The state_entry_time 5717 must equal the sum of the other four times. 5718 5719KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST 5720 This *adds* the contents of the .u.runstate members of the structure 5721 to the corresponding members of the given vCPU's runstate data, thus 5722 permitting atomic adjustments to the runstate times. The adjustment 5723 to the state_entry_time must equal the sum of the adjustments to the 5724 other four times. The state field must be set to -1, or to a valid 5725 runstate value (RUNSTATE_running, RUNSTATE_runnable, RUNSTATE_blocked 5726 or RUNSTATE_offline) to set the current accounted state as of the 5727 adjusted state_entry_time. 5728 5729KVM_XEN_VCPU_ATTR_TYPE_VCPU_ID 5730 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5731 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It sets the Xen 5732 vCPU ID of the given vCPU, to allow timer-related VCPU operations to 5733 be intercepted by KVM. 5734 5735KVM_XEN_VCPU_ATTR_TYPE_TIMER 5736 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5737 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It sets the 5738 event channel port/priority for the VIRQ_TIMER of the vCPU, as well 5739 as allowing a pending timer to be saved/restored. Setting the timer 5740 port to zero disables kernel handling of the singleshot timer. 5741 5742KVM_XEN_VCPU_ATTR_TYPE_UPCALL_VECTOR 5743 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5744 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It sets the 5745 per-vCPU local APIC upcall vector, configured by a Xen guest with 5746 the HVMOP_set_evtchn_upcall_vector hypercall. This is typically 5747 used by Windows guests, and is distinct from the HVM-wide upcall 5748 vector configured with HVM_PARAM_CALLBACK_IRQ. It is disabled by 5749 setting the vector to zero. 5750 5751 57524.129 KVM_XEN_VCPU_GET_ATTR 5753--------------------------- 5754 5755:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 5756:Architectures: x86 5757:Type: vcpu ioctl 5758:Parameters: struct kvm_xen_vcpu_attr 5759:Returns: 0 on success, < 0 on error 5760 5761Allows Xen vCPU attributes to be read. For the structure and types, 5762see KVM_XEN_VCPU_SET_ATTR above. 5763 5764The KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST type may not be used 5765with the KVM_XEN_VCPU_GET_ATTR ioctl. 5766 57674.130 KVM_ARM_MTE_COPY_TAGS 5768--------------------------- 5769 5770:Capability: KVM_CAP_ARM_MTE 5771:Architectures: arm64 5772:Type: vm ioctl 5773:Parameters: struct kvm_arm_copy_mte_tags 5774:Returns: number of bytes copied, < 0 on error (-EINVAL for incorrect 5775 arguments, -EFAULT if memory cannot be accessed). 5776 5777:: 5778 5779 struct kvm_arm_copy_mte_tags { 5780 __u64 guest_ipa; 5781 __u64 length; 5782 void __user *addr; 5783 __u64 flags; 5784 __u64 reserved[2]; 5785 }; 5786 5787Copies Memory Tagging Extension (MTE) tags to/from guest tag memory. The 5788``guest_ipa`` and ``length`` fields must be ``PAGE_SIZE`` aligned. 5789``length`` must not be bigger than 2^31 - PAGE_SIZE bytes. The ``addr`` 5790field must point to a buffer which the tags will be copied to or from. 5791 5792``flags`` specifies the direction of copy, either ``KVM_ARM_TAGS_TO_GUEST`` or 5793``KVM_ARM_TAGS_FROM_GUEST``. 5794 5795The size of the buffer to store the tags is ``(length / 16)`` bytes 5796(granules in MTE are 16 bytes long). Each byte contains a single tag 5797value. This matches the format of ``PTRACE_PEEKMTETAGS`` and 5798``PTRACE_POKEMTETAGS``. 5799 5800If an error occurs before any data is copied then a negative error code is 5801returned. If some tags have been copied before an error occurs then the number 5802of bytes successfully copied is returned. If the call completes successfully 5803then ``length`` is returned. 5804 58054.131 KVM_GET_SREGS2 5806-------------------- 5807 5808:Capability: KVM_CAP_SREGS2 5809:Architectures: x86 5810:Type: vcpu ioctl 5811:Parameters: struct kvm_sregs2 (out) 5812:Returns: 0 on success, -1 on error 5813 5814Reads special registers from the vcpu. 5815This ioctl (when supported) replaces the KVM_GET_SREGS. 5816 5817:: 5818 5819 struct kvm_sregs2 { 5820 /* out (KVM_GET_SREGS2) / in (KVM_SET_SREGS2) */ 5821 struct kvm_segment cs, ds, es, fs, gs, ss; 5822 struct kvm_segment tr, ldt; 5823 struct kvm_dtable gdt, idt; 5824 __u64 cr0, cr2, cr3, cr4, cr8; 5825 __u64 efer; 5826 __u64 apic_base; 5827 __u64 flags; 5828 __u64 pdptrs[4]; 5829 }; 5830 5831flags values for ``kvm_sregs2``: 5832 5833``KVM_SREGS2_FLAGS_PDPTRS_VALID`` 5834 5835 Indicates that the struct contains valid PDPTR values. 5836 5837 58384.132 KVM_SET_SREGS2 5839-------------------- 5840 5841:Capability: KVM_CAP_SREGS2 5842:Architectures: x86 5843:Type: vcpu ioctl 5844:Parameters: struct kvm_sregs2 (in) 5845:Returns: 0 on success, -1 on error 5846 5847Writes special registers into the vcpu. 5848See KVM_GET_SREGS2 for the data structures. 5849This ioctl (when supported) replaces the KVM_SET_SREGS. 5850 58514.133 KVM_GET_STATS_FD 5852---------------------- 5853 5854:Capability: KVM_CAP_STATS_BINARY_FD 5855:Architectures: all 5856:Type: vm ioctl, vcpu ioctl 5857:Parameters: none 5858:Returns: statistics file descriptor on success, < 0 on error 5859 5860Errors: 5861 5862 ====== ====================================================== 5863 ENOMEM if the fd could not be created due to lack of memory 5864 EMFILE if the number of opened files exceeds the limit 5865 ====== ====================================================== 5866 5867The returned file descriptor can be used to read VM/vCPU statistics data in 5868binary format. The data in the file descriptor consists of four blocks 5869organized as follows: 5870 5871+-------------+ 5872| Header | 5873+-------------+ 5874| id string | 5875+-------------+ 5876| Descriptors | 5877+-------------+ 5878| Stats Data | 5879+-------------+ 5880 5881Apart from the header starting at offset 0, please be aware that it is 5882not guaranteed that the four blocks are adjacent or in the above order; 5883the offsets of the id, descriptors and data blocks are found in the 5884header. However, all four blocks are aligned to 64 bit offsets in the 5885file and they do not overlap. 5886 5887All blocks except the data block are immutable. Userspace can read them 5888only one time after retrieving the file descriptor, and then use ``pread`` or 5889``lseek`` to read the statistics repeatedly. 5890 5891All data is in system endianness. 5892 5893The format of the header is as follows:: 5894 5895 struct kvm_stats_header { 5896 __u32 flags; 5897 __u32 name_size; 5898 __u32 num_desc; 5899 __u32 id_offset; 5900 __u32 desc_offset; 5901 __u32 data_offset; 5902 }; 5903 5904The ``flags`` field is not used at the moment. It is always read as 0. 5905 5906The ``name_size`` field is the size (in byte) of the statistics name string 5907(including trailing '\0') which is contained in the "id string" block and 5908appended at the end of every descriptor. 5909 5910The ``num_desc`` field is the number of descriptors that are included in the 5911descriptor block. (The actual number of values in the data block may be 5912larger, since each descriptor may comprise more than one value). 5913 5914The ``id_offset`` field is the offset of the id string from the start of the 5915file indicated by the file descriptor. It is a multiple of 8. 5916 5917The ``desc_offset`` field is the offset of the Descriptors block from the start 5918of the file indicated by the file descriptor. It is a multiple of 8. 5919 5920The ``data_offset`` field is the offset of the Stats Data block from the start 5921of the file indicated by the file descriptor. It is a multiple of 8. 5922 5923The id string block contains a string which identifies the file descriptor on 5924which KVM_GET_STATS_FD was invoked. The size of the block, including the 5925trailing ``'\0'``, is indicated by the ``name_size`` field in the header. 5926 5927The descriptors block is only needed to be read once for the lifetime of the 5928file descriptor contains a sequence of ``struct kvm_stats_desc``, each followed 5929by a string of size ``name_size``. 5930:: 5931 5932 #define KVM_STATS_TYPE_SHIFT 0 5933 #define KVM_STATS_TYPE_MASK (0xF << KVM_STATS_TYPE_SHIFT) 5934 #define KVM_STATS_TYPE_CUMULATIVE (0x0 << KVM_STATS_TYPE_SHIFT) 5935 #define KVM_STATS_TYPE_INSTANT (0x1 << KVM_STATS_TYPE_SHIFT) 5936 #define KVM_STATS_TYPE_PEAK (0x2 << KVM_STATS_TYPE_SHIFT) 5937 #define KVM_STATS_TYPE_LINEAR_HIST (0x3 << KVM_STATS_TYPE_SHIFT) 5938 #define KVM_STATS_TYPE_LOG_HIST (0x4 << KVM_STATS_TYPE_SHIFT) 5939 #define KVM_STATS_TYPE_MAX KVM_STATS_TYPE_LOG_HIST 5940 5941 #define KVM_STATS_UNIT_SHIFT 4 5942 #define KVM_STATS_UNIT_MASK (0xF << KVM_STATS_UNIT_SHIFT) 5943 #define KVM_STATS_UNIT_NONE (0x0 << KVM_STATS_UNIT_SHIFT) 5944 #define KVM_STATS_UNIT_BYTES (0x1 << KVM_STATS_UNIT_SHIFT) 5945 #define KVM_STATS_UNIT_SECONDS (0x2 << KVM_STATS_UNIT_SHIFT) 5946 #define KVM_STATS_UNIT_CYCLES (0x3 << KVM_STATS_UNIT_SHIFT) 5947 #define KVM_STATS_UNIT_BOOLEAN (0x4 << KVM_STATS_UNIT_SHIFT) 5948 #define KVM_STATS_UNIT_MAX KVM_STATS_UNIT_BOOLEAN 5949 5950 #define KVM_STATS_BASE_SHIFT 8 5951 #define KVM_STATS_BASE_MASK (0xF << KVM_STATS_BASE_SHIFT) 5952 #define KVM_STATS_BASE_POW10 (0x0 << KVM_STATS_BASE_SHIFT) 5953 #define KVM_STATS_BASE_POW2 (0x1 << KVM_STATS_BASE_SHIFT) 5954 #define KVM_STATS_BASE_MAX KVM_STATS_BASE_POW2 5955 5956 struct kvm_stats_desc { 5957 __u32 flags; 5958 __s16 exponent; 5959 __u16 size; 5960 __u32 offset; 5961 __u32 bucket_size; 5962 char name[]; 5963 }; 5964 5965The ``flags`` field contains the type and unit of the statistics data described 5966by this descriptor. Its endianness is CPU native. 5967The following flags are supported: 5968 5969Bits 0-3 of ``flags`` encode the type: 5970 5971 * ``KVM_STATS_TYPE_CUMULATIVE`` 5972 The statistics reports a cumulative count. The value of data can only be increased. 5973 Most of the counters used in KVM are of this type. 5974 The corresponding ``size`` field for this type is always 1. 5975 All cumulative statistics data are read/write. 5976 * ``KVM_STATS_TYPE_INSTANT`` 5977 The statistics reports an instantaneous value. Its value can be increased or 5978 decreased. This type is usually used as a measurement of some resources, 5979 like the number of dirty pages, the number of large pages, etc. 5980 All instant statistics are read only. 5981 The corresponding ``size`` field for this type is always 1. 5982 * ``KVM_STATS_TYPE_PEAK`` 5983 The statistics data reports a peak value, for example the maximum number 5984 of items in a hash table bucket, the longest time waited and so on. 5985 The value of data can only be increased. 5986 The corresponding ``size`` field for this type is always 1. 5987 * ``KVM_STATS_TYPE_LINEAR_HIST`` 5988 The statistic is reported as a linear histogram. The number of 5989 buckets is specified by the ``size`` field. The size of buckets is specified 5990 by the ``hist_param`` field. The range of the Nth bucket (1 <= N < ``size``) 5991 is [``hist_param``*(N-1), ``hist_param``*N), while the range of the last 5992 bucket is [``hist_param``*(``size``-1), +INF). (+INF means positive infinity 5993 value.) 5994 * ``KVM_STATS_TYPE_LOG_HIST`` 5995 The statistic is reported as a logarithmic histogram. The number of 5996 buckets is specified by the ``size`` field. The range of the first bucket is 5997 [0, 1), while the range of the last bucket is [pow(2, ``size``-2), +INF). 5998 Otherwise, The Nth bucket (1 < N < ``size``) covers 5999 [pow(2, N-2), pow(2, N-1)). 6000 6001Bits 4-7 of ``flags`` encode the unit: 6002 6003 * ``KVM_STATS_UNIT_NONE`` 6004 There is no unit for the value of statistics data. This usually means that 6005 the value is a simple counter of an event. 6006 * ``KVM_STATS_UNIT_BYTES`` 6007 It indicates that the statistics data is used to measure memory size, in the 6008 unit of Byte, KiByte, MiByte, GiByte, etc. The unit of the data is 6009 determined by the ``exponent`` field in the descriptor. 6010 * ``KVM_STATS_UNIT_SECONDS`` 6011 It indicates that the statistics data is used to measure time or latency. 6012 * ``KVM_STATS_UNIT_CYCLES`` 6013 It indicates that the statistics data is used to measure CPU clock cycles. 6014 * ``KVM_STATS_UNIT_BOOLEAN`` 6015 It indicates that the statistic will always be either 0 or 1. Boolean 6016 statistics of "peak" type will never go back from 1 to 0. Boolean 6017 statistics can be linear histograms (with two buckets) but not logarithmic 6018 histograms. 6019 6020Note that, in the case of histograms, the unit applies to the bucket 6021ranges, while the bucket value indicates how many samples fell in the 6022bucket's range. 6023 6024Bits 8-11 of ``flags``, together with ``exponent``, encode the scale of the 6025unit: 6026 6027 * ``KVM_STATS_BASE_POW10`` 6028 The scale is based on power of 10. It is used for measurement of time and 6029 CPU clock cycles. For example, an exponent of -9 can be used with 6030 ``KVM_STATS_UNIT_SECONDS`` to express that the unit is nanoseconds. 6031 * ``KVM_STATS_BASE_POW2`` 6032 The scale is based on power of 2. It is used for measurement of memory size. 6033 For example, an exponent of 20 can be used with ``KVM_STATS_UNIT_BYTES`` to 6034 express that the unit is MiB. 6035 6036The ``size`` field is the number of values of this statistics data. Its 6037value is usually 1 for most of simple statistics. 1 means it contains an 6038unsigned 64bit data. 6039 6040The ``offset`` field is the offset from the start of Data Block to the start of 6041the corresponding statistics data. 6042 6043The ``bucket_size`` field is used as a parameter for histogram statistics data. 6044It is only used by linear histogram statistics data, specifying the size of a 6045bucket in the unit expressed by bits 4-11 of ``flags`` together with ``exponent``. 6046 6047The ``name`` field is the name string of the statistics data. The name string 6048starts at the end of ``struct kvm_stats_desc``. The maximum length including 6049the trailing ``'\0'``, is indicated by ``name_size`` in the header. 6050 6051The Stats Data block contains an array of 64-bit values in the same order 6052as the descriptors in Descriptors block. 6053 60544.134 KVM_GET_XSAVE2 6055-------------------- 6056 6057:Capability: KVM_CAP_XSAVE2 6058:Architectures: x86 6059:Type: vcpu ioctl 6060:Parameters: struct kvm_xsave (out) 6061:Returns: 0 on success, -1 on error 6062 6063 6064:: 6065 6066 struct kvm_xsave { 6067 __u32 region[1024]; 6068 __u32 extra[0]; 6069 }; 6070 6071This ioctl would copy current vcpu's xsave struct to the userspace. It 6072copies as many bytes as are returned by KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) 6073when invoked on the vm file descriptor. The size value returned by 6074KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) will always be at least 4096. 6075Currently, it is only greater than 4096 if a dynamic feature has been 6076enabled with ``arch_prctl()``, but this may change in the future. 6077 6078The offsets of the state save areas in struct kvm_xsave follow the contents 6079of CPUID leaf 0xD on the host. 6080 60814.135 KVM_XEN_HVM_EVTCHN_SEND 6082----------------------------- 6083 6084:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_EVTCHN_SEND 6085:Architectures: x86 6086:Type: vm ioctl 6087:Parameters: struct kvm_irq_routing_xen_evtchn 6088:Returns: 0 on success, < 0 on error 6089 6090 6091:: 6092 6093 struct kvm_irq_routing_xen_evtchn { 6094 __u32 port; 6095 __u32 vcpu; 6096 __u32 priority; 6097 }; 6098 6099This ioctl injects an event channel interrupt directly to the guest vCPU. 6100 61014.136 KVM_S390_PV_CPU_COMMAND 6102----------------------------- 6103 6104:Capability: KVM_CAP_S390_PROTECTED_DUMP 6105:Architectures: s390 6106:Type: vcpu ioctl 6107:Parameters: none 6108:Returns: 0 on success, < 0 on error 6109 6110This ioctl closely mirrors `KVM_S390_PV_COMMAND` but handles requests 6111for vcpus. It re-uses the kvm_s390_pv_dmp struct and hence also shares 6112the command ids. 6113 6114**command:** 6115 6116KVM_PV_DUMP 6117 Presents an API that provides calls which facilitate dumping a vcpu 6118 of a protected VM. 6119 6120**subcommand:** 6121 6122KVM_PV_DUMP_CPU 6123 Provides encrypted dump data like register values. 6124 The length of the returned data is provided by uv_info.guest_cpu_stor_len. 6125 61264.137 KVM_S390_ZPCI_OP 6127---------------------- 6128 6129:Capability: KVM_CAP_S390_ZPCI_OP 6130:Architectures: s390 6131:Type: vm ioctl 6132:Parameters: struct kvm_s390_zpci_op (in) 6133:Returns: 0 on success, <0 on error 6134 6135Used to manage hardware-assisted virtualization features for zPCI devices. 