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 GICv2/GICv3: SPI, irq_id between 32 and 1019 (incl.) 911 (the vcpu_index field is ignored) 912 in-kernel GICv5: SPI, irq_id between 0 and 65535 (incl.) 913- KVM_ARM_IRQ_TYPE_PPI: 914 in-kernel GICv2/GICv3: PPI, irq_id between 16 and 31 (incl.) 915 in-kernel GICv5: PPI, irq_id between 0 and 127 (incl.) 916 917(The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs) 918 919In both cases, level is used to assert/deassert the line. 920 921When KVM_CAP_ARM_IRQ_LINE_LAYOUT_2 is supported, the target vcpu is 922identified as (256 * vcpu2_index + vcpu_index). Otherwise, vcpu2_index 923must be zero. 924 925Note that on arm64, the KVM_CAP_IRQCHIP capability only conditions 926injection of interrupts for the in-kernel irqchip. KVM_IRQ_LINE can always 927be used for a userspace interrupt controller. 928 929:: 930 931 struct kvm_irq_level { 932 union { 933 __u32 irq; /* GSI */ 934 __s32 status; /* not used for KVM_IRQ_LEVEL */ 935 }; 936 __u32 level; /* 0 or 1 */ 937 }; 938 939 9404.26 KVM_GET_IRQCHIP 941-------------------- 942 943:Capability: KVM_CAP_IRQCHIP 944:Architectures: x86 945:Type: vm ioctl 946:Parameters: struct kvm_irqchip (in/out) 947:Returns: 0 on success, -1 on error 948 949Reads the state of a kernel interrupt controller created with 950KVM_CREATE_IRQCHIP into a buffer provided by the caller. 951 952:: 953 954 struct kvm_irqchip { 955 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */ 956 __u32 pad; 957 union { 958 char dummy[512]; /* reserving space */ 959 struct kvm_pic_state pic; 960 struct kvm_ioapic_state ioapic; 961 } chip; 962 }; 963 964 9654.27 KVM_SET_IRQCHIP 966-------------------- 967 968:Capability: KVM_CAP_IRQCHIP 969:Architectures: x86 970:Type: vm ioctl 971:Parameters: struct kvm_irqchip (in) 972:Returns: 0 on success, -1 on error 973 974Sets the state of a kernel interrupt controller created with 975KVM_CREATE_IRQCHIP from a buffer provided by the caller. 976 977:: 978 979 struct kvm_irqchip { 980 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */ 981 __u32 pad; 982 union { 983 char dummy[512]; /* reserving space */ 984 struct kvm_pic_state pic; 985 struct kvm_ioapic_state ioapic; 986 } chip; 987 }; 988 989 9904.28 KVM_XEN_HVM_CONFIG 991----------------------- 992 993:Capability: KVM_CAP_XEN_HVM 994:Architectures: x86 995:Type: vm ioctl 996:Parameters: struct kvm_xen_hvm_config (in) 997:Returns: 0 on success, -1 on error 998 999Sets the MSR that the Xen HVM guest uses to initialize its hypercall 1000page, and provides the starting address and size of the hypercall 1001blobs in userspace. When the guest writes the MSR, kvm copies one 1002page of a blob (32- or 64-bit, depending on the vcpu mode) to guest 1003memory. 1004 1005The MSR index must be in the range [0x40000000, 0x4fffffff], i.e. must reside 1006in the range that is unofficially reserved for use by hypervisors. The min/max 1007values are enumerated via KVM_XEN_MSR_MIN_INDEX and KVM_XEN_MSR_MAX_INDEX. 1008 1009:: 1010 1011 struct kvm_xen_hvm_config { 1012 __u32 flags; 1013 __u32 msr; 1014 __u64 blob_addr_32; 1015 __u64 blob_addr_64; 1016 __u8 blob_size_32; 1017 __u8 blob_size_64; 1018 __u8 pad2[30]; 1019 }; 1020 1021If certain flags are returned from the KVM_CAP_XEN_HVM check, they may 1022be set in the flags field of this ioctl: 1023 1024The KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL flag requests KVM to generate 1025the contents of the hypercall page automatically; hypercalls will be 1026intercepted and passed to userspace through KVM_EXIT_XEN. In this 1027case, all of the blob size and address fields must be zero. 1028 1029The KVM_XEN_HVM_CONFIG_EVTCHN_SEND flag indicates to KVM that userspace 1030will always use the KVM_XEN_HVM_EVTCHN_SEND ioctl to deliver event 1031channel interrupts rather than manipulating the guest's shared_info 1032structures directly. This, in turn, may allow KVM to enable features 1033such as intercepting the SCHEDOP_poll hypercall to accelerate PV 1034spinlock operation for the guest. Userspace may still use the ioctl 1035to deliver events if it was advertised, even if userspace does not 1036send this indication that it will always do so 1037 1038No other flags are currently valid in the struct kvm_xen_hvm_config. 1039 10404.29 KVM_GET_CLOCK 1041------------------ 1042 1043:Capability: KVM_CAP_ADJUST_CLOCK 1044:Architectures: x86 1045:Type: vm ioctl 1046:Parameters: struct kvm_clock_data (out) 1047:Returns: 0 on success, -1 on error 1048 1049Gets the current timestamp of kvmclock as seen by the current guest. In 1050conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios 1051such as migration. 1052 1053When KVM_CAP_ADJUST_CLOCK is passed to KVM_CHECK_EXTENSION, it returns the 1054set of bits that KVM can return in struct kvm_clock_data's flag member. 1055 1056The following flags are defined: 1057 1058KVM_CLOCK_TSC_STABLE 1059 If set, the returned value is the exact kvmclock 1060 value seen by all VCPUs at the instant when KVM_GET_CLOCK was called. 1061 If clear, the returned value is simply CLOCK_MONOTONIC plus a constant 1062 offset; the offset can be modified with KVM_SET_CLOCK. KVM will try 1063 to make all VCPUs follow this clock, but the exact value read by each 1064 VCPU could differ, because the host TSC is not stable. 1065 1066KVM_CLOCK_REALTIME 1067 If set, the `realtime` field in the kvm_clock_data 1068 structure is populated with the value of the host's real time 1069 clocksource at the instant when KVM_GET_CLOCK was called. If clear, 1070 the `realtime` field does not contain a value. 1071 1072KVM_CLOCK_HOST_TSC 1073 If set, the `host_tsc` field in the kvm_clock_data 1074 structure is populated with the value of the host's timestamp counter (TSC) 1075 at the instant when KVM_GET_CLOCK was called. If clear, the `host_tsc` field 1076 does not contain a value. 1077 1078:: 1079 1080 struct kvm_clock_data { 1081 __u64 clock; /* kvmclock current value */ 1082 __u32 flags; 1083 __u32 pad0; 1084 __u64 realtime; 1085 __u64 host_tsc; 1086 __u32 pad[4]; 1087 }; 1088 1089 10904.30 KVM_SET_CLOCK 1091------------------ 1092 1093:Capability: KVM_CAP_ADJUST_CLOCK 1094:Architectures: x86 1095:Type: vm ioctl 1096:Parameters: struct kvm_clock_data (in) 1097:Returns: 0 on success, -1 on error 1098 1099Sets the current timestamp of kvmclock to the value specified in its parameter. 1100In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios 1101such as migration. 1102 1103The following flags can be passed: 1104 1105KVM_CLOCK_REALTIME 1106 If set, KVM will compare the value of the `realtime` field 1107 with the value of the host's real time clocksource at the instant when 1108 KVM_SET_CLOCK was called. The difference in elapsed time is added to the final 1109 kvmclock value that will be provided to guests. 1110 1111Other flags returned by ``KVM_GET_CLOCK`` are accepted but ignored. 1112 1113:: 1114 1115 struct kvm_clock_data { 1116 __u64 clock; /* kvmclock current value */ 1117 __u32 flags; 1118 __u32 pad0; 1119 __u64 realtime; 1120 __u64 host_tsc; 1121 __u32 pad[4]; 1122 }; 1123 1124 11254.31 KVM_GET_VCPU_EVENTS 1126------------------------ 1127 1128:Capability: KVM_CAP_VCPU_EVENTS 1129:Extended by: KVM_CAP_INTR_SHADOW 1130:Architectures: x86, arm64 1131:Type: vcpu ioctl 1132:Parameters: struct kvm_vcpu_events (out) 1133:Returns: 0 on success, -1 on error 1134 1135X86: 1136^^^^ 1137 1138Gets currently pending exceptions, interrupts, and NMIs as well as related 1139states of the vcpu. 1140 1141:: 1142 1143 struct kvm_vcpu_events { 1144 struct { 1145 __u8 injected; 1146 __u8 nr; 1147 __u8 has_error_code; 1148 __u8 pending; 1149 __u32 error_code; 1150 } exception; 1151 struct { 1152 __u8 injected; 1153 __u8 nr; 1154 __u8 soft; 1155 __u8 shadow; 1156 } interrupt; 1157 struct { 1158 __u8 injected; 1159 __u8 pending; 1160 __u8 masked; 1161 __u8 pad; 1162 } nmi; 1163 __u32 sipi_vector; 1164 __u32 flags; 1165 struct { 1166 __u8 smm; 1167 __u8 pending; 1168 __u8 smm_inside_nmi; 1169 __u8 latched_init; 1170 } smi; 1171 __u8 reserved[27]; 1172 __u8 exception_has_payload; 1173 __u64 exception_payload; 1174 }; 1175 1176The following bits are defined in the flags field: 1177 1178- KVM_VCPUEVENT_VALID_SHADOW may be set to signal that 1179 interrupt.shadow contains a valid state. 1180 1181- KVM_VCPUEVENT_VALID_SMM may be set to signal that smi contains a 1182 valid state. 1183 1184- KVM_VCPUEVENT_VALID_PAYLOAD may be set to signal that the 1185 exception_has_payload, exception_payload, and exception.pending 1186 fields contain a valid state. This bit will be set whenever 1187 KVM_CAP_EXCEPTION_PAYLOAD is enabled. 1188 1189- KVM_VCPUEVENT_VALID_TRIPLE_FAULT may be set to signal that the 1190 triple_fault_pending field contains a valid state. This bit will 1191 be set whenever KVM_CAP_X86_TRIPLE_FAULT_EVENT is enabled. 1192 1193ARM64: 1194^^^^^^ 1195 1196If the guest accesses a device that is being emulated by the host kernel in 1197such a way that a real device would generate a physical SError, KVM may make 1198a virtual SError pending for that VCPU. This system error interrupt remains 1199pending until the guest takes the exception by unmasking PSTATE.A. 1200 1201Running the VCPU may cause it to take a pending SError, or make an access that 1202causes an SError to become pending. The event's description is only valid while 1203the VPCU is not running. 1204 1205This API provides a way to read and write the pending 'event' state that is not 1206visible to the guest. To save, restore or migrate a VCPU the struct representing 1207the state can be read then written using this GET/SET API, along with the other 1208guest-visible registers. It is not possible to 'cancel' an SError that has been 1209made pending. 1210 1211A device being emulated in user-space may also wish to generate an SError. To do 1212this the events structure can be populated by user-space. The current state 1213should be read first, to ensure no existing SError is pending. If an existing 1214SError is pending, the architecture's 'Multiple SError interrupts' rules should 1215be followed. (2.5.3 of DDI0587.a "ARM Reliability, Availability, and 1216Serviceability (RAS) Specification"). 1217 1218SError exceptions always have an ESR value. Some CPUs have the ability to 1219specify what the virtual SError's ESR value should be. These systems will 1220advertise KVM_CAP_ARM_INJECT_SERROR_ESR. In this case exception.has_esr will 1221always have a non-zero value when read, and the agent making an SError pending 1222should specify the ISS field in the lower 24 bits of exception.serror_esr. If 1223the system supports KVM_CAP_ARM_INJECT_SERROR_ESR, but user-space sets the events 1224with exception.has_esr as zero, KVM will choose an ESR. 1225 1226Specifying exception.has_esr on a system that does not support it will return 1227-EINVAL. Setting anything other than the lower 24bits of exception.serror_esr 1228will return -EINVAL. 1229 1230It is not possible to read back a pending external abort (injected via 1231KVM_SET_VCPU_EVENTS or otherwise) because such an exception is always delivered 1232directly to the virtual CPU). 1233 1234Calling this ioctl on a vCPU that hasn't been initialized will return 1235-ENOEXEC. 1236 1237:: 1238 1239 struct kvm_vcpu_events { 1240 struct { 1241 __u8 serror_pending; 1242 __u8 serror_has_esr; 1243 __u8 ext_dabt_pending; 1244 /* Align it to 8 bytes */ 1245 __u8 pad[5]; 1246 __u64 serror_esr; 1247 } exception; 1248 __u32 reserved[12]; 1249 }; 1250 12514.32 KVM_SET_VCPU_EVENTS 1252------------------------ 1253 1254:Capability: KVM_CAP_VCPU_EVENTS 1255:Extended by: KVM_CAP_INTR_SHADOW 1256:Architectures: x86, arm64 1257:Type: vcpu ioctl 1258:Parameters: struct kvm_vcpu_events (in) 1259:Returns: 0 on success, -1 on error 1260 1261X86: 1262^^^^ 1263 1264Set pending exceptions, interrupts, and NMIs as well as related states of the 1265vcpu. 1266 1267See KVM_GET_VCPU_EVENTS for the data structure. 1268 1269Fields that may be modified asynchronously by running VCPUs can be excluded 1270from the update. These fields are nmi.pending, sipi_vector, smi.smm, 1271smi.pending. Keep the corresponding bits in the flags field cleared to 1272suppress overwriting the current in-kernel state. The bits are: 1273 1274=============================== ================================== 1275KVM_VCPUEVENT_VALID_NMI_PENDING transfer nmi.pending to the kernel 1276KVM_VCPUEVENT_VALID_SIPI_VECTOR transfer sipi_vector 1277KVM_VCPUEVENT_VALID_SMM transfer the smi sub-struct. 1278=============================== ================================== 1279 1280If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in 1281the flags field to signal that interrupt.shadow contains a valid state and 1282shall be written into the VCPU. 1283 1284KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available. 1285 1286If KVM_CAP_EXCEPTION_PAYLOAD is enabled, KVM_VCPUEVENT_VALID_PAYLOAD 1287can be set in the flags field to signal that the 1288exception_has_payload, exception_payload, and exception.pending fields 1289contain a valid state and shall be written into the VCPU. 1290 1291If KVM_CAP_X86_TRIPLE_FAULT_EVENT is enabled, KVM_VCPUEVENT_VALID_TRIPLE_FAULT 1292can be set in flags field to signal that the triple_fault field contains 1293a valid state and shall be written into the VCPU. 1294 1295ARM64: 1296^^^^^^ 1297 1298User space may need to inject several types of events to the guest. 1299 1300Set the pending SError exception state for this VCPU. It is not possible to 1301'cancel' an Serror that has been made pending. 1302 1303If the guest performed an access to I/O memory which could not be handled by 1304userspace, for example because of missing instruction syndrome decode 1305information or because there is no device mapped at the accessed IPA, then 1306userspace can ask the kernel to inject an external abort using the address 1307from the exiting fault on the VCPU. It is a programming error to set 1308ext_dabt_pending after an exit which was not either KVM_EXIT_MMIO, 1309KVM_EXIT_ARM_NISV, or KVM_EXIT_ARM_LDST64B. This feature is only available if 1310the system supports KVM_CAP_ARM_INJECT_EXT_DABT. This is a helper which 1311provides commonality in how userspace reports accesses for the above cases to 1312guests, across different userspace implementations. Nevertheless, userspace 1313can still emulate all Arm exceptions by manipulating individual registers 1314using the KVM_SET_ONE_REG API. 1315 1316See KVM_GET_VCPU_EVENTS for the data structure. 1317 1318Calling this ioctl on a vCPU that hasn't been initialized will return 1319-ENOEXEC. 1320 13214.33 KVM_GET_DEBUGREGS 1322---------------------- 1323 1324:Capability: KVM_CAP_DEBUGREGS 1325:Architectures: x86 1326:Type: vcpu ioctl 1327:Parameters: struct kvm_debugregs (out) 1328:Returns: 0 on success, -1 on error 1329 1330Reads debug registers from the vcpu. 1331 1332:: 1333 1334 struct kvm_debugregs { 1335 __u64 db[4]; 1336 __u64 dr6; 1337 __u64 dr7; 1338 __u64 flags; 1339 __u64 reserved[9]; 1340 }; 1341 1342 13434.34 KVM_SET_DEBUGREGS 1344---------------------- 1345 1346:Capability: KVM_CAP_DEBUGREGS 1347:Architectures: x86 1348:Type: vcpu ioctl 1349:Parameters: struct kvm_debugregs (in) 1350:Returns: 0 on success, -1 on error 1351 1352Writes debug registers into the vcpu. 1353 1354See KVM_GET_DEBUGREGS for the data structure. The flags field is unused 1355yet and must be cleared on entry. 1356 1357 13584.35 KVM_SET_USER_MEMORY_REGION 1359------------------------------- 1360 1361:Capability: KVM_CAP_USER_MEMORY 1362:Architectures: all 1363:Type: vm ioctl 1364:Parameters: struct kvm_userspace_memory_region (in) 1365:Returns: 0 on success, -1 on error 1366 1367:: 1368 1369 struct kvm_userspace_memory_region { 1370 __u32 slot; 1371 __u32 flags; 1372 __u64 guest_phys_addr; 1373 __u64 memory_size; /* bytes */ 1374 __u64 userspace_addr; /* start of the userspace allocated memory */ 1375 }; 1376 1377 /* for kvm_userspace_memory_region::flags */ 1378 #define KVM_MEM_LOG_DIRTY_PAGES (1UL << 0) 1379 #define KVM_MEM_READONLY (1UL << 1) 1380 1381This ioctl allows the user to create, modify or delete a guest physical 1382memory slot. Bits 0-15 of "slot" specify the slot id and this value 1383should be less than the maximum number of user memory slots supported per 1384VM. The maximum allowed slots can be queried using KVM_CAP_NR_MEMSLOTS. 1385Slots may not overlap in guest physical address space. 1386 1387If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot" 1388specifies the address space which is being modified. They must be 1389less than the value that KVM_CHECK_EXTENSION returns for the 1390KVM_CAP_MULTI_ADDRESS_SPACE capability. Slots in separate address spaces 1391are unrelated; the restriction on overlapping slots only applies within 1392each address space. 1393 1394Deleting a slot is done by passing zero for memory_size. When changing 1395an existing slot, it may be moved in the guest physical memory space, 1396or its flags may be modified, but it may not be resized. 1397 1398Memory for the region is taken starting at the address denoted by the 1399field userspace_addr, which must point at user addressable memory for 1400the entire memory slot size. Any object may back this memory, including 1401anonymous memory, ordinary files, and hugetlbfs. Changes in the backing 1402of the memory region are automatically reflected into the guest. 1403For example, an mmap() that affects the region will be made visible 1404immediately. Another example is madvise(MADV_DROP). 1405 1406On architectures that support a form of address tagging, userspace_addr must 1407be an untagged address. 1408 1409It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr 1410be identical. This allows large pages in the guest to be backed by large 1411pages in the host. 1412 1413The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and 1414KVM_MEM_READONLY. The former can be set to instruct KVM to keep track of 1415writes to memory within the slot. See KVM_GET_DIRTY_LOG ioctl to know how to 1416use it. The latter can be set, if KVM_CAP_READONLY_MEM capability allows it, 1417to make a new slot read-only. In this case, writes to this memory will be 1418posted to userspace as KVM_EXIT_MMIO exits. 1419 1420For TDX guest, deleting/moving memory region loses guest memory contents. 1421Read only region isn't supported. Only as-id 0 is supported. 1422 1423Note: On arm64, a write generated by the page-table walker (to update 1424the Access and Dirty flags, for example) never results in a 1425KVM_EXIT_MMIO exit when the slot has the KVM_MEM_READONLY flag. This 1426is because KVM cannot provide the data that would be written by the 1427page-table walker, making it impossible to emulate the access. 1428Instead, an abort (data abort if the cause of the page-table update 1429was a load or a store, instruction abort if it was an instruction 1430fetch) is injected in the guest. 1431 1432S390: 1433^^^^^ 1434 1435Returns -EINVAL or -EEXIST if the VM has the KVM_VM_S390_UCONTROL flag set. 1436Returns -EINVAL if called on a protected VM. 1437 14384.36 KVM_SET_TSS_ADDR 1439--------------------- 1440 1441:Capability: KVM_CAP_SET_TSS_ADDR 1442:Architectures: x86 1443:Type: vm ioctl 1444:Parameters: unsigned long tss_address (in) 1445:Returns: 0 on success, -1 on error 1446 1447This ioctl defines the physical address of a three-page region in the guest 1448physical address space. The region must be within the first 4GB of the 1449guest physical address space and must not conflict with any memory slot 1450or any mmio address. The guest may malfunction if it accesses this memory 1451region. 1452 1453This ioctl is required on Intel-based hosts. This is needed on Intel hardware 1454because of a quirk in the virtualization implementation (see the internals 1455documentation when it pops into existence). 1456 1457 1458.. _KVM_ENABLE_CAP: 1459 14604.37 KVM_ENABLE_CAP 1461------------------- 1462 1463:Capability: KVM_CAP_ENABLE_CAP 1464:Architectures: mips, ppc, s390, x86, loongarch 1465:Type: vcpu ioctl 1466:Parameters: struct kvm_enable_cap (in) 1467:Returns: 0 on success; -1 on error 1468 1469:Capability: KVM_CAP_ENABLE_CAP_VM 1470:Architectures: all 1471:Type: vm ioctl 1472:Parameters: struct kvm_enable_cap (in) 1473:Returns: 0 on success; -1 on error 1474 1475.. note:: 1476 1477 Not all extensions are enabled by default. Using this ioctl the application 1478 can enable an extension, making it available to the guest. 1479 1480On systems that do not support this ioctl, it always fails. On systems that 1481do support it, it only works for extensions that are supported for enablement. 1482 1483To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should 1484be used. 1485 1486:: 1487 1488 struct kvm_enable_cap { 1489 /* in */ 1490 __u32 cap; 1491 1492The capability that is supposed to get enabled. 1493 1494:: 1495 1496 __u32 flags; 1497 1498A bitfield indicating future enhancements. Has to be 0 for now. 1499 1500:: 1501 1502 __u64 args[4]; 1503 1504Arguments for enabling a feature. If a feature needs initial values to 1505function properly, this is the place to put them. 1506 1507:: 1508 1509 __u8 pad[64]; 1510 }; 1511 1512The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl 1513for vm-wide capabilities. 1514 15154.38 KVM_GET_MP_STATE 1516--------------------- 1517 1518:Capability: KVM_CAP_MP_STATE 1519:Architectures: x86, s390, arm64, riscv, loongarch 1520:Type: vcpu ioctl 1521:Parameters: struct kvm_mp_state (out) 1522:Returns: 0 on success; -1 on error 1523 1524:: 1525 1526 struct kvm_mp_state { 1527 __u32 mp_state; 1528 }; 1529 1530Returns the vcpu's current "multiprocessing state" (though also valid on 1531uniprocessor guests). 1532 1533Possible values are: 1534 1535 ========================== =============================================== 1536 KVM_MP_STATE_RUNNABLE the vcpu is currently running 1537 [x86,arm64,riscv,loongarch] 1538 KVM_MP_STATE_UNINITIALIZED the vcpu is an application processor (AP) 1539 which has not yet received an INIT signal [x86] 1540 KVM_MP_STATE_INIT_RECEIVED the vcpu has received an INIT signal, and is 1541 now ready for a SIPI [x86] 1542 KVM_MP_STATE_HALTED the vcpu has executed a HLT instruction and 1543 is waiting for an interrupt [x86] 1544 KVM_MP_STATE_SIPI_RECEIVED the vcpu has just received a SIPI (vector 1545 accessible via KVM_GET_VCPU_EVENTS) [x86] 1546 KVM_MP_STATE_STOPPED the vcpu is stopped [s390,arm64,riscv] 1547 KVM_MP_STATE_CHECK_STOP the vcpu is in a special error state [s390] 1548 KVM_MP_STATE_OPERATING the vcpu is operating (running or halted) 1549 [s390] 1550 KVM_MP_STATE_LOAD the vcpu is in a special load/startup state 1551 [s390] 1552 KVM_MP_STATE_SUSPENDED the vcpu is in a suspend state and is waiting 1553 for a wakeup event [arm64] 1554 ========================== =============================================== 1555 1556On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an 1557in-kernel irqchip, the multiprocessing state must be maintained by userspace on 1558these architectures. 1559 1560For arm64: 1561^^^^^^^^^^ 1562 1563If a vCPU is in the KVM_MP_STATE_SUSPENDED state, KVM will emulate the 1564architectural execution of a WFI instruction. 1565 1566If a wakeup event is recognized, KVM will exit to userspace with a 1567KVM_SYSTEM_EVENT exit, where the event type is KVM_SYSTEM_EVENT_WAKEUP. If 1568userspace wants to honor the wakeup, it must set the vCPU's MP state to 1569KVM_MP_STATE_RUNNABLE. If it does not, KVM will continue to await a wakeup 1570event in subsequent calls to KVM_RUN. 1571 1572.. warning:: 1573 1574 If userspace intends to keep the vCPU in a SUSPENDED state, it is 1575 strongly recommended that userspace take action to suppress the 1576 wakeup event (such as masking an interrupt). Otherwise, subsequent 1577 calls to KVM_RUN will immediately exit with a KVM_SYSTEM_EVENT_WAKEUP 1578 event and inadvertently waste CPU cycles. 1579 1580 Additionally, if userspace takes action to suppress a wakeup event, 1581 it is strongly recommended that it also restores the vCPU to its 1582 original state when the vCPU is made RUNNABLE again. For example, 1583 if userspace masked a pending interrupt to suppress the wakeup, 1584 the interrupt should be unmasked before returning control to the 1585 guest. 1586 1587For riscv: 1588^^^^^^^^^^ 1589 1590The only states that are valid are KVM_MP_STATE_STOPPED and 1591KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not. 1592 1593On LoongArch, only the KVM_MP_STATE_RUNNABLE state is used to reflect 1594whether the vcpu is runnable. 1595 15964.39 KVM_SET_MP_STATE 1597--------------------- 1598 1599:Capability: KVM_CAP_MP_STATE 1600:Architectures: x86, s390, arm64, riscv, loongarch 1601:Type: vcpu ioctl 1602:Parameters: struct kvm_mp_state (in) 1603:Returns: 0 on success; -1 on error 1604 1605Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for 1606arguments. 1607 1608On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an 1609in-kernel irqchip, the multiprocessing state must be maintained by userspace on 1610these architectures. 1611 1612For arm64/riscv: 1613^^^^^^^^^^^^^^^^ 1614 1615The only states that are valid are KVM_MP_STATE_STOPPED and 1616KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not. 1617 1618On LoongArch, only the KVM_MP_STATE_RUNNABLE state is used to reflect 1619whether the vcpu is runnable. 1620 16214.40 KVM_SET_IDENTITY_MAP_ADDR 1622------------------------------ 1623 1624:Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR 1625:Architectures: x86 1626:Type: vm ioctl 1627:Parameters: unsigned long identity (in) 1628:Returns: 0 on success, -1 on error 1629 1630This ioctl defines the physical address of a one-page region in the guest 1631physical address space. The region must be within the first 4GB of the 1632guest physical address space and must not conflict with any memory slot 1633or any mmio address. The guest may malfunction if it accesses this memory 1634region. 1635 1636Setting the address to 0 will result in resetting the address to its default 1637(0xfffbc000). 1638 1639This ioctl is required on Intel-based hosts. This is needed on Intel hardware 1640because of a quirk in the virtualization implementation (see the internals 1641documentation when it pops into existence). 1642 1643Fails if any VCPU has already been created. 1644 16454.41 KVM_SET_BOOT_CPU_ID 1646------------------------ 1647 1648:Capability: KVM_CAP_SET_BOOT_CPU_ID 1649:Architectures: x86 1650:Type: vm ioctl 1651:Parameters: unsigned long vcpu_id 1652:Returns: 0 on success, -1 on error 1653 1654Define which vcpu is the Bootstrap Processor (BSP). Values are the same 1655as the vcpu id in KVM_CREATE_VCPU. If this ioctl is not called, the default 1656is vcpu 0. This ioctl has to be called before vcpu creation, 1657otherwise it will return EBUSY error. 1658 1659 16604.42 KVM_GET_XSAVE 1661------------------ 1662 1663:Capability: KVM_CAP_XSAVE 1664:Architectures: x86 1665:Type: vcpu ioctl 1666:Parameters: struct kvm_xsave (out) 1667:Returns: 0 on success, -1 on error 1668 1669 1670:: 1671 1672 struct kvm_xsave { 1673 __u32 region[1024]; 1674 __u32 extra[0]; 1675 }; 1676 1677This ioctl would copy current vcpu's xsave struct to the userspace. 1678 1679 16804.43 KVM_SET_XSAVE 1681------------------ 1682 1683:Capability: KVM_CAP_XSAVE and KVM_CAP_XSAVE2 1684:Architectures: x86 1685:Type: vcpu ioctl 1686:Parameters: struct kvm_xsave (in) 1687:Returns: 0 on success, -1 on error 1688 1689:: 1690 1691 1692 struct kvm_xsave { 1693 __u32 region[1024]; 1694 __u32 extra[0]; 1695 }; 1696 1697This ioctl would copy userspace's xsave struct to the kernel. It copies 1698as many bytes as are returned by KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2), 1699when invoked on the vm file descriptor. The size value returned by 1700KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) will always be at least 4096. 