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 1003The MSR index must be in the range [0x40000000, 0x4fffffff], i.e. must reside 1004in the range that is unofficially reserved for use by hypervisors. The min/max 1005values are enumerated via KVM_XEN_MSR_MIN_INDEX and KVM_XEN_MSR_MAX_INDEX. 1006 1007:: 1008 1009 struct kvm_xen_hvm_config { 1010 __u32 flags; 1011 __u32 msr; 1012 __u64 blob_addr_32; 1013 __u64 blob_addr_64; 1014 __u8 blob_size_32; 1015 __u8 blob_size_64; 1016 __u8 pad2[30]; 1017 }; 1018 1019If certain flags are returned from the KVM_CAP_XEN_HVM check, they may 1020be set in the flags field of this ioctl: 1021 1022The KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL flag requests KVM to generate 1023the contents of the hypercall page automatically; hypercalls will be 1024intercepted and passed to userspace through KVM_EXIT_XEN. In this 1025case, all of the blob size and address fields must be zero. 1026 1027The KVM_XEN_HVM_CONFIG_EVTCHN_SEND flag indicates to KVM that userspace 1028will always use the KVM_XEN_HVM_EVTCHN_SEND ioctl to deliver event 1029channel interrupts rather than manipulating the guest's shared_info 1030structures directly. This, in turn, may allow KVM to enable features 1031such as intercepting the SCHEDOP_poll hypercall to accelerate PV 1032spinlock operation for the guest. Userspace may still use the ioctl 1033to deliver events if it was advertised, even if userspace does not 1034send this indication that it will always do so 1035 1036No other flags are currently valid in the struct kvm_xen_hvm_config. 1037 10384.29 KVM_GET_CLOCK 1039------------------ 1040 1041:Capability: KVM_CAP_ADJUST_CLOCK 1042:Architectures: x86 1043:Type: vm ioctl 1044:Parameters: struct kvm_clock_data (out) 1045:Returns: 0 on success, -1 on error 1046 1047Gets the current timestamp of kvmclock as seen by the current guest. In 1048conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios 1049such as migration. 1050 1051When KVM_CAP_ADJUST_CLOCK is passed to KVM_CHECK_EXTENSION, it returns the 1052set of bits that KVM can return in struct kvm_clock_data's flag member. 1053 1054The following flags are defined: 1055 1056KVM_CLOCK_TSC_STABLE 1057 If set, the returned value is the exact kvmclock 1058 value seen by all VCPUs at the instant when KVM_GET_CLOCK was called. 1059 If clear, the returned value is simply CLOCK_MONOTONIC plus a constant 1060 offset; the offset can be modified with KVM_SET_CLOCK. KVM will try 1061 to make all VCPUs follow this clock, but the exact value read by each 1062 VCPU could differ, because the host TSC is not stable. 1063 1064KVM_CLOCK_REALTIME 1065 If set, the `realtime` field in the kvm_clock_data 1066 structure is populated with the value of the host's real time 1067 clocksource at the instant when KVM_GET_CLOCK was called. If clear, 1068 the `realtime` field does not contain a value. 1069 1070KVM_CLOCK_HOST_TSC 1071 If set, the `host_tsc` field in the kvm_clock_data 1072 structure is populated with the value of the host's timestamp counter (TSC) 1073 at the instant when KVM_GET_CLOCK was called. If clear, the `host_tsc` field 1074 does not contain a value. 1075 1076:: 1077 1078 struct kvm_clock_data { 1079 __u64 clock; /* kvmclock current value */ 1080 __u32 flags; 1081 __u32 pad0; 1082 __u64 realtime; 1083 __u64 host_tsc; 1084 __u32 pad[4]; 1085 }; 1086 1087 10884.30 KVM_SET_CLOCK 1089------------------ 1090 1091:Capability: KVM_CAP_ADJUST_CLOCK 1092:Architectures: x86 1093:Type: vm ioctl 1094:Parameters: struct kvm_clock_data (in) 1095:Returns: 0 on success, -1 on error 1096 1097Sets the current timestamp of kvmclock to the value specified in its parameter. 1098In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios 1099such as migration. 1100 1101The following flags can be passed: 1102 1103KVM_CLOCK_REALTIME 1104 If set, KVM will compare the value of the `realtime` field 1105 with the value of the host's real time clocksource at the instant when 1106 KVM_SET_CLOCK was called. The difference in elapsed time is added to the final 1107 kvmclock value that will be provided to guests. 1108 1109Other flags returned by ``KVM_GET_CLOCK`` are accepted but ignored. 1110 1111:: 1112 1113 struct kvm_clock_data { 1114 __u64 clock; /* kvmclock current value */ 1115 __u32 flags; 1116 __u32 pad0; 1117 __u64 realtime; 1118 __u64 host_tsc; 1119 __u32 pad[4]; 1120 }; 1121 1122 11234.31 KVM_GET_VCPU_EVENTS 1124------------------------ 1125 1126:Capability: KVM_CAP_VCPU_EVENTS 1127:Extended by: KVM_CAP_INTR_SHADOW 1128:Architectures: x86, arm64 1129:Type: vcpu ioctl 1130:Parameters: struct kvm_vcpu_events (out) 1131:Returns: 0 on success, -1 on error 1132 1133X86: 1134^^^^ 1135 1136Gets currently pending exceptions, interrupts, and NMIs as well as related 1137states of the vcpu. 1138 1139:: 1140 1141 struct kvm_vcpu_events { 1142 struct { 1143 __u8 injected; 1144 __u8 nr; 1145 __u8 has_error_code; 1146 __u8 pending; 1147 __u32 error_code; 1148 } exception; 1149 struct { 1150 __u8 injected; 1151 __u8 nr; 1152 __u8 soft; 1153 __u8 shadow; 1154 } interrupt; 1155 struct { 1156 __u8 injected; 1157 __u8 pending; 1158 __u8 masked; 1159 __u8 pad; 1160 } nmi; 1161 __u32 sipi_vector; 1162 __u32 flags; 1163 struct { 1164 __u8 smm; 1165 __u8 pending; 1166 __u8 smm_inside_nmi; 1167 __u8 latched_init; 1168 } smi; 1169 __u8 reserved[27]; 1170 __u8 exception_has_payload; 1171 __u64 exception_payload; 1172 }; 1173 1174The following bits are defined in the flags field: 1175 1176- KVM_VCPUEVENT_VALID_SHADOW may be set to signal that 1177 interrupt.shadow contains a valid state. 1178 1179- KVM_VCPUEVENT_VALID_SMM may be set to signal that smi contains a 1180 valid state. 1181 1182- KVM_VCPUEVENT_VALID_PAYLOAD may be set to signal that the 1183 exception_has_payload, exception_payload, and exception.pending 1184 fields contain a valid state. This bit will be set whenever 1185 KVM_CAP_EXCEPTION_PAYLOAD is enabled. 1186 1187- KVM_VCPUEVENT_VALID_TRIPLE_FAULT may be set to signal that the 1188 triple_fault_pending field contains a valid state. This bit will 1189 be set whenever KVM_CAP_X86_TRIPLE_FAULT_EVENT is enabled. 1190 1191ARM64: 1192^^^^^^ 1193 1194If the guest accesses a device that is being emulated by the host kernel in 1195such a way that a real device would generate a physical SError, KVM may make 1196a virtual SError pending for that VCPU. This system error interrupt remains 1197pending until the guest takes the exception by unmasking PSTATE.A. 1198 1199Running the VCPU may cause it to take a pending SError, or make an access that 1200causes an SError to become pending. The event's description is only valid while 1201the VPCU is not running. 1202 1203This API provides a way to read and write the pending 'event' state that is not 1204visible to the guest. To save, restore or migrate a VCPU the struct representing 1205the state can be read then written using this GET/SET API, along with the other 1206guest-visible registers. It is not possible to 'cancel' an SError that has been 1207made pending. 1208 1209A device being emulated in user-space may also wish to generate an SError. To do 1210this the events structure can be populated by user-space. The current state 1211should be read first, to ensure no existing SError is pending. If an existing 1212SError is pending, the architecture's 'Multiple SError interrupts' rules should 1213be followed. (2.5.3 of DDI0587.a "ARM Reliability, Availability, and 1214Serviceability (RAS) Specification"). 1215 1216SError exceptions always have an ESR value. Some CPUs have the ability to 1217specify what the virtual SError's ESR value should be. These systems will 1218advertise KVM_CAP_ARM_INJECT_SERROR_ESR. In this case exception.has_esr will 1219always have a non-zero value when read, and the agent making an SError pending 1220should specify the ISS field in the lower 24 bits of exception.serror_esr. If 1221the system supports KVM_CAP_ARM_INJECT_SERROR_ESR, but user-space sets the events 1222with exception.has_esr as zero, KVM will choose an ESR. 1223 1224Specifying exception.has_esr on a system that does not support it will return 1225-EINVAL. Setting anything other than the lower 24bits of exception.serror_esr 1226will return -EINVAL. 1227 1228It is not possible to read back a pending external abort (injected via 1229KVM_SET_VCPU_EVENTS or otherwise) because such an exception is always delivered 1230directly to the virtual CPU). 1231 1232:: 1233 1234 struct kvm_vcpu_events { 1235 struct { 1236 __u8 serror_pending; 1237 __u8 serror_has_esr; 1238 __u8 ext_dabt_pending; 1239 /* Align it to 8 bytes */ 1240 __u8 pad[5]; 1241 __u64 serror_esr; 1242 } exception; 1243 __u32 reserved[12]; 1244 }; 1245 12464.32 KVM_SET_VCPU_EVENTS 1247------------------------ 1248 1249:Capability: KVM_CAP_VCPU_EVENTS 1250:Extended by: KVM_CAP_INTR_SHADOW 1251:Architectures: x86, arm64 1252:Type: vcpu ioctl 1253:Parameters: struct kvm_vcpu_events (in) 1254:Returns: 0 on success, -1 on error 1255 1256X86: 1257^^^^ 1258 1259Set pending exceptions, interrupts, and NMIs as well as related states of the 1260vcpu. 1261 1262See KVM_GET_VCPU_EVENTS for the data structure. 1263 1264Fields that may be modified asynchronously by running VCPUs can be excluded 1265from the update. These fields are nmi.pending, sipi_vector, smi.smm, 1266smi.pending. Keep the corresponding bits in the flags field cleared to 1267suppress overwriting the current in-kernel state. The bits are: 1268 1269=============================== ================================== 1270KVM_VCPUEVENT_VALID_NMI_PENDING transfer nmi.pending to the kernel 1271KVM_VCPUEVENT_VALID_SIPI_VECTOR transfer sipi_vector 1272KVM_VCPUEVENT_VALID_SMM transfer the smi sub-struct. 1273=============================== ================================== 1274 1275If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in 1276the flags field to signal that interrupt.shadow contains a valid state and 1277shall be written into the VCPU. 1278 1279KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available. 1280 1281If KVM_CAP_EXCEPTION_PAYLOAD is enabled, KVM_VCPUEVENT_VALID_PAYLOAD 1282can be set in the flags field to signal that the 1283exception_has_payload, exception_payload, and exception.pending fields 1284contain a valid state and shall be written into the VCPU. 1285 1286If KVM_CAP_X86_TRIPLE_FAULT_EVENT is enabled, KVM_VCPUEVENT_VALID_TRIPLE_FAULT 1287can be set in flags field to signal that the triple_fault field contains 1288a valid state and shall be written into the VCPU. 1289 1290ARM64: 1291^^^^^^ 1292 1293User space may need to inject several types of events to the guest. 1294 1295Set the pending SError exception state for this VCPU. It is not possible to 1296'cancel' an Serror that has been made pending. 1297 1298If the guest performed an access to I/O memory which could not be handled by 1299userspace, for example because of missing instruction syndrome decode 1300information or because there is no device mapped at the accessed IPA, then 1301userspace can ask the kernel to inject an external abort using the address 1302from the exiting fault on the VCPU. It is a programming error to set 1303ext_dabt_pending after an exit which was not either KVM_EXIT_MMIO or 1304KVM_EXIT_ARM_NISV. This feature is only available if the system supports 1305KVM_CAP_ARM_INJECT_EXT_DABT. This is a helper which provides commonality in 1306how userspace reports accesses for the above cases to guests, across different 1307userspace implementations. Nevertheless, userspace can still emulate all Arm 1308exceptions by manipulating individual registers using the KVM_SET_ONE_REG API. 1309 1310See KVM_GET_VCPU_EVENTS for the data structure. 1311 1312 13134.33 KVM_GET_DEBUGREGS 1314---------------------- 1315 1316:Capability: KVM_CAP_DEBUGREGS 1317:Architectures: x86 1318:Type: vcpu ioctl 1319:Parameters: struct kvm_debugregs (out) 1320:Returns: 0 on success, -1 on error 1321 1322Reads debug registers from the vcpu. 1323 1324:: 1325 1326 struct kvm_debugregs { 1327 __u64 db[4]; 1328 __u64 dr6; 1329 __u64 dr7; 1330 __u64 flags; 1331 __u64 reserved[9]; 1332 }; 1333 1334 13354.34 KVM_SET_DEBUGREGS 1336---------------------- 1337 1338:Capability: KVM_CAP_DEBUGREGS 1339:Architectures: x86 1340:Type: vcpu ioctl 1341:Parameters: struct kvm_debugregs (in) 1342:Returns: 0 on success, -1 on error 1343 1344Writes debug registers into the vcpu. 1345 1346See KVM_GET_DEBUGREGS for the data structure. The flags field is unused 1347yet and must be cleared on entry. 1348 1349 13504.35 KVM_SET_USER_MEMORY_REGION 1351------------------------------- 1352 1353:Capability: KVM_CAP_USER_MEMORY 1354:Architectures: all 1355:Type: vm ioctl 1356:Parameters: struct kvm_userspace_memory_region (in) 1357:Returns: 0 on success, -1 on error 1358 1359:: 1360 1361 struct kvm_userspace_memory_region { 1362 __u32 slot; 1363 __u32 flags; 1364 __u64 guest_phys_addr; 1365 __u64 memory_size; /* bytes */ 1366 __u64 userspace_addr; /* start of the userspace allocated memory */ 1367 }; 1368 1369 /* for kvm_userspace_memory_region::flags */ 1370 #define KVM_MEM_LOG_DIRTY_PAGES (1UL << 0) 1371 #define KVM_MEM_READONLY (1UL << 1) 1372 1373This ioctl allows the user to create, modify or delete a guest physical 1374memory slot. Bits 0-15 of "slot" specify the slot id and this value 1375should be less than the maximum number of user memory slots supported per 1376VM. The maximum allowed slots can be queried using KVM_CAP_NR_MEMSLOTS. 1377Slots may not overlap in guest physical address space. 1378 1379If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot" 1380specifies the address space which is being modified. They must be 1381less than the value that KVM_CHECK_EXTENSION returns for the 1382KVM_CAP_MULTI_ADDRESS_SPACE capability. Slots in separate address spaces 1383are unrelated; the restriction on overlapping slots only applies within 1384each address space. 1385 1386Deleting a slot is done by passing zero for memory_size. When changing 1387an existing slot, it may be moved in the guest physical memory space, 1388or its flags may be modified, but it may not be resized. 1389 1390Memory for the region is taken starting at the address denoted by the 1391field userspace_addr, which must point at user addressable memory for 1392the entire memory slot size. Any object may back this memory, including 1393anonymous memory, ordinary files, and hugetlbfs. 1394 1395On architectures that support a form of address tagging, userspace_addr must 1396be an untagged address. 1397 1398It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr 1399be identical. This allows large pages in the guest to be backed by large 1400pages in the host. 1401 1402The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and 1403KVM_MEM_READONLY. The former can be set to instruct KVM to keep track of 1404writes to memory within the slot. See KVM_GET_DIRTY_LOG ioctl to know how to 1405use it. The latter can be set, if KVM_CAP_READONLY_MEM capability allows it, 1406to make a new slot read-only. In this case, writes to this memory will be 1407posted to userspace as KVM_EXIT_MMIO exits. 1408 1409When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of 1410the memory region are automatically reflected into the guest. For example, an 1411mmap() that affects the region will be made visible immediately. Another 1412example is madvise(MADV_DROP). 1413 1414Note: On arm64, a write generated by the page-table walker (to update 1415the Access and Dirty flags, for example) never results in a 1416KVM_EXIT_MMIO exit when the slot has the KVM_MEM_READONLY flag. This 1417is because KVM cannot provide the data that would be written by the 1418page-table walker, making it impossible to emulate the access. 1419Instead, an abort (data abort if the cause of the page-table update 1420was a load or a store, instruction abort if it was an instruction 1421fetch) is injected in the guest. 1422 1423S390: 1424^^^^^ 1425 1426Returns -EINVAL or -EEXIST if the VM has the KVM_VM_S390_UCONTROL flag set. 1427Returns -EINVAL if called on a protected VM. 1428 14294.36 KVM_SET_TSS_ADDR 1430--------------------- 1431 1432:Capability: KVM_CAP_SET_TSS_ADDR 1433:Architectures: x86 1434:Type: vm ioctl 1435:Parameters: unsigned long tss_address (in) 1436:Returns: 0 on success, -1 on error 1437 1438This ioctl defines the physical address of a three-page region in the guest 1439physical address space. The region must be within the first 4GB of the 1440guest physical address space and must not conflict with any memory slot 1441or any mmio address. The guest may malfunction if it accesses this memory 1442region. 1443 1444This ioctl is required on Intel-based hosts. This is needed on Intel hardware 1445because of a quirk in the virtualization implementation (see the internals 1446documentation when it pops into existence). 1447 1448 1449.. _KVM_ENABLE_CAP: 1450 14514.37 KVM_ENABLE_CAP 1452------------------- 1453 1454:Capability: KVM_CAP_ENABLE_CAP 1455:Architectures: mips, ppc, s390, x86, loongarch 1456:Type: vcpu ioctl 1457:Parameters: struct kvm_enable_cap (in) 1458:Returns: 0 on success; -1 on error 1459 1460:Capability: KVM_CAP_ENABLE_CAP_VM 1461:Architectures: all 1462:Type: vm ioctl 1463:Parameters: struct kvm_enable_cap (in) 1464:Returns: 0 on success; -1 on error 1465 1466.. note:: 1467 1468 Not all extensions are enabled by default. Using this ioctl the application 1469 can enable an extension, making it available to the guest. 1470 1471On systems that do not support this ioctl, it always fails. On systems that 1472do support it, it only works for extensions that are supported for enablement. 1473 1474To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should 1475be used. 1476 1477:: 1478 1479 struct kvm_enable_cap { 1480 /* in */ 1481 __u32 cap; 1482 1483The capability that is supposed to get enabled. 1484 1485:: 1486 1487 __u32 flags; 1488 1489A bitfield indicating future enhancements. Has to be 0 for now. 1490 1491:: 1492 1493 __u64 args[4]; 1494 1495Arguments for enabling a feature. If a feature needs initial values to 1496function properly, this is the place to put them. 1497 1498:: 1499 1500 __u8 pad[64]; 1501 }; 1502 1503The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl 1504for vm-wide capabilities. 1505 15064.38 KVM_GET_MP_STATE 1507--------------------- 1508 1509:Capability: KVM_CAP_MP_STATE 1510:Architectures: x86, s390, arm64, riscv, loongarch 1511:Type: vcpu ioctl 1512:Parameters: struct kvm_mp_state (out) 1513:Returns: 0 on success; -1 on error 1514 1515:: 1516 1517 struct kvm_mp_state { 1518 __u32 mp_state; 1519 }; 1520 1521Returns the vcpu's current "multiprocessing state" (though also valid on 1522uniprocessor guests). 1523 1524Possible values are: 1525 1526 ========================== =============================================== 1527 KVM_MP_STATE_RUNNABLE the vcpu is currently running 1528 [x86,arm64,riscv,loongarch] 1529 KVM_MP_STATE_UNINITIALIZED the vcpu is an application processor (AP) 1530 which has not yet received an INIT signal [x86] 1531 KVM_MP_STATE_INIT_RECEIVED the vcpu has received an INIT signal, and is 1532 now ready for a SIPI [x86] 1533 KVM_MP_STATE_HALTED the vcpu has executed a HLT instruction and 1534 is waiting for an interrupt [x86] 1535 KVM_MP_STATE_SIPI_RECEIVED the vcpu has just received a SIPI (vector 1536 accessible via KVM_GET_VCPU_EVENTS) [x86] 1537 KVM_MP_STATE_STOPPED the vcpu is stopped [s390,arm64,riscv] 1538 KVM_MP_STATE_CHECK_STOP the vcpu is in a special error state [s390] 1539 KVM_MP_STATE_OPERATING the vcpu is operating (running or halted) 1540 [s390] 1541 KVM_MP_STATE_LOAD the vcpu is in a special load/startup state 1542 [s390] 1543 KVM_MP_STATE_SUSPENDED the vcpu is in a suspend state and is waiting 1544 for a wakeup event [arm64] 1545 ========================== =============================================== 1546 1547On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an 1548in-kernel irqchip, the multiprocessing state must be maintained by userspace on 1549these architectures. 1550 1551For arm64: 1552^^^^^^^^^^ 1553 1554If a vCPU is in the KVM_MP_STATE_SUSPENDED state, KVM will emulate the 1555architectural execution of a WFI instruction. 1556 1557If a wakeup event is recognized, KVM will exit to userspace with a 1558KVM_SYSTEM_EVENT exit, where the event type is KVM_SYSTEM_EVENT_WAKEUP. If 1559userspace wants to honor the wakeup, it must set the vCPU's MP state to 1560KVM_MP_STATE_RUNNABLE. If it does not, KVM will continue to await a wakeup 1561event in subsequent calls to KVM_RUN. 1562 1563.. warning:: 1564 1565 If userspace intends to keep the vCPU in a SUSPENDED state, it is 1566 strongly recommended that userspace take action to suppress the 1567 wakeup event (such as masking an interrupt). Otherwise, subsequent 1568 calls to KVM_RUN will immediately exit with a KVM_SYSTEM_EVENT_WAKEUP 1569 event and inadvertently waste CPU cycles. 1570 1571 Additionally, if userspace takes action to suppress a wakeup event, 1572 it is strongly recommended that it also restores the vCPU to its 1573 original state when the vCPU is made RUNNABLE again. For example, 1574 if userspace masked a pending interrupt to suppress the wakeup, 1575 the interrupt should be unmasked before returning control to the 1576 guest. 1577 1578For riscv: 1579^^^^^^^^^^ 1580 1581The only states that are valid are KVM_MP_STATE_STOPPED and 1582KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not. 1583 1584On LoongArch, only the KVM_MP_STATE_RUNNABLE state is used to reflect 1585whether the vcpu is runnable. 1586 15874.39 KVM_SET_MP_STATE 1588--------------------- 1589 1590:Capability: KVM_CAP_MP_STATE 1591:Architectures: x86, s390, arm64, riscv, loongarch 1592:Type: vcpu ioctl 1593:Parameters: struct kvm_mp_state (in) 1594:Returns: 0 on success; -1 on error 1595 1596Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for 1597arguments. 1598 1599On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an 1600in-kernel irqchip, the multiprocessing state must be maintained by userspace on 1601these architectures. 1602 1603For arm64/riscv: 1604^^^^^^^^^^^^^^^^ 1605 1606The only states that are valid are KVM_MP_STATE_STOPPED and 1607KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not. 1608 1609On LoongArch, only the KVM_MP_STATE_RUNNABLE state is used to reflect 1610whether the vcpu is runnable. 1611 16124.40 KVM_SET_IDENTITY_MAP_ADDR 1613------------------------------ 1614 1615:Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR 1616:Architectures: x86 1617:Type: vm ioctl 1618:Parameters: unsigned long identity (in) 1619:Returns: 0 on success, -1 on error 1620 1621This ioctl defines the physical address of a one-page region in the guest 1622physical address space. The region must be within the first 4GB of the 1623guest physical address space and must not conflict with any memory slot 1624or any mmio address. The guest may malfunction if it accesses this memory 1625region. 1626 1627Setting the address to 0 will result in resetting the address to its default 1628(0xfffbc000). 1629 1630This ioctl is required on Intel-based hosts. This is needed on Intel hardware 1631because of a quirk in the virtualization implementation (see the internals 1632documentation when it pops into existence). 1633 1634Fails if any VCPU has already been created. 1635 16364.41 KVM_SET_BOOT_CPU_ID 1637------------------------ 1638 1639:Capability: KVM_CAP_SET_BOOT_CPU_ID 1640:Architectures: x86 1641:Type: vm ioctl 1642:Parameters: unsigned long vcpu_id 1643:Returns: 0 on success, -1 on error 1644 1645Define which vcpu is the Bootstrap Processor (BSP). Values are the same 1646as the vcpu id in KVM_CREATE_VCPU. If this ioctl is not called, the default 1647is vcpu 0. This ioctl has to be called before vcpu creation, 1648otherwise it will return EBUSY error. 1649 1650 16514.42 KVM_GET_XSAVE 1652------------------ 1653 1654:Capability: KVM_CAP_XSAVE 1655:Architectures: x86 1656:Type: vcpu ioctl 1657:Parameters: struct kvm_xsave (out) 1658:Returns: 0 on success, -1 on error 1659 1660 1661:: 1662 1663 struct kvm_xsave { 1664 __u32 region[1024]; 1665 __u32 extra[0]; 1666 }; 1667 1668This ioctl would copy current vcpu's xsave struct to the userspace. 1669 1670 16714.43 KVM_SET_XSAVE 1672------------------ 1673 1674:Capability: KVM_CAP_XSAVE and KVM_CAP_XSAVE2 1675:Architectures: x86 1676:Type: vcpu ioctl 1677:Parameters: struct kvm_xsave (in) 1678:Returns: 0 on success, -1 on error 1679 1680:: 1681 1682 1683 struct kvm_xsave { 1684 __u32 region[1024]; 1685 __u32 extra[0]; 1686 }; 1687 1688This ioctl would copy userspace's xsave struct to the kernel. It copies 1689as many bytes as are returned by KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2), 1690when invoked on the vm file descriptor. The size value returned by 1691KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) will always be at least 4096. 1692Currently, it is only greater than 4096 if a dynamic feature has been 1693enabled with ``arch_prctl()``, but this may change in the future. 1694 1695The offsets of the state save areas in struct kvm_xsave follow the 1696contents of CPUID leaf 0xD on the host. 1697 1698 16994.44 KVM_GET_XCRS 1700----------------- 1701 1702:Capability: KVM_CAP_XCRS 1703:Architectures: x86 1704:Type: vcpu ioctl 1705:Parameters: struct kvm_xcrs (out) 1706:Returns: 0 on success, -1 on error 1707 1708:: 1709 1710 struct kvm_xcr { 1711 __u32 xcr; 1712 __u32 reserved; 1713 __u64 value; 1714 }; 1715 1716 struct kvm_xcrs { 1717 __u32 nr_xcrs; 1718 __u32 flags; 1719 struct kvm_xcr xcrs[KVM_MAX_XCRS]; 1720 __u64 padding[16]; 1721 }; 1722 1723This ioctl would copy current vcpu's xcrs to the userspace. 1724 1725 17264.45 KVM_SET_XCRS 1727----------------- 1728 1729:Capability: KVM_CAP_XCRS 1730:Architectures: x86 1731:Type: vcpu ioctl 1732:Parameters: struct kvm_xcrs (in) 1733:Returns: 0 on success, -1 on error 1734 1735:: 1736 1737 struct kvm_xcr { 1738 __u32 xcr; 1739 __u32 reserved; 1740 __u64 value; 1741 }; 1742 1743 struct kvm_xcrs { 1744 __u32 nr_xcrs; 1745 __u32 flags; 1746 struct kvm_xcr xcrs[KVM_MAX_XCRS]; 1747 __u64 padding[16]; 1748 }; 1749 1750This ioctl would set vcpu's xcr to the value userspace specified. 1751 1752 17534.46 KVM_GET_SUPPORTED_CPUID 1754---------------------------- 1755 1756:Capability: KVM_CAP_EXT_CPUID 1757:Architectures: x86 1758:Type: system ioctl 1759:Parameters: struct kvm_cpuid2 (in/out) 1760:Returns: 0 on success, -1 on error 1761 1762:: 1763 1764 struct kvm_cpuid2 { 1765 __u32 nent; 1766 __u32 padding; 1767 struct kvm_cpuid_entry2 entries[0]; 1768 }; 1769 1770 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0) 1771 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */ 1772 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */ 1773 1774 struct kvm_cpuid_entry2 { 1775 __u32 function; 1776 __u32 index; 1777 __u32 flags; 1778 __u32 eax; 1779 __u32 ebx; 1780 __u32 ecx; 1781 __u32 edx; 1782 __u32 padding[3]; 1783 }; 1784 1785This ioctl returns x86 cpuid features which are supported by both the 1786hardware and kvm in its default configuration. Userspace can use the 1787information returned by this ioctl to construct cpuid information (for 1788KVM_SET_CPUID2) that is consistent with hardware, kernel, and 1789userspace capabilities, and with user requirements (for example, the 1790user may wish to constrain cpuid to emulate older hardware, or for 1791feature consistency across a cluster). 1792 1793Dynamically-enabled feature bits need to be requested with 1794``arch_prctl()`` before calling this ioctl. Feature bits that have not 1795been requested are excluded from the result. 1796 1797Note that certain capabilities, such as KVM_CAP_X86_DISABLE_EXITS, may 1798expose cpuid features (e.g. MONITOR) which are not supported by kvm in 1799its default configuration. If userspace enables such capabilities, it 1800is responsible for modifying the results of this ioctl appropriately. 1801 1802Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure 1803with the 'nent' field indicating the number of entries in the variable-size 1804array 'entries'. If the number of entries is too low to describe the cpu 1805capabilities, an error (E2BIG) is returned. If the number is too high, 1806the 'nent' field is adjusted and an error (ENOMEM) is returned. If the 1807number is just right, the 'nent' field is adjusted to the number of valid 1808entries in the 'entries' array, which is then filled. 1809 1810The entries returned are the host cpuid as returned by the cpuid instruction, 1811with unknown or unsupported features masked out. Some features (for example, 1812x2apic), may not be present in the host cpu, but are exposed by kvm if it can 1813emulate them efficiently. The fields in each entry are defined as follows: 1814 1815 function: 1816 the eax value used to obtain the entry 1817 1818 index: 1819 the ecx value used to obtain the entry (for entries that are 1820 affected by ecx) 1821 1822 flags: 1823 an OR of zero or more of the following: 1824 1825 KVM_CPUID_FLAG_SIGNIFCANT_INDEX: 1826 if the index field is valid 1827 1828 eax, ebx, ecx, edx: 1829 the values returned by the cpuid instruction for 1830 this function/index combination 1831 1832x2APIC (CPUID leaf 1, ecx[21) and TSC deadline timer (CPUID leaf 1, ecx[24]) 1833may be returned as true, but they depend on KVM_CREATE_IRQCHIP for in-kernel 1834emulation of the local APIC. TSC deadline timer support is also reported via:: 1835 1836 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER) 1837 1838if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the 1839feature in userspace, then you can enable the feature for KVM_SET_CPUID2. 1840 1841Enabling x2APIC in KVM_SET_CPUID2 requires KVM_CREATE_IRQCHIP as KVM doesn't 1842support forwarding x2APIC MSR accesses to userspace, i.e. KVM does not support 1843emulating x2APIC in userspace. 1844 18454.47 KVM_PPC_GET_PVINFO 1846----------------------- 1847 1848:Capability: KVM_CAP_PPC_GET_PVINFO 1849:Architectures: ppc 1850:Type: vm ioctl 1851:Parameters: struct kvm_ppc_pvinfo (out) 1852:Returns: 0 on success, !0 on error 1853 1854:: 1855 1856 struct kvm_ppc_pvinfo { 1857 __u32 flags; 1858 __u32 hcall[4]; 1859 __u8 pad[108]; 1860 }; 1861 1862This ioctl fetches PV specific information that need to be passed to the guest 1863using the device tree or other means from vm context. 1864 1865The hcall array defines 4 instructions that make up a hypercall. 1866 1867If any additional field gets added to this structure later on, a bit for that 1868additional piece of information will be set in the flags bitmap. 1869 1870The flags bitmap is defined as:: 1871 1872 /* the host supports the ePAPR idle hcall 1873 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1<<0) 1874 18754.52 KVM_SET_GSI_ROUTING 1876------------------------ 1877 1878:Capability: KVM_CAP_IRQ_ROUTING 1879:Architectures: x86 s390 arm64 1880:Type: vm ioctl 1881:Parameters: struct kvm_irq_routing (in) 1882:Returns: 0 on success, -1 on error 1883 1884Sets the GSI routing table entries, overwriting any previously set entries. 1885 1886On arm64, GSI routing has the following limitation: 1887 1888- GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD. 1889 1890:: 1891 1892 struct kvm_irq_routing { 1893 __u32 nr; 1894 __u32 flags; 1895 struct kvm_irq_routing_entry entries[0]; 1896 }; 1897 1898No flags are specified so far, the corresponding field must be set to zero. 1899 1900:: 1901 1902 struct kvm_irq_routing_entry { 1903 __u32 gsi; 1904 __u32 type; 1905 __u32 flags; 1906 __u32 pad; 1907 union { 1908 struct kvm_irq_routing_irqchip irqchip; 1909 struct kvm_irq_routing_msi msi; 1910 struct kvm_irq_routing_s390_adapter adapter; 1911 struct kvm_irq_routing_hv_sint hv_sint; 1912 struct kvm_irq_routing_xen_evtchn xen_evtchn; 1913 __u32 pad[8]; 1914 } u; 1915 }; 1916 1917 /* gsi routing entry types */ 1918 #define KVM_IRQ_ROUTING_IRQCHIP 1 1919 #define KVM_IRQ_ROUTING_MSI 2 1920 #define KVM_IRQ_ROUTING_S390_ADAPTER 3 1921 #define KVM_IRQ_ROUTING_HV_SINT 4 1922 #define KVM_IRQ_ROUTING_XEN_EVTCHN 5 1923 1924On s390, adding a KVM_IRQ_ROUTING_S390_ADAPTER is rejected on ucontrol VMs with 1925error -EINVAL. 1926 1927flags: 1928 1929- KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry 1930 type, specifies that the devid field contains a valid value. The per-VM 1931 KVM_CAP_MSI_DEVID capability advertises the requirement to provide 1932 the device ID. If this capability is not available, userspace should 1933 never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail. 1934- zero otherwise 1935 1936:: 1937 1938 struct kvm_irq_routing_irqchip { 1939 __u32 irqchip; 1940 __u32 pin; 1941 }; 1942 1943 struct kvm_irq_routing_msi { 1944 __u32 address_lo; 1945 __u32 address_hi; 1946 __u32 data; 1947 union { 1948 __u32 pad; 1949 __u32 devid; 1950 }; 1951 }; 1952 1953If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier 1954for the device that wrote the MSI message. For PCI, this is usually a 1955BDF identifier in the lower 16 bits. 1956 1957On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS 1958feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled, 1959address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of 1960address_hi must be zero. 1961 1962:: 1963 1964 struct kvm_irq_routing_s390_adapter { 1965 __u64 ind_addr; 1966 __u64 summary_addr; 1967 __u64 ind_offset; 1968 __u32 summary_offset; 1969 __u32 adapter_id; 1970 }; 1971 1972 struct kvm_irq_routing_hv_sint { 1973 __u32 vcpu; 1974 __u32 sint; 1975 }; 1976 1977 struct kvm_irq_routing_xen_evtchn { 1978 __u32 port; 1979 __u32 vcpu; 1980 __u32 priority; 1981 }; 1982 1983 1984When KVM_CAP_XEN_HVM includes the KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL bit 1985in its indication of supported features, routing to Xen event channels 1986is supported. Although the priority field is present, only the value 1987KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL is supported, which means delivery by 19882 level event channels. FIFO event channel support may be added in 1989the future. 1990 1991 19924.55 KVM_SET_TSC_KHZ 1993-------------------- 1994 1995:Capability: KVM_CAP_TSC_CONTROL / KVM_CAP_VM_TSC_CONTROL 1996:Architectures: x86 1997:Type: vcpu ioctl / vm ioctl 1998:Parameters: virtual tsc_khz 1999:Returns: 0 on success, -1 on error 2000 2001Specifies the tsc frequency for the virtual machine. The unit of the 2002frequency is KHz. 2003 2004If the KVM_CAP_VM_TSC_CONTROL capability is advertised, this can also 2005be used as a vm ioctl to set the initial tsc frequency of subsequently 2006created vCPUs. 2007 20084.56 KVM_GET_TSC_KHZ 2009-------------------- 2010 2011:Capability: KVM_CAP_GET_TSC_KHZ / KVM_CAP_VM_TSC_CONTROL 2012:Architectures: x86 2013:Type: vcpu ioctl / vm ioctl 2014:Parameters: none 2015:Returns: virtual tsc-khz on success, negative value on error 2016 2017Returns the tsc frequency of the guest. The unit of the return value is 2018KHz. If the host has unstable tsc this ioctl returns -EIO instead as an 2019error. 2020 2021 20224.57 KVM_GET_LAPIC 2023------------------ 2024 2025:Capability: KVM_CAP_IRQCHIP 2026:Architectures: x86 2027:Type: vcpu ioctl 2028:Parameters: struct kvm_lapic_state (out) 2029:Returns: 0 on success, -1 on error 2030 2031:: 2032 2033 #define KVM_APIC_REG_SIZE 0x400 2034 struct kvm_lapic_state { 2035 char regs[KVM_APIC_REG_SIZE]; 2036 }; 2037 2038Reads the Local APIC registers and copies them into the input argument. The 2039data format and layout are the same as documented in the architecture manual. 2040 2041If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is 2042enabled, then the format of APIC_ID register depends on the APIC mode 2043(reported by MSR_IA32_APICBASE) of its VCPU. x2APIC stores APIC ID in 2044the APIC_ID register (bytes 32-35). xAPIC only allows an 8-bit APIC ID 2045which is stored in bits 31-24 of the APIC register, or equivalently in 2046byte 35 of struct kvm_lapic_state's regs field. KVM_GET_LAPIC must then 2047be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR. 2048 2049If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state 2050always uses xAPIC format. 2051 2052 20534.58 KVM_SET_LAPIC 2054------------------ 2055 2056:Capability: KVM_CAP_IRQCHIP 2057:Architectures: x86 2058:Type: vcpu ioctl 2059:Parameters: struct kvm_lapic_state (in) 2060:Returns: 0 on success, -1 on error 2061 2062:: 2063 2064 #define KVM_APIC_REG_SIZE 0x400 2065 struct kvm_lapic_state { 2066 char regs[KVM_APIC_REG_SIZE]; 2067 }; 2068 2069Copies the input argument into the Local APIC registers. The data format 2070and layout are the same as documented in the architecture manual. 2071 2072The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's 2073regs field) depends on the state of the KVM_CAP_X2APIC_API capability. 2074See the note in KVM_GET_LAPIC. 2075 2076 20774.59 KVM_IOEVENTFD 2078------------------ 2079 2080:Capability: KVM_CAP_IOEVENTFD 2081:Architectures: all 2082:Type: vm ioctl 2083:Parameters: struct kvm_ioeventfd (in) 2084:Returns: 0 on success, !0 on error 2085 2086This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address 2087within the guest. A guest write in the registered address will signal the 2088provided event instead of triggering an exit. 2089 2090:: 2091 2092 struct kvm_ioeventfd { 2093 __u64 datamatch; 2094 __u64 addr; /* legal pio/mmio address */ 2095 __u32 len; /* 0, 1, 2, 4, or 8 bytes */ 2096 __s32 fd; 2097 __u32 flags; 2098 __u8 pad[36]; 2099 }; 2100 2101For the special case of virtio-ccw devices on s390, the ioevent is matched 2102to a subchannel/virtqueue tuple instead. 2103 2104The following flags are defined:: 2105 2106 #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch) 2107 #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio) 2108 #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign) 2109 #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \ 2110 (1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify) 2111 2112If datamatch flag is set, the event will be signaled only if the written value 2113to the registered address is equal to datamatch in struct kvm_ioeventfd. 2114 2115For virtio-ccw devices, addr contains the subchannel id and datamatch the 2116virtqueue index. 2117 2118With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and 2119the kernel will ignore the length of guest write and may get a faster vmexit. 2120The speedup may only apply to specific architectures, but the ioeventfd will 2121work anyway. 2122 21234.60 KVM_DIRTY_TLB 2124------------------ 2125 2126:Capability: KVM_CAP_SW_TLB 2127:Architectures: ppc 2128:Type: vcpu ioctl 2129:Parameters: struct kvm_dirty_tlb (in) 2130:Returns: 0 on success, -1 on error 2131 2132:: 2133 2134 struct kvm_dirty_tlb { 2135 __u64 bitmap; 2136 __u32 num_dirty; 2137 }; 2138 2139This must be called whenever userspace has changed an entry in the shared 2140TLB, prior to calling KVM_RUN on the associated vcpu. 2141 2142The "bitmap" field is the userspace address of an array. This array 2143consists of a number of bits, equal to the total number of TLB entries as 2144determined by the last successful call to ``KVM_ENABLE_CAP(KVM_CAP_SW_TLB)``, 2145rounded up to the nearest multiple of 64. 