6136 6137Parameters are specified via the following structure:: 6138 6139 struct kvm_s390_zpci_op { 6140 /* in */ 6141 __u32 fh; /* target device */ 6142 __u8 op; /* operation to perform */ 6143 __u8 pad[3]; 6144 union { 6145 /* for KVM_S390_ZPCIOP_REG_AEN */ 6146 struct { 6147 __u64 ibv; /* Guest addr of interrupt bit vector */ 6148 __u64 sb; /* Guest addr of summary bit */ 6149 __u32 flags; 6150 __u32 noi; /* Number of interrupts */ 6151 __u8 isc; /* Guest interrupt subclass */ 6152 __u8 sbo; /* Offset of guest summary bit vector */ 6153 __u16 pad; 6154 } reg_aen; 6155 __u64 reserved[8]; 6156 } u; 6157 }; 6158 6159The type of operation is specified in the "op" field. 6160KVM_S390_ZPCIOP_REG_AEN is used to register the VM for adapter event 6161notification interpretation, which will allow firmware delivery of adapter 6162events directly to the vm, with KVM providing a backup delivery mechanism; 6163KVM_S390_ZPCIOP_DEREG_AEN is used to subsequently disable interpretation of 6164adapter event notifications. 6165 6166The target zPCI function must also be specified via the "fh" field. For the 6167KVM_S390_ZPCIOP_REG_AEN operation, additional information to establish firmware 6168delivery must be provided via the "reg_aen" struct. 6169 6170The "pad" and "reserved" fields may be used for future extensions and should be 6171set to 0s by userspace. 6172 61734.138 KVM_ARM_SET_COUNTER_OFFSET 6174-------------------------------- 6175 6176:Capability: KVM_CAP_COUNTER_OFFSET 6177:Architectures: arm64 6178:Type: vm ioctl 6179:Parameters: struct kvm_arm_counter_offset (in) 6180:Returns: 0 on success, < 0 on error 6181 6182This capability indicates that userspace is able to apply a single VM-wide 6183offset to both the virtual and physical counters as viewed by the guest 6184using the KVM_ARM_SET_CNT_OFFSET ioctl and the following data structure: 6185 6186:: 6187 6188 struct kvm_arm_counter_offset { 6189 __u64 counter_offset; 6190 __u64 reserved; 6191 }; 6192 6193The offset describes a number of counter cycles that are subtracted from 6194both virtual and physical counter views (similar to the effects of the 6195CNTVOFF_EL2 and CNTPOFF_EL2 system registers, but only global). The offset 6196always applies to all vcpus (already created or created after this ioctl) 6197for this VM. 6198 6199It is userspace's responsibility to compute the offset based, for example, 6200on previous values of the guest counters. 6201 6202Any value other than 0 for the "reserved" field may result in an error 6203(-EINVAL) being returned. This ioctl can also return -EBUSY if any vcpu 6204ioctl is issued concurrently. 6205 6206Note that using this ioctl results in KVM ignoring subsequent userspace 6207writes to the CNTVCT_EL0 and CNTPCT_EL0 registers using the SET_ONE_REG 6208interface. No error will be returned, but the resulting offset will not be 6209applied. 6210 6211.. _KVM_ARM_GET_REG_WRITABLE_MASKS: 6212 62134.139 KVM_ARM_GET_REG_WRITABLE_MASKS 6214------------------------------------ 6215 6216:Capability: KVM_CAP_ARM_SUPPORTED_REG_MASK_RANGES 6217:Architectures: arm64 6218:Type: vm ioctl 6219:Parameters: struct reg_mask_range (in/out) 6220:Returns: 0 on success, < 0 on error 6221 6222 6223:: 6224 6225 #define KVM_ARM_FEATURE_ID_RANGE 0 6226 #define KVM_ARM_FEATURE_ID_RANGE_SIZE (3 * 8 * 8) 6227 6228 struct reg_mask_range { 6229 __u64 addr; /* Pointer to mask array */ 6230 __u32 range; /* Requested range */ 6231 __u32 reserved[13]; 6232 }; 6233 6234This ioctl copies the writable masks for a selected range of registers to 6235userspace. 6236 6237The ``addr`` field is a pointer to the destination array where KVM copies 6238the writable masks. 6239 6240The ``range`` field indicates the requested range of registers. 6241``KVM_CHECK_EXTENSION`` for the ``KVM_CAP_ARM_SUPPORTED_REG_MASK_RANGES`` 6242capability returns the supported ranges, expressed as a set of flags. Each 6243flag's bit index represents a possible value for the ``range`` field. 6244All other values are reserved for future use and KVM may return an error. 6245 6246The ``reserved[13]`` array is reserved for future use and should be 0, or 6247KVM may return an error. 6248 6249KVM_ARM_FEATURE_ID_RANGE (0) 6250^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6251 6252The Feature ID range is defined as the AArch64 System register space with 6253op0==3, op1=={0, 1, 3}, CRn==0, CRm=={0-7}, op2=={0-7}. 6254 6255The mask returned array pointed to by ``addr`` is indexed by the macro 6256``ARM64_FEATURE_ID_RANGE_IDX(op0, op1, crn, crm, op2)``, allowing userspace 6257to know what fields can be changed for the system register described by 6258``op0, op1, crn, crm, op2``. KVM rejects ID register values that describe a 6259superset of the features supported by the system. 6260 62614.140 KVM_SET_USER_MEMORY_REGION2 6262--------------------------------- 6263 6264:Capability: KVM_CAP_USER_MEMORY2 6265:Architectures: all 6266:Type: vm ioctl 6267:Parameters: struct kvm_userspace_memory_region2 (in) 6268:Returns: 0 on success, -1 on error 6269 6270KVM_SET_USER_MEMORY_REGION2 is an extension to KVM_SET_USER_MEMORY_REGION that 6271allows mapping guest_memfd memory into a guest. All fields shared with 6272KVM_SET_USER_MEMORY_REGION identically. Userspace can set KVM_MEM_GUEST_MEMFD 6273in flags to have KVM bind the memory region to a given guest_memfd range of 6274[guest_memfd_offset, guest_memfd_offset + memory_size]. The target guest_memfd 6275must point at a file created via KVM_CREATE_GUEST_MEMFD on the current VM, and 6276the target range must not be bound to any other memory region. All standard 6277bounds checks apply (use common sense). 6278 6279:: 6280 6281 struct kvm_userspace_memory_region2 { 6282 __u32 slot; 6283 __u32 flags; 6284 __u64 guest_phys_addr; 6285 __u64 memory_size; /* bytes */ 6286 __u64 userspace_addr; /* start of the userspace allocated memory */ 6287 __u64 guest_memfd_offset; 6288 __u32 guest_memfd; 6289 __u32 pad1; 6290 __u64 pad2[14]; 6291 }; 6292 6293A KVM_MEM_GUEST_MEMFD region _must_ have a valid guest_memfd (private memory) and 6294userspace_addr (shared memory). However, "valid" for userspace_addr simply 6295means that the address itself must be a legal userspace address. The backing 6296mapping for userspace_addr is not required to be valid/populated at the time of 6297KVM_SET_USER_MEMORY_REGION2, e.g. shared memory can be lazily mapped/allocated 6298on-demand. 6299 6300When mapping a gfn into the guest, KVM selects shared vs. private, i.e consumes 6301userspace_addr vs. guest_memfd, based on the gfn's KVM_MEMORY_ATTRIBUTE_PRIVATE 6302state. At VM creation time, all memory is shared, i.e. the PRIVATE attribute 6303is '0' for all gfns. Userspace can control whether memory is shared/private by 6304toggling KVM_MEMORY_ATTRIBUTE_PRIVATE via KVM_SET_MEMORY_ATTRIBUTES as needed. 6305 6306S390: 6307^^^^^ 6308 6309Returns -EINVAL if the VM has the KVM_VM_S390_UCONTROL flag set. 6310Returns -EINVAL if called on a protected VM. 6311 63124.141 KVM_SET_MEMORY_ATTRIBUTES 6313------------------------------- 6314 6315:Capability: KVM_CAP_MEMORY_ATTRIBUTES 6316:Architectures: x86 6317:Type: vm ioctl 6318:Parameters: struct kvm_memory_attributes (in) 6319:Returns: 0 on success, <0 on error 6320 6321KVM_SET_MEMORY_ATTRIBUTES allows userspace to set memory attributes for a range 6322of guest physical memory. 6323 6324:: 6325 6326 struct kvm_memory_attributes { 6327 __u64 address; 6328 __u64 size; 6329 __u64 attributes; 6330 __u64 flags; 6331 }; 6332 6333 #define KVM_MEMORY_ATTRIBUTE_PRIVATE (1ULL << 3) 6334 6335The address and size must be page aligned. The supported attributes can be 6336retrieved via ioctl(KVM_CHECK_EXTENSION) on KVM_CAP_MEMORY_ATTRIBUTES. If 6337executed on a VM, KVM_CAP_MEMORY_ATTRIBUTES precisely returns the attributes 6338supported by that VM. If executed at system scope, KVM_CAP_MEMORY_ATTRIBUTES 6339returns all attributes supported by KVM. The only attribute defined at this 6340time is KVM_MEMORY_ATTRIBUTE_PRIVATE, which marks the associated gfn as being 6341guest private memory. 6342 6343Note, there is no "get" API. Userspace is responsible for explicitly tracking 6344the state of a gfn/page as needed. 6345 6346The "flags" field is reserved for future extensions and must be '0'. 6347 63484.142 KVM_CREATE_GUEST_MEMFD 6349---------------------------- 6350 6351:Capability: KVM_CAP_GUEST_MEMFD 6352:Architectures: none 6353:Type: vm ioctl 6354:Parameters: struct kvm_create_guest_memfd(in) 6355:Returns: A file descriptor on success, <0 on error 6356 6357KVM_CREATE_GUEST_MEMFD creates an anonymous file and returns a file descriptor 6358that refers to it. guest_memfd files are roughly analogous to files created 6359via memfd_create(), e.g. guest_memfd files live in RAM, have volatile storage, 6360and are automatically released when the last reference is dropped. Unlike 6361"regular" memfd_create() files, guest_memfd files are bound to their owning 6362virtual machine (see below), cannot be mapped, read, or written by userspace, 6363and cannot be resized (guest_memfd files do however support PUNCH_HOLE). 6364 6365:: 6366 6367 struct kvm_create_guest_memfd { 6368 __u64 size; 6369 __u64 flags; 6370 __u64 reserved[6]; 6371 }; 6372 6373Conceptually, the inode backing a guest_memfd file represents physical memory, 6374i.e. is coupled to the virtual machine as a thing, not to a "struct kvm". The 6375file itself, which is bound to a "struct kvm", is that instance's view of the 6376underlying memory, e.g. effectively provides the translation of guest addresses 6377to host memory. This allows for use cases where multiple KVM structures are 6378used to manage a single virtual machine, e.g. when performing intrahost 6379migration of a virtual machine. 6380 6381KVM currently only supports mapping guest_memfd via KVM_SET_USER_MEMORY_REGION2, 6382and more specifically via the guest_memfd and guest_memfd_offset fields in 6383"struct kvm_userspace_memory_region2", where guest_memfd_offset is the offset 6384into the guest_memfd instance. For a given guest_memfd file, there can be at 6385most one mapping per page, i.e. binding multiple memory regions to a single 6386guest_memfd range is not allowed (any number of memory regions can be bound to 6387a single guest_memfd file, but the bound ranges must not overlap). 6388 6389See KVM_SET_USER_MEMORY_REGION2 for additional details. 6390 63914.143 KVM_PRE_FAULT_MEMORY 6392--------------------------- 6393 6394:Capability: KVM_CAP_PRE_FAULT_MEMORY 6395:Architectures: none 6396:Type: vcpu ioctl 6397:Parameters: struct kvm_pre_fault_memory (in/out) 6398:Returns: 0 if at least one page is processed, < 0 on error 6399 6400Errors: 6401 6402 ========== =============================================================== 6403 EINVAL The specified `gpa` and `size` were invalid (e.g. not 6404 page aligned, causes an overflow, or size is zero). 6405 ENOENT The specified `gpa` is outside defined memslots. 6406 EINTR An unmasked signal is pending and no page was processed. 6407 EFAULT The parameter address was invalid. 6408 EOPNOTSUPP Mapping memory for a GPA is unsupported by the 6409 hypervisor, and/or for the current vCPU state/mode. 6410 EIO unexpected error conditions (also causes a WARN) 6411 ========== =============================================================== 6412 6413:: 6414 6415 struct kvm_pre_fault_memory { 6416 /* in/out */ 6417 __u64 gpa; 6418 __u64 size; 6419 /* in */ 6420 __u64 flags; 6421 __u64 padding[5]; 6422 }; 6423 6424KVM_PRE_FAULT_MEMORY populates KVM's stage-2 page tables used to map memory 6425for the current vCPU state. KVM maps memory as if the vCPU generated a 6426stage-2 read page fault, e.g. faults in memory as needed, but doesn't break 6427CoW. However, KVM does not mark any newly created stage-2 PTE as Accessed. 6428 6429In the case of confidential VM types where there is an initial set up of 6430private guest memory before the guest is 'finalized'/measured, this ioctl 6431should only be issued after completing all the necessary setup to put the 6432guest into a 'finalized' state so that the above semantics can be reliably 6433ensured. 6434 6435In some cases, multiple vCPUs might share the page tables. In this 6436case, the ioctl can be called in parallel. 6437 6438When the ioctl returns, the input values are updated to point to the 6439remaining range. If `size` > 0 on return, the caller can just issue 6440the ioctl again with the same `struct kvm_map_memory` argument. 6441 6442Shadow page tables cannot support this ioctl because they 6443are indexed by virtual address or nested guest physical address. 6444Calling this ioctl when the guest is using shadow page tables (for 6445example because it is running a nested guest with nested page tables) 6446will fail with `EOPNOTSUPP` even if `KVM_CHECK_EXTENSION` reports 6447the capability to be present. 6448 6449`flags` must currently be zero. 6450 6451 6452.. _kvm_run: 6453 64545. The kvm_run structure 6455======================== 6456 6457Application code obtains a pointer to the kvm_run structure by 6458mmap()ing a vcpu fd. From that point, application code can control 6459execution by changing fields in kvm_run prior to calling the KVM_RUN 6460ioctl, and obtain information about the reason KVM_RUN returned by 6461looking up structure members. 6462 6463:: 6464 6465 struct kvm_run { 6466 /* in */ 6467 __u8 request_interrupt_window; 6468 6469Request that KVM_RUN return when it becomes possible to inject external 6470interrupts into the guest. Useful in conjunction with KVM_INTERRUPT. 6471 6472:: 6473 6474 __u8 immediate_exit; 6475 6476This field is polled once when KVM_RUN starts; if non-zero, KVM_RUN 6477exits immediately, returning -EINTR. In the common scenario where a 6478signal is used to "kick" a VCPU out of KVM_RUN, this field can be used 6479to avoid usage of KVM_SET_SIGNAL_MASK, which has worse scalability. 6480Rather than blocking the signal outside KVM_RUN, userspace can set up 6481a signal handler that sets run->immediate_exit to a non-zero value. 6482 6483This field is ignored if KVM_CAP_IMMEDIATE_EXIT is not available. 6484 6485:: 6486 6487 __u8 padding1[6]; 6488 6489 /* out */ 6490 __u32 exit_reason; 6491 6492When KVM_RUN has returned successfully (return value 0), this informs 6493application code why KVM_RUN has returned. Allowable values for this 6494field are detailed below. 6495 6496:: 6497 6498 __u8 ready_for_interrupt_injection; 6499 6500If request_interrupt_window has been specified, this field indicates 6501an interrupt can be injected now with KVM_INTERRUPT. 6502 6503:: 6504 6505 __u8 if_flag; 6506 6507The value of the current interrupt flag. Only valid if in-kernel 6508local APIC is not used. 6509 6510:: 6511 6512 __u16 flags; 6513 6514More architecture-specific flags detailing state of the VCPU that may 6515affect the device's behavior. Current defined flags:: 6516 6517 /* x86, set if the VCPU is in system management mode */ 6518 #define KVM_RUN_X86_SMM (1 << 0) 6519 /* x86, set if bus lock detected in VM */ 6520 #define KVM_RUN_X86_BUS_LOCK (1 << 1) 6521 /* x86, set if the VCPU is executing a nested (L2) guest */ 6522 #define KVM_RUN_X86_GUEST_MODE (1 << 2) 6523 6524 /* arm64, set for KVM_EXIT_DEBUG */ 6525 #define KVM_DEBUG_ARCH_HSR_HIGH_VALID (1 << 0) 6526 6527:: 6528 6529 /* in (pre_kvm_run), out (post_kvm_run) */ 6530 __u64 cr8; 6531 6532The value of the cr8 register. Only valid if in-kernel local APIC is 6533not used. Both input and output. 6534 6535:: 6536 6537 __u64 apic_base; 6538 6539The value of the APIC BASE msr. Only valid if in-kernel local 6540APIC is not used. Both input and output. 6541 6542:: 6543 6544 union { 6545 /* KVM_EXIT_UNKNOWN */ 6546 struct { 6547 __u64 hardware_exit_reason; 6548 } hw; 6549 6550If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown 6551reasons. Further architecture-specific information is available in 6552hardware_exit_reason. 6553 6554:: 6555 6556 /* KVM_EXIT_FAIL_ENTRY */ 6557 struct { 6558 __u64 hardware_entry_failure_reason; 6559 __u32 cpu; /* if KVM_LAST_CPU */ 6560 } fail_entry; 6561 6562If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due 6563to unknown reasons. Further architecture-specific information is 6564available in hardware_entry_failure_reason. 6565 6566:: 6567 6568 /* KVM_EXIT_EXCEPTION */ 6569 struct { 6570 __u32 exception; 6571 __u32 error_code; 6572 } ex; 6573 6574Unused. 6575 6576:: 6577 6578 /* KVM_EXIT_IO */ 6579 struct { 6580 #define KVM_EXIT_IO_IN 0 6581 #define KVM_EXIT_IO_OUT 1 6582 __u8 direction; 6583 __u8 size; /* bytes */ 6584 __u16 port; 6585 __u32 count; 6586 __u64 data_offset; /* relative to kvm_run start */ 6587 } io; 6588 6589If exit_reason is KVM_EXIT_IO, then the vcpu has 6590executed a port I/O instruction which could not be satisfied by kvm. 6591data_offset describes where the data is located (KVM_EXIT_IO_OUT) or 6592where kvm expects application code to place the data for the next 6593KVM_RUN invocation (KVM_EXIT_IO_IN). Data format is a packed array. 6594 6595:: 6596 6597 /* KVM_EXIT_DEBUG */ 6598 struct { 6599 struct kvm_debug_exit_arch arch; 6600 } debug; 6601 6602If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event 6603for which architecture specific information is returned. 