1701Currently, it is only greater than 4096 if a dynamic feature has been 1702enabled with ``arch_prctl()``, but this may change in the future. 1703 1704The offsets of the state save areas in struct kvm_xsave follow the 1705contents of CPUID leaf 0xD on the host. 1706 1707 17084.44 KVM_GET_XCRS 1709----------------- 1710 1711:Capability: KVM_CAP_XCRS 1712:Architectures: x86 1713:Type: vcpu ioctl 1714:Parameters: struct kvm_xcrs (out) 1715:Returns: 0 on success, -1 on error 1716 1717:: 1718 1719 struct kvm_xcr { 1720 __u32 xcr; 1721 __u32 reserved; 1722 __u64 value; 1723 }; 1724 1725 struct kvm_xcrs { 1726 __u32 nr_xcrs; 1727 __u32 flags; 1728 struct kvm_xcr xcrs[KVM_MAX_XCRS]; 1729 __u64 padding[16]; 1730 }; 1731 1732This ioctl would copy current vcpu's xcrs to the userspace. 1733 1734 17354.45 KVM_SET_XCRS 1736----------------- 1737 1738:Capability: KVM_CAP_XCRS 1739:Architectures: x86 1740:Type: vcpu ioctl 1741:Parameters: struct kvm_xcrs (in) 1742:Returns: 0 on success, -1 on error 1743 1744:: 1745 1746 struct kvm_xcr { 1747 __u32 xcr; 1748 __u32 reserved; 1749 __u64 value; 1750 }; 1751 1752 struct kvm_xcrs { 1753 __u32 nr_xcrs; 1754 __u32 flags; 1755 struct kvm_xcr xcrs[KVM_MAX_XCRS]; 1756 __u64 padding[16]; 1757 }; 1758 1759This ioctl would set vcpu's xcr to the value userspace specified. 1760 1761 17624.46 KVM_GET_SUPPORTED_CPUID 1763---------------------------- 1764 1765:Capability: KVM_CAP_EXT_CPUID 1766:Architectures: x86 1767:Type: system ioctl 1768:Parameters: struct kvm_cpuid2 (in/out) 1769:Returns: 0 on success, -1 on error 1770 1771:: 1772 1773 struct kvm_cpuid2 { 1774 __u32 nent; 1775 __u32 padding; 1776 struct kvm_cpuid_entry2 entries[0]; 1777 }; 1778 1779 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0) 1780 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */ 1781 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */ 1782 1783 struct kvm_cpuid_entry2 { 1784 __u32 function; 1785 __u32 index; 1786 __u32 flags; 1787 __u32 eax; 1788 __u32 ebx; 1789 __u32 ecx; 1790 __u32 edx; 1791 __u32 padding[3]; 1792 }; 1793 1794This ioctl returns x86 cpuid features which are supported by both the 1795hardware and kvm in its default configuration. Userspace can use the 1796information returned by this ioctl to construct cpuid information (for 1797KVM_SET_CPUID2) that is consistent with hardware, kernel, and 1798userspace capabilities, and with user requirements (for example, the 1799user may wish to constrain cpuid to emulate older hardware, or for 1800feature consistency across a cluster). 1801 1802Dynamically-enabled feature bits need to be requested with 1803``arch_prctl()`` before calling this ioctl. Feature bits that have not 1804been requested are excluded from the result. 1805 1806Note that certain capabilities, such as KVM_CAP_X86_DISABLE_EXITS, may 1807expose cpuid features (e.g. MONITOR) which are not supported by kvm in 1808its default configuration. If userspace enables such capabilities, it 1809is responsible for modifying the results of this ioctl appropriately. 1810 1811Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure 1812with the 'nent' field indicating the number of entries in the variable-size 1813array 'entries'. If the number of entries is too low to describe the cpu 1814capabilities, an error (E2BIG) is returned. If the number is too high, 1815the 'nent' field is adjusted and an error (ENOMEM) is returned. If the 1816number is just right, the 'nent' field is adjusted to the number of valid 1817entries in the 'entries' array, which is then filled. 1818 1819The entries returned are the host cpuid as returned by the cpuid instruction, 1820with unknown or unsupported features masked out. Some features (for example, 1821x2apic), may not be present in the host cpu, but are exposed by kvm if it can 1822emulate them efficiently. The fields in each entry are defined as follows: 1823 1824 function: 1825 the eax value used to obtain the entry 1826 1827 index: 1828 the ecx value used to obtain the entry (for entries that are 1829 affected by ecx) 1830 1831 flags: 1832 an OR of zero or more of the following: 1833 1834 KVM_CPUID_FLAG_SIGNIFCANT_INDEX: 1835 if the index field is valid 1836 1837 eax, ebx, ecx, edx: 1838 the values returned by the cpuid instruction for 1839 this function/index combination 1840 1841x2APIC (CPUID leaf 1, ecx[21) and TSC deadline timer (CPUID leaf 1, ecx[24]) 1842may be returned as true, but they depend on KVM_CREATE_IRQCHIP for in-kernel 1843emulation of the local APIC. TSC deadline timer support is also reported via:: 1844 1845 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER) 1846 1847if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the 1848feature in userspace, then you can enable the feature for KVM_SET_CPUID2. 1849 1850Enabling x2APIC in KVM_SET_CPUID2 requires KVM_CREATE_IRQCHIP as KVM doesn't 1851support forwarding x2APIC MSR accesses to userspace, i.e. KVM does not support 1852emulating x2APIC in userspace. 1853 18544.47 KVM_PPC_GET_PVINFO 1855----------------------- 1856 1857:Capability: KVM_CAP_PPC_GET_PVINFO 1858:Architectures: ppc 1859:Type: vm ioctl 1860:Parameters: struct kvm_ppc_pvinfo (out) 1861:Returns: 0 on success, !0 on error 1862 1863:: 1864 1865 struct kvm_ppc_pvinfo { 1866 __u32 flags; 1867 __u32 hcall[4]; 1868 __u8 pad[108]; 1869 }; 1870 1871This ioctl fetches PV specific information that need to be passed to the guest 1872using the device tree or other means from vm context. 1873 1874The hcall array defines 4 instructions that make up a hypercall. 1875 1876If any additional field gets added to this structure later on, a bit for that 1877additional piece of information will be set in the flags bitmap. 1878 1879The flags bitmap is defined as:: 1880 1881 /* the host supports the ePAPR idle hcall 1882 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1<<0) 1883 18844.52 KVM_SET_GSI_ROUTING 1885------------------------ 1886 1887:Capability: KVM_CAP_IRQ_ROUTING 1888:Architectures: x86 s390 arm64 1889:Type: vm ioctl 1890:Parameters: struct kvm_irq_routing (in) 1891:Returns: 0 on success, -1 on error 1892 1893Sets the GSI routing table entries, overwriting any previously set entries. 1894 1895On arm64, GSI routing has the following limitation: 1896 1897- GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD. 1898 1899:: 1900 1901 struct kvm_irq_routing { 1902 __u32 nr; 1903 __u32 flags; 1904 struct kvm_irq_routing_entry entries[0]; 1905 }; 1906 1907No flags are specified so far, the corresponding field must be set to zero. 1908 1909:: 1910 1911 struct kvm_irq_routing_entry { 1912 __u32 gsi; 1913 __u32 type; 1914 __u32 flags; 1915 __u32 pad; 1916 union { 1917 struct kvm_irq_routing_irqchip irqchip; 1918 struct kvm_irq_routing_msi msi; 1919 struct kvm_irq_routing_s390_adapter adapter; 1920 struct kvm_irq_routing_hv_sint hv_sint; 1921 struct kvm_irq_routing_xen_evtchn xen_evtchn; 1922 __u32 pad[8]; 1923 } u; 1924 }; 1925 1926 /* gsi routing entry types */ 1927 #define KVM_IRQ_ROUTING_IRQCHIP 1 1928 #define KVM_IRQ_ROUTING_MSI 2 1929 #define KVM_IRQ_ROUTING_S390_ADAPTER 3 1930 #define KVM_IRQ_ROUTING_HV_SINT 4 1931 #define KVM_IRQ_ROUTING_XEN_EVTCHN 5 1932 1933On s390, adding a KVM_IRQ_ROUTING_S390_ADAPTER is rejected on ucontrol VMs with 1934error -EINVAL. 1935 1936flags: 1937 1938- KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry 1939 type, specifies that the devid field contains a valid value. The per-VM 1940 KVM_CAP_MSI_DEVID capability advertises the requirement to provide 1941 the device ID. If this capability is not available, userspace should 1942 never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail. 1943- zero otherwise 1944 1945:: 1946 1947 struct kvm_irq_routing_irqchip { 1948 __u32 irqchip; 1949 __u32 pin; 1950 }; 1951 1952 struct kvm_irq_routing_msi { 1953 __u32 address_lo; 1954 __u32 address_hi; 1955 __u32 data; 1956 union { 1957 __u32 pad; 1958 __u32 devid; 1959 }; 1960 }; 1961 1962If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier 1963for the device that wrote the MSI message. For PCI, this is usually a 1964BDF identifier in the lower 16 bits. 1965 1966On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS 1967feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled, 1968address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of 1969address_hi must be zero. 1970 1971:: 1972 1973 struct kvm_irq_routing_s390_adapter { 1974 __u64 ind_addr; 1975 __u64 summary_addr; 1976 __u64 ind_offset; 1977 __u32 summary_offset; 1978 __u32 adapter_id; 1979 }; 1980 1981 struct kvm_irq_routing_hv_sint { 1982 __u32 vcpu; 1983 __u32 sint; 1984 }; 1985 1986 struct kvm_irq_routing_xen_evtchn { 1987 __u32 port; 1988 __u32 vcpu; 1989 __u32 priority; 1990 }; 1991 1992 1993When KVM_CAP_XEN_HVM includes the KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL bit 1994in its indication of supported features, routing to Xen event channels 1995is supported. Although the priority field is present, only the value 1996KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL is supported, which means delivery by 19972 level event channels. FIFO event channel support may be added in 1998the future. 1999 2000 20014.55 KVM_SET_TSC_KHZ 2002-------------------- 2003 2004:Capability: KVM_CAP_TSC_CONTROL / KVM_CAP_VM_TSC_CONTROL 2005:Architectures: x86 2006:Type: vcpu ioctl / vm ioctl 2007:Parameters: virtual tsc_khz 2008:Returns: 0 on success, -1 on error 2009 2010Specifies the tsc frequency for the virtual machine. The unit of the 2011frequency is KHz. 2012 2013If the KVM_CAP_VM_TSC_CONTROL capability is advertised, this can also 2014be used as a vm ioctl to set the initial tsc frequency of subsequently 2015created vCPUs. Note, the vm ioctl is only allowed prior to creating vCPUs. 2016 2017For TSC protected Confidential Computing (CoCo) VMs where TSC frequency 2018is configured once at VM scope and remains unchanged during VM's 2019lifetime, the vm ioctl should be used to configure the TSC frequency 2020and the vcpu ioctl is not supported. 2021 2022Example of such CoCo VMs: TDX guests. 2023 20244.56 KVM_GET_TSC_KHZ 2025-------------------- 2026 2027:Capability: KVM_CAP_GET_TSC_KHZ / KVM_CAP_VM_TSC_CONTROL 2028:Architectures: x86 2029:Type: vcpu ioctl / vm ioctl 2030:Parameters: none 2031:Returns: virtual tsc-khz on success, negative value on error 2032 2033Returns the tsc frequency of the guest. The unit of the return value is 2034KHz. If the host has unstable tsc this ioctl returns -EIO instead as an 2035error. 2036 2037 20384.57 KVM_GET_LAPIC 2039------------------ 2040 2041:Capability: KVM_CAP_IRQCHIP 2042:Architectures: x86 2043:Type: vcpu ioctl 2044:Parameters: struct kvm_lapic_state (out) 2045:Returns: 0 on success, -1 on error 2046 2047:: 2048 2049 #define KVM_APIC_REG_SIZE 0x400 2050 struct kvm_lapic_state { 2051 char regs[KVM_APIC_REG_SIZE]; 2052 }; 2053 2054Reads the Local APIC registers and copies them into the input argument. The 2055data format and layout are the same as documented in the architecture manual. 2056 2057If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is 2058enabled, then the format of APIC_ID register depends on the APIC mode 2059(reported by MSR_IA32_APICBASE) of its VCPU. x2APIC stores APIC ID in 2060the APIC_ID register (bytes 32-35). xAPIC only allows an 8-bit APIC ID 2061which is stored in bits 31-24 of the APIC register, or equivalently in 2062byte 35 of struct kvm_lapic_state's regs field. KVM_GET_LAPIC must then 2063be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR. 2064 2065If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state 2066always uses xAPIC format. 2067 2068 20694.58 KVM_SET_LAPIC 2070------------------ 2071 2072:Capability: KVM_CAP_IRQCHIP 2073:Architectures: x86 2074:Type: vcpu ioctl 2075:Parameters: struct kvm_lapic_state (in) 2076:Returns: 0 on success, -1 on error 2077 2078:: 2079 2080 #define KVM_APIC_REG_SIZE 0x400 2081 struct kvm_lapic_state { 2082 char regs[KVM_APIC_REG_SIZE]; 2083 }; 2084 2085Copies the input argument into the Local APIC registers. The data format 2086and layout are the same as documented in the architecture manual. 2087 2088The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's 2089regs field) depends on the state of the KVM_CAP_X2APIC_API capability. 2090See the note in KVM_GET_LAPIC. 2091 2092 20934.59 KVM_IOEVENTFD 2094------------------ 2095 2096:Capability: KVM_CAP_IOEVENTFD 2097:Architectures: all 2098:Type: vm ioctl 2099:Parameters: struct kvm_ioeventfd (in) 2100:Returns: 0 on success, !0 on error 2101 2102This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address 2103within the guest. A guest write in the registered address will signal the 2104provided event instead of triggering an exit. 2105 2106:: 2107 2108 struct kvm_ioeventfd { 2109 __u64 datamatch; 2110 __u64 addr; /* legal pio/mmio address */ 2111 __u32 len; /* 0, 1, 2, 4, or 8 bytes */ 2112 __s32 fd; 2113 __u32 flags; 2114 __u8 pad[36]; 2115 }; 2116 2117For the special case of virtio-ccw devices on s390, the ioevent is matched 2118to a subchannel/virtqueue tuple instead. 2119 2120The following flags are defined:: 2121 2122 #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch) 2123 #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio) 2124 #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign) 2125 #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \ 2126 (1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify) 2127 2128If datamatch flag is set, the event will be signaled only if the written value 2129to the registered address is equal to datamatch in struct kvm_ioeventfd. 2130 2131For virtio-ccw devices, addr contains the subchannel id and datamatch the 2132virtqueue index. 2133 2134With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and 2135the kernel will ignore the length of guest write and may get a faster vmexit. 2136The speedup may only apply to specific architectures, but the ioeventfd will 2137work anyway. 2138 21394.60 KVM_DIRTY_TLB 2140------------------ 2141 2142:Capability: KVM_CAP_SW_TLB 2143:Architectures: ppc 2144:Type: vcpu ioctl 2145:Parameters: struct kvm_dirty_tlb (in) 2146:Returns: 0 on success, -1 on error 2147 2148:: 2149 2150 struct kvm_dirty_tlb { 2151 __u64 bitmap; 2152 __u32 num_dirty; 2153 }; 2154 2155This must be called whenever userspace has changed an entry in the shared 2156TLB, prior to calling KVM_RUN on the associated vcpu. 2157 2158The "bitmap" field is the userspace address of an array. This array 2159consists of a number of bits, equal to the total number of TLB entries as 2160determined by the last successful call to ``KVM_ENABLE_CAP(KVM_CAP_SW_TLB)``, 2161rounded up to the nearest multiple of 64. 2162 2163Each bit corresponds to one TLB entry, ordered the same as in the shared TLB 2164array. 2165 2166The array is little-endian: the bit 0 is the least significant bit of the 2167first byte, bit 8 is the least significant bit of the second byte, etc. 2168This avoids any complications with differing word sizes. 2169 2170The "num_dirty" field is a performance hint for KVM to determine whether it 2171should skip processing the bitmap and just invalidate everything. It must 2172be set to the number of set bits in the bitmap. 2173 2174 21754.62 KVM_CREATE_SPAPR_TCE 2176------------------------- 2177 2178:Capability: KVM_CAP_SPAPR_TCE 2179:Architectures: powerpc 2180:Type: vm ioctl 2181:Parameters: struct kvm_create_spapr_tce (in) 2182:Returns: file descriptor for manipulating the created TCE table 2183 2184This creates a virtual TCE (translation control entry) table, which 2185is an IOMMU for PAPR-style virtual I/O. It is used to translate 2186logical addresses used in virtual I/O into guest physical addresses, 2187and provides a scatter/gather capability for PAPR virtual I/O. 2188 2189:: 2190 2191 /* for KVM_CAP_SPAPR_TCE */ 2192 struct kvm_create_spapr_tce { 2193 __u64 liobn; 2194 __u32 window_size; 2195 }; 2196 2197The liobn field gives the logical IO bus number for which to create a 2198TCE table. The window_size field specifies the size of the DMA window 2199which this TCE table will translate - the table will contain one 64 2200bit TCE entry for every 4kiB of the DMA window. 2201 2202When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE 2203table has been created using this ioctl(), the kernel will handle it 2204in real mode, updating the TCE table. H_PUT_TCE calls for other 2205liobns will cause a vm exit and must be handled by userspace. 2206 2207The return value is a file descriptor which can be passed to mmap(2) 2208to map the created TCE table into userspace. This lets userspace read 2209the entries written by kernel-handled H_PUT_TCE calls, and also lets 2210userspace update the TCE table directly which is useful in some 2211circumstances. 2212 2213 22144.64 KVM_NMI 2215------------ 2216 2217:Capability: KVM_CAP_USER_NMI 2218:Architectures: x86 2219:Type: vcpu ioctl 2220:Parameters: none 2221:Returns: 0 on success, -1 on error 2222 2223Queues an NMI on the thread's vcpu. Note this is well defined only 2224when KVM_CREATE_IRQCHIP has not been called, since this is an interface 2225between the virtual cpu core and virtual local APIC. After KVM_CREATE_IRQCHIP 2226has been called, this interface is completely emulated within the kernel. 2227 2228To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the 2229following algorithm: 2230 2231 - pause the vcpu 2232 - read the local APIC's state (KVM_GET_LAPIC) 2233 - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1) 2234 - if so, issue KVM_NMI 2235 - resume the vcpu 2236 2237Some guests configure the LINT1 NMI input to cause a panic, aiding in 2238debugging. 2239 2240 22414.65 KVM_S390_UCAS_MAP 2242---------------------- 2243 2244:Capability: KVM_CAP_S390_UCONTROL 2245:Architectures: s390 2246:Type: vcpu ioctl 2247:Parameters: struct kvm_s390_ucas_mapping (in) 2248:Returns: 0 in case of success 2249 2250The parameter is defined like this:: 2251 2252 struct kvm_s390_ucas_mapping { 2253 __u64 user_addr; 2254 __u64 vcpu_addr; 2255 __u64 length; 2256 }; 2257 2258This ioctl maps the memory at "user_addr" with the length "length" to 2259the vcpu's address space starting at "vcpu_addr". All parameters need to 2260be aligned by 1 megabyte. 2261 2262 22634.66 KVM_S390_UCAS_UNMAP 2264------------------------ 2265 2266:Capability: KVM_CAP_S390_UCONTROL 2267:Architectures: s390 2268:Type: vcpu ioctl 2269:Parameters: struct kvm_s390_ucas_mapping (in) 2270:Returns: 0 in case of success 2271 2272The parameter is defined like this:: 2273 2274 struct kvm_s390_ucas_mapping { 2275 __u64 user_addr; 2276 __u64 vcpu_addr; 2277 __u64 length; 2278 }; 2279 2280This ioctl unmaps the memory in the vcpu's address space starting at 2281"vcpu_addr" with the length "length". The field "user_addr" is ignored. 2282All parameters need to be aligned by 1 megabyte. 2283 2284 22854.67 KVM_S390_VCPU_FAULT 2286------------------------ 2287 2288:Capability: KVM_CAP_S390_UCONTROL 2289:Architectures: s390 2290:Type: vcpu ioctl 2291:Parameters: vcpu absolute address (in) 2292:Returns: 0 in case of success 2293 2294This call creates a page table entry on the virtual cpu's address space 2295(for user controlled virtual machines) or the virtual machine's address 2296space (for regular virtual machines). This only works for minor faults, 2297thus it's recommended to access subject memory page via the user page 2298table upfront. This is useful to handle validity intercepts for user 2299controlled virtual machines to fault in the virtual cpu's lowcore pages 2300prior to calling the KVM_RUN ioctl. 2301 2302 23034.68 KVM_SET_ONE_REG 2304-------------------- 2305 2306:Capability: KVM_CAP_ONE_REG 2307:Architectures: all 2308:Type: vcpu ioctl 2309:Parameters: struct kvm_one_reg (in) 2310:Returns: 0 on success, negative value on failure 2311 2312Errors: 2313 2314 ====== ============================================================ 2315 ENOENT no such register 2316 EINVAL invalid register ID, or no such register or used with VMs in 2317 protected virtualization mode on s390 2318 EPERM (arm64) register access not allowed before vcpu finalization 2319 EBUSY (riscv) changing register value not allowed after the vcpu 2320 has run at least once 2321 ====== ============================================================ 2322 2323(These error codes are indicative only: do not rely on a specific error 2324code being returned in a specific situation.) 2325 2326:: 2327 2328 struct kvm_one_reg { 2329 __u64 id; 2330 __u64 addr; 2331 }; 2332 2333Using this ioctl, a single vcpu register can be set to a specific value 2334defined by user space with the passed in struct kvm_one_reg, where id 2335refers to the register identifier as described below and addr is a pointer 2336to a variable with the respective size. There can be architecture agnostic 2337and architecture specific registers. Each have their own range of operation 2338and their own constants and width. To keep track of the implemented 2339registers, find a list below: 2340 2341 ======= =============================== ============ 2342 Arch Register Width (bits) 2343 ======= =============================== ============ 2344 PPC KVM_REG_PPC_HIOR 64 2345 PPC KVM_REG_PPC_IAC1 64 2346 PPC KVM_REG_PPC_IAC2 64 2347 PPC KVM_REG_PPC_IAC3 64 2348 PPC KVM_REG_PPC_IAC4 64 2349 PPC KVM_REG_PPC_DAC1 64 2350 PPC KVM_REG_PPC_DAC2 64 2351 PPC KVM_REG_PPC_DABR 64 2352 PPC KVM_REG_PPC_DSCR 64 2353 PPC KVM_REG_PPC_PURR 64 2354 PPC KVM_REG_PPC_SPURR 64 2355 PPC KVM_REG_PPC_DAR 64 2356 PPC KVM_REG_PPC_DSISR 32 2357 PPC KVM_REG_PPC_AMR 64 2358 PPC KVM_REG_PPC_UAMOR 64 2359 PPC KVM_REG_PPC_MMCR0 64 2360 PPC KVM_REG_PPC_MMCR1 64 2361 PPC KVM_REG_PPC_MMCRA 64 2362 PPC KVM_REG_PPC_MMCR2 64 2363 PPC KVM_REG_PPC_MMCRS 64 2364 PPC KVM_REG_PPC_MMCR3 64 2365 PPC KVM_REG_PPC_SIAR 64 2366 PPC KVM_REG_PPC_SDAR 64 2367 PPC KVM_REG_PPC_SIER 64 2368 PPC KVM_REG_PPC_SIER2 64 2369 PPC KVM_REG_PPC_SIER3 64 2370 PPC KVM_REG_PPC_PMC1 32 2371 PPC KVM_REG_PPC_PMC2 32 2372 PPC KVM_REG_PPC_PMC3 32 2373 PPC KVM_REG_PPC_PMC4 32 2374 PPC KVM_REG_PPC_PMC5 32 2375 PPC KVM_REG_PPC_PMC6 32 2376 PPC KVM_REG_PPC_PMC7 32 2377 PPC KVM_REG_PPC_PMC8 32 2378 PPC KVM_REG_PPC_FPR0 64 2379 ... 2380 PPC KVM_REG_PPC_FPR31 64 2381 PPC KVM_REG_PPC_VR0 128 2382 ... 2383 PPC KVM_REG_PPC_VR31 128 2384 PPC KVM_REG_PPC_VSR0 128 2385 ... 2386 PPC KVM_REG_PPC_VSR31 128 2387 PPC KVM_REG_PPC_FPSCR 64 2388 PPC KVM_REG_PPC_VSCR 32 2389 PPC KVM_REG_PPC_VPA_ADDR 64 2390 PPC KVM_REG_PPC_VPA_SLB 128 2391 PPC KVM_REG_PPC_VPA_DTL 128 2392 PPC KVM_REG_PPC_EPCR 32 2393 PPC KVM_REG_PPC_EPR 32 2394 PPC KVM_REG_PPC_TCR 32 2395 PPC KVM_REG_PPC_TSR 32 2396 PPC KVM_REG_PPC_OR_TSR 32 2397 PPC KVM_REG_PPC_CLEAR_TSR 32 2398 PPC KVM_REG_PPC_MAS0 32 2399 PPC KVM_REG_PPC_MAS1 32 2400 PPC KVM_REG_PPC_MAS2 64 2401 PPC KVM_REG_PPC_MAS7_3 64 2402 PPC KVM_REG_PPC_MAS4 32 2403 PPC KVM_REG_PPC_MAS6 32 2404 PPC KVM_REG_PPC_MMUCFG 32 2405 PPC KVM_REG_PPC_TLB0CFG 32 2406 PPC KVM_REG_PPC_TLB1CFG 32 2407 PPC KVM_REG_PPC_TLB2CFG 32 2408 PPC KVM_REG_PPC_TLB3CFG 32 2409 PPC KVM_REG_PPC_TLB0PS 32 2410 PPC KVM_REG_PPC_TLB1PS 32 2411 PPC KVM_REG_PPC_TLB2PS 32 2412 PPC KVM_REG_PPC_TLB3PS 32 2413 PPC KVM_REG_PPC_EPTCFG 32 2414 PPC KVM_REG_PPC_ICP_STATE 64 2415 PPC KVM_REG_PPC_VP_STATE 128 2416 PPC KVM_REG_PPC_TB_OFFSET 64 2417 PPC KVM_REG_PPC_SPMC1 32 2418 PPC KVM_REG_PPC_SPMC2 32 2419 PPC KVM_REG_PPC_IAMR 64 2420 PPC KVM_REG_PPC_TFHAR 64 2421 PPC KVM_REG_PPC_TFIAR 64 2422 PPC KVM_REG_PPC_TEXASR 64 2423 PPC KVM_REG_PPC_FSCR 64 2424 PPC KVM_REG_PPC_PSPB 32 2425 PPC KVM_REG_PPC_EBBHR 64 2426 PPC KVM_REG_PPC_EBBRR 64 2427 PPC KVM_REG_PPC_BESCR 64 2428 PPC KVM_REG_PPC_TAR 64 2429 PPC KVM_REG_PPC_DPDES 64 2430 PPC KVM_REG_PPC_DAWR 64 2431 PPC KVM_REG_PPC_DAWRX 64 2432 PPC KVM_REG_PPC_CIABR 64 2433 PPC KVM_REG_PPC_IC 64 2434 PPC KVM_REG_PPC_VTB 64 2435 PPC KVM_REG_PPC_CSIGR 64 2436 PPC KVM_REG_PPC_TACR 64 2437 PPC KVM_REG_PPC_TCSCR 64 2438 PPC KVM_REG_PPC_PID 64 2439 PPC KVM_REG_PPC_ACOP 64 2440 PPC KVM_REG_PPC_VRSAVE 32 2441 PPC KVM_REG_PPC_LPCR 32 2442 PPC KVM_REG_PPC_LPCR_64 64 2443 PPC KVM_REG_PPC_PPR 64 2444 PPC KVM_REG_PPC_ARCH_COMPAT 32 2445 PPC KVM_REG_PPC_DABRX 32 2446 PPC KVM_REG_PPC_WORT 64 2447 PPC KVM_REG_PPC_SPRG9 64 2448 PPC KVM_REG_PPC_DBSR 32 2449 PPC KVM_REG_PPC_TIDR 64 2450 PPC KVM_REG_PPC_PSSCR 64 2451 PPC KVM_REG_PPC_DEC_EXPIRY 64 2452 PPC KVM_REG_PPC_PTCR 64 2453 PPC KVM_REG_PPC_HASHKEYR 64 2454 PPC KVM_REG_PPC_HASHPKEYR 64 2455 PPC KVM_REG_PPC_DAWR1 64 2456 PPC KVM_REG_PPC_DAWRX1 64 2457 PPC KVM_REG_PPC_DEXCR 64 2458 PPC KVM_REG_PPC_TM_GPR0 64 2459 ... 2460 PPC KVM_REG_PPC_TM_GPR31 64 2461 PPC KVM_REG_PPC_TM_VSR0 128 2462 ... 2463 PPC KVM_REG_PPC_TM_VSR63 128 2464 PPC KVM_REG_PPC_TM_CR 64 2465 PPC KVM_REG_PPC_TM_LR 64 2466 PPC KVM_REG_PPC_TM_CTR 64 2467 PPC KVM_REG_PPC_TM_FPSCR 64 2468 PPC KVM_REG_PPC_TM_AMR 64 2469 PPC KVM_REG_PPC_TM_PPR 64 2470 PPC KVM_REG_PPC_TM_VRSAVE 64 2471 PPC KVM_REG_PPC_TM_VSCR 32 2472 PPC KVM_REG_PPC_TM_DSCR 64 2473 PPC KVM_REG_PPC_TM_TAR 64 2474 PPC KVM_REG_PPC_TM_XER 64 2475 2476 MIPS KVM_REG_MIPS_R0 64 2477 ... 2478 MIPS KVM_REG_MIPS_R31 64 2479 MIPS KVM_REG_MIPS_HI 64 2480 MIPS KVM_REG_MIPS_LO 64 2481 MIPS KVM_REG_MIPS_PC 64 2482 MIPS KVM_REG_MIPS_CP0_INDEX 32 2483 MIPS KVM_REG_MIPS_CP0_ENTRYLO0 64 2484 MIPS KVM_REG_MIPS_CP0_ENTRYLO1 64 2485 MIPS KVM_REG_MIPS_CP0_CONTEXT 64 2486 MIPS KVM_REG_MIPS_CP0_CONTEXTCONFIG 32 2487 MIPS KVM_REG_MIPS_CP0_USERLOCAL 64 2488 MIPS KVM_REG_MIPS_CP0_XCONTEXTCONFIG 64 2489 MIPS KVM_REG_MIPS_CP0_PAGEMASK 32 2490 MIPS KVM_REG_MIPS_CP0_PAGEGRAIN 32 2491 MIPS KVM_REG_MIPS_CP0_SEGCTL0 64 2492 MIPS KVM_REG_MIPS_CP0_SEGCTL1 64 2493 MIPS KVM_REG_MIPS_CP0_SEGCTL2 64 2494 MIPS KVM_REG_MIPS_CP0_PWBASE 64 2495 MIPS KVM_REG_MIPS_CP0_PWFIELD 64 2496 MIPS KVM_REG_MIPS_CP0_PWSIZE 64 2497 MIPS KVM_REG_MIPS_CP0_WIRED 32 2498 MIPS KVM_REG_MIPS_CP0_PWCTL 32 2499 MIPS KVM_REG_MIPS_CP0_HWRENA 32 2500 MIPS KVM_REG_MIPS_CP0_BADVADDR 64 2501 MIPS KVM_REG_MIPS_CP0_BADINSTR 32 2502 MIPS KVM_REG_MIPS_CP0_BADINSTRP 32 2503 MIPS KVM_REG_MIPS_CP0_COUNT 32 2504 MIPS KVM_REG_MIPS_CP0_ENTRYHI 64 2505 MIPS KVM_REG_MIPS_CP0_COMPARE 32 2506 MIPS KVM_REG_MIPS_CP0_STATUS 32 2507 MIPS KVM_REG_MIPS_CP0_INTCTL 32 2508 MIPS KVM_REG_MIPS_CP0_CAUSE 32 2509 MIPS KVM_REG_MIPS_CP0_EPC 64 2510 MIPS KVM_REG_MIPS_CP0_PRID 32 2511 MIPS KVM_REG_MIPS_CP0_EBASE 64 2512 MIPS KVM_REG_MIPS_CP0_CONFIG 32 2513 MIPS KVM_REG_MIPS_CP0_CONFIG1 32 2514 MIPS KVM_REG_MIPS_CP0_CONFIG2 32 2515 MIPS KVM_REG_MIPS_CP0_CONFIG3 32 2516 MIPS KVM_REG_MIPS_CP0_CONFIG4 32 2517 MIPS KVM_REG_MIPS_CP0_CONFIG5 32 2518 MIPS KVM_REG_MIPS_CP0_CONFIG7 32 2519 MIPS KVM_REG_MIPS_CP0_XCONTEXT 64 2520 MIPS KVM_REG_MIPS_CP0_ERROREPC 64 2521 MIPS KVM_REG_MIPS_CP0_KSCRATCH1 64 2522 MIPS KVM_REG_MIPS_CP0_KSCRATCH2 64 2523 MIPS KVM_REG_MIPS_CP0_KSCRATCH3 64 2524 MIPS KVM_REG_MIPS_CP0_KSCRATCH4 64 2525 MIPS KVM_REG_MIPS_CP0_KSCRATCH5 64 2526 MIPS KVM_REG_MIPS_CP0_KSCRATCH6 64 2527 MIPS KVM_REG_MIPS_CP0_MAAR(0..63) 64 2528 MIPS KVM_REG_MIPS_COUNT_CTL 64 2529 MIPS KVM_REG_MIPS_COUNT_RESUME 64 2530 MIPS KVM_REG_MIPS_COUNT_HZ 64 2531 MIPS KVM_REG_MIPS_FPR_32(0..31) 32 2532 MIPS KVM_REG_MIPS_FPR_64(0..31) 64 2533 MIPS KVM_REG_MIPS_VEC_128(0..31) 128 2534 MIPS KVM_REG_MIPS_FCR_IR 32 2535 MIPS KVM_REG_MIPS_FCR_CSR 32 2536 MIPS KVM_REG_MIPS_MSA_IR 32 2537 MIPS KVM_REG_MIPS_MSA_CSR 32 2538 ======= =============================== ============ 2539 2540ARM registers are mapped using the lower 32 bits. The upper 16 of that 2541is the register group type, or coprocessor number: 2542 2543ARM core registers have the following id bit patterns:: 2544 2545 0x4020 0000 0010 <index into the kvm_regs struct:16> 2546 2547ARM 32-bit CP15 registers have the following id bit patterns:: 2548 2549 0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3> 2550 2551ARM 64-bit CP15 registers have the following id bit patterns:: 2552 2553 0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3> 2554 2555ARM CCSIDR registers are demultiplexed by CSSELR value:: 2556 2557 0x4020 0000 0011 00 <csselr:8> 2558 2559ARM 32-bit VFP control registers have the following id bit patterns:: 2560 2561 0x4020 0000 0012 1 <regno:12> 2562 2563ARM 64-bit FP registers have the following id bit patterns:: 2564 2565 0x4030 0000 0012 0 <regno:12> 2566 2567ARM firmware pseudo-registers have the following bit pattern:: 2568 2569 0x4030 0000 0014 <regno:16> 2570 2571 2572arm64 registers are mapped using the lower 32 bits. The upper 16 of 2573that is the register group type, or coprocessor number: 2574 2575arm64 core/FP-SIMD registers have the following id bit patterns. Note 2576that the size of the access is variable, as the kvm_regs structure 2577contains elements ranging from 32 to 128 bits. The index is a 32bit 2578value in the kvm_regs structure seen as a 32bit array:: 2579 2580 0x60x0 0000 0010 <index into the kvm_regs struct:16> 2581 2582Specifically: 2583 2584======================= ========= ===== ======================================= 2585 Encoding Register Bits kvm_regs member 2586======================= ========= ===== ======================================= 2587 0x6030 0000 0010 0000 X0 64 regs.regs[0] 2588 0x6030 0000 0010 0002 X1 64 regs.regs[1] 2589 ... 