2146 2147Each bit corresponds to one TLB entry, ordered the same as in the shared TLB 2148array. 2149 2150The array is little-endian: the bit 0 is the least significant bit of the 2151first byte, bit 8 is the least significant bit of the second byte, etc. 2152This avoids any complications with differing word sizes. 2153 2154The "num_dirty" field is a performance hint for KVM to determine whether it 2155should skip processing the bitmap and just invalidate everything. It must 2156be set to the number of set bits in the bitmap. 2157 2158 21594.62 KVM_CREATE_SPAPR_TCE 2160------------------------- 2161 2162:Capability: KVM_CAP_SPAPR_TCE 2163:Architectures: powerpc 2164:Type: vm ioctl 2165:Parameters: struct kvm_create_spapr_tce (in) 2166:Returns: file descriptor for manipulating the created TCE table 2167 2168This creates a virtual TCE (translation control entry) table, which 2169is an IOMMU for PAPR-style virtual I/O. It is used to translate 2170logical addresses used in virtual I/O into guest physical addresses, 2171and provides a scatter/gather capability for PAPR virtual I/O. 2172 2173:: 2174 2175 /* for KVM_CAP_SPAPR_TCE */ 2176 struct kvm_create_spapr_tce { 2177 __u64 liobn; 2178 __u32 window_size; 2179 }; 2180 2181The liobn field gives the logical IO bus number for which to create a 2182TCE table. The window_size field specifies the size of the DMA window 2183which this TCE table will translate - the table will contain one 64 2184bit TCE entry for every 4kiB of the DMA window. 2185 2186When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE 2187table has been created using this ioctl(), the kernel will handle it 2188in real mode, updating the TCE table. H_PUT_TCE calls for other 2189liobns will cause a vm exit and must be handled by userspace. 2190 2191The return value is a file descriptor which can be passed to mmap(2) 2192to map the created TCE table into userspace. This lets userspace read 2193the entries written by kernel-handled H_PUT_TCE calls, and also lets 2194userspace update the TCE table directly which is useful in some 2195circumstances. 2196 2197 21984.64 KVM_NMI 2199------------ 2200 2201:Capability: KVM_CAP_USER_NMI 2202:Architectures: x86 2203:Type: vcpu ioctl 2204:Parameters: none 2205:Returns: 0 on success, -1 on error 2206 2207Queues an NMI on the thread's vcpu. Note this is well defined only 2208when KVM_CREATE_IRQCHIP has not been called, since this is an interface 2209between the virtual cpu core and virtual local APIC. After KVM_CREATE_IRQCHIP 2210has been called, this interface is completely emulated within the kernel. 2211 2212To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the 2213following algorithm: 2214 2215 - pause the vcpu 2216 - read the local APIC's state (KVM_GET_LAPIC) 2217 - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1) 2218 - if so, issue KVM_NMI 2219 - resume the vcpu 2220 2221Some guests configure the LINT1 NMI input to cause a panic, aiding in 2222debugging. 2223 2224 22254.65 KVM_S390_UCAS_MAP 2226---------------------- 2227 2228:Capability: KVM_CAP_S390_UCONTROL 2229:Architectures: s390 2230:Type: vcpu ioctl 2231:Parameters: struct kvm_s390_ucas_mapping (in) 2232:Returns: 0 in case of success 2233 2234The parameter is defined like this:: 2235 2236 struct kvm_s390_ucas_mapping { 2237 __u64 user_addr; 2238 __u64 vcpu_addr; 2239 __u64 length; 2240 }; 2241 2242This ioctl maps the memory at "user_addr" with the length "length" to 2243the vcpu's address space starting at "vcpu_addr". All parameters need to 2244be aligned by 1 megabyte. 2245 2246 22474.66 KVM_S390_UCAS_UNMAP 2248------------------------ 2249 2250:Capability: KVM_CAP_S390_UCONTROL 2251:Architectures: s390 2252:Type: vcpu ioctl 2253:Parameters: struct kvm_s390_ucas_mapping (in) 2254:Returns: 0 in case of success 2255 2256The parameter is defined like this:: 2257 2258 struct kvm_s390_ucas_mapping { 2259 __u64 user_addr; 2260 __u64 vcpu_addr; 2261 __u64 length; 2262 }; 2263 2264This ioctl unmaps the memory in the vcpu's address space starting at 2265"vcpu_addr" with the length "length". The field "user_addr" is ignored. 2266All parameters need to be aligned by 1 megabyte. 2267 2268 22694.67 KVM_S390_VCPU_FAULT 2270------------------------ 2271 2272:Capability: KVM_CAP_S390_UCONTROL 2273:Architectures: s390 2274:Type: vcpu ioctl 2275:Parameters: vcpu absolute address (in) 2276:Returns: 0 in case of success 2277 2278This call creates a page table entry on the virtual cpu's address space 2279(for user controlled virtual machines) or the virtual machine's address 2280space (for regular virtual machines). This only works for minor faults, 2281thus it's recommended to access subject memory page via the user page 2282table upfront. This is useful to handle validity intercepts for user 2283controlled virtual machines to fault in the virtual cpu's lowcore pages 2284prior to calling the KVM_RUN ioctl. 2285 2286 22874.68 KVM_SET_ONE_REG 2288-------------------- 2289 2290:Capability: KVM_CAP_ONE_REG 2291:Architectures: all 2292:Type: vcpu ioctl 2293:Parameters: struct kvm_one_reg (in) 2294:Returns: 0 on success, negative value on failure 2295 2296Errors: 2297 2298 ====== ============================================================ 2299 ENOENT no such register 2300 EINVAL invalid register ID, or no such register or used with VMs in 2301 protected virtualization mode on s390 2302 EPERM (arm64) register access not allowed before vcpu finalization 2303 EBUSY (riscv) changing register value not allowed after the vcpu 2304 has run at least once 2305 ====== ============================================================ 2306 2307(These error codes are indicative only: do not rely on a specific error 2308code being returned in a specific situation.) 2309 2310:: 2311 2312 struct kvm_one_reg { 2313 __u64 id; 2314 __u64 addr; 2315 }; 2316 2317Using this ioctl, a single vcpu register can be set to a specific value 2318defined by user space with the passed in struct kvm_one_reg, where id 2319refers to the register identifier as described below and addr is a pointer 2320to a variable with the respective size. There can be architecture agnostic 2321and architecture specific registers. Each have their own range of operation 2322and their own constants and width. To keep track of the implemented 2323registers, find a list below: 2324 2325 ======= =============================== ============ 2326 Arch Register Width (bits) 2327 ======= =============================== ============ 2328 PPC KVM_REG_PPC_HIOR 64 2329 PPC KVM_REG_PPC_IAC1 64 2330 PPC KVM_REG_PPC_IAC2 64 2331 PPC KVM_REG_PPC_IAC3 64 2332 PPC KVM_REG_PPC_IAC4 64 2333 PPC KVM_REG_PPC_DAC1 64 2334 PPC KVM_REG_PPC_DAC2 64 2335 PPC KVM_REG_PPC_DABR 64 2336 PPC KVM_REG_PPC_DSCR 64 2337 PPC KVM_REG_PPC_PURR 64 2338 PPC KVM_REG_PPC_SPURR 64 2339 PPC KVM_REG_PPC_DAR 64 2340 PPC KVM_REG_PPC_DSISR 32 2341 PPC KVM_REG_PPC_AMR 64 2342 PPC KVM_REG_PPC_UAMOR 64 2343 PPC KVM_REG_PPC_MMCR0 64 2344 PPC KVM_REG_PPC_MMCR1 64 2345 PPC KVM_REG_PPC_MMCRA 64 2346 PPC KVM_REG_PPC_MMCR2 64 2347 PPC KVM_REG_PPC_MMCRS 64 2348 PPC KVM_REG_PPC_MMCR3 64 2349 PPC KVM_REG_PPC_SIAR 64 2350 PPC KVM_REG_PPC_SDAR 64 2351 PPC KVM_REG_PPC_SIER 64 2352 PPC KVM_REG_PPC_SIER2 64 2353 PPC KVM_REG_PPC_SIER3 64 2354 PPC KVM_REG_PPC_PMC1 32 2355 PPC KVM_REG_PPC_PMC2 32 2356 PPC KVM_REG_PPC_PMC3 32 2357 PPC KVM_REG_PPC_PMC4 32 2358 PPC KVM_REG_PPC_PMC5 32 2359 PPC KVM_REG_PPC_PMC6 32 2360 PPC KVM_REG_PPC_PMC7 32 2361 PPC KVM_REG_PPC_PMC8 32 2362 PPC KVM_REG_PPC_FPR0 64 2363 ... 2364 PPC KVM_REG_PPC_FPR31 64 2365 PPC KVM_REG_PPC_VR0 128 2366 ... 2367 PPC KVM_REG_PPC_VR31 128 2368 PPC KVM_REG_PPC_VSR0 128 2369 ... 2370 PPC KVM_REG_PPC_VSR31 128 2371 PPC KVM_REG_PPC_FPSCR 64 2372 PPC KVM_REG_PPC_VSCR 32 2373 PPC KVM_REG_PPC_VPA_ADDR 64 2374 PPC KVM_REG_PPC_VPA_SLB 128 2375 PPC KVM_REG_PPC_VPA_DTL 128 2376 PPC KVM_REG_PPC_EPCR 32 2377 PPC KVM_REG_PPC_EPR 32 2378 PPC KVM_REG_PPC_TCR 32 2379 PPC KVM_REG_PPC_TSR 32 2380 PPC KVM_REG_PPC_OR_TSR 32 2381 PPC KVM_REG_PPC_CLEAR_TSR 32 2382 PPC KVM_REG_PPC_MAS0 32 2383 PPC KVM_REG_PPC_MAS1 32 2384 PPC KVM_REG_PPC_MAS2 64 2385 PPC KVM_REG_PPC_MAS7_3 64 2386 PPC KVM_REG_PPC_MAS4 32 2387 PPC KVM_REG_PPC_MAS6 32 2388 PPC KVM_REG_PPC_MMUCFG 32 2389 PPC KVM_REG_PPC_TLB0CFG 32 2390 PPC KVM_REG_PPC_TLB1CFG 32 2391 PPC KVM_REG_PPC_TLB2CFG 32 2392 PPC KVM_REG_PPC_TLB3CFG 32 2393 PPC KVM_REG_PPC_TLB0PS 32 2394 PPC KVM_REG_PPC_TLB1PS 32 2395 PPC KVM_REG_PPC_TLB2PS 32 2396 PPC KVM_REG_PPC_TLB3PS 32 2397 PPC KVM_REG_PPC_EPTCFG 32 2398 PPC KVM_REG_PPC_ICP_STATE 64 2399 PPC KVM_REG_PPC_VP_STATE 128 2400 PPC KVM_REG_PPC_TB_OFFSET 64 2401 PPC KVM_REG_PPC_SPMC1 32 2402 PPC KVM_REG_PPC_SPMC2 32 2403 PPC KVM_REG_PPC_IAMR 64 2404 PPC KVM_REG_PPC_TFHAR 64 2405 PPC KVM_REG_PPC_TFIAR 64 2406 PPC KVM_REG_PPC_TEXASR 64 2407 PPC KVM_REG_PPC_FSCR 64 2408 PPC KVM_REG_PPC_PSPB 32 2409 PPC KVM_REG_PPC_EBBHR 64 2410 PPC KVM_REG_PPC_EBBRR 64 2411 PPC KVM_REG_PPC_BESCR 64 2412 PPC KVM_REG_PPC_TAR 64 2413 PPC KVM_REG_PPC_DPDES 64 2414 PPC KVM_REG_PPC_DAWR 64 2415 PPC KVM_REG_PPC_DAWRX 64 2416 PPC KVM_REG_PPC_CIABR 64 2417 PPC KVM_REG_PPC_IC 64 2418 PPC KVM_REG_PPC_VTB 64 2419 PPC KVM_REG_PPC_CSIGR 64 2420 PPC KVM_REG_PPC_TACR 64 2421 PPC KVM_REG_PPC_TCSCR 64 2422 PPC KVM_REG_PPC_PID 64 2423 PPC KVM_REG_PPC_ACOP 64 2424 PPC KVM_REG_PPC_VRSAVE 32 2425 PPC KVM_REG_PPC_LPCR 32 2426 PPC KVM_REG_PPC_LPCR_64 64 2427 PPC KVM_REG_PPC_PPR 64 2428 PPC KVM_REG_PPC_ARCH_COMPAT 32 2429 PPC KVM_REG_PPC_DABRX 32 2430 PPC KVM_REG_PPC_WORT 64 2431 PPC KVM_REG_PPC_SPRG9 64 2432 PPC KVM_REG_PPC_DBSR 32 2433 PPC KVM_REG_PPC_TIDR 64 2434 PPC KVM_REG_PPC_PSSCR 64 2435 PPC KVM_REG_PPC_DEC_EXPIRY 64 2436 PPC KVM_REG_PPC_PTCR 64 2437 PPC KVM_REG_PPC_HASHKEYR 64 2438 PPC KVM_REG_PPC_HASHPKEYR 64 2439 PPC KVM_REG_PPC_DAWR1 64 2440 PPC KVM_REG_PPC_DAWRX1 64 2441 PPC KVM_REG_PPC_DEXCR 64 2442 PPC KVM_REG_PPC_TM_GPR0 64 2443 ... 2444 PPC KVM_REG_PPC_TM_GPR31 64 2445 PPC KVM_REG_PPC_TM_VSR0 128 2446 ... 2447 PPC KVM_REG_PPC_TM_VSR63 128 2448 PPC KVM_REG_PPC_TM_CR 64 2449 PPC KVM_REG_PPC_TM_LR 64 2450 PPC KVM_REG_PPC_TM_CTR 64 2451 PPC KVM_REG_PPC_TM_FPSCR 64 2452 PPC KVM_REG_PPC_TM_AMR 64 2453 PPC KVM_REG_PPC_TM_PPR 64 2454 PPC KVM_REG_PPC_TM_VRSAVE 64 2455 PPC KVM_REG_PPC_TM_VSCR 32 2456 PPC KVM_REG_PPC_TM_DSCR 64 2457 PPC KVM_REG_PPC_TM_TAR 64 2458 PPC KVM_REG_PPC_TM_XER 64 2459 2460 MIPS KVM_REG_MIPS_R0 64 2461 ... 2462 MIPS KVM_REG_MIPS_R31 64 2463 MIPS KVM_REG_MIPS_HI 64 2464 MIPS KVM_REG_MIPS_LO 64 2465 MIPS KVM_REG_MIPS_PC 64 2466 MIPS KVM_REG_MIPS_CP0_INDEX 32 2467 MIPS KVM_REG_MIPS_CP0_ENTRYLO0 64 2468 MIPS KVM_REG_MIPS_CP0_ENTRYLO1 64 2469 MIPS KVM_REG_MIPS_CP0_CONTEXT 64 2470 MIPS KVM_REG_MIPS_CP0_CONTEXTCONFIG 32 2471 MIPS KVM_REG_MIPS_CP0_USERLOCAL 64 2472 MIPS KVM_REG_MIPS_CP0_XCONTEXTCONFIG 64 2473 MIPS KVM_REG_MIPS_CP0_PAGEMASK 32 2474 MIPS KVM_REG_MIPS_CP0_PAGEGRAIN 32 2475 MIPS KVM_REG_MIPS_CP0_SEGCTL0 64 2476 MIPS KVM_REG_MIPS_CP0_SEGCTL1 64 2477 MIPS KVM_REG_MIPS_CP0_SEGCTL2 64 2478 MIPS KVM_REG_MIPS_CP0_PWBASE 64 2479 MIPS KVM_REG_MIPS_CP0_PWFIELD 64 2480 MIPS KVM_REG_MIPS_CP0_PWSIZE 64 2481 MIPS KVM_REG_MIPS_CP0_WIRED 32 2482 MIPS KVM_REG_MIPS_CP0_PWCTL 32 2483 MIPS KVM_REG_MIPS_CP0_HWRENA 32 2484 MIPS KVM_REG_MIPS_CP0_BADVADDR 64 2485 MIPS KVM_REG_MIPS_CP0_BADINSTR 32 2486 MIPS KVM_REG_MIPS_CP0_BADINSTRP 32 2487 MIPS KVM_REG_MIPS_CP0_COUNT 32 2488 MIPS KVM_REG_MIPS_CP0_ENTRYHI 64 2489 MIPS KVM_REG_MIPS_CP0_COMPARE 32 2490 MIPS KVM_REG_MIPS_CP0_STATUS 32 2491 MIPS KVM_REG_MIPS_CP0_INTCTL 32 2492 MIPS KVM_REG_MIPS_CP0_CAUSE 32 2493 MIPS KVM_REG_MIPS_CP0_EPC 64 2494 MIPS KVM_REG_MIPS_CP0_PRID 32 2495 MIPS KVM_REG_MIPS_CP0_EBASE 64 2496 MIPS KVM_REG_MIPS_CP0_CONFIG 32 2497 MIPS KVM_REG_MIPS_CP0_CONFIG1 32 2498 MIPS KVM_REG_MIPS_CP0_CONFIG2 32 2499 MIPS KVM_REG_MIPS_CP0_CONFIG3 32 2500 MIPS KVM_REG_MIPS_CP0_CONFIG4 32 2501 MIPS KVM_REG_MIPS_CP0_CONFIG5 32 2502 MIPS KVM_REG_MIPS_CP0_CONFIG7 32 2503 MIPS KVM_REG_MIPS_CP0_XCONTEXT 64 2504 MIPS KVM_REG_MIPS_CP0_ERROREPC 64 2505 MIPS KVM_REG_MIPS_CP0_KSCRATCH1 64 2506 MIPS KVM_REG_MIPS_CP0_KSCRATCH2 64 2507 MIPS KVM_REG_MIPS_CP0_KSCRATCH3 64 2508 MIPS KVM_REG_MIPS_CP0_KSCRATCH4 64 2509 MIPS KVM_REG_MIPS_CP0_KSCRATCH5 64 2510 MIPS KVM_REG_MIPS_CP0_KSCRATCH6 64 2511 MIPS KVM_REG_MIPS_CP0_MAAR(0..63) 64 2512 MIPS KVM_REG_MIPS_COUNT_CTL 64 2513 MIPS KVM_REG_MIPS_COUNT_RESUME 64 2514 MIPS KVM_REG_MIPS_COUNT_HZ 64 2515 MIPS KVM_REG_MIPS_FPR_32(0..31) 32 2516 MIPS KVM_REG_MIPS_FPR_64(0..31) 64 2517 MIPS KVM_REG_MIPS_VEC_128(0..31) 128 2518 MIPS KVM_REG_MIPS_FCR_IR 32 2519 MIPS KVM_REG_MIPS_FCR_CSR 32 2520 MIPS KVM_REG_MIPS_MSA_IR 32 2521 MIPS KVM_REG_MIPS_MSA_CSR 32 2522 ======= =============================== ============ 2523 2524ARM registers are mapped using the lower 32 bits. The upper 16 of that 2525is the register group type, or coprocessor number: 2526 2527ARM core registers have the following id bit patterns:: 2528 2529 0x4020 0000 0010 <index into the kvm_regs struct:16> 2530 2531ARM 32-bit CP15 registers have the following id bit patterns:: 2532 2533 0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3> 2534 2535ARM 64-bit CP15 registers have the following id bit patterns:: 2536 2537 0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3> 2538 2539ARM CCSIDR registers are demultiplexed by CSSELR value:: 2540 2541 0x4020 0000 0011 00 <csselr:8> 2542 2543ARM 32-bit VFP control registers have the following id bit patterns:: 2544 2545 0x4020 0000 0012 1 <regno:12> 2546 2547ARM 64-bit FP registers have the following id bit patterns:: 2548 2549 0x4030 0000 0012 0 <regno:12> 2550 2551ARM firmware pseudo-registers have the following bit pattern:: 2552 2553 0x4030 0000 0014 <regno:16> 2554 2555 2556arm64 registers are mapped using the lower 32 bits. The upper 16 of 2557that is the register group type, or coprocessor number: 2558 2559arm64 core/FP-SIMD registers have the following id bit patterns. Note 2560that the size of the access is variable, as the kvm_regs structure 2561contains elements ranging from 32 to 128 bits. The index is a 32bit 2562value in the kvm_regs structure seen as a 32bit array:: 2563 2564 0x60x0 0000 0010 <index into the kvm_regs struct:16> 2565 2566Specifically: 2567 2568======================= ========= ===== ======================================= 2569 Encoding Register Bits kvm_regs member 2570======================= ========= ===== ======================================= 2571 0x6030 0000 0010 0000 X0 64 regs.regs[0] 2572 0x6030 0000 0010 0002 X1 64 regs.regs[1] 2573 ... 2574 0x6030 0000 0010 003c X30 64 regs.regs[30] 2575 0x6030 0000 0010 003e SP 64 regs.sp 2576 0x6030 0000 0010 0040 PC 64 regs.pc 2577 0x6030 0000 0010 0042 PSTATE 64 regs.pstate 2578 0x6030 0000 0010 0044 SP_EL1 64 sp_el1 2579 0x6030 0000 0010 0046 ELR_EL1 64 elr_el1 2580 0x6030 0000 0010 0048 SPSR_EL1 64 spsr[KVM_SPSR_EL1] (alias SPSR_SVC) 2581 0x6030 0000 0010 004a SPSR_ABT 64 spsr[KVM_SPSR_ABT] 2582 0x6030 0000 0010 004c SPSR_UND 64 spsr[KVM_SPSR_UND] 2583 0x6030 0000 0010 004e SPSR_IRQ 64 spsr[KVM_SPSR_IRQ] 2584 0x6030 0000 0010 0050 SPSR_FIQ 64 spsr[KVM_SPSR_FIQ] 2585 0x6040 0000 0010 0054 V0 128 fp_regs.vregs[0] [1]_ 2586 0x6040 0000 0010 0058 V1 128 fp_regs.vregs[1] [1]_ 2587 ... 2588 0x6040 0000 0010 00d0 V31 128 fp_regs.vregs[31] [1]_ 2589 0x6020 0000 0010 00d4 FPSR 32 fp_regs.fpsr 2590 0x6020 0000 0010 00d5 FPCR 32 fp_regs.fpcr 2591======================= ========= ===== ======================================= 2592 2593.. [1] These encodings are not accepted for SVE-enabled vcpus. See 2594 :ref:`KVM_ARM_VCPU_INIT`. 2595 2596 The equivalent register content can be accessed via bits [127:0] of 2597 the corresponding SVE Zn registers instead for vcpus that have SVE 2598 enabled (see below). 2599 2600arm64 CCSIDR registers are demultiplexed by CSSELR value:: 2601 2602 0x6020 0000 0011 00 <csselr:8> 2603 2604arm64 system registers have the following id bit patterns:: 2605 2606 0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3> 2607 2608.. warning:: 2609 2610 Two system register IDs do not follow the specified pattern. These 2611 are KVM_REG_ARM_TIMER_CVAL and KVM_REG_ARM_TIMER_CNT, which map to 2612 system registers CNTV_CVAL_EL0 and CNTVCT_EL0 respectively. These 2613 two had their values accidentally swapped, which means TIMER_CVAL is 2614 derived from the register encoding for CNTVCT_EL0 and TIMER_CNT is 2615 derived from the register encoding for CNTV_CVAL_EL0. As this is 2616 API, it must remain this way. 2617 2618arm64 firmware pseudo-registers have the following bit pattern:: 2619 2620 0x6030 0000 0014 <regno:16> 2621 2622arm64 SVE registers have the following bit patterns:: 2623 2624 0x6080 0000 0015 00 <n:5> <slice:5> Zn bits[2048*slice + 2047 : 2048*slice] 2625 0x6050 0000 0015 04 <n:4> <slice:5> Pn bits[256*slice + 255 : 256*slice] 2626 0x6050 0000 0015 060 <slice:5> FFR bits[256*slice + 255 : 256*slice] 2627 0x6060 0000 0015 ffff KVM_REG_ARM64_SVE_VLS pseudo-register 2628 2629Access to register IDs where 2048 * slice >= 128 * max_vq will fail with 2630ENOENT. max_vq is the vcpu's maximum supported vector length in 128-bit 2631quadwords: see [2]_ below. 2632 2633These registers are only accessible on vcpus for which SVE is enabled. 2634See KVM_ARM_VCPU_INIT for details. 2635 2636In addition, except for KVM_REG_ARM64_SVE_VLS, these registers are not 2637accessible until the vcpu's SVE configuration has been finalized 2638using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). See KVM_ARM_VCPU_INIT 2639and KVM_ARM_VCPU_FINALIZE for more information about this procedure. 2640 2641KVM_REG_ARM64_SVE_VLS is a pseudo-register that allows the set of vector 2642lengths supported by the vcpu to be discovered and configured by 2643userspace. When transferred to or from user memory via KVM_GET_ONE_REG 2644or KVM_SET_ONE_REG, the value of this register is of type 2645__u64[KVM_ARM64_SVE_VLS_WORDS], and encodes the set of vector lengths as 2646follows:: 2647 2648 __u64 vector_lengths[KVM_ARM64_SVE_VLS_WORDS]; 2649 2650 if (vq >= SVE_VQ_MIN && vq <= SVE_VQ_MAX && 2651 ((vector_lengths[(vq - KVM_ARM64_SVE_VQ_MIN) / 64] >> 2652 ((vq - KVM_ARM64_SVE_VQ_MIN) % 64)) & 1)) 2653 /* Vector length vq * 16 bytes supported */ 2654 else 2655 /* Vector length vq * 16 bytes not supported */ 2656 2657.. [2] The maximum value vq for which the above condition is true is 2658 max_vq. This is the maximum vector length available to the guest on 2659 this vcpu, and determines which register slices are visible through 2660 this ioctl interface. 2661 2662(See Documentation/arch/arm64/sve.rst for an explanation of the "vq" 2663nomenclature.) 2664 2665KVM_REG_ARM64_SVE_VLS is only accessible after KVM_ARM_VCPU_INIT. 2666KVM_ARM_VCPU_INIT initialises it to the best set of vector lengths that 2667the host supports. 2668 2669Userspace may subsequently modify it if desired until the vcpu's SVE 2670configuration is finalized using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). 2671 2672Apart from simply removing all vector lengths from the host set that 2673exceed some value, support for arbitrarily chosen sets of vector lengths 2674is hardware-dependent and may not be available. Attempting to configure 2675an invalid set of vector lengths via KVM_SET_ONE_REG will fail with 2676EINVAL. 2677 2678After the vcpu's SVE configuration is finalized, further attempts to 2679write this register will fail with EPERM. 2680 2681arm64 bitmap feature firmware pseudo-registers have the following bit pattern:: 2682 2683 0x6030 0000 0016 <regno:16> 2684 2685The bitmap feature firmware registers exposes the hypercall services that 2686are available for userspace to configure. The set bits corresponds to the 2687services that are available for the guests to access. By default, KVM 2688sets all the supported bits during VM initialization. The userspace can 2689discover the available services via KVM_GET_ONE_REG, and write back the 2690bitmap corresponding to the features that it wishes guests to see via 2691KVM_SET_ONE_REG. 2692 2693Note: These registers are immutable once any of the vCPUs of the VM has 2694run at least once. A KVM_SET_ONE_REG in such a scenario will return 2695a -EBUSY to userspace. 2696 2697(See Documentation/virt/kvm/arm/hypercalls.rst for more details.) 2698 2699 2700MIPS registers are mapped using the lower 32 bits. The upper 16 of that is 2701the register group type: 2702 2703MIPS core registers (see above) have the following id bit patterns:: 2704 2705 0x7030 0000 0000 <reg:16> 2706 2707MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit 2708patterns depending on whether they're 32-bit or 64-bit registers:: 2709 2710 0x7020 0000 0001 00 <reg:5> <sel:3> (32-bit) 2711 0x7030 0000 0001 00 <reg:5> <sel:3> (64-bit) 2712 2713Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_MIPS_CP0_ENTRYLO1 are the MIPS64 2714versions of the EntryLo registers regardless of the word size of the host 2715hardware, host kernel, guest, and whether XPA is present in the guest, i.e. 2716with the RI and XI bits (if they exist) in bits 63 and 62 respectively, and 2717the PFNX field starting at bit 30. 2718 2719MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) above) have the following id bit 2720patterns:: 2721 2722 0x7030 0000 0001 01 <reg:8> 2723 2724MIPS KVM control registers (see above) have the following id bit patterns:: 2725 2726 0x7030 0000 0002 <reg:16> 2727 2728MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following 2729id bit patterns depending on the size of the register being accessed. They are 2730always accessed according to the current guest FPU mode (Status.FR and 2731Config5.FRE), i.e. as the guest would see them, and they become unpredictable 2732if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector 2733registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they 2734overlap the FPU registers:: 2735 2736 0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers) 2737 0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers) 2738 0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers) 2739 2740MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the 2741following id bit patterns:: 2742 2743 0x7020 0000 0003 01 <0:3> <reg:5> 2744 2745MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the 2746following id bit patterns:: 2747 2748 0x7020 0000 0003 02 <0:3> <reg:5> 2749 2750RISC-V registers are mapped using the lower 32 bits. The upper 8 bits of 2751that is the register group type. 2752 2753RISC-V config registers are meant for configuring a Guest VCPU and it has 2754the following id bit patterns:: 2755 2756 0x8020 0000 01 <index into the kvm_riscv_config struct:24> (32bit Host) 2757 0x8030 0000 01 <index into the kvm_riscv_config struct:24> (64bit Host) 2758 2759Following are the RISC-V config registers: 2760 2761======================= ========= ============================================= 2762 Encoding Register Description 2763======================= ========= ============================================= 2764 0x80x0 0000 0100 0000 isa ISA feature bitmap of Guest VCPU 2765======================= ========= ============================================= 2766 2767The isa config register can be read anytime but can only be written before 2768a Guest VCPU runs. It will have ISA feature bits matching underlying host 2769set by default. 2770 2771RISC-V core registers represent the general execution state of a Guest VCPU 2772and it has the following id bit patterns:: 2773 2774 0x8020 0000 02 <index into the kvm_riscv_core struct:24> (32bit Host) 2775 0x8030 0000 02 <index into the kvm_riscv_core struct:24> (64bit Host) 2776 2777Following are the RISC-V core registers: 2778 2779======================= ========= ============================================= 2780 Encoding Register Description 2781======================= ========= ============================================= 2782 0x80x0 0000 0200 0000 regs.pc Program counter 2783 0x80x0 0000 0200 0001 regs.ra Return address 2784 0x80x0 0000 0200 0002 regs.sp Stack pointer 2785 0x80x0 0000 0200 0003 regs.gp Global pointer 2786 0x80x0 0000 0200 0004 regs.tp Task pointer 2787 0x80x0 0000 0200 0005 regs.t0 Caller saved register 0 2788 0x80x0 0000 0200 0006 regs.t1 Caller saved register 1 2789 0x80x0 0000 0200 0007 regs.t2 Caller saved register 2 2790 0x80x0 0000 0200 0008 regs.s0 Callee saved register 0 2791 0x80x0 0000 0200 0009 regs.s1 Callee saved register 1 2792 0x80x0 0000 0200 000a regs.a0 Function argument (or return value) 0 2793 0x80x0 0000 0200 000b regs.a1 Function argument (or return value) 1 2794 0x80x0 0000 0200 000c regs.a2 Function argument 2 2795 0x80x0 0000 0200 000d regs.a3 Function argument 3 2796 0x80x0 0000 0200 000e regs.a4 Function argument 4 2797 0x80x0 0000 0200 000f regs.a5 Function argument 5 2798 0x80x0 0000 0200 0010 regs.a6 Function argument 6 2799 0x80x0 0000 0200 0011 regs.a7 Function argument 7 2800 0x80x0 0000 0200 0012 regs.s2 Callee saved register 2 2801 0x80x0 0000 0200 0013 regs.s3 Callee saved register 3 2802 0x80x0 0000 0200 0014 regs.s4 Callee saved register 4 2803 0x80x0 0000 0200 0015 regs.s5 Callee saved register 5 2804 0x80x0 0000 0200 0016 regs.s6 Callee saved register 6 2805 0x80x0 0000 0200 0017 regs.s7 Callee saved register 7 2806 0x80x0 0000 0200 0018 regs.s8 Callee saved register 8 2807 0x80x0 0000 0200 0019 regs.s9 Callee saved register 9 2808 0x80x0 0000 0200 001a regs.s10 Callee saved register 10 2809 0x80x0 0000 0200 001b regs.s11 Callee saved register 11 2810 0x80x0 0000 0200 001c regs.t3 Caller saved register 3 2811 0x80x0 0000 0200 001d regs.t4 Caller saved register 4 2812 0x80x0 0000 0200 001e regs.t5 Caller saved register 5 2813 0x80x0 0000 0200 001f regs.t6 Caller saved register 6 2814 0x80x0 0000 0200 0020 mode Privilege mode (1 = S-mode or 0 = U-mode) 2815======================= ========= ============================================= 2816 2817RISC-V csr registers represent the supervisor mode control/status registers 2818of a Guest VCPU and it has the following id bit patterns:: 2819 2820 0x8020 0000 03 <index into the kvm_riscv_csr struct:24> (32bit Host) 2821 0x8030 0000 03 <index into the kvm_riscv_csr struct:24> (64bit Host) 2822 2823Following are the RISC-V csr registers: 2824 2825======================= ========= ============================================= 2826 Encoding Register Description 2827======================= ========= ============================================= 2828 0x80x0 0000 0300 0000 sstatus Supervisor status 2829 0x80x0 0000 0300 0001 sie Supervisor interrupt enable 2830 0x80x0 0000 0300 0002 stvec Supervisor trap vector base 2831 0x80x0 0000 0300 0003 sscratch Supervisor scratch register 2832 0x80x0 0000 0300 0004 sepc Supervisor exception program counter 2833 0x80x0 0000 0300 0005 scause Supervisor trap cause 2834 0x80x0 0000 0300 0006 stval Supervisor bad address or instruction 2835 0x80x0 0000 0300 0007 sip Supervisor interrupt pending 2836 0x80x0 0000 0300 0008 satp Supervisor address translation and protection 2837======================= ========= ============================================= 2838 2839RISC-V timer registers represent the timer state of a Guest VCPU and it has 2840the following id bit patterns:: 2841 2842 0x8030 0000 04 <index into the kvm_riscv_timer struct:24> 2843 2844Following are the RISC-V timer registers: 2845 2846======================= ========= ============================================= 2847 Encoding Register Description 2848======================= ========= ============================================= 2849 0x8030 0000 0400 0000 frequency Time base frequency (read-only) 2850 0x8030 0000 0400 0001 time Time value visible to Guest 2851 0x8030 0000 0400 0002 compare Time compare programmed by Guest 2852 0x8030 0000 0400 0003 state Time compare state (1 = ON or 0 = OFF) 2853======================= ========= ============================================= 2854 2855RISC-V F-extension registers represent the single precision floating point 2856state of a Guest VCPU and it has the following id bit patterns:: 2857 2858 0x8020 0000 05 <index into the __riscv_f_ext_state struct:24> 2859 2860Following are the RISC-V F-extension registers: 2861 2862======================= ========= ============================================= 2863 Encoding Register Description 2864======================= ========= ============================================= 2865 0x8020 0000 0500 0000 f[0] Floating point register 0 2866 ... 2867 0x8020 0000 0500 001f f[31] Floating point register 31 2868 0x8020 0000 0500 0020 fcsr Floating point control and status register 2869======================= ========= ============================================= 2870 2871RISC-V D-extension registers represent the double precision floating point 2872state of a Guest VCPU and it has the following id bit patterns:: 2873 2874 0x8020 0000 06 <index into the __riscv_d_ext_state struct:24> (fcsr) 2875 0x8030 0000 06 <index into the __riscv_d_ext_state struct:24> (non-fcsr) 2876 2877Following are the RISC-V D-extension registers: 2878 2879======================= ========= ============================================= 2880 Encoding Register Description 2881======================= ========= ============================================= 2882 0x8030 0000 0600 0000 f[0] Floating point register 0 2883 ... 2884 0x8030 0000 0600 001f f[31] Floating point register 31 2885 0x8020 0000 0600 0020 fcsr Floating point control and status register 2886======================= ========= ============================================= 2887 2888LoongArch registers are mapped using the lower 32 bits. The upper 16 bits of 2889that is the register group type. 2890 2891LoongArch csr registers are used to control guest cpu or get status of guest 2892cpu, and they have the following id bit patterns:: 2893 2894 0x9030 0000 0001 00 <reg:5> <sel:3> (64-bit) 2895 2896LoongArch KVM control registers are used to implement some new defined functions 2897such as set vcpu counter or reset vcpu, and they have the following id bit patterns:: 2898 2899 0x9030 0000 0002 <reg:16> 2900 2901 29024.69 KVM_GET_ONE_REG 2903-------------------- 2904 2905:Capability: KVM_CAP_ONE_REG 2906:Architectures: all 2907:Type: vcpu ioctl 2908:Parameters: struct kvm_one_reg (in and out) 2909:Returns: 0 on success, negative value on failure 2910 2911Errors include: 2912 2913 ======== ============================================================ 2914 ENOENT no such register 2915 EINVAL invalid register ID, or no such register or used with VMs in 2916 protected virtualization mode on s390 2917 EPERM (arm64) register access not allowed before vcpu finalization 2918 ======== ============================================================ 2919 2920(These error codes are indicative only: do not rely on a specific error 2921code being returned in a specific situation.) 2922 2923This ioctl allows to receive the value of a single register implemented 2924in a vcpu. The register to read is indicated by the "id" field of the 2925kvm_one_reg struct passed in. On success, the register value can be found 2926at the memory location pointed to by "addr". 2927 2928The list of registers accessible using this interface is identical to the 2929list in 4.68. 2930 2931 29324.70 KVM_KVMCLOCK_CTRL 2933---------------------- 2934 2935:Capability: KVM_CAP_KVMCLOCK_CTRL 2936:Architectures: Any that implement pvclocks (currently x86 only) 2937:Type: vcpu ioctl 2938:Parameters: None 2939:Returns: 0 on success, -1 on error 2940 2941This ioctl sets a flag accessible to the guest indicating that the specified 2942vCPU has been paused by the host userspace. 2943 2944The host will set a flag in the pvclock structure that is checked from the 2945soft lockup watchdog. The flag is part of the pvclock structure that is 2946shared between guest and host, specifically the second bit of the flags 2947field of the pvclock_vcpu_time_info structure. It will be set exclusively by 2948the host and read/cleared exclusively by the guest. The guest operation of 2949checking and clearing the flag must be an atomic operation so 2950load-link/store-conditional, or equivalent must be used. There are two cases 2951where the guest will clear the flag: when the soft lockup watchdog timer resets 2952itself or when a soft lockup is detected. This ioctl can be called any time 2953after pausing the vcpu, but before it is resumed. 2954 2955 29564.71 KVM_SIGNAL_MSI 2957------------------- 2958 2959:Capability: KVM_CAP_SIGNAL_MSI 2960:Architectures: x86 arm64 2961:Type: vm ioctl 2962:Parameters: struct kvm_msi (in) 2963:Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error 2964 2965Directly inject a MSI message. Only valid with in-kernel irqchip that handles 2966MSI messages. 2967 2968:: 2969 2970 struct kvm_msi { 2971 __u32 address_lo; 2972 __u32 address_hi; 2973 __u32 data; 2974 __u32 flags; 2975 __u32 devid; 2976 __u8 pad[12]; 2977 }; 2978 2979flags: 2980 KVM_MSI_VALID_DEVID: devid contains a valid value. The per-VM 2981 KVM_CAP_MSI_DEVID capability advertises the requirement to provide 2982 the device ID. If this capability is not available, userspace 2983 should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail. 2984 2985If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier 2986for the device that wrote the MSI message. For PCI, this is usually a 2987BDF identifier in the lower 16 bits. 2988 2989On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS 2990feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled, 2991address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of 2992address_hi must be zero. 2993 2994 29954.71 KVM_CREATE_PIT2 2996-------------------- 2997 2998:Capability: KVM_CAP_PIT2 2999:Architectures: x86 3000:Type: vm ioctl 3001:Parameters: struct kvm_pit_config (in) 3002:Returns: 0 on success, -1 on error 3003 3004Creates an in-kernel device model for the i8254 PIT. This call is only valid 3005after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following 3006parameters have to be passed:: 3007 3008 struct kvm_pit_config { 3009 __u32 flags; 3010 __u32 pad[15]; 3011 }; 3012 3013Valid flags are:: 3014 3015 #define KVM_PIT_SPEAKER_DUMMY 1 /* emulate speaker port stub */ 3016 3017PIT timer interrupts may use a per-VM kernel thread for injection. If it 3018exists, this thread will have a name of the following pattern:: 3019 3020 kvm-pit/<owner-process-pid> 3021 3022When running a guest with elevated priorities, the scheduling parameters of 3023this thread may have to be adjusted accordingly. 3024 3025This IOCTL replaces the obsolete KVM_CREATE_PIT. 3026 3027 30284.72 KVM_GET_PIT2 3029----------------- 3030 3031:Capability: KVM_CAP_PIT_STATE2 3032:Architectures: x86 3033:Type: vm ioctl 3034:Parameters: struct kvm_pit_state2 (out) 3035:Returns: 0 on success, -1 on error 3036 3037Retrieves the state of the in-kernel PIT model. Only valid after 3038KVM_CREATE_PIT2. The state is returned in the following structure:: 3039 3040 struct kvm_pit_state2 { 3041 struct kvm_pit_channel_state channels[3]; 3042 __u32 flags; 3043 __u32 reserved[9]; 3044 }; 3045 3046Valid flags are:: 3047 3048 /* disable PIT in HPET legacy mode */ 3049 #define KVM_PIT_FLAGS_HPET_LEGACY 0x00000001 3050 /* speaker port data bit enabled */ 3051 #define KVM_PIT_FLAGS_SPEAKER_DATA_ON 0x00000002 3052 3053This IOCTL replaces the obsolete KVM_GET_PIT. 3054 3055 30564.73 KVM_SET_PIT2 3057----------------- 3058 3059:Capability: KVM_CAP_PIT_STATE2 3060:Architectures: x86 3061:Type: vm ioctl 3062:Parameters: struct kvm_pit_state2 (in) 3063:Returns: 0 on success, -1 on error 3064 3065Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2. 3066See KVM_GET_PIT2 for details on struct kvm_pit_state2. 3067 3068This IOCTL replaces the obsolete KVM_SET_PIT. 3069 3070 30714.74 KVM_PPC_GET_SMMU_INFO 3072-------------------------- 3073 3074:Capability: KVM_CAP_PPC_GET_SMMU_INFO 3075:Architectures: powerpc 3076:Type: vm ioctl 3077:Parameters: None 3078:Returns: 0 on success, -1 on error 3079 3080This populates and returns a structure describing the features of 3081the "Server" class MMU emulation supported by KVM. 3082This can in turn be used by userspace to generate the appropriate 3083device-tree properties for the guest operating system. 3084 3085The structure contains some global information, followed by an 3086array of supported segment page sizes:: 3087 3088 struct kvm_ppc_smmu_info { 3089 __u64 flags; 3090 __u32 slb_size; 3091 __u32 pad; 3092 struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ]; 3093 }; 3094 3095The supported flags are: 3096 3097 - KVM_PPC_PAGE_SIZES_REAL: 3098 When that flag is set, guest page sizes must "fit" the backing 3099 store page sizes. When not set, any page size in the list can 3100 be used regardless of how they are backed by userspace. 3101 3102 - KVM_PPC_1T_SEGMENTS 3103 The emulated MMU supports 1T segments in addition to the 3104 standard 256M ones. 3105 3106 - KVM_PPC_NO_HASH 3107 This flag indicates that HPT guests are not supported by KVM, 3108 thus all guests must use radix MMU mode. 3109 3110The "slb_size" field indicates how many SLB entries are supported 3111 3112The "sps" array contains 8 entries indicating the supported base 3113page sizes for a segment in increasing order. Each entry is defined 3114as follow:: 3115 3116 struct kvm_ppc_one_seg_page_size { 3117 __u32 page_shift; /* Base page shift of segment (or 0) */ 3118 __u32 slb_enc; /* SLB encoding for BookS */ 3119 struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ]; 3120 }; 3121 3122An entry with a "page_shift" of 0 is unused. Because the array is 3123organized in increasing order, a lookup can stop when encountering 3124such an entry. 3125 3126The "slb_enc" field provides the encoding to use in the SLB for the 3127page size. The bits are in positions such as the value can directly 3128be OR'ed into the "vsid" argument of the slbmte instruction. 3129 3130The "enc" array is a list which for each of those segment base page 3131size provides the list of supported actual page sizes (which can be 3132only larger or equal to the base page size), along with the 3133corresponding encoding in the hash PTE. Similarly, the array is 31348 entries sorted by increasing sizes and an entry with a "0" shift 3135is an empty entry and a terminator:: 3136 3137 struct kvm_ppc_one_page_size { 3138 __u32 page_shift; /* Page shift (or 0) */ 3139 __u32 pte_enc; /* Encoding in the HPTE (>>12) */ 3140 }; 3141 3142The "pte_enc" field provides a value that can OR'ed into the hash 3143PTE's RPN field (ie, it needs to be shifted left by 12 to OR it 3144into the hash PTE second double word). 3145 31464.75 KVM_IRQFD 3147-------------- 3148 3149:Capability: KVM_CAP_IRQFD 3150:Architectures: x86 s390 arm64 3151:Type: vm ioctl 3152:Parameters: struct kvm_irqfd (in) 3153:Returns: 0 on success, -1 on error 3154 3155Allows setting an eventfd to directly trigger a guest interrupt. 