6604 6605:: 6606 6607 /* KVM_EXIT_MMIO */ 6608 struct { 6609 __u64 phys_addr; 6610 __u8 data[8]; 6611 __u32 len; 6612 __u8 is_write; 6613 } mmio; 6614 6615If exit_reason is KVM_EXIT_MMIO, then the vcpu has 6616executed a memory-mapped I/O instruction which could not be satisfied 6617by kvm. The 'data' member contains the written data if 'is_write' is 6618true, and should be filled by application code otherwise. 6619 6620The 'data' member contains, in its first 'len' bytes, the value as it would 6621appear if the VCPU performed a load or store of the appropriate width directly 6622to the byte array. 6623 6624.. note:: 6625 6626 For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR, KVM_EXIT_XEN, 6627 KVM_EXIT_EPR, KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR the corresponding 6628 operations are complete (and guest state is consistent) only after userspace 6629 has re-entered the kernel with KVM_RUN. The kernel side will first finish 6630 incomplete operations and then check for pending signals. 6631 6632 The pending state of the operation is not preserved in state which is 6633 visible to userspace, thus userspace should ensure that the operation is 6634 completed before performing a live migration. Userspace can re-enter the 6635 guest with an unmasked signal pending or with the immediate_exit field set 6636 to complete pending operations without allowing any further instructions 6637 to be executed. 6638 6639:: 6640 6641 /* KVM_EXIT_HYPERCALL */ 6642 struct { 6643 __u64 nr; 6644 __u64 args[6]; 6645 __u64 ret; 6646 __u64 flags; 6647 } hypercall; 6648 6649 6650It is strongly recommended that userspace use ``KVM_EXIT_IO`` (x86) or 6651``KVM_EXIT_MMIO`` (all except s390) to implement functionality that 6652requires a guest to interact with host userspace. 6653 6654.. note:: KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO. 6655 6656For arm64: 6657---------- 6658 6659SMCCC exits can be enabled depending on the configuration of the SMCCC 6660filter. See the Documentation/virt/kvm/devices/vm.rst 6661``KVM_ARM_SMCCC_FILTER`` for more details. 6662 6663``nr`` contains the function ID of the guest's SMCCC call. Userspace is 6664expected to use the ``KVM_GET_ONE_REG`` ioctl to retrieve the call 6665parameters from the vCPU's GPRs. 6666 6667Definition of ``flags``: 6668 - ``KVM_HYPERCALL_EXIT_SMC``: Indicates that the guest used the SMC 6669 conduit to initiate the SMCCC call. If this bit is 0 then the guest 6670 used the HVC conduit for the SMCCC call. 6671 6672 - ``KVM_HYPERCALL_EXIT_16BIT``: Indicates that the guest used a 16bit 6673 instruction to initiate the SMCCC call. If this bit is 0 then the 6674 guest used a 32bit instruction. An AArch64 guest always has this 6675 bit set to 0. 6676 6677At the point of exit, PC points to the instruction immediately following 6678the trapping instruction. 6679 6680:: 6681 6682 /* KVM_EXIT_TPR_ACCESS */ 6683 struct { 6684 __u64 rip; 6685 __u32 is_write; 6686 __u32 pad; 6687 } tpr_access; 6688 6689To be documented (KVM_TPR_ACCESS_REPORTING). 6690 6691:: 6692 6693 /* KVM_EXIT_S390_SIEIC */ 6694 struct { 6695 __u8 icptcode; 6696 __u64 mask; /* psw upper half */ 6697 __u64 addr; /* psw lower half */ 6698 __u16 ipa; 6699 __u32 ipb; 6700 } s390_sieic; 6701 6702s390 specific. 6703 6704:: 6705 6706 /* KVM_EXIT_S390_RESET */ 6707 #define KVM_S390_RESET_POR 1 6708 #define KVM_S390_RESET_CLEAR 2 6709 #define KVM_S390_RESET_SUBSYSTEM 4 6710 #define KVM_S390_RESET_CPU_INIT 8 6711 #define KVM_S390_RESET_IPL 16 6712 __u64 s390_reset_flags; 6713 6714s390 specific. 6715 6716:: 6717 6718 /* KVM_EXIT_S390_UCONTROL */ 6719 struct { 6720 __u64 trans_exc_code; 6721 __u32 pgm_code; 6722 } s390_ucontrol; 6723 6724s390 specific. A page fault has occurred for a user controlled virtual 6725machine (KVM_VM_S390_UNCONTROL) on its host page table that cannot be 6726resolved by the kernel. 6727The program code and the translation exception code that were placed 6728in the cpu's lowcore are presented here as defined by the z Architecture 6729Principles of Operation Book in the Chapter for Dynamic Address Translation 6730(DAT) 6731 6732:: 6733 6734 /* KVM_EXIT_DCR */ 6735 struct { 6736 __u32 dcrn; 6737 __u32 data; 6738 __u8 is_write; 6739 } dcr; 6740 6741Deprecated - was used for 440 KVM. 6742 6743:: 6744 6745 /* KVM_EXIT_OSI */ 6746 struct { 6747 __u64 gprs[32]; 6748 } osi; 6749 6750MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch 6751hypercalls and exit with this exit struct that contains all the guest gprs. 6752 6753If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall. 6754Userspace can now handle the hypercall and when it's done modify the gprs as 6755necessary. Upon guest entry all guest GPRs will then be replaced by the values 6756in this struct. 6757 6758:: 6759 6760 /* KVM_EXIT_PAPR_HCALL */ 6761 struct { 6762 __u64 nr; 6763 __u64 ret; 6764 __u64 args[9]; 6765 } papr_hcall; 6766 6767This is used on 64-bit PowerPC when emulating a pSeries partition, 6768e.g. with the 'pseries' machine type in qemu. It occurs when the 6769guest does a hypercall using the 'sc 1' instruction. The 'nr' field 6770contains the hypercall number (from the guest R3), and 'args' contains 6771the arguments (from the guest R4 - R12). Userspace should put the 6772return code in 'ret' and any extra returned values in args[]. 6773The possible hypercalls are defined in the Power Architecture Platform 6774Requirements (PAPR) document available from www.power.org (free 6775developer registration required to access it). 6776 6777:: 6778 6779 /* KVM_EXIT_S390_TSCH */ 6780 struct { 6781 __u16 subchannel_id; 6782 __u16 subchannel_nr; 6783 __u32 io_int_parm; 6784 __u32 io_int_word; 6785 __u32 ipb; 6786 __u8 dequeued; 6787 } s390_tsch; 6788 6789s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled 6790and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O 6791interrupt for the target subchannel has been dequeued and subchannel_id, 6792subchannel_nr, io_int_parm and io_int_word contain the parameters for that 6793interrupt. ipb is needed for instruction parameter decoding. 6794 6795:: 6796 6797 /* KVM_EXIT_EPR */ 6798 struct { 6799 __u32 epr; 6800 } epr; 6801 6802On FSL BookE PowerPC chips, the interrupt controller has a fast patch 6803interrupt acknowledge path to the core. When the core successfully 6804delivers an interrupt, it automatically populates the EPR register with 6805the interrupt vector number and acknowledges the interrupt inside 6806the interrupt controller. 6807 6808In case the interrupt controller lives in user space, we need to do 6809the interrupt acknowledge cycle through it to fetch the next to be 6810delivered interrupt vector using this exit. 6811 6812It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an 6813external interrupt has just been delivered into the guest. User space 6814should put the acknowledged interrupt vector into the 'epr' field. 6815 6816:: 6817 6818 /* KVM_EXIT_SYSTEM_EVENT */ 6819 struct { 6820 #define KVM_SYSTEM_EVENT_SHUTDOWN 1 6821 #define KVM_SYSTEM_EVENT_RESET 2 6822 #define KVM_SYSTEM_EVENT_CRASH 3 6823 #define KVM_SYSTEM_EVENT_WAKEUP 4 6824 #define KVM_SYSTEM_EVENT_SUSPEND 5 6825 #define KVM_SYSTEM_EVENT_SEV_TERM 6 6826 __u32 type; 6827 __u32 ndata; 6828 __u64 data[16]; 6829 } system_event; 6830 6831If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered 6832a system-level event using some architecture specific mechanism (hypercall 6833or some special instruction). In case of ARM64, this is triggered using 6834HVC instruction based PSCI call from the vcpu. 6835 6836The 'type' field describes the system-level event type. 6837Valid values for 'type' are: 6838 6839 - KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the 6840 VM. Userspace is not obliged to honour this, and if it does honour 6841 this does not need to destroy the VM synchronously (ie it may call 6842 KVM_RUN again before shutdown finally occurs). 6843 - KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM. 6844 As with SHUTDOWN, userspace can choose to ignore the request, or 6845 to schedule the reset to occur in the future and may call KVM_RUN again. 6846 - KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest 6847 has requested a crash condition maintenance. Userspace can choose 6848 to ignore the request, or to gather VM memory core dump and/or 6849 reset/shutdown of the VM. 6850 - KVM_SYSTEM_EVENT_SEV_TERM -- an AMD SEV guest requested termination. 6851 The guest physical address of the guest's GHCB is stored in `data[0]`. 6852 - KVM_SYSTEM_EVENT_WAKEUP -- the exiting vCPU is in a suspended state and 6853 KVM has recognized a wakeup event. Userspace may honor this event by 6854 marking the exiting vCPU as runnable, or deny it and call KVM_RUN again. 6855 - KVM_SYSTEM_EVENT_SUSPEND -- the guest has requested a suspension of 6856 the VM. 6857 6858If KVM_CAP_SYSTEM_EVENT_DATA is present, the 'data' field can contain 6859architecture specific information for the system-level event. Only 6860the first `ndata` items (possibly zero) of the data array are valid. 6861 6862 - for arm64, data[0] is set to KVM_SYSTEM_EVENT_RESET_FLAG_PSCI_RESET2 if 6863 the guest issued a SYSTEM_RESET2 call according to v1.1 of the PSCI 6864 specification. 6865 6866 - for arm64, data[0] is set to KVM_SYSTEM_EVENT_SHUTDOWN_FLAG_PSCI_OFF2 6867 if the guest issued a SYSTEM_OFF2 call according to v1.3 of the PSCI 6868 specification. 6869 6870 - for RISC-V, data[0] is set to the value of the second argument of the 6871 ``sbi_system_reset`` call. 6872 6873Previous versions of Linux defined a `flags` member in this struct. The 6874field is now aliased to `data[0]`. Userspace can assume that it is only 6875written if ndata is greater than 0. 6876 6877For arm/arm64: 6878-------------- 6879 6880KVM_SYSTEM_EVENT_SUSPEND exits are enabled with the 6881KVM_CAP_ARM_SYSTEM_SUSPEND VM capability. If a guest invokes the PSCI 6882SYSTEM_SUSPEND function, KVM will exit to userspace with this event 6883type. 6884 6885It is the sole responsibility of userspace to implement the PSCI 6886SYSTEM_SUSPEND call according to ARM DEN0022D.b 5.19 "SYSTEM_SUSPEND". 6887KVM does not change the vCPU's state before exiting to userspace, so 6888the call parameters are left in-place in the vCPU registers. 6889 6890Userspace is _required_ to take action for such an exit. It must 6891either: 6892 6893 - Honor the guest request to suspend the VM. Userspace can request 6894 in-kernel emulation of suspension by setting the calling vCPU's 6895 state to KVM_MP_STATE_SUSPENDED. Userspace must configure the vCPU's 6896 state according to the parameters passed to the PSCI function when 6897 the calling vCPU is resumed. See ARM DEN0022D.b 5.19.1 "Intended use" 6898 for details on the function parameters. 6899 6900 - Deny the guest request to suspend the VM. See ARM DEN0022D.b 5.19.2 6901 "Caller responsibilities" for possible return values. 6902 6903Hibernation using the PSCI SYSTEM_OFF2 call is enabled when PSCI v1.3 6904is enabled. If a guest invokes the PSCI SYSTEM_OFF2 function, KVM will 6905exit to userspace with the KVM_SYSTEM_EVENT_SHUTDOWN event type and with 6906data[0] set to KVM_SYSTEM_EVENT_SHUTDOWN_FLAG_PSCI_OFF2. The only 6907supported hibernate type for the SYSTEM_OFF2 function is HIBERNATE_OFF. 6908 6909:: 6910 6911 /* KVM_EXIT_IOAPIC_EOI */ 6912 struct { 6913 __u8 vector; 6914 } eoi; 6915 6916Indicates that the VCPU's in-kernel local APIC received an EOI for a 6917level-triggered IOAPIC interrupt. This exit only triggers when the 6918IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled); 6919the userspace IOAPIC should process the EOI and retrigger the interrupt if 6920it is still asserted. Vector is the LAPIC interrupt vector for which the 6921EOI was received. 6922 6923:: 6924 6925 struct kvm_hyperv_exit { 6926 #define KVM_EXIT_HYPERV_SYNIC 1 6927 #define KVM_EXIT_HYPERV_HCALL 2 6928 #define KVM_EXIT_HYPERV_SYNDBG 3 6929 __u32 type; 6930 __u32 pad1; 6931 union { 6932 struct { 6933 __u32 msr; 6934 __u32 pad2; 6935 __u64 control; 6936 __u64 evt_page; 6937 __u64 msg_page; 6938 } synic; 6939 struct { 6940 __u64 input; 6941 __u64 result; 6942 __u64 params[2]; 6943 } hcall; 6944 struct { 6945 __u32 msr; 6946 __u32 pad2; 6947 __u64 control; 6948 __u64 status; 6949 __u64 send_page; 6950 __u64 recv_page; 6951 __u64 pending_page; 6952 } syndbg; 6953 } u; 6954 }; 6955 /* KVM_EXIT_HYPERV */ 6956 struct kvm_hyperv_exit hyperv; 6957 6958Indicates that the VCPU exits into userspace to process some tasks 6959related to Hyper-V emulation. 6960 6961Valid values for 'type' are: 6962 6963 - KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about 6964 6965Hyper-V SynIC state change. Notification is used to remap SynIC 6966event/message pages and to enable/disable SynIC messages/events processing 6967in userspace. 6968 6969 - KVM_EXIT_HYPERV_SYNDBG -- synchronously notify user-space about 6970 6971Hyper-V Synthetic debugger state change. Notification is used to either update 6972the pending_page location or to send a control command (send the buffer located 6973in send_page or recv a buffer to recv_page). 6974 6975:: 6976 6977 /* KVM_EXIT_ARM_NISV */ 6978 struct { 6979 __u64 esr_iss; 6980 __u64 fault_ipa; 6981 } arm_nisv; 6982 6983Used on arm64 systems. If a guest accesses memory not in a memslot, 6984KVM will typically return to userspace and ask it to do MMIO emulation on its 6985behalf. However, for certain classes of instructions, no instruction decode 6986(direction, length of memory access) is provided, and fetching and decoding 6987the instruction from the VM is overly complicated to live in the kernel. 6988 6989Historically, when this situation occurred, KVM would print a warning and kill 6990the VM. KVM assumed that if the guest accessed non-memslot memory, it was 6991trying to do I/O, which just couldn't be emulated, and the warning message was 6992phrased accordingly. However, what happened more often was that a guest bug 6993caused access outside the guest memory areas which should lead to a more 6994meaningful warning message and an external abort in the guest, if the access 6995did not fall within an I/O window. 6996 6997Userspace implementations can query for KVM_CAP_ARM_NISV_TO_USER, and enable 6998this capability at VM creation. Once this is done, these types of errors will 6999instead return to userspace with KVM_EXIT_ARM_NISV, with the valid bits from 7000the ESR_EL2 in the esr_iss field, and the faulting IPA in the fault_ipa field. 7001Userspace can either fix up the access if it's actually an I/O access by 7002decoding the instruction from guest memory (if it's very brave) and continue 7003executing the guest, or it can decide to suspend, dump, or restart the guest. 7004 7005Note that KVM does not skip the faulting instruction as it does for 7006KVM_EXIT_MMIO, but userspace has to emulate any change to the processing state 7007if it decides to decode and emulate the instruction. 7008 7009This feature isn't available to protected VMs, as userspace does not 7010have access to the state that is required to perform the emulation. 7011Instead, a data abort exception is directly injected in the guest. 7012Note that although KVM_CAP_ARM_NISV_TO_USER will be reported if 7013queried outside of a protected VM context, the feature will not be 7014exposed if queried on a protected VM file descriptor. 7015 7016:: 7017 7018 /* KVM_EXIT_X86_RDMSR / KVM_EXIT_X86_WRMSR */ 7019 struct { 7020 __u8 error; /* user -> kernel */ 7021 __u8 pad[7]; 7022 __u32 reason; /* kernel -> user */ 7023 __u32 index; /* kernel -> user */ 7024 __u64 data; /* kernel <-> user */ 7025 } msr; 7026 7027Used on x86 systems. When the VM capability KVM_CAP_X86_USER_SPACE_MSR is 7028enabled, MSR accesses to registers that would invoke a #GP by KVM kernel code 7029may instead trigger a KVM_EXIT_X86_RDMSR exit for reads and KVM_EXIT_X86_WRMSR 7030exit for writes. 7031 7032The "reason" field specifies why the MSR interception occurred. Userspace will 7033only receive MSR exits when a particular reason was requested during through 7034ENABLE_CAP. Currently valid exit reasons are: 7035 7036============================ ======================================== 7037 KVM_MSR_EXIT_REASON_UNKNOWN access to MSR that is unknown to KVM 7038 KVM_MSR_EXIT_REASON_INVAL access to invalid MSRs or reserved bits 7039 KVM_MSR_EXIT_REASON_FILTER access blocked by KVM_X86_SET_MSR_FILTER 7040============================ ======================================== 7041 7042For KVM_EXIT_X86_RDMSR, the "index" field tells userspace which MSR the guest 7043wants to read. To respond to this request with a successful read, userspace 7044writes the respective data into the "data" field and must continue guest 7045execution to ensure the read data is transferred into guest register state. 7046 7047If the RDMSR request was unsuccessful, userspace indicates that with a "1" in 7048the "error" field. This will inject a #GP into the guest when the VCPU is 7049executed again. 