2590 0x6030 0000 0010 003c X30 64 regs.regs[30] 2591 0x6030 0000 0010 003e SP 64 regs.sp 2592 0x6030 0000 0010 0040 PC 64 regs.pc 2593 0x6030 0000 0010 0042 PSTATE 64 regs.pstate 2594 0x6030 0000 0010 0044 SP_EL1 64 sp_el1 2595 0x6030 0000 0010 0046 ELR_EL1 64 elr_el1 2596 0x6030 0000 0010 0048 SPSR_EL1 64 spsr[KVM_SPSR_EL1] (alias SPSR_SVC) 2597 0x6030 0000 0010 004a SPSR_ABT 64 spsr[KVM_SPSR_ABT] 2598 0x6030 0000 0010 004c SPSR_UND 64 spsr[KVM_SPSR_UND] 2599 0x6030 0000 0010 004e SPSR_IRQ 64 spsr[KVM_SPSR_IRQ] 2600 0x6030 0000 0010 0050 SPSR_FIQ 64 spsr[KVM_SPSR_FIQ] 2601 0x6040 0000 0010 0054 V0 128 fp_regs.vregs[0] [1]_ 2602 0x6040 0000 0010 0058 V1 128 fp_regs.vregs[1] [1]_ 2603 ... 2604 0x6040 0000 0010 00d0 V31 128 fp_regs.vregs[31] [1]_ 2605 0x6020 0000 0010 00d4 FPSR 32 fp_regs.fpsr 2606 0x6020 0000 0010 00d5 FPCR 32 fp_regs.fpcr 2607======================= ========= ===== ======================================= 2608 2609.. [1] These encodings are not accepted for SVE-enabled vcpus. See 2610 :ref:`KVM_ARM_VCPU_INIT`. 2611 2612 The equivalent register content can be accessed via bits [127:0] of 2613 the corresponding SVE Zn registers instead for vcpus that have SVE 2614 enabled (see below). 2615 2616arm64 CCSIDR registers are demultiplexed by CSSELR value:: 2617 2618 0x6020 0000 0011 00 <csselr:8> 2619 2620arm64 system registers have the following id bit patterns:: 2621 2622 0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3> 2623 2624.. warning:: 2625 2626 Two system register IDs do not follow the specified pattern. These 2627 are KVM_REG_ARM_TIMER_CVAL and KVM_REG_ARM_TIMER_CNT, which map to 2628 system registers CNTV_CVAL_EL0 and CNTVCT_EL0 respectively. These 2629 two had their values accidentally swapped, which means TIMER_CVAL is 2630 derived from the register encoding for CNTVCT_EL0 and TIMER_CNT is 2631 derived from the register encoding for CNTV_CVAL_EL0. As this is 2632 API, it must remain this way. 2633 2634arm64 firmware pseudo-registers have the following bit pattern:: 2635 2636 0x6030 0000 0014 <regno:16> 2637 2638arm64 SVE registers have the following bit patterns:: 2639 2640 0x6080 0000 0015 00 <n:5> <slice:5> Zn bits[2048*slice + 2047 : 2048*slice] 2641 0x6050 0000 0015 04 <n:4> <slice:5> Pn bits[256*slice + 255 : 256*slice] 2642 0x6050 0000 0015 060 <slice:5> FFR bits[256*slice + 255 : 256*slice] 2643 0x6060 0000 0015 ffff KVM_REG_ARM64_SVE_VLS pseudo-register 2644 2645Access to register IDs where 2048 * slice >= 128 * max_vq will fail with 2646ENOENT. max_vq is the vcpu's maximum supported vector length in 128-bit 2647quadwords: see [2]_ below. 2648 2649These registers are only accessible on vcpus for which SVE is enabled. 2650See KVM_ARM_VCPU_INIT for details. 2651 2652In addition, except for KVM_REG_ARM64_SVE_VLS, these registers are not 2653accessible until the vcpu's SVE configuration has been finalized 2654using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). See KVM_ARM_VCPU_INIT 2655and KVM_ARM_VCPU_FINALIZE for more information about this procedure. 2656 2657KVM_REG_ARM64_SVE_VLS is a pseudo-register that allows the set of vector 2658lengths supported by the vcpu to be discovered and configured by 2659userspace. When transferred to or from user memory via KVM_GET_ONE_REG 2660or KVM_SET_ONE_REG, the value of this register is of type 2661__u64[KVM_ARM64_SVE_VLS_WORDS], and encodes the set of vector lengths as 2662follows:: 2663 2664 __u64 vector_lengths[KVM_ARM64_SVE_VLS_WORDS]; 2665 2666 if (vq >= SVE_VQ_MIN && vq <= SVE_VQ_MAX && 2667 ((vector_lengths[(vq - KVM_ARM64_SVE_VQ_MIN) / 64] >> 2668 ((vq - KVM_ARM64_SVE_VQ_MIN) % 64)) & 1)) 2669 /* Vector length vq * 16 bytes supported */ 2670 else 2671 /* Vector length vq * 16 bytes not supported */ 2672 2673.. [2] The maximum value vq for which the above condition is true is 2674 max_vq. This is the maximum vector length available to the guest on 2675 this vcpu, and determines which register slices are visible through 2676 this ioctl interface. 2677 2678(See Documentation/arch/arm64/sve.rst for an explanation of the "vq" 2679nomenclature.) 2680 2681KVM_REG_ARM64_SVE_VLS is only accessible after KVM_ARM_VCPU_INIT. 2682KVM_ARM_VCPU_INIT initialises it to the best set of vector lengths that 2683the host supports. 2684 2685Userspace may subsequently modify it if desired until the vcpu's SVE 2686configuration is finalized using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). 2687 2688Apart from simply removing all vector lengths from the host set that 2689exceed some value, support for arbitrarily chosen sets of vector lengths 2690is hardware-dependent and may not be available. Attempting to configure 2691an invalid set of vector lengths via KVM_SET_ONE_REG will fail with 2692EINVAL. 2693 2694After the vcpu's SVE configuration is finalized, further attempts to 2695write this register will fail with EPERM. 2696 2697arm64 bitmap feature firmware pseudo-registers have the following bit pattern:: 2698 2699 0x6030 0000 0016 <regno:16> 2700 2701The bitmap feature firmware registers exposes the hypercall services that 2702are available for userspace to configure. The set bits corresponds to the 2703services that are available for the guests to access. By default, KVM 2704sets all the supported bits during VM initialization. The userspace can 2705discover the available services via KVM_GET_ONE_REG, and write back the 2706bitmap corresponding to the features that it wishes guests to see via 2707KVM_SET_ONE_REG. 2708 2709Note: These registers are immutable once any of the vCPUs of the VM has 2710run at least once. A KVM_SET_ONE_REG in such a scenario will return 2711a -EBUSY to userspace. 2712 2713(See Documentation/virt/kvm/arm/hypercalls.rst for more details.) 2714 2715 2716MIPS registers are mapped using the lower 32 bits. The upper 16 of that is 2717the register group type: 2718 2719MIPS core registers (see above) have the following id bit patterns:: 2720 2721 0x7030 0000 0000 <reg:16> 2722 2723MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit 2724patterns depending on whether they're 32-bit or 64-bit registers:: 2725 2726 0x7020 0000 0001 00 <reg:5> <sel:3> (32-bit) 2727 0x7030 0000 0001 00 <reg:5> <sel:3> (64-bit) 2728 2729Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_MIPS_CP0_ENTRYLO1 are the MIPS64 2730versions of the EntryLo registers regardless of the word size of the host 2731hardware, host kernel, guest, and whether XPA is present in the guest, i.e. 2732with the RI and XI bits (if they exist) in bits 63 and 62 respectively, and 2733the PFNX field starting at bit 30. 2734 2735MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) above) have the following id bit 2736patterns:: 2737 2738 0x7030 0000 0001 01 <reg:8> 2739 2740MIPS KVM control registers (see above) have the following id bit patterns:: 2741 2742 0x7030 0000 0002 <reg:16> 2743 2744MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following 2745id bit patterns depending on the size of the register being accessed. They are 2746always accessed according to the current guest FPU mode (Status.FR and 2747Config5.FRE), i.e. as the guest would see them, and they become unpredictable 2748if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector 2749registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they 2750overlap the FPU registers:: 2751 2752 0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers) 2753 0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers) 2754 0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers) 2755 2756MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the 2757following id bit patterns:: 2758 2759 0x7020 0000 0003 01 <0:3> <reg:5> 2760 2761MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the 2762following id bit patterns:: 2763 2764 0x7020 0000 0003 02 <0:3> <reg:5> 2765 2766RISC-V registers are mapped using the lower 32 bits. The upper 8 bits of 2767that is the register group type. 2768 2769RISC-V config registers are meant for configuring a Guest VCPU and it has 2770the following id bit patterns:: 2771 2772 0x8020 0000 01 <index into the kvm_riscv_config struct:24> (32bit Host) 2773 0x8030 0000 01 <index into the kvm_riscv_config struct:24> (64bit Host) 2774 2775Following are the RISC-V config registers: 2776 2777======================= ========= ============================================= 2778 Encoding Register Description 2779======================= ========= ============================================= 2780 0x80x0 0000 0100 0000 isa ISA feature bitmap of Guest VCPU 2781======================= ========= ============================================= 2782 2783The isa config register can be read anytime but can only be written before 2784a Guest VCPU runs. It will have ISA feature bits matching underlying host 2785set by default. 2786 2787RISC-V core registers represent the general execution state of a Guest VCPU 2788and it has the following id bit patterns:: 2789 2790 0x8020 0000 02 <index into the kvm_riscv_core struct:24> (32bit Host) 2791 0x8030 0000 02 <index into the kvm_riscv_core struct:24> (64bit Host) 2792 2793Following are the RISC-V core registers: 2794 2795======================= ========= ============================================= 2796 Encoding Register Description 2797======================= ========= ============================================= 2798 0x80x0 0000 0200 0000 regs.pc Program counter 2799 0x80x0 0000 0200 0001 regs.ra Return address 2800 0x80x0 0000 0200 0002 regs.sp Stack pointer 2801 0x80x0 0000 0200 0003 regs.gp Global pointer 2802 0x80x0 0000 0200 0004 regs.tp Task pointer 2803 0x80x0 0000 0200 0005 regs.t0 Caller saved register 0 2804 0x80x0 0000 0200 0006 regs.t1 Caller saved register 1 2805 0x80x0 0000 0200 0007 regs.t2 Caller saved register 2 2806 0x80x0 0000 0200 0008 regs.s0 Callee saved register 0 2807 0x80x0 0000 0200 0009 regs.s1 Callee saved register 1 2808 0x80x0 0000 0200 000a regs.a0 Function argument (or return value) 0 2809 0x80x0 0000 0200 000b regs.a1 Function argument (or return value) 1 2810 0x80x0 0000 0200 000c regs.a2 Function argument 2 2811 0x80x0 0000 0200 000d regs.a3 Function argument 3 2812 0x80x0 0000 0200 000e regs.a4 Function argument 4 2813 0x80x0 0000 0200 000f regs.a5 Function argument 5 2814 0x80x0 0000 0200 0010 regs.a6 Function argument 6 2815 0x80x0 0000 0200 0011 regs.a7 Function argument 7 2816 0x80x0 0000 0200 0012 regs.s2 Callee saved register 2 2817 0x80x0 0000 0200 0013 regs.s3 Callee saved register 3 2818 0x80x0 0000 0200 0014 regs.s4 Callee saved register 4 2819 0x80x0 0000 0200 0015 regs.s5 Callee saved register 5 2820 0x80x0 0000 0200 0016 regs.s6 Callee saved register 6 2821 0x80x0 0000 0200 0017 regs.s7 Callee saved register 7 2822 0x80x0 0000 0200 0018 regs.s8 Callee saved register 8 2823 0x80x0 0000 0200 0019 regs.s9 Callee saved register 9 2824 0x80x0 0000 0200 001a regs.s10 Callee saved register 10 2825 0x80x0 0000 0200 001b regs.s11 Callee saved register 11 2826 0x80x0 0000 0200 001c regs.t3 Caller saved register 3 2827 0x80x0 0000 0200 001d regs.t4 Caller saved register 4 2828 0x80x0 0000 0200 001e regs.t5 Caller saved register 5 2829 0x80x0 0000 0200 001f regs.t6 Caller saved register 6 2830 0x80x0 0000 0200 0020 mode Privilege mode (1 = S-mode or 0 = U-mode) 2831======================= ========= ============================================= 2832 2833RISC-V csr registers represent the supervisor mode control/status registers 2834of a Guest VCPU and it has the following id bit patterns:: 2835 2836 0x8020 0000 03 <index into the kvm_riscv_csr struct:24> (32bit Host) 2837 0x8030 0000 03 <index into the kvm_riscv_csr struct:24> (64bit Host) 2838 2839Following are the RISC-V csr registers: 2840 2841======================= ========= ============================================= 2842 Encoding Register Description 2843======================= ========= ============================================= 2844 0x80x0 0000 0300 0000 sstatus Supervisor status 2845 0x80x0 0000 0300 0001 sie Supervisor interrupt enable 2846 0x80x0 0000 0300 0002 stvec Supervisor trap vector base 2847 0x80x0 0000 0300 0003 sscratch Supervisor scratch register 2848 0x80x0 0000 0300 0004 sepc Supervisor exception program counter 2849 0x80x0 0000 0300 0005 scause Supervisor trap cause 2850 0x80x0 0000 0300 0006 stval Supervisor bad address or instruction 2851 0x80x0 0000 0300 0007 sip Supervisor interrupt pending 2852 0x80x0 0000 0300 0008 satp Supervisor address translation and protection 2853======================= ========= ============================================= 2854 2855RISC-V timer registers represent the timer state of a Guest VCPU and it has 2856the following id bit patterns:: 2857 2858 0x8030 0000 04 <index into the kvm_riscv_timer struct:24> 2859 2860Following are the RISC-V timer registers: 2861 2862======================= ========= ============================================= 2863 Encoding Register Description 2864======================= ========= ============================================= 2865 0x8030 0000 0400 0000 frequency Time base frequency (read-only) 2866 0x8030 0000 0400 0001 time Time value visible to Guest 2867 0x8030 0000 0400 0002 compare Time compare programmed by Guest 2868 0x8030 0000 0400 0003 state Time compare state (1 = ON or 0 = OFF) 2869======================= ========= ============================================= 2870 2871RISC-V F-extension registers represent the single precision floating point 2872state of a Guest VCPU and it has the following id bit patterns:: 2873 2874 0x8020 0000 05 <index into the __riscv_f_ext_state struct:24> 2875 2876Following are the RISC-V F-extension registers: 2877 2878======================= ========= ============================================= 2879 Encoding Register Description 2880======================= ========= ============================================= 2881 0x8020 0000 0500 0000 f[0] Floating point register 0 2882 ... 2883 0x8020 0000 0500 001f f[31] Floating point register 31 2884 0x8020 0000 0500 0020 fcsr Floating point control and status register 2885======================= ========= ============================================= 2886 2887RISC-V D-extension registers represent the double precision floating point 2888state of a Guest VCPU and it has the following id bit patterns:: 2889 2890 0x8020 0000 06 <index into the __riscv_d_ext_state struct:24> (fcsr) 2891 0x8030 0000 06 <index into the __riscv_d_ext_state struct:24> (non-fcsr) 2892 2893Following are the RISC-V D-extension registers: 2894 2895======================= ========= ============================================= 2896 Encoding Register Description 2897======================= ========= ============================================= 2898 0x8030 0000 0600 0000 f[0] Floating point register 0 2899 ... 2900 0x8030 0000 0600 001f f[31] Floating point register 31 2901 0x8020 0000 0600 0020 fcsr Floating point control and status register 2902======================= ========= ============================================= 2903 2904LoongArch registers are mapped using the lower 32 bits. The upper 16 bits of 2905that is the register group type. 2906 2907LoongArch csr registers are used to control guest cpu or get status of guest 2908cpu, and they have the following id bit patterns:: 2909 2910 0x9030 0000 0001 00 <reg:5> <sel:3> (64-bit) 2911 2912LoongArch KVM control registers are used to implement some new defined functions 2913such as set vcpu counter or reset vcpu, and they have the following id bit patterns:: 2914 2915 0x9030 0000 0002 <reg:16> 2916 2917x86 MSR registers have the following id bit patterns:: 2918 0x2030 0002 <msr number:32> 2919 2920Following are the KVM-defined registers for x86: 2921 2922======================= ========= ============================================= 2923 Encoding Register Description 2924======================= ========= ============================================= 2925 0x2030 0003 0000 0000 SSP Shadow Stack Pointer 2926======================= ========= ============================================= 2927 29284.69 KVM_GET_ONE_REG 2929-------------------- 2930 2931:Capability: KVM_CAP_ONE_REG 2932:Architectures: all 2933:Type: vcpu ioctl 2934:Parameters: struct kvm_one_reg (in and out) 2935:Returns: 0 on success, negative value on failure 2936 2937Errors include: 2938 2939 ======== ============================================================ 2940 ENOENT no such register 2941 EINVAL invalid register ID, or no such register or used with VMs in 2942 protected virtualization mode on s390 2943 EPERM (arm64) register access not allowed before vcpu finalization 2944 ======== ============================================================ 2945 2946(These error codes are indicative only: do not rely on a specific error 2947code being returned in a specific situation.) 2948 2949This ioctl allows to receive the value of a single register implemented 2950in a vcpu. The register to read is indicated by the "id" field of the 2951kvm_one_reg struct passed in. On success, the register value can be found 2952at the memory location pointed to by "addr". 2953 2954The list of registers accessible using this interface is identical to the 2955list in 4.68. 2956 2957 29584.70 KVM_KVMCLOCK_CTRL 2959---------------------- 2960 2961:Capability: KVM_CAP_KVMCLOCK_CTRL 2962:Architectures: Any that implement pvclocks (currently x86 only) 2963:Type: vcpu ioctl 2964:Parameters: None 2965:Returns: 0 on success, -1 on error 2966 2967This ioctl sets a flag accessible to the guest indicating that the specified 2968vCPU has been paused by the host userspace. 2969 2970The host will set a flag in the pvclock structure that is checked from the 2971soft lockup watchdog. The flag is part of the pvclock structure that is 2972shared between guest and host, specifically the second bit of the flags 2973field of the pvclock_vcpu_time_info structure. It will be set exclusively by 2974the host and read/cleared exclusively by the guest. The guest operation of 2975checking and clearing the flag must be an atomic operation so 2976load-link/store-conditional, or equivalent must be used. There are two cases 2977where the guest will clear the flag: when the soft lockup watchdog timer resets 2978itself or when a soft lockup is detected. This ioctl can be called any time 2979after pausing the vcpu, but before it is resumed. 2980 2981 29824.71 KVM_SIGNAL_MSI 2983------------------- 2984 2985:Capability: KVM_CAP_SIGNAL_MSI 2986:Architectures: x86 arm64 2987:Type: vm ioctl 2988:Parameters: struct kvm_msi (in) 2989:Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error 2990 2991Directly inject a MSI message. Only valid with in-kernel irqchip that handles 2992MSI messages. 2993 2994:: 2995 2996 struct kvm_msi { 2997 __u32 address_lo; 2998 __u32 address_hi; 2999 __u32 data; 3000 __u32 flags; 3001 __u32 devid; 3002 __u8 pad[12]; 3003 }; 3004 3005flags: 3006 KVM_MSI_VALID_DEVID: devid contains a valid value. The per-VM 3007 KVM_CAP_MSI_DEVID capability advertises the requirement to provide 3008 the device ID. If this capability is not available, userspace 3009 should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail. 3010 3011If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier 3012for the device that wrote the MSI message. For PCI, this is usually a 3013BDF identifier in the lower 16 bits. 3014 3015On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS 3016feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled, 3017address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of 3018address_hi must be zero. 3019 3020 30214.71 KVM_CREATE_PIT2 3022-------------------- 3023 3024:Capability: KVM_CAP_PIT2 3025:Architectures: x86 3026:Type: vm ioctl 3027:Parameters: struct kvm_pit_config (in) 3028:Returns: 0 on success, -1 on error 3029 3030Creates an in-kernel device model for the i8254 PIT. This call is only valid 3031after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following 3032parameters have to be passed:: 3033 3034 struct kvm_pit_config { 3035 __u32 flags; 3036 __u32 pad[15]; 3037 }; 3038 3039Valid flags are:: 3040 3041 #define KVM_PIT_SPEAKER_DUMMY 1 /* emulate speaker port stub */ 3042 3043PIT timer interrupts may use a per-VM kernel thread for injection. If it 3044exists, this thread will have a name of the following pattern:: 3045 3046 kvm-pit/<owner-process-pid> 3047 3048When running a guest with elevated priorities, the scheduling parameters of 3049this thread may have to be adjusted accordingly. 3050 3051This IOCTL replaces the obsolete KVM_CREATE_PIT. 3052 3053 30544.72 KVM_GET_PIT2 3055----------------- 3056 3057:Capability: KVM_CAP_PIT_STATE2 3058:Architectures: x86 3059:Type: vm ioctl 3060:Parameters: struct kvm_pit_state2 (out) 3061:Returns: 0 on success, -1 on error 3062 3063Retrieves the state of the in-kernel PIT model. Only valid after 3064KVM_CREATE_PIT2. The state is returned in the following structure:: 3065 3066 struct kvm_pit_state2 { 3067 struct kvm_pit_channel_state channels[3]; 3068 __u32 flags; 3069 __u32 reserved[9]; 3070 }; 3071 3072Valid flags are:: 3073 3074 /* disable PIT in HPET legacy mode */ 3075 #define KVM_PIT_FLAGS_HPET_LEGACY 0x00000001 3076 /* speaker port data bit enabled */ 3077 #define KVM_PIT_FLAGS_SPEAKER_DATA_ON 0x00000002 3078 3079This IOCTL replaces the obsolete KVM_GET_PIT. 3080 3081 30824.73 KVM_SET_PIT2 3083----------------- 3084 3085:Capability: KVM_CAP_PIT_STATE2 3086:Architectures: x86 3087:Type: vm ioctl 3088:Parameters: struct kvm_pit_state2 (in) 3089:Returns: 0 on success, -1 on error 3090 3091Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2. 3092See KVM_GET_PIT2 for details on struct kvm_pit_state2. 3093 3094.. Tip:: 3095 ``KVM_SET_PIT2`` strictly adheres to the spec of Intel 8254 PIT. For example, 3096 a ``count`` value of 0 in ``struct kvm_pit_channel_state`` is interpreted as 3097 65536, which is the maximum count value. Refer to `Intel 8254 programmable 3098 interval timer <https://www.scs.stanford.edu/10wi-cs140/pintos/specs/8254.pdf>`_. 3099 3100This IOCTL replaces the obsolete KVM_SET_PIT. 3101 3102 31034.74 KVM_PPC_GET_SMMU_INFO 3104-------------------------- 3105 3106:Capability: KVM_CAP_PPC_GET_SMMU_INFO 3107:Architectures: powerpc 3108:Type: vm ioctl 3109:Parameters: None 3110:Returns: 0 on success, -1 on error 3111 3112This populates and returns a structure describing the features of 3113the "Server" class MMU emulation supported by KVM. 3114This can in turn be used by userspace to generate the appropriate 3115device-tree properties for the guest operating system. 3116 3117The structure contains some global information, followed by an 3118array of supported segment page sizes:: 3119 3120 struct kvm_ppc_smmu_info { 3121 __u64 flags; 3122 __u32 slb_size; 3123 __u32 pad; 3124 struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ]; 3125 }; 3126 3127The supported flags are: 3128 3129 - KVM_PPC_PAGE_SIZES_REAL: 3130 When that flag is set, guest page sizes must "fit" the backing 3131 store page sizes. When not set, any page size in the list can 3132 be used regardless of how they are backed by userspace. 3133 3134 - KVM_PPC_1T_SEGMENTS 3135 The emulated MMU supports 1T segments in addition to the 3136 standard 256M ones. 3137 3138 - KVM_PPC_NO_HASH 3139 This flag indicates that HPT guests are not supported by KVM, 3140 thus all guests must use radix MMU mode. 3141 3142The "slb_size" field indicates how many SLB entries are supported 3143 3144The "sps" array contains 8 entries indicating the supported base 3145page sizes for a segment in increasing order. Each entry is defined 3146as follow:: 3147 3148 struct kvm_ppc_one_seg_page_size { 3149 __u32 page_shift; /* Base page shift of segment (or 0) */ 3150 __u32 slb_enc; /* SLB encoding for BookS */ 3151 struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ]; 3152 }; 3153 3154An entry with a "page_shift" of 0 is unused. Because the array is 3155organized in increasing order, a lookup can stop when encountering 3156such an entry. 3157 3158The "slb_enc" field provides the encoding to use in the SLB for the 3159page size. The bits are in positions such as the value can directly 3160be OR'ed into the "vsid" argument of the slbmte instruction. 3161 3162The "enc" array is a list which for each of those segment base page 3163size provides the list of supported actual page sizes (which can be 3164only larger or equal to the base page size), along with the 3165corresponding encoding in the hash PTE. Similarly, the array is 31668 entries sorted by increasing sizes and an entry with a "0" shift 3167is an empty entry and a terminator:: 3168 3169 struct kvm_ppc_one_page_size { 3170 __u32 page_shift; /* Page shift (or 0) */ 3171 __u32 pte_enc; /* Encoding in the HPTE (>>12) */ 3172 }; 3173 3174The "pte_enc" field provides a value that can OR'ed into the hash 3175PTE's RPN field (ie, it needs to be shifted left by 12 to OR it 3176into the hash PTE second double word). 3177 31784.75 KVM_IRQFD 3179-------------- 3180 3181:Capability: KVM_CAP_IRQFD 3182:Architectures: x86 s390 arm64 3183:Type: vm ioctl 3184:Parameters: struct kvm_irqfd (in) 3185:Returns: 0 on success, -1 on error 3186 3187Allows setting an eventfd to directly trigger a guest interrupt. 3188kvm_irqfd.fd specifies the file descriptor to use as the eventfd and 3189kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When 3190an event is triggered on the eventfd, an interrupt is injected into 3191the guest using the specified gsi pin. The irqfd is removed using 3192the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd 3193and kvm_irqfd.gsi. 3194 3195With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify 3196mechanism allowing emulation of level-triggered, irqfd-based 3197interrupts. When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an 3198additional eventfd in the kvm_irqfd.resamplefd field. When operating 3199in resample mode, posting of an interrupt through kvm_irq.fd asserts 3200the specified gsi in the irqchip. When the irqchip is resampled, such 3201as from an EOI, the gsi is de-asserted and the user is notified via 3202kvm_irqfd.resamplefd. It is the user's responsibility to re-queue 3203the interrupt if the device making use of it still requires service. 3204Note that closing the resamplefd is not sufficient to disable the 3205irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment 3206and need not be specified with KVM_IRQFD_FLAG_DEASSIGN. 