3156kvm_irqfd.fd specifies the file descriptor to use as the eventfd and 3157kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When 3158an event is triggered on the eventfd, an interrupt is injected into 3159the guest using the specified gsi pin. The irqfd is removed using 3160the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd 3161and kvm_irqfd.gsi. 3162 3163With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify 3164mechanism allowing emulation of level-triggered, irqfd-based 3165interrupts. When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an 3166additional eventfd in the kvm_irqfd.resamplefd field. When operating 3167in resample mode, posting of an interrupt through kvm_irq.fd asserts 3168the specified gsi in the irqchip. When the irqchip is resampled, such 3169as from an EOI, the gsi is de-asserted and the user is notified via 3170kvm_irqfd.resamplefd. It is the user's responsibility to re-queue 3171the interrupt if the device making use of it still requires service. 3172Note that closing the resamplefd is not sufficient to disable the 3173irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment 3174and need not be specified with KVM_IRQFD_FLAG_DEASSIGN. 3175 3176On arm64, gsi routing being supported, the following can happen: 3177 3178- in case no routing entry is associated to this gsi, injection fails 3179- in case the gsi is associated to an irqchip routing entry, 3180 irqchip.pin + 32 corresponds to the injected SPI ID. 3181- in case the gsi is associated to an MSI routing entry, the MSI 3182 message and device ID are translated into an LPI (support restricted 3183 to GICv3 ITS in-kernel emulation). 3184 31854.76 KVM_PPC_ALLOCATE_HTAB 3186-------------------------- 3187 3188:Capability: KVM_CAP_PPC_ALLOC_HTAB 3189:Architectures: powerpc 3190:Type: vm ioctl 3191:Parameters: Pointer to u32 containing hash table order (in/out) 3192:Returns: 0 on success, -1 on error 3193 3194This requests the host kernel to allocate an MMU hash table for a 3195guest using the PAPR paravirtualization interface. This only does 3196anything if the kernel is configured to use the Book 3S HV style of 3197virtualization. Otherwise the capability doesn't exist and the ioctl 3198returns an ENOTTY error. The rest of this description assumes Book 3S 3199HV. 3200 3201There must be no vcpus running when this ioctl is called; if there 3202are, it will do nothing and return an EBUSY error. 3203 3204The parameter is a pointer to a 32-bit unsigned integer variable 3205containing the order (log base 2) of the desired size of the hash 3206table, which must be between 18 and 46. On successful return from the 3207ioctl, the value will not be changed by the kernel. 3208 3209If no hash table has been allocated when any vcpu is asked to run 3210(with the KVM_RUN ioctl), the host kernel will allocate a 3211default-sized hash table (16 MB). 3212 3213If this ioctl is called when a hash table has already been allocated, 3214with a different order from the existing hash table, the existing hash 3215table will be freed and a new one allocated. If this is ioctl is 3216called when a hash table has already been allocated of the same order 3217as specified, the kernel will clear out the existing hash table (zero 3218all HPTEs). In either case, if the guest is using the virtualized 3219real-mode area (VRMA) facility, the kernel will re-create the VMRA 3220HPTEs on the next KVM_RUN of any vcpu. 3221 32224.77 KVM_S390_INTERRUPT 3223----------------------- 3224 3225:Capability: basic 3226:Architectures: s390 3227:Type: vm ioctl, vcpu ioctl 3228:Parameters: struct kvm_s390_interrupt (in) 3229:Returns: 0 on success, -1 on error 3230 3231Allows to inject an interrupt to the guest. Interrupts can be floating 3232(vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type. 3233 3234Interrupt parameters are passed via kvm_s390_interrupt:: 3235 3236 struct kvm_s390_interrupt { 3237 __u32 type; 3238 __u32 parm; 3239 __u64 parm64; 3240 }; 3241 3242type can be one of the following: 3243 3244KVM_S390_SIGP_STOP (vcpu) 3245 - sigp stop; optional flags in parm 3246KVM_S390_PROGRAM_INT (vcpu) 3247 - program check; code in parm 3248KVM_S390_SIGP_SET_PREFIX (vcpu) 3249 - sigp set prefix; prefix address in parm 3250KVM_S390_RESTART (vcpu) 3251 - restart 3252KVM_S390_INT_CLOCK_COMP (vcpu) 3253 - clock comparator interrupt 3254KVM_S390_INT_CPU_TIMER (vcpu) 3255 - CPU timer interrupt 3256KVM_S390_INT_VIRTIO (vm) 3257 - virtio external interrupt; external interrupt 3258 parameters in parm and parm64 3259KVM_S390_INT_SERVICE (vm) 3260 - sclp external interrupt; sclp parameter in parm 3261KVM_S390_INT_EMERGENCY (vcpu) 3262 - sigp emergency; source cpu in parm 3263KVM_S390_INT_EXTERNAL_CALL (vcpu) 3264 - sigp external call; source cpu in parm 3265KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm) 3266 - compound value to indicate an 3267 I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel); 3268 I/O interruption parameters in parm (subchannel) and parm64 (intparm, 3269 interruption subclass) 3270KVM_S390_MCHK (vm, vcpu) 3271 - machine check interrupt; cr 14 bits in parm, machine check interrupt 3272 code in parm64 (note that machine checks needing further payload are not 3273 supported by this ioctl) 3274 3275This is an asynchronous vcpu ioctl and can be invoked from any thread. 3276 32774.78 KVM_PPC_GET_HTAB_FD 3278------------------------ 3279 3280:Capability: KVM_CAP_PPC_HTAB_FD 3281:Architectures: powerpc 3282:Type: vm ioctl 3283:Parameters: Pointer to struct kvm_get_htab_fd (in) 3284:Returns: file descriptor number (>= 0) on success, -1 on error 3285 3286This returns a file descriptor that can be used either to read out the 3287entries in the guest's hashed page table (HPT), or to write entries to 3288initialize the HPT. The returned fd can only be written to if the 3289KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and 3290can only be read if that bit is clear. The argument struct looks like 3291this:: 3292 3293 /* For KVM_PPC_GET_HTAB_FD */ 3294 struct kvm_get_htab_fd { 3295 __u64 flags; 3296 __u64 start_index; 3297 __u64 reserved[2]; 3298 }; 3299 3300 /* Values for kvm_get_htab_fd.flags */ 3301 #define KVM_GET_HTAB_BOLTED_ONLY ((__u64)0x1) 3302 #define KVM_GET_HTAB_WRITE ((__u64)0x2) 3303 3304The 'start_index' field gives the index in the HPT of the entry at 3305which to start reading. It is ignored when writing. 3306 3307Reads on the fd will initially supply information about all 3308"interesting" HPT entries. Interesting entries are those with the 3309bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise 3310all entries. When the end of the HPT is reached, the read() will 3311return. If read() is called again on the fd, it will start again from 3312the beginning of the HPT, but will only return HPT entries that have 3313changed since they were last read. 3314 3315Data read or written is structured as a header (8 bytes) followed by a 3316series of valid HPT entries (16 bytes) each. The header indicates how 3317many valid HPT entries there are and how many invalid entries follow 3318the valid entries. The invalid entries are not represented explicitly 3319in the stream. The header format is:: 3320 3321 struct kvm_get_htab_header { 3322 __u32 index; 3323 __u16 n_valid; 3324 __u16 n_invalid; 3325 }; 3326 3327Writes to the fd create HPT entries starting at the index given in the 3328header; first 'n_valid' valid entries with contents from the data 3329written, then 'n_invalid' invalid entries, invalidating any previously 3330valid entries found. 3331 33324.79 KVM_CREATE_DEVICE 3333---------------------- 3334 3335:Capability: KVM_CAP_DEVICE_CTRL 3336:Architectures: all 3337:Type: vm ioctl 3338:Parameters: struct kvm_create_device (in/out) 3339:Returns: 0 on success, -1 on error 3340 3341Errors: 3342 3343 ====== ======================================================= 3344 ENODEV The device type is unknown or unsupported 3345 EEXIST Device already created, and this type of device may not 3346 be instantiated multiple times 3347 ====== ======================================================= 3348 3349 Other error conditions may be defined by individual device types or 3350 have their standard meanings. 3351 3352Creates an emulated device in the kernel. The file descriptor returned 3353in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR. 3354 3355If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the 3356device type is supported (not necessarily whether it can be created 3357in the current vm). 3358 3359Individual devices should not define flags. Attributes should be used 3360for specifying any behavior that is not implied by the device type 3361number. 3362 3363:: 3364 3365 struct kvm_create_device { 3366 __u32 type; /* in: KVM_DEV_TYPE_xxx */ 3367 __u32 fd; /* out: device handle */ 3368 __u32 flags; /* in: KVM_CREATE_DEVICE_xxx */ 3369 }; 3370 33714.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR 3372-------------------------------------------- 3373 3374:Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device, 3375 KVM_CAP_VCPU_ATTRIBUTES for vcpu device 3376 KVM_CAP_SYS_ATTRIBUTES for system (/dev/kvm) device (no set) 3377:Architectures: x86, arm64, s390 3378:Type: device ioctl, vm ioctl, vcpu ioctl 3379:Parameters: struct kvm_device_attr 3380:Returns: 0 on success, -1 on error 3381 3382Errors: 3383 3384 ===== ============================================================= 3385 ENXIO The group or attribute is unknown/unsupported for this device 3386 or hardware support is missing. 3387 EPERM The attribute cannot (currently) be accessed this way 3388 (e.g. read-only attribute, or attribute that only makes 3389 sense when the device is in a different state) 3390 ===== ============================================================= 3391 3392 Other error conditions may be defined by individual device types. 3393 3394Gets/sets a specified piece of device configuration and/or state. The 3395semantics are device-specific. See individual device documentation in 3396the "devices" directory. As with ONE_REG, the size of the data 3397transferred is defined by the particular attribute. 3398 3399:: 3400 3401 struct kvm_device_attr { 3402 __u32 flags; /* no flags currently defined */ 3403 __u32 group; /* device-defined */ 3404 __u64 attr; /* group-defined */ 3405 __u64 addr; /* userspace address of attr data */ 3406 }; 3407 34084.81 KVM_HAS_DEVICE_ATTR 3409------------------------ 3410 3411:Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device, 3412 KVM_CAP_VCPU_ATTRIBUTES for vcpu device 3413 KVM_CAP_SYS_ATTRIBUTES for system (/dev/kvm) device 3414:Type: device ioctl, vm ioctl, vcpu ioctl 3415:Parameters: struct kvm_device_attr 3416:Returns: 0 on success, -1 on error 3417 3418Errors: 3419 3420 ===== ============================================================= 3421 ENXIO The group or attribute is unknown/unsupported for this device 3422 or hardware support is missing. 3423 ===== ============================================================= 3424 3425Tests whether a device supports a particular attribute. A successful 3426return indicates the attribute is implemented. It does not necessarily 3427indicate that the attribute can be read or written in the device's 3428current state. "addr" is ignored. 3429 3430.. _KVM_ARM_VCPU_INIT: 3431 34324.82 KVM_ARM_VCPU_INIT 3433---------------------- 3434 3435:Capability: basic 3436:Architectures: arm64 3437:Type: vcpu ioctl 3438:Parameters: struct kvm_vcpu_init (in) 3439:Returns: 0 on success; -1 on error 3440 3441Errors: 3442 3443 ====== ================================================================= 3444 EINVAL the target is unknown, or the combination of features is invalid. 3445 ENOENT a features bit specified is unknown. 3446 ====== ================================================================= 3447 3448This tells KVM what type of CPU to present to the guest, and what 3449optional features it should have. This will cause a reset of the cpu 3450registers to their initial values. If this is not called, KVM_RUN will 3451return ENOEXEC for that vcpu. 3452 3453The initial values are defined as: 3454 - Processor state: 3455 * AArch64: EL1h, D, A, I and F bits set. All other bits 3456 are cleared. 3457 * AArch32: SVC, A, I and F bits set. All other bits are 3458 cleared. 3459 - General Purpose registers, including PC and SP: set to 0 3460 - FPSIMD/NEON registers: set to 0 3461 - SVE registers: set to 0 3462 - System registers: Reset to their architecturally defined 3463 values as for a warm reset to EL1 (resp. SVC) 3464 3465Note that because some registers reflect machine topology, all vcpus 3466should be created before this ioctl is invoked. 3467 3468Userspace can call this function multiple times for a given vcpu, including 3469after the vcpu has been run. This will reset the vcpu to its initial 3470state. All calls to this function after the initial call must use the same 3471target and same set of feature flags, otherwise EINVAL will be returned. 3472 3473Possible features: 3474 3475 - KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state. 3476 Depends on KVM_CAP_ARM_PSCI. If not set, the CPU will be powered on 3477 and execute guest code when KVM_RUN is called. 3478 - KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode. 3479 Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only). 3480 - KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 (or a future revision 3481 backward compatible with v0.2) for the CPU. 3482 Depends on KVM_CAP_ARM_PSCI_0_2. 3483 - KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU. 3484 Depends on KVM_CAP_ARM_PMU_V3. 3485 3486 - KVM_ARM_VCPU_PTRAUTH_ADDRESS: Enables Address Pointer authentication 3487 for arm64 only. 3488 Depends on KVM_CAP_ARM_PTRAUTH_ADDRESS. 3489 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are 3490 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and 3491 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be 3492 requested. 3493 3494 - KVM_ARM_VCPU_PTRAUTH_GENERIC: Enables Generic Pointer authentication 3495 for arm64 only. 3496 Depends on KVM_CAP_ARM_PTRAUTH_GENERIC. 3497 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are 3498 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and 3499 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be 3500 requested. 3501 3502 - KVM_ARM_VCPU_SVE: Enables SVE for the CPU (arm64 only). 3503 Depends on KVM_CAP_ARM_SVE. 3504 Requires KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3505 3506 * After KVM_ARM_VCPU_INIT: 3507 3508 - KVM_REG_ARM64_SVE_VLS may be read using KVM_GET_ONE_REG: the 3509 initial value of this pseudo-register indicates the best set of 3510 vector lengths possible for a vcpu on this host. 3511 3512 * Before KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3513 3514 - KVM_RUN and KVM_GET_REG_LIST are not available; 3515 3516 - KVM_GET_ONE_REG and KVM_SET_ONE_REG cannot be used to access 3517 the scalable architectural SVE registers 3518 KVM_REG_ARM64_SVE_ZREG(), KVM_REG_ARM64_SVE_PREG() or 3519 KVM_REG_ARM64_SVE_FFR; 3520 3521 - KVM_REG_ARM64_SVE_VLS may optionally be written using 3522 KVM_SET_ONE_REG, to modify the set of vector lengths available 3523 for the vcpu. 3524 3525 * After KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3526 3527 - the KVM_REG_ARM64_SVE_VLS pseudo-register is immutable, and can 3528 no longer be written using KVM_SET_ONE_REG. 3529 35304.83 KVM_ARM_PREFERRED_TARGET 3531----------------------------- 3532 3533:Capability: basic 3534:Architectures: arm64 3535:Type: vm ioctl 3536:Parameters: struct kvm_vcpu_init (out) 3537:Returns: 0 on success; -1 on error 3538 3539Errors: 3540 3541 ====== ========================================== 3542 ENODEV no preferred target available for the host 3543 ====== ========================================== 3544 3545This queries KVM for preferred CPU target type which can be emulated 3546by KVM on underlying host. 3547 3548The ioctl returns struct kvm_vcpu_init instance containing information 3549about preferred CPU target type and recommended features for it. The 3550kvm_vcpu_init->features bitmap returned will have feature bits set if 3551the preferred target recommends setting these features, but this is 3552not mandatory. 3553 3554The information returned by this ioctl can be used to prepare an instance 3555of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in 3556VCPU matching underlying host. 3557 3558 35594.84 KVM_GET_REG_LIST 3560--------------------- 3561 3562:Capability: basic 3563:Architectures: arm64, mips, riscv 3564:Type: vcpu ioctl 3565:Parameters: struct kvm_reg_list (in/out) 3566:Returns: 0 on success; -1 on error 3567 3568Errors: 3569 3570 ===== ============================================================== 3571 E2BIG the reg index list is too big to fit in the array specified by 3572 the user (the number required will be written into n). 3573 ===== ============================================================== 3574 3575:: 3576 3577 struct kvm_reg_list { 3578 __u64 n; /* number of registers in reg[] */ 3579 __u64 reg[0]; 3580 }; 3581 3582This ioctl returns the guest registers that are supported for the 3583KVM_GET_ONE_REG/KVM_SET_ONE_REG calls. 3584 3585Note that s390 does not support KVM_GET_REG_LIST for historical reasons 3586(read: nobody cared). The set of registers in kernels 4.x and newer is: 3587 3588- KVM_REG_S390_TODPR 3589 3590- KVM_REG_S390_EPOCHDIFF 3591 3592- KVM_REG_S390_CPU_TIMER 3593 3594- KVM_REG_S390_CLOCK_COMP 3595 3596- KVM_REG_S390_PFTOKEN 3597 3598- KVM_REG_S390_PFCOMPARE 3599 3600- KVM_REG_S390_PFSELECT 3601 3602- KVM_REG_S390_PP 3603 3604- KVM_REG_S390_GBEA 3605 3606 36074.85 KVM_ARM_SET_DEVICE_ADDR (deprecated) 3608----------------------------------------- 3609 3610:Capability: KVM_CAP_ARM_SET_DEVICE_ADDR 3611:Architectures: arm64 3612:Type: vm ioctl 3613:Parameters: struct kvm_arm_device_address (in) 3614:Returns: 0 on success, -1 on error 3615 3616Errors: 3617 3618 ====== ============================================ 3619 ENODEV The device id is unknown 3620 ENXIO Device not supported on current system 3621 EEXIST Address already set 3622 E2BIG Address outside guest physical address space 3623 EBUSY Address overlaps with other device range 3624 ====== ============================================ 3625 3626:: 3627 3628 struct kvm_arm_device_addr { 3629 __u64 id; 3630 __u64 addr; 3631 }; 3632 3633Specify a device address in the guest's physical address space where guests 3634can access emulated or directly exposed devices, which the host kernel needs 3635to know about. The id field is an architecture specific identifier for a 3636specific device. 3637 3638arm64 divides the id field into two parts, a device id and an 3639address type id specific to the individual device:: 3640 3641 bits: | 63 ... 32 | 31 ... 16 | 15 ... 0 | 3642 field: | 0x00000000 | device id | addr type id | 3643 3644arm64 currently only require this when using the in-kernel GIC 3645support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2 3646as the device id. When setting the base address for the guest's 3647mapping of the VGIC virtual CPU and distributor interface, the ioctl 3648must be called after calling KVM_CREATE_IRQCHIP, but before calling 3649KVM_RUN on any of the VCPUs. Calling this ioctl twice for any of the 3650base addresses will return -EEXIST. 3651 3652Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API 3653should be used instead. 3654 3655 36564.86 KVM_PPC_RTAS_DEFINE_TOKEN 3657------------------------------ 3658 3659:Capability: KVM_CAP_PPC_RTAS 3660:Architectures: ppc 3661:Type: vm ioctl 3662:Parameters: struct kvm_rtas_token_args 3663:Returns: 0 on success, -1 on error 3664 3665Defines a token value for a RTAS (Run Time Abstraction Services) 3666service in order to allow it to be handled in the kernel. The 3667argument struct gives the name of the service, which must be the name 3668of a service that has a kernel-side implementation. If the token 3669value is non-zero, it will be associated with that service, and 3670subsequent RTAS calls by the guest specifying that token will be 3671handled by the kernel. If the token value is 0, then any token 3672associated with the service will be forgotten, and subsequent RTAS 3673calls by the guest for that service will be passed to userspace to be 3674handled. 3675 36764.87 KVM_SET_GUEST_DEBUG 3677------------------------ 3678 3679:Capability: KVM_CAP_SET_GUEST_DEBUG 3680:Architectures: x86, s390, ppc, arm64 3681:Type: vcpu ioctl 3682:Parameters: struct kvm_guest_debug (in) 3683:Returns: 0 on success; -1 on error 3684 3685:: 3686 3687 struct kvm_guest_debug { 3688 __u32 control; 3689 __u32 pad; 3690 struct kvm_guest_debug_arch arch; 3691 }; 3692 3693Set up the processor specific debug registers and configure vcpu for 3694handling guest debug events. There are two parts to the structure, the 3695first a control bitfield indicates the type of debug events to handle 3696when running. Common control bits are: 3697 3698 - KVM_GUESTDBG_ENABLE: guest debugging is enabled 3699 - KVM_GUESTDBG_SINGLESTEP: the next run should single-step 3700 3701The top 16 bits of the control field are architecture specific control 3702flags which can include the following: 3703 3704 - KVM_GUESTDBG_USE_SW_BP: using software breakpoints [x86, arm64] 3705 - KVM_GUESTDBG_USE_HW_BP: using hardware breakpoints [x86, s390] 3706 - KVM_GUESTDBG_USE_HW: using hardware debug events [arm64] 3707 - KVM_GUESTDBG_INJECT_DB: inject DB type exception [x86] 3708 - KVM_GUESTDBG_INJECT_BP: inject BP type exception [x86] 3709 - KVM_GUESTDBG_EXIT_PENDING: trigger an immediate guest exit [s390] 3710 - KVM_GUESTDBG_BLOCKIRQ: avoid injecting interrupts/NMI/SMI [x86] 3711 3712For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints 3713are enabled in memory so we need to ensure breakpoint exceptions are 3714correctly trapped and the KVM run loop exits at the breakpoint and not 3715running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP 3716we need to ensure the guest vCPUs architecture specific registers are 3717updated to the correct (supplied) values. 3718 3719The second part of the structure is architecture specific and 3720typically contains a set of debug registers. 3721 3722For arm64 the number of debug registers is implementation defined and 3723can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and 3724KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number 3725indicating the number of supported registers. 3726 3727For ppc, the KVM_CAP_PPC_GUEST_DEBUG_SSTEP capability indicates whether 3728the single-step debug event (KVM_GUESTDBG_SINGLESTEP) is supported. 3729 3730Also when supported, KVM_CAP_SET_GUEST_DEBUG2 capability indicates the 3731supported KVM_GUESTDBG_* bits in the control field. 3732 3733When debug events exit the main run loop with the reason 3734KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run 3735structure containing architecture specific debug information. 3736 37374.88 KVM_GET_EMULATED_CPUID 3738--------------------------- 3739 3740:Capability: KVM_CAP_EXT_EMUL_CPUID 3741:Architectures: x86 3742:Type: system ioctl 3743:Parameters: struct kvm_cpuid2 (in/out) 3744:Returns: 0 on success, -1 on error 3745 3746:: 3747 3748 struct kvm_cpuid2 { 3749 __u32 nent; 3750 __u32 flags; 3751 struct kvm_cpuid_entry2 entries[0]; 3752 }; 3753 3754The member 'flags' is used for passing flags from userspace. 3755 3756:: 3757 3758 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0) 3759 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */ 3760 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */ 3761 3762 struct kvm_cpuid_entry2 { 3763 __u32 function; 3764 __u32 index; 3765 __u32 flags; 3766 __u32 eax; 3767 __u32 ebx; 3768 __u32 ecx; 3769 __u32 edx; 3770 __u32 padding[3]; 3771 }; 3772 3773This ioctl returns x86 cpuid features which are emulated by 3774kvm.Userspace can use the information returned by this ioctl to query 3775which features are emulated by kvm instead of being present natively. 3776 3777Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2 3778structure with the 'nent' field indicating the number of entries in 3779the variable-size array 'entries'. If the number of entries is too low 3780to describe the cpu capabilities, an error (E2BIG) is returned. If the 3781number is too high, the 'nent' field is adjusted and an error (ENOMEM) 3782is returned. If the number is just right, the 'nent' field is adjusted 3783to the number of valid entries in the 'entries' array, which is then 3784filled. 3785 3786The entries returned are the set CPUID bits of the respective features 3787which kvm emulates, as returned by the CPUID instruction, with unknown 3788or unsupported feature bits cleared. 3789 3790Features like x2apic, for example, may not be present in the host cpu 3791but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be 3792emulated efficiently and thus not included here. 3793 3794The fields in each entry are defined as follows: 3795 3796 function: 3797 the eax value used to obtain the entry 3798 index: 3799 the ecx value used to obtain the entry (for entries that are 3800 affected by ecx) 3801 flags: 3802 an OR of zero or more of the following: 3803 3804 KVM_CPUID_FLAG_SIGNIFCANT_INDEX: 3805 if the index field is valid 3806 3807 eax, ebx, ecx, edx: 3808 3809 the values returned by the cpuid instruction for 3810 this function/index combination 3811 38124.89 KVM_S390_MEM_OP 3813-------------------- 3814 3815:Capability: KVM_CAP_S390_MEM_OP, KVM_CAP_S390_PROTECTED, KVM_CAP_S390_MEM_OP_EXTENSION 3816:Architectures: s390 3817:Type: vm ioctl, vcpu ioctl 3818:Parameters: struct kvm_s390_mem_op (in) 3819:Returns: = 0 on success, 3820 < 0 on generic error (e.g. -EFAULT or -ENOMEM), 3821 16 bit program exception code if the access causes such an exception 3822 3823Read or write data from/to the VM's memory. 3824The KVM_CAP_S390_MEM_OP_EXTENSION capability specifies what functionality is 3825supported. 3826 3827Parameters are specified via the following structure:: 3828 3829 struct kvm_s390_mem_op { 3830 __u64 gaddr; /* the guest address */ 3831 __u64 flags; /* flags */ 3832 __u32 size; /* amount of bytes */ 3833 __u32 op; /* type of operation */ 3834 __u64 buf; /* buffer in userspace */ 3835 union { 3836 struct { 3837 __u8 ar; /* the access register number */ 3838 __u8 key; /* access key, ignored if flag unset */ 3839 __u8 pad1[6]; /* ignored */ 3840 __u64 old_addr; /* ignored if flag unset */ 3841 }; 3842 __u32 sida_offset; /* offset into the sida */ 3843 __u8 reserved[32]; /* ignored */ 3844 }; 3845 }; 3846 3847The start address of the memory region has to be specified in the "gaddr" 3848field, and the length of the region in the "size" field (which must not 3849be 0). The maximum value for "size" can be obtained by checking the 3850KVM_CAP_S390_MEM_OP capability. "buf" is the buffer supplied by the 3851userspace application where the read data should be written to for 3852a read access, or where the data that should be written is stored for 3853a write access. The "reserved" field is meant for future extensions. 3854Reserved and unused values are ignored. Future extension that add members must 3855introduce new flags. 3856 3857The type of operation is specified in the "op" field. Flags modifying 3858their behavior can be set in the "flags" field. Undefined flag bits must 3859be set to 0. 3860 3861Possible operations are: 3862 * ``KVM_S390_MEMOP_LOGICAL_READ`` 3863 * ``KVM_S390_MEMOP_LOGICAL_WRITE`` 3864 * ``KVM_S390_MEMOP_ABSOLUTE_READ`` 3865 * ``KVM_S390_MEMOP_ABSOLUTE_WRITE`` 3866 * ``KVM_S390_MEMOP_SIDA_READ`` 3867 * ``KVM_S390_MEMOP_SIDA_WRITE`` 3868 * ``KVM_S390_MEMOP_ABSOLUTE_CMPXCHG`` 3869 3870Logical read/write: 3871^^^^^^^^^^^^^^^^^^^ 3872 3873Access logical memory, i.e. translate the given guest address to an absolute 3874address given the state of the VCPU and use the absolute address as target of 3875the access. "ar" designates the access register number to be used; the valid 3876range is 0..15. 3877Logical accesses are permitted for the VCPU ioctl only. 3878Logical accesses are permitted for non-protected guests only. 3879 3880Supported flags: 3881 * ``KVM_S390_MEMOP_F_CHECK_ONLY`` 3882 * ``KVM_S390_MEMOP_F_INJECT_EXCEPTION`` 3883 * ``KVM_S390_MEMOP_F_SKEY_PROTECTION`` 3884 3885The KVM_S390_MEMOP_F_CHECK_ONLY flag can be set to check whether the 3886corresponding memory access would cause an access exception; however, 3887no actual access to the data in memory at the destination is performed. 3888In this case, "buf" is unused and can be NULL. 3889 3890In case an access exception occurred during the access (or would occur 3891in case of KVM_S390_MEMOP_F_CHECK_ONLY), the ioctl returns a positive 3892error number indicating the type of exception. This exception is also 3893raised directly at the corresponding VCPU if the flag 3894KVM_S390_MEMOP_F_INJECT_EXCEPTION is set. 3895On protection exceptions, unless specified otherwise, the injected 3896translation-exception identifier (TEID) indicates suppression. 3897 3898If the KVM_S390_MEMOP_F_SKEY_PROTECTION flag is set, storage key 3899protection is also in effect and may cause exceptions if accesses are 3900prohibited given the access key designated by "key"; the valid range is 0..15. 3901KVM_S390_MEMOP_F_SKEY_PROTECTION is available if KVM_CAP_S390_MEM_OP_EXTENSION 3902is > 0. 3903Since the accessed memory may span multiple pages and those pages might have 3904different storage keys, it is possible that a protection exception occurs 3905after memory has been modified. In this case, if the exception is injected, 3906the TEID does not indicate suppression. 3907 3908Absolute read/write: 3909^^^^^^^^^^^^^^^^^^^^ 3910 3911Access absolute memory. This operation is intended to be used with the 3912KVM_S390_MEMOP_F_SKEY_PROTECTION flag, to allow accessing memory and performing 3913the checks required for storage key protection as one operation (as opposed to 3914user space getting the storage keys, performing the checks, and accessing 3915memory thereafter, which could lead to a delay between check and access). 3916Absolute accesses are permitted for the VM ioctl if KVM_CAP_S390_MEM_OP_EXTENSION 3917has the KVM_S390_MEMOP_EXTENSION_CAP_BASE bit set. 3918Currently absolute accesses are not permitted for VCPU ioctls. 3919Absolute accesses are permitted for non-protected guests only. 3920 3921Supported flags: 3922 * ``KVM_S390_MEMOP_F_CHECK_ONLY`` 3923 * ``KVM_S390_MEMOP_F_SKEY_PROTECTION`` 3924 3925The semantics of the flags common with logical accesses are as for logical 3926accesses. 3927 3928Absolute cmpxchg: 3929^^^^^^^^^^^^^^^^^ 3930 3931Perform cmpxchg on absolute guest memory. Intended for use with the 3932KVM_S390_MEMOP_F_SKEY_PROTECTION flag. 3933Instead of doing an unconditional write, the access occurs only if the target 3934location contains the value pointed to by "old_addr". 3935This is performed as an atomic cmpxchg with the length specified by the "size" 3936parameter. "size" must be a power of two up to and including 16. 3937If the exchange did not take place because the target value doesn't match the 3938old value, the value "old_addr" points to is replaced by the target value. 3939User space can tell if an exchange took place by checking if this replacement 3940occurred. The cmpxchg op is permitted for the VM ioctl if 3941KVM_CAP_S390_MEM_OP_EXTENSION has flag KVM_S390_MEMOP_EXTENSION_CAP_CMPXCHG set. 3942 3943Supported flags: 3944 * ``KVM_S390_MEMOP_F_SKEY_PROTECTION`` 3945 3946SIDA read/write: 3947^^^^^^^^^^^^^^^^ 3948 3949Access the secure instruction data area which contains memory operands necessary 3950for instruction emulation for protected guests. 3951SIDA accesses are available if the KVM_CAP_S390_PROTECTED capability is available. 3952SIDA accesses are permitted for the VCPU ioctl only. 3953SIDA accesses are permitted for protected guests only. 3954 3955No flags are supported. 3956 39574.90 KVM_S390_GET_SKEYS 3958----------------------- 3959 3960:Capability: KVM_CAP_S390_SKEYS 3961:Architectures: s390 3962:Type: vm ioctl 3963:Parameters: struct kvm_s390_skeys 3964:Returns: 0 on success, KVM_S390_GET_SKEYS_NONE if guest is not using storage 3965 keys, negative value on error 3966 3967This ioctl is used to get guest storage key values on the s390 3968architecture. The ioctl takes parameters via the kvm_s390_skeys struct:: 3969 3970 struct kvm_s390_skeys { 3971 __u64 start_gfn; 3972 __u64 count; 3973 __u64 skeydata_addr; 3974 __u32 flags; 3975 __u32 reserved[9]; 3976 }; 3977 3978The start_gfn field is the number of the first guest frame whose storage keys 3979you want to get. 3980 3981The count field is the number of consecutive frames (starting from start_gfn) 3982whose storage keys to get. The count field must be at least 1 and the maximum 3983allowed value is defined as KVM_S390_SKEYS_MAX. Values outside this range 3984will cause the ioctl to return -EINVAL. 3985 3986The skeydata_addr field is the address to a buffer large enough to hold count 3987bytes. This buffer will be filled with storage key data by the ioctl. 3988 39894.91 KVM_S390_SET_SKEYS 3990----------------------- 3991 3992:Capability: KVM_CAP_S390_SKEYS 3993:Architectures: s390 3994:Type: vm ioctl 3995:Parameters: struct kvm_s390_skeys 3996:Returns: 0 on success, negative value on error 3997 3998This ioctl is used to set guest storage key values on the s390 3999architecture. The ioctl takes parameters via the kvm_s390_skeys struct. 4000See section on KVM_S390_GET_SKEYS for struct definition. 4001 4002The start_gfn field is the number of the first guest frame whose storage keys 4003you want to set. 4004 4005The count field is the number of consecutive frames (starting from start_gfn) 4006whose storage keys to get. The count field must be at least 1 and the maximum 4007allowed value is defined as KVM_S390_SKEYS_MAX. Values outside this range 4008will cause the ioctl to return -EINVAL. 4009 4010The skeydata_addr field is the address to a buffer containing count bytes of 4011storage keys. Each byte in the buffer will be set as the storage key for a 4012single frame starting at start_gfn for count frames. 4013 4014Note: If any architecturally invalid key value is found in the given data then 4015the ioctl will return -EINVAL. 4016 40174.92 KVM_S390_IRQ 4018----------------- 4019 4020:Capability: KVM_CAP_S390_INJECT_IRQ 4021:Architectures: s390 4022:Type: vcpu ioctl 4023:Parameters: struct kvm_s390_irq (in) 4024:Returns: 0 on success, -1 on error 4025 4026Errors: 4027 4028 4029 ====== ================================================================= 4030 EINVAL interrupt type is invalid 4031 type is KVM_S390_SIGP_STOP and flag parameter is invalid value, 4032 type is KVM_S390_INT_EXTERNAL_CALL and code is bigger 4033 than the maximum of VCPUs 4034 EBUSY type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped, 4035 type is KVM_S390_SIGP_STOP and a stop irq is already pending, 4036 type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt 4037 is already pending 4038 ====== ================================================================= 4039 4040Allows to inject an interrupt to the guest. 4041 4042Using struct kvm_s390_irq as a parameter allows 4043to inject additional payload which is not 4044possible via KVM_S390_INTERRUPT. 4045 4046Interrupt parameters are passed via kvm_s390_irq:: 4047 4048 struct kvm_s390_irq { 4049 __u64 type; 4050 union { 4051 struct kvm_s390_io_info io; 4052 struct kvm_s390_ext_info ext; 4053 struct kvm_s390_pgm_info pgm; 4054 struct kvm_s390_emerg_info emerg; 4055 struct kvm_s390_extcall_info extcall; 4056 struct kvm_s390_prefix_info prefix; 4057 struct kvm_s390_stop_info stop; 4058 struct kvm_s390_mchk_info mchk; 4059 char reserved[64]; 4060 } u; 4061 }; 4062 4063type can be one of the following: 4064 4065- KVM_S390_SIGP_STOP - sigp stop; parameter in .stop 4066- KVM_S390_PROGRAM_INT - program check; parameters in .pgm 4067- KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix 4068- KVM_S390_RESTART - restart; no parameters 4069- KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters 4070- KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters 4071- KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg 4072- KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall 4073- KVM_S390_MCHK - machine check interrupt; parameters in .mchk 4074 4075This is an asynchronous vcpu ioctl and can be invoked from any thread. 4076 40774.94 KVM_S390_GET_IRQ_STATE 4078--------------------------- 4079 4080:Capability: KVM_CAP_S390_IRQ_STATE 4081:Architectures: s390 4082:Type: vcpu ioctl 4083:Parameters: struct kvm_s390_irq_state (out) 4084:Returns: >= number of bytes copied into buffer, 4085 -EINVAL if buffer size is 0, 4086 -ENOBUFS if buffer size is too small to fit all pending interrupts, 4087 -EFAULT if the buffer address was invalid 4088 4089This ioctl allows userspace to retrieve the complete state of all currently 4090pending interrupts in a single buffer. Use cases include migration 4091and introspection. The parameter structure contains the address of a 4092userspace buffer and its length:: 4093 4094 struct kvm_s390_irq_state { 4095 __u64 buf; 4096 __u32 flags; /* will stay unused for compatibility reasons */ 4097 __u32 len; 4098 __u32 reserved[4]; /* will stay unused for compatibility reasons */ 4099 }; 4100 4101Userspace passes in the above struct and for each pending interrupt a 4102struct kvm_s390_irq is copied to the provided buffer. 4103 4104The structure contains a flags and a reserved field for future extensions. As 4105the kernel never checked for flags == 0 and QEMU never pre-zeroed flags and 4106reserved, these fields can not be used in the future without breaking 4107compatibility. 4108 4109If -ENOBUFS is returned the buffer provided was too small and userspace 4110may retry with a bigger buffer. 4111 41124.95 KVM_S390_SET_IRQ_STATE 4113--------------------------- 4114 4115:Capability: KVM_CAP_S390_IRQ_STATE 4116:Architectures: s390 4117:Type: vcpu ioctl 4118:Parameters: struct kvm_s390_irq_state (in) 4119:Returns: 0 on success, 4120 -EFAULT if the buffer address was invalid, 4121 -EINVAL for an invalid buffer length (see below), 4122 -EBUSY if there were already interrupts pending, 4123 errors occurring when actually injecting the 4124 interrupt. See KVM_S390_IRQ. 4125 4126This ioctl allows userspace to set the complete state of all cpu-local 4127interrupts currently pending for the vcpu. It is intended for restoring 4128interrupt state after a migration. The input parameter is a userspace buffer 4129containing a struct kvm_s390_irq_state:: 4130 4131 struct kvm_s390_irq_state { 4132 __u64 buf; 4133 __u32 flags; /* will stay unused for compatibility reasons */ 4134 __u32 len; 4135 __u32 reserved[4]; /* will stay unused for compatibility reasons */ 4136 }; 4137 4138The restrictions for flags and reserved apply as well. 4139(see KVM_S390_GET_IRQ_STATE) 4140 4141The userspace memory referenced by buf contains a struct kvm_s390_irq 4142for each interrupt to be injected into the guest. 4143If one of the interrupts could not be injected for some reason the 4144ioctl aborts. 4145 4146len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0 4147and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq), 4148which is the maximum number of possibly pending cpu-local interrupts. 