7050 7051For KVM_EXIT_X86_WRMSR, the "index" field tells userspace which MSR the guest 7052wants to write. Once finished processing the event, userspace must continue 7053vCPU execution. If the MSR write was unsuccessful, userspace also sets the 7054"error" field to "1". 7055 7056See KVM_X86_SET_MSR_FILTER for details on the interaction with MSR filtering. 7057 7058:: 7059 7060 7061 struct kvm_xen_exit { 7062 #define KVM_EXIT_XEN_HCALL 1 7063 __u32 type; 7064 union { 7065 struct { 7066 __u32 longmode; 7067 __u32 cpl; 7068 __u64 input; 7069 __u64 result; 7070 __u64 params[6]; 7071 } hcall; 7072 } u; 7073 }; 7074 /* KVM_EXIT_XEN */ 7075 struct kvm_hyperv_exit xen; 7076 7077Indicates that the VCPU exits into userspace to process some tasks 7078related to Xen emulation. 7079 7080Valid values for 'type' are: 7081 7082 - KVM_EXIT_XEN_HCALL -- synchronously notify user-space about Xen hypercall. 7083 Userspace is expected to place the hypercall result into the appropriate 7084 field before invoking KVM_RUN again. 7085 7086:: 7087 7088 /* KVM_EXIT_RISCV_SBI */ 7089 struct { 7090 unsigned long extension_id; 7091 unsigned long function_id; 7092 unsigned long args[6]; 7093 unsigned long ret[2]; 7094 } riscv_sbi; 7095 7096If exit reason is KVM_EXIT_RISCV_SBI then it indicates that the VCPU has 7097done a SBI call which is not handled by KVM RISC-V kernel module. The details 7098of the SBI call are available in 'riscv_sbi' member of kvm_run structure. The 7099'extension_id' field of 'riscv_sbi' represents SBI extension ID whereas the 7100'function_id' field represents function ID of given SBI extension. The 'args' 7101array field of 'riscv_sbi' represents parameters for the SBI call and 'ret' 7102array field represents return values. The userspace should update the return 7103values of SBI call before resuming the VCPU. For more details on RISC-V SBI 7104spec refer, https://github.com/riscv/riscv-sbi-doc. 7105 7106:: 7107 7108 /* KVM_EXIT_MEMORY_FAULT */ 7109 struct { 7110 #define KVM_MEMORY_EXIT_FLAG_PRIVATE (1ULL << 3) 7111 __u64 flags; 7112 __u64 gpa; 7113 __u64 size; 7114 } memory_fault; 7115 7116KVM_EXIT_MEMORY_FAULT indicates the vCPU has encountered a memory fault that 7117could not be resolved by KVM. The 'gpa' and 'size' (in bytes) describe the 7118guest physical address range [gpa, gpa + size) of the fault. The 'flags' field 7119describes properties of the faulting access that are likely pertinent: 7120 7121 - KVM_MEMORY_EXIT_FLAG_PRIVATE - When set, indicates the memory fault occurred 7122 on a private memory access. When clear, indicates the fault occurred on a 7123 shared access. 7124 7125Note! KVM_EXIT_MEMORY_FAULT is unique among all KVM exit reasons in that it 7126accompanies a return code of '-1', not '0'! errno will always be set to EFAULT 7127or EHWPOISON when KVM exits with KVM_EXIT_MEMORY_FAULT, userspace should assume 7128kvm_run.exit_reason is stale/undefined for all other error numbers. 7129 7130:: 7131 7132 /* KVM_EXIT_NOTIFY */ 7133 struct { 7134 #define KVM_NOTIFY_CONTEXT_INVALID (1 << 0) 7135 __u32 flags; 7136 } notify; 7137 7138Used on x86 systems. When the VM capability KVM_CAP_X86_NOTIFY_VMEXIT is 7139enabled, a VM exit generated if no event window occurs in VM non-root mode 7140for a specified amount of time. Once KVM_X86_NOTIFY_VMEXIT_USER is set when 7141enabling the cap, it would exit to userspace with the exit reason 7142KVM_EXIT_NOTIFY for further handling. The "flags" field contains more 7143detailed info. 7144 7145The valid value for 'flags' is: 7146 7147 - KVM_NOTIFY_CONTEXT_INVALID -- the VM context is corrupted and not valid 7148 in VMCS. It would run into unknown result if resume the target VM. 7149 7150:: 7151 7152 /* Fix the size of the union. */ 7153 char padding[256]; 7154 }; 7155 7156 /* 7157 * shared registers between kvm and userspace. 7158 * kvm_valid_regs specifies the register classes set by the host 7159 * kvm_dirty_regs specified the register classes dirtied by userspace 7160 * struct kvm_sync_regs is architecture specific, as well as the 7161 * bits for kvm_valid_regs and kvm_dirty_regs 7162 */ 7163 __u64 kvm_valid_regs; 7164 __u64 kvm_dirty_regs; 7165 union { 7166 struct kvm_sync_regs regs; 7167 char padding[SYNC_REGS_SIZE_BYTES]; 7168 } s; 7169 7170If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access 7171certain guest registers without having to call SET/GET_*REGS. Thus we can 7172avoid some system call overhead if userspace has to handle the exit. 7173Userspace can query the validity of the structure by checking 7174kvm_valid_regs for specific bits. These bits are architecture specific 7175and usually define the validity of a groups of registers. (e.g. one bit 7176for general purpose registers) 7177 7178Please note that the kernel is allowed to use the kvm_run structure as the 7179primary storage for certain register types. Therefore, the kernel may use the 7180values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set. 7181 7182 7183.. _cap_enable: 7184 71856. Capabilities that can be enabled on vCPUs 7186============================================ 7187 7188There are certain capabilities that change the behavior of the virtual CPU or 7189the virtual machine when enabled. To enable them, please see 7190:ref:`KVM_ENABLE_CAP`. 7191 7192Below you can find a list of capabilities and what their effect on the vCPU or 7193the virtual machine is when enabling them. 7194 7195The following information is provided along with the description: 7196 7197 Architectures: 7198 which instruction set architectures provide this ioctl. 7199 x86 includes both i386 and x86_64. 7200 7201 Target: 7202 whether this is a per-vcpu or per-vm capability. 7203 7204 Parameters: 7205 what parameters are accepted by the capability. 7206 7207 Returns: 7208 the return value. General error numbers (EBADF, ENOMEM, EINVAL) 7209 are not detailed, but errors with specific meanings are. 7210 7211 72126.1 KVM_CAP_PPC_OSI 7213------------------- 7214 7215:Architectures: ppc 7216:Target: vcpu 7217:Parameters: none 7218:Returns: 0 on success; -1 on error 7219 7220This capability enables interception of OSI hypercalls that otherwise would 7221be treated as normal system calls to be injected into the guest. OSI hypercalls 7222were invented by Mac-on-Linux to have a standardized communication mechanism 7223between the guest and the host. 7224 7225When this capability is enabled, KVM_EXIT_OSI can occur. 7226 7227 72286.2 KVM_CAP_PPC_PAPR 7229-------------------- 7230 7231:Architectures: ppc 7232:Target: vcpu 7233:Parameters: none 7234:Returns: 0 on success; -1 on error 7235 7236This capability enables interception of PAPR hypercalls. PAPR hypercalls are 7237done using the hypercall instruction "sc 1". 7238 7239It also sets the guest privilege level to "supervisor" mode. Usually the guest 7240runs in "hypervisor" privilege mode with a few missing features. 7241 7242In addition to the above, it changes the semantics of SDR1. In this mode, the 7243HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the 7244HTAB invisible to the guest. 7245 7246When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur. 7247 7248 72496.3 KVM_CAP_SW_TLB 7250------------------ 7251 7252:Architectures: ppc 7253:Target: vcpu 7254:Parameters: args[0] is the address of a struct kvm_config_tlb 7255:Returns: 0 on success; -1 on error 7256 7257:: 7258 7259 struct kvm_config_tlb { 7260 __u64 params; 7261 __u64 array; 7262 __u32 mmu_type; 7263 __u32 array_len; 7264 }; 7265 7266Configures the virtual CPU's TLB array, establishing a shared memory area 7267between userspace and KVM. The "params" and "array" fields are userspace 7268addresses of mmu-type-specific data structures. The "array_len" field is an 7269safety mechanism, and should be set to the size in bytes of the memory that 7270userspace has reserved for the array. It must be at least the size dictated 7271by "mmu_type" and "params". 7272 7273While KVM_RUN is active, the shared region is under control of KVM. Its 7274contents are undefined, and any modification by userspace results in 7275boundedly undefined behavior. 7276 7277On return from KVM_RUN, the shared region will reflect the current state of 7278the guest's TLB. If userspace makes any changes, it must call KVM_DIRTY_TLB 7279to tell KVM which entries have been changed, prior to calling KVM_RUN again 7280on this vcpu. 7281 7282For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV: 7283 7284 - The "params" field is of type "struct kvm_book3e_206_tlb_params". 7285 - The "array" field points to an array of type "struct 7286 kvm_book3e_206_tlb_entry". 7287 - The array consists of all entries in the first TLB, followed by all 7288 entries in the second TLB. 7289 - Within a TLB, entries are ordered first by increasing set number. Within a 7290 set, entries are ordered by way (increasing ESEL). 7291 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1) 7292 where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value. 7293 - The tsize field of mas1 shall be set to 4K on TLB0, even though the 7294 hardware ignores this value for TLB0. 7295 72966.4 KVM_CAP_S390_CSS_SUPPORT 7297---------------------------- 7298 7299:Architectures: s390 7300:Target: vcpu 7301:Parameters: none 7302:Returns: 0 on success; -1 on error 7303 7304This capability enables support for handling of channel I/O instructions. 7305 7306TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are 7307handled in-kernel, while the other I/O instructions are passed to userspace. 7308 7309When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST 7310SUBCHANNEL intercepts. 7311 7312Note that even though this capability is enabled per-vcpu, the complete 7313virtual machine is affected. 7314 73156.5 KVM_CAP_PPC_EPR 7316------------------- 7317 7318:Architectures: ppc 7319:Target: vcpu 7320:Parameters: args[0] defines whether the proxy facility is active 7321:Returns: 0 on success; -1 on error 7322 7323This capability enables or disables the delivery of interrupts through the 7324external proxy facility. 7325 7326When enabled (args[0] != 0), every time the guest gets an external interrupt 7327delivered, it automatically exits into user space with a KVM_EXIT_EPR exit 7328to receive the topmost interrupt vector. 7329 7330When disabled (args[0] == 0), behavior is as if this facility is unsupported. 7331 7332When this capability is enabled, KVM_EXIT_EPR can occur. 7333 73346.6 KVM_CAP_IRQ_MPIC 7335-------------------- 7336 7337:Architectures: ppc 7338:Parameters: args[0] is the MPIC device fd; 7339 args[1] is the MPIC CPU number for this vcpu 7340 7341This capability connects the vcpu to an in-kernel MPIC device. 7342 73436.7 KVM_CAP_IRQ_XICS 7344-------------------- 7345 7346:Architectures: ppc 7347:Target: vcpu 7348:Parameters: args[0] is the XICS device fd; 7349 args[1] is the XICS CPU number (server ID) for this vcpu 7350 7351This capability connects the vcpu to an in-kernel XICS device. 7352 73536.8 KVM_CAP_S390_IRQCHIP 7354------------------------ 7355 7356:Architectures: s390 7357:Target: vm 7358:Parameters: none 7359 7360This capability enables the in-kernel irqchip for s390. Please refer to 7361"4.24 KVM_CREATE_IRQCHIP" for details. 7362 73636.9 KVM_CAP_MIPS_FPU 7364-------------------- 7365 7366:Architectures: mips 7367:Target: vcpu 7368:Parameters: args[0] is reserved for future use (should be 0). 7369 7370This capability allows the use of the host Floating Point Unit by the guest. It 7371allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is 7372done the ``KVM_REG_MIPS_FPR_*`` and ``KVM_REG_MIPS_FCR_*`` registers can be 7373accessed (depending on the current guest FPU register mode), and the Status.FR, 7374Config5.FRE bits are accessible via the KVM API and also from the guest, 7375depending on them being supported by the FPU. 7376 73776.10 KVM_CAP_MIPS_MSA 7378--------------------- 7379 7380:Architectures: mips 7381:Target: vcpu 7382:Parameters: args[0] is reserved for future use (should be 0). 7383 7384This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest. 7385It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest. 7386Once this is done the ``KVM_REG_MIPS_VEC_*`` and ``KVM_REG_MIPS_MSA_*`` 7387registers can be accessed, and the Config5.MSAEn bit is accessible via the 7388KVM API and also from the guest. 7389 73906.74 KVM_CAP_SYNC_REGS 7391---------------------- 7392 7393:Architectures: s390, x86 7394:Target: s390: always enabled, x86: vcpu 7395:Parameters: none 7396:Returns: x86: KVM_CHECK_EXTENSION returns a bit-array indicating which register 7397 sets are supported 7398 (bitfields defined in arch/x86/include/uapi/asm/kvm.h). 7399 7400As described above in the kvm_sync_regs struct info in section :ref:`kvm_run`, 7401KVM_CAP_SYNC_REGS "allow[s] userspace to access certain guest registers 7402without having to call SET/GET_*REGS". This reduces overhead by eliminating 7403repeated ioctl calls for setting and/or getting register values. This is 7404particularly important when userspace is making synchronous guest state 7405modifications, e.g. when emulating and/or intercepting instructions in 7406userspace. 7407 7408For s390 specifics, please refer to the source code. 7409 7410For x86: 7411 7412- the register sets to be copied out to kvm_run are selectable 7413 by userspace (rather that all sets being copied out for every exit). 7414- vcpu_events are available in addition to regs and sregs. 7415 7416For x86, the 'kvm_valid_regs' field of struct kvm_run is overloaded to 7417function as an input bit-array field set by userspace to indicate the 7418specific register sets to be copied out on the next exit. 7419 7420To indicate when userspace has modified values that should be copied into 7421the vCPU, the all architecture bitarray field, 'kvm_dirty_regs' must be set. 7422This is done using the same bitflags as for the 'kvm_valid_regs' field. 7423If the dirty bit is not set, then the register set values will not be copied 7424into the vCPU even if they've been modified. 7425 7426Unused bitfields in the bitarrays must be set to zero. 7427 7428:: 7429 7430 struct kvm_sync_regs { 7431 struct kvm_regs regs; 7432 struct kvm_sregs sregs; 7433 struct kvm_vcpu_events events; 7434 }; 7435 74366.75 KVM_CAP_PPC_IRQ_XIVE 7437------------------------- 7438 7439:Architectures: ppc 7440:Target: vcpu 7441:Parameters: args[0] is the XIVE device fd; 7442 args[1] is the XIVE CPU number (server ID) for this vcpu 7443 7444This capability connects the vcpu to an in-kernel XIVE device. 7445 7446.. _cap_enable_vm: 7447 74487. Capabilities that can be enabled on VMs 7449========================================== 7450 7451There are certain capabilities that change the behavior of the virtual 7452machine when enabled. To enable them, please see section 7453:ref:`KVM_ENABLE_CAP`. Below you can find a list of capabilities and 7454what their effect on the VM is when enabling them. 7455 7456The following information is provided along with the description: 7457 7458 Architectures: 7459 which instruction set architectures provide this ioctl. 7460 x86 includes both i386 and x86_64. 7461 7462 Parameters: 7463 what parameters are accepted by the capability. 7464 7465 Returns: 7466 the return value. General error numbers (EBADF, ENOMEM, EINVAL) 7467 are not detailed, but errors with specific meanings are. 7468 7469 74707.1 KVM_CAP_PPC_ENABLE_HCALL 7471---------------------------- 7472 7473:Architectures: ppc 7474:Parameters: args[0] is the sPAPR hcall number; 7475 args[1] is 0 to disable, 1 to enable in-kernel handling 7476 7477This capability controls whether individual sPAPR hypercalls (hcalls) 7478get handled by the kernel or not. Enabling or disabling in-kernel 7479handling of an hcall is effective across the VM. On creation, an 7480initial set of hcalls are enabled for in-kernel handling, which 7481consists of those hcalls for which in-kernel handlers were implemented 7482before this capability was implemented. If disabled, the kernel will 7483not to attempt to handle the hcall, but will always exit to userspace 7484to handle it. Note that it may not make sense to enable some and 7485disable others of a group of related hcalls, but KVM does not prevent 7486userspace from doing that. 7487 7488If the hcall number specified is not one that has an in-kernel 7489implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL 7490error. 7491 74927.2 KVM_CAP_S390_USER_SIGP 7493-------------------------- 7494 7495:Architectures: s390 7496:Parameters: none 7497 7498This capability controls which SIGP orders will be handled completely in user 7499space. With this capability enabled, all fast orders will be handled completely 7500in the kernel: 7501 7502- SENSE 7503- SENSE RUNNING 7504- EXTERNAL CALL 7505- EMERGENCY SIGNAL 7506- CONDITIONAL EMERGENCY SIGNAL 7507 7508All other orders will be handled completely in user space. 7509 7510Only privileged operation exceptions will be checked for in the kernel (or even 7511in the hardware prior to interception). If this capability is not enabled, the 7512old way of handling SIGP orders is used (partially in kernel and user space). 