3207 3208On arm64, gsi routing being supported, the following can happen: 3209 3210- in case no routing entry is associated to this gsi, injection fails 3211- in case the gsi is associated to an irqchip routing entry, 3212 irqchip.pin + 32 corresponds to the injected SPI ID. 3213- in case the gsi is associated to an MSI routing entry, the MSI 3214 message and device ID are translated into an LPI (support restricted 3215 to GICv3 ITS in-kernel emulation). 3216 32174.76 KVM_PPC_ALLOCATE_HTAB 3218-------------------------- 3219 3220:Capability: KVM_CAP_PPC_ALLOC_HTAB 3221:Architectures: powerpc 3222:Type: vm ioctl 3223:Parameters: Pointer to u32 containing hash table order (in/out) 3224:Returns: 0 on success, -1 on error 3225 3226This requests the host kernel to allocate an MMU hash table for a 3227guest using the PAPR paravirtualization interface. This only does 3228anything if the kernel is configured to use the Book 3S HV style of 3229virtualization. Otherwise the capability doesn't exist and the ioctl 3230returns an ENOTTY error. The rest of this description assumes Book 3S 3231HV. 3232 3233There must be no vcpus running when this ioctl is called; if there 3234are, it will do nothing and return an EBUSY error. 3235 3236The parameter is a pointer to a 32-bit unsigned integer variable 3237containing the order (log base 2) of the desired size of the hash 3238table, which must be between 18 and 46. On successful return from the 3239ioctl, the value will not be changed by the kernel. 3240 3241If no hash table has been allocated when any vcpu is asked to run 3242(with the KVM_RUN ioctl), the host kernel will allocate a 3243default-sized hash table (16 MB). 3244 3245If this ioctl is called when a hash table has already been allocated, 3246with a different order from the existing hash table, the existing hash 3247table will be freed and a new one allocated. If this is ioctl is 3248called when a hash table has already been allocated of the same order 3249as specified, the kernel will clear out the existing hash table (zero 3250all HPTEs). In either case, if the guest is using the virtualized 3251real-mode area (VRMA) facility, the kernel will re-create the VMRA 3252HPTEs on the next KVM_RUN of any vcpu. 3253 32544.77 KVM_S390_INTERRUPT 3255----------------------- 3256 3257:Capability: basic 3258:Architectures: s390 3259:Type: vm ioctl, vcpu ioctl 3260:Parameters: struct kvm_s390_interrupt (in) 3261:Returns: 0 on success, -1 on error 3262 3263Allows to inject an interrupt to the guest. Interrupts can be floating 3264(vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type. 3265 3266Interrupt parameters are passed via kvm_s390_interrupt:: 3267 3268 struct kvm_s390_interrupt { 3269 __u32 type; 3270 __u32 parm; 3271 __u64 parm64; 3272 }; 3273 3274type can be one of the following: 3275 3276KVM_S390_SIGP_STOP (vcpu) 3277 - sigp stop; optional flags in parm 3278KVM_S390_PROGRAM_INT (vcpu) 3279 - program check; code in parm 3280KVM_S390_SIGP_SET_PREFIX (vcpu) 3281 - sigp set prefix; prefix address in parm 3282KVM_S390_RESTART (vcpu) 3283 - restart 3284KVM_S390_INT_CLOCK_COMP (vcpu) 3285 - clock comparator interrupt 3286KVM_S390_INT_CPU_TIMER (vcpu) 3287 - CPU timer interrupt 3288KVM_S390_INT_VIRTIO (vm) 3289 - virtio external interrupt; external interrupt 3290 parameters in parm and parm64 3291KVM_S390_INT_SERVICE (vm) 3292 - sclp external interrupt; sclp parameter in parm 3293KVM_S390_INT_EMERGENCY (vcpu) 3294 - sigp emergency; source cpu in parm 3295KVM_S390_INT_EXTERNAL_CALL (vcpu) 3296 - sigp external call; source cpu in parm 3297KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm) 3298 - compound value to indicate an 3299 I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel); 3300 I/O interruption parameters in parm (subchannel) and parm64 (intparm, 3301 interruption subclass) 3302KVM_S390_MCHK (vm, vcpu) 3303 - machine check interrupt; cr 14 bits in parm, machine check interrupt 3304 code in parm64 (note that machine checks needing further payload are not 3305 supported by this ioctl) 3306 3307This is an asynchronous vcpu ioctl and can be invoked from any thread. 3308 33094.78 KVM_PPC_GET_HTAB_FD 3310------------------------ 3311 3312:Capability: KVM_CAP_PPC_HTAB_FD 3313:Architectures: powerpc 3314:Type: vm ioctl 3315:Parameters: Pointer to struct kvm_get_htab_fd (in) 3316:Returns: file descriptor number (>= 0) on success, -1 on error 3317 3318This returns a file descriptor that can be used either to read out the 3319entries in the guest's hashed page table (HPT), or to write entries to 3320initialize the HPT. The returned fd can only be written to if the 3321KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and 3322can only be read if that bit is clear. The argument struct looks like 3323this:: 3324 3325 /* For KVM_PPC_GET_HTAB_FD */ 3326 struct kvm_get_htab_fd { 3327 __u64 flags; 3328 __u64 start_index; 3329 __u64 reserved[2]; 3330 }; 3331 3332 /* Values for kvm_get_htab_fd.flags */ 3333 #define KVM_GET_HTAB_BOLTED_ONLY ((__u64)0x1) 3334 #define KVM_GET_HTAB_WRITE ((__u64)0x2) 3335 3336The 'start_index' field gives the index in the HPT of the entry at 3337which to start reading. It is ignored when writing. 3338 3339Reads on the fd will initially supply information about all 3340"interesting" HPT entries. Interesting entries are those with the 3341bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise 3342all entries. When the end of the HPT is reached, the read() will 3343return. If read() is called again on the fd, it will start again from 3344the beginning of the HPT, but will only return HPT entries that have 3345changed since they were last read. 3346 3347Data read or written is structured as a header (8 bytes) followed by a 3348series of valid HPT entries (16 bytes) each. The header indicates how 3349many valid HPT entries there are and how many invalid entries follow 3350the valid entries. The invalid entries are not represented explicitly 3351in the stream. The header format is:: 3352 3353 struct kvm_get_htab_header { 3354 __u32 index; 3355 __u16 n_valid; 3356 __u16 n_invalid; 3357 }; 3358 3359Writes to the fd create HPT entries starting at the index given in the 3360header; first 'n_valid' valid entries with contents from the data 3361written, then 'n_invalid' invalid entries, invalidating any previously 3362valid entries found. 3363 33644.79 KVM_CREATE_DEVICE 3365---------------------- 3366 3367:Capability: KVM_CAP_DEVICE_CTRL 3368:Architectures: all 3369:Type: vm ioctl 3370:Parameters: struct kvm_create_device (in/out) 3371:Returns: 0 on success, -1 on error 3372 3373Errors: 3374 3375 ====== ======================================================= 3376 ENODEV The device type is unknown or unsupported 3377 EEXIST Device already created, and this type of device may not 3378 be instantiated multiple times 3379 ====== ======================================================= 3380 3381 Other error conditions may be defined by individual device types or 3382 have their standard meanings. 3383 3384Creates an emulated device in the kernel. The file descriptor returned 3385in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR. 3386 3387If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the 3388device type is supported (not necessarily whether it can be created 3389in the current vm). 3390 3391Individual devices should not define flags. Attributes should be used 3392for specifying any behavior that is not implied by the device type 3393number. 3394 3395:: 3396 3397 struct kvm_create_device { 3398 __u32 type; /* in: KVM_DEV_TYPE_xxx */ 3399 __u32 fd; /* out: device handle */ 3400 __u32 flags; /* in: KVM_CREATE_DEVICE_xxx */ 3401 }; 3402 34034.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR 3404-------------------------------------------- 3405 3406:Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device, 3407 KVM_CAP_VCPU_ATTRIBUTES for vcpu device 3408 KVM_CAP_SYS_ATTRIBUTES for system (/dev/kvm) device (no set) 3409:Architectures: x86, arm64, s390 3410:Type: device ioctl, vm ioctl, vcpu ioctl 3411:Parameters: struct kvm_device_attr 3412:Returns: 0 on success, -1 on error 3413 3414Errors: 3415 3416 ===== ============================================================= 3417 ENXIO The group or attribute is unknown/unsupported for this device 3418 or hardware support is missing. 3419 EPERM The attribute cannot (currently) be accessed this way 3420 (e.g. read-only attribute, or attribute that only makes 3421 sense when the device is in a different state) 3422 ===== ============================================================= 3423 3424 Other error conditions may be defined by individual device types. 3425 3426Gets/sets a specified piece of device configuration and/or state. The 3427semantics are device-specific. See individual device documentation in 3428the "devices" directory. As with ONE_REG, the size of the data 3429transferred is defined by the particular attribute. 3430 3431:: 3432 3433 struct kvm_device_attr { 3434 __u32 flags; /* no flags currently defined */ 3435 __u32 group; /* device-defined */ 3436 __u64 attr; /* group-defined */ 3437 __u64 addr; /* userspace address of attr data */ 3438 }; 3439 34404.81 KVM_HAS_DEVICE_ATTR 3441------------------------ 3442 3443:Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device, 3444 KVM_CAP_VCPU_ATTRIBUTES for vcpu device 3445 KVM_CAP_SYS_ATTRIBUTES for system (/dev/kvm) device 3446:Type: device ioctl, vm ioctl, vcpu ioctl 3447:Parameters: struct kvm_device_attr 3448:Returns: 0 on success, -1 on error 3449 3450Errors: 3451 3452 ===== ============================================================= 3453 ENXIO The group or attribute is unknown/unsupported for this device 3454 or hardware support is missing. 3455 ===== ============================================================= 3456 3457Tests whether a device supports a particular attribute. A successful 3458return indicates the attribute is implemented. It does not necessarily 3459indicate that the attribute can be read or written in the device's 3460current state. "addr" is ignored. 3461 3462.. _KVM_ARM_VCPU_INIT: 3463 34644.82 KVM_ARM_VCPU_INIT 3465---------------------- 3466 3467:Capability: basic 3468:Architectures: arm64 3469:Type: vcpu ioctl 3470:Parameters: struct kvm_vcpu_init (in) 3471:Returns: 0 on success; -1 on error 3472 3473Errors: 3474 3475 ====== ================================================================= 3476 EINVAL the target is unknown, or the combination of features is invalid. 3477 ENOENT a features bit specified is unknown. 3478 ====== ================================================================= 3479 3480This tells KVM what type of CPU to present to the guest, and what 3481optional features it should have. This will cause a reset of the cpu 3482registers to their initial values. If this is not called, KVM_RUN will 3483return ENOEXEC for that vcpu. 3484 3485The initial values are defined as: 3486 - Processor state: 3487 * AArch64: EL1h, D, A, I and F bits set. All other bits 3488 are cleared. 3489 * AArch32: SVC, A, I and F bits set. All other bits are 3490 cleared. 3491 - General Purpose registers, including PC and SP: set to 0 3492 - FPSIMD/NEON registers: set to 0 3493 - SVE registers: set to 0 3494 - System registers: Reset to their architecturally defined 3495 values as for a warm reset to EL1 (resp. SVC) or EL2 (in the 3496 case of EL2 being enabled). 3497 3498Note that because some registers reflect machine topology, all vcpus 3499should be created before this ioctl is invoked. 3500 3501Userspace can call this function multiple times for a given vcpu, including 3502after the vcpu has been run. This will reset the vcpu to its initial 3503state. All calls to this function after the initial call must use the same 3504target and same set of feature flags, otherwise EINVAL will be returned. 3505 3506Possible features: 3507 3508 - KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state. 3509 Depends on KVM_CAP_ARM_PSCI. If not set, the CPU will be powered on 3510 and execute guest code when KVM_RUN is called. 3511 - KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode. 3512 Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only). 3513 - KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 (or a future revision 3514 backward compatible with v0.2) for the CPU. 3515 Depends on KVM_CAP_ARM_PSCI_0_2. 3516 - KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU. 3517 Depends on KVM_CAP_ARM_PMU_V3. 3518 3519 - KVM_ARM_VCPU_PTRAUTH_ADDRESS: Enables Address Pointer authentication 3520 for arm64 only. 3521 Depends on KVM_CAP_ARM_PTRAUTH_ADDRESS. 3522 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are 3523 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and 3524 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be 3525 requested. 3526 3527 - KVM_ARM_VCPU_PTRAUTH_GENERIC: Enables Generic Pointer authentication 3528 for arm64 only. 3529 Depends on KVM_CAP_ARM_PTRAUTH_GENERIC. 3530 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are 3531 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and 3532 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be 3533 requested. 3534 3535 - KVM_ARM_VCPU_SVE: Enables SVE for the CPU (arm64 only). 3536 Depends on KVM_CAP_ARM_SVE. 3537 Requires KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3538 3539 * After KVM_ARM_VCPU_INIT: 3540 3541 - KVM_REG_ARM64_SVE_VLS may be read using KVM_GET_ONE_REG: the 3542 initial value of this pseudo-register indicates the best set of 3543 vector lengths possible for a vcpu on this host. 3544 3545 * Before KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3546 3547 - KVM_RUN and KVM_GET_REG_LIST are not available; 3548 3549 - KVM_GET_ONE_REG and KVM_SET_ONE_REG cannot be used to access 3550 the scalable architectural SVE registers 3551 KVM_REG_ARM64_SVE_ZREG(), KVM_REG_ARM64_SVE_PREG() or 3552 KVM_REG_ARM64_SVE_FFR; 3553 3554 - KVM_REG_ARM64_SVE_VLS may optionally be written using 3555 KVM_SET_ONE_REG, to modify the set of vector lengths available 3556 for the vcpu. 3557 3558 * After KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3559 3560 - the KVM_REG_ARM64_SVE_VLS pseudo-register is immutable, and can 3561 no longer be written using KVM_SET_ONE_REG. 3562 3563 - KVM_ARM_VCPU_HAS_EL2: Enable Nested Virtualisation support, 3564 booting the guest from EL2 instead of EL1. 3565 Depends on KVM_CAP_ARM_EL2. 3566 The VM is running with HCR_EL2.E2H being RES1 (VHE) unless 3567 KVM_ARM_VCPU_HAS_EL2_E2H0 is also set. 3568 3569 - KVM_ARM_VCPU_HAS_EL2_E2H0: Restrict Nested Virtualisation 3570 support to HCR_EL2.E2H being RES0 (non-VHE). 3571 Depends on KVM_CAP_ARM_EL2_E2H0. 3572 KVM_ARM_VCPU_HAS_EL2 must also be set. 3573 35744.83 KVM_ARM_PREFERRED_TARGET 3575----------------------------- 3576 3577:Capability: basic 3578:Architectures: arm64 3579:Type: vm ioctl 3580:Parameters: struct kvm_vcpu_init (out) 3581:Returns: 0 on success; -1 on error 3582 3583Errors: 3584 3585 ====== ========================================== 3586 ENODEV no preferred target available for the host 3587 ====== ========================================== 3588 3589This queries KVM for preferred CPU target type which can be emulated 3590by KVM on underlying host. 3591 3592The ioctl returns struct kvm_vcpu_init instance containing information 3593about preferred CPU target type and recommended features for it. The 3594kvm_vcpu_init->features bitmap returned will have feature bits set if 3595the preferred target recommends setting these features, but this is 3596not mandatory. 3597 3598The information returned by this ioctl can be used to prepare an instance 3599of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in 3600VCPU matching underlying host. 3601 3602 36034.84 KVM_GET_REG_LIST 3604--------------------- 3605 3606:Capability: basic 3607:Architectures: arm64, mips, riscv, x86 (if KVM_CAP_ONE_REG) 3608:Type: vcpu ioctl 3609:Parameters: struct kvm_reg_list (in/out) 3610:Returns: 0 on success; -1 on error 3611 3612Errors: 3613 3614 ===== ============================================================== 3615 E2BIG the reg index list is too big to fit in the array specified by 3616 the user (the number required will be written into n). 3617 ===== ============================================================== 3618 3619:: 3620 3621 struct kvm_reg_list { 3622 __u64 n; /* number of registers in reg[] */ 3623 __u64 reg[0]; 3624 }; 3625 3626This ioctl returns the guest registers that are supported for the 3627KVM_GET_ONE_REG/KVM_SET_ONE_REG calls. 3628 3629Note that s390 does not support KVM_GET_REG_LIST for historical reasons 3630(read: nobody cared). The set of registers in kernels 4.x and newer is: 3631 3632- KVM_REG_S390_TODPR 3633 3634- KVM_REG_S390_EPOCHDIFF 3635 3636- KVM_REG_S390_CPU_TIMER 3637 3638- KVM_REG_S390_CLOCK_COMP 3639 3640- KVM_REG_S390_PFTOKEN 3641 3642- KVM_REG_S390_PFCOMPARE 3643 3644- KVM_REG_S390_PFSELECT 3645 3646- KVM_REG_S390_PP 3647 3648- KVM_REG_S390_GBEA 3649 3650Note, for x86, all MSRs enumerated by KVM_GET_MSR_INDEX_LIST are supported as 3651type KVM_X86_REG_TYPE_MSR, but are NOT enumerated via KVM_GET_REG_LIST. 3652 36534.85 KVM_ARM_SET_DEVICE_ADDR (deprecated) 3654----------------------------------------- 3655 3656:Capability: KVM_CAP_ARM_SET_DEVICE_ADDR 3657:Architectures: arm64 3658:Type: vm ioctl 3659:Parameters: struct kvm_arm_device_address (in) 3660:Returns: 0 on success, -1 on error 3661 3662Errors: 3663 3664 ====== ============================================ 3665 ENODEV The device id is unknown 3666 ENXIO Device not supported on current system 3667 EEXIST Address already set 3668 E2BIG Address outside guest physical address space 3669 EBUSY Address overlaps with other device range 3670 ====== ============================================ 3671 3672:: 3673 3674 struct kvm_arm_device_addr { 3675 __u64 id; 3676 __u64 addr; 3677 }; 3678 3679Specify a device address in the guest's physical address space where guests 3680can access emulated or directly exposed devices, which the host kernel needs 3681to know about. The id field is an architecture specific identifier for a 3682specific device. 3683 3684arm64 divides the id field into two parts, a device id and an 3685address type id specific to the individual device:: 3686 3687 bits: | 63 ... 32 | 31 ... 16 | 15 ... 0 | 3688 field: | 0x00000000 | device id | addr type id | 3689 3690arm64 currently only require this when using the in-kernel GIC 3691support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2 3692as the device id. When setting the base address for the guest's 3693mapping of the VGIC virtual CPU and distributor interface, the ioctl 3694must be called after calling KVM_CREATE_IRQCHIP, but before calling 3695KVM_RUN on any of the VCPUs. Calling this ioctl twice for any of the 3696base addresses will return -EEXIST. 3697 3698Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API 3699should be used instead. 3700 3701 37024.86 KVM_PPC_RTAS_DEFINE_TOKEN 3703------------------------------ 3704 3705:Capability: KVM_CAP_PPC_RTAS 3706:Architectures: ppc 3707:Type: vm ioctl 3708:Parameters: struct kvm_rtas_token_args 3709:Returns: 0 on success, -1 on error 3710 3711Defines a token value for a RTAS (Run Time Abstraction Services) 3712service in order to allow it to be handled in the kernel. The 3713argument struct gives the name of the service, which must be the name 3714of a service that has a kernel-side implementation. If the token 3715value is non-zero, it will be associated with that service, and 3716subsequent RTAS calls by the guest specifying that token will be 3717handled by the kernel. If the token value is 0, then any token 3718associated with the service will be forgotten, and subsequent RTAS 3719calls by the guest for that service will be passed to userspace to be 3720handled. 3721 37224.87 KVM_SET_GUEST_DEBUG 3723------------------------ 3724 3725:Capability: KVM_CAP_SET_GUEST_DEBUG 3726:Architectures: x86, s390, ppc, arm64 3727:Type: vcpu ioctl 3728:Parameters: struct kvm_guest_debug (in) 3729:Returns: 0 on success; -1 on error 3730 3731:: 3732 3733 struct kvm_guest_debug { 3734 __u32 control; 3735 __u32 pad; 3736 struct kvm_guest_debug_arch arch; 3737 }; 3738 3739Set up the processor specific debug registers and configure vcpu for 3740handling guest debug events. There are two parts to the structure, the 3741first a control bitfield indicates the type of debug events to handle 3742when running. Common control bits are: 3743 3744 - KVM_GUESTDBG_ENABLE: guest debugging is enabled 3745 - KVM_GUESTDBG_SINGLESTEP: the next run should single-step 3746 3747The top 16 bits of the control field are architecture specific control 3748flags which can include the following: 3749 3750 - KVM_GUESTDBG_USE_SW_BP: using software breakpoints [x86, arm64] 3751 - KVM_GUESTDBG_USE_HW_BP: using hardware breakpoints [x86, s390] 3752 - KVM_GUESTDBG_USE_HW: using hardware debug events [arm64] 3753 - KVM_GUESTDBG_INJECT_DB: inject DB type exception [x86] 3754 - KVM_GUESTDBG_INJECT_BP: inject BP type exception [x86] 3755 - KVM_GUESTDBG_EXIT_PENDING: trigger an immediate guest exit [s390] 3756 - KVM_GUESTDBG_BLOCKIRQ: avoid injecting interrupts/NMI/SMI [x86] 3757 3758For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints 3759are enabled in memory so we need to ensure breakpoint exceptions are 3760correctly trapped and the KVM run loop exits at the breakpoint and not 3761running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP 3762we need to ensure the guest vCPUs architecture specific registers are 3763updated to the correct (supplied) values. 3764 3765The second part of the structure is architecture specific and 3766typically contains a set of debug registers. 3767 3768For arm64 the number of debug registers is implementation defined and 3769can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and 3770KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number 3771indicating the number of supported registers. 3772 3773For ppc, the KVM_CAP_PPC_GUEST_DEBUG_SSTEP capability indicates whether 3774the single-step debug event (KVM_GUESTDBG_SINGLESTEP) is supported. 3775 3776Also when supported, KVM_CAP_SET_GUEST_DEBUG2 capability indicates the 3777supported KVM_GUESTDBG_* bits in the control field. 3778 3779When debug events exit the main run loop with the reason 3780KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run 3781structure containing architecture specific debug information. 3782 37834.88 KVM_GET_EMULATED_CPUID 3784--------------------------- 3785 3786:Capability: KVM_CAP_EXT_EMUL_CPUID 3787:Architectures: x86 3788:Type: system ioctl 3789:Parameters: struct kvm_cpuid2 (in/out) 3790:Returns: 0 on success, -1 on error 3791 3792:: 3793 3794 struct kvm_cpuid2 { 3795 __u32 nent; 3796 __u32 flags; 3797 struct kvm_cpuid_entry2 entries[0]; 3798 }; 3799 3800The member 'flags' is used for passing flags from userspace. 3801 3802:: 3803 3804 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0) 3805 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */ 3806 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */ 3807 3808 struct kvm_cpuid_entry2 { 3809 __u32 function; 3810 __u32 index; 3811 __u32 flags; 3812 __u32 eax; 3813 __u32 ebx; 3814 __u32 ecx; 3815 __u32 edx; 3816 __u32 padding[3]; 3817 }; 3818 3819This ioctl returns x86 cpuid features which are emulated by 3820kvm.Userspace can use the information returned by this ioctl to query 3821which features are emulated by kvm instead of being present natively. 3822 3823Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2 3824structure with the 'nent' field indicating the number of entries in 3825the variable-size array 'entries'. If the number of entries is too low 3826to describe the cpu capabilities, an error (E2BIG) is returned. If the 3827number is too high, the 'nent' field is adjusted and an error (ENOMEM) 3828is returned. If the number is just right, the 'nent' field is adjusted 3829to the number of valid entries in the 'entries' array, which is then 3830filled. 3831 3832The entries returned are the set CPUID bits of the respective features 3833which kvm emulates, as returned by the CPUID instruction, with unknown 3834or unsupported feature bits cleared. 3835 3836Features like x2apic, for example, may not be present in the host cpu 3837but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be 3838emulated efficiently and thus not included here. 3839 3840The fields in each entry are defined as follows: 3841 3842 function: 3843 the eax value used to obtain the entry 3844 index: 3845 the ecx value used to obtain the entry (for entries that are 3846 affected by ecx) 3847 flags: 3848 an OR of zero or more of the following: 3849 3850 KVM_CPUID_FLAG_SIGNIFCANT_INDEX: 3851 if the index field is valid 3852 3853 eax, ebx, ecx, edx: 3854 3855 the values returned by the cpuid instruction for 3856 this function/index combination 3857 38584.89 KVM_S390_MEM_OP 3859-------------------- 3860 3861:Capability: KVM_CAP_S390_MEM_OP, KVM_CAP_S390_PROTECTED, KVM_CAP_S390_MEM_OP_EXTENSION 3862:Architectures: s390 3863:Type: vm ioctl, vcpu ioctl 3864:Parameters: struct kvm_s390_mem_op (in) 3865:Returns: = 0 on success, 3866 < 0 on generic error (e.g. -EFAULT or -ENOMEM), 3867 16 bit program exception code if the access causes such an exception 3868 3869Read or write data from/to the VM's memory. 3870The KVM_CAP_S390_MEM_OP_EXTENSION capability specifies what functionality is 3871supported. 3872 3873Parameters are specified via the following structure:: 3874 3875 struct kvm_s390_mem_op { 3876 __u64 gaddr; /* the guest address */ 3877 __u64 flags; /* flags */ 3878 __u32 size; /* amount of bytes */ 3879 __u32 op; /* type of operation */ 3880 __u64 buf; /* buffer in userspace */ 3881 union { 3882 struct { 3883 __u8 ar; /* the access register number */ 3884 __u8 key; /* access key, ignored if flag unset */ 3885 __u8 pad1[6]; /* ignored */ 3886 __u64 old_addr; /* ignored if flag unset */ 3887 }; 3888 __u32 sida_offset; /* offset into the sida */ 3889 __u8 reserved[32]; /* ignored */ 3890 }; 3891 }; 3892 3893The start address of the memory region has to be specified in the "gaddr" 3894field, and the length of the region in the "size" field (which must not 3895be 0). The maximum value for "size" can be obtained by checking the 3896KVM_CAP_S390_MEM_OP capability. "buf" is the buffer supplied by the 3897userspace application where the read data should be written to for 3898a read access, or where the data that should be written is stored for 3899a write access. The "reserved" field is meant for future extensions. 3900Reserved and unused values are ignored. Future extension that add members must 3901introduce new flags. 3902 3903The type of operation is specified in the "op" field. Flags modifying 3904their behavior can be set in the "flags" field. Undefined flag bits must 3905be set to 0. 3906 3907Possible operations are: 3908 * ``KVM_S390_MEMOP_LOGICAL_READ`` 3909 * ``KVM_S390_MEMOP_LOGICAL_WRITE`` 3910 * ``KVM_S390_MEMOP_ABSOLUTE_READ`` 3911 * ``KVM_S390_MEMOP_ABSOLUTE_WRITE`` 3912 * ``KVM_S390_MEMOP_SIDA_READ`` 3913 * ``KVM_S390_MEMOP_SIDA_WRITE`` 3914 * ``KVM_S390_MEMOP_ABSOLUTE_CMPXCHG`` 3915 3916Logical read/write: 3917^^^^^^^^^^^^^^^^^^^ 3918 3919Access logical memory, i.e. translate the given guest address to an absolute 3920address given the state of the VCPU and use the absolute address as target of 3921the access. "ar" designates the access register number to be used; the valid 3922range is 0..15. 3923Logical accesses are permitted for the VCPU ioctl only. 3924Logical accesses are permitted for non-protected guests only. 3925 3926Supported flags: 3927 * ``KVM_S390_MEMOP_F_CHECK_ONLY`` 3928 * ``KVM_S390_MEMOP_F_INJECT_EXCEPTION`` 3929 * ``KVM_S390_MEMOP_F_SKEY_PROTECTION`` 3930 3931The KVM_S390_MEMOP_F_CHECK_ONLY flag can be set to check whether the 3932corresponding memory access would cause an access exception; however, 3933no actual access to the data in memory at the destination is performed. 