4149 41504.96 KVM_SMI 4151------------ 4152 4153:Capability: KVM_CAP_X86_SMM 4154:Architectures: x86 4155:Type: vcpu ioctl 4156:Parameters: none 4157:Returns: 0 on success, -1 on error 4158 4159Queues an SMI on the thread's vcpu. 4160 41614.97 KVM_X86_SET_MSR_FILTER 4162---------------------------- 4163 4164:Capability: KVM_CAP_X86_MSR_FILTER 4165:Architectures: x86 4166:Type: vm ioctl 4167:Parameters: struct kvm_msr_filter 4168:Returns: 0 on success, < 0 on error 4169 4170:: 4171 4172 struct kvm_msr_filter_range { 4173 #define KVM_MSR_FILTER_READ (1 << 0) 4174 #define KVM_MSR_FILTER_WRITE (1 << 1) 4175 __u32 flags; 4176 __u32 nmsrs; /* number of msrs in bitmap */ 4177 __u32 base; /* MSR index the bitmap starts at */ 4178 __u8 *bitmap; /* a 1 bit allows the operations in flags, 0 denies */ 4179 }; 4180 4181 #define KVM_MSR_FILTER_MAX_RANGES 16 4182 struct kvm_msr_filter { 4183 #define KVM_MSR_FILTER_DEFAULT_ALLOW (0 << 0) 4184 #define KVM_MSR_FILTER_DEFAULT_DENY (1 << 0) 4185 __u32 flags; 4186 struct kvm_msr_filter_range ranges[KVM_MSR_FILTER_MAX_RANGES]; 4187 }; 4188 4189flags values for ``struct kvm_msr_filter_range``: 4190 4191``KVM_MSR_FILTER_READ`` 4192 4193 Filter read accesses to MSRs using the given bitmap. A 0 in the bitmap 4194 indicates that read accesses should be denied, while a 1 indicates that 4195 a read for a particular MSR should be allowed regardless of the default 4196 filter action. 4197 4198``KVM_MSR_FILTER_WRITE`` 4199 4200 Filter write accesses to MSRs using the given bitmap. A 0 in the bitmap 4201 indicates that write accesses should be denied, while a 1 indicates that 4202 a write for a particular MSR should be allowed regardless of the default 4203 filter action. 4204 4205flags values for ``struct kvm_msr_filter``: 4206 4207``KVM_MSR_FILTER_DEFAULT_ALLOW`` 4208 4209 If no filter range matches an MSR index that is getting accessed, KVM will 4210 allow accesses to all MSRs by default. 4211 4212``KVM_MSR_FILTER_DEFAULT_DENY`` 4213 4214 If no filter range matches an MSR index that is getting accessed, KVM will 4215 deny accesses to all MSRs by default. 4216 4217This ioctl allows userspace to define up to 16 bitmaps of MSR ranges to deny 4218guest MSR accesses that would normally be allowed by KVM. If an MSR is not 4219covered by a specific range, the "default" filtering behavior applies. Each 4220bitmap range covers MSRs from [base .. base+nmsrs). 4221 4222If an MSR access is denied by userspace, the resulting KVM behavior depends on 4223whether or not KVM_CAP_X86_USER_SPACE_MSR's KVM_MSR_EXIT_REASON_FILTER is 4224enabled. If KVM_MSR_EXIT_REASON_FILTER is enabled, KVM will exit to userspace 4225on denied accesses, i.e. userspace effectively intercepts the MSR access. If 4226KVM_MSR_EXIT_REASON_FILTER is not enabled, KVM will inject a #GP into the guest 4227on denied accesses. Note, if an MSR access is denied during emulation of MSR 4228load/stores during VMX transitions, KVM ignores KVM_MSR_EXIT_REASON_FILTER. 4229See the below warning for full details. 4230 4231If an MSR access is allowed by userspace, KVM will emulate and/or virtualize 4232the access in accordance with the vCPU model. Note, KVM may still ultimately 4233inject a #GP if an access is allowed by userspace, e.g. if KVM doesn't support 4234the MSR, or to follow architectural behavior for the MSR. 4235 4236By default, KVM operates in KVM_MSR_FILTER_DEFAULT_ALLOW mode with no MSR range 4237filters. 4238 4239Calling this ioctl with an empty set of ranges (all nmsrs == 0) disables MSR 4240filtering. In that mode, ``KVM_MSR_FILTER_DEFAULT_DENY`` is invalid and causes 4241an error. 4242 4243.. warning:: 4244 MSR accesses that are side effects of instruction execution (emulated or 4245 native) are not filtered as hardware does not honor MSR bitmaps outside of 4246 RDMSR and WRMSR, and KVM mimics that behavior when emulating instructions 4247 to avoid pointless divergence from hardware. E.g. RDPID reads MSR_TSC_AUX, 4248 SYSENTER reads the SYSENTER MSRs, etc. 4249 4250 MSRs that are loaded/stored via dedicated VMCS fields are not filtered as 4251 part of VM-Enter/VM-Exit emulation. 4252 4253 MSRs that are loaded/store via VMX's load/store lists _are_ filtered as part 4254 of VM-Enter/VM-Exit emulation. If an MSR access is denied on VM-Enter, KVM 4255 synthesizes a consistency check VM-Exit(EXIT_REASON_MSR_LOAD_FAIL). If an 4256 MSR access is denied on VM-Exit, KVM synthesizes a VM-Abort. In short, KVM 4257 extends Intel's architectural list of MSRs that cannot be loaded/saved via 4258 the VM-Enter/VM-Exit MSR list. It is platform owner's responsibility to 4259 to communicate any such restrictions to their end users. 4260 4261 x2APIC MSR accesses cannot be filtered (KVM silently ignores filters that 4262 cover any x2APIC MSRs). 4263 4264Note, invoking this ioctl while a vCPU is running is inherently racy. However, 4265KVM does guarantee that vCPUs will see either the previous filter or the new 4266filter, e.g. MSRs with identical settings in both the old and new filter will 4267have deterministic behavior. 4268 4269Similarly, if userspace wishes to intercept on denied accesses, 4270KVM_MSR_EXIT_REASON_FILTER must be enabled before activating any filters, and 4271left enabled until after all filters are deactivated. Failure to do so may 4272result in KVM injecting a #GP instead of exiting to userspace. 4273 42744.98 KVM_CREATE_SPAPR_TCE_64 4275---------------------------- 4276 4277:Capability: KVM_CAP_SPAPR_TCE_64 4278:Architectures: powerpc 4279:Type: vm ioctl 4280:Parameters: struct kvm_create_spapr_tce_64 (in) 4281:Returns: file descriptor for manipulating the created TCE table 4282 4283This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit 4284windows, described in 4.62 KVM_CREATE_SPAPR_TCE 4285 4286This capability uses extended struct in ioctl interface:: 4287 4288 /* for KVM_CAP_SPAPR_TCE_64 */ 4289 struct kvm_create_spapr_tce_64 { 4290 __u64 liobn; 4291 __u32 page_shift; 4292 __u32 flags; 4293 __u64 offset; /* in pages */ 4294 __u64 size; /* in pages */ 4295 }; 4296 4297The aim of extension is to support an additional bigger DMA window with 4298a variable page size. 4299KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and 4300a bus offset of the corresponding DMA window, @size and @offset are numbers 4301of IOMMU pages. 4302 4303@flags are not used at the moment. 4304 4305The rest of functionality is identical to KVM_CREATE_SPAPR_TCE. 4306 43074.99 KVM_REINJECT_CONTROL 4308------------------------- 4309 4310:Capability: KVM_CAP_REINJECT_CONTROL 4311:Architectures: x86 4312:Type: vm ioctl 4313:Parameters: struct kvm_reinject_control (in) 4314:Returns: 0 on success, 4315 -EFAULT if struct kvm_reinject_control cannot be read, 4316 -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier. 4317 4318i8254 (PIT) has two modes, reinject and !reinject. The default is reinject, 4319where KVM queues elapsed i8254 ticks and monitors completion of interrupt from 4320vector(s) that i8254 injects. Reinject mode dequeues a tick and injects its 4321interrupt whenever there isn't a pending interrupt from i8254. 4322!reinject mode injects an interrupt as soon as a tick arrives. 4323 4324:: 4325 4326 struct kvm_reinject_control { 4327 __u8 pit_reinject; 4328 __u8 reserved[31]; 4329 }; 4330 4331pit_reinject = 0 (!reinject mode) is recommended, unless running an old 4332operating system that uses the PIT for timing (e.g. Linux 2.4.x). 4333 43344.100 KVM_PPC_CONFIGURE_V3_MMU 4335------------------------------ 4336 4337:Capability: KVM_CAP_PPC_MMU_RADIX or KVM_CAP_PPC_MMU_HASH_V3 4338:Architectures: ppc 4339:Type: vm ioctl 4340:Parameters: struct kvm_ppc_mmuv3_cfg (in) 4341:Returns: 0 on success, 4342 -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read, 4343 -EINVAL if the configuration is invalid 4344 4345This ioctl controls whether the guest will use radix or HPT (hashed 4346page table) translation, and sets the pointer to the process table for 4347the guest. 4348 4349:: 4350 4351 struct kvm_ppc_mmuv3_cfg { 4352 __u64 flags; 4353 __u64 process_table; 4354 }; 4355 4356There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and 4357KVM_PPC_MMUV3_GTSE. KVM_PPC_MMUV3_RADIX, if set, configures the guest 4358to use radix tree translation, and if clear, to use HPT translation. 4359KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest 4360to be able to use the global TLB and SLB invalidation instructions; 4361if clear, the guest may not use these instructions. 4362 4363The process_table field specifies the address and size of the guest 4364process table, which is in the guest's space. This field is formatted 4365as the second doubleword of the partition table entry, as defined in 4366the Power ISA V3.00, Book III section 5.7.6.1. 4367 43684.101 KVM_PPC_GET_RMMU_INFO 4369--------------------------- 4370 4371:Capability: KVM_CAP_PPC_MMU_RADIX 4372:Architectures: ppc 4373:Type: vm ioctl 4374:Parameters: struct kvm_ppc_rmmu_info (out) 4375:Returns: 0 on success, 4376 -EFAULT if struct kvm_ppc_rmmu_info cannot be written, 4377 -EINVAL if no useful information can be returned 4378 4379This ioctl returns a structure containing two things: (a) a list 4380containing supported radix tree geometries, and (b) a list that maps 4381page sizes to put in the "AP" (actual page size) field for the tlbie 4382(TLB invalidate entry) instruction. 4383 4384:: 4385 4386 struct kvm_ppc_rmmu_info { 4387 struct kvm_ppc_radix_geom { 4388 __u8 page_shift; 4389 __u8 level_bits[4]; 4390 __u8 pad[3]; 4391 } geometries[8]; 4392 __u32 ap_encodings[8]; 4393 }; 4394 4395The geometries[] field gives up to 8 supported geometries for the 4396radix page table, in terms of the log base 2 of the smallest page 4397size, and the number of bits indexed at each level of the tree, from 4398the PTE level up to the PGD level in that order. Any unused entries 4399will have 0 in the page_shift field. 4400 4401The ap_encodings gives the supported page sizes and their AP field 4402encodings, encoded with the AP value in the top 3 bits and the log 4403base 2 of the page size in the bottom 6 bits. 4404 44054.102 KVM_PPC_RESIZE_HPT_PREPARE 4406-------------------------------- 4407 4408:Capability: KVM_CAP_SPAPR_RESIZE_HPT 4409:Architectures: powerpc 4410:Type: vm ioctl 4411:Parameters: struct kvm_ppc_resize_hpt (in) 4412:Returns: 0 on successful completion, 4413 >0 if a new HPT is being prepared, the value is an estimated 4414 number of milliseconds until preparation is complete, 4415 -EFAULT if struct kvm_reinject_control cannot be read, 4416 -EINVAL if the supplied shift or flags are invalid, 4417 -ENOMEM if unable to allocate the new HPT, 4418 4419Used to implement the PAPR extension for runtime resizing of a guest's 4420Hashed Page Table (HPT). Specifically this starts, stops or monitors 4421the preparation of a new potential HPT for the guest, essentially 4422implementing the H_RESIZE_HPT_PREPARE hypercall. 4423 4424:: 4425 4426 struct kvm_ppc_resize_hpt { 4427 __u64 flags; 4428 __u32 shift; 4429 __u32 pad; 4430 }; 4431 4432If called with shift > 0 when there is no pending HPT for the guest, 4433this begins preparation of a new pending HPT of size 2^(shift) bytes. 4434It then returns a positive integer with the estimated number of 4435milliseconds until preparation is complete. 4436 4437If called when there is a pending HPT whose size does not match that 4438requested in the parameters, discards the existing pending HPT and 4439creates a new one as above. 4440 4441If called when there is a pending HPT of the size requested, will: 4442 4443 * If preparation of the pending HPT is already complete, return 0 4444 * If preparation of the pending HPT has failed, return an error 4445 code, then discard the pending HPT. 4446 * If preparation of the pending HPT is still in progress, return an 4447 estimated number of milliseconds until preparation is complete. 4448 4449If called with shift == 0, discards any currently pending HPT and 4450returns 0 (i.e. cancels any in-progress preparation). 4451 4452flags is reserved for future expansion, currently setting any bits in 4453flags will result in an -EINVAL. 4454 4455Normally this will be called repeatedly with the same parameters until 4456it returns <= 0. The first call will initiate preparation, subsequent 4457ones will monitor preparation until it completes or fails. 4458 44594.103 KVM_PPC_RESIZE_HPT_COMMIT 4460------------------------------- 4461 4462:Capability: KVM_CAP_SPAPR_RESIZE_HPT 4463:Architectures: powerpc 4464:Type: vm ioctl 4465:Parameters: struct kvm_ppc_resize_hpt (in) 4466:Returns: 0 on successful completion, 4467 -EFAULT if struct kvm_reinject_control cannot be read, 4468 -EINVAL if the supplied shift or flags are invalid, 4469 -ENXIO is there is no pending HPT, or the pending HPT doesn't 4470 have the requested size, 4471 -EBUSY if the pending HPT is not fully prepared, 4472 -ENOSPC if there was a hash collision when moving existing 4473 HPT entries to the new HPT, 4474 -EIO on other error conditions 4475 4476Used to implement the PAPR extension for runtime resizing of a guest's 4477Hashed Page Table (HPT). Specifically this requests that the guest be 4478transferred to working with the new HPT, essentially implementing the 4479H_RESIZE_HPT_COMMIT hypercall. 4480 4481:: 4482 4483 struct kvm_ppc_resize_hpt { 4484 __u64 flags; 4485 __u32 shift; 4486 __u32 pad; 4487 }; 4488 4489This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has 4490returned 0 with the same parameters. In other cases 4491KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or 4492-EBUSY, though others may be possible if the preparation was started, 4493but failed). 4494 4495This will have undefined effects on the guest if it has not already 4496placed itself in a quiescent state where no vcpu will make MMU enabled 4497memory accesses. 4498 4499On successful completion, the pending HPT will become the guest's active 4500HPT and the previous HPT will be discarded. 4501 4502On failure, the guest will still be operating on its previous HPT. 4503 45044.104 KVM_X86_GET_MCE_CAP_SUPPORTED 4505----------------------------------- 4506 4507:Capability: KVM_CAP_MCE 4508:Architectures: x86 4509:Type: system ioctl 4510:Parameters: u64 mce_cap (out) 4511:Returns: 0 on success, -1 on error 4512 4513Returns supported MCE capabilities. The u64 mce_cap parameter 4514has the same format as the MSR_IA32_MCG_CAP register. Supported 4515capabilities will have the corresponding bits set. 4516 45174.105 KVM_X86_SETUP_MCE 4518----------------------- 4519 4520:Capability: KVM_CAP_MCE 4521:Architectures: x86 4522:Type: vcpu ioctl 4523:Parameters: u64 mcg_cap (in) 4524:Returns: 0 on success, 4525 -EFAULT if u64 mcg_cap cannot be read, 4526 -EINVAL if the requested number of banks is invalid, 4527 -EINVAL if requested MCE capability is not supported. 4528 4529Initializes MCE support for use. The u64 mcg_cap parameter 4530has the same format as the MSR_IA32_MCG_CAP register and 4531specifies which capabilities should be enabled. The maximum 4532supported number of error-reporting banks can be retrieved when 4533checking for KVM_CAP_MCE. The supported capabilities can be 4534retrieved with KVM_X86_GET_MCE_CAP_SUPPORTED. 4535 45364.106 KVM_X86_SET_MCE 4537--------------------- 4538 4539:Capability: KVM_CAP_MCE 4540:Architectures: x86 4541:Type: vcpu ioctl 4542:Parameters: struct kvm_x86_mce (in) 4543:Returns: 0 on success, 4544 -EFAULT if struct kvm_x86_mce cannot be read, 4545 -EINVAL if the bank number is invalid, 4546 -EINVAL if VAL bit is not set in status field. 4547 4548Inject a machine check error (MCE) into the guest. The input 4549parameter is:: 4550 4551 struct kvm_x86_mce { 4552 __u64 status; 4553 __u64 addr; 4554 __u64 misc; 4555 __u64 mcg_status; 4556 __u8 bank; 4557 __u8 pad1[7]; 4558 __u64 pad2[3]; 4559 }; 4560 4561If the MCE being reported is an uncorrected error, KVM will 4562inject it as an MCE exception into the guest. If the guest 4563MCG_STATUS register reports that an MCE is in progress, KVM 4564causes an KVM_EXIT_SHUTDOWN vmexit. 4565 4566Otherwise, if the MCE is a corrected error, KVM will just 4567store it in the corresponding bank (provided this bank is 4568not holding a previously reported uncorrected error). 4569 45704.107 KVM_S390_GET_CMMA_BITS 4571---------------------------- 4572 4573:Capability: KVM_CAP_S390_CMMA_MIGRATION 4574:Architectures: s390 4575:Type: vm ioctl 4576:Parameters: struct kvm_s390_cmma_log (in, out) 4577:Returns: 0 on success, a negative value on error 4578 4579Errors: 4580 4581 ====== ============================================================= 4582 ENOMEM not enough memory can be allocated to complete the task 4583 ENXIO if CMMA is not enabled 4584 EINVAL if KVM_S390_CMMA_PEEK is not set but migration mode was not enabled 4585 EINVAL if KVM_S390_CMMA_PEEK is not set but dirty tracking has been 4586 disabled (and thus migration mode was automatically disabled) 4587 EFAULT if the userspace address is invalid or if no page table is 4588 present for the addresses (e.g. when using hugepages). 4589 ====== ============================================================= 4590 4591This ioctl is used to get the values of the CMMA bits on the s390 4592architecture. It is meant to be used in two scenarios: 4593 4594- During live migration to save the CMMA values. Live migration needs 4595 to be enabled via the KVM_REQ_START_MIGRATION VM property. 4596- To non-destructively peek at the CMMA values, with the flag 4597 KVM_S390_CMMA_PEEK set. 4598 4599The ioctl takes parameters via the kvm_s390_cmma_log struct. The desired 4600values are written to a buffer whose location is indicated via the "values" 4601member in the kvm_s390_cmma_log struct. The values in the input struct are 4602also updated as needed. 4603 4604Each CMMA value takes up one byte. 4605 4606:: 4607 4608 struct kvm_s390_cmma_log { 4609 __u64 start_gfn; 4610 __u32 count; 4611 __u32 flags; 4612 union { 4613 __u64 remaining; 4614 __u64 mask; 4615 }; 4616 __u64 values; 4617 }; 4618 4619start_gfn is the number of the first guest frame whose CMMA values are 4620to be retrieved, 4621 4622count is the length of the buffer in bytes, 4623 4624values points to the buffer where the result will be written to. 4625 4626If count is greater than KVM_S390_SKEYS_MAX, then it is considered to be 4627KVM_S390_SKEYS_MAX. KVM_S390_SKEYS_MAX is re-used for consistency with 4628other ioctls. 4629 4630The result is written in the buffer pointed to by the field values, and 4631the values of the input parameter are updated as follows. 4632 4633Depending on the flags, different actions are performed. The only 4634supported flag so far is KVM_S390_CMMA_PEEK. 4635 4636The default behaviour if KVM_S390_CMMA_PEEK is not set is: 4637start_gfn will indicate the first page frame whose CMMA bits were dirty. 4638It is not necessarily the same as the one passed as input, as clean pages 4639are skipped. 4640 4641count will indicate the number of bytes actually written in the buffer. 4642It can (and very often will) be smaller than the input value, since the 4643buffer is only filled until 16 bytes of clean values are found (which 4644are then not copied in the buffer). Since a CMMA migration block needs 4645the base address and the length, for a total of 16 bytes, we will send 4646back some clean data if there is some dirty data afterwards, as long as 4647the size of the clean data does not exceed the size of the header. This 4648allows to minimize the amount of data to be saved or transferred over 4649the network at the expense of more roundtrips to userspace. The next 4650invocation of the ioctl will skip over all the clean values, saving 4651potentially more than just the 16 bytes we found. 4652 4653If KVM_S390_CMMA_PEEK is set: 4654the existing storage attributes are read even when not in migration 4655mode, and no other action is performed; 4656 4657the output start_gfn will be equal to the input start_gfn, 4658 4659the output count will be equal to the input count, except if the end of 4660memory has been reached. 4661 4662In both cases: 4663the field "remaining" will indicate the total number of dirty CMMA values 4664still remaining, or 0 if KVM_S390_CMMA_PEEK is set and migration mode is 4665not enabled. 4666 4667mask is unused. 4668 4669values points to the userspace buffer where the result will be stored. 4670 46714.108 KVM_S390_SET_CMMA_BITS 4672---------------------------- 4673 4674:Capability: KVM_CAP_S390_CMMA_MIGRATION 4675:Architectures: s390 4676:Type: vm ioctl 4677:Parameters: struct kvm_s390_cmma_log (in) 4678:Returns: 0 on success, a negative value on error 4679 4680This ioctl is used to set the values of the CMMA bits on the s390 4681architecture. It is meant to be used during live migration to restore 4682the CMMA values, but there are no restrictions on its use. 4683The ioctl takes parameters via the kvm_s390_cmma_values struct. 4684Each CMMA value takes up one byte. 4685 4686:: 4687 4688 struct kvm_s390_cmma_log { 4689 __u64 start_gfn; 4690 __u32 count; 4691 __u32 flags; 4692 union { 4693 __u64 remaining; 4694 __u64 mask; 4695 }; 4696 __u64 values; 4697 }; 4698 4699start_gfn indicates the starting guest frame number, 4700 4701count indicates how many values are to be considered in the buffer, 4702 4703flags is not used and must be 0. 4704 4705mask indicates which PGSTE bits are to be considered. 4706 4707remaining is not used. 4708 4709values points to the buffer in userspace where to store the values. 4710 4711This ioctl can fail with -ENOMEM if not enough memory can be allocated to 4712complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if 4713the count field is too large (e.g. more than KVM_S390_CMMA_SIZE_MAX) or 4714if the flags field was not 0, with -EFAULT if the userspace address is 4715invalid, if invalid pages are written to (e.g. after the end of memory) 4716or if no page table is present for the addresses (e.g. when using 4717hugepages). 4718 47194.109 KVM_PPC_GET_CPU_CHAR 4720-------------------------- 4721 4722:Capability: KVM_CAP_PPC_GET_CPU_CHAR 4723:Architectures: powerpc 4724:Type: vm ioctl 4725:Parameters: struct kvm_ppc_cpu_char (out) 4726:Returns: 0 on successful completion, 4727 -EFAULT if struct kvm_ppc_cpu_char cannot be written 4728 4729This ioctl gives userspace information about certain characteristics 4730of the CPU relating to speculative execution of instructions and 4731possible information leakage resulting from speculative execution (see 4732CVE-2017-5715, CVE-2017-5753 and CVE-2017-5754). The information is 4733returned in struct kvm_ppc_cpu_char, which looks like this:: 4734 4735 struct kvm_ppc_cpu_char { 4736 __u64 character; /* characteristics of the CPU */ 4737 __u64 behaviour; /* recommended software behaviour */ 4738 __u64 character_mask; /* valid bits in character */ 4739 __u64 behaviour_mask; /* valid bits in behaviour */ 4740 }; 4741 4742For extensibility, the character_mask and behaviour_mask fields 4743indicate which bits of character and behaviour have been filled in by 4744the kernel. If the set of defined bits is extended in future then 4745userspace will be able to tell whether it is running on a kernel that 4746knows about the new bits. 4747 4748The character field describes attributes of the CPU which can help 4749with preventing inadvertent information disclosure - specifically, 4750whether there is an instruction to flash-invalidate the L1 data cache 4751(ori 30,30,0 or mtspr SPRN_TRIG2,rN), whether the L1 data cache is set 4752to a mode where entries can only be used by the thread that created 4753them, whether the bcctr[l] instruction prevents speculation, and 4754whether a speculation barrier instruction (ori 31,31,0) is provided. 4755 4756The behaviour field describes actions that software should take to 4757prevent inadvertent information disclosure, and thus describes which 4758vulnerabilities the hardware is subject to; specifically whether the 4759L1 data cache should be flushed when returning to user mode from the 4760kernel, and whether a speculation barrier should be placed between an 4761array bounds check and the array access. 4762 4763These fields use the same bit definitions as the new 4764H_GET_CPU_CHARACTERISTICS hypercall. 4765 47664.110 KVM_MEMORY_ENCRYPT_OP 4767--------------------------- 4768 4769:Capability: basic 4770:Architectures: x86 4771:Type: vm 4772:Parameters: an opaque platform specific structure (in/out) 4773:Returns: 0 on success; -1 on error 4774 4775If the platform supports creating encrypted VMs then this ioctl can be used 4776for issuing platform-specific memory encryption commands to manage those 4777encrypted VMs. 4778 4779Currently, this ioctl is used for issuing Secure Encrypted Virtualization 4780(SEV) commands on AMD Processors. The SEV commands are defined in 4781Documentation/virt/kvm/x86/amd-memory-encryption.rst. 4782 47834.111 KVM_MEMORY_ENCRYPT_REG_REGION 4784----------------------------------- 4785 4786:Capability: basic 4787:Architectures: x86 4788:Type: system 4789:Parameters: struct kvm_enc_region (in) 4790:Returns: 0 on success; -1 on error 4791 4792This ioctl can be used to register a guest memory region which may 4793contain encrypted data (e.g. guest RAM, SMRAM etc). 4794 4795It is used in the SEV-enabled guest. When encryption is enabled, a guest 4796memory region may contain encrypted data. The SEV memory encryption 4797engine uses a tweak such that two identical plaintext pages, each at 4798different locations will have differing ciphertexts. So swapping or 4799moving ciphertext of those pages will not result in plaintext being 4800swapped. So relocating (or migrating) physical backing pages for the SEV 4801guest will require some additional steps. 4802 4803Note: The current SEV key management spec does not provide commands to 4804swap or migrate (move) ciphertext pages. Hence, for now we pin the guest 4805memory region registered with the ioctl. 4806 48074.112 KVM_MEMORY_ENCRYPT_UNREG_REGION 4808------------------------------------- 4809 4810:Capability: basic 4811:Architectures: x86 4812:Type: system 4813:Parameters: struct kvm_enc_region (in) 4814:Returns: 0 on success; -1 on error 4815 4816This ioctl can be used to unregister the guest memory region registered 4817with KVM_MEMORY_ENCRYPT_REG_REGION ioctl above. 4818 48194.113 KVM_HYPERV_EVENTFD 4820------------------------ 4821 4822:Capability: KVM_CAP_HYPERV_EVENTFD 4823:Architectures: x86 4824:Type: vm ioctl 4825:Parameters: struct kvm_hyperv_eventfd (in) 4826 4827This ioctl (un)registers an eventfd to receive notifications from the guest on 4828the specified Hyper-V connection id through the SIGNAL_EVENT hypercall, without 4829causing a user exit. SIGNAL_EVENT hypercall with non-zero event flag number 4830(bits 24-31) still triggers a KVM_EXIT_HYPERV_HCALL user exit. 4831 4832:: 4833 4834 struct kvm_hyperv_eventfd { 4835 __u32 conn_id; 4836 __s32 fd; 4837 __u32 flags; 4838 __u32 padding[3]; 4839 }; 4840 4841The conn_id field should fit within 24 bits:: 4842 4843 #define KVM_HYPERV_CONN_ID_MASK 0x00ffffff 4844 4845The acceptable values for the flags field are:: 4846 4847 #define KVM_HYPERV_EVENTFD_DEASSIGN (1 << 0) 4848 4849:Returns: 0 on success, 4850 -EINVAL if conn_id or flags is outside the allowed range, 4851 -ENOENT on deassign if the conn_id isn't registered, 4852 -EEXIST on assign if the conn_id is already registered 4853 48544.114 KVM_GET_NESTED_STATE 4855-------------------------- 4856 4857:Capability: KVM_CAP_NESTED_STATE 4858:Architectures: x86 4859:Type: vcpu ioctl 4860:Parameters: struct kvm_nested_state (in/out) 4861:Returns: 0 on success, -1 on error 4862 4863Errors: 4864 4865 ===== ============================================================= 4866 E2BIG the total state size exceeds the value of 'size' specified by 4867 the user; the size required will be written into size. 4868 ===== ============================================================= 4869 4870:: 4871 4872 struct kvm_nested_state { 4873 __u16 flags; 4874 __u16 format; 4875 __u32 size; 4876 4877 union { 4878 struct kvm_vmx_nested_state_hdr vmx; 4879 struct kvm_svm_nested_state_hdr svm; 4880 4881 /* Pad the header to 128 bytes. */ 4882 __u8 pad[120]; 4883 } hdr; 4884 4885 union { 4886 struct kvm_vmx_nested_state_data vmx[0]; 4887 struct kvm_svm_nested_state_data svm[0]; 4888 } data; 4889 }; 4890 4891 #define KVM_STATE_NESTED_GUEST_MODE 0x00000001 4892 #define KVM_STATE_NESTED_RUN_PENDING 0x00000002 4893 #define KVM_STATE_NESTED_EVMCS 0x00000004 4894 4895 #define KVM_STATE_NESTED_FORMAT_VMX 0 4896 #define KVM_STATE_NESTED_FORMAT_SVM 1 4897 4898 #define KVM_STATE_NESTED_VMX_VMCS_SIZE 0x1000 4899 4900 #define KVM_STATE_NESTED_VMX_SMM_GUEST_MODE 0x00000001 4901 #define KVM_STATE_NESTED_VMX_SMM_VMXON 0x00000002 4902 4903 #define KVM_STATE_VMX_PREEMPTION_TIMER_DEADLINE 0x00000001 4904 4905 struct kvm_vmx_nested_state_hdr { 4906 __u64 vmxon_pa; 4907 __u64 vmcs12_pa; 4908 4909 struct { 4910 __u16 flags; 4911 } smm; 4912 4913 __u32 flags; 4914 __u64 preemption_timer_deadline; 4915 }; 4916 4917 struct kvm_vmx_nested_state_data { 4918 __u8 vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE]; 4919 __u8 shadow_vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE]; 4920 }; 4921 4922This ioctl copies the vcpu's nested virtualization state from the kernel to 4923userspace. 4924 4925The maximum size of the state can be retrieved by passing KVM_CAP_NESTED_STATE 4926to the KVM_CHECK_EXTENSION ioctl(). 4927 49284.115 KVM_SET_NESTED_STATE 4929-------------------------- 4930 4931:Capability: KVM_CAP_NESTED_STATE 4932:Architectures: x86 4933:Type: vcpu ioctl 4934:Parameters: struct kvm_nested_state (in) 4935:Returns: 0 on success, -1 on error 4936 4937This copies the vcpu's kvm_nested_state struct from userspace to the kernel. 4938For the definition of struct kvm_nested_state, see KVM_GET_NESTED_STATE. 4939 49404.116 KVM_(UN)REGISTER_COALESCED_MMIO 4941------------------------------------- 4942 4943:Capability: KVM_CAP_COALESCED_MMIO (for coalesced mmio) 4944 KVM_CAP_COALESCED_PIO (for coalesced pio) 4945:Architectures: all 4946:Type: vm ioctl 4947:Parameters: struct kvm_coalesced_mmio_zone 4948:Returns: 0 on success, < 0 on error 4949 4950Coalesced I/O is a performance optimization that defers hardware 4951register write emulation so that userspace exits are avoided. It is 4952typically used to reduce the overhead of emulating frequently accessed 4953hardware registers. 4954 4955When a hardware register is configured for coalesced I/O, write accesses 4956do not exit to userspace and their value is recorded in a ring buffer 4957that is shared between kernel and userspace. 4958 4959Coalesced I/O is used if one or more write accesses to a hardware 4960register can be deferred until a read or a write to another hardware 4961register on the same device. This last access will cause a vmexit and 4962userspace will process accesses from the ring buffer before emulating 4963it. That will avoid exiting to userspace on repeated writes. 4964 4965Coalesced pio is based on coalesced mmio. There is little difference 4966between coalesced mmio and pio except that coalesced pio records accesses 4967to I/O ports. 4968 49694.117 KVM_CLEAR_DIRTY_LOG 4970------------------------- 4971 4972:Capability: KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 4973:Architectures: x86, arm64, mips 4974:Type: vm ioctl 4975:Parameters: struct kvm_clear_dirty_log (in) 4976:Returns: 0 on success, -1 on error 4977 4978:: 4979 4980 /* for KVM_CLEAR_DIRTY_LOG */ 4981 struct kvm_clear_dirty_log { 4982 __u32 slot; 4983 __u32 num_pages; 4984 __u64 first_page; 4985 union { 4986 void __user *dirty_bitmap; /* one bit per page */ 4987 __u64 padding; 4988 }; 4989 }; 4990 4991The ioctl clears the dirty status of pages in a memory slot, according to 4992the bitmap that is passed in struct kvm_clear_dirty_log's dirty_bitmap 4993field. Bit 0 of the bitmap corresponds to page "first_page" in the 4994memory slot, and num_pages is the size in bits of the input bitmap. 4995first_page must be a multiple of 64; num_pages must also be a multiple of 499664 unless first_page + num_pages is the size of the memory slot. For each 4997bit that is set in the input bitmap, the corresponding page is marked "clean" 4998in KVM's dirty bitmap, and dirty tracking is re-enabled for that page 4999(for example via write-protection, or by clearing the dirty bit in 5000a page table entry). 5001 5002If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies 5003the address space for which you want to clear the dirty status. See 5004KVM_SET_USER_MEMORY_REGION for details on the usage of slot field. 5005 5006This ioctl is mostly useful when KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 5007is enabled; for more information, see the description of the capability. 5008However, it can always be used as long as KVM_CHECK_EXTENSION confirms 5009that KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is present. 5010 50114.118 KVM_GET_SUPPORTED_HV_CPUID 5012-------------------------------- 5013 5014:Capability: KVM_CAP_HYPERV_CPUID (vcpu), KVM_CAP_SYS_HYPERV_CPUID (system) 5015:Architectures: x86 5016:Type: system ioctl, vcpu ioctl 5017:Parameters: struct kvm_cpuid2 (in/out) 5018:Returns: 0 on success, -1 on error 5019 5020:: 5021 5022 struct kvm_cpuid2 { 5023 __u32 nent; 5024 __u32 padding; 5025 struct kvm_cpuid_entry2 entries[0]; 5026 }; 5027 5028 struct kvm_cpuid_entry2 { 5029 __u32 function; 5030 __u32 index; 5031 __u32 flags; 5032 __u32 eax; 5033 __u32 ebx; 5034 __u32 ecx; 5035 __u32 edx; 5036 __u32 padding[3]; 5037 }; 5038 5039This ioctl returns x86 cpuid features leaves related to Hyper-V emulation in 5040KVM. Userspace can use the information returned by this ioctl to construct 5041cpuid information presented to guests consuming Hyper-V enlightenments (e.g. 5042Windows or Hyper-V guests). 5043 5044CPUID feature leaves returned by this ioctl are defined by Hyper-V Top Level 5045Functional Specification (TLFS). These leaves can't be obtained with 5046KVM_GET_SUPPORTED_CPUID ioctl because some of them intersect with KVM feature 5047leaves (0x40000000, 0x40000001). 5048 5049Currently, the following list of CPUID leaves are returned: 5050 5051 - HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS 5052 - HYPERV_CPUID_INTERFACE 5053 - HYPERV_CPUID_VERSION 5054 - HYPERV_CPUID_FEATURES 5055 - HYPERV_CPUID_ENLIGHTMENT_INFO 5056 - HYPERV_CPUID_IMPLEMENT_LIMITS 5057 - HYPERV_CPUID_NESTED_FEATURES 5058 - HYPERV_CPUID_SYNDBG_VENDOR_AND_MAX_FUNCTIONS 5059 - HYPERV_CPUID_SYNDBG_INTERFACE 5060 - HYPERV_CPUID_SYNDBG_PLATFORM_CAPABILITIES 5061 5062Userspace invokes KVM_GET_SUPPORTED_HV_CPUID by passing a kvm_cpuid2 structure 5063with the 'nent' field indicating the number of entries in the variable-size 5064array 'entries'. If the number of entries is too low to describe all Hyper-V 5065feature leaves, an error (E2BIG) is returned. If the number is more or equal 5066to the number of Hyper-V feature leaves, the 'nent' field is adjusted to the 5067number of valid entries in the 'entries' array, which is then filled. 5068 5069'index' and 'flags' fields in 'struct kvm_cpuid_entry2' are currently reserved, 5070userspace should not expect to get any particular value there. 5071 5072Note, vcpu version of KVM_GET_SUPPORTED_HV_CPUID is currently deprecated. Unlike 5073system ioctl which exposes all supported feature bits unconditionally, vcpu 5074version has the following quirks: 5075 5076- HYPERV_CPUID_NESTED_FEATURES leaf and HV_X64_ENLIGHTENED_VMCS_RECOMMENDED 5077 feature bit are only exposed when Enlightened VMCS was previously enabled 5078 on the corresponding vCPU (KVM_CAP_HYPERV_ENLIGHTENED_VMCS). 5079- HV_STIMER_DIRECT_MODE_AVAILABLE bit is only exposed with in-kernel LAPIC. 5080 (presumes KVM_CREATE_IRQCHIP has already been called). 5081 50824.119 KVM_ARM_VCPU_FINALIZE 5083--------------------------- 5084 5085:Architectures: arm64 5086:Type: vcpu ioctl 5087:Parameters: int feature (in) 5088:Returns: 0 on success, -1 on error 5089 5090Errors: 5091 5092 ====== ============================================================== 5093 EPERM feature not enabled, needs configuration, or already finalized 5094 EINVAL feature unknown or not present 5095 ====== ============================================================== 5096 5097Recognised values for feature: 5098 5099 ===== =========================================== 5100 arm64 KVM_ARM_VCPU_SVE (requires KVM_CAP_ARM_SVE) 5101 ===== =========================================== 5102 5103Finalizes the configuration of the specified vcpu feature. 5104 5105The vcpu must already have been initialised, enabling the affected feature, by 5106means of a successful :ref:`KVM_ARM_VCPU_INIT <KVM_ARM_VCPU_INIT>` call with the 5107appropriate flag set in features[]. 5108 5109For affected vcpu features, this is a mandatory step that must be performed 5110before the vcpu is fully usable. 5111 5112Between KVM_ARM_VCPU_INIT and KVM_ARM_VCPU_FINALIZE, the feature may be 5113configured by use of ioctls such as KVM_SET_ONE_REG. The exact configuration 5114that should be performed and how to do it are feature-dependent. 5115 5116Other calls that depend on a particular feature being finalized, such as 5117KVM_RUN, KVM_GET_REG_LIST, KVM_GET_ONE_REG and KVM_SET_ONE_REG, will fail with 5118-EPERM unless the feature has already been finalized by means of a 5119KVM_ARM_VCPU_FINALIZE call. 5120 5121See KVM_ARM_VCPU_INIT for details of vcpu features that require finalization 5122using this ioctl. 5123 51244.120 KVM_SET_PMU_EVENT_FILTER 5125------------------------------ 5126 5127:Capability: KVM_CAP_PMU_EVENT_FILTER 5128:Architectures: x86 5129:Type: vm ioctl 5130:Parameters: struct kvm_pmu_event_filter (in) 5131:Returns: 0 on success, -1 on error 5132 5133Errors: 5134 5135 ====== ============================================================ 5136 EFAULT args[0] cannot be accessed 5137 EINVAL args[0] contains invalid data in the filter or filter events 5138 E2BIG nevents is too large 5139 EBUSY not enough memory to allocate the filter 5140 ====== ============================================================ 5141 5142:: 5143 5144 struct kvm_pmu_event_filter { 5145 __u32 action; 5146 __u32 nevents; 5147 __u32 fixed_counter_bitmap; 5148 __u32 flags; 5149 __u32 pad[4]; 5150 __u64 events[0]; 5151 }; 5152 5153This ioctl restricts the set of PMU events the guest can program by limiting 5154which event select and unit mask combinations are permitted. 5155 5156The argument holds a list of filter events which will be allowed or denied. 5157 5158Filter events only control general purpose counters; fixed purpose counters 5159are controlled by the fixed_counter_bitmap. 5160 5161Valid values for 'flags':: 5162 5163``0`` 5164 5165To use this mode, clear the 'flags' field. 5166 5167In this mode each event will contain an event select + unit mask. 5168 5169When the guest attempts to program the PMU the guest's event select + 5170unit mask is compared against the filter events to determine whether the 5171guest should have access. 