7513 75147.3 KVM_CAP_S390_VECTOR_REGISTERS 7515--------------------------------- 7516 7517:Architectures: s390 7518:Parameters: none 7519:Returns: 0 on success, negative value on error 7520 7521Allows use of the vector registers introduced with z13 processor, and 7522provides for the synchronization between host and user space. Will 7523return -EINVAL if the machine does not support vectors. 7524 75257.4 KVM_CAP_S390_USER_STSI 7526-------------------------- 7527 7528:Architectures: s390 7529:Parameters: none 7530 7531This capability allows post-handlers for the STSI instruction. After 7532initial handling in the kernel, KVM exits to user space with 7533KVM_EXIT_S390_STSI to allow user space to insert further data. 7534 7535Before exiting to userspace, kvm handlers should fill in s390_stsi field of 7536vcpu->run:: 7537 7538 struct { 7539 __u64 addr; 7540 __u8 ar; 7541 __u8 reserved; 7542 __u8 fc; 7543 __u8 sel1; 7544 __u16 sel2; 7545 } s390_stsi; 7546 7547 @addr - guest address of STSI SYSIB 7548 @fc - function code 7549 @sel1 - selector 1 7550 @sel2 - selector 2 7551 @ar - access register number 7552 7553KVM handlers should exit to userspace with rc = -EREMOTE. 7554 75557.5 KVM_CAP_SPLIT_IRQCHIP 7556------------------------- 7557 7558:Architectures: x86 7559:Parameters: args[0] - number of routes reserved for userspace IOAPICs 7560:Returns: 0 on success, -1 on error 7561 7562Create a local apic for each processor in the kernel. This can be used 7563instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the 7564IOAPIC and PIC (and also the PIT, even though this has to be enabled 7565separately). 7566 7567This capability also enables in kernel routing of interrupt requests; 7568when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are 7569used in the IRQ routing table. The first args[0] MSI routes are reserved 7570for the IOAPIC pins. Whenever the LAPIC receives an EOI for these routes, 7571a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace. 7572 7573Fails if VCPU has already been created, or if the irqchip is already in the 7574kernel (i.e. KVM_CREATE_IRQCHIP has already been called). 7575 75767.6 KVM_CAP_S390_RI 7577------------------- 7578 7579:Architectures: s390 7580:Parameters: none 7581 7582Allows use of runtime-instrumentation introduced with zEC12 processor. 7583Will return -EINVAL if the machine does not support runtime-instrumentation. 7584Will return -EBUSY if a VCPU has already been created. 7585 75867.7 KVM_CAP_X2APIC_API 7587---------------------- 7588 7589:Architectures: x86 7590:Parameters: args[0] - features that should be enabled 7591:Returns: 0 on success, -EINVAL when args[0] contains invalid features 7592 7593Valid feature flags in args[0] are:: 7594 7595 #define KVM_X2APIC_API_USE_32BIT_IDS (1ULL << 0) 7596 #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK (1ULL << 1) 7597 7598Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of 7599KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC, 7600allowing the use of 32-bit APIC IDs. See KVM_CAP_X2APIC_API in their 7601respective sections. 7602 7603KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work 7604in logical mode or with more than 255 VCPUs. Otherwise, KVM treats 0xff 7605as a broadcast even in x2APIC mode in order to support physical x2APIC 7606without interrupt remapping. This is undesirable in logical mode, 7607where 0xff represents CPUs 0-7 in cluster 0. 7608 76097.8 KVM_CAP_S390_USER_INSTR0 7610---------------------------- 7611 7612:Architectures: s390 7613:Parameters: none 7614 7615With this capability enabled, all illegal instructions 0x0000 (2 bytes) will 7616be intercepted and forwarded to user space. User space can use this 7617mechanism e.g. to realize 2-byte software breakpoints. The kernel will 7618not inject an operating exception for these instructions, user space has 7619to take care of that. 7620 7621This capability can be enabled dynamically even if VCPUs were already 7622created and are running. 7623 76247.9 KVM_CAP_S390_GS 7625------------------- 7626 7627:Architectures: s390 7628:Parameters: none 7629:Returns: 0 on success; -EINVAL if the machine does not support 7630 guarded storage; -EBUSY if a VCPU has already been created. 7631 7632Allows use of guarded storage for the KVM guest. 7633 76347.10 KVM_CAP_S390_AIS 7635--------------------- 7636 7637:Architectures: s390 7638:Parameters: none 7639 7640Allow use of adapter-interruption suppression. 7641:Returns: 0 on success; -EBUSY if a VCPU has already been created. 7642 76437.11 KVM_CAP_PPC_SMT 7644-------------------- 7645 7646:Architectures: ppc 7647:Parameters: vsmt_mode, flags 7648 7649Enabling this capability on a VM provides userspace with a way to set 7650the desired virtual SMT mode (i.e. the number of virtual CPUs per 7651virtual core). The virtual SMT mode, vsmt_mode, must be a power of 2 7652between 1 and 8. On POWER8, vsmt_mode must also be no greater than 7653the number of threads per subcore for the host. Currently flags must 7654be 0. A successful call to enable this capability will result in 7655vsmt_mode being returned when the KVM_CAP_PPC_SMT capability is 7656subsequently queried for the VM. This capability is only supported by 7657HV KVM, and can only be set before any VCPUs have been created. 7658The KVM_CAP_PPC_SMT_POSSIBLE capability indicates which virtual SMT 7659modes are available. 7660 76617.12 KVM_CAP_PPC_FWNMI 7662---------------------- 7663 7664:Architectures: ppc 7665:Parameters: none 7666 7667With this capability a machine check exception in the guest address 7668space will cause KVM to exit the guest with NMI exit reason. This 7669enables QEMU to build error log and branch to guest kernel registered 7670machine check handling routine. Without this capability KVM will 7671branch to guests' 0x200 interrupt vector. 7672 76737.13 KVM_CAP_X86_DISABLE_EXITS 7674------------------------------ 7675 7676:Architectures: x86 7677:Parameters: args[0] defines which exits are disabled 7678:Returns: 0 on success, -EINVAL when args[0] contains invalid exits 7679 or if any vCPUs have already been created 7680 7681Valid bits in args[0] are:: 7682 7683 #define KVM_X86_DISABLE_EXITS_MWAIT (1 << 0) 7684 #define KVM_X86_DISABLE_EXITS_HLT (1 << 1) 7685 #define KVM_X86_DISABLE_EXITS_PAUSE (1 << 2) 7686 #define KVM_X86_DISABLE_EXITS_CSTATE (1 << 3) 7687 7688Enabling this capability on a VM provides userspace with a way to no 7689longer intercept some instructions for improved latency in some 7690workloads, and is suggested when vCPUs are associated to dedicated 7691physical CPUs. More bits can be added in the future; userspace can 7692just pass the KVM_CHECK_EXTENSION result to KVM_ENABLE_CAP to disable 7693all such vmexits. 7694 7695Do not enable KVM_FEATURE_PV_UNHALT if you disable HLT exits. 7696 76977.14 KVM_CAP_S390_HPAGE_1M 7698-------------------------- 7699 7700:Architectures: s390 7701:Parameters: none 7702:Returns: 0 on success, -EINVAL if hpage module parameter was not set 7703 or cmma is enabled, or the VM has the KVM_VM_S390_UCONTROL 7704 flag set 7705 7706With this capability the KVM support for memory backing with 1m pages 7707through hugetlbfs can be enabled for a VM. After the capability is 7708enabled, cmma can't be enabled anymore and pfmfi and the storage key 7709interpretation are disabled. If cmma has already been enabled or the 7710hpage module parameter is not set to 1, -EINVAL is returned. 7711 7712While it is generally possible to create a huge page backed VM without 7713this capability, the VM will not be able to run. 7714 77157.15 KVM_CAP_MSR_PLATFORM_INFO 7716------------------------------ 7717 7718:Architectures: x86 7719:Parameters: args[0] whether feature should be enabled or not 7720 7721With this capability, a guest may read the MSR_PLATFORM_INFO MSR. Otherwise, 7722a #GP would be raised when the guest tries to access. Currently, this 7723capability does not enable write permissions of this MSR for the guest. 7724 77257.16 KVM_CAP_PPC_NESTED_HV 7726-------------------------- 7727 7728:Architectures: ppc 7729:Parameters: none 7730:Returns: 0 on success, -EINVAL when the implementation doesn't support 7731 nested-HV virtualization. 7732 7733HV-KVM on POWER9 and later systems allows for "nested-HV" 7734virtualization, which provides a way for a guest VM to run guests that 7735can run using the CPU's supervisor mode (privileged non-hypervisor 7736state). Enabling this capability on a VM depends on the CPU having 7737the necessary functionality and on the facility being enabled with a 7738kvm-hv module parameter. 7739 77407.17 KVM_CAP_EXCEPTION_PAYLOAD 7741------------------------------ 7742 7743:Architectures: x86 7744:Parameters: args[0] whether feature should be enabled or not 7745 7746With this capability enabled, CR2 will not be modified prior to the 7747emulated VM-exit when L1 intercepts a #PF exception that occurs in 7748L2. Similarly, for kvm-intel only, DR6 will not be modified prior to 7749the emulated VM-exit when L1 intercepts a #DB exception that occurs in 7750L2. As a result, when KVM_GET_VCPU_EVENTS reports a pending #PF (or 7751#DB) exception for L2, exception.has_payload will be set and the 7752faulting address (or the new DR6 bits*) will be reported in the 7753exception_payload field. Similarly, when userspace injects a #PF (or 7754#DB) into L2 using KVM_SET_VCPU_EVENTS, it is expected to set 7755exception.has_payload and to put the faulting address - or the new DR6 7756bits\ [#]_ - in the exception_payload field. 7757 7758This capability also enables exception.pending in struct 7759kvm_vcpu_events, which allows userspace to distinguish between pending 7760and injected exceptions. 7761 7762 7763.. [#] For the new DR6 bits, note that bit 16 is set iff the #DB exception 7764 will clear DR6.RTM. 7765 77667.18 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 7767-------------------------------------- 7768 7769:Architectures: x86, arm64, mips 7770:Parameters: args[0] whether feature should be enabled or not 7771 7772Valid flags are:: 7773 7774 #define KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (1 << 0) 7775 #define KVM_DIRTY_LOG_INITIALLY_SET (1 << 1) 7776 7777With KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE is set, KVM_GET_DIRTY_LOG will not 7778automatically clear and write-protect all pages that are returned as dirty. 7779Rather, userspace will have to do this operation separately using 7780KVM_CLEAR_DIRTY_LOG. 7781 7782At the cost of a slightly more complicated operation, this provides better 7783scalability and responsiveness for two reasons. First, 7784KVM_CLEAR_DIRTY_LOG ioctl can operate on a 64-page granularity rather 7785than requiring to sync a full memslot; this ensures that KVM does not 7786take spinlocks for an extended period of time. Second, in some cases a 7787large amount of time can pass between a call to KVM_GET_DIRTY_LOG and 7788userspace actually using the data in the page. Pages can be modified 7789during this time, which is inefficient for both the guest and userspace: 7790the guest will incur a higher penalty due to write protection faults, 7791while userspace can see false reports of dirty pages. Manual reprotection 7792helps reducing this time, improving guest performance and reducing the 7793number of dirty log false positives. 7794 7795With KVM_DIRTY_LOG_INITIALLY_SET set, all the bits of the dirty bitmap 7796will be initialized to 1 when created. This also improves performance because 7797dirty logging can be enabled gradually in small chunks on the first call 7798to KVM_CLEAR_DIRTY_LOG. KVM_DIRTY_LOG_INITIALLY_SET depends on 7799KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (it is also only available on 7800x86 and arm64 for now). 7801 7802KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 was previously available under the name 7803KVM_CAP_MANUAL_DIRTY_LOG_PROTECT, but the implementation had bugs that make 7804it hard or impossible to use it correctly. The availability of 7805KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 signals that those bugs are fixed. 7806Userspace should not try to use KVM_CAP_MANUAL_DIRTY_LOG_PROTECT. 7807 78087.19 KVM_CAP_PPC_SECURE_GUEST 7809------------------------------ 7810 7811:Architectures: ppc 7812 7813This capability indicates that KVM is running on a host that has 7814ultravisor firmware and thus can support a secure guest. On such a 7815system, a guest can ask the ultravisor to make it a secure guest, 7816one whose memory is inaccessible to the host except for pages which 7817are explicitly requested to be shared with the host. The ultravisor 7818notifies KVM when a guest requests to become a secure guest, and KVM 7819has the opportunity to veto the transition. 7820 7821If present, this capability can be enabled for a VM, meaning that KVM 7822will allow the transition to secure guest mode. Otherwise KVM will 7823veto the transition. 7824 78257.20 KVM_CAP_HALT_POLL 7826---------------------- 7827 7828:Architectures: all 7829:Target: VM 7830:Parameters: args[0] is the maximum poll time in nanoseconds 7831:Returns: 0 on success; -1 on error 7832 7833KVM_CAP_HALT_POLL overrides the kvm.halt_poll_ns module parameter to set the 7834maximum halt-polling time for all vCPUs in the target VM. This capability can 7835be invoked at any time and any number of times to dynamically change the 7836maximum halt-polling time. 7837 7838See Documentation/virt/kvm/halt-polling.rst for more information on halt 7839polling. 7840 78417.21 KVM_CAP_X86_USER_SPACE_MSR 7842------------------------------- 7843 7844:Architectures: x86 7845:Target: VM 7846:Parameters: args[0] contains the mask of KVM_MSR_EXIT_REASON_* events to report 7847:Returns: 0 on success; -1 on error 7848 7849This capability allows userspace to intercept RDMSR and WRMSR instructions if 7850access to an MSR is denied. By default, KVM injects #GP on denied accesses. 7851 7852When a guest requests to read or write an MSR, KVM may not implement all MSRs 7853that are relevant to a respective system. It also does not differentiate by 7854CPU type. 7855 7856To allow more fine grained control over MSR handling, userspace may enable 7857this capability. With it enabled, MSR accesses that match the mask specified in 7858args[0] and would trigger a #GP inside the guest will instead trigger 7859KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR exit notifications. Userspace 7860can then implement model specific MSR handling and/or user notifications 7861to inform a user that an MSR was not emulated/virtualized by KVM. 7862 7863The valid mask flags are: 7864 7865============================ =============================================== 7866 KVM_MSR_EXIT_REASON_UNKNOWN intercept accesses to unknown (to KVM) MSRs 7867 KVM_MSR_EXIT_REASON_INVAL intercept accesses that are architecturally 7868 invalid according to the vCPU model and/or mode 7869 KVM_MSR_EXIT_REASON_FILTER intercept accesses that are denied by userspace 7870 via KVM_X86_SET_MSR_FILTER 7871============================ =============================================== 7872 78737.22 KVM_CAP_X86_BUS_LOCK_EXIT 7874------------------------------- 7875 7876:Architectures: x86 7877:Target: VM 7878:Parameters: args[0] defines the policy used when bus locks detected in guest 7879:Returns: 0 on success, -EINVAL when args[0] contains invalid bits 7880 7881Valid bits in args[0] are:: 7882 7883 #define KVM_BUS_LOCK_DETECTION_OFF (1 << 0) 7884 #define KVM_BUS_LOCK_DETECTION_EXIT (1 << 1) 7885 7886Enabling this capability on a VM provides userspace with a way to select a 7887policy to handle the bus locks detected in guest. Userspace can obtain the 7888supported modes from the result of KVM_CHECK_EXTENSION and define it through 7889the KVM_ENABLE_CAP. The supported modes are mutually-exclusive. 7890 7891This capability allows userspace to force VM exits on bus locks detected in the 7892guest, irrespective whether or not the host has enabled split-lock detection 7893(which triggers an #AC exception that KVM intercepts). This capability is 7894intended to mitigate attacks where a malicious/buggy guest can exploit bus 7895locks to degrade the performance of the whole system. 7896 7897If KVM_BUS_LOCK_DETECTION_OFF is set, KVM doesn't force guest bus locks to VM 7898exit, although the host kernel's split-lock #AC detection still applies, if 7899enabled. 7900 7901If KVM_BUS_LOCK_DETECTION_EXIT is set, KVM enables a CPU feature that ensures 7902bus locks in the guest trigger a VM exit, and KVM exits to userspace for all 7903such VM exits, e.g. to allow userspace to throttle the offending guest and/or 7904apply some other policy-based mitigation. When exiting to userspace, KVM sets 7905KVM_RUN_X86_BUS_LOCK in vcpu-run->flags, and conditionally sets the exit_reason 7906to KVM_EXIT_X86_BUS_LOCK. 7907 7908Note! Detected bus locks may be coincident with other exits to userspace, i.e. 7909KVM_RUN_X86_BUS_LOCK should be checked regardless of the primary exit reason if 7910userspace wants to take action on all detected bus locks. 