3934In this case, "buf" is unused and can be NULL. 3935 3936In case an access exception occurred during the access (or would occur 3937in case of KVM_S390_MEMOP_F_CHECK_ONLY), the ioctl returns a positive 3938error number indicating the type of exception. This exception is also 3939raised directly at the corresponding VCPU if the flag 3940KVM_S390_MEMOP_F_INJECT_EXCEPTION is set. 3941On protection exceptions, unless specified otherwise, the injected 3942translation-exception identifier (TEID) indicates suppression. 3943 3944If the KVM_S390_MEMOP_F_SKEY_PROTECTION flag is set, storage key 3945protection is also in effect and may cause exceptions if accesses are 3946prohibited given the access key designated by "key"; the valid range is 0..15. 3947KVM_S390_MEMOP_F_SKEY_PROTECTION is available if KVM_CAP_S390_MEM_OP_EXTENSION 3948is > 0. 3949Since the accessed memory may span multiple pages and those pages might have 3950different storage keys, it is possible that a protection exception occurs 3951after memory has been modified. In this case, if the exception is injected, 3952the TEID does not indicate suppression. 3953 3954Absolute read/write: 3955^^^^^^^^^^^^^^^^^^^^ 3956 3957Access absolute memory. This operation is intended to be used with the 3958KVM_S390_MEMOP_F_SKEY_PROTECTION flag, to allow accessing memory and performing 3959the checks required for storage key protection as one operation (as opposed to 3960user space getting the storage keys, performing the checks, and accessing 3961memory thereafter, which could lead to a delay between check and access). 3962Absolute accesses are permitted for the VM ioctl if KVM_CAP_S390_MEM_OP_EXTENSION 3963has the KVM_S390_MEMOP_EXTENSION_CAP_BASE bit set. 3964Currently absolute accesses are not permitted for VCPU ioctls. 3965Absolute accesses are permitted for non-protected guests only. 3966 3967Supported flags: 3968 * ``KVM_S390_MEMOP_F_CHECK_ONLY`` 3969 * ``KVM_S390_MEMOP_F_SKEY_PROTECTION`` 3970 3971The semantics of the flags common with logical accesses are as for logical 3972accesses. 3973 3974Absolute cmpxchg: 3975^^^^^^^^^^^^^^^^^ 3976 3977Perform cmpxchg on absolute guest memory. Intended for use with the 3978KVM_S390_MEMOP_F_SKEY_PROTECTION flag. 3979Instead of doing an unconditional write, the access occurs only if the target 3980location contains the value pointed to by "old_addr". 3981This is performed as an atomic cmpxchg with the length specified by the "size" 3982parameter. "size" must be a power of two up to and including 16. 3983If the exchange did not take place because the target value doesn't match the 3984old value, the value "old_addr" points to is replaced by the target value. 3985User space can tell if an exchange took place by checking if this replacement 3986occurred. The cmpxchg op is permitted for the VM ioctl if 3987KVM_CAP_S390_MEM_OP_EXTENSION has flag KVM_S390_MEMOP_EXTENSION_CAP_CMPXCHG set. 3988 3989Supported flags: 3990 * ``KVM_S390_MEMOP_F_SKEY_PROTECTION`` 3991 3992SIDA read/write: 3993^^^^^^^^^^^^^^^^ 3994 3995Access the secure instruction data area which contains memory operands necessary 3996for instruction emulation for protected guests. 3997SIDA accesses are available if the KVM_CAP_S390_PROTECTED capability is available. 3998SIDA accesses are permitted for the VCPU ioctl only. 3999SIDA accesses are permitted for protected guests only. 4000 4001No flags are supported. 4002 40034.90 KVM_S390_GET_SKEYS 4004----------------------- 4005 4006:Capability: KVM_CAP_S390_SKEYS 4007:Architectures: s390 4008:Type: vm ioctl 4009:Parameters: struct kvm_s390_skeys 4010:Returns: 0 on success, KVM_S390_GET_SKEYS_NONE if guest is not using storage 4011 keys, negative value on error 4012 4013This ioctl is used to get guest storage key values on the s390 4014architecture. The ioctl takes parameters via the kvm_s390_skeys struct:: 4015 4016 struct kvm_s390_skeys { 4017 __u64 start_gfn; 4018 __u64 count; 4019 __u64 skeydata_addr; 4020 __u32 flags; 4021 __u32 reserved[9]; 4022 }; 4023 4024The start_gfn field is the number of the first guest frame whose storage keys 4025you want to get. 4026 4027The count field is the number of consecutive frames (starting from start_gfn) 4028whose storage keys to get. The count field must be at least 1 and the maximum 4029allowed value is defined as KVM_S390_SKEYS_MAX. Values outside this range 4030will cause the ioctl to return -EINVAL. 4031 4032The skeydata_addr field is the address to a buffer large enough to hold count 4033bytes. This buffer will be filled with storage key data by the ioctl. 4034 40354.91 KVM_S390_SET_SKEYS 4036----------------------- 4037 4038:Capability: KVM_CAP_S390_SKEYS 4039:Architectures: s390 4040:Type: vm ioctl 4041:Parameters: struct kvm_s390_skeys 4042:Returns: 0 on success, negative value on error 4043 4044This ioctl is used to set guest storage key values on the s390 4045architecture. The ioctl takes parameters via the kvm_s390_skeys struct. 4046See section on KVM_S390_GET_SKEYS for struct definition. 4047 4048The start_gfn field is the number of the first guest frame whose storage keys 4049you want to set. 4050 4051The count field is the number of consecutive frames (starting from start_gfn) 4052whose storage keys to get. The count field must be at least 1 and the maximum 4053allowed value is defined as KVM_S390_SKEYS_MAX. Values outside this range 4054will cause the ioctl to return -EINVAL. 4055 4056The skeydata_addr field is the address to a buffer containing count bytes of 4057storage keys. Each byte in the buffer will be set as the storage key for a 4058single frame starting at start_gfn for count frames. 4059 4060Note: If any architecturally invalid key value is found in the given data then 4061the ioctl will return -EINVAL. 4062 40634.92 KVM_S390_IRQ 4064----------------- 4065 4066:Capability: KVM_CAP_S390_INJECT_IRQ 4067:Architectures: s390 4068:Type: vcpu ioctl 4069:Parameters: struct kvm_s390_irq (in) 4070:Returns: 0 on success, -1 on error 4071 4072Errors: 4073 4074 4075 ====== ================================================================= 4076 EINVAL interrupt type is invalid 4077 type is KVM_S390_SIGP_STOP and flag parameter is invalid value, 4078 type is KVM_S390_INT_EXTERNAL_CALL and code is bigger 4079 than the maximum of VCPUs 4080 EBUSY type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped, 4081 type is KVM_S390_SIGP_STOP and a stop irq is already pending, 4082 type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt 4083 is already pending 4084 ====== ================================================================= 4085 4086Allows to inject an interrupt to the guest. 4087 4088Using struct kvm_s390_irq as a parameter allows 4089to inject additional payload which is not 4090possible via KVM_S390_INTERRUPT. 4091 4092Interrupt parameters are passed via kvm_s390_irq:: 4093 4094 struct kvm_s390_irq { 4095 __u64 type; 4096 union { 4097 struct kvm_s390_io_info io; 4098 struct kvm_s390_ext_info ext; 4099 struct kvm_s390_pgm_info pgm; 4100 struct kvm_s390_emerg_info emerg; 4101 struct kvm_s390_extcall_info extcall; 4102 struct kvm_s390_prefix_info prefix; 4103 struct kvm_s390_stop_info stop; 4104 struct kvm_s390_mchk_info mchk; 4105 char reserved[64]; 4106 } u; 4107 }; 4108 4109type can be one of the following: 4110 4111- KVM_S390_SIGP_STOP - sigp stop; parameter in .stop 4112- KVM_S390_PROGRAM_INT - program check; parameters in .pgm 4113- KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix 4114- KVM_S390_RESTART - restart; no parameters 4115- KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters 4116- KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters 4117- KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg 4118- KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall 4119- KVM_S390_MCHK - machine check interrupt; parameters in .mchk 4120 4121This is an asynchronous vcpu ioctl and can be invoked from any thread. 4122 41234.94 KVM_S390_GET_IRQ_STATE 4124--------------------------- 4125 4126:Capability: KVM_CAP_S390_IRQ_STATE 4127:Architectures: s390 4128:Type: vcpu ioctl 4129:Parameters: struct kvm_s390_irq_state (out) 4130:Returns: >= number of bytes copied into buffer, 4131 -EINVAL if buffer size is 0, 4132 -ENOBUFS if buffer size is too small to fit all pending interrupts, 4133 -EFAULT if the buffer address was invalid 4134 4135This ioctl allows userspace to retrieve the complete state of all currently 4136pending interrupts in a single buffer. Use cases include migration 4137and introspection. The parameter structure contains the address of a 4138userspace buffer and its length:: 4139 4140 struct kvm_s390_irq_state { 4141 __u64 buf; 4142 __u32 flags; /* will stay unused for compatibility reasons */ 4143 __u32 len; 4144 __u32 reserved[4]; /* will stay unused for compatibility reasons */ 4145 }; 4146 4147Userspace passes in the above struct and for each pending interrupt a 4148struct kvm_s390_irq is copied to the provided buffer. 4149 4150The structure contains a flags and a reserved field for future extensions. As 4151the kernel never checked for flags == 0 and QEMU never pre-zeroed flags and 4152reserved, these fields can not be used in the future without breaking 4153compatibility. 4154 4155If -ENOBUFS is returned the buffer provided was too small and userspace 4156may retry with a bigger buffer. 4157 41584.95 KVM_S390_SET_IRQ_STATE 4159--------------------------- 4160 4161:Capability: KVM_CAP_S390_IRQ_STATE 4162:Architectures: s390 4163:Type: vcpu ioctl 4164:Parameters: struct kvm_s390_irq_state (in) 4165:Returns: 0 on success, 4166 -EFAULT if the buffer address was invalid, 4167 -EINVAL for an invalid buffer length (see below), 4168 -EBUSY if there were already interrupts pending, 4169 errors occurring when actually injecting the 4170 interrupt. See KVM_S390_IRQ. 4171 4172This ioctl allows userspace to set the complete state of all cpu-local 4173interrupts currently pending for the vcpu. It is intended for restoring 4174interrupt state after a migration. The input parameter is a userspace buffer 4175containing a struct kvm_s390_irq_state:: 4176 4177 struct kvm_s390_irq_state { 4178 __u64 buf; 4179 __u32 flags; /* will stay unused for compatibility reasons */ 4180 __u32 len; 4181 __u32 reserved[4]; /* will stay unused for compatibility reasons */ 4182 }; 4183 4184The restrictions for flags and reserved apply as well. 4185(see KVM_S390_GET_IRQ_STATE) 4186 4187The userspace memory referenced by buf contains a struct kvm_s390_irq 4188for each interrupt to be injected into the guest. 4189If one of the interrupts could not be injected for some reason the 4190ioctl aborts. 4191 4192len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0 4193and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq), 4194which is the maximum number of possibly pending cpu-local interrupts. 4195 41964.96 KVM_SMI 4197------------ 4198 4199:Capability: KVM_CAP_X86_SMM 4200:Architectures: x86 4201:Type: vcpu ioctl 4202:Parameters: none 4203:Returns: 0 on success, -1 on error 4204 4205Queues an SMI on the thread's vcpu. 4206 42074.97 KVM_X86_SET_MSR_FILTER 4208---------------------------- 4209 4210:Capability: KVM_CAP_X86_MSR_FILTER 4211:Architectures: x86 4212:Type: vm ioctl 4213:Parameters: struct kvm_msr_filter 4214:Returns: 0 on success, < 0 on error 4215 4216:: 4217 4218 struct kvm_msr_filter_range { 4219 #define KVM_MSR_FILTER_READ (1 << 0) 4220 #define KVM_MSR_FILTER_WRITE (1 << 1) 4221 __u32 flags; 4222 __u32 nmsrs; /* number of msrs in bitmap */ 4223 __u32 base; /* MSR index the bitmap starts at */ 4224 __u8 *bitmap; /* a 1 bit allows the operations in flags, 0 denies */ 4225 }; 4226 4227 #define KVM_MSR_FILTER_MAX_RANGES 16 4228 struct kvm_msr_filter { 4229 #define KVM_MSR_FILTER_DEFAULT_ALLOW (0 << 0) 4230 #define KVM_MSR_FILTER_DEFAULT_DENY (1 << 0) 4231 __u32 flags; 4232 struct kvm_msr_filter_range ranges[KVM_MSR_FILTER_MAX_RANGES]; 4233 }; 4234 4235flags values for ``struct kvm_msr_filter_range``: 4236 4237``KVM_MSR_FILTER_READ`` 4238 4239 Filter read accesses to MSRs using the given bitmap. A 0 in the bitmap 4240 indicates that read accesses should be denied, while a 1 indicates that 4241 a read for a particular MSR should be allowed regardless of the default 4242 filter action. 4243 4244``KVM_MSR_FILTER_WRITE`` 4245 4246 Filter write accesses to MSRs using the given bitmap. A 0 in the bitmap 4247 indicates that write accesses should be denied, while a 1 indicates that 4248 a write for a particular MSR should be allowed regardless of the default 4249 filter action. 4250 4251flags values for ``struct kvm_msr_filter``: 4252 4253``KVM_MSR_FILTER_DEFAULT_ALLOW`` 4254 4255 If no filter range matches an MSR index that is getting accessed, KVM will 4256 allow accesses to all MSRs by default. 4257 4258``KVM_MSR_FILTER_DEFAULT_DENY`` 4259 4260 If no filter range matches an MSR index that is getting accessed, KVM will 4261 deny accesses to all MSRs by default. 4262 4263This ioctl allows userspace to define up to 16 bitmaps of MSR ranges to deny 4264guest MSR accesses that would normally be allowed by KVM. If an MSR is not 4265covered by a specific range, the "default" filtering behavior applies. Each 4266bitmap range covers MSRs from [base .. base+nmsrs). 4267 4268If an MSR access is denied by userspace, the resulting KVM behavior depends on 4269whether or not KVM_CAP_X86_USER_SPACE_MSR's KVM_MSR_EXIT_REASON_FILTER is 4270enabled. If KVM_MSR_EXIT_REASON_FILTER is enabled, KVM will exit to userspace 4271on denied accesses, i.e. userspace effectively intercepts the MSR access. If 4272KVM_MSR_EXIT_REASON_FILTER is not enabled, KVM will inject a #GP into the guest 4273on denied accesses. Note, if an MSR access is denied during emulation of MSR 4274load/stores during VMX transitions, KVM ignores KVM_MSR_EXIT_REASON_FILTER. 4275See the below warning for full details. 4276 4277If an MSR access is allowed by userspace, KVM will emulate and/or virtualize 4278the access in accordance with the vCPU model. Note, KVM may still ultimately 4279inject a #GP if an access is allowed by userspace, e.g. if KVM doesn't support 4280the MSR, or to follow architectural behavior for the MSR. 4281 4282By default, KVM operates in KVM_MSR_FILTER_DEFAULT_ALLOW mode with no MSR range 4283filters. 4284 4285Calling this ioctl with an empty set of ranges (all nmsrs == 0) disables MSR 4286filtering. In that mode, ``KVM_MSR_FILTER_DEFAULT_DENY`` is invalid and causes 4287an error. 4288 4289.. warning:: 4290 MSR accesses that are side effects of instruction execution (emulated or 4291 native) are not filtered as hardware does not honor MSR bitmaps outside of 4292 RDMSR and WRMSR, and KVM mimics that behavior when emulating instructions 4293 to avoid pointless divergence from hardware. E.g. RDPID reads MSR_TSC_AUX, 4294 SYSENTER reads the SYSENTER MSRs, etc. 4295 4296 MSRs that are loaded/stored via dedicated VMCS fields are not filtered as 4297 part of VM-Enter/VM-Exit emulation. 4298 4299 MSRs that are loaded/store via VMX's load/store lists _are_ filtered as part 4300 of VM-Enter/VM-Exit emulation. If an MSR access is denied on VM-Enter, KVM 4301 synthesizes a consistency check VM-Exit(EXIT_REASON_MSR_LOAD_FAIL). If an 4302 MSR access is denied on VM-Exit, KVM synthesizes a VM-Abort. In short, KVM 4303 extends Intel's architectural list of MSRs that cannot be loaded/saved via 4304 the VM-Enter/VM-Exit MSR list. It is platform owner's responsibility to 4305 to communicate any such restrictions to their end users. 4306 4307 x2APIC MSR accesses cannot be filtered (KVM silently ignores filters that 4308 cover any x2APIC MSRs). 4309 4310Note, invoking this ioctl while a vCPU is running is inherently racy. However, 4311KVM does guarantee that vCPUs will see either the previous filter or the new 4312filter, e.g. MSRs with identical settings in both the old and new filter will 4313have deterministic behavior. 4314 4315Similarly, if userspace wishes to intercept on denied accesses, 4316KVM_MSR_EXIT_REASON_FILTER must be enabled before activating any filters, and 4317left enabled until after all filters are deactivated. Failure to do so may 4318result in KVM injecting a #GP instead of exiting to userspace. 4319 43204.98 KVM_CREATE_SPAPR_TCE_64 4321---------------------------- 4322 4323:Capability: KVM_CAP_SPAPR_TCE_64 4324:Architectures: powerpc 4325:Type: vm ioctl 4326:Parameters: struct kvm_create_spapr_tce_64 (in) 4327:Returns: file descriptor for manipulating the created TCE table 4328 4329This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit 4330windows, described in 4.62 KVM_CREATE_SPAPR_TCE 4331 4332This capability uses extended struct in ioctl interface:: 4333 4334 /* for KVM_CAP_SPAPR_TCE_64 */ 4335 struct kvm_create_spapr_tce_64 { 4336 __u64 liobn; 4337 __u32 page_shift; 4338 __u32 flags; 4339 __u64 offset; /* in pages */ 4340 __u64 size; /* in pages */ 4341 }; 4342 4343The aim of extension is to support an additional bigger DMA window with 4344a variable page size. 4345KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and 4346a bus offset of the corresponding DMA window, @size and @offset are numbers 4347of IOMMU pages. 4348 4349@flags are not used at the moment. 4350 4351The rest of functionality is identical to KVM_CREATE_SPAPR_TCE. 4352 43534.99 KVM_REINJECT_CONTROL 4354------------------------- 4355 4356:Capability: KVM_CAP_REINJECT_CONTROL 4357:Architectures: x86 4358:Type: vm ioctl 4359:Parameters: struct kvm_reinject_control (in) 4360:Returns: 0 on success, 4361 -EFAULT if struct kvm_reinject_control cannot be read, 4362 -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier. 4363 4364i8254 (PIT) has two modes, reinject and !reinject. The default is reinject, 4365where KVM queues elapsed i8254 ticks and monitors completion of interrupt from 4366vector(s) that i8254 injects. Reinject mode dequeues a tick and injects its 4367interrupt whenever there isn't a pending interrupt from i8254. 4368!reinject mode injects an interrupt as soon as a tick arrives. 4369 4370:: 4371 4372 struct kvm_reinject_control { 4373 __u8 pit_reinject; 4374 __u8 reserved[31]; 4375 }; 4376 4377pit_reinject = 0 (!reinject mode) is recommended, unless running an old 4378operating system that uses the PIT for timing (e.g. Linux 2.4.x). 4379 43804.100 KVM_PPC_CONFIGURE_V3_MMU 4381------------------------------ 4382 4383:Capability: KVM_CAP_PPC_MMU_RADIX or KVM_CAP_PPC_MMU_HASH_V3 4384:Architectures: ppc 4385:Type: vm ioctl 4386:Parameters: struct kvm_ppc_mmuv3_cfg (in) 4387:Returns: 0 on success, 4388 -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read, 4389 -EINVAL if the configuration is invalid 4390 4391This ioctl controls whether the guest will use radix or HPT (hashed 4392page table) translation, and sets the pointer to the process table for 4393the guest. 4394 4395:: 4396 4397 struct kvm_ppc_mmuv3_cfg { 4398 __u64 flags; 4399 __u64 process_table; 4400 }; 4401 4402There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and 4403KVM_PPC_MMUV3_GTSE. KVM_PPC_MMUV3_RADIX, if set, configures the guest 4404to use radix tree translation, and if clear, to use HPT translation. 4405KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest 4406to be able to use the global TLB and SLB invalidation instructions; 4407if clear, the guest may not use these instructions. 4408 4409The process_table field specifies the address and size of the guest 4410process table, which is in the guest's space. This field is formatted 4411as the second doubleword of the partition table entry, as defined in 4412the Power ISA V3.00, Book III section 5.7.6.1. 4413 44144.101 KVM_PPC_GET_RMMU_INFO 4415--------------------------- 4416 4417:Capability: KVM_CAP_PPC_MMU_RADIX 4418:Architectures: ppc 4419:Type: vm ioctl 4420:Parameters: struct kvm_ppc_rmmu_info (out) 4421:Returns: 0 on success, 4422 -EFAULT if struct kvm_ppc_rmmu_info cannot be written, 4423 -EINVAL if no useful information can be returned 4424 4425This ioctl returns a structure containing two things: (a) a list 4426containing supported radix tree geometries, and (b) a list that maps 4427page sizes to put in the "AP" (actual page size) field for the tlbie 4428(TLB invalidate entry) instruction. 4429 4430:: 4431 4432 struct kvm_ppc_rmmu_info { 4433 struct kvm_ppc_radix_geom { 4434 __u8 page_shift; 4435 __u8 level_bits[4]; 4436 __u8 pad[3]; 4437 } geometries[8]; 4438 __u32 ap_encodings[8]; 4439 }; 4440 4441The geometries[] field gives up to 8 supported geometries for the 4442radix page table, in terms of the log base 2 of the smallest page 4443size, and the number of bits indexed at each level of the tree, from 4444the PTE level up to the PGD level in that order. Any unused entries 4445will have 0 in the page_shift field. 4446 4447The ap_encodings gives the supported page sizes and their AP field 4448encodings, encoded with the AP value in the top 3 bits and the log 4449base 2 of the page size in the bottom 6 bits. 4450 44514.102 KVM_PPC_RESIZE_HPT_PREPARE 4452-------------------------------- 4453 4454:Capability: KVM_CAP_SPAPR_RESIZE_HPT 4455:Architectures: powerpc 4456:Type: vm ioctl 4457:Parameters: struct kvm_ppc_resize_hpt (in) 4458:Returns: 0 on successful completion, 4459 >0 if a new HPT is being prepared, the value is an estimated 4460 number of milliseconds until preparation is complete, 4461 -EFAULT if struct kvm_reinject_control cannot be read, 4462 -EINVAL if the supplied shift or flags are invalid, 4463 -ENOMEM if unable to allocate the new HPT, 4464 4465Used to implement the PAPR extension for runtime resizing of a guest's 4466Hashed Page Table (HPT). Specifically this starts, stops or monitors 4467the preparation of a new potential HPT for the guest, essentially 4468implementing the H_RESIZE_HPT_PREPARE hypercall. 4469 4470:: 4471 4472 struct kvm_ppc_resize_hpt { 4473 __u64 flags; 4474 __u32 shift; 4475 __u32 pad; 4476 }; 4477 4478If called with shift > 0 when there is no pending HPT for the guest, 4479this begins preparation of a new pending HPT of size 2^(shift) bytes. 4480It then returns a positive integer with the estimated number of 4481milliseconds until preparation is complete. 4482 4483If called when there is a pending HPT whose size does not match that 4484requested in the parameters, discards the existing pending HPT and 4485creates a new one as above. 4486 4487If called when there is a pending HPT of the size requested, will: 4488 4489 * If preparation of the pending HPT is already complete, return 0 4490 * If preparation of the pending HPT has failed, return an error 4491 code, then discard the pending HPT. 4492 * If preparation of the pending HPT is still in progress, return an 4493 estimated number of milliseconds until preparation is complete. 4494 4495If called with shift == 0, discards any currently pending HPT and 4496returns 0 (i.e. cancels any in-progress preparation). 4497 4498flags is reserved for future expansion, currently setting any bits in 4499flags will result in an -EINVAL. 4500 4501Normally this will be called repeatedly with the same parameters until 4502it returns <= 0. The first call will initiate preparation, subsequent 4503ones will monitor preparation until it completes or fails. 4504 45054.103 KVM_PPC_RESIZE_HPT_COMMIT 4506------------------------------- 4507 4508:Capability: KVM_CAP_SPAPR_RESIZE_HPT 4509:Architectures: powerpc 4510:Type: vm ioctl 4511:Parameters: struct kvm_ppc_resize_hpt (in) 4512:Returns: 0 on successful completion, 4513 -EFAULT if struct kvm_reinject_control cannot be read, 4514 -EINVAL if the supplied shift or flags are invalid, 4515 -ENXIO is there is no pending HPT, or the pending HPT doesn't 4516 have the requested size, 4517 -EBUSY if the pending HPT is not fully prepared, 4518 -ENOSPC if there was a hash collision when moving existing 4519 HPT entries to the new HPT, 4520 -EIO on other error conditions 4521 4522Used to implement the PAPR extension for runtime resizing of a guest's 4523Hashed Page Table (HPT). Specifically this requests that the guest be 4524transferred to working with the new HPT, essentially implementing the 4525H_RESIZE_HPT_COMMIT hypercall. 4526 4527:: 4528 4529 struct kvm_ppc_resize_hpt { 4530 __u64 flags; 4531 __u32 shift; 4532 __u32 pad; 4533 }; 4534 4535This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has 4536returned 0 with the same parameters. In other cases 4537KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or 4538-EBUSY, though others may be possible if the preparation was started, 4539but failed). 4540 4541This will have undefined effects on the guest if it has not already 4542placed itself in a quiescent state where no vcpu will make MMU enabled 4543memory accesses. 4544 4545On successful completion, the pending HPT will become the guest's active 4546HPT and the previous HPT will be discarded. 4547 4548On failure, the guest will still be operating on its previous HPT. 4549 45504.104 KVM_X86_GET_MCE_CAP_SUPPORTED 4551----------------------------------- 4552 4553:Capability: KVM_CAP_MCE 4554:Architectures: x86 4555:Type: system ioctl 4556:Parameters: u64 mce_cap (out) 4557:Returns: 0 on success, -1 on error 4558 4559Returns supported MCE capabilities. The u64 mce_cap parameter 4560has the same format as the MSR_IA32_MCG_CAP register. Supported 4561capabilities will have the corresponding bits set. 4562 45634.105 KVM_X86_SETUP_MCE 4564----------------------- 4565 4566:Capability: KVM_CAP_MCE 4567:Architectures: x86 4568:Type: vcpu ioctl 4569:Parameters: u64 mcg_cap (in) 4570:Returns: 0 on success, 4571 -EFAULT if u64 mcg_cap cannot be read, 4572 -EINVAL if the requested number of banks is invalid, 4573 -EINVAL if requested MCE capability is not supported. 4574 4575Initializes MCE support for use. The u64 mcg_cap parameter 4576has the same format as the MSR_IA32_MCG_CAP register and 4577specifies which capabilities should be enabled. The maximum 4578supported number of error-reporting banks can be retrieved when 4579checking for KVM_CAP_MCE. The supported capabilities can be 4580retrieved with KVM_X86_GET_MCE_CAP_SUPPORTED. 4581 45824.106 KVM_X86_SET_MCE 4583--------------------- 4584 4585:Capability: KVM_CAP_MCE 4586:Architectures: x86 4587:Type: vcpu ioctl 4588:Parameters: struct kvm_x86_mce (in) 4589:Returns: 0 on success, 4590 -EFAULT if struct kvm_x86_mce cannot be read, 4591 -EINVAL if the bank number is invalid, 4592 -EINVAL if VAL bit is not set in status field. 4593 4594Inject a machine check error (MCE) into the guest. The input 4595parameter is:: 4596 4597 struct kvm_x86_mce { 4598 __u64 status; 4599 __u64 addr; 4600 __u64 misc; 4601 __u64 mcg_status; 4602 __u8 bank; 4603 __u8 pad1[7]; 4604 __u64 pad2[3]; 4605 }; 4606 4607If the MCE being reported is an uncorrected error, KVM will 4608inject it as an MCE exception into the guest. If the guest 4609MCG_STATUS register reports that an MCE is in progress, KVM 4610causes an KVM_EXIT_SHUTDOWN vmexit. 4611 4612Otherwise, if the MCE is a corrected error, KVM will just 4613store it in the corresponding bank (provided this bank is 4614not holding a previously reported uncorrected error). 4615 46164.107 KVM_S390_GET_CMMA_BITS 4617---------------------------- 4618 4619:Capability: KVM_CAP_S390_CMMA_MIGRATION 4620:Architectures: s390 4621:Type: vm ioctl 4622:Parameters: struct kvm_s390_cmma_log (in, out) 4623:Returns: 0 on success, a negative value on error 4624 4625Errors: 4626 4627 ====== ============================================================= 4628 ENOMEM not enough memory can be allocated to complete the task 4629 ENXIO if CMMA is not enabled 4630 EINVAL if KVM_S390_CMMA_PEEK is not set but migration mode was not enabled 4631 EINVAL if KVM_S390_CMMA_PEEK is not set but dirty tracking has been 4632 disabled (and thus migration mode was automatically disabled) 4633 EFAULT if the userspace address is invalid or if no page table is 4634 present for the addresses (e.g. when using hugepages). 