5172 5173``KVM_PMU_EVENT_FLAG_MASKED_EVENTS`` 5174:Capability: KVM_CAP_PMU_EVENT_MASKED_EVENTS 5175 5176In this mode each filter event will contain an event select, mask, match, and 5177exclude value. To encode a masked event use:: 5178 5179 KVM_PMU_ENCODE_MASKED_ENTRY() 5180 5181An encoded event will follow this layout:: 5182 5183 Bits Description 5184 ---- ----------- 5185 7:0 event select (low bits) 5186 15:8 umask match 5187 31:16 unused 5188 35:32 event select (high bits) 5189 36:54 unused 5190 55 exclude bit 5191 63:56 umask mask 5192 5193When the guest attempts to program the PMU, these steps are followed in 5194determining if the guest should have access: 5195 5196 1. Match the event select from the guest against the filter events. 5197 2. If a match is found, match the guest's unit mask to the mask and match 5198 values of the included filter events. 5199 I.e. (unit mask & mask) == match && !exclude. 5200 3. If a match is found, match the guest's unit mask to the mask and match 5201 values of the excluded filter events. 5202 I.e. (unit mask & mask) == match && exclude. 5203 4. 5204 a. If an included match is found and an excluded match is not found, filter 5205 the event. 5206 b. For everything else, do not filter the event. 5207 5. 5208 a. If the event is filtered and it's an allow list, allow the guest to 5209 program the event. 5210 b. If the event is filtered and it's a deny list, do not allow the guest to 5211 program the event. 5212 5213When setting a new pmu event filter, -EINVAL will be returned if any of the 5214unused fields are set or if any of the high bits (35:32) in the event 5215select are set when called on Intel. 5216 5217Valid values for 'action':: 5218 5219 #define KVM_PMU_EVENT_ALLOW 0 5220 #define KVM_PMU_EVENT_DENY 1 5221 5222Via this API, KVM userspace can also control the behavior of the VM's fixed 5223counters (if any) by configuring the "action" and "fixed_counter_bitmap" fields. 5224 5225Specifically, KVM follows the following pseudo-code when determining whether to 5226allow the guest FixCtr[i] to count its pre-defined fixed event:: 5227 5228 FixCtr[i]_is_allowed = (action == ALLOW) && (bitmap & BIT(i)) || 5229 (action == DENY) && !(bitmap & BIT(i)); 5230 FixCtr[i]_is_denied = !FixCtr[i]_is_allowed; 5231 5232KVM always consumes fixed_counter_bitmap, it's userspace's responsibility to 5233ensure fixed_counter_bitmap is set correctly, e.g. if userspace wants to define 5234a filter that only affects general purpose counters. 5235 5236Note, the "events" field also applies to fixed counters' hardcoded event_select 5237and unit_mask values. "fixed_counter_bitmap" has higher priority than "events" 5238if there is a contradiction between the two. 5239 52404.121 KVM_PPC_SVM_OFF 5241--------------------- 5242 5243:Capability: basic 5244:Architectures: powerpc 5245:Type: vm ioctl 5246:Parameters: none 5247:Returns: 0 on successful completion, 5248 5249Errors: 5250 5251 ====== ================================================================ 5252 EINVAL if ultravisor failed to terminate the secure guest 5253 ENOMEM if hypervisor failed to allocate new radix page tables for guest 5254 ====== ================================================================ 5255 5256This ioctl is used to turn off the secure mode of the guest or transition 5257the guest from secure mode to normal mode. This is invoked when the guest 5258is reset. This has no effect if called for a normal guest. 5259 5260This ioctl issues an ultravisor call to terminate the secure guest, 5261unpins the VPA pages and releases all the device pages that are used to 5262track the secure pages by hypervisor. 5263 52644.122 KVM_S390_NORMAL_RESET 5265--------------------------- 5266 5267:Capability: KVM_CAP_S390_VCPU_RESETS 5268:Architectures: s390 5269:Type: vcpu ioctl 5270:Parameters: none 5271:Returns: 0 5272 5273This ioctl resets VCPU registers and control structures according to 5274the cpu reset definition in the POP (Principles Of Operation). 5275 52764.123 KVM_S390_INITIAL_RESET 5277---------------------------- 5278 5279:Capability: basic 5280:Architectures: s390 5281:Type: vcpu ioctl 5282:Parameters: none 5283:Returns: 0 5284 5285This ioctl resets VCPU registers and control structures according to 5286the initial cpu reset definition in the POP. However, the cpu is not 5287put into ESA mode. This reset is a superset of the normal reset. 5288 52894.124 KVM_S390_CLEAR_RESET 5290-------------------------- 5291 5292:Capability: KVM_CAP_S390_VCPU_RESETS 5293:Architectures: s390 5294:Type: vcpu ioctl 5295:Parameters: none 5296:Returns: 0 5297 5298This ioctl resets VCPU registers and control structures according to 5299the clear cpu reset definition in the POP. However, the cpu is not put 5300into ESA mode. This reset is a superset of the initial reset. 5301 5302 53034.125 KVM_S390_PV_COMMAND 5304------------------------- 5305 5306:Capability: KVM_CAP_S390_PROTECTED 5307:Architectures: s390 5308:Type: vm ioctl 5309:Parameters: struct kvm_pv_cmd 5310:Returns: 0 on success, < 0 on error 5311 5312:: 5313 5314 struct kvm_pv_cmd { 5315 __u32 cmd; /* Command to be executed */ 5316 __u16 rc; /* Ultravisor return code */ 5317 __u16 rrc; /* Ultravisor return reason code */ 5318 __u64 data; /* Data or address */ 5319 __u32 flags; /* flags for future extensions. Must be 0 for now */ 5320 __u32 reserved[3]; 5321 }; 5322 5323**Ultravisor return codes** 5324The Ultravisor return (reason) codes are provided by the kernel if a 5325Ultravisor call has been executed to achieve the results expected by 5326the command. Therefore they are independent of the IOCTL return 5327code. If KVM changes `rc`, its value will always be greater than 0 5328hence setting it to 0 before issuing a PV command is advised to be 5329able to detect a change of `rc`. 5330 5331**cmd values:** 5332 5333KVM_PV_ENABLE 5334 Allocate memory and register the VM with the Ultravisor, thereby 5335 donating memory to the Ultravisor that will become inaccessible to 5336 KVM. All existing CPUs are converted to protected ones. After this 5337 command has succeeded, any CPU added via hotplug will become 5338 protected during its creation as well. 5339 5340 Errors: 5341 5342 ===== ============================= 5343 EINTR an unmasked signal is pending 5344 ===== ============================= 5345 5346KVM_PV_DISABLE 5347 Deregister the VM from the Ultravisor and reclaim the memory that had 5348 been donated to the Ultravisor, making it usable by the kernel again. 5349 All registered VCPUs are converted back to non-protected ones. If a 5350 previous protected VM had been prepared for asynchronous teardown with 5351 KVM_PV_ASYNC_CLEANUP_PREPARE and not subsequently torn down with 5352 KVM_PV_ASYNC_CLEANUP_PERFORM, it will be torn down in this call 5353 together with the current protected VM. 5354 5355KVM_PV_VM_SET_SEC_PARMS 5356 Pass the image header from VM memory to the Ultravisor in 5357 preparation of image unpacking and verification. 5358 5359KVM_PV_VM_UNPACK 5360 Unpack (protect and decrypt) a page of the encrypted boot image. 5361 5362KVM_PV_VM_VERIFY 5363 Verify the integrity of the unpacked image. Only if this succeeds, 5364 KVM is allowed to start protected VCPUs. 5365 5366KVM_PV_INFO 5367 :Capability: KVM_CAP_S390_PROTECTED_DUMP 5368 5369 Presents an API that provides Ultravisor related data to userspace 5370 via subcommands. len_max is the size of the user space buffer, 5371 len_written is KVM's indication of how much bytes of that buffer 5372 were actually written to. len_written can be used to determine the 5373 valid fields if more response fields are added in the future. 5374 5375 :: 5376 5377 enum pv_cmd_info_id { 5378 KVM_PV_INFO_VM, 5379 KVM_PV_INFO_DUMP, 5380 }; 5381 5382 struct kvm_s390_pv_info_header { 5383 __u32 id; 5384 __u32 len_max; 5385 __u32 len_written; 5386 __u32 reserved; 5387 }; 5388 5389 struct kvm_s390_pv_info { 5390 struct kvm_s390_pv_info_header header; 5391 struct kvm_s390_pv_info_dump dump; 5392 struct kvm_s390_pv_info_vm vm; 5393 }; 5394 5395**subcommands:** 5396 5397 KVM_PV_INFO_VM 5398 This subcommand provides basic Ultravisor information for PV 5399 hosts. These values are likely also exported as files in the sysfs 5400 firmware UV query interface but they are more easily available to 5401 programs in this API. 5402 5403 The installed calls and feature_indication members provide the 5404 installed UV calls and the UV's other feature indications. 5405 5406 The max_* members provide information about the maximum number of PV 5407 vcpus, PV guests and PV guest memory size. 5408 5409 :: 5410 5411 struct kvm_s390_pv_info_vm { 5412 __u64 inst_calls_list[4]; 5413 __u64 max_cpus; 5414 __u64 max_guests; 5415 __u64 max_guest_addr; 5416 __u64 feature_indication; 5417 }; 5418 5419 5420 KVM_PV_INFO_DUMP 5421 This subcommand provides information related to dumping PV guests. 5422 5423 :: 5424 5425 struct kvm_s390_pv_info_dump { 5426 __u64 dump_cpu_buffer_len; 5427 __u64 dump_config_mem_buffer_per_1m; 5428 __u64 dump_config_finalize_len; 5429 }; 5430 5431KVM_PV_DUMP 5432 :Capability: KVM_CAP_S390_PROTECTED_DUMP 5433 5434 Presents an API that provides calls which facilitate dumping a 5435 protected VM. 5436 5437 :: 5438 5439 struct kvm_s390_pv_dmp { 5440 __u64 subcmd; 5441 __u64 buff_addr; 5442 __u64 buff_len; 5443 __u64 gaddr; /* For dump storage state */ 5444 }; 5445 5446 **subcommands:** 5447 5448 KVM_PV_DUMP_INIT 5449 Initializes the dump process of a protected VM. If this call does 5450 not succeed all other subcommands will fail with -EINVAL. This 5451 subcommand will return -EINVAL if a dump process has not yet been 5452 completed. 5453 5454 Not all PV vms can be dumped, the owner needs to set `dump 5455 allowed` PCF bit 34 in the SE header to allow dumping. 5456 5457 KVM_PV_DUMP_CONFIG_STOR_STATE 5458 Stores `buff_len` bytes of tweak component values starting with 5459 the 1MB block specified by the absolute guest address 5460 (`gaddr`). `buff_len` needs to be `conf_dump_storage_state_len` 5461 aligned and at least >= the `conf_dump_storage_state_len` value 5462 provided by the dump uv_info data. buff_user might be written to 5463 even if an error rc is returned. For instance if we encounter a 5464 fault after writing the first page of data. 5465 5466 KVM_PV_DUMP_COMPLETE 5467 If the subcommand succeeds it completes the dump process and lets 5468 KVM_PV_DUMP_INIT be called again. 5469 5470 On success `conf_dump_finalize_len` bytes of completion data will be 5471 stored to the `buff_addr`. The completion data contains a key 5472 derivation seed, IV, tweak nonce and encryption keys as well as an 5473 authentication tag all of which are needed to decrypt the dump at a 5474 later time. 5475 5476KVM_PV_ASYNC_CLEANUP_PREPARE 5477 :Capability: KVM_CAP_S390_PROTECTED_ASYNC_DISABLE 5478 5479 Prepare the current protected VM for asynchronous teardown. Most 5480 resources used by the current protected VM will be set aside for a 5481 subsequent asynchronous teardown. The current protected VM will then 5482 resume execution immediately as non-protected. There can be at most 5483 one protected VM prepared for asynchronous teardown at any time. If 5484 a protected VM had already been prepared for teardown without 5485 subsequently calling KVM_PV_ASYNC_CLEANUP_PERFORM, this call will 5486 fail. In that case, the userspace process should issue a normal 5487 KVM_PV_DISABLE. The resources set aside with this call will need to 5488 be cleaned up with a subsequent call to KVM_PV_ASYNC_CLEANUP_PERFORM 5489 or KVM_PV_DISABLE, otherwise they will be cleaned up when KVM 5490 terminates. KVM_PV_ASYNC_CLEANUP_PREPARE can be called again as soon 5491 as cleanup starts, i.e. before KVM_PV_ASYNC_CLEANUP_PERFORM finishes. 5492 5493KVM_PV_ASYNC_CLEANUP_PERFORM 5494 :Capability: KVM_CAP_S390_PROTECTED_ASYNC_DISABLE 5495 5496 Tear down the protected VM previously prepared for teardown with 5497 KVM_PV_ASYNC_CLEANUP_PREPARE. The resources that had been set aside 5498 will be freed during the execution of this command. This PV command 5499 should ideally be issued by userspace from a separate thread. If a 5500 fatal signal is received (or the process terminates naturally), the 5501 command will terminate immediately without completing, and the normal 5502 KVM shutdown procedure will take care of cleaning up all remaining 5503 protected VMs, including the ones whose teardown was interrupted by 5504 process termination. 5505 55064.126 KVM_XEN_HVM_SET_ATTR 5507-------------------------- 5508 5509:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 5510:Architectures: x86 5511:Type: vm ioctl 5512:Parameters: struct kvm_xen_hvm_attr 5513:Returns: 0 on success, < 0 on error 5514 5515:: 5516 5517 struct kvm_xen_hvm_attr { 5518 __u16 type; 5519 __u16 pad[3]; 5520 union { 5521 __u8 long_mode; 5522 __u8 vector; 5523 __u8 runstate_update_flag; 5524 union { 5525 __u64 gfn; 5526 __u64 hva; 5527 } shared_info; 5528 struct { 5529 __u32 send_port; 5530 __u32 type; /* EVTCHNSTAT_ipi / EVTCHNSTAT_interdomain */ 5531 __u32 flags; 5532 union { 5533 struct { 5534 __u32 port; 5535 __u32 vcpu; 5536 __u32 priority; 5537 } port; 5538 struct { 5539 __u32 port; /* Zero for eventfd */ 5540 __s32 fd; 5541 } eventfd; 5542 __u32 padding[4]; 5543 } deliver; 5544 } evtchn; 5545 __u32 xen_version; 5546 __u64 pad[8]; 5547 } u; 5548 }; 5549 5550type values: 5551 5552KVM_XEN_ATTR_TYPE_LONG_MODE 5553 Sets the ABI mode of the VM to 32-bit or 64-bit (long mode). This 5554 determines the layout of the shared_info page exposed to the VM. 5555 5556KVM_XEN_ATTR_TYPE_SHARED_INFO 5557 Sets the guest physical frame number at which the Xen shared_info 5558 page resides. Note that although Xen places vcpu_info for the first 5559 32 vCPUs in the shared_info page, KVM does not automatically do so 5560 and instead requires that KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO or 5561 KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO_HVA be used explicitly even when 5562 the vcpu_info for a given vCPU resides at the "default" location 5563 in the shared_info page. This is because KVM may not be aware of 5564 the Xen CPU id which is used as the index into the vcpu_info[] 5565 array, so may know the correct default location. 5566 5567 Note that the shared_info page may be constantly written to by KVM; 5568 it contains the event channel bitmap used to deliver interrupts to 5569 a Xen guest, amongst other things. It is exempt from dirty tracking 5570 mechanisms — KVM will not explicitly mark the page as dirty each 5571 time an event channel interrupt is delivered to the guest! Thus, 5572 userspace should always assume that the designated GFN is dirty if 5573 any vCPU has been running or any event channel interrupts can be 5574 routed to the guest. 5575 5576 Setting the gfn to KVM_XEN_INVALID_GFN will disable the shared_info 5577 page. 5578 5579KVM_XEN_ATTR_TYPE_SHARED_INFO_HVA 5580 If the KVM_XEN_HVM_CONFIG_SHARED_INFO_HVA flag is also set in the 5581 Xen capabilities, then this attribute may be used to set the 5582 userspace address at which the shared_info page resides, which 5583 will always be fixed in the VMM regardless of where it is mapped 5584 in guest physical address space. This attribute should be used in 5585 preference to KVM_XEN_ATTR_TYPE_SHARED_INFO as it avoids 5586 unnecessary invalidation of an internal cache when the page is 5587 re-mapped in guest physical address space. 5588 5589 Setting the hva to zero will disable the shared_info page. 5590 5591KVM_XEN_ATTR_TYPE_UPCALL_VECTOR 5592 Sets the exception vector used to deliver Xen event channel upcalls. 5593 This is the HVM-wide vector injected directly by the hypervisor 5594 (not through the local APIC), typically configured by a guest via 5595 HVM_PARAM_CALLBACK_IRQ. This can be disabled again (e.g. for guest 5596 SHUTDOWN_soft_reset) by setting it to zero. 5597 5598KVM_XEN_ATTR_TYPE_EVTCHN 5599 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5600 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It configures 5601 an outbound port number for interception of EVTCHNOP_send requests 5602 from the guest. A given sending port number may be directed back to 5603 a specified vCPU (by APIC ID) / port / priority on the guest, or to 5604 trigger events on an eventfd. The vCPU and priority can be changed 5605 by setting KVM_XEN_EVTCHN_UPDATE in a subsequent call, but other 5606 fields cannot change for a given sending port. A port mapping is 5607 removed by using KVM_XEN_EVTCHN_DEASSIGN in the flags field. Passing 5608 KVM_XEN_EVTCHN_RESET in the flags field removes all interception of 5609 outbound event channels. The values of the flags field are mutually 5610 exclusive and cannot be combined as a bitmask. 5611 5612KVM_XEN_ATTR_TYPE_XEN_VERSION 5613 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5614 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It configures 5615 the 32-bit version code returned to the guest when it invokes the 5616 XENVER_version call; typically (XEN_MAJOR << 16 | XEN_MINOR). PV 5617 Xen guests will often use this to as a dummy hypercall to trigger 5618 event channel delivery, so responding within the kernel without 5619 exiting to userspace is beneficial. 5620 5621KVM_XEN_ATTR_TYPE_RUNSTATE_UPDATE_FLAG 5622 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5623 support for KVM_XEN_HVM_CONFIG_RUNSTATE_UPDATE_FLAG. It enables the 5624 XEN_RUNSTATE_UPDATE flag which allows guest vCPUs to safely read 5625 other vCPUs' vcpu_runstate_info. Xen guests enable this feature via 5626 the VMASST_TYPE_runstate_update_flag of the HYPERVISOR_vm_assist 5627 hypercall. 5628 56294.127 KVM_XEN_HVM_GET_ATTR 5630-------------------------- 5631 5632:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 5633:Architectures: x86 5634:Type: vm ioctl 5635:Parameters: struct kvm_xen_hvm_attr 5636:Returns: 0 on success, < 0 on error 5637 5638Allows Xen VM attributes to be read. For the structure and types, 5639see KVM_XEN_HVM_SET_ATTR above. The KVM_XEN_ATTR_TYPE_EVTCHN 5640attribute cannot be read. 5641 56424.128 KVM_XEN_VCPU_SET_ATTR 5643--------------------------- 5644 5645:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 5646:Architectures: x86 5647:Type: vcpu ioctl 5648:Parameters: struct kvm_xen_vcpu_attr 5649:Returns: 0 on success, < 0 on error 5650 5651:: 5652 5653 struct kvm_xen_vcpu_attr { 5654 __u16 type; 5655 __u16 pad[3]; 5656 union { 5657 __u64 gpa; 5658 __u64 pad[4]; 5659 struct { 5660 __u64 state; 5661 __u64 state_entry_time; 5662 __u64 time_running; 5663 __u64 time_runnable; 5664 __u64 time_blocked; 5665 __u64 time_offline; 5666 } runstate; 5667 __u32 vcpu_id; 5668 struct { 5669 __u32 port; 5670 __u32 priority; 5671 __u64 expires_ns; 5672 } timer; 5673 __u8 vector; 5674 } u; 5675 }; 5676 5677type values: 5678 5679KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO 5680 Sets the guest physical address of the vcpu_info for a given vCPU. 5681 As with the shared_info page for the VM, the corresponding page may be 5682 dirtied at any time if event channel interrupt delivery is enabled, so 5683 userspace should always assume that the page is dirty without relying 5684 on dirty logging. Setting the gpa to KVM_XEN_INVALID_GPA will disable 5685 the vcpu_info. 5686 5687KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO_HVA 5688 If the KVM_XEN_HVM_CONFIG_SHARED_INFO_HVA flag is also set in the 5689 Xen capabilities, then this attribute may be used to set the 5690 userspace address of the vcpu_info for a given vCPU. It should 5691 only be used when the vcpu_info resides at the "default" location 5692 in the shared_info page. In this case it is safe to assume the 5693 userspace address will not change, because the shared_info page is 5694 an overlay on guest memory and remains at a fixed host address 5695 regardless of where it is mapped in guest physical address space 5696 and hence unnecessary invalidation of an internal cache may be 5697 avoided if the guest memory layout is modified. 5698 If the vcpu_info does not reside at the "default" location then 5699 it is not guaranteed to remain at the same host address and 5700 hence the aforementioned cache invalidation is required. 5701 5702KVM_XEN_VCPU_ATTR_TYPE_VCPU_TIME_INFO 5703 Sets the guest physical address of an additional pvclock structure 5704 for a given vCPU. This is typically used for guest vsyscall support. 5705 Setting the gpa to KVM_XEN_INVALID_GPA will disable the structure. 5706 5707KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR 5708 Sets the guest physical address of the vcpu_runstate_info for a given 5709 vCPU. This is how a Xen guest tracks CPU state such as steal time. 5710 Setting the gpa to KVM_XEN_INVALID_GPA will disable the runstate area. 5711 5712KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_CURRENT 5713 Sets the runstate (RUNSTATE_running/_runnable/_blocked/_offline) of 5714 the given vCPU from the .u.runstate.state member of the structure. 5715 KVM automatically accounts running and runnable time but blocked 5716 and offline states are only entered explicitly. 5717 5718KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_DATA 5719 Sets all fields of the vCPU runstate data from the .u.runstate member 5720 of the structure, including the current runstate. The state_entry_time 5721 must equal the sum of the other four times. 5722 5723KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST 5724 This *adds* the contents of the .u.runstate members of the structure 5725 to the corresponding members of the given vCPU's runstate data, thus 5726 permitting atomic adjustments to the runstate times. The adjustment 5727 to the state_entry_time must equal the sum of the adjustments to the 5728 other four times. The state field must be set to -1, or to a valid 5729 runstate value (RUNSTATE_running, RUNSTATE_runnable, RUNSTATE_blocked 5730 or RUNSTATE_offline) to set the current accounted state as of the 5731 adjusted state_entry_time. 5732 5733KVM_XEN_VCPU_ATTR_TYPE_VCPU_ID 5734 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5735 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It sets the Xen 5736 vCPU ID of the given vCPU, to allow timer-related VCPU operations to 5737 be intercepted by KVM. 5738 5739KVM_XEN_VCPU_ATTR_TYPE_TIMER 5740 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5741 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It sets the 5742 event channel port/priority for the VIRQ_TIMER of the vCPU, as well 5743 as allowing a pending timer to be saved/restored. Setting the timer 5744 port to zero disables kernel handling of the singleshot timer. 5745 5746KVM_XEN_VCPU_ATTR_TYPE_UPCALL_VECTOR 5747 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5748 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It sets the 5749 per-vCPU local APIC upcall vector, configured by a Xen guest with 5750 the HVMOP_set_evtchn_upcall_vector hypercall. This is typically 5751 used by Windows guests, and is distinct from the HVM-wide upcall 5752 vector configured with HVM_PARAM_CALLBACK_IRQ. It is disabled by 5753 setting the vector to zero. 5754 5755 57564.129 KVM_XEN_VCPU_GET_ATTR 5757--------------------------- 5758 5759:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 5760:Architectures: x86 5761:Type: vcpu ioctl 5762:Parameters: struct kvm_xen_vcpu_attr 5763:Returns: 0 on success, < 0 on error 5764 5765Allows Xen vCPU attributes to be read. For the structure and types, 5766see KVM_XEN_VCPU_SET_ATTR above. 5767 5768The KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST type may not be used 5769with the KVM_XEN_VCPU_GET_ATTR ioctl. 5770 57714.130 KVM_ARM_MTE_COPY_TAGS 5772--------------------------- 5773 5774:Capability: KVM_CAP_ARM_MTE 5775:Architectures: arm64 5776:Type: vm ioctl 5777:Parameters: struct kvm_arm_copy_mte_tags 5778:Returns: number of bytes copied, < 0 on error (-EINVAL for incorrect 5779 arguments, -EFAULT if memory cannot be accessed). 5780 5781:: 5782 5783 struct kvm_arm_copy_mte_tags { 5784 __u64 guest_ipa; 5785 __u64 length; 5786 void __user *addr; 5787 __u64 flags; 5788 __u64 reserved[2]; 5789 }; 5790 5791Copies Memory Tagging Extension (MTE) tags to/from guest tag memory. The 5792``guest_ipa`` and ``length`` fields must be ``PAGE_SIZE`` aligned. 5793``length`` must not be bigger than 2^31 - PAGE_SIZE bytes. The ``addr`` 5794field must point to a buffer which the tags will be copied to or from. 5795 5796``flags`` specifies the direction of copy, either ``KVM_ARM_TAGS_TO_GUEST`` or 5797``KVM_ARM_TAGS_FROM_GUEST``. 5798 5799The size of the buffer to store the tags is ``(length / 16)`` bytes 5800(granules in MTE are 16 bytes long). Each byte contains a single tag 5801value. This matches the format of ``PTRACE_PEEKMTETAGS`` and 5802``PTRACE_POKEMTETAGS``. 5803 5804If an error occurs before any data is copied then a negative error code is 5805returned. If some tags have been copied before an error occurs then the number 5806of bytes successfully copied is returned. If the call completes successfully 5807then ``length`` is returned. 5808 58094.131 KVM_GET_SREGS2 5810-------------------- 5811 5812:Capability: KVM_CAP_SREGS2 5813:Architectures: x86 5814:Type: vcpu ioctl 5815:Parameters: struct kvm_sregs2 (out) 5816:Returns: 0 on success, -1 on error 5817 5818Reads special registers from the vcpu. 5819This ioctl (when supported) replaces the KVM_GET_SREGS. 5820 5821:: 5822 5823 struct kvm_sregs2 { 5824 /* out (KVM_GET_SREGS2) / in (KVM_SET_SREGS2) */ 5825 struct kvm_segment cs, ds, es, fs, gs, ss; 5826 struct kvm_segment tr, ldt; 5827 struct kvm_dtable gdt, idt; 5828 __u64 cr0, cr2, cr3, cr4, cr8; 5829 __u64 efer; 5830 __u64 apic_base; 5831 __u64 flags; 5832 __u64 pdptrs[4]; 5833 }; 5834 5835flags values for ``kvm_sregs2``: 5836 5837``KVM_SREGS2_FLAGS_PDPTRS_VALID`` 5838 5839 Indicates that the struct contains valid PDPTR values. 5840 5841 58424.132 KVM_SET_SREGS2 5843-------------------- 5844 5845:Capability: KVM_CAP_SREGS2 5846:Architectures: x86 5847:Type: vcpu ioctl 5848:Parameters: struct kvm_sregs2 (in) 5849:Returns: 0 on success, -1 on error 5850 5851Writes special registers into the vcpu. 5852See KVM_GET_SREGS2 for the data structures. 5853This ioctl (when supported) replaces the KVM_SET_SREGS. 5854 58554.133 KVM_GET_STATS_FD 5856---------------------- 5857 5858:Capability: KVM_CAP_STATS_BINARY_FD 5859:Architectures: all 5860:Type: vm ioctl, vcpu ioctl 5861:Parameters: none 5862:Returns: statistics file descriptor on success, < 0 on error 5863 5864Errors: 5865 5866 ====== ====================================================== 5867 ENOMEM if the fd could not be created due to lack of memory 5868 EMFILE if the number of opened files exceeds the limit 5869 ====== ====================================================== 5870 5871The returned file descriptor can be used to read VM/vCPU statistics data in 5872binary format. The data in the file descriptor consists of four blocks 5873organized as follows: 5874 5875+-------------+ 5876| Header | 5877+-------------+ 5878| id string | 5879+-------------+ 5880| Descriptors | 5881+-------------+ 5882| Stats Data | 5883+-------------+ 5884 5885Apart from the header starting at offset 0, please be aware that it is 5886not guaranteed that the four blocks are adjacent or in the above order; 5887the offsets of the id, descriptors and data blocks are found in the 5888header. However, all four blocks are aligned to 64 bit offsets in the 5889file and they do not overlap. 5890 5891All blocks except the data block are immutable. Userspace can read them 5892only one time after retrieving the file descriptor, and then use ``pread`` or 5893``lseek`` to read the statistics repeatedly. 5894 5895All data is in system endianness. 5896 5897The format of the header is as follows:: 5898 5899 struct kvm_stats_header { 5900 __u32 flags; 5901 __u32 name_size; 5902 __u32 num_desc; 5903 __u32 id_offset; 5904 __u32 desc_offset; 5905 __u32 data_offset; 5906 }; 5907 5908The ``flags`` field is not used at the moment. It is always read as 0. 5909 5910The ``name_size`` field is the size (in byte) of the statistics name string 5911(including trailing '\0') which is contained in the "id string" block and 5912appended at the end of every descriptor. 5913 5914The ``num_desc`` field is the number of descriptors that are included in the 5915descriptor block. (The actual number of values in the data block may be 5916larger, since each descriptor may comprise more than one value). 5917 5918The ``id_offset`` field is the offset of the id string from the start of the 5919file indicated by the file descriptor. It is a multiple of 8. 5920 5921The ``desc_offset`` field is the offset of the Descriptors block from the start 5922of the file indicated by the file descriptor. It is a multiple of 8. 5923 5924The ``data_offset`` field is the offset of the Stats Data block from the start 5925of the file indicated by the file descriptor. It is a multiple of 8. 5926 5927The id string block contains a string which identifies the file descriptor on 5928which KVM_GET_STATS_FD was invoked. The size of the block, including the 5929trailing ``'\0'``, is indicated by the ``name_size`` field in the header. 5930 5931The descriptors block is only needed to be read once for the lifetime of the 5932file descriptor contains a sequence of ``struct kvm_stats_desc``, each followed 5933by a string of size ``name_size``. 5934:: 5935 5936 #define KVM_STATS_TYPE_SHIFT 0 5937 #define KVM_STATS_TYPE_MASK (0xF << KVM_STATS_TYPE_SHIFT) 5938 #define KVM_STATS_TYPE_CUMULATIVE (0x0 << KVM_STATS_TYPE_SHIFT) 5939 #define KVM_STATS_TYPE_INSTANT (0x1 << KVM_STATS_TYPE_SHIFT) 5940 #define KVM_STATS_TYPE_PEAK (0x2 << KVM_STATS_TYPE_SHIFT) 5941 #define KVM_STATS_TYPE_LINEAR_HIST (0x3 << KVM_STATS_TYPE_SHIFT) 5942 #define KVM_STATS_TYPE_LOG_HIST (0x4 << KVM_STATS_TYPE_SHIFT) 5943 #define KVM_STATS_TYPE_MAX KVM_STATS_TYPE_LOG_HIST 5944 5945 #define KVM_STATS_UNIT_SHIFT 4 5946 #define KVM_STATS_UNIT_MASK (0xF << KVM_STATS_UNIT_SHIFT) 5947 #define KVM_STATS_UNIT_NONE (0x0 << KVM_STATS_UNIT_SHIFT) 5948 #define KVM_STATS_UNIT_BYTES (0x1 << KVM_STATS_UNIT_SHIFT) 5949 #define KVM_STATS_UNIT_SECONDS (0x2 << KVM_STATS_UNIT_SHIFT) 5950 #define KVM_STATS_UNIT_CYCLES (0x3 << KVM_STATS_UNIT_SHIFT) 5951 #define KVM_STATS_UNIT_BOOLEAN (0x4 << KVM_STATS_UNIT_SHIFT) 5952 #define KVM_STATS_UNIT_MAX KVM_STATS_UNIT_BOOLEAN 5953 5954 #define KVM_STATS_BASE_SHIFT 8 5955 #define KVM_STATS_BASE_MASK (0xF << KVM_STATS_BASE_SHIFT) 5956 #define KVM_STATS_BASE_POW10 (0x0 << KVM_STATS_BASE_SHIFT) 5957 #define KVM_STATS_BASE_POW2 (0x1 << KVM_STATS_BASE_SHIFT) 5958 #define KVM_STATS_BASE_MAX KVM_STATS_BASE_POW2 5959 5960 struct kvm_stats_desc { 5961 __u32 flags; 5962 __s16 exponent; 5963 __u16 size; 5964 __u32 offset; 5965 __u32 bucket_size; 5966 char name[]; 5967 }; 5968 5969The ``flags`` field contains the type and unit of the statistics data described 5970by this descriptor. Its endianness is CPU native. 5971The following flags are supported: 5972 5973Bits 0-3 of ``flags`` encode the type: 5974 5975 * ``KVM_STATS_TYPE_CUMULATIVE`` 5976 The statistics reports a cumulative count. The value of data can only be increased. 5977 Most of the counters used in KVM are of this type. 5978 The corresponding ``size`` field for this type is always 1. 5979 All cumulative statistics data are read/write. 5980 * ``KVM_STATS_TYPE_INSTANT`` 5981 The statistics reports an instantaneous value. Its value can be increased or 5982 decreased. This type is usually used as a measurement of some resources, 5983 like the number of dirty pages, the number of large pages, etc. 5984 All instant statistics are read only. 5985 The corresponding ``size`` field for this type is always 1. 5986 * ``KVM_STATS_TYPE_PEAK`` 5987 The statistics data reports a peak value, for example the maximum number 5988 of items in a hash table bucket, the longest time waited and so on. 5989 The value of data can only be increased. 5990 The corresponding ``size`` field for this type is always 1. 5991 * ``KVM_STATS_TYPE_LINEAR_HIST`` 5992 The statistic is reported as a linear histogram. The number of 5993 buckets is specified by the ``size`` field. The size of buckets is specified 5994 by the ``hist_param`` field. The range of the Nth bucket (1 <= N < ``size``) 5995 is [``hist_param``*(N-1), ``hist_param``*N), while the range of the last 5996 bucket is [``hist_param``*(``size``-1), +INF). (+INF means positive infinity 5997 value.) 5998 * ``KVM_STATS_TYPE_LOG_HIST`` 5999 The statistic is reported as a logarithmic histogram. The number of 6000 buckets is specified by the ``size`` field. The range of the first bucket is 6001 [0, 1), while the range of the last bucket is [pow(2, ``size``-2), +INF). 6002 Otherwise, The Nth bucket (1 < N < ``size``) covers 6003 [pow(2, N-2), pow(2, N-1)). 6004 6005Bits 4-7 of ``flags`` encode the unit: 6006 6007 * ``KVM_STATS_UNIT_NONE`` 6008 There is no unit for the value of statistics data. This usually means that 6009 the value is a simple counter of an event. 6010 * ``KVM_STATS_UNIT_BYTES`` 6011 It indicates that the statistics data is used to measure memory size, in the 6012 unit of Byte, KiByte, MiByte, GiByte, etc. The unit of the data is 6013 determined by the ``exponent`` field in the descriptor. 6014 * ``KVM_STATS_UNIT_SECONDS`` 6015 It indicates that the statistics data is used to measure time or latency. 6016 * ``KVM_STATS_UNIT_CYCLES`` 6017 It indicates that the statistics data is used to measure CPU clock cycles. 6018 * ``KVM_STATS_UNIT_BOOLEAN`` 6019 It indicates that the statistic will always be either 0 or 1. Boolean 6020 statistics of "peak" type will never go back from 1 to 0. Boolean 6021 statistics can be linear histograms (with two buckets) but not logarithmic 6022 histograms. 6023 6024Note that, in the case of histograms, the unit applies to the bucket 6025ranges, while the bucket value indicates how many samples fell in the 6026bucket's range. 6027 6028Bits 8-11 of ``flags``, together with ``exponent``, encode the scale of the 6029unit: 6030 6031 * ``KVM_STATS_BASE_POW10`` 6032 The scale is based on power of 10. It is used for measurement of time and 6033 CPU clock cycles. For example, an exponent of -9 can be used with 6034 ``KVM_STATS_UNIT_SECONDS`` to express that the unit is nanoseconds. 6035 * ``KVM_STATS_BASE_POW2`` 6036 The scale is based on power of 2. It is used for measurement of memory size. 6037 For example, an exponent of 20 can be used with ``KVM_STATS_UNIT_BYTES`` to 6038 express that the unit is MiB. 6039 6040The ``size`` field is the number of values of this statistics data. Its 6041value is usually 1 for most of simple statistics. 1 means it contains an 6042unsigned 64bit data. 6043 6044The ``offset`` field is the offset from the start of Data Block to the start of 6045the corresponding statistics data. 6046 6047The ``bucket_size`` field is used as a parameter for histogram statistics data. 6048It is only used by linear histogram statistics data, specifying the size of a 6049bucket in the unit expressed by bits 4-11 of ``flags`` together with ``exponent``. 6050 6051The ``name`` field is the name string of the statistics data. The name string 6052starts at the end of ``struct kvm_stats_desc``. The maximum length including 6053the trailing ``'\0'``, is indicated by ``name_size`` in the header. 6054 6055The Stats Data block contains an array of 64-bit values in the same order 6056as the descriptors in Descriptors block. 6057 60584.134 KVM_GET_XSAVE2 6059-------------------- 6060 6061:Capability: KVM_CAP_XSAVE2 6062:Architectures: x86 6063:Type: vcpu ioctl 6064:Parameters: struct kvm_xsave (out) 6065:Returns: 0 on success, -1 on error 6066 6067 6068:: 6069 6070 struct kvm_xsave { 6071 __u32 region[1024]; 6072 __u32 extra[0]; 6073 }; 6074 6075This ioctl would copy current vcpu's xsave struct to the userspace. It 6076copies as many bytes as are returned by KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) 6077when invoked on the vm file descriptor. The size value returned by 6078KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) will always be at least 4096. 6079Currently, it is only greater than 4096 if a dynamic feature has been 6080enabled with ``arch_prctl()``, but this may change in the future. 6081 6082The offsets of the state save areas in struct kvm_xsave follow the contents 6083of CPUID leaf 0xD on the host. 6084 60854.135 KVM_XEN_HVM_EVTCHN_SEND 6086----------------------------- 6087 6088:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_EVTCHN_SEND 6089:Architectures: x86 6090:Type: vm ioctl 6091:Parameters: struct kvm_irq_routing_xen_evtchn 6092:Returns: 0 on success, < 0 on error 6093 6094 6095:: 6096 6097 struct kvm_irq_routing_xen_evtchn { 6098 __u32 port; 6099 __u32 vcpu; 6100 __u32 priority; 6101 }; 6102 6103This ioctl injects an event channel interrupt directly to the guest vCPU. 6104 61054.136 KVM_S390_PV_CPU_COMMAND 6106----------------------------- 6107 6108:Capability: KVM_CAP_S390_PROTECTED_DUMP 6109:Architectures: s390 6110:Type: vcpu ioctl 6111:Parameters: none 6112:Returns: 0 on success, < 0 on error 6113 6114This ioctl closely mirrors `KVM_S390_PV_COMMAND` but handles requests 6115for vcpus. It re-uses the kvm_s390_pv_dmp struct and hence also shares 6116the command ids. 6117 6118**command:** 6119 6120KVM_PV_DUMP 6121 Presents an API that provides calls which facilitate dumping a vcpu 6122 of a protected VM. 6123 6124**subcommand:** 6125 6126KVM_PV_DUMP_CPU 6127 Provides encrypted dump data like register values. 6128 The length of the returned data is provided by uv_info.guest_cpu_stor_len. 6129 61304.137 KVM_S390_ZPCI_OP 6131---------------------- 6132 6133:Capability: KVM_CAP_S390_ZPCI_OP 6134:Architectures: s390 6135:Type: vm ioctl 6136:Parameters: struct kvm_s390_zpci_op (in) 6137:Returns: 0 on success, <0 on error 6138 6139Used to manage hardware-assisted virtualization features for zPCI devices. 6140 6141Parameters are specified via the following structure:: 6142 6143 struct kvm_s390_zpci_op { 6144 /* in */ 6145 __u32 fh; /* target device */ 6146 __u8 op; /* operation to perform */ 6147 __u8 pad[3]; 6148 union { 6149 /* for KVM_S390_ZPCIOP_REG_AEN */ 6150 struct { 6151 __u64 ibv; /* Guest addr of interrupt bit vector */ 6152 __u64 sb; /* Guest addr of summary bit */ 6153 __u32 flags; 6154 __u32 noi; /* Number of interrupts */ 6155 __u8 isc; /* Guest interrupt subclass */ 6156 __u8 sbo; /* Offset of guest summary bit vector */ 6157 __u16 pad; 6158 } reg_aen; 6159 __u64 reserved[8]; 6160 } u; 6161 }; 6162 6163The type of operation is specified in the "op" field. 6164KVM_S390_ZPCIOP_REG_AEN is used to register the VM for adapter event 6165notification interpretation, which will allow firmware delivery of adapter 6166events directly to the vm, with KVM providing a backup delivery mechanism; 6167KVM_S390_ZPCIOP_DEREG_AEN is used to subsequently disable interpretation of 6168adapter event notifications. 6169 6170The target zPCI function must also be specified via the "fh" field. For the 6171KVM_S390_ZPCIOP_REG_AEN operation, additional information to establish firmware 6172delivery must be provided via the "reg_aen" struct. 6173 6174The "pad" and "reserved" fields may be used for future extensions and should be 6175set to 0s by userspace. 