7911 79127.23 KVM_CAP_PPC_DAWR1 7913---------------------- 7914 7915:Architectures: ppc 7916:Parameters: none 7917:Returns: 0 on success, -EINVAL when CPU doesn't support 2nd DAWR 7918 7919This capability can be used to check / enable 2nd DAWR feature provided 7920by POWER10 processor. 7921 7922 79237.24 KVM_CAP_VM_COPY_ENC_CONTEXT_FROM 7924------------------------------------- 7925 7926Architectures: x86 SEV enabled 7927Type: vm 7928Parameters: args[0] is the fd of the source vm 7929Returns: 0 on success; ENOTTY on error 7930 7931This capability enables userspace to copy encryption context from the vm 7932indicated by the fd to the vm this is called on. 7933 7934This is intended to support in-guest workloads scheduled by the host. This 7935allows the in-guest workload to maintain its own NPTs and keeps the two vms 7936from accidentally clobbering each other with interrupts and the like (separate 7937APIC/MSRs/etc). 7938 79397.25 KVM_CAP_SGX_ATTRIBUTE 7940-------------------------- 7941 7942:Architectures: x86 7943:Target: VM 7944:Parameters: args[0] is a file handle of a SGX attribute file in securityfs 7945:Returns: 0 on success, -EINVAL if the file handle is invalid or if a requested 7946 attribute is not supported by KVM. 7947 7948KVM_CAP_SGX_ATTRIBUTE enables a userspace VMM to grant a VM access to one or 7949more privileged enclave attributes. args[0] must hold a file handle to a valid 7950SGX attribute file corresponding to an attribute that is supported/restricted 7951by KVM (currently only PROVISIONKEY). 7952 7953The SGX subsystem restricts access to a subset of enclave attributes to provide 7954additional security for an uncompromised kernel, e.g. use of the PROVISIONKEY 7955is restricted to deter malware from using the PROVISIONKEY to obtain a stable 7956system fingerprint. To prevent userspace from circumventing such restrictions 7957by running an enclave in a VM, KVM prevents access to privileged attributes by 7958default. 7959 7960See Documentation/arch/x86/sgx.rst for more details. 7961 79627.26 KVM_CAP_PPC_RPT_INVALIDATE 7963------------------------------- 7964 7965:Capability: KVM_CAP_PPC_RPT_INVALIDATE 7966:Architectures: ppc 7967:Type: vm 7968 7969This capability indicates that the kernel is capable of handling 7970H_RPT_INVALIDATE hcall. 7971 7972In order to enable the use of H_RPT_INVALIDATE in the guest, 7973user space might have to advertise it for the guest. For example, 7974IBM pSeries (sPAPR) guest starts using it if "hcall-rpt-invalidate" is 7975present in the "ibm,hypertas-functions" device-tree property. 7976 7977This capability is enabled for hypervisors on platforms like POWER9 7978that support radix MMU. 7979 79807.27 KVM_CAP_EXIT_ON_EMULATION_FAILURE 7981-------------------------------------- 7982 7983:Architectures: x86 7984:Parameters: args[0] whether the feature should be enabled or not 7985 7986When this capability is enabled, an emulation failure will result in an exit 7987to userspace with KVM_INTERNAL_ERROR (except when the emulator was invoked 7988to handle a VMware backdoor instruction). Furthermore, KVM will now provide up 7989to 15 instruction bytes for any exit to userspace resulting from an emulation 7990failure. When these exits to userspace occur use the emulation_failure struct 7991instead of the internal struct. They both have the same layout, but the 7992emulation_failure struct matches the content better. It also explicitly 7993defines the 'flags' field which is used to describe the fields in the struct 7994that are valid (ie: if KVM_INTERNAL_ERROR_EMULATION_FLAG_INSTRUCTION_BYTES is 7995set in the 'flags' field then both 'insn_size' and 'insn_bytes' have valid data 7996in them.) 7997 79987.28 KVM_CAP_ARM_MTE 7999-------------------- 8000 8001:Architectures: arm64 8002:Parameters: none 8003 8004This capability indicates that KVM (and the hardware) supports exposing the 8005Memory Tagging Extensions (MTE) to the guest. It must also be enabled by the 8006VMM before creating any VCPUs to allow the guest access. Note that MTE is only 8007available to a guest running in AArch64 mode and enabling this capability will 8008cause attempts to create AArch32 VCPUs to fail. 8009 8010When enabled the guest is able to access tags associated with any memory given 8011to the guest. KVM will ensure that the tags are maintained during swap or 8012hibernation of the host; however the VMM needs to manually save/restore the 8013tags as appropriate if the VM is migrated. 8014 8015When this capability is enabled all memory in memslots must be mapped as 8016``MAP_ANONYMOUS`` or with a RAM-based file mapping (``tmpfs``, ``memfd``), 8017attempts to create a memslot with an invalid mmap will result in an 8018-EINVAL return. 8019 8020When enabled the VMM may make use of the ``KVM_ARM_MTE_COPY_TAGS`` ioctl to 8021perform a bulk copy of tags to/from the guest. 8022 80237.29 KVM_CAP_VM_MOVE_ENC_CONTEXT_FROM 8024------------------------------------- 8025 8026:Architectures: x86 SEV enabled 8027:Type: vm 8028:Parameters: args[0] is the fd of the source vm 8029:Returns: 0 on success 8030 8031This capability enables userspace to migrate the encryption context from the VM 8032indicated by the fd to the VM this is called on. 8033 8034This is intended to support intra-host migration of VMs between userspace VMMs, 8035upgrading the VMM process without interrupting the guest. 8036 80377.30 KVM_CAP_PPC_AIL_MODE_3 8038------------------------------- 8039 8040:Capability: KVM_CAP_PPC_AIL_MODE_3 8041:Architectures: ppc 8042:Type: vm 8043 8044This capability indicates that the kernel supports the mode 3 setting for the 8045"Address Translation Mode on Interrupt" aka "Alternate Interrupt Location" 8046resource that is controlled with the H_SET_MODE hypercall. 8047 8048This capability allows a guest kernel to use a better-performance mode for 8049handling interrupts and system calls. 8050 80517.31 KVM_CAP_DISABLE_QUIRKS2 8052---------------------------- 8053 8054:Capability: KVM_CAP_DISABLE_QUIRKS2 8055:Parameters: args[0] - set of KVM quirks to disable 8056:Architectures: x86 8057:Type: vm 8058 8059This capability, if enabled, will cause KVM to disable some behavior 8060quirks. 8061 8062Calling KVM_CHECK_EXTENSION for this capability returns a bitmask of 8063quirks that can be disabled in KVM. 8064 8065The argument to KVM_ENABLE_CAP for this capability is a bitmask of 8066quirks to disable, and must be a subset of the bitmask returned by 8067KVM_CHECK_EXTENSION. 8068 8069The valid bits in cap.args[0] are: 8070 8071=================================== ============================================ 8072 KVM_X86_QUIRK_LINT0_REENABLED By default, the reset value for the LVT 8073 LINT0 register is 0x700 (APIC_MODE_EXTINT). 8074 When this quirk is disabled, the reset value 8075 is 0x10000 (APIC_LVT_MASKED). 8076 8077 KVM_X86_QUIRK_CD_NW_CLEARED By default, KVM clears CR0.CD and CR0.NW on 8078 AMD CPUs to workaround buggy guest firmware 8079 that runs in perpetuity with CR0.CD, i.e. 8080 with caches in "no fill" mode. 8081 8082 When this quirk is disabled, KVM does not 8083 change the value of CR0.CD and CR0.NW. 8084 8085 KVM_X86_QUIRK_LAPIC_MMIO_HOLE By default, the MMIO LAPIC interface is 8086 available even when configured for x2APIC 8087 mode. When this quirk is disabled, KVM 8088 disables the MMIO LAPIC interface if the 8089 LAPIC is in x2APIC mode. 8090 8091 KVM_X86_QUIRK_OUT_7E_INC_RIP By default, KVM pre-increments %rip before 8092 exiting to userspace for an OUT instruction 8093 to port 0x7e. When this quirk is disabled, 8094 KVM does not pre-increment %rip before 8095 exiting to userspace. 8096 8097 KVM_X86_QUIRK_MISC_ENABLE_NO_MWAIT When this quirk is disabled, KVM sets 8098 CPUID.01H:ECX[bit 3] (MONITOR/MWAIT) if 8099 IA32_MISC_ENABLE[bit 18] (MWAIT) is set. 8100 Additionally, when this quirk is disabled, 8101 KVM clears CPUID.01H:ECX[bit 3] if 8102 IA32_MISC_ENABLE[bit 18] is cleared. 8103 8104 KVM_X86_QUIRK_FIX_HYPERCALL_INSN By default, KVM rewrites guest 8105 VMMCALL/VMCALL instructions to match the 8106 vendor's hypercall instruction for the 8107 system. When this quirk is disabled, KVM 8108 will no longer rewrite invalid guest 8109 hypercall instructions. Executing the 8110 incorrect hypercall instruction will 8111 generate a #UD within the guest. 8112 8113KVM_X86_QUIRK_MWAIT_NEVER_UD_FAULTS By default, KVM emulates MONITOR/MWAIT (if 8114 they are intercepted) as NOPs regardless of 8115 whether or not MONITOR/MWAIT are supported 8116 according to guest CPUID. When this quirk 8117 is disabled and KVM_X86_DISABLE_EXITS_MWAIT 8118 is not set (MONITOR/MWAIT are intercepted), 8119 KVM will inject a #UD on MONITOR/MWAIT if 8120 they're unsupported per guest CPUID. Note, 8121 KVM will modify MONITOR/MWAIT support in 8122 guest CPUID on writes to MISC_ENABLE if 8123 KVM_X86_QUIRK_MISC_ENABLE_NO_MWAIT is 8124 disabled. 8125 8126KVM_X86_QUIRK_SLOT_ZAP_ALL By default, for KVM_X86_DEFAULT_VM VMs, KVM 8127 invalidates all SPTEs in all memslots and 8128 address spaces when a memslot is deleted or 8129 moved. When this quirk is disabled (or the 8130 VM type isn't KVM_X86_DEFAULT_VM), KVM only 8131 ensures the backing memory of the deleted 8132 or moved memslot isn't reachable, i.e KVM 8133 _may_ invalidate only SPTEs related to the 8134 memslot. 8135 8136KVM_X86_QUIRK_STUFF_FEATURE_MSRS By default, at vCPU creation, KVM sets the 8137 vCPU's MSR_IA32_PERF_CAPABILITIES (0x345), 8138 MSR_IA32_ARCH_CAPABILITIES (0x10a), 8139 MSR_PLATFORM_INFO (0xce), and all VMX MSRs 8140 (0x480..0x492) to the maximal capabilities 8141 supported by KVM. KVM also sets 8142 MSR_IA32_UCODE_REV (0x8b) to an arbitrary 8143 value (which is different for Intel vs. 8144 AMD). Lastly, when guest CPUID is set (by 8145 userspace), KVM modifies select VMX MSR 8146 fields to force consistency between guest 8147 CPUID and L2's effective ISA. When this 8148 quirk is disabled, KVM zeroes the vCPU's MSR 8149 values (with two exceptions, see below), 8150 i.e. treats the feature MSRs like CPUID 8151 leaves and gives userspace full control of 8152 the vCPU model definition. This quirk does 8153 not affect VMX MSRs CR0/CR4_FIXED1 (0x487 8154 and 0x489), as KVM does now allow them to 8155 be set by userspace (KVM sets them based on 8156 guest CPUID, for safety purposes). 8157=================================== ============================================ 8158 81597.32 KVM_CAP_MAX_VCPU_ID 8160------------------------ 8161 8162:Architectures: x86 8163:Target: VM 8164:Parameters: args[0] - maximum APIC ID value set for current VM 8165:Returns: 0 on success, -EINVAL if args[0] is beyond KVM_MAX_VCPU_IDS 8166 supported in KVM or if it has been set. 8167 8168This capability allows userspace to specify maximum possible APIC ID 8169assigned for current VM session prior to the creation of vCPUs, saving 8170memory for data structures indexed by the APIC ID. Userspace is able 8171to calculate the limit to APIC ID values from designated 8172CPU topology. 8173 8174The value can be changed only until KVM_ENABLE_CAP is set to a nonzero 8175value or until a vCPU is created. Upon creation of the first vCPU, 8176if the value was set to zero or KVM_ENABLE_CAP was not invoked, KVM 8177uses the return value of KVM_CHECK_EXTENSION(KVM_CAP_MAX_VCPU_ID) as 8178the maximum APIC ID. 8179 81807.33 KVM_CAP_X86_NOTIFY_VMEXIT 8181------------------------------ 8182 8183:Architectures: x86 8184:Target: VM 8185:Parameters: args[0] is the value of notify window as well as some flags 8186:Returns: 0 on success, -EINVAL if args[0] contains invalid flags or notify 8187 VM exit is unsupported. 8188 8189Bits 63:32 of args[0] are used for notify window. 8190Bits 31:0 of args[0] are for some flags. Valid bits are:: 8191 8192 #define KVM_X86_NOTIFY_VMEXIT_ENABLED (1 << 0) 8193 #define KVM_X86_NOTIFY_VMEXIT_USER (1 << 1) 8194 8195This capability allows userspace to configure the notify VM exit on/off 8196in per-VM scope during VM creation. Notify VM exit is disabled by default. 8197When userspace sets KVM_X86_NOTIFY_VMEXIT_ENABLED bit in args[0], VMM will 8198enable this feature with the notify window provided, which will generate 8199a VM exit if no event window occurs in VM non-root mode for a specified of 8200time (notify window). 8201 8202If KVM_X86_NOTIFY_VMEXIT_USER is set in args[0], upon notify VM exits happen, 8203KVM would exit to userspace for handling. 8204 8205This capability is aimed to mitigate the threat that malicious VMs can 8206cause CPU stuck (due to event windows don't open up) and make the CPU 8207unavailable to host or other VMs. 8208 82097.34 KVM_CAP_MEMORY_FAULT_INFO 8210------------------------------ 8211 8212:Architectures: x86 8213:Returns: Informational only, -EINVAL on direct KVM_ENABLE_CAP. 8214 8215The presence of this capability indicates that KVM_RUN will fill 8216kvm_run.memory_fault if KVM cannot resolve a guest page fault VM-Exit, e.g. if 8217there is a valid memslot but no backing VMA for the corresponding host virtual 8218address. 8219 8220The information in kvm_run.memory_fault is valid if and only if KVM_RUN returns 8221an error with errno=EFAULT or errno=EHWPOISON *and* kvm_run.exit_reason is set 8222to KVM_EXIT_MEMORY_FAULT. 8223 8224Note: Userspaces which attempt to resolve memory faults so that they can retry 8225KVM_RUN are encouraged to guard against repeatedly receiving the same 8226error/annotated fault. 8227 8228See KVM_EXIT_MEMORY_FAULT for more information. 8229 82307.35 KVM_CAP_X86_APIC_BUS_CYCLES_NS 8231----------------------------------- 8232 8233:Architectures: x86 8234:Target: VM 8235:Parameters: args[0] is the desired APIC bus clock rate, in nanoseconds 8236:Returns: 0 on success, -EINVAL if args[0] contains an invalid value for the 8237 frequency or if any vCPUs have been created, -ENXIO if a virtual 8238 local APIC has not been created using KVM_CREATE_IRQCHIP. 8239 8240This capability sets the VM's APIC bus clock frequency, used by KVM's in-kernel 8241virtual APIC when emulating APIC timers. KVM's default value can be retrieved 8242by KVM_CHECK_EXTENSION. 8243 8244Note: Userspace is responsible for correctly configuring CPUID 0x15, a.k.a. the 8245core crystal clock frequency, if a non-zero CPUID 0x15 is exposed to the guest. 8246 82477.36 KVM_CAP_X86_GUEST_MODE 8248------------------------------ 8249 8250:Architectures: x86 8251:Returns: Informational only, -EINVAL on direct KVM_ENABLE_CAP. 8252 8253The presence of this capability indicates that KVM_RUN will update the 8254KVM_RUN_X86_GUEST_MODE bit in kvm_run.flags to indicate whether the 8255vCPU was executing nested guest code when it exited. 8256 8257KVM exits with the register state of either the L1 or L2 guest 8258depending on which executed at the time of an exit. Userspace must 8259take care to differentiate between these cases. 8260 82618. Other capabilities. 8262====================== 8263 8264This section lists capabilities that give information about other 8265features of the KVM implementation. 8266 82678.1 KVM_CAP_PPC_HWRNG 8268--------------------- 8269 8270:Architectures: ppc 8271 8272This capability, if KVM_CHECK_EXTENSION indicates that it is 8273available, means that the kernel has an implementation of the 8274H_RANDOM hypercall backed by a hardware random-number generator. 8275If present, the kernel H_RANDOM handler can be enabled for guest use 8276with the KVM_CAP_PPC_ENABLE_HCALL capability. 8277 82788.2 KVM_CAP_HYPERV_SYNIC 8279------------------------ 8280 8281:Architectures: x86 8282 8283This capability, if KVM_CHECK_EXTENSION indicates that it is 8284available, means that the kernel has an implementation of the 8285Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is 8286used to support Windows Hyper-V based guest paravirt drivers(VMBus). 8287 8288In order to use SynIC, it has to be activated by setting this 8289capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this 8290will disable the use of APIC hardware virtualization even if supported 8291by the CPU, as it's incompatible with SynIC auto-EOI behavior. 8292 82938.3 KVM_CAP_PPC_MMU_RADIX 8294------------------------- 8295 8296:Architectures: ppc 8297 8298This capability, if KVM_CHECK_EXTENSION indicates that it is 8299available, means that the kernel can support guests using the 8300radix MMU defined in Power ISA V3.00 (as implemented in the POWER9 8301processor). 8302 83038.4 KVM_CAP_PPC_MMU_HASH_V3 8304--------------------------- 8305 8306:Architectures: ppc 8307 8308This capability, if KVM_CHECK_EXTENSION indicates that it is 8309available, means that the kernel can support guests using the 8310hashed page table MMU defined in Power ISA V3.00 (as implemented in 8311the POWER9 processor), including in-memory segment tables. 