4635 ====== ============================================================= 4636 4637This ioctl is used to get the values of the CMMA bits on the s390 4638architecture. It is meant to be used in two scenarios: 4639 4640- During live migration to save the CMMA values. Live migration needs 4641 to be enabled via the KVM_REQ_START_MIGRATION VM property. 4642- To non-destructively peek at the CMMA values, with the flag 4643 KVM_S390_CMMA_PEEK set. 4644 4645The ioctl takes parameters via the kvm_s390_cmma_log struct. The desired 4646values are written to a buffer whose location is indicated via the "values" 4647member in the kvm_s390_cmma_log struct. The values in the input struct are 4648also updated as needed. 4649 4650Each CMMA value takes up one byte. 4651 4652:: 4653 4654 struct kvm_s390_cmma_log { 4655 __u64 start_gfn; 4656 __u32 count; 4657 __u32 flags; 4658 union { 4659 __u64 remaining; 4660 __u64 mask; 4661 }; 4662 __u64 values; 4663 }; 4664 4665start_gfn is the number of the first guest frame whose CMMA values are 4666to be retrieved, 4667 4668count is the length of the buffer in bytes, 4669 4670values points to the buffer where the result will be written to. 4671 4672If count is greater than KVM_S390_SKEYS_MAX, then it is considered to be 4673KVM_S390_SKEYS_MAX. KVM_S390_SKEYS_MAX is re-used for consistency with 4674other ioctls. 4675 4676The result is written in the buffer pointed to by the field values, and 4677the values of the input parameter are updated as follows. 4678 4679Depending on the flags, different actions are performed. The only 4680supported flag so far is KVM_S390_CMMA_PEEK. 4681 4682The default behaviour if KVM_S390_CMMA_PEEK is not set is: 4683start_gfn will indicate the first page frame whose CMMA bits were dirty. 4684It is not necessarily the same as the one passed as input, as clean pages 4685are skipped. 4686 4687count will indicate the number of bytes actually written in the buffer. 4688It can (and very often will) be smaller than the input value, since the 4689buffer is only filled until 16 bytes of clean values are found (which 4690are then not copied in the buffer). Since a CMMA migration block needs 4691the base address and the length, for a total of 16 bytes, we will send 4692back some clean data if there is some dirty data afterwards, as long as 4693the size of the clean data does not exceed the size of the header. This 4694allows to minimize the amount of data to be saved or transferred over 4695the network at the expense of more roundtrips to userspace. The next 4696invocation of the ioctl will skip over all the clean values, saving 4697potentially more than just the 16 bytes we found. 4698 4699If KVM_S390_CMMA_PEEK is set: 4700the existing storage attributes are read even when not in migration 4701mode, and no other action is performed; 4702 4703the output start_gfn will be equal to the input start_gfn, 4704 4705the output count will be equal to the input count, except if the end of 4706memory has been reached. 4707 4708In both cases: 4709the field "remaining" will indicate the total number of dirty CMMA values 4710still remaining, or 0 if KVM_S390_CMMA_PEEK is set and migration mode is 4711not enabled. 4712 4713mask is unused. 4714 4715values points to the userspace buffer where the result will be stored. 4716 47174.108 KVM_S390_SET_CMMA_BITS 4718---------------------------- 4719 4720:Capability: KVM_CAP_S390_CMMA_MIGRATION 4721:Architectures: s390 4722:Type: vm ioctl 4723:Parameters: struct kvm_s390_cmma_log (in) 4724:Returns: 0 on success, a negative value on error 4725 4726This ioctl is used to set the values of the CMMA bits on the s390 4727architecture. It is meant to be used during live migration to restore 4728the CMMA values, but there are no restrictions on its use. 4729The ioctl takes parameters via the kvm_s390_cmma_values struct. 4730Each CMMA value takes up one byte. 4731 4732:: 4733 4734 struct kvm_s390_cmma_log { 4735 __u64 start_gfn; 4736 __u32 count; 4737 __u32 flags; 4738 union { 4739 __u64 remaining; 4740 __u64 mask; 4741 }; 4742 __u64 values; 4743 }; 4744 4745start_gfn indicates the starting guest frame number, 4746 4747count indicates how many values are to be considered in the buffer, 4748 4749flags is not used and must be 0. 4750 4751mask indicates which PGSTE bits are to be considered. 4752 4753remaining is not used. 4754 4755values points to the buffer in userspace where to store the values. 4756 4757This ioctl can fail with -ENOMEM if not enough memory can be allocated to 4758complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if 4759the count field is too large (e.g. more than KVM_S390_CMMA_SIZE_MAX) or 4760if the flags field was not 0, with -EFAULT if the userspace address is 4761invalid, if invalid pages are written to (e.g. after the end of memory) 4762or if no page table is present for the addresses (e.g. when using 4763hugepages). 4764 47654.109 KVM_PPC_GET_CPU_CHAR 4766-------------------------- 4767 4768:Capability: KVM_CAP_PPC_GET_CPU_CHAR 4769:Architectures: powerpc 4770:Type: vm ioctl 4771:Parameters: struct kvm_ppc_cpu_char (out) 4772:Returns: 0 on successful completion, 4773 -EFAULT if struct kvm_ppc_cpu_char cannot be written 4774 4775This ioctl gives userspace information about certain characteristics 4776of the CPU relating to speculative execution of instructions and 4777possible information leakage resulting from speculative execution (see 4778CVE-2017-5715, CVE-2017-5753 and CVE-2017-5754). The information is 4779returned in struct kvm_ppc_cpu_char, which looks like this:: 4780 4781 struct kvm_ppc_cpu_char { 4782 __u64 character; /* characteristics of the CPU */ 4783 __u64 behaviour; /* recommended software behaviour */ 4784 __u64 character_mask; /* valid bits in character */ 4785 __u64 behaviour_mask; /* valid bits in behaviour */ 4786 }; 4787 4788For extensibility, the character_mask and behaviour_mask fields 4789indicate which bits of character and behaviour have been filled in by 4790the kernel. If the set of defined bits is extended in future then 4791userspace will be able to tell whether it is running on a kernel that 4792knows about the new bits. 4793 4794The character field describes attributes of the CPU which can help 4795with preventing inadvertent information disclosure - specifically, 4796whether there is an instruction to flash-invalidate the L1 data cache 4797(ori 30,30,0 or mtspr SPRN_TRIG2,rN), whether the L1 data cache is set 4798to a mode where entries can only be used by the thread that created 4799them, whether the bcctr[l] instruction prevents speculation, and 4800whether a speculation barrier instruction (ori 31,31,0) is provided. 4801 4802The behaviour field describes actions that software should take to 4803prevent inadvertent information disclosure, and thus describes which 4804vulnerabilities the hardware is subject to; specifically whether the 4805L1 data cache should be flushed when returning to user mode from the 4806kernel, and whether a speculation barrier should be placed between an 4807array bounds check and the array access. 4808 4809These fields use the same bit definitions as the new 4810H_GET_CPU_CHARACTERISTICS hypercall. 4811 48124.110 KVM_MEMORY_ENCRYPT_OP 4813--------------------------- 4814 4815:Capability: basic 4816:Architectures: x86 4817:Type: vm ioctl, vcpu ioctl 4818:Parameters: an opaque platform specific structure (in/out) 4819:Returns: 0 on success; -1 on error 4820 4821If the platform supports creating encrypted VMs then this ioctl can be used 4822for issuing platform-specific memory encryption commands to manage those 4823encrypted VMs. 4824 4825Currently, this ioctl is used for issuing both Secure Encrypted Virtualization 4826(SEV) commands on AMD Processors and Trusted Domain Extensions (TDX) commands 4827on Intel Processors. The detailed commands are defined in 4828Documentation/virt/kvm/x86/amd-memory-encryption.rst and 4829Documentation/virt/kvm/x86/intel-tdx.rst. 4830 48314.111 KVM_MEMORY_ENCRYPT_REG_REGION 4832----------------------------------- 4833 4834:Capability: basic 4835:Architectures: x86 4836:Type: system 4837:Parameters: struct kvm_enc_region (in) 4838:Returns: 0 on success; -1 on error 4839 4840This ioctl can be used to register a guest memory region which may 4841contain encrypted data (e.g. guest RAM, SMRAM etc). 4842 4843It is used in the SEV-enabled guest. When encryption is enabled, a guest 4844memory region may contain encrypted data. The SEV memory encryption 4845engine uses a tweak such that two identical plaintext pages, each at 4846different locations will have differing ciphertexts. So swapping or 4847moving ciphertext of those pages will not result in plaintext being 4848swapped. So relocating (or migrating) physical backing pages for the SEV 4849guest will require some additional steps. 4850 4851Note: The current SEV key management spec does not provide commands to 4852swap or migrate (move) ciphertext pages. Hence, for now we pin the guest 4853memory region registered with the ioctl. 4854 48554.112 KVM_MEMORY_ENCRYPT_UNREG_REGION 4856------------------------------------- 4857 4858:Capability: basic 4859:Architectures: x86 4860:Type: system 4861:Parameters: struct kvm_enc_region (in) 4862:Returns: 0 on success; -1 on error 4863 4864This ioctl can be used to unregister the guest memory region registered 4865with KVM_MEMORY_ENCRYPT_REG_REGION ioctl above. 4866 48674.113 KVM_HYPERV_EVENTFD 4868------------------------ 4869 4870:Capability: KVM_CAP_HYPERV_EVENTFD 4871:Architectures: x86 4872:Type: vm ioctl 4873:Parameters: struct kvm_hyperv_eventfd (in) 4874 4875This ioctl (un)registers an eventfd to receive notifications from the guest on 4876the specified Hyper-V connection id through the SIGNAL_EVENT hypercall, without 4877causing a user exit. SIGNAL_EVENT hypercall with non-zero event flag number 4878(bits 24-31) still triggers a KVM_EXIT_HYPERV_HCALL user exit. 4879 4880:: 4881 4882 struct kvm_hyperv_eventfd { 4883 __u32 conn_id; 4884 __s32 fd; 4885 __u32 flags; 4886 __u32 padding[3]; 4887 }; 4888 4889The conn_id field should fit within 24 bits:: 4890 4891 #define KVM_HYPERV_CONN_ID_MASK 0x00ffffff 4892 4893The acceptable values for the flags field are:: 4894 4895 #define KVM_HYPERV_EVENTFD_DEASSIGN (1 << 0) 4896 4897:Returns: 0 on success, 4898 -EINVAL if conn_id or flags is outside the allowed range, 4899 -ENOENT on deassign if the conn_id isn't registered, 4900 -EEXIST on assign if the conn_id is already registered 4901 49024.114 KVM_GET_NESTED_STATE 4903-------------------------- 4904 4905:Capability: KVM_CAP_NESTED_STATE 4906:Architectures: x86 4907:Type: vcpu ioctl 4908:Parameters: struct kvm_nested_state (in/out) 4909:Returns: 0 on success, -1 on error 4910 4911Errors: 4912 4913 ===== ============================================================= 4914 E2BIG the total state size exceeds the value of 'size' specified by 4915 the user; the size required will be written into size. 4916 ===== ============================================================= 4917 4918:: 4919 4920 struct kvm_nested_state { 4921 __u16 flags; 4922 __u16 format; 4923 __u32 size; 4924 4925 union { 4926 struct kvm_vmx_nested_state_hdr vmx; 4927 struct kvm_svm_nested_state_hdr svm; 4928 4929 /* Pad the header to 128 bytes. */ 4930 __u8 pad[120]; 4931 } hdr; 4932 4933 union { 4934 struct kvm_vmx_nested_state_data vmx[0]; 4935 struct kvm_svm_nested_state_data svm[0]; 4936 } data; 4937 }; 4938 4939 #define KVM_STATE_NESTED_GUEST_MODE 0x00000001 4940 #define KVM_STATE_NESTED_RUN_PENDING 0x00000002 4941 #define KVM_STATE_NESTED_EVMCS 0x00000004 4942 4943 #define KVM_STATE_NESTED_FORMAT_VMX 0 4944 #define KVM_STATE_NESTED_FORMAT_SVM 1 4945 4946 #define KVM_STATE_NESTED_VMX_VMCS_SIZE 0x1000 4947 #define KVM_STATE_NESTED_SVM_VMCB_SIZE 0x1000 4948 4949 #define KVM_STATE_NESTED_VMX_SMM_GUEST_MODE 0x00000001 4950 #define KVM_STATE_NESTED_VMX_SMM_VMXON 0x00000002 4951 4952 #define KVM_STATE_NESTED_GIF_SET 0x00000100 4953 4954 #define KVM_STATE_VMX_PREEMPTION_TIMER_DEADLINE 0x00000001 4955 4956 struct kvm_vmx_nested_state_hdr { 4957 __u64 vmxon_pa; 4958 __u64 vmcs12_pa; 4959 4960 struct { 4961 __u16 flags; 4962 } smm; 4963 4964 __u32 flags; 4965 __u64 preemption_timer_deadline; 4966 }; 4967 4968 struct kvm_svm_nested_state_hdr { 4969 __u64 vmcb_pa; 4970 __u64 gpat; 4971 }; 4972 4973 struct kvm_vmx_nested_state_data { 4974 __u8 vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE]; 4975 __u8 shadow_vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE]; 4976 }; 4977 4978 struct kvm_svm_nested_state_data { 4979 __u8 vmcb12[KVM_STATE_NESTED_SVM_VMCB_SIZE]; 4980 }; 4981 4982This ioctl copies the vcpu's nested virtualization state from the kernel to 4983userspace. 4984 4985The maximum size of the state can be retrieved by passing KVM_CAP_NESTED_STATE 4986to the KVM_CHECK_EXTENSION ioctl(). 4987 49884.115 KVM_SET_NESTED_STATE 4989-------------------------- 4990 4991:Capability: KVM_CAP_NESTED_STATE 4992:Architectures: x86 4993:Type: vcpu ioctl 4994:Parameters: struct kvm_nested_state (in) 4995:Returns: 0 on success, -1 on error 4996 4997This copies the vcpu's kvm_nested_state struct from userspace to the kernel. 4998For the definition of struct kvm_nested_state, see KVM_GET_NESTED_STATE. 4999 50004.116 KVM_(UN)REGISTER_COALESCED_MMIO 5001------------------------------------- 5002 5003:Capability: KVM_CAP_COALESCED_MMIO (for coalesced mmio) 5004 KVM_CAP_COALESCED_PIO (for coalesced pio) 5005:Architectures: all 5006:Type: vm ioctl 5007:Parameters: struct kvm_coalesced_mmio_zone 5008:Returns: 0 on success, < 0 on error 5009 5010Coalesced I/O is a performance optimization that defers hardware 5011register write emulation so that userspace exits are avoided. It is 5012typically used to reduce the overhead of emulating frequently accessed 5013hardware registers. 5014 5015When a hardware register is configured for coalesced I/O, write accesses 5016do not exit to userspace and their value is recorded in a ring buffer 5017that is shared between kernel and userspace. 5018 5019Coalesced I/O is used if one or more write accesses to a hardware 5020register can be deferred until a read or a write to another hardware 5021register on the same device. This last access will cause a vmexit and 5022userspace will process accesses from the ring buffer before emulating 5023it. That will avoid exiting to userspace on repeated writes. 5024 5025Coalesced pio is based on coalesced mmio. There is little difference 5026between coalesced mmio and pio except that coalesced pio records accesses 5027to I/O ports. 5028 50294.117 KVM_CLEAR_DIRTY_LOG 5030------------------------- 5031 5032:Capability: KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 5033:Architectures: x86, arm64, mips 5034:Type: vm ioctl 5035:Parameters: struct kvm_clear_dirty_log (in) 5036:Returns: 0 on success, -1 on error 5037 5038:: 5039 5040 /* for KVM_CLEAR_DIRTY_LOG */ 5041 struct kvm_clear_dirty_log { 5042 __u32 slot; 5043 __u32 num_pages; 5044 __u64 first_page; 5045 union { 5046 void __user *dirty_bitmap; /* one bit per page */ 5047 __u64 padding; 5048 }; 5049 }; 5050 5051The ioctl clears the dirty status of pages in a memory slot, according to 5052the bitmap that is passed in struct kvm_clear_dirty_log's dirty_bitmap 5053field. Bit 0 of the bitmap corresponds to page "first_page" in the 5054memory slot, and num_pages is the size in bits of the input bitmap. 5055first_page must be a multiple of 64; num_pages must also be a multiple of 505664 unless first_page + num_pages is the size of the memory slot. For each 5057bit that is set in the input bitmap, the corresponding page is marked "clean" 5058in KVM's dirty bitmap, and dirty tracking is re-enabled for that page 5059(for example via write-protection, or by clearing the dirty bit in 5060a page table entry). 5061 5062If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies 5063the address space for which you want to clear the dirty status. See 5064KVM_SET_USER_MEMORY_REGION for details on the usage of slot field. 5065 5066This ioctl is mostly useful when KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 5067is enabled; for more information, see the description of the capability. 5068However, it can always be used as long as KVM_CHECK_EXTENSION confirms 5069that KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is present. 5070 50714.118 KVM_GET_SUPPORTED_HV_CPUID 5072-------------------------------- 5073 5074:Capability: KVM_CAP_HYPERV_CPUID (vcpu), KVM_CAP_SYS_HYPERV_CPUID (system) 5075:Architectures: x86 5076:Type: system ioctl, vcpu ioctl 5077:Parameters: struct kvm_cpuid2 (in/out) 5078:Returns: 0 on success, -1 on error 5079 5080:: 5081 5082 struct kvm_cpuid2 { 5083 __u32 nent; 5084 __u32 padding; 5085 struct kvm_cpuid_entry2 entries[0]; 5086 }; 5087 5088 struct kvm_cpuid_entry2 { 5089 __u32 function; 5090 __u32 index; 5091 __u32 flags; 5092 __u32 eax; 5093 __u32 ebx; 5094 __u32 ecx; 5095 __u32 edx; 5096 __u32 padding[3]; 5097 }; 5098 5099This ioctl returns x86 cpuid features leaves related to Hyper-V emulation in 5100KVM. Userspace can use the information returned by this ioctl to construct 5101cpuid information presented to guests consuming Hyper-V enlightenments (e.g. 5102Windows or Hyper-V guests). 5103 5104CPUID feature leaves returned by this ioctl are defined by Hyper-V Top Level 5105Functional Specification (TLFS). These leaves can't be obtained with 5106KVM_GET_SUPPORTED_CPUID ioctl because some of them intersect with KVM feature 5107leaves (0x40000000, 0x40000001). 5108 5109Currently, the following list of CPUID leaves are returned: 5110 5111 - HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS 5112 - HYPERV_CPUID_INTERFACE 5113 - HYPERV_CPUID_VERSION 5114 - HYPERV_CPUID_FEATURES 5115 - HYPERV_CPUID_ENLIGHTMENT_INFO 5116 - HYPERV_CPUID_IMPLEMENT_LIMITS 5117 - HYPERV_CPUID_NESTED_FEATURES 5118 - HYPERV_CPUID_SYNDBG_VENDOR_AND_MAX_FUNCTIONS 5119 - HYPERV_CPUID_SYNDBG_INTERFACE 5120 - HYPERV_CPUID_SYNDBG_PLATFORM_CAPABILITIES 5121 5122Userspace invokes KVM_GET_SUPPORTED_HV_CPUID by passing a kvm_cpuid2 structure 5123with the 'nent' field indicating the number of entries in the variable-size 5124array 'entries'. If the number of entries is too low to describe all Hyper-V 5125feature leaves, an error (E2BIG) is returned. If the number is more or equal 5126to the number of Hyper-V feature leaves, the 'nent' field is adjusted to the 5127number of valid entries in the 'entries' array, which is then filled. 5128 5129'index' and 'flags' fields in 'struct kvm_cpuid_entry2' are currently reserved, 5130userspace should not expect to get any particular value there. 5131 5132Note, vcpu version of KVM_GET_SUPPORTED_HV_CPUID is currently deprecated. Unlike 5133system ioctl which exposes all supported feature bits unconditionally, vcpu 5134version has the following quirks: 5135 5136- HYPERV_CPUID_NESTED_FEATURES leaf and HV_X64_ENLIGHTENED_VMCS_RECOMMENDED 5137 feature bit are only exposed when Enlightened VMCS was previously enabled 5138 on the corresponding vCPU (KVM_CAP_HYPERV_ENLIGHTENED_VMCS). 5139- HV_STIMER_DIRECT_MODE_AVAILABLE bit is only exposed with in-kernel LAPIC. 5140 (presumes KVM_CREATE_IRQCHIP has already been called). 5141 51424.119 KVM_ARM_VCPU_FINALIZE 5143--------------------------- 5144 5145:Architectures: arm64 5146:Type: vcpu ioctl 5147:Parameters: int feature (in) 5148:Returns: 0 on success, -1 on error 5149 5150Errors: 5151 5152 ====== ============================================================== 5153 EPERM feature not enabled, needs configuration, or already finalized 5154 EINVAL feature unknown or not present 5155 ====== ============================================================== 5156 5157Recognised values for feature: 5158 5159 ===== =========================================== 5160 arm64 KVM_ARM_VCPU_SVE (requires KVM_CAP_ARM_SVE) 5161 ===== =========================================== 5162 5163Finalizes the configuration of the specified vcpu feature. 5164 5165The vcpu must already have been initialised, enabling the affected feature, by 5166means of a successful :ref:`KVM_ARM_VCPU_INIT <KVM_ARM_VCPU_INIT>` call with the 5167appropriate flag set in features[]. 5168 5169For affected vcpu features, this is a mandatory step that must be performed 5170before the vcpu is fully usable. 5171 5172Between KVM_ARM_VCPU_INIT and KVM_ARM_VCPU_FINALIZE, the feature may be 5173configured by use of ioctls such as KVM_SET_ONE_REG. The exact configuration 5174that should be performed and how to do it are feature-dependent. 5175 5176Other calls that depend on a particular feature being finalized, such as 5177KVM_RUN, KVM_GET_REG_LIST, KVM_GET_ONE_REG and KVM_SET_ONE_REG, will fail with 5178-EPERM unless the feature has already been finalized by means of a 5179KVM_ARM_VCPU_FINALIZE call. 5180 5181See KVM_ARM_VCPU_INIT for details of vcpu features that require finalization 5182using this ioctl. 5183 51844.120 KVM_SET_PMU_EVENT_FILTER 5185------------------------------ 5186 5187:Capability: KVM_CAP_PMU_EVENT_FILTER 5188:Architectures: x86 5189:Type: vm ioctl 5190:Parameters: struct kvm_pmu_event_filter (in) 5191:Returns: 0 on success, -1 on error 5192 5193Errors: 5194 5195 ====== ============================================================ 5196 EFAULT args[0] cannot be accessed 5197 EINVAL args[0] contains invalid data in the filter or filter events 5198 E2BIG nevents is too large 5199 EBUSY not enough memory to allocate the filter 5200 ====== ============================================================ 5201 5202:: 5203 5204 struct kvm_pmu_event_filter { 5205 __u32 action; 5206 __u32 nevents; 5207 __u32 fixed_counter_bitmap; 5208 __u32 flags; 5209 __u32 pad[4]; 5210 __u64 events[0]; 5211 }; 5212 5213This ioctl restricts the set of PMU events the guest can program by limiting 5214which event select and unit mask combinations are permitted. 5215 5216The argument holds a list of filter events which will be allowed or denied. 5217 5218Filter events only control general purpose counters; fixed purpose counters 5219are controlled by the fixed_counter_bitmap. 5220 5221Valid values for 'flags':: 5222 5223``0`` 5224 5225To use this mode, clear the 'flags' field. 5226 5227In this mode each event will contain an event select + unit mask. 5228 5229When the guest attempts to program the PMU the guest's event select + 5230unit mask is compared against the filter events to determine whether the 5231guest should have access. 5232 5233``KVM_PMU_EVENT_FLAG_MASKED_EVENTS`` 5234:Capability: KVM_CAP_PMU_EVENT_MASKED_EVENTS 5235 5236In this mode each filter event will contain an event select, mask, match, and 5237exclude value. To encode a masked event use:: 5238 5239 KVM_PMU_ENCODE_MASKED_ENTRY() 5240 5241An encoded event will follow this layout:: 5242 5243 Bits Description 5244 ---- ----------- 5245 7:0 event select (low bits) 5246 15:8 umask match 5247 31:16 unused 5248 35:32 event select (high bits) 5249 36:54 unused 5250 55 exclude bit 5251 63:56 umask mask 5252 5253When the guest attempts to program the PMU, these steps are followed in 5254determining if the guest should have access: 5255 5256 1. Match the event select from the guest against the filter events. 5257 2. If a match is found, match the guest's unit mask to the mask and match 5258 values of the included filter events. 5259 I.e. (unit mask & mask) == match && !exclude. 5260 3. If a match is found, match the guest's unit mask to the mask and match 5261 values of the excluded filter events. 5262 I.e. (unit mask & mask) == match && exclude. 5263 4. 5264 a. If an included match is found and an excluded match is not found, filter 5265 the event. 5266 b. For everything else, do not filter the event. 5267 5. 5268 a. If the event is filtered and it's an allow list, allow the guest to 5269 program the event. 5270 b. If the event is filtered and it's a deny list, do not allow the guest to 5271 program the event. 5272 5273When setting a new pmu event filter, -EINVAL will be returned if any of the 5274unused fields are set or if any of the high bits (35:32) in the event 5275select are set when called on Intel. 5276 5277Valid values for 'action':: 5278 5279 #define KVM_PMU_EVENT_ALLOW 0 5280 #define KVM_PMU_EVENT_DENY 1 5281 5282Via this API, KVM userspace can also control the behavior of the VM's fixed 5283counters (if any) by configuring the "action" and "fixed_counter_bitmap" fields. 5284 5285Specifically, KVM follows the following pseudo-code when determining whether to 5286allow the guest FixCtr[i] to count its pre-defined fixed event:: 5287 5288 FixCtr[i]_is_allowed = (action == ALLOW) && (bitmap & BIT(i)) || 5289 (action == DENY) && !(bitmap & BIT(i)); 5290 FixCtr[i]_is_denied = !FixCtr[i]_is_allowed; 5291 5292KVM always consumes fixed_counter_bitmap, it's userspace's responsibility to 5293ensure fixed_counter_bitmap is set correctly, e.g. if userspace wants to define 5294a filter that only affects general purpose counters. 5295 5296Note, the "events" field also applies to fixed counters' hardcoded event_select 5297and unit_mask values. "fixed_counter_bitmap" has higher priority than "events" 5298if there is a contradiction between the two. 5299 53004.121 KVM_PPC_SVM_OFF 5301--------------------- 5302 5303:Capability: basic 5304:Architectures: powerpc 5305:Type: vm ioctl 5306:Parameters: none 5307:Returns: 0 on successful completion, 5308 5309Errors: 5310 5311 ====== ================================================================ 5312 EINVAL if ultravisor failed to terminate the secure guest 5313 ENOMEM if hypervisor failed to allocate new radix page tables for guest 5314 ====== ================================================================ 5315 5316This ioctl is used to turn off the secure mode of the guest or transition 5317the guest from secure mode to normal mode. This is invoked when the guest 5318is reset. This has no effect if called for a normal guest. 5319 5320This ioctl issues an ultravisor call to terminate the secure guest, 5321unpins the VPA pages and releases all the device pages that are used to 5322track the secure pages by hypervisor. 5323 53244.122 KVM_S390_NORMAL_RESET 5325--------------------------- 5326 5327:Capability: KVM_CAP_S390_VCPU_RESETS 5328:Architectures: s390 5329:Type: vcpu ioctl 5330:Parameters: none 5331:Returns: 0 5332 5333This ioctl resets VCPU registers and control structures according to 5334the cpu reset definition in the POP (Principles Of Operation). 5335 53364.123 KVM_S390_INITIAL_RESET 5337---------------------------- 5338 5339:Capability: basic 5340:Architectures: s390 5341:Type: vcpu ioctl 5342:Parameters: none 5343:Returns: 0 5344 5345This ioctl resets VCPU registers and control structures according to 5346the initial cpu reset definition in the POP. However, the cpu is not 5347put into ESA mode. This reset is a superset of the normal reset. 5348 53494.124 KVM_S390_CLEAR_RESET 5350-------------------------- 5351 5352:Capability: KVM_CAP_S390_VCPU_RESETS 5353:Architectures: s390 5354:Type: vcpu ioctl 5355:Parameters: none 5356:Returns: 0 5357 5358This ioctl resets VCPU registers and control structures according to 5359the clear cpu reset definition in the POP. However, the cpu is not put 5360into ESA mode. This reset is a superset of the initial reset. 