6176 61774.138 KVM_ARM_SET_COUNTER_OFFSET 6178-------------------------------- 6179 6180:Capability: KVM_CAP_COUNTER_OFFSET 6181:Architectures: arm64 6182:Type: vm ioctl 6183:Parameters: struct kvm_arm_counter_offset (in) 6184:Returns: 0 on success, < 0 on error 6185 6186This capability indicates that userspace is able to apply a single VM-wide 6187offset to both the virtual and physical counters as viewed by the guest 6188using the KVM_ARM_SET_CNT_OFFSET ioctl and the following data structure: 6189 6190:: 6191 6192 struct kvm_arm_counter_offset { 6193 __u64 counter_offset; 6194 __u64 reserved; 6195 }; 6196 6197The offset describes a number of counter cycles that are subtracted from 6198both virtual and physical counter views (similar to the effects of the 6199CNTVOFF_EL2 and CNTPOFF_EL2 system registers, but only global). The offset 6200always applies to all vcpus (already created or created after this ioctl) 6201for this VM. 6202 6203It is userspace's responsibility to compute the offset based, for example, 6204on previous values of the guest counters. 6205 6206Any value other than 0 for the "reserved" field may result in an error 6207(-EINVAL) being returned. This ioctl can also return -EBUSY if any vcpu 6208ioctl is issued concurrently. 6209 6210Note that using this ioctl results in KVM ignoring subsequent userspace 6211writes to the CNTVCT_EL0 and CNTPCT_EL0 registers using the SET_ONE_REG 6212interface. No error will be returned, but the resulting offset will not be 6213applied. 6214 6215.. _KVM_ARM_GET_REG_WRITABLE_MASKS: 6216 62174.139 KVM_ARM_GET_REG_WRITABLE_MASKS 6218------------------------------------ 6219 6220:Capability: KVM_CAP_ARM_SUPPORTED_REG_MASK_RANGES 6221:Architectures: arm64 6222:Type: vm ioctl 6223:Parameters: struct reg_mask_range (in/out) 6224:Returns: 0 on success, < 0 on error 6225 6226 6227:: 6228 6229 #define KVM_ARM_FEATURE_ID_RANGE 0 6230 #define KVM_ARM_FEATURE_ID_RANGE_SIZE (3 * 8 * 8) 6231 6232 struct reg_mask_range { 6233 __u64 addr; /* Pointer to mask array */ 6234 __u32 range; /* Requested range */ 6235 __u32 reserved[13]; 6236 }; 6237 6238This ioctl copies the writable masks for a selected range of registers to 6239userspace. 6240 6241The ``addr`` field is a pointer to the destination array where KVM copies 6242the writable masks. 6243 6244The ``range`` field indicates the requested range of registers. 6245``KVM_CHECK_EXTENSION`` for the ``KVM_CAP_ARM_SUPPORTED_REG_MASK_RANGES`` 6246capability returns the supported ranges, expressed as a set of flags. Each 6247flag's bit index represents a possible value for the ``range`` field. 6248All other values are reserved for future use and KVM may return an error. 6249 6250The ``reserved[13]`` array is reserved for future use and should be 0, or 6251KVM may return an error. 6252 6253KVM_ARM_FEATURE_ID_RANGE (0) 6254^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6255 6256The Feature ID range is defined as the AArch64 System register space with 6257op0==3, op1=={0, 1, 3}, CRn==0, CRm=={0-7}, op2=={0-7}. 6258 6259The mask returned array pointed to by ``addr`` is indexed by the macro 6260``ARM64_FEATURE_ID_RANGE_IDX(op0, op1, crn, crm, op2)``, allowing userspace 6261to know what fields can be changed for the system register described by 6262``op0, op1, crn, crm, op2``. KVM rejects ID register values that describe a 6263superset of the features supported by the system. 6264 62654.140 KVM_SET_USER_MEMORY_REGION2 6266--------------------------------- 6267 6268:Capability: KVM_CAP_USER_MEMORY2 6269:Architectures: all 6270:Type: vm ioctl 6271:Parameters: struct kvm_userspace_memory_region2 (in) 6272:Returns: 0 on success, -1 on error 6273 6274KVM_SET_USER_MEMORY_REGION2 is an extension to KVM_SET_USER_MEMORY_REGION that 6275allows mapping guest_memfd memory into a guest. All fields shared with 6276KVM_SET_USER_MEMORY_REGION identically. Userspace can set KVM_MEM_GUEST_MEMFD 6277in flags to have KVM bind the memory region to a given guest_memfd range of 6278[guest_memfd_offset, guest_memfd_offset + memory_size]. The target guest_memfd 6279must point at a file created via KVM_CREATE_GUEST_MEMFD on the current VM, and 6280the target range must not be bound to any other memory region. All standard 6281bounds checks apply (use common sense). 6282 6283:: 6284 6285 struct kvm_userspace_memory_region2 { 6286 __u32 slot; 6287 __u32 flags; 6288 __u64 guest_phys_addr; 6289 __u64 memory_size; /* bytes */ 6290 __u64 userspace_addr; /* start of the userspace allocated memory */ 6291 __u64 guest_memfd_offset; 6292 __u32 guest_memfd; 6293 __u32 pad1; 6294 __u64 pad2[14]; 6295 }; 6296 6297A KVM_MEM_GUEST_MEMFD region _must_ have a valid guest_memfd (private memory) and 6298userspace_addr (shared memory). However, "valid" for userspace_addr simply 6299means that the address itself must be a legal userspace address. The backing 6300mapping for userspace_addr is not required to be valid/populated at the time of 6301KVM_SET_USER_MEMORY_REGION2, e.g. shared memory can be lazily mapped/allocated 6302on-demand. 6303 6304When mapping a gfn into the guest, KVM selects shared vs. private, i.e consumes 6305userspace_addr vs. guest_memfd, based on the gfn's KVM_MEMORY_ATTRIBUTE_PRIVATE 6306state. At VM creation time, all memory is shared, i.e. the PRIVATE attribute 6307is '0' for all gfns. Userspace can control whether memory is shared/private by 6308toggling KVM_MEMORY_ATTRIBUTE_PRIVATE via KVM_SET_MEMORY_ATTRIBUTES as needed. 6309 6310S390: 6311^^^^^ 6312 6313Returns -EINVAL if the VM has the KVM_VM_S390_UCONTROL flag set. 6314Returns -EINVAL if called on a protected VM. 6315 63164.141 KVM_SET_MEMORY_ATTRIBUTES 6317------------------------------- 6318 6319:Capability: KVM_CAP_MEMORY_ATTRIBUTES 6320:Architectures: x86 6321:Type: vm ioctl 6322:Parameters: struct kvm_memory_attributes (in) 6323:Returns: 0 on success, <0 on error 6324 6325KVM_SET_MEMORY_ATTRIBUTES allows userspace to set memory attributes for a range 6326of guest physical memory. 6327 6328:: 6329 6330 struct kvm_memory_attributes { 6331 __u64 address; 6332 __u64 size; 6333 __u64 attributes; 6334 __u64 flags; 6335 }; 6336 6337 #define KVM_MEMORY_ATTRIBUTE_PRIVATE (1ULL << 3) 6338 6339The address and size must be page aligned. The supported attributes can be 6340retrieved via ioctl(KVM_CHECK_EXTENSION) on KVM_CAP_MEMORY_ATTRIBUTES. If 6341executed on a VM, KVM_CAP_MEMORY_ATTRIBUTES precisely returns the attributes 6342supported by that VM. If executed at system scope, KVM_CAP_MEMORY_ATTRIBUTES 6343returns all attributes supported by KVM. The only attribute defined at this 6344time is KVM_MEMORY_ATTRIBUTE_PRIVATE, which marks the associated gfn as being 6345guest private memory. 6346 6347Note, there is no "get" API. Userspace is responsible for explicitly tracking 6348the state of a gfn/page as needed. 6349 6350The "flags" field is reserved for future extensions and must be '0'. 6351 63524.142 KVM_CREATE_GUEST_MEMFD 6353---------------------------- 6354 6355:Capability: KVM_CAP_GUEST_MEMFD 6356:Architectures: none 6357:Type: vm ioctl 6358:Parameters: struct kvm_create_guest_memfd(in) 6359:Returns: A file descriptor on success, <0 on error 6360 6361KVM_CREATE_GUEST_MEMFD creates an anonymous file and returns a file descriptor 6362that refers to it. guest_memfd files are roughly analogous to files created 6363via memfd_create(), e.g. guest_memfd files live in RAM, have volatile storage, 6364and are automatically released when the last reference is dropped. Unlike 6365"regular" memfd_create() files, guest_memfd files are bound to their owning 6366virtual machine (see below), cannot be mapped, read, or written by userspace, 6367and cannot be resized (guest_memfd files do however support PUNCH_HOLE). 6368 6369:: 6370 6371 struct kvm_create_guest_memfd { 6372 __u64 size; 6373 __u64 flags; 6374 __u64 reserved[6]; 6375 }; 6376 6377Conceptually, the inode backing a guest_memfd file represents physical memory, 6378i.e. is coupled to the virtual machine as a thing, not to a "struct kvm". The 6379file itself, which is bound to a "struct kvm", is that instance's view of the 6380underlying memory, e.g. effectively provides the translation of guest addresses 6381to host memory. This allows for use cases where multiple KVM structures are 6382used to manage a single virtual machine, e.g. when performing intrahost 6383migration of a virtual machine. 6384 6385KVM currently only supports mapping guest_memfd via KVM_SET_USER_MEMORY_REGION2, 6386and more specifically via the guest_memfd and guest_memfd_offset fields in 6387"struct kvm_userspace_memory_region2", where guest_memfd_offset is the offset 6388into the guest_memfd instance. For a given guest_memfd file, there can be at 6389most one mapping per page, i.e. binding multiple memory regions to a single 6390guest_memfd range is not allowed (any number of memory regions can be bound to 6391a single guest_memfd file, but the bound ranges must not overlap). 6392 6393See KVM_SET_USER_MEMORY_REGION2 for additional details. 6394 63954.143 KVM_PRE_FAULT_MEMORY 6396--------------------------- 6397 6398:Capability: KVM_CAP_PRE_FAULT_MEMORY 6399:Architectures: none 6400:Type: vcpu ioctl 6401:Parameters: struct kvm_pre_fault_memory (in/out) 6402:Returns: 0 if at least one page is processed, < 0 on error 6403 6404Errors: 6405 6406 ========== =============================================================== 6407 EINVAL The specified `gpa` and `size` were invalid (e.g. not 6408 page aligned, causes an overflow, or size is zero). 6409 ENOENT The specified `gpa` is outside defined memslots. 6410 EINTR An unmasked signal is pending and no page was processed. 6411 EFAULT The parameter address was invalid. 6412 EOPNOTSUPP Mapping memory for a GPA is unsupported by the 6413 hypervisor, and/or for the current vCPU state/mode. 6414 EIO unexpected error conditions (also causes a WARN) 6415 ========== =============================================================== 6416 6417:: 6418 6419 struct kvm_pre_fault_memory { 6420 /* in/out */ 6421 __u64 gpa; 6422 __u64 size; 6423 /* in */ 6424 __u64 flags; 6425 __u64 padding[5]; 6426 }; 6427 6428KVM_PRE_FAULT_MEMORY populates KVM's stage-2 page tables used to map memory 6429for the current vCPU state. KVM maps memory as if the vCPU generated a 6430stage-2 read page fault, e.g. faults in memory as needed, but doesn't break 6431CoW. However, KVM does not mark any newly created stage-2 PTE as Accessed. 6432 6433In the case of confidential VM types where there is an initial set up of 6434private guest memory before the guest is 'finalized'/measured, this ioctl 6435should only be issued after completing all the necessary setup to put the 6436guest into a 'finalized' state so that the above semantics can be reliably 6437ensured. 6438 6439In some cases, multiple vCPUs might share the page tables. In this 6440case, the ioctl can be called in parallel. 6441 6442When the ioctl returns, the input values are updated to point to the 6443remaining range. If `size` > 0 on return, the caller can just issue 6444the ioctl again with the same `struct kvm_map_memory` argument. 6445 6446Shadow page tables cannot support this ioctl because they 6447are indexed by virtual address or nested guest physical address. 6448Calling this ioctl when the guest is using shadow page tables (for 6449example because it is running a nested guest with nested page tables) 6450will fail with `EOPNOTSUPP` even if `KVM_CHECK_EXTENSION` reports 6451the capability to be present. 6452 6453`flags` must currently be zero. 6454 6455 6456.. _kvm_run: 6457 64585. The kvm_run structure 6459======================== 6460 6461Application code obtains a pointer to the kvm_run structure by 6462mmap()ing a vcpu fd. From that point, application code can control 6463execution by changing fields in kvm_run prior to calling the KVM_RUN 6464ioctl, and obtain information about the reason KVM_RUN returned by 6465looking up structure members. 6466 6467:: 6468 6469 struct kvm_run { 6470 /* in */ 6471 __u8 request_interrupt_window; 6472 6473Request that KVM_RUN return when it becomes possible to inject external 6474interrupts into the guest. Useful in conjunction with KVM_INTERRUPT. 6475 6476:: 6477 6478 __u8 immediate_exit; 6479 6480This field is polled once when KVM_RUN starts; if non-zero, KVM_RUN 6481exits immediately, returning -EINTR. In the common scenario where a 6482signal is used to "kick" a VCPU out of KVM_RUN, this field can be used 6483to avoid usage of KVM_SET_SIGNAL_MASK, which has worse scalability. 6484Rather than blocking the signal outside KVM_RUN, userspace can set up 6485a signal handler that sets run->immediate_exit to a non-zero value. 6486 6487This field is ignored if KVM_CAP_IMMEDIATE_EXIT is not available. 6488 6489:: 6490 6491 __u8 padding1[6]; 6492 6493 /* out */ 6494 __u32 exit_reason; 6495 6496When KVM_RUN has returned successfully (return value 0), this informs 6497application code why KVM_RUN has returned. Allowable values for this 6498field are detailed below. 6499 6500:: 6501 6502 __u8 ready_for_interrupt_injection; 6503 6504If request_interrupt_window has been specified, this field indicates 6505an interrupt can be injected now with KVM_INTERRUPT. 6506 6507:: 6508 6509 __u8 if_flag; 6510 6511The value of the current interrupt flag. Only valid if in-kernel 6512local APIC is not used. 6513 6514:: 6515 6516 __u16 flags; 6517 6518More architecture-specific flags detailing state of the VCPU that may 6519affect the device's behavior. Current defined flags:: 6520 6521 /* x86, set if the VCPU is in system management mode */ 6522 #define KVM_RUN_X86_SMM (1 << 0) 6523 /* x86, set if bus lock detected in VM */ 6524 #define KVM_RUN_X86_BUS_LOCK (1 << 1) 6525 /* x86, set if the VCPU is executing a nested (L2) guest */ 6526 #define KVM_RUN_X86_GUEST_MODE (1 << 2) 6527 6528 /* arm64, set for KVM_EXIT_DEBUG */ 6529 #define KVM_DEBUG_ARCH_HSR_HIGH_VALID (1 << 0) 6530 6531:: 6532 6533 /* in (pre_kvm_run), out (post_kvm_run) */ 6534 __u64 cr8; 6535 6536The value of the cr8 register. Only valid if in-kernel local APIC is 6537not used. Both input and output. 6538 6539:: 6540 6541 __u64 apic_base; 6542 6543The value of the APIC BASE msr. Only valid if in-kernel local 6544APIC is not used. Both input and output. 6545 6546:: 6547 6548 union { 6549 /* KVM_EXIT_UNKNOWN */ 6550 struct { 6551 __u64 hardware_exit_reason; 6552 } hw; 6553 6554If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown 6555reasons. Further architecture-specific information is available in 6556hardware_exit_reason. 6557 6558:: 6559 6560 /* KVM_EXIT_FAIL_ENTRY */ 6561 struct { 6562 __u64 hardware_entry_failure_reason; 6563 __u32 cpu; /* if KVM_LAST_CPU */ 6564 } fail_entry; 6565 6566If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due 6567to unknown reasons. Further architecture-specific information is 6568available in hardware_entry_failure_reason. 6569 6570:: 6571 6572 /* KVM_EXIT_EXCEPTION */ 6573 struct { 6574 __u32 exception; 6575 __u32 error_code; 6576 } ex; 6577 6578Unused. 6579 6580:: 6581 6582 /* KVM_EXIT_IO */ 6583 struct { 6584 #define KVM_EXIT_IO_IN 0 6585 #define KVM_EXIT_IO_OUT 1 6586 __u8 direction; 6587 __u8 size; /* bytes */ 6588 __u16 port; 6589 __u32 count; 6590 __u64 data_offset; /* relative to kvm_run start */ 6591 } io; 6592 6593If exit_reason is KVM_EXIT_IO, then the vcpu has 6594executed a port I/O instruction which could not be satisfied by kvm. 6595data_offset describes where the data is located (KVM_EXIT_IO_OUT) or 6596where kvm expects application code to place the data for the next 6597KVM_RUN invocation (KVM_EXIT_IO_IN). Data format is a packed array. 6598 6599:: 6600 6601 /* KVM_EXIT_DEBUG */ 6602 struct { 6603 struct kvm_debug_exit_arch arch; 6604 } debug; 6605 6606If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event 6607for which architecture specific information is returned. 6608 6609:: 6610 6611 /* KVM_EXIT_MMIO */ 6612 struct { 6613 __u64 phys_addr; 6614 __u8 data[8]; 6615 __u32 len; 6616 __u8 is_write; 6617 } mmio; 6618 6619If exit_reason is KVM_EXIT_MMIO, then the vcpu has 6620executed a memory-mapped I/O instruction which could not be satisfied 6621by kvm. The 'data' member contains the written data if 'is_write' is 6622true, and should be filled by application code otherwise. 6623 6624The 'data' member contains, in its first 'len' bytes, the value as it would 6625appear if the VCPU performed a load or store of the appropriate width directly 6626to the byte array. 6627 6628.. note:: 6629 6630 For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR, KVM_EXIT_XEN, 6631 KVM_EXIT_EPR, KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR the corresponding 6632 operations are complete (and guest state is consistent) only after userspace 6633 has re-entered the kernel with KVM_RUN. The kernel side will first finish 6634 incomplete operations and then check for pending signals. 6635 6636 The pending state of the operation is not preserved in state which is 6637 visible to userspace, thus userspace should ensure that the operation is 6638 completed before performing a live migration. Userspace can re-enter the 6639 guest with an unmasked signal pending or with the immediate_exit field set 6640 to complete pending operations without allowing any further instructions 6641 to be executed. 6642 6643:: 6644 6645 /* KVM_EXIT_HYPERCALL */ 6646 struct { 6647 __u64 nr; 6648 __u64 args[6]; 6649 __u64 ret; 6650 __u64 flags; 6651 } hypercall; 6652 6653 6654It is strongly recommended that userspace use ``KVM_EXIT_IO`` (x86) or 6655``KVM_EXIT_MMIO`` (all except s390) to implement functionality that 6656requires a guest to interact with host userspace. 6657 6658.. note:: KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO. 6659 6660For arm64: 6661---------- 6662 6663SMCCC exits can be enabled depending on the configuration of the SMCCC 6664filter. See the Documentation/virt/kvm/devices/vm.rst 6665``KVM_ARM_SMCCC_FILTER`` for more details. 6666 6667``nr`` contains the function ID of the guest's SMCCC call. Userspace is 6668expected to use the ``KVM_GET_ONE_REG`` ioctl to retrieve the call 6669parameters from the vCPU's GPRs. 6670 6671Definition of ``flags``: 6672 - ``KVM_HYPERCALL_EXIT_SMC``: Indicates that the guest used the SMC 6673 conduit to initiate the SMCCC call. If this bit is 0 then the guest 6674 used the HVC conduit for the SMCCC call. 6675 6676 - ``KVM_HYPERCALL_EXIT_16BIT``: Indicates that the guest used a 16bit 6677 instruction to initiate the SMCCC call. If this bit is 0 then the 6678 guest used a 32bit instruction. An AArch64 guest always has this 6679 bit set to 0. 6680 6681At the point of exit, PC points to the instruction immediately following 6682the trapping instruction. 6683 6684:: 6685 6686 /* KVM_EXIT_TPR_ACCESS */ 6687 struct { 6688 __u64 rip; 6689 __u32 is_write; 6690 __u32 pad; 6691 } tpr_access; 6692 6693To be documented (KVM_TPR_ACCESS_REPORTING). 6694 6695:: 6696 6697 /* KVM_EXIT_S390_SIEIC */ 6698 struct { 6699 __u8 icptcode; 6700 __u64 mask; /* psw upper half */ 6701 __u64 addr; /* psw lower half */ 6702 __u16 ipa; 6703 __u32 ipb; 6704 } s390_sieic; 6705 6706s390 specific. 6707 6708:: 6709 6710 /* KVM_EXIT_S390_RESET */ 6711 #define KVM_S390_RESET_POR 1 6712 #define KVM_S390_RESET_CLEAR 2 6713 #define KVM_S390_RESET_SUBSYSTEM 4 6714 #define KVM_S390_RESET_CPU_INIT 8 6715 #define KVM_S390_RESET_IPL 16 6716 __u64 s390_reset_flags; 6717 6718s390 specific. 6719 6720:: 6721 6722 /* KVM_EXIT_S390_UCONTROL */ 6723 struct { 6724 __u64 trans_exc_code; 6725 __u32 pgm_code; 6726 } s390_ucontrol; 6727 6728s390 specific. A page fault has occurred for a user controlled virtual 6729machine (KVM_VM_S390_UNCONTROL) on its host page table that cannot be 6730resolved by the kernel. 6731The program code and the translation exception code that were placed 6732in the cpu's lowcore are presented here as defined by the z Architecture 6733Principles of Operation Book in the Chapter for Dynamic Address Translation 6734(DAT) 6735 6736:: 6737 6738 /* KVM_EXIT_DCR */ 6739 struct { 6740 __u32 dcrn; 6741 __u32 data; 6742 __u8 is_write; 6743 } dcr; 6744 6745Deprecated - was used for 440 KVM. 6746 6747:: 6748 6749 /* KVM_EXIT_OSI */ 6750 struct { 6751 __u64 gprs[32]; 6752 } osi; 6753 6754MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch 6755hypercalls and exit with this exit struct that contains all the guest gprs. 6756 6757If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall. 6758Userspace can now handle the hypercall and when it's done modify the gprs as 6759necessary. Upon guest entry all guest GPRs will then be replaced by the values 6760in this struct. 6761 6762:: 6763 6764 /* KVM_EXIT_PAPR_HCALL */ 6765 struct { 6766 __u64 nr; 6767 __u64 ret; 6768 __u64 args[9]; 6769 } papr_hcall; 6770 6771This is used on 64-bit PowerPC when emulating a pSeries partition, 6772e.g. with the 'pseries' machine type in qemu. It occurs when the 6773guest does a hypercall using the 'sc 1' instruction. The 'nr' field 6774contains the hypercall number (from the guest R3), and 'args' contains 6775the arguments (from the guest R4 - R12). Userspace should put the 6776return code in 'ret' and any extra returned values in args[]. 6777The possible hypercalls are defined in the Power Architecture Platform 6778Requirements (PAPR) document available from www.power.org (free 6779developer registration required to access it). 6780 6781:: 6782 6783 /* KVM_EXIT_S390_TSCH */ 6784 struct { 6785 __u16 subchannel_id; 6786 __u16 subchannel_nr; 6787 __u32 io_int_parm; 6788 __u32 io_int_word; 6789 __u32 ipb; 6790 __u8 dequeued; 6791 } s390_tsch; 6792 6793s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled 6794and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O 6795interrupt for the target subchannel has been dequeued and subchannel_id, 6796subchannel_nr, io_int_parm and io_int_word contain the parameters for that 6797interrupt. ipb is needed for instruction parameter decoding. 6798 6799:: 6800 6801 /* KVM_EXIT_EPR */ 6802 struct { 6803 __u32 epr; 6804 } epr; 6805 6806On FSL BookE PowerPC chips, the interrupt controller has a fast patch 6807interrupt acknowledge path to the core. When the core successfully 6808delivers an interrupt, it automatically populates the EPR register with 6809the interrupt vector number and acknowledges the interrupt inside 6810the interrupt controller. 6811 6812In case the interrupt controller lives in user space, we need to do 6813the interrupt acknowledge cycle through it to fetch the next to be 6814delivered interrupt vector using this exit. 6815 6816It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an 6817external interrupt has just been delivered into the guest. User space 6818should put the acknowledged interrupt vector into the 'epr' field. 6819 6820:: 6821 6822 /* KVM_EXIT_SYSTEM_EVENT */ 6823 struct { 6824 #define KVM_SYSTEM_EVENT_SHUTDOWN 1 6825 #define KVM_SYSTEM_EVENT_RESET 2 6826 #define KVM_SYSTEM_EVENT_CRASH 3 6827 #define KVM_SYSTEM_EVENT_WAKEUP 4 6828 #define KVM_SYSTEM_EVENT_SUSPEND 5 6829 #define KVM_SYSTEM_EVENT_SEV_TERM 6 6830 __u32 type; 6831 __u32 ndata; 6832 __u64 data[16]; 6833 } system_event; 6834 6835If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered 6836a system-level event using some architecture specific mechanism (hypercall 6837or some special instruction). In case of ARM64, this is triggered using 6838HVC instruction based PSCI call from the vcpu. 6839 6840The 'type' field describes the system-level event type. 6841Valid values for 'type' are: 6842 6843 - KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the 6844 VM. Userspace is not obliged to honour this, and if it does honour 6845 this does not need to destroy the VM synchronously (ie it may call 6846 KVM_RUN again before shutdown finally occurs). 6847 - KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM. 6848 As with SHUTDOWN, userspace can choose to ignore the request, or 6849 to schedule the reset to occur in the future and may call KVM_RUN again. 6850 - KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest 6851 has requested a crash condition maintenance. Userspace can choose 6852 to ignore the request, or to gather VM memory core dump and/or 6853 reset/shutdown of the VM. 6854 - KVM_SYSTEM_EVENT_SEV_TERM -- an AMD SEV guest requested termination. 6855 The guest physical address of the guest's GHCB is stored in `data[0]`. 6856 - KVM_SYSTEM_EVENT_WAKEUP -- the exiting vCPU is in a suspended state and 6857 KVM has recognized a wakeup event. Userspace may honor this event by 6858 marking the exiting vCPU as runnable, or deny it and call KVM_RUN again. 6859 - KVM_SYSTEM_EVENT_SUSPEND -- the guest has requested a suspension of 6860 the VM. 6861 6862If KVM_CAP_SYSTEM_EVENT_DATA is present, the 'data' field can contain 6863architecture specific information for the system-level event. Only 6864the first `ndata` items (possibly zero) of the data array are valid. 6865 6866 - for arm64, data[0] is set to KVM_SYSTEM_EVENT_RESET_FLAG_PSCI_RESET2 if 6867 the guest issued a SYSTEM_RESET2 call according to v1.1 of the PSCI 6868 specification. 6869 6870 - for arm64, data[0] is set to KVM_SYSTEM_EVENT_SHUTDOWN_FLAG_PSCI_OFF2 6871 if the guest issued a SYSTEM_OFF2 call according to v1.3 of the PSCI 6872 specification. 6873 6874 - for RISC-V, data[0] is set to the value of the second argument of the 6875 ``sbi_system_reset`` call. 6876 6877Previous versions of Linux defined a `flags` member in this struct. The 6878field is now aliased to `data[0]`. Userspace can assume that it is only 6879written if ndata is greater than 0. 6880 6881For arm/arm64: 6882-------------- 6883 6884KVM_SYSTEM_EVENT_SUSPEND exits are enabled with the 6885KVM_CAP_ARM_SYSTEM_SUSPEND VM capability. If a guest invokes the PSCI 6886SYSTEM_SUSPEND function, KVM will exit to userspace with this event 6887type. 6888 6889It is the sole responsibility of userspace to implement the PSCI 6890SYSTEM_SUSPEND call according to ARM DEN0022D.b 5.19 "SYSTEM_SUSPEND". 6891KVM does not change the vCPU's state before exiting to userspace, so 6892the call parameters are left in-place in the vCPU registers. 6893 6894Userspace is _required_ to take action for such an exit. It must 6895either: 6896 6897 - Honor the guest request to suspend the VM. Userspace can request 6898 in-kernel emulation of suspension by setting the calling vCPU's 6899 state to KVM_MP_STATE_SUSPENDED. Userspace must configure the vCPU's 6900 state according to the parameters passed to the PSCI function when 6901 the calling vCPU is resumed. See ARM DEN0022D.b 5.19.1 "Intended use" 6902 for details on the function parameters. 6903 6904 - Deny the guest request to suspend the VM. See ARM DEN0022D.b 5.19.2 6905 "Caller responsibilities" for possible return values. 6906 6907Hibernation using the PSCI SYSTEM_OFF2 call is enabled when PSCI v1.3 6908is enabled. If a guest invokes the PSCI SYSTEM_OFF2 function, KVM will 6909exit to userspace with the KVM_SYSTEM_EVENT_SHUTDOWN event type and with 6910data[0] set to KVM_SYSTEM_EVENT_SHUTDOWN_FLAG_PSCI_OFF2. The only 6911supported hibernate type for the SYSTEM_OFF2 function is HIBERNATE_OFF. 6912 6913:: 6914 6915 /* KVM_EXIT_IOAPIC_EOI */ 6916 struct { 6917 __u8 vector; 6918 } eoi; 6919 6920Indicates that the VCPU's in-kernel local APIC received an EOI for a 6921level-triggered IOAPIC interrupt. This exit only triggers when the 6922IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled); 6923the userspace IOAPIC should process the EOI and retrigger the interrupt if 6924it is still asserted. Vector is the LAPIC interrupt vector for which the 6925EOI was received. 6926 6927:: 6928 6929 struct kvm_hyperv_exit { 6930 #define KVM_EXIT_HYPERV_SYNIC 1 6931 #define KVM_EXIT_HYPERV_HCALL 2 6932 #define KVM_EXIT_HYPERV_SYNDBG 3 6933 __u32 type; 6934 __u32 pad1; 6935 union { 6936 struct { 6937 __u32 msr; 6938 __u32 pad2; 6939 __u64 control; 6940 __u64 evt_page; 6941 __u64 msg_page; 6942 } synic; 6943 struct { 6944 __u64 input; 6945 __u64 result; 6946 __u64 params[2]; 6947 } hcall; 6948 struct { 6949 __u32 msr; 6950 __u32 pad2; 6951 __u64 control; 6952 __u64 status; 6953 __u64 send_page; 6954 __u64 recv_page; 6955 __u64 pending_page; 6956 } syndbg; 6957 } u; 6958 }; 6959 /* KVM_EXIT_HYPERV */ 6960 struct kvm_hyperv_exit hyperv; 6961 6962Indicates that the VCPU exits into userspace to process some tasks 6963related to Hyper-V emulation. 6964 6965Valid values for 'type' are: 6966 6967 - KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about 6968 6969Hyper-V SynIC state change. Notification is used to remap SynIC 6970event/message pages and to enable/disable SynIC messages/events processing 6971in userspace. 6972 6973 - KVM_EXIT_HYPERV_SYNDBG -- synchronously notify user-space about 6974 6975Hyper-V Synthetic debugger state change. Notification is used to either update 6976the pending_page location or to send a control command (send the buffer located 6977in send_page or recv a buffer to recv_page). 6978 6979:: 6980 6981 /* KVM_EXIT_ARM_NISV */ 6982 struct { 6983 __u64 esr_iss; 6984 __u64 fault_ipa; 6985 } arm_nisv; 6986 6987Used on arm64 systems. If a guest accesses memory not in a memslot, 6988KVM will typically return to userspace and ask it to do MMIO emulation on its 6989behalf. However, for certain classes of instructions, no instruction decode 6990(direction, length of memory access) is provided, and fetching and decoding 6991the instruction from the VM is overly complicated to live in the kernel. 6992 6993Historically, when this situation occurred, KVM would print a warning and kill 6994the VM. KVM assumed that if the guest accessed non-memslot memory, it was 6995trying to do I/O, which just couldn't be emulated, and the warning message was 6996phrased accordingly. However, what happened more often was that a guest bug 6997caused access outside the guest memory areas which should lead to a more 6998meaningful warning message and an external abort in the guest, if the access 6999did not fall within an I/O window. 7000 7001Userspace implementations can query for KVM_CAP_ARM_NISV_TO_USER, and enable 7002this capability at VM creation. Once this is done, these types of errors will 7003instead return to userspace with KVM_EXIT_ARM_NISV, with the valid bits from 7004the ESR_EL2 in the esr_iss field, and the faulting IPA in the fault_ipa field. 7005Userspace can either fix up the access if it's actually an I/O access by 7006decoding the instruction from guest memory (if it's very brave) and continue 7007executing the guest, or it can decide to suspend, dump, or restart the guest. 7008 7009Note that KVM does not skip the faulting instruction as it does for 7010KVM_EXIT_MMIO, but userspace has to emulate any change to the processing state 7011if it decides to decode and emulate the instruction. 7012 7013This feature isn't available to protected VMs, as userspace does not 7014have access to the state that is required to perform the emulation. 7015Instead, a data abort exception is directly injected in the guest. 7016Note that although KVM_CAP_ARM_NISV_TO_USER will be reported if 7017queried outside of a protected VM context, the feature will not be 7018exposed if queried on a protected VM file descriptor. 7019 7020:: 7021 7022 /* KVM_EXIT_X86_RDMSR / KVM_EXIT_X86_WRMSR */ 7023 struct { 7024 __u8 error; /* user -> kernel */ 7025 __u8 pad[7]; 7026 __u32 reason; /* kernel -> user */ 7027 __u32 index; /* kernel -> user */ 7028 __u64 data; /* kernel <-> user */ 7029 } msr; 7030 7031Used on x86 systems. When the VM capability KVM_CAP_X86_USER_SPACE_MSR is 7032enabled, MSR accesses to registers that would invoke a #GP by KVM kernel code 7033may instead trigger a KVM_EXIT_X86_RDMSR exit for reads and KVM_EXIT_X86_WRMSR 7034exit for writes. 7035 7036The "reason" field specifies why the MSR interception occurred. Userspace will 7037only receive MSR exits when a particular reason was requested during through 7038ENABLE_CAP. Currently valid exit reasons are: 7039 7040============================ ======================================== 7041 KVM_MSR_EXIT_REASON_UNKNOWN access to MSR that is unknown to KVM 7042 KVM_MSR_EXIT_REASON_INVAL access to invalid MSRs or reserved bits 7043 KVM_MSR_EXIT_REASON_FILTER access blocked by KVM_X86_SET_MSR_FILTER 7044============================ ======================================== 7045 7046For KVM_EXIT_X86_RDMSR, the "index" field tells userspace which MSR the guest 7047wants to read. To respond to this request with a successful read, userspace 7048writes the respective data into the "data" field and must continue guest 7049execution to ensure the read data is transferred into guest register state. 7050 7051If the RDMSR request was unsuccessful, userspace indicates that with a "1" in 7052the "error" field. This will inject a #GP into the guest when the VCPU is 7053executed again. 7054 7055For KVM_EXIT_X86_WRMSR, the "index" field tells userspace which MSR the guest 7056wants to write. Once finished processing the event, userspace must continue 7057vCPU execution. If the MSR write was unsuccessful, userspace also sets the 7058"error" field to "1". 7059 7060See KVM_X86_SET_MSR_FILTER for details on the interaction with MSR filtering. 7061 7062:: 7063 7064 7065 struct kvm_xen_exit { 7066 #define KVM_EXIT_XEN_HCALL 1 7067 __u32 type; 7068 union { 7069 struct { 7070 __u32 longmode; 7071 __u32 cpl; 7072 __u64 input; 7073 __u64 result; 7074 __u64 params[6]; 7075 } hcall; 7076 } u; 7077 }; 7078 /* KVM_EXIT_XEN */ 7079 struct kvm_hyperv_exit xen; 7080 7081Indicates that the VCPU exits into userspace to process some tasks 7082related to Xen emulation. 7083 7084Valid values for 'type' are: 7085 7086 - KVM_EXIT_XEN_HCALL -- synchronously notify user-space about Xen hypercall. 7087 Userspace is expected to place the hypercall result into the appropriate 7088 field before invoking KVM_RUN again. 7089 7090:: 7091 7092 /* KVM_EXIT_RISCV_SBI */ 7093 struct { 7094 unsigned long extension_id; 7095 unsigned long function_id; 7096 unsigned long args[6]; 7097 unsigned long ret[2]; 7098 } riscv_sbi; 7099 7100If exit reason is KVM_EXIT_RISCV_SBI then it indicates that the VCPU has 7101done a SBI call which is not handled by KVM RISC-V kernel module. The details 7102of the SBI call are available in 'riscv_sbi' member of kvm_run structure. The 7103'extension_id' field of 'riscv_sbi' represents SBI extension ID whereas the 7104'function_id' field represents function ID of given SBI extension. The 'args' 7105array field of 'riscv_sbi' represents parameters for the SBI call and 'ret' 7106array field represents return values. The userspace should update the return 7107values of SBI call before resuming the VCPU. For more details on RISC-V SBI 7108spec refer, https://github.com/riscv/riscv-sbi-doc. 7109 7110:: 7111 7112 /* KVM_EXIT_MEMORY_FAULT */ 7113 struct { 7114 #define KVM_MEMORY_EXIT_FLAG_PRIVATE (1ULL << 3) 7115 __u64 flags; 7116 __u64 gpa; 7117 __u64 size; 7118 } memory_fault; 7119 7120KVM_EXIT_MEMORY_FAULT indicates the vCPU has encountered a memory fault that 7121could not be resolved by KVM. The 'gpa' and 'size' (in bytes) describe the 7122guest physical address range [gpa, gpa + size) of the fault. The 'flags' field 7123describes properties of the faulting access that are likely pertinent: 7124 7125 - KVM_MEMORY_EXIT_FLAG_PRIVATE - When set, indicates the memory fault occurred 7126 on a private memory access. When clear, indicates the fault occurred on a 7127 shared access. 7128 7129Note! KVM_EXIT_MEMORY_FAULT is unique among all KVM exit reasons in that it 7130accompanies a return code of '-1', not '0'! errno will always be set to EFAULT 7131or EHWPOISON when KVM exits with KVM_EXIT_MEMORY_FAULT, userspace should assume 7132kvm_run.exit_reason is stale/undefined for all other error numbers. 7133 7134:: 7135 7136 /* KVM_EXIT_NOTIFY */ 7137 struct { 7138 #define KVM_NOTIFY_CONTEXT_INVALID (1 << 0) 7139 __u32 flags; 7140 } notify; 7141 7142Used on x86 systems. When the VM capability KVM_CAP_X86_NOTIFY_VMEXIT is 7143enabled, a VM exit generated if no event window occurs in VM non-root mode 7144for a specified amount of time. Once KVM_X86_NOTIFY_VMEXIT_USER is set when 7145enabling the cap, it would exit to userspace with the exit reason 7146KVM_EXIT_NOTIFY for further handling. The "flags" field contains more 7147detailed info. 7148 7149The valid value for 'flags' is: 7150 7151 - KVM_NOTIFY_CONTEXT_INVALID -- the VM context is corrupted and not valid 7152 in VMCS. It would run into unknown result if resume the target VM. 7153 7154:: 7155 7156 /* Fix the size of the union. */ 7157 char padding[256]; 7158 }; 7159 7160 /* 7161 * shared registers between kvm and userspace. 7162 * kvm_valid_regs specifies the register classes set by the host 7163 * kvm_dirty_regs specified the register classes dirtied by userspace 7164 * struct kvm_sync_regs is architecture specific, as well as the 7165 * bits for kvm_valid_regs and kvm_dirty_regs 7166 */ 7167 __u64 kvm_valid_regs; 7168 __u64 kvm_dirty_regs; 7169 union { 7170 struct kvm_sync_regs regs; 7171 char padding[SYNC_REGS_SIZE_BYTES]; 7172 } s; 7173 7174If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access 7175certain guest registers without having to call SET/GET_*REGS. Thus we can 7176avoid some system call overhead if userspace has to handle the exit. 7177Userspace can query the validity of the structure by checking 7178kvm_valid_regs for specific bits. These bits are architecture specific 7179and usually define the validity of a groups of registers. (e.g. one bit 7180for general purpose registers) 7181 7182Please note that the kernel is allowed to use the kvm_run structure as the 7183primary storage for certain register types. Therefore, the kernel may use the 7184values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set. 7185 7186 7187.. _cap_enable: 7188 71896. Capabilities that can be enabled on vCPUs 7190============================================ 7191 7192There are certain capabilities that change the behavior of the virtual CPU or 7193the virtual machine when enabled. To enable them, please see 7194:ref:`KVM_ENABLE_CAP`. 7195 7196Below you can find a list of capabilities and what their effect on the vCPU or 7197the virtual machine is when enabling them. 7198 7199The following information is provided along with the description: 7200 7201 Architectures: 7202 which instruction set architectures provide this ioctl. 7203 x86 includes both i386 and x86_64. 7204 7205 Target: 7206 whether this is a per-vcpu or per-vm capability. 7207 7208 Parameters: 7209 what parameters are accepted by the capability. 7210 7211 Returns: 7212 the return value. General error numbers (EBADF, ENOMEM, EINVAL) 7213 are not detailed, but errors with specific meanings are. 7214 7215 72166.1 KVM_CAP_PPC_OSI 7217------------------- 7218 7219:Architectures: ppc 7220:Target: vcpu 7221:Parameters: none 7222:Returns: 0 on success; -1 on error 7223 7224This capability enables interception of OSI hypercalls that otherwise would 7225be treated as normal system calls to be injected into the guest. OSI hypercalls 7226were invented by Mac-on-Linux to have a standardized communication mechanism 7227between the guest and the host. 7228 7229When this capability is enabled, KVM_EXIT_OSI can occur. 