8312 83138.5 KVM_CAP_MIPS_VZ 8314------------------- 8315 8316:Architectures: mips 8317 8318This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that 8319it is available, means that full hardware assisted virtualization capabilities 8320of the hardware are available for use through KVM. An appropriate 8321KVM_VM_MIPS_* type must be passed to KVM_CREATE_VM to create a VM which 8322utilises it. 8323 8324If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is 8325available, it means that the VM is using full hardware assisted virtualization 8326capabilities of the hardware. This is useful to check after creating a VM with 8327KVM_VM_MIPS_DEFAULT. 8328 8329The value returned by KVM_CHECK_EXTENSION should be compared against known 8330values (see below). All other values are reserved. This is to allow for the 8331possibility of other hardware assisted virtualization implementations which 8332may be incompatible with the MIPS VZ ASE. 8333 8334== ========================================================================== 8335 0 The trap & emulate implementation is in use to run guest code in user 8336 mode. Guest virtual memory segments are rearranged to fit the guest in the 8337 user mode address space. 8338 8339 1 The MIPS VZ ASE is in use, providing full hardware assisted 8340 virtualization, including standard guest virtual memory segments. 8341== ========================================================================== 8342 83438.6 KVM_CAP_MIPS_TE 8344------------------- 8345 8346:Architectures: mips 8347 8348This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that 8349it is available, means that the trap & emulate implementation is available to 8350run guest code in user mode, even if KVM_CAP_MIPS_VZ indicates that hardware 8351assisted virtualisation is also available. KVM_VM_MIPS_TE (0) must be passed 8352to KVM_CREATE_VM to create a VM which utilises it. 8353 8354If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is 8355available, it means that the VM is using trap & emulate. 8356 83578.7 KVM_CAP_MIPS_64BIT 8358---------------------- 8359 8360:Architectures: mips 8361 8362This capability indicates the supported architecture type of the guest, i.e. the 8363supported register and address width. 8364 8365The values returned when this capability is checked by KVM_CHECK_EXTENSION on a 8366kvm VM handle correspond roughly to the CP0_Config.AT register field, and should 8367be checked specifically against known values (see below). All other values are 8368reserved. 8369 8370== ======================================================================== 8371 0 MIPS32 or microMIPS32. 8372 Both registers and addresses are 32-bits wide. 8373 It will only be possible to run 32-bit guest code. 8374 8375 1 MIPS64 or microMIPS64 with access only to 32-bit compatibility segments. 8376 Registers are 64-bits wide, but addresses are 32-bits wide. 8377 64-bit guest code may run but cannot access MIPS64 memory segments. 8378 It will also be possible to run 32-bit guest code. 8379 8380 2 MIPS64 or microMIPS64 with access to all address segments. 8381 Both registers and addresses are 64-bits wide. 8382 It will be possible to run 64-bit or 32-bit guest code. 8383== ======================================================================== 8384 83858.9 KVM_CAP_ARM_USER_IRQ 8386------------------------ 8387 8388:Architectures: arm64 8389 8390This capability, if KVM_CHECK_EXTENSION indicates that it is available, means 8391that if userspace creates a VM without an in-kernel interrupt controller, it 8392will be notified of changes to the output level of in-kernel emulated devices, 8393which can generate virtual interrupts, presented to the VM. 8394For such VMs, on every return to userspace, the kernel 8395updates the vcpu's run->s.regs.device_irq_level field to represent the actual 8396output level of the device. 8397 8398Whenever kvm detects a change in the device output level, kvm guarantees at 8399least one return to userspace before running the VM. This exit could either 8400be a KVM_EXIT_INTR or any other exit event, like KVM_EXIT_MMIO. This way, 8401userspace can always sample the device output level and re-compute the state of 8402the userspace interrupt controller. Userspace should always check the state 8403of run->s.regs.device_irq_level on every kvm exit. 8404The value in run->s.regs.device_irq_level can represent both level and edge 8405triggered interrupt signals, depending on the device. Edge triggered interrupt 8406signals will exit to userspace with the bit in run->s.regs.device_irq_level 8407set exactly once per edge signal. 8408 8409The field run->s.regs.device_irq_level is available independent of 8410run->kvm_valid_regs or run->kvm_dirty_regs bits. 8411 8412If KVM_CAP_ARM_USER_IRQ is supported, the KVM_CHECK_EXTENSION ioctl returns a 8413number larger than 0 indicating the version of this capability is implemented 8414and thereby which bits in run->s.regs.device_irq_level can signal values. 8415 8416Currently the following bits are defined for the device_irq_level bitmap:: 8417 8418 KVM_CAP_ARM_USER_IRQ >= 1: 8419 8420 KVM_ARM_DEV_EL1_VTIMER - EL1 virtual timer 8421 KVM_ARM_DEV_EL1_PTIMER - EL1 physical timer 8422 KVM_ARM_DEV_PMU - ARM PMU overflow interrupt signal 8423 8424Future versions of kvm may implement additional events. These will get 8425indicated by returning a higher number from KVM_CHECK_EXTENSION and will be 8426listed above. 8427 84288.10 KVM_CAP_PPC_SMT_POSSIBLE 8429----------------------------- 8430 8431:Architectures: ppc 8432 8433Querying this capability returns a bitmap indicating the possible 8434virtual SMT modes that can be set using KVM_CAP_PPC_SMT. If bit N 8435(counting from the right) is set, then a virtual SMT mode of 2^N is 8436available. 8437 84388.11 KVM_CAP_HYPERV_SYNIC2 8439-------------------------- 8440 8441:Architectures: x86 8442 8443This capability enables a newer version of Hyper-V Synthetic interrupt 8444controller (SynIC). The only difference with KVM_CAP_HYPERV_SYNIC is that KVM 8445doesn't clear SynIC message and event flags pages when they are enabled by 8446writing to the respective MSRs. 8447 84488.12 KVM_CAP_HYPERV_VP_INDEX 8449---------------------------- 8450 8451:Architectures: x86 8452 8453This capability indicates that userspace can load HV_X64_MSR_VP_INDEX msr. Its 8454value is used to denote the target vcpu for a SynIC interrupt. For 8455compatibility, KVM initializes this msr to KVM's internal vcpu index. When this 8456capability is absent, userspace can still query this msr's value. 8457 84588.13 KVM_CAP_S390_AIS_MIGRATION 8459------------------------------- 8460 8461:Architectures: s390 8462:Parameters: none 8463 8464This capability indicates if the flic device will be able to get/set the 8465AIS states for migration via the KVM_DEV_FLIC_AISM_ALL attribute and allows 8466to discover this without having to create a flic device. 8467 84688.14 KVM_CAP_S390_PSW 8469--------------------- 8470 8471:Architectures: s390 8472 8473This capability indicates that the PSW is exposed via the kvm_run structure. 8474 84758.15 KVM_CAP_S390_GMAP 8476---------------------- 8477 8478:Architectures: s390 8479 8480This capability indicates that the user space memory used as guest mapping can 8481be anywhere in the user memory address space, as long as the memory slots are 8482aligned and sized to a segment (1MB) boundary. 8483 84848.16 KVM_CAP_S390_COW 8485--------------------- 8486 8487:Architectures: s390 8488 8489This capability indicates that the user space memory used as guest mapping can 8490use copy-on-write semantics as well as dirty pages tracking via read-only page 8491tables. 8492 84938.17 KVM_CAP_S390_BPB 8494--------------------- 8495 8496:Architectures: s390 8497 8498This capability indicates that kvm will implement the interfaces to handle 8499reset, migration and nested KVM for branch prediction blocking. The stfle 8500facility 82 should not be provided to the guest without this capability. 8501 85028.18 KVM_CAP_HYPERV_TLBFLUSH 8503---------------------------- 8504 8505:Architectures: x86 8506 8507This capability indicates that KVM supports paravirtualized Hyper-V TLB Flush 8508hypercalls: 8509HvFlushVirtualAddressSpace, HvFlushVirtualAddressSpaceEx, 8510HvFlushVirtualAddressList, HvFlushVirtualAddressListEx. 8511 85128.19 KVM_CAP_ARM_INJECT_SERROR_ESR 8513---------------------------------- 8514 8515:Architectures: arm64 8516 8517This capability indicates that userspace can specify (via the 8518KVM_SET_VCPU_EVENTS ioctl) the syndrome value reported to the guest when it 8519takes a virtual SError interrupt exception. 8520If KVM advertises this capability, userspace can only specify the ISS field for 8521the ESR syndrome. Other parts of the ESR, such as the EC are generated by the 8522CPU when the exception is taken. If this virtual SError is taken to EL1 using 8523AArch64, this value will be reported in the ISS field of ESR_ELx. 8524 8525See KVM_CAP_VCPU_EVENTS for more details. 8526 85278.20 KVM_CAP_HYPERV_SEND_IPI 8528---------------------------- 8529 8530:Architectures: x86 8531 8532This capability indicates that KVM supports paravirtualized Hyper-V IPI send 8533hypercalls: 8534HvCallSendSyntheticClusterIpi, HvCallSendSyntheticClusterIpiEx. 8535 85368.21 KVM_CAP_HYPERV_DIRECT_TLBFLUSH 8537----------------------------------- 8538 8539:Architectures: x86 8540 8541This capability indicates that KVM running on top of Hyper-V hypervisor 8542enables Direct TLB flush for its guests meaning that TLB flush 8543hypercalls are handled by Level 0 hypervisor (Hyper-V) bypassing KVM. 8544Due to the different ABI for hypercall parameters between Hyper-V and 8545KVM, enabling this capability effectively disables all hypercall 8546handling by KVM (as some KVM hypercall may be mistakenly treated as TLB 8547flush hypercalls by Hyper-V) so userspace should disable KVM identification 8548in CPUID and only exposes Hyper-V identification. In this case, guest 8549thinks it's running on Hyper-V and only use Hyper-V hypercalls. 8550 85518.22 KVM_CAP_S390_VCPU_RESETS 8552----------------------------- 8553 8554:Architectures: s390 8555 8556This capability indicates that the KVM_S390_NORMAL_RESET and 8557KVM_S390_CLEAR_RESET ioctls are available. 8558 85598.23 KVM_CAP_S390_PROTECTED 8560--------------------------- 8561 8562:Architectures: s390 8563 8564This capability indicates that the Ultravisor has been initialized and 8565KVM can therefore start protected VMs. 8566This capability governs the KVM_S390_PV_COMMAND ioctl and the 8567KVM_MP_STATE_LOAD MP_STATE. KVM_SET_MP_STATE can fail for protected 8568guests when the state change is invalid. 8569 85708.24 KVM_CAP_STEAL_TIME 8571----------------------- 8572 8573:Architectures: arm64, x86 8574 8575This capability indicates that KVM supports steal time accounting. 8576When steal time accounting is supported it may be enabled with 8577architecture-specific interfaces. This capability and the architecture- 8578specific interfaces must be consistent, i.e. if one says the feature 8579is supported, than the other should as well and vice versa. For arm64 8580see Documentation/virt/kvm/devices/vcpu.rst "KVM_ARM_VCPU_PVTIME_CTRL". 8581For x86 see Documentation/virt/kvm/x86/msr.rst "MSR_KVM_STEAL_TIME". 8582 85838.25 KVM_CAP_S390_DIAG318 8584------------------------- 8585 8586:Architectures: s390 8587 8588This capability enables a guest to set information about its control program 8589(i.e. guest kernel type and version). The information is helpful during 8590system/firmware service events, providing additional data about the guest 8591environments running on the machine. 8592 8593The information is associated with the DIAGNOSE 0x318 instruction, which sets 8594an 8-byte value consisting of a one-byte Control Program Name Code (CPNC) and 8595a 7-byte Control Program Version Code (CPVC). The CPNC determines what 8596environment the control program is running in (e.g. Linux, z/VM...), and the 8597CPVC is used for information specific to OS (e.g. Linux version, Linux 8598distribution...) 8599 8600If this capability is available, then the CPNC and CPVC can be synchronized 8601between KVM and userspace via the sync regs mechanism (KVM_SYNC_DIAG318). 8602 86038.26 KVM_CAP_X86_USER_SPACE_MSR 8604------------------------------- 8605 8606:Architectures: x86 8607 8608This capability indicates that KVM supports deflection of MSR reads and 8609writes to user space. It can be enabled on a VM level. If enabled, MSR 8610accesses that would usually trigger a #GP by KVM into the guest will 8611instead get bounced to user space through the KVM_EXIT_X86_RDMSR and 8612KVM_EXIT_X86_WRMSR exit notifications. 8613 86148.27 KVM_CAP_X86_MSR_FILTER 8615--------------------------- 8616 8617:Architectures: x86 8618 8619This capability indicates that KVM supports that accesses to user defined MSRs 8620may be rejected. With this capability exposed, KVM exports new VM ioctl 8621KVM_X86_SET_MSR_FILTER which user space can call to specify bitmaps of MSR 8622ranges that KVM should deny access to. 8623 8624In combination with KVM_CAP_X86_USER_SPACE_MSR, this allows user space to 8625trap and emulate MSRs that are outside of the scope of KVM as well as 8626limit the attack surface on KVM's MSR emulation code. 8627 86288.28 KVM_CAP_ENFORCE_PV_FEATURE_CPUID 8629------------------------------------- 8630 8631Architectures: x86 8632 8633When enabled, KVM will disable paravirtual features provided to the 8634guest according to the bits in the KVM_CPUID_FEATURES CPUID leaf 8635(0x40000001). Otherwise, a guest may use the paravirtual features 8636regardless of what has actually been exposed through the CPUID leaf. 8637 8638.. _KVM_CAP_DIRTY_LOG_RING: 8639 86408.29 KVM_CAP_DIRTY_LOG_RING/KVM_CAP_DIRTY_LOG_RING_ACQ_REL 8641---------------------------------------------------------- 8642 8643:Architectures: x86, arm64 8644:Parameters: args[0] - size of the dirty log ring 8645 8646KVM is capable of tracking dirty memory using ring buffers that are 8647mmapped into userspace; there is one dirty ring per vcpu. 8648 8649The dirty ring is available to userspace as an array of 8650``struct kvm_dirty_gfn``. Each dirty entry is defined as:: 8651 8652 struct kvm_dirty_gfn { 8653 __u32 flags; 8654 __u32 slot; /* as_id | slot_id */ 8655 __u64 offset; 8656 }; 8657 8658The following values are defined for the flags field to define the 8659current state of the entry:: 8660 8661 #define KVM_DIRTY_GFN_F_DIRTY BIT(0) 8662 #define KVM_DIRTY_GFN_F_RESET BIT(1) 8663 #define KVM_DIRTY_GFN_F_MASK 0x3 8664 8665Userspace should call KVM_ENABLE_CAP ioctl right after KVM_CREATE_VM 8666ioctl to enable this capability for the new guest and set the size of 8667the rings. Enabling the capability is only allowed before creating any 8668vCPU, and the size of the ring must be a power of two. The larger the 8669ring buffer, the less likely the ring is full and the VM is forced to 8670exit to userspace. The optimal size depends on the workload, but it is 8671recommended that it be at least 64 KiB (4096 entries). 8672 8673Just like for dirty page bitmaps, the buffer tracks writes to 8674all user memory regions for which the KVM_MEM_LOG_DIRTY_PAGES flag was 8675set in KVM_SET_USER_MEMORY_REGION. Once a memory region is registered 8676with the flag set, userspace can start harvesting dirty pages from the 8677ring buffer. 8678 8679An entry in the ring buffer can be unused (flag bits ``00``), 8680dirty (flag bits ``01``) or harvested (flag bits ``1X``). The 8681state machine for the entry is as follows:: 8682 8683 dirtied harvested reset 8684 00 -----------> 01 -------------> 1X -------+ 8685 ^ | 8686 | | 8687 +------------------------------------------+ 8688 8689To harvest the dirty pages, userspace accesses the mmapped ring buffer 8690to read the dirty GFNs. If the flags has the DIRTY bit set (at this stage 8691the RESET bit must be cleared), then it means this GFN is a dirty GFN. 8692The userspace should harvest this GFN and mark the flags from state 8693``01b`` to ``1Xb`` (bit 0 will be ignored by KVM, but bit 1 must be set 8694to show that this GFN is harvested and waiting for a reset), and move 8695on to the next GFN. The userspace should continue to do this until the 8696flags of a GFN have the DIRTY bit cleared, meaning that it has harvested 8697all the dirty GFNs that were available. 8698 8699Note that on weakly ordered architectures, userspace accesses to the 8700ring buffer (and more specifically the 'flags' field) must be ordered, 8701using load-acquire/store-release accessors when available, or any 8702other memory barrier that will ensure this ordering. 8703 8704It's not necessary for userspace to harvest the all dirty GFNs at once. 8705However it must collect the dirty GFNs in sequence, i.e., the userspace 8706program cannot skip one dirty GFN to collect the one next to it. 8707 8708After processing one or more entries in the ring buffer, userspace 8709calls the VM ioctl KVM_RESET_DIRTY_RINGS to notify the kernel about 8710it, so that the kernel will reprotect those collected GFNs. 8711Therefore, the ioctl must be called *before* reading the content of 8712the dirty pages. 