5361 5362 53634.125 KVM_S390_PV_COMMAND 5364------------------------- 5365 5366:Capability: KVM_CAP_S390_PROTECTED 5367:Architectures: s390 5368:Type: vm ioctl 5369:Parameters: struct kvm_pv_cmd 5370:Returns: 0 on success, < 0 on error 5371 5372:: 5373 5374 struct kvm_pv_cmd { 5375 __u32 cmd; /* Command to be executed */ 5376 __u16 rc; /* Ultravisor return code */ 5377 __u16 rrc; /* Ultravisor return reason code */ 5378 __u64 data; /* Data or address */ 5379 __u32 flags; /* flags for future extensions. Must be 0 for now */ 5380 __u32 reserved[3]; 5381 }; 5382 5383**Ultravisor return codes** 5384The Ultravisor return (reason) codes are provided by the kernel if a 5385Ultravisor call has been executed to achieve the results expected by 5386the command. Therefore they are independent of the IOCTL return 5387code. If KVM changes `rc`, its value will always be greater than 0 5388hence setting it to 0 before issuing a PV command is advised to be 5389able to detect a change of `rc`. 5390 5391**cmd values:** 5392 5393KVM_PV_ENABLE 5394 Allocate memory and register the VM with the Ultravisor, thereby 5395 donating memory to the Ultravisor that will become inaccessible to 5396 KVM. All existing CPUs are converted to protected ones. After this 5397 command has succeeded, any CPU added via hotplug will become 5398 protected during its creation as well. 5399 5400 Errors: 5401 5402 ===== ============================= 5403 EINTR an unmasked signal is pending 5404 ===== ============================= 5405 5406KVM_PV_DISABLE 5407 Deregister the VM from the Ultravisor and reclaim the memory that had 5408 been donated to the Ultravisor, making it usable by the kernel again. 5409 All registered VCPUs are converted back to non-protected ones. If a 5410 previous protected VM had been prepared for asynchronous teardown with 5411 KVM_PV_ASYNC_CLEANUP_PREPARE and not subsequently torn down with 5412 KVM_PV_ASYNC_CLEANUP_PERFORM, it will be torn down in this call 5413 together with the current protected VM. 5414 5415KVM_PV_VM_SET_SEC_PARMS 5416 Pass the image header from VM memory to the Ultravisor in 5417 preparation of image unpacking and verification. 5418 5419KVM_PV_VM_UNPACK 5420 Unpack (protect and decrypt) a page of the encrypted boot image. 5421 5422KVM_PV_VM_VERIFY 5423 Verify the integrity of the unpacked image. Only if this succeeds, 5424 KVM is allowed to start protected VCPUs. 5425 5426KVM_PV_INFO 5427 :Capability: KVM_CAP_S390_PROTECTED_DUMP 5428 5429 Presents an API that provides Ultravisor related data to userspace 5430 via subcommands. len_max is the size of the user space buffer, 5431 len_written is KVM's indication of how much bytes of that buffer 5432 were actually written to. len_written can be used to determine the 5433 valid fields if more response fields are added in the future. 5434 5435 :: 5436 5437 enum pv_cmd_info_id { 5438 KVM_PV_INFO_VM, 5439 KVM_PV_INFO_DUMP, 5440 }; 5441 5442 struct kvm_s390_pv_info_header { 5443 __u32 id; 5444 __u32 len_max; 5445 __u32 len_written; 5446 __u32 reserved; 5447 }; 5448 5449 struct kvm_s390_pv_info { 5450 struct kvm_s390_pv_info_header header; 5451 struct kvm_s390_pv_info_dump dump; 5452 struct kvm_s390_pv_info_vm vm; 5453 }; 5454 5455**subcommands:** 5456 5457 KVM_PV_INFO_VM 5458 This subcommand provides basic Ultravisor information for PV 5459 hosts. These values are likely also exported as files in the sysfs 5460 firmware UV query interface but they are more easily available to 5461 programs in this API. 5462 5463 The installed calls and feature_indication members provide the 5464 installed UV calls and the UV's other feature indications. 5465 5466 The max_* members provide information about the maximum number of PV 5467 vcpus, PV guests and PV guest memory size. 5468 5469 :: 5470 5471 struct kvm_s390_pv_info_vm { 5472 __u64 inst_calls_list[4]; 5473 __u64 max_cpus; 5474 __u64 max_guests; 5475 __u64 max_guest_addr; 5476 __u64 feature_indication; 5477 }; 5478 5479 5480 KVM_PV_INFO_DUMP 5481 This subcommand provides information related to dumping PV guests. 5482 5483 :: 5484 5485 struct kvm_s390_pv_info_dump { 5486 __u64 dump_cpu_buffer_len; 5487 __u64 dump_config_mem_buffer_per_1m; 5488 __u64 dump_config_finalize_len; 5489 }; 5490 5491KVM_PV_DUMP 5492 :Capability: KVM_CAP_S390_PROTECTED_DUMP 5493 5494 Presents an API that provides calls which facilitate dumping a 5495 protected VM. 5496 5497 :: 5498 5499 struct kvm_s390_pv_dmp { 5500 __u64 subcmd; 5501 __u64 buff_addr; 5502 __u64 buff_len; 5503 __u64 gaddr; /* For dump storage state */ 5504 }; 5505 5506 **subcommands:** 5507 5508 KVM_PV_DUMP_INIT 5509 Initializes the dump process of a protected VM. If this call does 5510 not succeed all other subcommands will fail with -EINVAL. This 5511 subcommand will return -EINVAL if a dump process has not yet been 5512 completed. 5513 5514 Not all PV vms can be dumped, the owner needs to set `dump 5515 allowed` PCF bit 34 in the SE header to allow dumping. 5516 5517 KVM_PV_DUMP_CONFIG_STOR_STATE 5518 Stores `buff_len` bytes of tweak component values starting with 5519 the 1MB block specified by the absolute guest address 5520 (`gaddr`). `buff_len` needs to be `conf_dump_storage_state_len` 5521 aligned and at least >= the `conf_dump_storage_state_len` value 5522 provided by the dump uv_info data. buff_user might be written to 5523 even if an error rc is returned. For instance if we encounter a 5524 fault after writing the first page of data. 5525 5526 KVM_PV_DUMP_COMPLETE 5527 If the subcommand succeeds it completes the dump process and lets 5528 KVM_PV_DUMP_INIT be called again. 5529 5530 On success `conf_dump_finalize_len` bytes of completion data will be 5531 stored to the `buff_addr`. The completion data contains a key 5532 derivation seed, IV, tweak nonce and encryption keys as well as an 5533 authentication tag all of which are needed to decrypt the dump at a 5534 later time. 5535 5536KVM_PV_ASYNC_CLEANUP_PREPARE 5537 :Capability: KVM_CAP_S390_PROTECTED_ASYNC_DISABLE 5538 5539 Prepare the current protected VM for asynchronous teardown. Most 5540 resources used by the current protected VM will be set aside for a 5541 subsequent asynchronous teardown. The current protected VM will then 5542 resume execution immediately as non-protected. There can be at most 5543 one protected VM prepared for asynchronous teardown at any time. If 5544 a protected VM had already been prepared for teardown without 5545 subsequently calling KVM_PV_ASYNC_CLEANUP_PERFORM, this call will 5546 fail. In that case, the userspace process should issue a normal 5547 KVM_PV_DISABLE. The resources set aside with this call will need to 5548 be cleaned up with a subsequent call to KVM_PV_ASYNC_CLEANUP_PERFORM 5549 or KVM_PV_DISABLE, otherwise they will be cleaned up when KVM 5550 terminates. KVM_PV_ASYNC_CLEANUP_PREPARE can be called again as soon 5551 as cleanup starts, i.e. before KVM_PV_ASYNC_CLEANUP_PERFORM finishes. 5552 5553KVM_PV_ASYNC_CLEANUP_PERFORM 5554 :Capability: KVM_CAP_S390_PROTECTED_ASYNC_DISABLE 5555 5556 Tear down the protected VM previously prepared for teardown with 5557 KVM_PV_ASYNC_CLEANUP_PREPARE. The resources that had been set aside 5558 will be freed during the execution of this command. This PV command 5559 should ideally be issued by userspace from a separate thread. If a 5560 fatal signal is received (or the process terminates naturally), the 5561 command will terminate immediately without completing, and the normal 5562 KVM shutdown procedure will take care of cleaning up all remaining 5563 protected VMs, including the ones whose teardown was interrupted by 5564 process termination. 5565 55664.126 KVM_XEN_HVM_SET_ATTR 5567-------------------------- 5568 5569:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 5570:Architectures: x86 5571:Type: vm ioctl 5572:Parameters: struct kvm_xen_hvm_attr 5573:Returns: 0 on success, < 0 on error 5574 5575:: 5576 5577 struct kvm_xen_hvm_attr { 5578 __u16 type; 5579 __u16 pad[3]; 5580 union { 5581 __u8 long_mode; 5582 __u8 vector; 5583 __u8 runstate_update_flag; 5584 union { 5585 __u64 gfn; 5586 __u64 hva; 5587 } shared_info; 5588 struct { 5589 __u32 send_port; 5590 __u32 type; /* EVTCHNSTAT_ipi / EVTCHNSTAT_interdomain */ 5591 __u32 flags; 5592 union { 5593 struct { 5594 __u32 port; 5595 __u32 vcpu; 5596 __u32 priority; 5597 } port; 5598 struct { 5599 __u32 port; /* Zero for eventfd */ 5600 __s32 fd; 5601 } eventfd; 5602 __u32 padding[4]; 5603 } deliver; 5604 } evtchn; 5605 __u32 xen_version; 5606 __u64 pad[8]; 5607 } u; 5608 }; 5609 5610type values: 5611 5612KVM_XEN_ATTR_TYPE_LONG_MODE 5613 Sets the ABI mode of the VM to 32-bit or 64-bit (long mode). This 5614 determines the layout of the shared_info page exposed to the VM. 5615 5616KVM_XEN_ATTR_TYPE_SHARED_INFO 5617 Sets the guest physical frame number at which the Xen shared_info 5618 page resides. Note that although Xen places vcpu_info for the first 5619 32 vCPUs in the shared_info page, KVM does not automatically do so 5620 and instead requires that KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO or 5621 KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO_HVA be used explicitly even when 5622 the vcpu_info for a given vCPU resides at the "default" location 5623 in the shared_info page. This is because KVM may not be aware of 5624 the Xen CPU id which is used as the index into the vcpu_info[] 5625 array, so may know the correct default location. 5626 5627 Note that the shared_info page may be constantly written to by KVM; 5628 it contains the event channel bitmap used to deliver interrupts to 5629 a Xen guest, amongst other things. It is exempt from dirty tracking 5630 mechanisms — KVM will not explicitly mark the page as dirty each 5631 time an event channel interrupt is delivered to the guest! Thus, 5632 userspace should always assume that the designated GFN is dirty if 5633 any vCPU has been running or any event channel interrupts can be 5634 routed to the guest. 5635 5636 Setting the gfn to KVM_XEN_INVALID_GFN will disable the shared_info 5637 page. 5638 5639KVM_XEN_ATTR_TYPE_SHARED_INFO_HVA 5640 If the KVM_XEN_HVM_CONFIG_SHARED_INFO_HVA flag is also set in the 5641 Xen capabilities, then this attribute may be used to set the 5642 userspace address at which the shared_info page resides, which 5643 will always be fixed in the VMM regardless of where it is mapped 5644 in guest physical address space. This attribute should be used in 5645 preference to KVM_XEN_ATTR_TYPE_SHARED_INFO as it avoids 5646 unnecessary invalidation of an internal cache when the page is 5647 re-mapped in guest physical address space. 5648 5649 Setting the hva to zero will disable the shared_info page. 5650 5651KVM_XEN_ATTR_TYPE_UPCALL_VECTOR 5652 Sets the exception vector used to deliver Xen event channel upcalls. 5653 This is the HVM-wide vector injected directly by the hypervisor 5654 (not through the local APIC), typically configured by a guest via 5655 HVM_PARAM_CALLBACK_IRQ. This can be disabled again (e.g. for guest 5656 SHUTDOWN_soft_reset) by setting it to zero. 5657 5658KVM_XEN_ATTR_TYPE_EVTCHN 5659 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5660 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It configures 5661 an outbound port number for interception of EVTCHNOP_send requests 5662 from the guest. A given sending port number may be directed back to 5663 a specified vCPU (by APIC ID) / port / priority on the guest, or to 5664 trigger events on an eventfd. The vCPU and priority can be changed 5665 by setting KVM_XEN_EVTCHN_UPDATE in a subsequent call, but other 5666 fields cannot change for a given sending port. A port mapping is 5667 removed by using KVM_XEN_EVTCHN_DEASSIGN in the flags field. Passing 5668 KVM_XEN_EVTCHN_RESET in the flags field removes all interception of 5669 outbound event channels. The values of the flags field are mutually 5670 exclusive and cannot be combined as a bitmask. 5671 5672KVM_XEN_ATTR_TYPE_XEN_VERSION 5673 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5674 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It configures 5675 the 32-bit version code returned to the guest when it invokes the 5676 XENVER_version call; typically (XEN_MAJOR << 16 | XEN_MINOR). PV 5677 Xen guests will often use this to as a dummy hypercall to trigger 5678 event channel delivery, so responding within the kernel without 5679 exiting to userspace is beneficial. 5680 5681KVM_XEN_ATTR_TYPE_RUNSTATE_UPDATE_FLAG 5682 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5683 support for KVM_XEN_HVM_CONFIG_RUNSTATE_UPDATE_FLAG. It enables the 5684 XEN_RUNSTATE_UPDATE flag which allows guest vCPUs to safely read 5685 other vCPUs' vcpu_runstate_info. Xen guests enable this feature via 5686 the VMASST_TYPE_runstate_update_flag of the HYPERVISOR_vm_assist 5687 hypercall. 5688 56894.127 KVM_XEN_HVM_GET_ATTR 5690-------------------------- 5691 5692:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 5693:Architectures: x86 5694:Type: vm ioctl 5695:Parameters: struct kvm_xen_hvm_attr 5696:Returns: 0 on success, < 0 on error 5697 5698Allows Xen VM attributes to be read. For the structure and types, 5699see KVM_XEN_HVM_SET_ATTR above. The KVM_XEN_ATTR_TYPE_EVTCHN 5700attribute cannot be read. 5701 57024.128 KVM_XEN_VCPU_SET_ATTR 5703--------------------------- 5704 5705:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 5706:Architectures: x86 5707:Type: vcpu ioctl 5708:Parameters: struct kvm_xen_vcpu_attr 5709:Returns: 0 on success, < 0 on error 5710 5711:: 5712 5713 struct kvm_xen_vcpu_attr { 5714 __u16 type; 5715 __u16 pad[3]; 5716 union { 5717 __u64 gpa; 5718 __u64 pad[4]; 5719 struct { 5720 __u64 state; 5721 __u64 state_entry_time; 5722 __u64 time_running; 5723 __u64 time_runnable; 5724 __u64 time_blocked; 5725 __u64 time_offline; 5726 } runstate; 5727 __u32 vcpu_id; 5728 struct { 5729 __u32 port; 5730 __u32 priority; 5731 __u64 expires_ns; 5732 } timer; 5733 __u8 vector; 5734 } u; 5735 }; 5736 5737type values: 5738 5739KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO 5740 Sets the guest physical address of the vcpu_info for a given vCPU. 5741 As with the shared_info page for the VM, the corresponding page may be 5742 dirtied at any time if event channel interrupt delivery is enabled, so 5743 userspace should always assume that the page is dirty without relying 5744 on dirty logging. Setting the gpa to KVM_XEN_INVALID_GPA will disable 5745 the vcpu_info. 5746 5747KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO_HVA 5748 If the KVM_XEN_HVM_CONFIG_SHARED_INFO_HVA flag is also set in the 5749 Xen capabilities, then this attribute may be used to set the 5750 userspace address of the vcpu_info for a given vCPU. It should 5751 only be used when the vcpu_info resides at the "default" location 5752 in the shared_info page. In this case it is safe to assume the 5753 userspace address will not change, because the shared_info page is 5754 an overlay on guest memory and remains at a fixed host address 5755 regardless of where it is mapped in guest physical address space 5756 and hence unnecessary invalidation of an internal cache may be 5757 avoided if the guest memory layout is modified. 5758 If the vcpu_info does not reside at the "default" location then 5759 it is not guaranteed to remain at the same host address and 5760 hence the aforementioned cache invalidation is required. 5761 5762KVM_XEN_VCPU_ATTR_TYPE_VCPU_TIME_INFO 5763 Sets the guest physical address of an additional pvclock structure 5764 for a given vCPU. This is typically used for guest vsyscall support. 5765 Setting the gpa to KVM_XEN_INVALID_GPA will disable the structure. 5766 5767KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR 5768 Sets the guest physical address of the vcpu_runstate_info for a given 5769 vCPU. This is how a Xen guest tracks CPU state such as steal time. 5770 Setting the gpa to KVM_XEN_INVALID_GPA will disable the runstate area. 5771 5772KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_CURRENT 5773 Sets the runstate (RUNSTATE_running/_runnable/_blocked/_offline) of 5774 the given vCPU from the .u.runstate.state member of the structure. 5775 KVM automatically accounts running and runnable time but blocked 5776 and offline states are only entered explicitly. 5777 5778KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_DATA 5779 Sets all fields of the vCPU runstate data from the .u.runstate member 5780 of the structure, including the current runstate. The state_entry_time 5781 must equal the sum of the other four times. 5782 5783KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST 5784 This *adds* the contents of the .u.runstate members of the structure 5785 to the corresponding members of the given vCPU's runstate data, thus 5786 permitting atomic adjustments to the runstate times. The adjustment 5787 to the state_entry_time must equal the sum of the adjustments to the 5788 other four times. The state field must be set to -1, or to a valid 5789 runstate value (RUNSTATE_running, RUNSTATE_runnable, RUNSTATE_blocked 5790 or RUNSTATE_offline) to set the current accounted state as of the 5791 adjusted state_entry_time. 5792 5793KVM_XEN_VCPU_ATTR_TYPE_VCPU_ID 5794 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5795 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It sets the Xen 5796 vCPU ID of the given vCPU, to allow timer-related VCPU operations to 5797 be intercepted by KVM. 5798 5799KVM_XEN_VCPU_ATTR_TYPE_TIMER 5800 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5801 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It sets the 5802 event channel port/priority for the VIRQ_TIMER of the vCPU, as well 5803 as allowing a pending timer to be saved/restored. Setting the timer 5804 port to zero disables kernel handling of the singleshot timer. 5805 5806KVM_XEN_VCPU_ATTR_TYPE_UPCALL_VECTOR 5807 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5808 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It sets the 5809 per-vCPU local APIC upcall vector, configured by a Xen guest with 5810 the HVMOP_set_evtchn_upcall_vector hypercall. This is typically 5811 used by Windows guests, and is distinct from the HVM-wide upcall 5812 vector configured with HVM_PARAM_CALLBACK_IRQ. It is disabled by 5813 setting the vector to zero. 5814 5815 58164.129 KVM_XEN_VCPU_GET_ATTR 5817--------------------------- 5818 5819:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 5820:Architectures: x86 5821:Type: vcpu ioctl 5822:Parameters: struct kvm_xen_vcpu_attr 5823:Returns: 0 on success, < 0 on error 5824 5825Allows Xen vCPU attributes to be read. For the structure and types, 5826see KVM_XEN_VCPU_SET_ATTR above. 5827 5828The KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST type may not be used 5829with the KVM_XEN_VCPU_GET_ATTR ioctl. 5830 58314.130 KVM_ARM_MTE_COPY_TAGS 5832--------------------------- 5833 5834:Capability: KVM_CAP_ARM_MTE 5835:Architectures: arm64 5836:Type: vm ioctl 5837:Parameters: struct kvm_arm_copy_mte_tags 5838:Returns: number of bytes copied, < 0 on error (-EINVAL for incorrect 5839 arguments, -EFAULT if memory cannot be accessed). 5840 5841:: 5842 5843 struct kvm_arm_copy_mte_tags { 5844 __u64 guest_ipa; 5845 __u64 length; 5846 void __user *addr; 5847 __u64 flags; 5848 __u64 reserved[2]; 5849 }; 5850 5851Copies Memory Tagging Extension (MTE) tags to/from guest tag memory. The 5852``guest_ipa`` and ``length`` fields must be ``PAGE_SIZE`` aligned. 5853``length`` must not be bigger than 2^31 - PAGE_SIZE bytes. The ``addr`` 5854field must point to a buffer which the tags will be copied to or from. 5855 5856``flags`` specifies the direction of copy, either ``KVM_ARM_TAGS_TO_GUEST`` or 5857``KVM_ARM_TAGS_FROM_GUEST``. 5858 5859The size of the buffer to store the tags is ``(length / 16)`` bytes 5860(granules in MTE are 16 bytes long). Each byte contains a single tag 5861value. This matches the format of ``PTRACE_PEEKMTETAGS`` and 5862``PTRACE_POKEMTETAGS``. 5863 5864If an error occurs before any data is copied then a negative error code is 5865returned. If some tags have been copied before an error occurs then the number 5866of bytes successfully copied is returned. If the call completes successfully 5867then ``length`` is returned. 5868 58694.131 KVM_GET_SREGS2 5870-------------------- 5871 5872:Capability: KVM_CAP_SREGS2 5873:Architectures: x86 5874:Type: vcpu ioctl 5875:Parameters: struct kvm_sregs2 (out) 5876:Returns: 0 on success, -1 on error 5877 5878Reads special registers from the vcpu. 5879This ioctl (when supported) replaces the KVM_GET_SREGS. 5880 5881:: 5882 5883 struct kvm_sregs2 { 5884 /* out (KVM_GET_SREGS2) / in (KVM_SET_SREGS2) */ 5885 struct kvm_segment cs, ds, es, fs, gs, ss; 5886 struct kvm_segment tr, ldt; 5887 struct kvm_dtable gdt, idt; 5888 __u64 cr0, cr2, cr3, cr4, cr8; 5889 __u64 efer; 5890 __u64 apic_base; 5891 __u64 flags; 5892 __u64 pdptrs[4]; 5893 }; 5894 5895flags values for ``kvm_sregs2``: 5896 5897``KVM_SREGS2_FLAGS_PDPTRS_VALID`` 5898 5899 Indicates that the struct contains valid PDPTR values. 5900 5901 59024.132 KVM_SET_SREGS2 5903-------------------- 5904 5905:Capability: KVM_CAP_SREGS2 5906:Architectures: x86 5907:Type: vcpu ioctl 5908:Parameters: struct kvm_sregs2 (in) 5909:Returns: 0 on success, -1 on error 5910 5911Writes special registers into the vcpu. 5912See KVM_GET_SREGS2 for the data structures. 5913This ioctl (when supported) replaces the KVM_SET_SREGS. 5914 59154.133 KVM_GET_STATS_FD 5916---------------------- 5917 5918:Capability: KVM_CAP_STATS_BINARY_FD 5919:Architectures: all 5920:Type: vm ioctl, vcpu ioctl 5921:Parameters: none 5922:Returns: statistics file descriptor on success, < 0 on error 5923 5924Errors: 5925 5926 ====== ====================================================== 5927 ENOMEM if the fd could not be created due to lack of memory 5928 EMFILE if the number of opened files exceeds the limit 5929 ====== ====================================================== 5930 5931The returned file descriptor can be used to read VM/vCPU statistics data in 5932binary format. The data in the file descriptor consists of four blocks 5933organized as follows: 5934 5935+-------------+ 5936| Header | 5937+-------------+ 5938| id string | 5939+-------------+ 5940| Descriptors | 5941+-------------+ 5942| Stats Data | 5943+-------------+ 5944 5945Apart from the header starting at offset 0, please be aware that it is 5946not guaranteed that the four blocks are adjacent or in the above order; 5947the offsets of the id, descriptors and data blocks are found in the 5948header. However, all four blocks are aligned to 64 bit offsets in the 5949file and they do not overlap. 5950 5951All blocks except the data block are immutable. Userspace can read them 5952only one time after retrieving the file descriptor, and then use ``pread`` or 5953``lseek`` to read the statistics repeatedly. 5954 5955All data is in system endianness. 5956 5957The format of the header is as follows:: 5958 5959 struct kvm_stats_header { 5960 __u32 flags; 5961 __u32 name_size; 5962 __u32 num_desc; 5963 __u32 id_offset; 5964 __u32 desc_offset; 5965 __u32 data_offset; 5966 }; 5967 5968The ``flags`` field is not used at the moment. It is always read as 0. 5969 5970The ``name_size`` field is the size (in byte) of the statistics name string 5971(including trailing '\0') which is contained in the "id string" block and 5972appended at the end of every descriptor. 5973 5974The ``num_desc`` field is the number of descriptors that are included in the 5975descriptor block. (The actual number of values in the data block may be 5976larger, since each descriptor may comprise more than one value). 5977 5978The ``id_offset`` field is the offset of the id string from the start of the 5979file indicated by the file descriptor. It is a multiple of 8. 5980 5981The ``desc_offset`` field is the offset of the Descriptors block from the start 5982of the file indicated by the file descriptor. It is a multiple of 8. 5983 5984The ``data_offset`` field is the offset of the Stats Data block from the start 5985of the file indicated by the file descriptor. It is a multiple of 8. 5986 5987The id string block contains a string which identifies the file descriptor on 5988which KVM_GET_STATS_FD was invoked. The size of the block, including the 5989trailing ``'\0'``, is indicated by the ``name_size`` field in the header. 5990 5991The descriptors block is only needed to be read once for the lifetime of the 5992file descriptor contains a sequence of ``struct kvm_stats_desc``, each followed 5993by a string of size ``name_size``. 5994:: 5995 5996 #define KVM_STATS_TYPE_SHIFT 0 5997 #define KVM_STATS_TYPE_MASK (0xF << KVM_STATS_TYPE_SHIFT) 5998 #define KVM_STATS_TYPE_CUMULATIVE (0x0 << KVM_STATS_TYPE_SHIFT) 5999 #define KVM_STATS_TYPE_INSTANT (0x1 << KVM_STATS_TYPE_SHIFT) 6000 #define KVM_STATS_TYPE_PEAK (0x2 << KVM_STATS_TYPE_SHIFT) 6001 #define KVM_STATS_TYPE_LINEAR_HIST (0x3 << KVM_STATS_TYPE_SHIFT) 6002 #define KVM_STATS_TYPE_LOG_HIST (0x4 << KVM_STATS_TYPE_SHIFT) 6003 #define KVM_STATS_TYPE_MAX KVM_STATS_TYPE_LOG_HIST 6004 6005 #define KVM_STATS_UNIT_SHIFT 4 6006 #define KVM_STATS_UNIT_MASK (0xF << KVM_STATS_UNIT_SHIFT) 6007 #define KVM_STATS_UNIT_NONE (0x0 << KVM_STATS_UNIT_SHIFT) 6008 #define KVM_STATS_UNIT_BYTES (0x1 << KVM_STATS_UNIT_SHIFT) 6009 #define KVM_STATS_UNIT_SECONDS (0x2 << KVM_STATS_UNIT_SHIFT) 6010 #define KVM_STATS_UNIT_CYCLES (0x3 << KVM_STATS_UNIT_SHIFT) 6011 #define KVM_STATS_UNIT_BOOLEAN (0x4 << KVM_STATS_UNIT_SHIFT) 6012 #define KVM_STATS_UNIT_MAX KVM_STATS_UNIT_BOOLEAN 6013 6014 #define KVM_STATS_BASE_SHIFT 8 6015 #define KVM_STATS_BASE_MASK (0xF << KVM_STATS_BASE_SHIFT) 6016 #define KVM_STATS_BASE_POW10 (0x0 << KVM_STATS_BASE_SHIFT) 6017 #define KVM_STATS_BASE_POW2 (0x1 << KVM_STATS_BASE_SHIFT) 6018 #define KVM_STATS_BASE_MAX KVM_STATS_BASE_POW2 6019 6020 struct kvm_stats_desc { 6021 __u32 flags; 6022 __s16 exponent; 6023 __u16 size; 6024 __u32 offset; 6025 __u32 bucket_size; 6026 char name[]; 6027 }; 6028 6029The ``flags`` field contains the type and unit of the statistics data described 6030by this descriptor. Its endianness is CPU native. 6031The following flags are supported: 6032 6033Bits 0-3 of ``flags`` encode the type: 6034 6035 * ``KVM_STATS_TYPE_CUMULATIVE`` 6036 The statistics reports a cumulative count. The value of data can only be increased. 6037 Most of the counters used in KVM are of this type. 6038 The corresponding ``size`` field for this type is always 1. 6039 All cumulative statistics data are read/write. 6040 * ``KVM_STATS_TYPE_INSTANT`` 6041 The statistics reports an instantaneous value. Its value can be increased or 6042 decreased. This type is usually used as a measurement of some resources, 6043 like the number of dirty pages, the number of large pages, etc. 6044 All instant statistics are read only. 6045 The corresponding ``size`` field for this type is always 1. 6046 * ``KVM_STATS_TYPE_PEAK`` 6047 The statistics data reports a peak value, for example the maximum number 6048 of items in a hash table bucket, the longest time waited and so on. 6049 The value of data can only be increased. 6050 The corresponding ``size`` field for this type is always 1. 6051 * ``KVM_STATS_TYPE_LINEAR_HIST`` 6052 The statistic is reported as a linear histogram. The number of 6053 buckets is specified by the ``size`` field. The size of buckets is specified 6054 by the ``hist_param`` field. The range of the Nth bucket (1 <= N < ``size``) 6055 is [``hist_param``*(N-1), ``hist_param``*N), while the range of the last 6056 bucket is [``hist_param``*(``size``-1), +INF). (+INF means positive infinity 6057 value.) 6058 * ``KVM_STATS_TYPE_LOG_HIST`` 6059 The statistic is reported as a logarithmic histogram. The number of 6060 buckets is specified by the ``size`` field. The range of the first bucket is 6061 [0, 1), while the range of the last bucket is [pow(2, ``size``-2), +INF). 6062 Otherwise, The Nth bucket (1 < N < ``size``) covers 6063 [pow(2, N-2), pow(2, N-1)). 6064 6065Bits 4-7 of ``flags`` encode the unit: 6066 6067 * ``KVM_STATS_UNIT_NONE`` 6068 There is no unit for the value of statistics data. This usually means that 6069 the value is a simple counter of an event. 