7230 7231 72326.2 KVM_CAP_PPC_PAPR 7233-------------------- 7234 7235:Architectures: ppc 7236:Target: vcpu 7237:Parameters: none 7238:Returns: 0 on success; -1 on error 7239 7240This capability enables interception of PAPR hypercalls. PAPR hypercalls are 7241done using the hypercall instruction "sc 1". 7242 7243It also sets the guest privilege level to "supervisor" mode. Usually the guest 7244runs in "hypervisor" privilege mode with a few missing features. 7245 7246In addition to the above, it changes the semantics of SDR1. In this mode, the 7247HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the 7248HTAB invisible to the guest. 7249 7250When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur. 7251 7252 72536.3 KVM_CAP_SW_TLB 7254------------------ 7255 7256:Architectures: ppc 7257:Target: vcpu 7258:Parameters: args[0] is the address of a struct kvm_config_tlb 7259:Returns: 0 on success; -1 on error 7260 7261:: 7262 7263 struct kvm_config_tlb { 7264 __u64 params; 7265 __u64 array; 7266 __u32 mmu_type; 7267 __u32 array_len; 7268 }; 7269 7270Configures the virtual CPU's TLB array, establishing a shared memory area 7271between userspace and KVM. The "params" and "array" fields are userspace 7272addresses of mmu-type-specific data structures. The "array_len" field is an 7273safety mechanism, and should be set to the size in bytes of the memory that 7274userspace has reserved for the array. It must be at least the size dictated 7275by "mmu_type" and "params". 7276 7277While KVM_RUN is active, the shared region is under control of KVM. Its 7278contents are undefined, and any modification by userspace results in 7279boundedly undefined behavior. 7280 7281On return from KVM_RUN, the shared region will reflect the current state of 7282the guest's TLB. If userspace makes any changes, it must call KVM_DIRTY_TLB 7283to tell KVM which entries have been changed, prior to calling KVM_RUN again 7284on this vcpu. 7285 7286For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV: 7287 7288 - The "params" field is of type "struct kvm_book3e_206_tlb_params". 7289 - The "array" field points to an array of type "struct 7290 kvm_book3e_206_tlb_entry". 7291 - The array consists of all entries in the first TLB, followed by all 7292 entries in the second TLB. 7293 - Within a TLB, entries are ordered first by increasing set number. Within a 7294 set, entries are ordered by way (increasing ESEL). 7295 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1) 7296 where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value. 7297 - The tsize field of mas1 shall be set to 4K on TLB0, even though the 7298 hardware ignores this value for TLB0. 7299 73006.4 KVM_CAP_S390_CSS_SUPPORT 7301---------------------------- 7302 7303:Architectures: s390 7304:Target: vcpu 7305:Parameters: none 7306:Returns: 0 on success; -1 on error 7307 7308This capability enables support for handling of channel I/O instructions. 7309 7310TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are 7311handled in-kernel, while the other I/O instructions are passed to userspace. 7312 7313When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST 7314SUBCHANNEL intercepts. 7315 7316Note that even though this capability is enabled per-vcpu, the complete 7317virtual machine is affected. 7318 73196.5 KVM_CAP_PPC_EPR 7320------------------- 7321 7322:Architectures: ppc 7323:Target: vcpu 7324:Parameters: args[0] defines whether the proxy facility is active 7325:Returns: 0 on success; -1 on error 7326 7327This capability enables or disables the delivery of interrupts through the 7328external proxy facility. 7329 7330When enabled (args[0] != 0), every time the guest gets an external interrupt 7331delivered, it automatically exits into user space with a KVM_EXIT_EPR exit 7332to receive the topmost interrupt vector. 7333 7334When disabled (args[0] == 0), behavior is as if this facility is unsupported. 7335 7336When this capability is enabled, KVM_EXIT_EPR can occur. 7337 73386.6 KVM_CAP_IRQ_MPIC 7339-------------------- 7340 7341:Architectures: ppc 7342:Parameters: args[0] is the MPIC device fd; 7343 args[1] is the MPIC CPU number for this vcpu 7344 7345This capability connects the vcpu to an in-kernel MPIC device. 7346 73476.7 KVM_CAP_IRQ_XICS 7348-------------------- 7349 7350:Architectures: ppc 7351:Target: vcpu 7352:Parameters: args[0] is the XICS device fd; 7353 args[1] is the XICS CPU number (server ID) for this vcpu 7354 7355This capability connects the vcpu to an in-kernel XICS device. 7356 73576.8 KVM_CAP_S390_IRQCHIP 7358------------------------ 7359 7360:Architectures: s390 7361:Target: vm 7362:Parameters: none 7363 7364This capability enables the in-kernel irqchip for s390. Please refer to 7365"4.24 KVM_CREATE_IRQCHIP" for details. 7366 73676.9 KVM_CAP_MIPS_FPU 7368-------------------- 7369 7370:Architectures: mips 7371:Target: vcpu 7372:Parameters: args[0] is reserved for future use (should be 0). 7373 7374This capability allows the use of the host Floating Point Unit by the guest. It 7375allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is 7376done the ``KVM_REG_MIPS_FPR_*`` and ``KVM_REG_MIPS_FCR_*`` registers can be 7377accessed (depending on the current guest FPU register mode), and the Status.FR, 7378Config5.FRE bits are accessible via the KVM API and also from the guest, 7379depending on them being supported by the FPU. 7380 73816.10 KVM_CAP_MIPS_MSA 7382--------------------- 7383 7384:Architectures: mips 7385:Target: vcpu 7386:Parameters: args[0] is reserved for future use (should be 0). 7387 7388This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest. 7389It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest. 7390Once this is done the ``KVM_REG_MIPS_VEC_*`` and ``KVM_REG_MIPS_MSA_*`` 7391registers can be accessed, and the Config5.MSAEn bit is accessible via the 7392KVM API and also from the guest. 7393 73946.74 KVM_CAP_SYNC_REGS 7395---------------------- 7396 7397:Architectures: s390, x86 7398:Target: s390: always enabled, x86: vcpu 7399:Parameters: none 7400:Returns: x86: KVM_CHECK_EXTENSION returns a bit-array indicating which register 7401 sets are supported 7402 (bitfields defined in arch/x86/include/uapi/asm/kvm.h). 7403 7404As described above in the kvm_sync_regs struct info in section :ref:`kvm_run`, 7405KVM_CAP_SYNC_REGS "allow[s] userspace to access certain guest registers 7406without having to call SET/GET_*REGS". This reduces overhead by eliminating 7407repeated ioctl calls for setting and/or getting register values. This is 7408particularly important when userspace is making synchronous guest state 7409modifications, e.g. when emulating and/or intercepting instructions in 7410userspace. 7411 7412For s390 specifics, please refer to the source code. 7413 7414For x86: 7415 7416- the register sets to be copied out to kvm_run are selectable 7417 by userspace (rather that all sets being copied out for every exit). 7418- vcpu_events are available in addition to regs and sregs. 7419 7420For x86, the 'kvm_valid_regs' field of struct kvm_run is overloaded to 7421function as an input bit-array field set by userspace to indicate the 7422specific register sets to be copied out on the next exit. 7423 7424To indicate when userspace has modified values that should be copied into 7425the vCPU, the all architecture bitarray field, 'kvm_dirty_regs' must be set. 7426This is done using the same bitflags as for the 'kvm_valid_regs' field. 7427If the dirty bit is not set, then the register set values will not be copied 7428into the vCPU even if they've been modified. 7429 7430Unused bitfields in the bitarrays must be set to zero. 7431 7432:: 7433 7434 struct kvm_sync_regs { 7435 struct kvm_regs regs; 7436 struct kvm_sregs sregs; 7437 struct kvm_vcpu_events events; 7438 }; 7439 74406.75 KVM_CAP_PPC_IRQ_XIVE 7441------------------------- 7442 7443:Architectures: ppc 7444:Target: vcpu 7445:Parameters: args[0] is the XIVE device fd; 7446 args[1] is the XIVE CPU number (server ID) for this vcpu 7447 7448This capability connects the vcpu to an in-kernel XIVE device. 7449 74506.76 KVM_CAP_HYPERV_SYNIC 7451------------------------- 7452 7453:Architectures: x86 7454:Target: vcpu 7455 7456This capability, if KVM_CHECK_EXTENSION indicates that it is 7457available, means that the kernel has an implementation of the 7458Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is 7459used to support Windows Hyper-V based guest paravirt drivers(VMBus). 7460 7461In order to use SynIC, it has to be activated by setting this 7462capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this 7463will disable the use of APIC hardware virtualization even if supported 7464by the CPU, as it's incompatible with SynIC auto-EOI behavior. 7465 74666.77 KVM_CAP_HYPERV_SYNIC2 7467-------------------------- 7468 7469:Architectures: x86 7470:Target: vcpu 7471 7472This capability enables a newer version of Hyper-V Synthetic interrupt 7473controller (SynIC). The only difference with KVM_CAP_HYPERV_SYNIC is that KVM 7474doesn't clear SynIC message and event flags pages when they are enabled by 7475writing to the respective MSRs. 7476 74776.78 KVM_CAP_HYPERV_DIRECT_TLBFLUSH 7478----------------------------------- 7479 7480:Architectures: x86 7481:Target: vcpu 7482 7483This capability indicates that KVM running on top of Hyper-V hypervisor 7484enables Direct TLB flush for its guests meaning that TLB flush 7485hypercalls are handled by Level 0 hypervisor (Hyper-V) bypassing KVM. 7486Due to the different ABI for hypercall parameters between Hyper-V and 7487KVM, enabling this capability effectively disables all hypercall 7488handling by KVM (as some KVM hypercall may be mistakenly treated as TLB 7489flush hypercalls by Hyper-V) so userspace should disable KVM identification 7490in CPUID and only exposes Hyper-V identification. In this case, guest 7491thinks it's running on Hyper-V and only use Hyper-V hypercalls. 7492 74936.79 KVM_CAP_HYPERV_ENFORCE_CPUID 7494--------------------------------- 7495 7496:Architectures: x86 7497:Target: vcpu 7498 7499When enabled, KVM will disable emulated Hyper-V features provided to the 7500guest according to the bits Hyper-V CPUID feature leaves. Otherwise, all 7501currently implemented Hyper-V features are provided unconditionally when 7502Hyper-V identification is set in the HYPERV_CPUID_INTERFACE (0x40000001) 7503leaf. 7504 75056.80 KVM_CAP_ENFORCE_PV_FEATURE_CPUID 7506------------------------------------- 7507 7508:Architectures: x86 7509:Target: vcpu 7510 7511When enabled, KVM will disable paravirtual features provided to the 7512guest according to the bits in the KVM_CPUID_FEATURES CPUID leaf 7513(0x40000001). Otherwise, a guest may use the paravirtual features 7514regardless of what has actually been exposed through the CPUID leaf. 7515 7516.. _KVM_CAP_DIRTY_LOG_RING: 7517 7518 7519.. _cap_enable_vm: 7520 75217. Capabilities that can be enabled on VMs 7522========================================== 7523 7524There are certain capabilities that change the behavior of the virtual 7525machine when enabled. To enable them, please see section 7526:ref:`KVM_ENABLE_CAP`. Below you can find a list of capabilities and 7527what their effect on the VM is when enabling them. 7528 7529The following information is provided along with the description: 7530 7531 Architectures: 7532 which instruction set architectures provide this ioctl. 7533 x86 includes both i386 and x86_64. 7534 7535 Parameters: 7536 what parameters are accepted by the capability. 7537 7538 Returns: 7539 the return value. General error numbers (EBADF, ENOMEM, EINVAL) 7540 are not detailed, but errors with specific meanings are. 7541 7542 75437.1 KVM_CAP_PPC_ENABLE_HCALL 7544---------------------------- 7545 7546:Architectures: ppc 7547:Parameters: args[0] is the sPAPR hcall number; 7548 args[1] is 0 to disable, 1 to enable in-kernel handling 7549 7550This capability controls whether individual sPAPR hypercalls (hcalls) 7551get handled by the kernel or not. Enabling or disabling in-kernel 7552handling of an hcall is effective across the VM. On creation, an 7553initial set of hcalls are enabled for in-kernel handling, which 7554consists of those hcalls for which in-kernel handlers were implemented 7555before this capability was implemented. If disabled, the kernel will 7556not to attempt to handle the hcall, but will always exit to userspace 7557to handle it. Note that it may not make sense to enable some and 7558disable others of a group of related hcalls, but KVM does not prevent 7559userspace from doing that. 7560 7561If the hcall number specified is not one that has an in-kernel 7562implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL 7563error. 7564 75657.2 KVM_CAP_S390_USER_SIGP 7566-------------------------- 7567 7568:Architectures: s390 7569:Parameters: none 7570 7571This capability controls which SIGP orders will be handled completely in user 7572space. With this capability enabled, all fast orders will be handled completely 7573in the kernel: 7574 7575- SENSE 7576- SENSE RUNNING 7577- EXTERNAL CALL 7578- EMERGENCY SIGNAL 7579- CONDITIONAL EMERGENCY SIGNAL 7580 7581All other orders will be handled completely in user space. 7582 7583Only privileged operation exceptions will be checked for in the kernel (or even 7584in the hardware prior to interception). If this capability is not enabled, the 7585old way of handling SIGP orders is used (partially in kernel and user space). 7586 75877.3 KVM_CAP_S390_VECTOR_REGISTERS 7588--------------------------------- 7589 7590:Architectures: s390 7591:Parameters: none 7592:Returns: 0 on success, negative value on error 7593 7594Allows use of the vector registers introduced with z13 processor, and 7595provides for the synchronization between host and user space. Will 7596return -EINVAL if the machine does not support vectors. 7597 75987.4 KVM_CAP_S390_USER_STSI 7599-------------------------- 7600 7601:Architectures: s390 7602:Parameters: none 7603 7604This capability allows post-handlers for the STSI instruction. After 7605initial handling in the kernel, KVM exits to user space with 7606KVM_EXIT_S390_STSI to allow user space to insert further data. 7607 7608Before exiting to userspace, kvm handlers should fill in s390_stsi field of 7609vcpu->run:: 7610 7611 struct { 7612 __u64 addr; 7613 __u8 ar; 7614 __u8 reserved; 7615 __u8 fc; 7616 __u8 sel1; 7617 __u16 sel2; 7618 } s390_stsi; 7619 7620 @addr - guest address of STSI SYSIB 7621 @fc - function code 7622 @sel1 - selector 1 7623 @sel2 - selector 2 7624 @ar - access register number 7625 7626KVM handlers should exit to userspace with rc = -EREMOTE. 7627 76287.5 KVM_CAP_SPLIT_IRQCHIP 7629------------------------- 7630 7631:Architectures: x86 7632:Parameters: args[0] - number of routes reserved for userspace IOAPICs 7633:Returns: 0 on success, -1 on error 7634 7635Create a local apic for each processor in the kernel. This can be used 7636instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the 7637IOAPIC and PIC (and also the PIT, even though this has to be enabled 7638separately). 7639 7640This capability also enables in kernel routing of interrupt requests; 7641when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are 7642used in the IRQ routing table. The first args[0] MSI routes are reserved 7643for the IOAPIC pins. Whenever the LAPIC receives an EOI for these routes, 7644a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace. 7645 7646Fails if VCPU has already been created, or if the irqchip is already in the 7647kernel (i.e. KVM_CREATE_IRQCHIP has already been called). 7648 76497.6 KVM_CAP_S390_RI 7650------------------- 7651 7652:Architectures: s390 7653:Parameters: none 7654 7655Allows use of runtime-instrumentation introduced with zEC12 processor. 7656Will return -EINVAL if the machine does not support runtime-instrumentation. 7657Will return -EBUSY if a VCPU has already been created. 7658 76597.7 KVM_CAP_X2APIC_API 7660---------------------- 7661 7662:Architectures: x86 7663:Parameters: args[0] - features that should be enabled 7664:Returns: 0 on success, -EINVAL when args[0] contains invalid features 7665 7666Valid feature flags in args[0] are:: 7667 7668 #define KVM_X2APIC_API_USE_32BIT_IDS (1ULL << 0) 7669 #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK (1ULL << 1) 7670 7671Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of 7672KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC, 7673allowing the use of 32-bit APIC IDs. See KVM_CAP_X2APIC_API in their 7674respective sections. 7675 7676KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work 7677in logical mode or with more than 255 VCPUs. Otherwise, KVM treats 0xff 7678as a broadcast even in x2APIC mode in order to support physical x2APIC 7679without interrupt remapping. This is undesirable in logical mode, 7680where 0xff represents CPUs 0-7 in cluster 0. 7681 76827.8 KVM_CAP_S390_USER_INSTR0 7683---------------------------- 7684 7685:Architectures: s390 7686:Parameters: none 7687 7688With this capability enabled, all illegal instructions 0x0000 (2 bytes) will 7689be intercepted and forwarded to user space. User space can use this 7690mechanism e.g. to realize 2-byte software breakpoints. The kernel will 7691not inject an operating exception for these instructions, user space has 7692to take care of that. 7693 7694This capability can be enabled dynamically even if VCPUs were already 7695created and are running. 7696 76977.9 KVM_CAP_S390_GS 7698------------------- 7699 7700:Architectures: s390 7701:Parameters: none 7702:Returns: 0 on success; -EINVAL if the machine does not support 7703 guarded storage; -EBUSY if a VCPU has already been created. 7704 7705Allows use of guarded storage for the KVM guest. 7706 77077.10 KVM_CAP_S390_AIS 7708--------------------- 7709 7710:Architectures: s390 7711:Parameters: none 7712 7713Allow use of adapter-interruption suppression. 7714:Returns: 0 on success; -EBUSY if a VCPU has already been created. 7715 77167.11 KVM_CAP_PPC_SMT 7717-------------------- 7718 7719:Architectures: ppc 7720:Parameters: vsmt_mode, flags 7721 7722Enabling this capability on a VM provides userspace with a way to set 7723the desired virtual SMT mode (i.e. the number of virtual CPUs per 7724virtual core). The virtual SMT mode, vsmt_mode, must be a power of 2 7725between 1 and 8. On POWER8, vsmt_mode must also be no greater than 7726the number of threads per subcore for the host. Currently flags must 7727be 0. A successful call to enable this capability will result in 7728vsmt_mode being returned when the KVM_CAP_PPC_SMT capability is 7729subsequently queried for the VM. This capability is only supported by 7730HV KVM, and can only be set before any VCPUs have been created. 7731The KVM_CAP_PPC_SMT_POSSIBLE capability indicates which virtual SMT 7732modes are available. 7733 77347.12 KVM_CAP_PPC_FWNMI 7735---------------------- 7736 7737:Architectures: ppc 7738:Parameters: none 7739 7740With this capability a machine check exception in the guest address 7741space will cause KVM to exit the guest with NMI exit reason. This 7742enables QEMU to build error log and branch to guest kernel registered 7743machine check handling routine. Without this capability KVM will 7744branch to guests' 0x200 interrupt vector. 7745 77467.13 KVM_CAP_X86_DISABLE_EXITS 7747------------------------------ 7748 7749:Architectures: x86 7750:Parameters: args[0] defines which exits are disabled 7751:Returns: 0 on success, -EINVAL when args[0] contains invalid exits 7752 or if any vCPUs have already been created 7753 7754Valid bits in args[0] are:: 7755 7756 #define KVM_X86_DISABLE_EXITS_MWAIT (1 << 0) 7757 #define KVM_X86_DISABLE_EXITS_HLT (1 << 1) 7758 #define KVM_X86_DISABLE_EXITS_PAUSE (1 << 2) 7759 #define KVM_X86_DISABLE_EXITS_CSTATE (1 << 3) 7760 7761Enabling this capability on a VM provides userspace with a way to no 7762longer intercept some instructions for improved latency in some 7763workloads, and is suggested when vCPUs are associated to dedicated 7764physical CPUs. More bits can be added in the future; userspace can 7765just pass the KVM_CHECK_EXTENSION result to KVM_ENABLE_CAP to disable 7766all such vmexits. 7767 7768Do not enable KVM_FEATURE_PV_UNHALT if you disable HLT exits. 7769 77707.14 KVM_CAP_S390_HPAGE_1M 7771-------------------------- 7772 7773:Architectures: s390 7774:Parameters: none 7775:Returns: 0 on success, -EINVAL if hpage module parameter was not set 7776 or cmma is enabled, or the VM has the KVM_VM_S390_UCONTROL 7777 flag set 7778 7779With this capability the KVM support for memory backing with 1m pages 7780through hugetlbfs can be enabled for a VM. After the capability is 7781enabled, cmma can't be enabled anymore and pfmfi and the storage key 7782interpretation are disabled. If cmma has already been enabled or the 7783hpage module parameter is not set to 1, -EINVAL is returned. 7784 7785While it is generally possible to create a huge page backed VM without 7786this capability, the VM will not be able to run. 7787 77887.15 KVM_CAP_MSR_PLATFORM_INFO 7789------------------------------ 7790 7791:Architectures: x86 7792:Parameters: args[0] whether feature should be enabled or not 7793 7794With this capability, a guest may read the MSR_PLATFORM_INFO MSR. Otherwise, 7795a #GP would be raised when the guest tries to access. Currently, this 7796capability does not enable write permissions of this MSR for the guest. 7797 77987.16 KVM_CAP_PPC_NESTED_HV 7799-------------------------- 7800 7801:Architectures: ppc 7802:Parameters: none 7803:Returns: 0 on success, -EINVAL when the implementation doesn't support 7804 nested-HV virtualization. 7805 7806HV-KVM on POWER9 and later systems allows for "nested-HV" 7807virtualization, which provides a way for a guest VM to run guests that 7808can run using the CPU's supervisor mode (privileged non-hypervisor 7809state). Enabling this capability on a VM depends on the CPU having 7810the necessary functionality and on the facility being enabled with a 7811kvm-hv module parameter. 7812 78137.17 KVM_CAP_EXCEPTION_PAYLOAD 7814------------------------------ 7815 7816:Architectures: x86 7817:Parameters: args[0] whether feature should be enabled or not 7818 7819With this capability enabled, CR2 will not be modified prior to the 7820emulated VM-exit when L1 intercepts a #PF exception that occurs in 7821L2. Similarly, for kvm-intel only, DR6 will not be modified prior to 7822the emulated VM-exit when L1 intercepts a #DB exception that occurs in 7823L2. As a result, when KVM_GET_VCPU_EVENTS reports a pending #PF (or 7824#DB) exception for L2, exception.has_payload will be set and the 7825faulting address (or the new DR6 bits*) will be reported in the 7826exception_payload field. Similarly, when userspace injects a #PF (or 7827#DB) into L2 using KVM_SET_VCPU_EVENTS, it is expected to set 7828exception.has_payload and to put the faulting address - or the new DR6 7829bits\ [#]_ - in the exception_payload field. 7830 7831This capability also enables exception.pending in struct 7832kvm_vcpu_events, which allows userspace to distinguish between pending 7833and injected exceptions. 7834 7835 7836.. [#] For the new DR6 bits, note that bit 16 is set iff the #DB exception 7837 will clear DR6.RTM. 7838 78397.18 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 7840-------------------------------------- 7841 7842:Architectures: x86, arm64, mips 7843:Parameters: args[0] whether feature should be enabled or not 7844 7845Valid flags are:: 7846 7847 #define KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (1 << 0) 7848 #define KVM_DIRTY_LOG_INITIALLY_SET (1 << 1) 7849 7850With KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE is set, KVM_GET_DIRTY_LOG will not 7851automatically clear and write-protect all pages that are returned as dirty. 7852Rather, userspace will have to do this operation separately using 7853KVM_CLEAR_DIRTY_LOG. 7854 7855At the cost of a slightly more complicated operation, this provides better 7856scalability and responsiveness for two reasons. First, 7857KVM_CLEAR_DIRTY_LOG ioctl can operate on a 64-page granularity rather 7858than requiring to sync a full memslot; this ensures that KVM does not 7859take spinlocks for an extended period of time. Second, in some cases a 7860large amount of time can pass between a call to KVM_GET_DIRTY_LOG and 7861userspace actually using the data in the page. Pages can be modified 7862during this time, which is inefficient for both the guest and userspace: 7863the guest will incur a higher penalty due to write protection faults, 7864while userspace can see false reports of dirty pages. Manual reprotection 7865helps reducing this time, improving guest performance and reducing the 7866number of dirty log false positives. 7867 7868With KVM_DIRTY_LOG_INITIALLY_SET set, all the bits of the dirty bitmap 7869will be initialized to 1 when created. This also improves performance because 7870dirty logging can be enabled gradually in small chunks on the first call 7871to KVM_CLEAR_DIRTY_LOG. KVM_DIRTY_LOG_INITIALLY_SET depends on 7872KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (it is also only available on 7873x86 and arm64 for now). 7874 7875KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 was previously available under the name 7876KVM_CAP_MANUAL_DIRTY_LOG_PROTECT, but the implementation had bugs that make 7877it hard or impossible to use it correctly. The availability of 7878KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 signals that those bugs are fixed. 7879Userspace should not try to use KVM_CAP_MANUAL_DIRTY_LOG_PROTECT. 7880 78817.19 KVM_CAP_PPC_SECURE_GUEST 7882------------------------------ 7883 7884:Architectures: ppc 7885 7886This capability indicates that KVM is running on a host that has 7887ultravisor firmware and thus can support a secure guest. On such a 7888system, a guest can ask the ultravisor to make it a secure guest, 7889one whose memory is inaccessible to the host except for pages which 7890are explicitly requested to be shared with the host. The ultravisor 7891notifies KVM when a guest requests to become a secure guest, and KVM 7892has the opportunity to veto the transition. 7893 7894If present, this capability can be enabled for a VM, meaning that KVM 7895will allow the transition to secure guest mode. Otherwise KVM will 7896veto the transition. 7897 78987.20 KVM_CAP_HALT_POLL 7899---------------------- 7900 7901:Architectures: all 7902:Target: VM 7903:Parameters: args[0] is the maximum poll time in nanoseconds 7904:Returns: 0 on success; -1 on error 7905 7906KVM_CAP_HALT_POLL overrides the kvm.halt_poll_ns module parameter to set the 7907maximum halt-polling time for all vCPUs in the target VM. This capability can 7908be invoked at any time and any number of times to dynamically change the 7909maximum halt-polling time. 7910 7911See Documentation/virt/kvm/halt-polling.rst for more information on halt 7912polling. 7913 79147.21 KVM_CAP_X86_USER_SPACE_MSR 7915------------------------------- 7916 7917:Architectures: x86 7918:Target: VM 7919:Parameters: args[0] contains the mask of KVM_MSR_EXIT_REASON_* events to report 7920:Returns: 0 on success; -1 on error 7921 7922This capability allows userspace to intercept RDMSR and WRMSR instructions if 7923access to an MSR is denied. By default, KVM injects #GP on denied accesses. 7924 7925When a guest requests to read or write an MSR, KVM may not implement all MSRs 7926that are relevant to a respective system. It also does not differentiate by 7927CPU type. 7928 7929To allow more fine grained control over MSR handling, userspace may enable 7930this capability. With it enabled, MSR accesses that match the mask specified in 7931args[0] and would trigger a #GP inside the guest will instead trigger 7932KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR exit notifications. Userspace 7933can then implement model specific MSR handling and/or user notifications 7934to inform a user that an MSR was not emulated/virtualized by KVM. 7935 7936The valid mask flags are: 7937 7938============================ =============================================== 7939 KVM_MSR_EXIT_REASON_UNKNOWN intercept accesses to unknown (to KVM) MSRs 7940 KVM_MSR_EXIT_REASON_INVAL intercept accesses that are architecturally 7941 invalid according to the vCPU model and/or mode 7942 KVM_MSR_EXIT_REASON_FILTER intercept accesses that are denied by userspace 7943 via KVM_X86_SET_MSR_FILTER 7944============================ =============================================== 7945 79467.22 KVM_CAP_X86_BUS_LOCK_EXIT 7947------------------------------- 7948 7949:Architectures: x86 7950:Target: VM 7951:Parameters: args[0] defines the policy used when bus locks detected in guest 7952:Returns: 0 on success, -EINVAL when args[0] contains invalid bits 7953 7954Valid bits in args[0] are:: 7955 7956 #define KVM_BUS_LOCK_DETECTION_OFF (1 << 0) 7957 #define KVM_BUS_LOCK_DETECTION_EXIT (1 << 1) 7958 7959Enabling this capability on a VM provides userspace with a way to select a 7960policy to handle the bus locks detected in guest. Userspace can obtain the 7961supported modes from the result of KVM_CHECK_EXTENSION and define it through 7962the KVM_ENABLE_CAP. The supported modes are mutually-exclusive. 7963 7964This capability allows userspace to force VM exits on bus locks detected in the 7965guest, irrespective whether or not the host has enabled split-lock detection 7966(which triggers an #AC exception that KVM intercepts). This capability is 7967intended to mitigate attacks where a malicious/buggy guest can exploit bus 7968locks to degrade the performance of the whole system. 7969 7970If KVM_BUS_LOCK_DETECTION_OFF is set, KVM doesn't force guest bus locks to VM 7971exit, although the host kernel's split-lock #AC detection still applies, if 7972enabled. 7973 7974If KVM_BUS_LOCK_DETECTION_EXIT is set, KVM enables a CPU feature that ensures 7975bus locks in the guest trigger a VM exit, and KVM exits to userspace for all 7976such VM exits, e.g. to allow userspace to throttle the offending guest and/or 7977apply some other policy-based mitigation. When exiting to userspace, KVM sets 7978KVM_RUN_X86_BUS_LOCK in vcpu-run->flags, and conditionally sets the exit_reason 7979to KVM_EXIT_X86_BUS_LOCK. 7980 7981Note! Detected bus locks may be coincident with other exits to userspace, i.e. 7982KVM_RUN_X86_BUS_LOCK should be checked regardless of the primary exit reason if 7983userspace wants to take action on all detected bus locks. 7984 79857.23 KVM_CAP_PPC_DAWR1 7986---------------------- 7987 7988:Architectures: ppc 7989:Parameters: none 7990:Returns: 0 on success, -EINVAL when CPU doesn't support 2nd DAWR 7991 7992This capability can be used to check / enable 2nd DAWR feature provided 7993by POWER10 processor. 7994 7995 79967.24 KVM_CAP_VM_COPY_ENC_CONTEXT_FROM 7997------------------------------------- 7998 7999:Architectures: x86 SEV enabled 8000:Type: vm 8001:Parameters: args[0] is the fd of the source vm 8002:Returns: 0 on success; ENOTTY on error 8003 8004This capability enables userspace to copy encryption context from the vm 8005indicated by the fd to the vm this is called on. 8006 8007This is intended to support in-guest workloads scheduled by the host. This 8008allows the in-guest workload to maintain its own NPTs and keeps the two vms 8009from accidentally clobbering each other with interrupts and the like (separate 8010APIC/MSRs/etc). 8011 80127.25 KVM_CAP_SGX_ATTRIBUTE 8013-------------------------- 8014 8015:Architectures: x86 8016:Target: VM 8017:Parameters: args[0] is a file handle of a SGX attribute file in securityfs 8018:Returns: 0 on success, -EINVAL if the file handle is invalid or if a requested 8019 attribute is not supported by KVM. 8020 8021KVM_CAP_SGX_ATTRIBUTE enables a userspace VMM to grant a VM access to one or 8022more privileged enclave attributes. args[0] must hold a file handle to a valid 8023SGX attribute file corresponding to an attribute that is supported/restricted 8024by KVM (currently only PROVISIONKEY). 8025 8026The SGX subsystem restricts access to a subset of enclave attributes to provide 8027additional security for an uncompromised kernel, e.g. use of the PROVISIONKEY 8028is restricted to deter malware from using the PROVISIONKEY to obtain a stable 8029system fingerprint. To prevent userspace from circumventing such restrictions 8030by running an enclave in a VM, KVM prevents access to privileged attributes by 8031default. 8032 8033See Documentation/arch/x86/sgx.rst for more details. 8034 80357.27 KVM_CAP_EXIT_ON_EMULATION_FAILURE 8036-------------------------------------- 8037 8038:Architectures: x86 8039:Parameters: args[0] whether the feature should be enabled or not 8040 8041When this capability is enabled, an emulation failure will result in an exit 8042to userspace with KVM_INTERNAL_ERROR (except when the emulator was invoked 8043to handle a VMware backdoor instruction). Furthermore, KVM will now provide up 8044to 15 instruction bytes for any exit to userspace resulting from an emulation 8045failure. When these exits to userspace occur use the emulation_failure struct 8046instead of the internal struct. They both have the same layout, but the 8047emulation_failure struct matches the content better. It also explicitly 8048defines the 'flags' field which is used to describe the fields in the struct 8049that are valid (ie: if KVM_INTERNAL_ERROR_EMULATION_FLAG_INSTRUCTION_BYTES is 8050set in the 'flags' field then both 'insn_size' and 'insn_bytes' have valid data 8051in them.) 8052 80537.28 KVM_CAP_ARM_MTE 8054-------------------- 8055 8056:Architectures: arm64 8057:Parameters: none 8058 8059This capability indicates that KVM (and the hardware) supports exposing the 8060Memory Tagging Extensions (MTE) to the guest. It must also be enabled by the 8061VMM before creating any VCPUs to allow the guest access. Note that MTE is only 8062available to a guest running in AArch64 mode and enabling this capability will 8063cause attempts to create AArch32 VCPUs to fail. 8064 8065When enabled the guest is able to access tags associated with any memory given 8066to the guest. KVM will ensure that the tags are maintained during swap or 8067hibernation of the host; however the VMM needs to manually save/restore the 8068tags as appropriate if the VM is migrated. 8069 8070When this capability is enabled all memory in memslots must be mapped as 8071``MAP_ANONYMOUS`` or with a RAM-based file mapping (``tmpfs``, ``memfd``), 8072attempts to create a memslot with an invalid mmap will result in an 8073-EINVAL return. 8074 8075When enabled the VMM may make use of the ``KVM_ARM_MTE_COPY_TAGS`` ioctl to 8076perform a bulk copy of tags to/from the guest. 8077 80787.29 KVM_CAP_VM_MOVE_ENC_CONTEXT_FROM 8079------------------------------------- 8080 8081:Architectures: x86 SEV enabled 8082:Type: vm 8083:Parameters: args[0] is the fd of the source vm 8084:Returns: 0 on success 8085 8086This capability enables userspace to migrate the encryption context from the VM 8087indicated by the fd to the VM this is called on. 8088 8089This is intended to support intra-host migration of VMs between userspace VMMs, 8090upgrading the VMM process without interrupting the guest. 8091 80927.31 KVM_CAP_DISABLE_QUIRKS2 8093---------------------------- 8094 8095:Parameters: args[0] - set of KVM quirks to disable 8096:Architectures: x86 8097:Type: vm 8098 8099This capability, if enabled, will cause KVM to disable some behavior 8100quirks. 8101 8102Calling KVM_CHECK_EXTENSION for this capability returns a bitmask of 8103quirks that can be disabled in KVM. 8104 8105The argument to KVM_ENABLE_CAP for this capability is a bitmask of 8106quirks to disable, and must be a subset of the bitmask returned by 8107KVM_CHECK_EXTENSION. 8108 8109The valid bits in cap.args[0] are: 8110 8111=================================== ============================================ 8112 KVM_X86_QUIRK_LINT0_REENABLED By default, the reset value for the LVT 8113 LINT0 register is 0x700 (APIC_MODE_EXTINT). 8114 When this quirk is disabled, the reset value 8115 is 0x10000 (APIC_LVT_MASKED). 8116 8117 KVM_X86_QUIRK_CD_NW_CLEARED By default, KVM clears CR0.CD and CR0.NW on 8118 AMD CPUs to workaround buggy guest firmware 8119 that runs in perpetuity with CR0.CD, i.e. 8120 with caches in "no fill" mode. 8121 8122 When this quirk is disabled, KVM does not 8123 change the value of CR0.CD and CR0.NW. 8124 8125 KVM_X86_QUIRK_LAPIC_MMIO_HOLE By default, the MMIO LAPIC interface is 8126 available even when configured for x2APIC 8127 mode. When this quirk is disabled, KVM 8128 disables the MMIO LAPIC interface if the 8129 LAPIC is in x2APIC mode. 8130 8131 KVM_X86_QUIRK_OUT_7E_INC_RIP By default, KVM pre-increments %rip before 8132 exiting to userspace for an OUT instruction 8133 to port 0x7e. When this quirk is disabled, 8134 KVM does not pre-increment %rip before 8135 exiting to userspace. 8136 8137 KVM_X86_QUIRK_MISC_ENABLE_NO_MWAIT When this quirk is disabled, KVM sets 8138 CPUID.01H:ECX[bit 3] (MONITOR/MWAIT) if 8139 IA32_MISC_ENABLE[bit 18] (MWAIT) is set. 8140 Additionally, when this quirk is disabled, 8141 KVM clears CPUID.01H:ECX[bit 3] if 8142 IA32_MISC_ENABLE[bit 18] is cleared. 8143 8144 KVM_X86_QUIRK_FIX_HYPERCALL_INSN By default, KVM rewrites guest 8145 VMMCALL/VMCALL instructions to match the 8146 vendor's hypercall instruction for the 8147 system. When this quirk is disabled, KVM 8148 will no longer rewrite invalid guest 8149 hypercall instructions. Executing the 8150 incorrect hypercall instruction will 8151 generate a #UD within the guest. 8152 8153KVM_X86_QUIRK_MWAIT_NEVER_UD_FAULTS By default, KVM emulates MONITOR/MWAIT (if 8154 they are intercepted) as NOPs regardless of 8155 whether or not MONITOR/MWAIT are supported 8156 according to guest CPUID. When this quirk 8157 is disabled and KVM_X86_DISABLE_EXITS_MWAIT 8158 is not set (MONITOR/MWAIT are intercepted), 8159 KVM will inject a #UD on MONITOR/MWAIT if 8160 they're unsupported per guest CPUID. Note, 8161 KVM will modify MONITOR/MWAIT support in 8162 guest CPUID on writes to MISC_ENABLE if 8163 KVM_X86_QUIRK_MISC_ENABLE_NO_MWAIT is 8164 disabled. 8165 8166KVM_X86_QUIRK_SLOT_ZAP_ALL By default, for KVM_X86_DEFAULT_VM VMs, KVM 8167 invalidates all SPTEs in all memslots and 8168 address spaces when a memslot is deleted or 8169 moved. When this quirk is disabled (or the 8170 VM type isn't KVM_X86_DEFAULT_VM), KVM only 8171 ensures the backing memory of the deleted 8172 or moved memslot isn't reachable, i.e KVM 8173 _may_ invalidate only SPTEs related to the 8174 memslot. 