8713 8714The dirty ring can get full. When it happens, the KVM_RUN of the 8715vcpu will return with exit reason KVM_EXIT_DIRTY_LOG_FULL. 8716 8717The dirty ring interface has a major difference comparing to the 8718KVM_GET_DIRTY_LOG interface in that, when reading the dirty ring from 8719userspace, it's still possible that the kernel has not yet flushed the 8720processor's dirty page buffers into the kernel buffer (with dirty bitmaps, the 8721flushing is done by the KVM_GET_DIRTY_LOG ioctl). To achieve that, one 8722needs to kick the vcpu out of KVM_RUN using a signal. The resulting 8723vmexit ensures that all dirty GFNs are flushed to the dirty rings. 8724 8725NOTE: KVM_CAP_DIRTY_LOG_RING_ACQ_REL is the only capability that 8726should be exposed by weakly ordered architecture, in order to indicate 8727the additional memory ordering requirements imposed on userspace when 8728reading the state of an entry and mutating it from DIRTY to HARVESTED. 8729Architecture with TSO-like ordering (such as x86) are allowed to 8730expose both KVM_CAP_DIRTY_LOG_RING and KVM_CAP_DIRTY_LOG_RING_ACQ_REL 8731to userspace. 8732 8733After enabling the dirty rings, the userspace needs to detect the 8734capability of KVM_CAP_DIRTY_LOG_RING_WITH_BITMAP to see whether the 8735ring structures can be backed by per-slot bitmaps. With this capability 8736advertised, it means the architecture can dirty guest pages without 8737vcpu/ring context, so that some of the dirty information will still be 8738maintained in the bitmap structure. KVM_CAP_DIRTY_LOG_RING_WITH_BITMAP 8739can't be enabled if the capability of KVM_CAP_DIRTY_LOG_RING_ACQ_REL 8740hasn't been enabled, or any memslot has been existing. 8741 8742Note that the bitmap here is only a backup of the ring structure. The 8743use of the ring and bitmap combination is only beneficial if there is 8744only a very small amount of memory that is dirtied out of vcpu/ring 8745context. Otherwise, the stand-alone per-slot bitmap mechanism needs to 8746be considered. 8747 8748To collect dirty bits in the backup bitmap, userspace can use the same 8749KVM_GET_DIRTY_LOG ioctl. KVM_CLEAR_DIRTY_LOG isn't needed as long as all 8750the generation of the dirty bits is done in a single pass. Collecting 8751the dirty bitmap should be the very last thing that the VMM does before 8752considering the state as complete. VMM needs to ensure that the dirty 8753state is final and avoid missing dirty pages from another ioctl ordered 8754after the bitmap collection. 8755 8756NOTE: Multiple examples of using the backup bitmap: (1) save vgic/its 8757tables through command KVM_DEV_ARM_{VGIC_GRP_CTRL, ITS_SAVE_TABLES} on 8758KVM device "kvm-arm-vgic-its". (2) restore vgic/its tables through 8759command KVM_DEV_ARM_{VGIC_GRP_CTRL, ITS_RESTORE_TABLES} on KVM device 8760"kvm-arm-vgic-its". VGICv3 LPI pending status is restored. (3) save 8761vgic3 pending table through KVM_DEV_ARM_VGIC_{GRP_CTRL, SAVE_PENDING_TABLES} 8762command on KVM device "kvm-arm-vgic-v3". 8763 87648.30 KVM_CAP_XEN_HVM 8765-------------------- 8766 8767:Architectures: x86 8768 8769This capability indicates the features that Xen supports for hosting Xen 8770PVHVM guests. Valid flags are:: 8771 8772 #define KVM_XEN_HVM_CONFIG_HYPERCALL_MSR (1 << 0) 8773 #define KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL (1 << 1) 8774 #define KVM_XEN_HVM_CONFIG_SHARED_INFO (1 << 2) 8775 #define KVM_XEN_HVM_CONFIG_RUNSTATE (1 << 3) 8776 #define KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL (1 << 4) 8777 #define KVM_XEN_HVM_CONFIG_EVTCHN_SEND (1 << 5) 8778 #define KVM_XEN_HVM_CONFIG_RUNSTATE_UPDATE_FLAG (1 << 6) 8779 #define KVM_XEN_HVM_CONFIG_PVCLOCK_TSC_UNSTABLE (1 << 7) 8780 8781The KVM_XEN_HVM_CONFIG_HYPERCALL_MSR flag indicates that the KVM_XEN_HVM_CONFIG 8782ioctl is available, for the guest to set its hypercall page. 8783 8784If KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL is also set, the same flag may also be 8785provided in the flags to KVM_XEN_HVM_CONFIG, without providing hypercall page 8786contents, to request that KVM generate hypercall page content automatically 8787and also enable interception of guest hypercalls with KVM_EXIT_XEN. 8788 8789The KVM_XEN_HVM_CONFIG_SHARED_INFO flag indicates the availability of the 8790KVM_XEN_HVM_SET_ATTR, KVM_XEN_HVM_GET_ATTR, KVM_XEN_VCPU_SET_ATTR and 8791KVM_XEN_VCPU_GET_ATTR ioctls, as well as the delivery of exception vectors 8792for event channel upcalls when the evtchn_upcall_pending field of a vcpu's 8793vcpu_info is set. 8794 8795The KVM_XEN_HVM_CONFIG_RUNSTATE flag indicates that the runstate-related 8796features KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR/_CURRENT/_DATA/_ADJUST are 8797supported by the KVM_XEN_VCPU_SET_ATTR/KVM_XEN_VCPU_GET_ATTR ioctls. 8798 8799The KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL flag indicates that IRQ routing entries 8800of the type KVM_IRQ_ROUTING_XEN_EVTCHN are supported, with the priority 8801field set to indicate 2 level event channel delivery. 8802 8803The KVM_XEN_HVM_CONFIG_EVTCHN_SEND flag indicates that KVM supports 8804injecting event channel events directly into the guest with the 8805KVM_XEN_HVM_EVTCHN_SEND ioctl. It also indicates support for the 8806KVM_XEN_ATTR_TYPE_EVTCHN/XEN_VERSION HVM attributes and the 8807KVM_XEN_VCPU_ATTR_TYPE_VCPU_ID/TIMER/UPCALL_VECTOR vCPU attributes. 8808related to event channel delivery, timers, and the XENVER_version 8809interception. 8810 8811The KVM_XEN_HVM_CONFIG_RUNSTATE_UPDATE_FLAG flag indicates that KVM supports 8812the KVM_XEN_ATTR_TYPE_RUNSTATE_UPDATE_FLAG attribute in the KVM_XEN_SET_ATTR 8813and KVM_XEN_GET_ATTR ioctls. This controls whether KVM will set the 8814XEN_RUNSTATE_UPDATE flag in guest memory mapped vcpu_runstate_info during 8815updates of the runstate information. Note that versions of KVM which support 8816the RUNSTATE feature above, but not the RUNSTATE_UPDATE_FLAG feature, will 8817always set the XEN_RUNSTATE_UPDATE flag when updating the guest structure, 8818which is perhaps counterintuitive. When this flag is advertised, KVM will 8819behave more correctly, not using the XEN_RUNSTATE_UPDATE flag until/unless 8820specifically enabled (by the guest making the hypercall, causing the VMM 8821to enable the KVM_XEN_ATTR_TYPE_RUNSTATE_UPDATE_FLAG attribute). 8822 8823The KVM_XEN_HVM_CONFIG_PVCLOCK_TSC_UNSTABLE flag indicates that KVM supports 8824clearing the PVCLOCK_TSC_STABLE_BIT flag in Xen pvclock sources. This will be 8825done when the KVM_CAP_XEN_HVM ioctl sets the 8826KVM_XEN_HVM_CONFIG_PVCLOCK_TSC_UNSTABLE flag. 8827 88288.31 KVM_CAP_PPC_MULTITCE 8829------------------------- 8830 8831:Capability: KVM_CAP_PPC_MULTITCE 8832:Architectures: ppc 8833:Type: vm 8834 8835This capability means the kernel is capable of handling hypercalls 8836H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user 8837space. This significantly accelerates DMA operations for PPC KVM guests. 8838User space should expect that its handlers for these hypercalls 8839are not going to be called if user space previously registered LIOBN 8840in KVM (via KVM_CREATE_SPAPR_TCE or similar calls). 8841 8842In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest, 8843user space might have to advertise it for the guest. For example, 8844IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is 8845present in the "ibm,hypertas-functions" device-tree property. 8846 8847The hypercalls mentioned above may or may not be processed successfully 8848in the kernel based fast path. If they can not be handled by the kernel, 8849they will get passed on to user space. So user space still has to have 8850an implementation for these despite the in kernel acceleration. 8851 8852This capability is always enabled. 8853 88548.32 KVM_CAP_PTP_KVM 8855-------------------- 8856 8857:Architectures: arm64 8858 8859This capability indicates that the KVM virtual PTP service is 8860supported in the host. A VMM can check whether the service is 8861available to the guest on migration. 8862 88638.33 KVM_CAP_HYPERV_ENFORCE_CPUID 8864--------------------------------- 8865 8866Architectures: x86 8867 8868When enabled, KVM will disable emulated Hyper-V features provided to the 8869guest according to the bits Hyper-V CPUID feature leaves. Otherwise, all 8870currently implemented Hyper-V features are provided unconditionally when 8871Hyper-V identification is set in the HYPERV_CPUID_INTERFACE (0x40000001) 8872leaf. 8873 88748.34 KVM_CAP_EXIT_HYPERCALL 8875--------------------------- 8876 8877:Capability: KVM_CAP_EXIT_HYPERCALL 8878:Architectures: x86 8879:Type: vm 8880 8881This capability, if enabled, will cause KVM to exit to userspace 8882with KVM_EXIT_HYPERCALL exit reason to process some hypercalls. 8883 8884Calling KVM_CHECK_EXTENSION for this capability will return a bitmask 8885of hypercalls that can be configured to exit to userspace. 8886Right now, the only such hypercall is KVM_HC_MAP_GPA_RANGE. 8887 8888The argument to KVM_ENABLE_CAP is also a bitmask, and must be a subset 8889of the result of KVM_CHECK_EXTENSION. KVM will forward to userspace 8890the hypercalls whose corresponding bit is in the argument, and return 8891ENOSYS for the others. 8892 88938.35 KVM_CAP_PMU_CAPABILITY 8894--------------------------- 8895 8896:Capability: KVM_CAP_PMU_CAPABILITY 8897:Architectures: x86 8898:Type: vm 8899:Parameters: arg[0] is bitmask of PMU virtualization capabilities. 8900:Returns: 0 on success, -EINVAL when arg[0] contains invalid bits 8901 8902This capability alters PMU virtualization in KVM. 8903 8904Calling KVM_CHECK_EXTENSION for this capability returns a bitmask of 8905PMU virtualization capabilities that can be adjusted on a VM. 8906 8907The argument to KVM_ENABLE_CAP is also a bitmask and selects specific 8908PMU virtualization capabilities to be applied to the VM. This can 8909only be invoked on a VM prior to the creation of VCPUs. 8910 8911At this time, KVM_PMU_CAP_DISABLE is the only capability. Setting 8912this capability will disable PMU virtualization for that VM. Usermode 8913should adjust CPUID leaf 0xA to reflect that the PMU is disabled. 8914 89158.36 KVM_CAP_ARM_SYSTEM_SUSPEND 8916------------------------------- 8917 8918:Capability: KVM_CAP_ARM_SYSTEM_SUSPEND 8919:Architectures: arm64 8920:Type: vm 8921 8922When enabled, KVM will exit to userspace with KVM_EXIT_SYSTEM_EVENT of 8923type KVM_SYSTEM_EVENT_SUSPEND to process the guest suspend request. 8924 89258.37 KVM_CAP_S390_PROTECTED_DUMP 8926-------------------------------- 8927 8928:Capability: KVM_CAP_S390_PROTECTED_DUMP 8929:Architectures: s390 8930:Type: vm 8931 8932This capability indicates that KVM and the Ultravisor support dumping 8933PV guests. The `KVM_PV_DUMP` command is available for the 8934`KVM_S390_PV_COMMAND` ioctl and the `KVM_PV_INFO` command provides 8935dump related UV data. Also the vcpu ioctl `KVM_S390_PV_CPU_COMMAND` is 8936available and supports the `KVM_PV_DUMP_CPU` subcommand. 8937 89388.38 KVM_CAP_VM_DISABLE_NX_HUGE_PAGES 8939------------------------------------- 8940 8941:Capability: KVM_CAP_VM_DISABLE_NX_HUGE_PAGES 8942:Architectures: x86 8943:Type: vm 8944:Parameters: arg[0] must be 0. 8945:Returns: 0 on success, -EPERM if the userspace process does not 8946 have CAP_SYS_BOOT, -EINVAL if args[0] is not 0 or any vCPUs have been 8947 created. 8948 8949This capability disables the NX huge pages mitigation for iTLB MULTIHIT. 8950 8951The capability has no effect if the nx_huge_pages module parameter is not set. 8952 8953This capability may only be set before any vCPUs are created. 8954 89558.39 KVM_CAP_S390_CPU_TOPOLOGY 8956------------------------------ 8957 8958:Capability: KVM_CAP_S390_CPU_TOPOLOGY 8959:Architectures: s390 8960:Type: vm 8961 8962This capability indicates that KVM will provide the S390 CPU Topology 8963facility which consist of the interpretation of the PTF instruction for 8964the function code 2 along with interception and forwarding of both the 8965PTF instruction with function codes 0 or 1 and the STSI(15,1,x) 8966instruction to the userland hypervisor. 8967 8968The stfle facility 11, CPU Topology facility, should not be indicated 8969to the guest without this capability. 8970 8971When this capability is present, KVM provides a new attribute group 8972on vm fd, KVM_S390_VM_CPU_TOPOLOGY. 8973This new attribute allows to get, set or clear the Modified Change 8974Topology Report (MTCR) bit of the SCA through the kvm_device_attr 8975structure. 8976 8977When getting the Modified Change Topology Report value, the attr->addr 8978must point to a byte where the value will be stored or retrieved from. 8979 89808.40 KVM_CAP_ARM_EAGER_SPLIT_CHUNK_SIZE 8981--------------------------------------- 8982 8983:Capability: KVM_CAP_ARM_EAGER_SPLIT_CHUNK_SIZE 8984:Architectures: arm64 8985:Type: vm 8986:Parameters: arg[0] is the new split chunk size. 8987:Returns: 0 on success, -EINVAL if any memslot was already created. 8988 8989This capability sets the chunk size used in Eager Page Splitting. 8990 8991Eager Page Splitting improves the performance of dirty-logging (used 8992in live migrations) when guest memory is backed by huge-pages. It 8993avoids splitting huge-pages (into PAGE_SIZE pages) on fault, by doing 8994it eagerly when enabling dirty logging (with the 8995KVM_MEM_LOG_DIRTY_PAGES flag for a memory region), or when using 8996KVM_CLEAR_DIRTY_LOG. 8997 8998The chunk size specifies how many pages to break at a time, using a 8999single allocation for each chunk. Bigger the chunk size, more pages 9000need to be allocated ahead of time. 9001 9002The chunk size needs to be a valid block size. The list of acceptable 9003block sizes is exposed in KVM_CAP_ARM_SUPPORTED_BLOCK_SIZES as a 900464-bit bitmap (each bit describing a block size). The default value is 90050, to disable the eager page splitting. 9006 90078.41 KVM_CAP_VM_TYPES 9008--------------------- 9009 9010:Capability: KVM_CAP_MEMORY_ATTRIBUTES 9011:Architectures: x86 9012:Type: system ioctl 9013 9014This capability returns a bitmap of support VM types. The 1-setting of bit @n 9015means the VM type with value @n is supported. Possible values of @n are:: 9016 9017 #define KVM_X86_DEFAULT_VM 0 9018 #define KVM_X86_SW_PROTECTED_VM 1 9019 #define KVM_X86_SEV_VM 2 9020 #define KVM_X86_SEV_ES_VM 3 9021 9022Note, KVM_X86_SW_PROTECTED_VM is currently only for development and testing. 9023Do not use KVM_X86_SW_PROTECTED_VM for "real" VMs, and especially not in 9024production. The behavior and effective ABI for software-protected VMs is 9025unstable. 9026 90279. Known KVM API problems 9028========================= 9029 9030In some cases, KVM's API has some inconsistencies or common pitfalls 9031that userspace need to be aware of. This section details some of 9032these issues. 9033 9034Most of them are architecture specific, so the section is split by 9035architecture. 9036 90379.1. x86 9038-------- 9039 9040``KVM_GET_SUPPORTED_CPUID`` issues 9041^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9042 9043In general, ``KVM_GET_SUPPORTED_CPUID`` is designed so that it is possible 9044to take its result and pass it directly to ``KVM_SET_CPUID2``. This section 9045documents some cases in which that requires some care. 9046 9047Local APIC features 9048~~~~~~~~~~~~~~~~~~~ 9049 9050CPU[EAX=1]:ECX[21] (X2APIC) is reported by ``KVM_GET_SUPPORTED_CPUID``, 9051but it can only be enabled if ``KVM_CREATE_IRQCHIP`` or 9052``KVM_ENABLE_CAP(KVM_CAP_IRQCHIP_SPLIT)`` are used to enable in-kernel emulation of 9053the local APIC. 9054 9055The same is true for the ``KVM_FEATURE_PV_UNHALT`` paravirtualized feature. 9056 9057CPU[EAX=1]:ECX[24] (TSC_DEADLINE) is not reported by ``KVM_GET_SUPPORTED_CPUID``. 9058It can be enabled if ``KVM_CAP_TSC_DEADLINE_TIMER`` is present and the kernel 9059has enabled in-kernel emulation of the local APIC. 9060 9061CPU topology 9062~~~~~~~~~~~~ 9063 9064Several CPUID values include topology information for the host CPU: 90650x0b and 0x1f for Intel systems, 0x8000001e for AMD systems. Different 9066versions of KVM return different values for this information and userspace 9067should not rely on it. Currently they return all zeroes. 9068 9069If userspace wishes to set up a guest topology, it should be careful that 9070the values of these three leaves differ for each CPU. In particular, 9071the APIC ID is found in EDX for all subleaves of 0x0b and 0x1f, and in EAX 9072for 0x8000001e; the latter also encodes the core id and node id in bits 90737:0 of EBX and ECX respectively. 9074 9075Obsolete ioctls and capabilities 9076^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9077 9078KVM_CAP_DISABLE_QUIRKS does not let userspace know which quirks are actually 9079available. Use ``KVM_CHECK_EXTENSION(KVM_CAP_DISABLE_QUIRKS2)`` instead if 9080available. 9081 9082Ordering of KVM_GET_*/KVM_SET_* ioctls 9083^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9084 9085TBD 9086