8175 8176KVM_X86_QUIRK_STUFF_FEATURE_MSRS By default, at vCPU creation, KVM sets the 8177 vCPU's MSR_IA32_PERF_CAPABILITIES (0x345), 8178 MSR_IA32_ARCH_CAPABILITIES (0x10a), 8179 MSR_PLATFORM_INFO (0xce), and all VMX MSRs 8180 (0x480..0x492) to the maximal capabilities 8181 supported by KVM. KVM also sets 8182 MSR_IA32_UCODE_REV (0x8b) to an arbitrary 8183 value (which is different for Intel vs. 8184 AMD). Lastly, when guest CPUID is set (by 8185 userspace), KVM modifies select VMX MSR 8186 fields to force consistency between guest 8187 CPUID and L2's effective ISA. When this 8188 quirk is disabled, KVM zeroes the vCPU's MSR 8189 values (with two exceptions, see below), 8190 i.e. treats the feature MSRs like CPUID 8191 leaves and gives userspace full control of 8192 the vCPU model definition. This quirk does 8193 not affect VMX MSRs CR0/CR4_FIXED1 (0x487 8194 and 0x489), as KVM does now allow them to 8195 be set by userspace (KVM sets them based on 8196 guest CPUID, for safety purposes). 8197=================================== ============================================ 8198 81997.32 KVM_CAP_MAX_VCPU_ID 8200------------------------ 8201 8202:Architectures: x86 8203:Target: VM 8204:Parameters: args[0] - maximum APIC ID value set for current VM 8205:Returns: 0 on success, -EINVAL if args[0] is beyond KVM_MAX_VCPU_IDS 8206 supported in KVM or if it has been set. 8207 8208This capability allows userspace to specify maximum possible APIC ID 8209assigned for current VM session prior to the creation of vCPUs, saving 8210memory for data structures indexed by the APIC ID. Userspace is able 8211to calculate the limit to APIC ID values from designated 8212CPU topology. 8213 8214The value can be changed only until KVM_ENABLE_CAP is set to a nonzero 8215value or until a vCPU is created. Upon creation of the first vCPU, 8216if the value was set to zero or KVM_ENABLE_CAP was not invoked, KVM 8217uses the return value of KVM_CHECK_EXTENSION(KVM_CAP_MAX_VCPU_ID) as 8218the maximum APIC ID. 8219 82207.33 KVM_CAP_X86_NOTIFY_VMEXIT 8221------------------------------ 8222 8223:Architectures: x86 8224:Target: VM 8225:Parameters: args[0] is the value of notify window as well as some flags 8226:Returns: 0 on success, -EINVAL if args[0] contains invalid flags or notify 8227 VM exit is unsupported. 8228 8229Bits 63:32 of args[0] are used for notify window. 8230Bits 31:0 of args[0] are for some flags. Valid bits are:: 8231 8232 #define KVM_X86_NOTIFY_VMEXIT_ENABLED (1 << 0) 8233 #define KVM_X86_NOTIFY_VMEXIT_USER (1 << 1) 8234 8235This capability allows userspace to configure the notify VM exit on/off 8236in per-VM scope during VM creation. Notify VM exit is disabled by default. 8237When userspace sets KVM_X86_NOTIFY_VMEXIT_ENABLED bit in args[0], VMM will 8238enable this feature with the notify window provided, which will generate 8239a VM exit if no event window occurs in VM non-root mode for a specified of 8240time (notify window). 8241 8242If KVM_X86_NOTIFY_VMEXIT_USER is set in args[0], upon notify VM exits happen, 8243KVM would exit to userspace for handling. 8244 8245This capability is aimed to mitigate the threat that malicious VMs can 8246cause CPU stuck (due to event windows don't open up) and make the CPU 8247unavailable to host or other VMs. 8248 82497.35 KVM_CAP_X86_APIC_BUS_CYCLES_NS 8250----------------------------------- 8251 8252:Architectures: x86 8253:Target: VM 8254:Parameters: args[0] is the desired APIC bus clock rate, in nanoseconds 8255:Returns: 0 on success, -EINVAL if args[0] contains an invalid value for the 8256 frequency or if any vCPUs have been created, -ENXIO if a virtual 8257 local APIC has not been created using KVM_CREATE_IRQCHIP. 8258 8259This capability sets the VM's APIC bus clock frequency, used by KVM's in-kernel 8260virtual APIC when emulating APIC timers. KVM's default value can be retrieved 8261by KVM_CHECK_EXTENSION. 8262 8263Note: Userspace is responsible for correctly configuring CPUID 0x15, a.k.a. the 8264core crystal clock frequency, if a non-zero CPUID 0x15 is exposed to the guest. 8265 82667.36 KVM_CAP_DIRTY_LOG_RING/KVM_CAP_DIRTY_LOG_RING_ACQ_REL 8267---------------------------------------------------------- 8268 8269:Architectures: x86, arm64 8270:Type: vm 8271:Parameters: args[0] - size of the dirty log ring 8272 8273KVM is capable of tracking dirty memory using ring buffers that are 8274mmapped into userspace; there is one dirty ring per vcpu. 8275 8276The dirty ring is available to userspace as an array of 8277``struct kvm_dirty_gfn``. Each dirty entry is defined as:: 8278 8279 struct kvm_dirty_gfn { 8280 __u32 flags; 8281 __u32 slot; /* as_id | slot_id */ 8282 __u64 offset; 8283 }; 8284 8285The following values are defined for the flags field to define the 8286current state of the entry:: 8287 8288 #define KVM_DIRTY_GFN_F_DIRTY BIT(0) 8289 #define KVM_DIRTY_GFN_F_RESET BIT(1) 8290 #define KVM_DIRTY_GFN_F_MASK 0x3 8291 8292Userspace should call KVM_ENABLE_CAP ioctl right after KVM_CREATE_VM 8293ioctl to enable this capability for the new guest and set the size of 8294the rings. Enabling the capability is only allowed before creating any 8295vCPU, and the size of the ring must be a power of two. The larger the 8296ring buffer, the less likely the ring is full and the VM is forced to 8297exit to userspace. The optimal size depends on the workload, but it is 8298recommended that it be at least 64 KiB (4096 entries). 8299 8300Just like for dirty page bitmaps, the buffer tracks writes to 8301all user memory regions for which the KVM_MEM_LOG_DIRTY_PAGES flag was 8302set in KVM_SET_USER_MEMORY_REGION. Once a memory region is registered 8303with the flag set, userspace can start harvesting dirty pages from the 8304ring buffer. 8305 8306An entry in the ring buffer can be unused (flag bits ``00``), 8307dirty (flag bits ``01``) or harvested (flag bits ``1X``). The 8308state machine for the entry is as follows:: 8309 8310 dirtied harvested reset 8311 00 -----------> 01 -------------> 1X -------+ 8312 ^ | 8313 | | 8314 +------------------------------------------+ 8315 8316To harvest the dirty pages, userspace accesses the mmapped ring buffer 8317to read the dirty GFNs. If the flags has the DIRTY bit set (at this stage 8318the RESET bit must be cleared), then it means this GFN is a dirty GFN. 8319The userspace should harvest this GFN and mark the flags from state 8320``01b`` to ``1Xb`` (bit 0 will be ignored by KVM, but bit 1 must be set 8321to show that this GFN is harvested and waiting for a reset), and move 8322on to the next GFN. The userspace should continue to do this until the 8323flags of a GFN have the DIRTY bit cleared, meaning that it has harvested 8324all the dirty GFNs that were available. 8325 8326Note that on weakly ordered architectures, userspace accesses to the 8327ring buffer (and more specifically the 'flags' field) must be ordered, 8328using load-acquire/store-release accessors when available, or any 8329other memory barrier that will ensure this ordering. 8330 8331It's not necessary for userspace to harvest the all dirty GFNs at once. 8332However it must collect the dirty GFNs in sequence, i.e., the userspace 8333program cannot skip one dirty GFN to collect the one next to it. 8334 8335After processing one or more entries in the ring buffer, userspace 8336calls the VM ioctl KVM_RESET_DIRTY_RINGS to notify the kernel about 8337it, so that the kernel will reprotect those collected GFNs. 8338Therefore, the ioctl must be called *before* reading the content of 8339the dirty pages. 8340 8341The dirty ring can get full. When it happens, the KVM_RUN of the 8342vcpu will return with exit reason KVM_EXIT_DIRTY_LOG_FULL. 8343 8344The dirty ring interface has a major difference comparing to the 8345KVM_GET_DIRTY_LOG interface in that, when reading the dirty ring from 8346userspace, it's still possible that the kernel has not yet flushed the 8347processor's dirty page buffers into the kernel buffer (with dirty bitmaps, the 8348flushing is done by the KVM_GET_DIRTY_LOG ioctl). To achieve that, one 8349needs to kick the vcpu out of KVM_RUN using a signal. The resulting 8350vmexit ensures that all dirty GFNs are flushed to the dirty rings. 8351 8352NOTE: KVM_CAP_DIRTY_LOG_RING_ACQ_REL is the only capability that 8353should be exposed by weakly ordered architecture, in order to indicate 8354the additional memory ordering requirements imposed on userspace when 8355reading the state of an entry and mutating it from DIRTY to HARVESTED. 8356Architecture with TSO-like ordering (such as x86) are allowed to 8357expose both KVM_CAP_DIRTY_LOG_RING and KVM_CAP_DIRTY_LOG_RING_ACQ_REL 8358to userspace. 8359 8360After enabling the dirty rings, the userspace needs to detect the 8361capability of KVM_CAP_DIRTY_LOG_RING_WITH_BITMAP to see whether the 8362ring structures can be backed by per-slot bitmaps. With this capability 8363advertised, it means the architecture can dirty guest pages without 8364vcpu/ring context, so that some of the dirty information will still be 8365maintained in the bitmap structure. KVM_CAP_DIRTY_LOG_RING_WITH_BITMAP 8366can't be enabled if the capability of KVM_CAP_DIRTY_LOG_RING_ACQ_REL 8367hasn't been enabled, or any memslot has been existing. 8368 8369Note that the bitmap here is only a backup of the ring structure. The 8370use of the ring and bitmap combination is only beneficial if there is 8371only a very small amount of memory that is dirtied out of vcpu/ring 8372context. Otherwise, the stand-alone per-slot bitmap mechanism needs to 8373be considered. 8374 8375To collect dirty bits in the backup bitmap, userspace can use the same 8376KVM_GET_DIRTY_LOG ioctl. KVM_CLEAR_DIRTY_LOG isn't needed as long as all 8377the generation of the dirty bits is done in a single pass. Collecting 8378the dirty bitmap should be the very last thing that the VMM does before 8379considering the state as complete. VMM needs to ensure that the dirty 8380state is final and avoid missing dirty pages from another ioctl ordered 8381after the bitmap collection. 8382 8383NOTE: Multiple examples of using the backup bitmap: (1) save vgic/its 8384tables through command KVM_DEV_ARM_{VGIC_GRP_CTRL, ITS_SAVE_TABLES} on 8385KVM device "kvm-arm-vgic-its". (2) restore vgic/its tables through 8386command KVM_DEV_ARM_{VGIC_GRP_CTRL, ITS_RESTORE_TABLES} on KVM device 8387"kvm-arm-vgic-its". VGICv3 LPI pending status is restored. (3) save 8388vgic3 pending table through KVM_DEV_ARM_VGIC_{GRP_CTRL, SAVE_PENDING_TABLES} 8389command on KVM device "kvm-arm-vgic-v3". 8390 83917.37 KVM_CAP_PMU_CAPABILITY 8392--------------------------- 8393 8394:Architectures: x86 8395:Type: vm 8396:Parameters: arg[0] is bitmask of PMU virtualization capabilities. 8397:Returns: 0 on success, -EINVAL when arg[0] contains invalid bits 8398 8399This capability alters PMU virtualization in KVM. 8400 8401Calling KVM_CHECK_EXTENSION for this capability returns a bitmask of 8402PMU virtualization capabilities that can be adjusted on a VM. 8403 8404The argument to KVM_ENABLE_CAP is also a bitmask and selects specific 8405PMU virtualization capabilities to be applied to the VM. This can 8406only be invoked on a VM prior to the creation of VCPUs. 8407 8408At this time, KVM_PMU_CAP_DISABLE is the only capability. Setting 8409this capability will disable PMU virtualization for that VM. Usermode 8410should adjust CPUID leaf 0xA to reflect that the PMU is disabled. 8411 84127.38 KVM_CAP_VM_DISABLE_NX_HUGE_PAGES 8413------------------------------------- 8414 8415:Architectures: x86 8416:Type: vm 8417:Parameters: arg[0] must be 0. 8418:Returns: 0 on success, -EPERM if the userspace process does not 8419 have CAP_SYS_BOOT, -EINVAL if args[0] is not 0 or any vCPUs have been 8420 created. 8421 8422This capability disables the NX huge pages mitigation for iTLB MULTIHIT. 8423 8424The capability has no effect if the nx_huge_pages module parameter is not set. 8425 8426This capability may only be set before any vCPUs are created. 8427 84287.39 KVM_CAP_ARM_EAGER_SPLIT_CHUNK_SIZE 8429--------------------------------------- 8430 8431:Architectures: arm64 8432:Type: vm 8433:Parameters: arg[0] is the new split chunk size. 8434:Returns: 0 on success, -EINVAL if any memslot was already created. 8435 8436This capability sets the chunk size used in Eager Page Splitting. 8437 8438Eager Page Splitting improves the performance of dirty-logging (used 8439in live migrations) when guest memory is backed by huge-pages. It 8440avoids splitting huge-pages (into PAGE_SIZE pages) on fault, by doing 8441it eagerly when enabling dirty logging (with the 8442KVM_MEM_LOG_DIRTY_PAGES flag for a memory region), or when using 8443KVM_CLEAR_DIRTY_LOG. 8444 8445The chunk size specifies how many pages to break at a time, using a 8446single allocation for each chunk. Bigger the chunk size, more pages 8447need to be allocated ahead of time. 8448 8449The chunk size needs to be a valid block size. The list of acceptable 8450block sizes is exposed in KVM_CAP_ARM_SUPPORTED_BLOCK_SIZES as a 845164-bit bitmap (each bit describing a block size). The default value is 84520, to disable the eager page splitting. 8453 84547.40 KVM_CAP_EXIT_HYPERCALL 8455--------------------------- 8456 8457:Architectures: x86 8458:Type: vm 8459 8460This capability, if enabled, will cause KVM to exit to userspace 8461with KVM_EXIT_HYPERCALL exit reason to process some hypercalls. 8462 8463Calling KVM_CHECK_EXTENSION for this capability will return a bitmask 8464of hypercalls that can be configured to exit to userspace. 8465Right now, the only such hypercall is KVM_HC_MAP_GPA_RANGE. 8466 8467The argument to KVM_ENABLE_CAP is also a bitmask, and must be a subset 8468of the result of KVM_CHECK_EXTENSION. KVM will forward to userspace 8469the hypercalls whose corresponding bit is in the argument, and return 8470ENOSYS for the others. 8471 84727.41 KVM_CAP_ARM_SYSTEM_SUSPEND 8473------------------------------- 8474 8475:Architectures: arm64 8476:Type: vm 8477 8478When enabled, KVM will exit to userspace with KVM_EXIT_SYSTEM_EVENT of 8479type KVM_SYSTEM_EVENT_SUSPEND to process the guest suspend request. 8480 84817.37 KVM_CAP_ARM_WRITABLE_IMP_ID_REGS 8482------------------------------------- 8483 8484:Architectures: arm64 8485:Target: VM 8486:Parameters: None 8487:Returns: 0 on success, -EINVAL if vCPUs have been created before enabling this 8488 capability. 8489 8490This capability changes the behavior of the registers that identify a PE 8491implementation of the Arm architecture: MIDR_EL1, REVIDR_EL1, and AIDR_EL1. 8492By default, these registers are visible to userspace but treated as invariant. 8493 8494When this capability is enabled, KVM allows userspace to change the 8495aforementioned registers before the first KVM_RUN. These registers are VM 8496scoped, meaning that the same set of values are presented on all vCPUs in a 8497given VM. 8498 84998. Other capabilities. 8500====================== 8501 8502This section lists capabilities that give information about other 8503features of the KVM implementation. 8504 85058.1 KVM_CAP_PPC_HWRNG 8506--------------------- 8507 8508:Architectures: ppc 8509 8510This capability, if KVM_CHECK_EXTENSION indicates that it is 8511available, means that the kernel has an implementation of the 8512H_RANDOM hypercall backed by a hardware random-number generator. 8513If present, the kernel H_RANDOM handler can be enabled for guest use 8514with the KVM_CAP_PPC_ENABLE_HCALL capability. 8515 85168.3 KVM_CAP_PPC_MMU_RADIX 8517------------------------- 8518 8519:Architectures: ppc 8520 8521This capability, if KVM_CHECK_EXTENSION indicates that it is 8522available, means that the kernel can support guests using the 8523radix MMU defined in Power ISA V3.00 (as implemented in the POWER9 8524processor). 8525 85268.4 KVM_CAP_PPC_MMU_HASH_V3 8527--------------------------- 8528 8529:Architectures: ppc 8530 8531This capability, if KVM_CHECK_EXTENSION indicates that it is 8532available, means that the kernel can support guests using the 8533hashed page table MMU defined in Power ISA V3.00 (as implemented in 8534the POWER9 processor), including in-memory segment tables. 8535 85368.5 KVM_CAP_MIPS_VZ 8537------------------- 8538 8539:Architectures: mips 8540 8541This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that 8542it is available, means that full hardware assisted virtualization capabilities 8543of the hardware are available for use through KVM. An appropriate 8544KVM_VM_MIPS_* type must be passed to KVM_CREATE_VM to create a VM which 8545utilises it. 8546 8547If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is 8548available, it means that the VM is using full hardware assisted virtualization 8549capabilities of the hardware. This is useful to check after creating a VM with 8550KVM_VM_MIPS_DEFAULT. 8551 8552The value returned by KVM_CHECK_EXTENSION should be compared against known 8553values (see below). All other values are reserved. This is to allow for the 8554possibility of other hardware assisted virtualization implementations which 8555may be incompatible with the MIPS VZ ASE. 8556 8557== ========================================================================== 8558 0 The trap & emulate implementation is in use to run guest code in user 8559 mode. Guest virtual memory segments are rearranged to fit the guest in the 8560 user mode address space. 8561 8562 1 The MIPS VZ ASE is in use, providing full hardware assisted 8563 virtualization, including standard guest virtual memory segments. 8564== ========================================================================== 8565 85668.7 KVM_CAP_MIPS_64BIT 8567---------------------- 8568 8569:Architectures: mips 8570 8571This capability indicates the supported architecture type of the guest, i.e. the 8572supported register and address width. 8573 8574The values returned when this capability is checked by KVM_CHECK_EXTENSION on a 8575kvm VM handle correspond roughly to the CP0_Config.AT register field, and should 8576be checked specifically against known values (see below). All other values are 8577reserved. 8578 8579== ======================================================================== 8580 0 MIPS32 or microMIPS32. 8581 Both registers and addresses are 32-bits wide. 8582 It will only be possible to run 32-bit guest code. 8583 8584 1 MIPS64 or microMIPS64 with access only to 32-bit compatibility segments. 8585 Registers are 64-bits wide, but addresses are 32-bits wide. 8586 64-bit guest code may run but cannot access MIPS64 memory segments. 8587 It will also be possible to run 32-bit guest code. 8588 8589 2 MIPS64 or microMIPS64 with access to all address segments. 8590 Both registers and addresses are 64-bits wide. 8591 It will be possible to run 64-bit or 32-bit guest code. 8592== ======================================================================== 8593 85948.9 KVM_CAP_ARM_USER_IRQ 8595------------------------ 8596 8597:Architectures: arm64 8598 8599This capability, if KVM_CHECK_EXTENSION indicates that it is available, means 8600that if userspace creates a VM without an in-kernel interrupt controller, it 8601will be notified of changes to the output level of in-kernel emulated devices, 8602which can generate virtual interrupts, presented to the VM. 8603For such VMs, on every return to userspace, the kernel 8604updates the vcpu's run->s.regs.device_irq_level field to represent the actual 8605output level of the device. 8606 8607Whenever kvm detects a change in the device output level, kvm guarantees at 8608least one return to userspace before running the VM. This exit could either 8609be a KVM_EXIT_INTR or any other exit event, like KVM_EXIT_MMIO. This way, 8610userspace can always sample the device output level and re-compute the state of 8611the userspace interrupt controller. Userspace should always check the state 8612of run->s.regs.device_irq_level on every kvm exit. 8613The value in run->s.regs.device_irq_level can represent both level and edge 8614triggered interrupt signals, depending on the device. Edge triggered interrupt 8615signals will exit to userspace with the bit in run->s.regs.device_irq_level 8616set exactly once per edge signal. 8617 8618The field run->s.regs.device_irq_level is available independent of 8619run->kvm_valid_regs or run->kvm_dirty_regs bits. 8620 8621If KVM_CAP_ARM_USER_IRQ is supported, the KVM_CHECK_EXTENSION ioctl returns a 8622number larger than 0 indicating the version of this capability is implemented 8623and thereby which bits in run->s.regs.device_irq_level can signal values. 8624 8625Currently the following bits are defined for the device_irq_level bitmap:: 8626 8627 KVM_CAP_ARM_USER_IRQ >= 1: 8628 8629 KVM_ARM_DEV_EL1_VTIMER - EL1 virtual timer 8630 KVM_ARM_DEV_EL1_PTIMER - EL1 physical timer 8631 KVM_ARM_DEV_PMU - ARM PMU overflow interrupt signal 8632 8633Future versions of kvm may implement additional events. These will get 8634indicated by returning a higher number from KVM_CHECK_EXTENSION and will be 8635listed above. 8636 86378.10 KVM_CAP_PPC_SMT_POSSIBLE 8638----------------------------- 8639 8640:Architectures: ppc 8641 8642Querying this capability returns a bitmap indicating the possible 8643virtual SMT modes that can be set using KVM_CAP_PPC_SMT. If bit N 8644(counting from the right) is set, then a virtual SMT mode of 2^N is 8645available. 8646 86478.12 KVM_CAP_HYPERV_VP_INDEX 8648---------------------------- 8649 8650:Architectures: x86 8651 8652This capability indicates that userspace can load HV_X64_MSR_VP_INDEX msr. Its 8653value is used to denote the target vcpu for a SynIC interrupt. For 8654compatibility, KVM initializes this msr to KVM's internal vcpu index. When this 8655capability is absent, userspace can still query this msr's value. 8656 86578.13 KVM_CAP_S390_AIS_MIGRATION 8658------------------------------- 8659 8660:Architectures: s390 8661 8662This capability indicates if the flic device will be able to get/set the 8663AIS states for migration via the KVM_DEV_FLIC_AISM_ALL attribute and allows 8664to discover this without having to create a flic device. 8665 86668.14 KVM_CAP_S390_PSW 8667--------------------- 8668 8669:Architectures: s390 8670 8671This capability indicates that the PSW is exposed via the kvm_run structure. 8672 86738.15 KVM_CAP_S390_GMAP 8674---------------------- 8675 8676:Architectures: s390 8677 8678This capability indicates that the user space memory used as guest mapping can 8679be anywhere in the user memory address space, as long as the memory slots are 8680aligned and sized to a segment (1MB) boundary. 8681 86828.16 KVM_CAP_S390_COW 8683--------------------- 8684 8685:Architectures: s390 8686 8687This capability indicates that the user space memory used as guest mapping can 8688use copy-on-write semantics as well as dirty pages tracking via read-only page 8689tables. 8690 86918.17 KVM_CAP_S390_BPB 8692--------------------- 8693 8694:Architectures: s390 8695 8696This capability indicates that kvm will implement the interfaces to handle 8697reset, migration and nested KVM for branch prediction blocking. The stfle 8698facility 82 should not be provided to the guest without this capability. 8699 87008.18 KVM_CAP_HYPERV_TLBFLUSH 8701---------------------------- 8702 8703:Architectures: x86 8704 8705This capability indicates that KVM supports paravirtualized Hyper-V TLB Flush 8706hypercalls: 8707HvFlushVirtualAddressSpace, HvFlushVirtualAddressSpaceEx, 8708HvFlushVirtualAddressList, HvFlushVirtualAddressListEx. 8709 87108.19 KVM_CAP_ARM_INJECT_SERROR_ESR 8711---------------------------------- 8712 8713:Architectures: arm64 8714 8715This capability indicates that userspace can specify (via the 8716KVM_SET_VCPU_EVENTS ioctl) the syndrome value reported to the guest when it 8717takes a virtual SError interrupt exception. 8718If KVM advertises this capability, userspace can only specify the ISS field for 8719the ESR syndrome. Other parts of the ESR, such as the EC are generated by the 8720CPU when the exception is taken. If this virtual SError is taken to EL1 using 8721AArch64, this value will be reported in the ISS field of ESR_ELx. 8722 8723See KVM_CAP_VCPU_EVENTS for more details. 8724 87258.20 KVM_CAP_HYPERV_SEND_IPI 8726---------------------------- 8727 8728:Architectures: x86 8729 8730This capability indicates that KVM supports paravirtualized Hyper-V IPI send 8731hypercalls: 8732HvCallSendSyntheticClusterIpi, HvCallSendSyntheticClusterIpiEx. 8733 87348.22 KVM_CAP_S390_VCPU_RESETS 8735----------------------------- 8736 8737:Architectures: s390 8738 8739This capability indicates that the KVM_S390_NORMAL_RESET and 8740KVM_S390_CLEAR_RESET ioctls are available. 8741 87428.23 KVM_CAP_S390_PROTECTED 8743--------------------------- 8744 8745:Architectures: s390 8746 8747This capability indicates that the Ultravisor has been initialized and 8748KVM can therefore start protected VMs. 8749This capability governs the KVM_S390_PV_COMMAND ioctl and the 8750KVM_MP_STATE_LOAD MP_STATE. KVM_SET_MP_STATE can fail for protected 8751guests when the state change is invalid. 8752 87538.24 KVM_CAP_STEAL_TIME 8754----------------------- 8755 8756:Architectures: arm64, x86 8757 8758This capability indicates that KVM supports steal time accounting. 8759When steal time accounting is supported it may be enabled with 8760architecture-specific interfaces. This capability and the architecture- 8761specific interfaces must be consistent, i.e. if one says the feature 8762is supported, than the other should as well and vice versa. For arm64 8763see Documentation/virt/kvm/devices/vcpu.rst "KVM_ARM_VCPU_PVTIME_CTRL". 8764For x86 see Documentation/virt/kvm/x86/msr.rst "MSR_KVM_STEAL_TIME". 8765 87668.25 KVM_CAP_S390_DIAG318 8767------------------------- 8768 8769:Architectures: s390 8770 8771This capability enables a guest to set information about its control program 8772(i.e. guest kernel type and version). The information is helpful during 8773system/firmware service events, providing additional data about the guest 8774environments running on the machine. 8775 8776The information is associated with the DIAGNOSE 0x318 instruction, which sets 8777an 8-byte value consisting of a one-byte Control Program Name Code (CPNC) and 8778a 7-byte Control Program Version Code (CPVC). The CPNC determines what 8779environment the control program is running in (e.g. Linux, z/VM...), and the 8780CPVC is used for information specific to OS (e.g. Linux version, Linux 8781distribution...) 8782 8783If this capability is available, then the CPNC and CPVC can be synchronized 8784between KVM and userspace via the sync regs mechanism (KVM_SYNC_DIAG318). 8785 87868.26 KVM_CAP_X86_USER_SPACE_MSR 8787------------------------------- 8788 8789:Architectures: x86 8790 8791This capability indicates that KVM supports deflection of MSR reads and 8792writes to user space. It can be enabled on a VM level. If enabled, MSR 8793accesses that would usually trigger a #GP by KVM into the guest will 8794instead get bounced to user space through the KVM_EXIT_X86_RDMSR and 8795KVM_EXIT_X86_WRMSR exit notifications. 8796 87978.27 KVM_CAP_X86_MSR_FILTER 8798--------------------------- 8799 8800:Architectures: x86 8801 8802This capability indicates that KVM supports that accesses to user defined MSRs 8803may be rejected. With this capability exposed, KVM exports new VM ioctl 8804KVM_X86_SET_MSR_FILTER which user space can call to specify bitmaps of MSR 8805ranges that KVM should deny access to. 8806 8807In combination with KVM_CAP_X86_USER_SPACE_MSR, this allows user space to 8808trap and emulate MSRs that are outside of the scope of KVM as well as 8809limit the attack surface on KVM's MSR emulation code. 8810 88118.30 KVM_CAP_XEN_HVM 8812-------------------- 8813 8814:Architectures: x86 8815 8816This capability indicates the features that Xen supports for hosting Xen 8817PVHVM guests. Valid flags are:: 8818 8819 #define KVM_XEN_HVM_CONFIG_HYPERCALL_MSR (1 << 0) 8820 #define KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL (1 << 1) 8821 #define KVM_XEN_HVM_CONFIG_SHARED_INFO (1 << 2) 8822 #define KVM_XEN_HVM_CONFIG_RUNSTATE (1 << 3) 8823 #define KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL (1 << 4) 8824 #define KVM_XEN_HVM_CONFIG_EVTCHN_SEND (1 << 5) 8825 #define KVM_XEN_HVM_CONFIG_RUNSTATE_UPDATE_FLAG (1 << 6) 8826 #define KVM_XEN_HVM_CONFIG_PVCLOCK_TSC_UNSTABLE (1 << 7) 8827 8828The KVM_XEN_HVM_CONFIG_HYPERCALL_MSR flag indicates that the KVM_XEN_HVM_CONFIG 8829ioctl is available, for the guest to set its hypercall page. 8830 8831If KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL is also set, the same flag may also be 8832provided in the flags to KVM_XEN_HVM_CONFIG, without providing hypercall page 8833contents, to request that KVM generate hypercall page content automatically 8834and also enable interception of guest hypercalls with KVM_EXIT_XEN. 8835 8836The KVM_XEN_HVM_CONFIG_SHARED_INFO flag indicates the availability of the 8837KVM_XEN_HVM_SET_ATTR, KVM_XEN_HVM_GET_ATTR, KVM_XEN_VCPU_SET_ATTR and 8838KVM_XEN_VCPU_GET_ATTR ioctls, as well as the delivery of exception vectors 8839for event channel upcalls when the evtchn_upcall_pending field of a vcpu's 8840vcpu_info is set. 8841 8842The KVM_XEN_HVM_CONFIG_RUNSTATE flag indicates that the runstate-related 8843features KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR/_CURRENT/_DATA/_ADJUST are 8844supported by the KVM_XEN_VCPU_SET_ATTR/KVM_XEN_VCPU_GET_ATTR ioctls. 8845 8846The KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL flag indicates that IRQ routing entries 8847of the type KVM_IRQ_ROUTING_XEN_EVTCHN are supported, with the priority 8848field set to indicate 2 level event channel delivery. 8849 8850The KVM_XEN_HVM_CONFIG_EVTCHN_SEND flag indicates that KVM supports 8851injecting event channel events directly into the guest with the 8852KVM_XEN_HVM_EVTCHN_SEND ioctl. It also indicates support for the 8853KVM_XEN_ATTR_TYPE_EVTCHN/XEN_VERSION HVM attributes and the 8854KVM_XEN_VCPU_ATTR_TYPE_VCPU_ID/TIMER/UPCALL_VECTOR vCPU attributes. 8855related to event channel delivery, timers, and the XENVER_version 8856interception. 8857 8858The KVM_XEN_HVM_CONFIG_RUNSTATE_UPDATE_FLAG flag indicates that KVM supports 8859the KVM_XEN_ATTR_TYPE_RUNSTATE_UPDATE_FLAG attribute in the KVM_XEN_SET_ATTR 8860and KVM_XEN_GET_ATTR ioctls. This controls whether KVM will set the 8861XEN_RUNSTATE_UPDATE flag in guest memory mapped vcpu_runstate_info during 8862updates of the runstate information. Note that versions of KVM which support 8863the RUNSTATE feature above, but not the RUNSTATE_UPDATE_FLAG feature, will 8864always set the XEN_RUNSTATE_UPDATE flag when updating the guest structure, 8865which is perhaps counterintuitive. When this flag is advertised, KVM will 8866behave more correctly, not using the XEN_RUNSTATE_UPDATE flag until/unless 8867specifically enabled (by the guest making the hypercall, causing the VMM 8868to enable the KVM_XEN_ATTR_TYPE_RUNSTATE_UPDATE_FLAG attribute). 8869 8870The KVM_XEN_HVM_CONFIG_PVCLOCK_TSC_UNSTABLE flag indicates that KVM supports 8871clearing the PVCLOCK_TSC_STABLE_BIT flag in Xen pvclock sources. This will be 8872done when the KVM_CAP_XEN_HVM ioctl sets the 8873KVM_XEN_HVM_CONFIG_PVCLOCK_TSC_UNSTABLE flag. 8874 88758.31 KVM_CAP_SPAPR_MULTITCE 8876--------------------------- 8877 8878:Architectures: ppc 8879:Type: vm 8880 8881This capability means the kernel is capable of handling hypercalls 8882H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user 8883space. This significantly accelerates DMA operations for PPC KVM guests. 8884User space should expect that its handlers for these hypercalls 8885are not going to be called if user space previously registered LIOBN 8886in KVM (via KVM_CREATE_SPAPR_TCE or similar calls). 8887 8888In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest, 8889user space might have to advertise it for the guest. For example, 8890IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is 8891present in the "ibm,hypertas-functions" device-tree property. 8892 8893The hypercalls mentioned above may or may not be processed successfully 8894in the kernel based fast path. If they can not be handled by the kernel, 8895they will get passed on to user space. So user space still has to have 8896an implementation for these despite the in kernel acceleration. 8897 8898This capability is always enabled. 8899 89008.32 KVM_CAP_PTP_KVM 8901-------------------- 8902 8903:Architectures: arm64 8904 8905This capability indicates that the KVM virtual PTP service is 8906supported in the host. A VMM can check whether the service is 8907available to the guest on migration. 8908 89098.37 KVM_CAP_S390_PROTECTED_DUMP 8910-------------------------------- 8911 8912:Architectures: s390 8913:Type: vm 8914 8915This capability indicates that KVM and the Ultravisor support dumping 8916PV guests. The `KVM_PV_DUMP` command is available for the 8917`KVM_S390_PV_COMMAND` ioctl and the `KVM_PV_INFO` command provides 8918dump related UV data. Also the vcpu ioctl `KVM_S390_PV_CPU_COMMAND` is 8919available and supports the `KVM_PV_DUMP_CPU` subcommand. 8920 89218.39 KVM_CAP_S390_CPU_TOPOLOGY 8922------------------------------ 8923 8924:Architectures: s390 8925:Type: vm 8926 8927This capability indicates that KVM will provide the S390 CPU Topology 8928facility which consist of the interpretation of the PTF instruction for 8929the function code 2 along with interception and forwarding of both the 8930PTF instruction with function codes 0 or 1 and the STSI(15,1,x) 8931instruction to the userland hypervisor. 8932 8933The stfle facility 11, CPU Topology facility, should not be indicated 8934to the guest without this capability. 8935 8936When this capability is present, KVM provides a new attribute group 8937on vm fd, KVM_S390_VM_CPU_TOPOLOGY. 8938This new attribute allows to get, set or clear the Modified Change 8939Topology Report (MTCR) bit of the SCA through the kvm_device_attr 8940structure. 8941 8942When getting the Modified Change Topology Report value, the attr->addr 8943must point to a byte where the value will be stored or retrieved from. 8944 89458.41 KVM_CAP_VM_TYPES 8946--------------------- 8947 8948:Architectures: x86 8949:Type: system ioctl 8950 8951This capability returns a bitmap of support VM types. The 1-setting of bit @n 8952means the VM type with value @n is supported. Possible values of @n are:: 8953 8954 #define KVM_X86_DEFAULT_VM 0 8955 #define KVM_X86_SW_PROTECTED_VM 1 8956 #define KVM_X86_SEV_VM 2 8957 #define KVM_X86_SEV_ES_VM 3 8958 8959Note, KVM_X86_SW_PROTECTED_VM is currently only for development and testing. 8960Do not use KVM_X86_SW_PROTECTED_VM for "real" VMs, and especially not in 8961production. The behavior and effective ABI for software-protected VMs is 8962unstable. 8963 89648.42 KVM_CAP_PPC_RPT_INVALIDATE 8965------------------------------- 8966 8967:Architectures: ppc 8968 8969This capability indicates that the kernel is capable of handling 8970H_RPT_INVALIDATE hcall. 8971 8972In order to enable the use of H_RPT_INVALIDATE in the guest, 8973user space might have to advertise it for the guest. For example, 8974IBM pSeries (sPAPR) guest starts using it if "hcall-rpt-invalidate" is 8975present in the "ibm,hypertas-functions" device-tree property. 8976 8977This capability is enabled for hypervisors on platforms like POWER9 8978that support radix MMU. 8979 89808.43 KVM_CAP_PPC_AIL_MODE_3 8981--------------------------- 8982 8983:Architectures: ppc 8984 8985This capability indicates that the kernel supports the mode 3 setting for the 8986"Address Translation Mode on Interrupt" aka "Alternate Interrupt Location" 8987resource that is controlled with the H_SET_MODE hypercall. 8988 8989This capability allows a guest kernel to use a better-performance mode for 8990handling interrupts and system calls. 8991 89928.44 KVM_CAP_MEMORY_FAULT_INFO 8993------------------------------ 8994 8995:Architectures: x86 8996 8997The presence of this capability indicates that KVM_RUN will fill 8998kvm_run.memory_fault if KVM cannot resolve a guest page fault VM-Exit, e.g. if 8999there is a valid memslot but no backing VMA for the corresponding host virtual 9000address. 9001 9002The information in kvm_run.memory_fault is valid if and only if KVM_RUN returns 9003an error with errno=EFAULT or errno=EHWPOISON *and* kvm_run.exit_reason is set 9004to KVM_EXIT_MEMORY_FAULT. 9005 9006Note: Userspaces which attempt to resolve memory faults so that they can retry 9007KVM_RUN are encouraged to guard against repeatedly receiving the same 9008error/annotated fault. 9009 9010See KVM_EXIT_MEMORY_FAULT for more information. 9011 90128.45 KVM_CAP_X86_GUEST_MODE 9013--------------------------- 9014 9015:Architectures: x86 9016 9017The presence of this capability indicates that KVM_RUN will update the 9018KVM_RUN_X86_GUEST_MODE bit in kvm_run.flags to indicate whether the 9019vCPU was executing nested guest code when it exited. 9020 9021KVM exits with the register state of either the L1 or L2 guest 9022depending on which executed at the time of an exit. Userspace must 9023take care to differentiate between these cases. 9024 90259. Known KVM API problems 9026========================= 9027 9028In some cases, KVM's API has some inconsistencies or common pitfalls 9029that userspace need to be aware of. This section details some of 9030these issues. 9031 9032Most of them are architecture specific, so the section is split by 9033architecture. 9034 90359.1. x86 9036-------- 9037 9038``KVM_GET_SUPPORTED_CPUID`` issues 9039^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9040 9041In general, ``KVM_GET_SUPPORTED_CPUID`` is designed so that it is possible 9042to take its result and pass it directly to ``KVM_SET_CPUID2``. This section 9043documents some cases in which that requires some care. 9044 9045Local APIC features 9046~~~~~~~~~~~~~~~~~~~ 9047 9048CPU[EAX=1]:ECX[21] (X2APIC) is reported by ``KVM_GET_SUPPORTED_CPUID``, 9049but it can only be enabled if ``KVM_CREATE_IRQCHIP`` or 9050``KVM_ENABLE_CAP(KVM_CAP_IRQCHIP_SPLIT)`` are used to enable in-kernel emulation of 9051the local APIC. 9052 9053The same is true for the ``KVM_FEATURE_PV_UNHALT`` paravirtualized feature. 9054 9055On older versions of Linux, CPU[EAX=1]:ECX[24] (TSC_DEADLINE) is not reported by 9056``KVM_GET_SUPPORTED_CPUID``, but it can be enabled if ``KVM_CAP_TSC_DEADLINE_TIMER`` 9057is present and the kernel has enabled in-kernel emulation of the local APIC. 9058On newer versions, ``KVM_GET_SUPPORTED_CPUID`` does report the bit as available. 9059 9060CPU topology 9061~~~~~~~~~~~~ 9062 9063Several CPUID values include topology information for the host CPU: 90640x0b and 0x1f for Intel systems, 0x8000001e for AMD systems. Different 9065versions of KVM return different values for this information and userspace 9066should not rely on it. Currently they return all zeroes. 9067 9068If userspace wishes to set up a guest topology, it should be careful that 9069the values of these three leaves differ for each CPU. In particular, 9070the APIC ID is found in EDX for all subleaves of 0x0b and 0x1f, and in EAX 9071for 0x8000001e; the latter also encodes the core id and node id in bits 90727:0 of EBX and ECX respectively. 9073 9074Obsolete ioctls and capabilities 9075^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9076 9077KVM_CAP_DISABLE_QUIRKS does not let userspace know which quirks are actually 9078available. Use ``KVM_CHECK_EXTENSION(KVM_CAP_DISABLE_QUIRKS2)`` instead if 9079available. 9080 9081Ordering of KVM_GET_*/KVM_SET_* ioctls 9082^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9083 9084TBD 9085