6070 * ``KVM_STATS_UNIT_BYTES`` 6071 It indicates that the statistics data is used to measure memory size, in the 6072 unit of Byte, KiByte, MiByte, GiByte, etc. The unit of the data is 6073 determined by the ``exponent`` field in the descriptor. 6074 * ``KVM_STATS_UNIT_SECONDS`` 6075 It indicates that the statistics data is used to measure time or latency. 6076 * ``KVM_STATS_UNIT_CYCLES`` 6077 It indicates that the statistics data is used to measure CPU clock cycles. 6078 * ``KVM_STATS_UNIT_BOOLEAN`` 6079 It indicates that the statistic will always be either 0 or 1. Boolean 6080 statistics of "peak" type will never go back from 1 to 0. Boolean 6081 statistics can be linear histograms (with two buckets) but not logarithmic 6082 histograms. 6083 6084Note that, in the case of histograms, the unit applies to the bucket 6085ranges, while the bucket value indicates how many samples fell in the 6086bucket's range. 6087 6088Bits 8-11 of ``flags``, together with ``exponent``, encode the scale of the 6089unit: 6090 6091 * ``KVM_STATS_BASE_POW10`` 6092 The scale is based on power of 10. It is used for measurement of time and 6093 CPU clock cycles. For example, an exponent of -9 can be used with 6094 ``KVM_STATS_UNIT_SECONDS`` to express that the unit is nanoseconds. 6095 * ``KVM_STATS_BASE_POW2`` 6096 The scale is based on power of 2. It is used for measurement of memory size. 6097 For example, an exponent of 20 can be used with ``KVM_STATS_UNIT_BYTES`` to 6098 express that the unit is MiB. 6099 6100The ``size`` field is the number of values of this statistics data. Its 6101value is usually 1 for most of simple statistics. 1 means it contains an 6102unsigned 64bit data. 6103 6104The ``offset`` field is the offset from the start of Data Block to the start of 6105the corresponding statistics data. 6106 6107The ``bucket_size`` field is used as a parameter for histogram statistics data. 6108It is only used by linear histogram statistics data, specifying the size of a 6109bucket in the unit expressed by bits 4-11 of ``flags`` together with ``exponent``. 6110 6111The ``name`` field is the name string of the statistics data. The name string 6112starts at the end of ``struct kvm_stats_desc``. The maximum length including 6113the trailing ``'\0'``, is indicated by ``name_size`` in the header. 6114 6115The Stats Data block contains an array of 64-bit values in the same order 6116as the descriptors in Descriptors block. 6117 61184.134 KVM_GET_XSAVE2 6119-------------------- 6120 6121:Capability: KVM_CAP_XSAVE2 6122:Architectures: x86 6123:Type: vcpu ioctl 6124:Parameters: struct kvm_xsave (out) 6125:Returns: 0 on success, -1 on error 6126 6127 6128:: 6129 6130 struct kvm_xsave { 6131 __u32 region[1024]; 6132 __u32 extra[0]; 6133 }; 6134 6135This ioctl would copy current vcpu's xsave struct to the userspace. It 6136copies as many bytes as are returned by KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) 6137when invoked on the vm file descriptor. The size value returned by 6138KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) will always be at least 4096. 6139Currently, it is only greater than 4096 if a dynamic feature has been 6140enabled with ``arch_prctl()``, but this may change in the future. 6141 6142The offsets of the state save areas in struct kvm_xsave follow the contents 6143of CPUID leaf 0xD on the host. 6144 61454.135 KVM_XEN_HVM_EVTCHN_SEND 6146----------------------------- 6147 6148:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_EVTCHN_SEND 6149:Architectures: x86 6150:Type: vm ioctl 6151:Parameters: struct kvm_irq_routing_xen_evtchn 6152:Returns: 0 on success, < 0 on error 6153 6154 6155:: 6156 6157 struct kvm_irq_routing_xen_evtchn { 6158 __u32 port; 6159 __u32 vcpu; 6160 __u32 priority; 6161 }; 6162 6163This ioctl injects an event channel interrupt directly to the guest vCPU. 6164 61654.136 KVM_S390_PV_CPU_COMMAND 6166----------------------------- 6167 6168:Capability: KVM_CAP_S390_PROTECTED_DUMP 6169:Architectures: s390 6170:Type: vcpu ioctl 6171:Parameters: none 6172:Returns: 0 on success, < 0 on error 6173 6174This ioctl closely mirrors `KVM_S390_PV_COMMAND` but handles requests 6175for vcpus. It re-uses the kvm_s390_pv_dmp struct and hence also shares 6176the command ids. 6177 6178**command:** 6179 6180KVM_PV_DUMP 6181 Presents an API that provides calls which facilitate dumping a vcpu 6182 of a protected VM. 6183 6184**subcommand:** 6185 6186KVM_PV_DUMP_CPU 6187 Provides encrypted dump data like register values. 6188 The length of the returned data is provided by uv_info.guest_cpu_stor_len. 6189 61904.137 KVM_S390_ZPCI_OP 6191---------------------- 6192 6193:Capability: KVM_CAP_S390_ZPCI_OP 6194:Architectures: s390 6195:Type: vm ioctl 6196:Parameters: struct kvm_s390_zpci_op (in) 6197:Returns: 0 on success, <0 on error 6198 6199Used to manage hardware-assisted virtualization features for zPCI devices. 6200 6201Parameters are specified via the following structure:: 6202 6203 struct kvm_s390_zpci_op { 6204 /* in */ 6205 __u32 fh; /* target device */ 6206 __u8 op; /* operation to perform */ 6207 __u8 pad[3]; 6208 union { 6209 /* for KVM_S390_ZPCIOP_REG_AEN */ 6210 struct { 6211 __u64 ibv; /* Guest addr of interrupt bit vector */ 6212 __u64 sb; /* Guest addr of summary bit */ 6213 __u32 flags; 6214 __u32 noi; /* Number of interrupts */ 6215 __u8 isc; /* Guest interrupt subclass */ 6216 __u8 sbo; /* Offset of guest summary bit vector */ 6217 __u16 pad; 6218 } reg_aen; 6219 __u64 reserved[8]; 6220 } u; 6221 }; 6222 6223The type of operation is specified in the "op" field. 6224KVM_S390_ZPCIOP_REG_AEN is used to register the VM for adapter event 6225notification interpretation, which will allow firmware delivery of adapter 6226events directly to the vm, with KVM providing a backup delivery mechanism; 6227KVM_S390_ZPCIOP_DEREG_AEN is used to subsequently disable interpretation of 6228adapter event notifications. 6229 6230The target zPCI function must also be specified via the "fh" field. For the 6231KVM_S390_ZPCIOP_REG_AEN operation, additional information to establish firmware 6232delivery must be provided via the "reg_aen" struct. 6233 6234The "pad" and "reserved" fields may be used for future extensions and should be 6235set to 0s by userspace. 6236 62374.138 KVM_ARM_SET_COUNTER_OFFSET 6238-------------------------------- 6239 6240:Capability: KVM_CAP_COUNTER_OFFSET 6241:Architectures: arm64 6242:Type: vm ioctl 6243:Parameters: struct kvm_arm_counter_offset (in) 6244:Returns: 0 on success, < 0 on error 6245 6246This capability indicates that userspace is able to apply a single VM-wide 6247offset to both the virtual and physical counters as viewed by the guest 6248using the KVM_ARM_SET_CNT_OFFSET ioctl and the following data structure: 6249 6250:: 6251 6252 struct kvm_arm_counter_offset { 6253 __u64 counter_offset; 6254 __u64 reserved; 6255 }; 6256 6257The offset describes a number of counter cycles that are subtracted from 6258both virtual and physical counter views (similar to the effects of the 6259CNTVOFF_EL2 and CNTPOFF_EL2 system registers, but only global). The offset 6260always applies to all vcpus (already created or created after this ioctl) 6261for this VM. 6262 6263It is userspace's responsibility to compute the offset based, for example, 6264on previous values of the guest counters. 6265 6266Any value other than 0 for the "reserved" field may result in an error 6267(-EINVAL) being returned. This ioctl can also return -EBUSY if any vcpu 6268ioctl is issued concurrently. 6269 6270Note that using this ioctl results in KVM ignoring subsequent userspace 6271writes to the CNTVCT_EL0 and CNTPCT_EL0 registers using the SET_ONE_REG 6272interface. No error will be returned, but the resulting offset will not be 6273applied. 6274 6275.. _KVM_ARM_GET_REG_WRITABLE_MASKS: 6276 62774.139 KVM_ARM_GET_REG_WRITABLE_MASKS 6278------------------------------------ 6279 6280:Capability: KVM_CAP_ARM_SUPPORTED_REG_MASK_RANGES 6281:Architectures: arm64 6282:Type: vm ioctl 6283:Parameters: struct reg_mask_range (in/out) 6284:Returns: 0 on success, < 0 on error 6285 6286 6287:: 6288 6289 #define KVM_ARM_FEATURE_ID_RANGE 0 6290 #define KVM_ARM_FEATURE_ID_RANGE_SIZE (3 * 8 * 8) 6291 6292 struct reg_mask_range { 6293 __u64 addr; /* Pointer to mask array */ 6294 __u32 range; /* Requested range */ 6295 __u32 reserved[13]; 6296 }; 6297 6298This ioctl copies the writable masks for a selected range of registers to 6299userspace. 6300 6301The ``addr`` field is a pointer to the destination array where KVM copies 6302the writable masks. 6303 6304The ``range`` field indicates the requested range of registers. 6305``KVM_CHECK_EXTENSION`` for the ``KVM_CAP_ARM_SUPPORTED_REG_MASK_RANGES`` 6306capability returns the supported ranges, expressed as a set of flags. Each 6307flag's bit index represents a possible value for the ``range`` field. 6308All other values are reserved for future use and KVM may return an error. 6309 6310The ``reserved[13]`` array is reserved for future use and should be 0, or 6311KVM may return an error. 6312 6313KVM_ARM_FEATURE_ID_RANGE (0) 6314^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6315 6316The Feature ID range is defined as the AArch64 System register space with 6317op0==3, op1=={0, 1, 3}, CRn==0, CRm=={0-7}, op2=={0-7}. 6318 6319The mask returned array pointed to by ``addr`` is indexed by the macro 6320``ARM64_FEATURE_ID_RANGE_IDX(op0, op1, crn, crm, op2)``, allowing userspace 6321to know what fields can be changed for the system register described by 6322``op0, op1, crn, crm, op2``. KVM rejects ID register values that describe a 6323superset of the features supported by the system. 6324 63254.140 KVM_SET_USER_MEMORY_REGION2 6326--------------------------------- 6327 6328:Capability: KVM_CAP_USER_MEMORY2 6329:Architectures: all 6330:Type: vm ioctl 6331:Parameters: struct kvm_userspace_memory_region2 (in) 6332:Returns: 0 on success, -1 on error 6333 6334KVM_SET_USER_MEMORY_REGION2 is an extension to KVM_SET_USER_MEMORY_REGION that 6335allows mapping guest_memfd memory into a guest. All fields shared with 6336KVM_SET_USER_MEMORY_REGION identically. Userspace can set KVM_MEM_GUEST_MEMFD 6337in flags to have KVM bind the memory region to a given guest_memfd range of 6338[guest_memfd_offset, guest_memfd_offset + memory_size]. The target guest_memfd 6339must point at a file created via KVM_CREATE_GUEST_MEMFD on the current VM, and 6340the target range must not be bound to any other memory region. All standard 6341bounds checks apply (use common sense). 6342 6343:: 6344 6345 struct kvm_userspace_memory_region2 { 6346 __u32 slot; 6347 __u32 flags; 6348 __u64 guest_phys_addr; 6349 __u64 memory_size; /* bytes */ 6350 __u64 userspace_addr; /* start of the userspace allocated memory */ 6351 __u64 guest_memfd_offset; 6352 __u32 guest_memfd; 6353 __u32 pad1; 6354 __u64 pad2[14]; 6355 }; 6356 6357A KVM_MEM_GUEST_MEMFD region _must_ have a valid guest_memfd (private memory) and 6358userspace_addr (shared memory). However, "valid" for userspace_addr simply 6359means that the address itself must be a legal userspace address. The backing 6360mapping for userspace_addr is not required to be valid/populated at the time of 6361KVM_SET_USER_MEMORY_REGION2, e.g. shared memory can be lazily mapped/allocated 6362on-demand. 6363 6364When mapping a gfn into the guest, KVM selects shared vs. private, i.e consumes 6365userspace_addr vs. guest_memfd, based on the gfn's KVM_MEMORY_ATTRIBUTE_PRIVATE 6366state. At VM creation time, all memory is shared, i.e. the PRIVATE attribute 6367is '0' for all gfns. Userspace can control whether memory is shared/private by 6368toggling KVM_MEMORY_ATTRIBUTE_PRIVATE via KVM_SET_MEMORY_ATTRIBUTES as needed. 6369 6370S390: 6371^^^^^ 6372 6373Returns -EINVAL if the VM has the KVM_VM_S390_UCONTROL flag set. 6374Returns -EINVAL if called on a protected VM. 6375 63764.141 KVM_SET_MEMORY_ATTRIBUTES 6377------------------------------- 6378 6379:Capability: KVM_CAP_MEMORY_ATTRIBUTES 6380:Architectures: x86 6381:Type: vm ioctl 6382:Parameters: struct kvm_memory_attributes (in) 6383:Returns: 0 on success, <0 on error 6384 6385KVM_SET_MEMORY_ATTRIBUTES allows userspace to set memory attributes for a range 6386of guest physical memory. 6387 6388:: 6389 6390 struct kvm_memory_attributes { 6391 __u64 address; 6392 __u64 size; 6393 __u64 attributes; 6394 __u64 flags; 6395 }; 6396 6397 #define KVM_MEMORY_ATTRIBUTE_PRIVATE (1ULL << 3) 6398 6399The address and size must be page aligned. The supported attributes can be 6400retrieved via ioctl(KVM_CHECK_EXTENSION) on KVM_CAP_MEMORY_ATTRIBUTES. If 6401executed on a VM, KVM_CAP_MEMORY_ATTRIBUTES precisely returns the attributes 6402supported by that VM. If executed at system scope, KVM_CAP_MEMORY_ATTRIBUTES 6403returns all attributes supported by KVM. The only attribute defined at this 6404time is KVM_MEMORY_ATTRIBUTE_PRIVATE, which marks the associated gfn as being 6405guest private memory. 6406 6407Note, there is no "get" API. Userspace is responsible for explicitly tracking 6408the state of a gfn/page as needed. 6409 6410The "flags" field is reserved for future extensions and must be '0'. 6411 64124.142 KVM_CREATE_GUEST_MEMFD 6413---------------------------- 6414 6415:Capability: KVM_CAP_GUEST_MEMFD 6416:Architectures: none 6417:Type: vm ioctl 6418:Parameters: struct kvm_create_guest_memfd(in) 6419:Returns: A file descriptor on success, <0 on error 6420 6421KVM_CREATE_GUEST_MEMFD creates an anonymous file and returns a file descriptor 6422that refers to it. guest_memfd files are roughly analogous to files created 6423via memfd_create(), e.g. guest_memfd files live in RAM, have volatile storage, 6424and are automatically released when the last reference is dropped. Unlike 6425"regular" memfd_create() files, guest_memfd files are bound to their owning 6426virtual machine (see below), cannot be mapped, read, or written by userspace, 6427and cannot be resized (guest_memfd files do however support PUNCH_HOLE). 6428 6429:: 6430 6431 struct kvm_create_guest_memfd { 6432 __u64 size; 6433 __u64 flags; 6434 __u64 reserved[6]; 6435 }; 6436 6437Conceptually, the inode backing a guest_memfd file represents physical memory, 6438i.e. is coupled to the virtual machine as a thing, not to a "struct kvm". The 6439file itself, which is bound to a "struct kvm", is that instance's view of the 6440underlying memory, e.g. effectively provides the translation of guest addresses 6441to host memory. This allows for use cases where multiple KVM structures are 6442used to manage a single virtual machine, e.g. when performing intrahost 6443migration of a virtual machine. 6444 6445KVM currently only supports mapping guest_memfd via KVM_SET_USER_MEMORY_REGION2, 6446and more specifically via the guest_memfd and guest_memfd_offset fields in 6447"struct kvm_userspace_memory_region2", where guest_memfd_offset is the offset 6448into the guest_memfd instance. For a given guest_memfd file, there can be at 6449most one mapping per page, i.e. binding multiple memory regions to a single 6450guest_memfd range is not allowed (any number of memory regions can be bound to 6451a single guest_memfd file, but the bound ranges must not overlap). 6452 6453The capability KVM_CAP_GUEST_MEMFD_FLAGS enumerates the `flags` that can be 6454specified via KVM_CREATE_GUEST_MEMFD. Currently defined flags: 6455 6456 ============================ ================================================ 6457 GUEST_MEMFD_FLAG_MMAP Enable using mmap() on the guest_memfd file 6458 descriptor. 6459 GUEST_MEMFD_FLAG_INIT_SHARED Make all memory in the file shared during 6460 KVM_CREATE_GUEST_MEMFD (memory files created 6461 without INIT_SHARED will be marked private). 6462 Shared memory can be faulted into host userspace 6463 page tables. Private memory cannot. 6464 ============================ ================================================ 6465 6466When the KVM MMU performs a PFN lookup to service a guest fault and the backing 6467guest_memfd has the GUEST_MEMFD_FLAG_MMAP set, then the fault will always be 6468consumed from guest_memfd, regardless of whether it is a shared or a private 6469fault. 6470 6471See KVM_SET_USER_MEMORY_REGION2 for additional details. 6472 64734.143 KVM_PRE_FAULT_MEMORY 6474--------------------------- 6475 6476:Capability: KVM_CAP_PRE_FAULT_MEMORY 6477:Architectures: none 6478:Type: vcpu ioctl 6479:Parameters: struct kvm_pre_fault_memory (in/out) 6480:Returns: 0 if at least one page is processed, < 0 on error 6481 6482Errors: 6483 6484 ========== =============================================================== 6485 EINVAL The specified `gpa` and `size` were invalid (e.g. not 6486 page aligned, causes an overflow, or size is zero), or the VM 6487 is UCONTROL (s390). 6488 ENOENT The specified `gpa` is outside defined memslots. 6489 EINTR An unmasked signal is pending and no page was processed. 6490 EFAULT The parameter address was invalid. 6491 EOPNOTSUPP Mapping memory for a GPA is unsupported by the 6492 hypervisor, and/or for the current vCPU state/mode. 6493 EIO unexpected error conditions (also causes a WARN) 6494 ========== =============================================================== 6495 6496:: 6497 6498 struct kvm_pre_fault_memory { 6499 /* in/out */ 6500 __u64 gpa; 6501 __u64 size; 6502 /* in */ 6503 __u64 flags; 6504 __u64 padding[5]; 6505 }; 6506 6507KVM_PRE_FAULT_MEMORY populates KVM's stage-2 page tables used to map memory 6508for the current vCPU state. KVM maps memory as if the vCPU generated a 6509stage-2 read page fault, e.g. faults in memory as needed, but doesn't break 6510CoW. On x86, KVM does not mark any newly created stage-2 PTE as Accessed. 6511 6512In the case of confidential VM types where there is an initial set up of 6513private guest memory before the guest is 'finalized'/measured, this ioctl 6514should only be issued after completing all the necessary setup to put the 6515guest into a 'finalized' state so that the above semantics can be reliably 6516ensured. 6517 6518In some cases, multiple vCPUs might share the page tables. In this 6519case, the ioctl can be called in parallel. 6520 6521When the ioctl returns, the input values are updated to point to the 6522remaining range. If `size` > 0 on return, the caller can just issue 6523the ioctl again with the same `struct kvm_map_memory` argument. 6524 6525Shadow page tables cannot support this ioctl because they 6526are indexed by virtual address or nested guest physical address. 6527Calling this ioctl when the guest is using shadow page tables (for 6528example because it is running a nested guest with nested page tables) 6529will fail with `EOPNOTSUPP` even if `KVM_CHECK_EXTENSION` reports 6530the capability to be present. 6531 6532`flags` must currently be zero. 6533 65344.144 KVM_S390_KEYOP 6535-------------------- 6536 6537:Capability: KVM_CAP_S390_KEYOP 6538:Architectures: s390 6539:Type: vm ioctl 6540:Parameters: struct kvm_s390_keyop (in/out) 6541:Returns: 0 in case of success, < 0 on error 6542 6543The specified key operation is performed on the given guest address. The 6544previous storage key (or the relevant part thereof) will be returned in 6545`key`. 6546 6547:: 6548 6549 struct kvm_s390_keyop { 6550 __u64 guest_addr; 6551 __u8 key; 6552 __u8 operation; 6553 }; 6554 6555Currently supported values for ``operation``: 6556 6557KVM_S390_KEYOP_ISKE 6558 Returns the storage key for the guest address ``guest_addr`` in ``key``. 6559 6560KVM_S390_KEYOP_RRBE 6561 Resets the reference bit for the guest address ``guest_addr``, returning the 6562 R and C bits of the old storage key in ``key``; the remaining fields of 6563 the storage key will be set to 0. 6564 6565KVM_S390_KEYOP_SSKE 6566 Sets the storage key for the guest address ``guest_addr`` to the key 6567 specified in ``key``, returning the previous value in ``key``. 6568 6569.. _kvm_run: 6570 65715. The kvm_run structure 6572======================== 6573 6574Application code obtains a pointer to the kvm_run structure by 6575mmap()ing a vcpu fd. From that point, application code can control 6576execution by changing fields in kvm_run prior to calling the KVM_RUN 6577ioctl, and obtain information about the reason KVM_RUN returned by 6578looking up structure members. 6579 6580:: 6581 6582 struct kvm_run { 6583 /* in */ 6584 __u8 request_interrupt_window; 6585 6586Request that KVM_RUN return when it becomes possible to inject external 6587interrupts into the guest. Useful in conjunction with KVM_INTERRUPT. 6588 6589:: 6590 6591 __u8 immediate_exit; 6592 6593This field is polled once when KVM_RUN starts; if non-zero, KVM_RUN 6594exits immediately, returning -EINTR. In the common scenario where a 6595signal is used to "kick" a VCPU out of KVM_RUN, this field can be used 6596to avoid usage of KVM_SET_SIGNAL_MASK, which has worse scalability. 6597Rather than blocking the signal outside KVM_RUN, userspace can set up 6598a signal handler that sets run->immediate_exit to a non-zero value. 6599 6600This field is ignored if KVM_CAP_IMMEDIATE_EXIT is not available. 6601 6602:: 6603 6604 __u8 padding1[6]; 6605 6606 /* out */ 6607 __u32 exit_reason; 6608 6609When KVM_RUN has returned successfully (return value 0), this informs 6610application code why KVM_RUN has returned. Allowable values for this 6611field are detailed below. 6612 6613:: 6614 6615 __u8 ready_for_interrupt_injection; 6616 6617If request_interrupt_window has been specified, this field indicates 6618an interrupt can be injected now with KVM_INTERRUPT. 6619 6620:: 6621 6622 __u8 if_flag; 6623 6624The value of the current interrupt flag. Only valid if in-kernel 6625local APIC is not used. 6626 6627:: 6628 6629 __u16 flags; 6630 6631More architecture-specific flags detailing state of the VCPU that may 6632affect the device's behavior. Current defined flags:: 6633 6634 /* x86, set if the VCPU is in system management mode */ 6635 #define KVM_RUN_X86_SMM (1 << 0) 6636 /* x86, set if bus lock detected in VM */ 6637 #define KVM_RUN_X86_BUS_LOCK (1 << 1) 6638 /* x86, set if the VCPU is executing a nested (L2) guest */ 6639 #define KVM_RUN_X86_GUEST_MODE (1 << 2) 6640 6641 /* arm64, set for KVM_EXIT_DEBUG */ 6642 #define KVM_DEBUG_ARCH_HSR_HIGH_VALID (1 << 0) 6643 6644:: 6645 6646 /* in (pre_kvm_run), out (post_kvm_run) */ 6647 __u64 cr8; 6648 6649The value of the cr8 register. Only valid if in-kernel local APIC is 6650not used. Both input and output. 6651 6652:: 6653 6654 __u64 apic_base; 6655 6656The value of the APIC BASE msr. Only valid if in-kernel local 6657APIC is not used. Both input and output. 6658 6659:: 6660 6661 union { 6662 /* KVM_EXIT_UNKNOWN */ 6663 struct { 6664 __u64 hardware_exit_reason; 6665 } hw; 6666 6667If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown 6668reasons. Further architecture-specific information is available in 6669hardware_exit_reason. 6670 6671:: 6672 6673 /* KVM_EXIT_FAIL_ENTRY */ 6674 struct { 6675 __u64 hardware_entry_failure_reason; 6676 __u32 cpu; /* if KVM_LAST_CPU */ 6677 } fail_entry; 6678 6679If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due 6680to unknown reasons. Further architecture-specific information is 6681available in hardware_entry_failure_reason. 6682 6683:: 6684 6685 /* KVM_EXIT_EXCEPTION */ 6686 struct { 6687 __u32 exception; 6688 __u32 error_code; 6689 } ex; 6690 6691Unused. 6692 6693:: 6694 6695 /* KVM_EXIT_IO */ 6696 struct { 6697 #define KVM_EXIT_IO_IN 0 6698 #define KVM_EXIT_IO_OUT 1 6699 __u8 direction; 6700 __u8 size; /* bytes */ 6701 __u16 port; 6702 __u32 count; 6703 __u64 data_offset; /* relative to kvm_run start */ 6704 } io; 6705 6706If exit_reason is KVM_EXIT_IO, then the vcpu has 6707executed a port I/O instruction which could not be satisfied by kvm. 6708data_offset describes where the data is located (KVM_EXIT_IO_OUT) or 6709where kvm expects application code to place the data for the next 6710KVM_RUN invocation (KVM_EXIT_IO_IN). Data format is a packed array. 6711 6712:: 6713 6714 /* KVM_EXIT_DEBUG */ 6715 struct { 6716 struct kvm_debug_exit_arch arch; 6717 } debug; 6718 6719If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event 6720for which architecture specific information is returned. 6721 6722:: 6723 6724 /* KVM_EXIT_MMIO */ 6725 struct { 6726 __u64 phys_addr; 6727 __u8 data[8]; 6728 __u32 len; 6729 __u8 is_write; 6730 } mmio; 6731 6732If exit_reason is KVM_EXIT_MMIO, then the vcpu has 6733executed a memory-mapped I/O instruction which could not be satisfied 6734by kvm. The 'data' member contains the written data if 'is_write' is 6735true, and should be filled by application code otherwise. 6736 6737The 'data' member contains, in its first 'len' bytes, the value as it would 6738appear if the VCPU performed a load or store of the appropriate width directly 6739to the byte array. 6740 6741.. note:: 6742 6743 For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR, KVM_EXIT_XEN, 6744 KVM_EXIT_EPR, KVM_EXIT_HYPERCALL, KVM_EXIT_TDX, 6745 KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR the corresponding 6746 operations are complete (and guest state is consistent) only after userspace 6747 has re-entered the kernel with KVM_RUN. The kernel side will first finish 6748 incomplete operations and then check for pending signals. 6749 6750 The pending state of the operation is not preserved in state which is 6751 visible to userspace, thus userspace should ensure that the operation is 6752 completed before performing a live migration. Userspace can re-enter the 6753 guest with an unmasked signal pending or with the immediate_exit field set 6754 to complete pending operations without allowing any further instructions 6755 to be executed. 6756 6757:: 6758 6759 /* KVM_EXIT_HYPERCALL */ 6760 struct { 6761 __u64 nr; 6762 __u64 args[6]; 6763 __u64 ret; 6764 __u64 flags; 6765 } hypercall; 6766 6767 6768It is strongly recommended that userspace use ``KVM_EXIT_IO`` (x86) or 6769``KVM_EXIT_MMIO`` (all except s390) to implement functionality that 6770requires a guest to interact with host userspace. 6771 6772.. note:: KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO. 6773 6774For arm64: 6775---------- 6776 6777SMCCC exits can be enabled depending on the configuration of the SMCCC 6778filter. See the Documentation/virt/kvm/devices/vm.rst 6779``KVM_ARM_SMCCC_FILTER`` for more details. 6780 6781``nr`` contains the function ID of the guest's SMCCC call. Userspace is 6782expected to use the ``KVM_GET_ONE_REG`` ioctl to retrieve the call 6783parameters from the vCPU's GPRs. 6784 6785Definition of ``flags``: 6786 - ``KVM_HYPERCALL_EXIT_SMC``: Indicates that the guest used the SMC 6787 conduit to initiate the SMCCC call. If this bit is 0 then the guest 6788 used the HVC conduit for the SMCCC call. 6789 6790 - ``KVM_HYPERCALL_EXIT_16BIT``: Indicates that the guest used a 16bit 6791 instruction to initiate the SMCCC call. If this bit is 0 then the 6792 guest used a 32bit instruction. An AArch64 guest always has this 6793 bit set to 0. 6794 6795At the point of exit, PC points to the instruction immediately following 6796the trapping instruction. 6797 6798:: 6799 6800 /* KVM_EXIT_TPR_ACCESS */ 6801 struct { 6802 __u64 rip; 6803 __u32 is_write; 6804 __u32 pad; 6805 } tpr_access; 6806 6807To be documented (KVM_TPR_ACCESS_REPORTING). 6808 6809:: 6810 6811 /* KVM_EXIT_S390_SIEIC */ 6812 struct { 6813 __u8 icptcode; 6814 __u64 mask; /* psw upper half */ 6815 __u64 addr; /* psw lower half */ 6816 __u16 ipa; 6817 __u32 ipb; 6818 } s390_sieic; 6819 6820s390 specific. 6821 6822:: 6823 6824 /* KVM_EXIT_S390_RESET */ 6825 #define KVM_S390_RESET_POR 1 6826 #define KVM_S390_RESET_CLEAR 2 6827 #define KVM_S390_RESET_SUBSYSTEM 4 6828 #define KVM_S390_RESET_CPU_INIT 8 6829 #define KVM_S390_RESET_IPL 16 6830 __u64 s390_reset_flags; 6831 6832s390 specific. 6833 6834:: 6835 6836 /* KVM_EXIT_S390_UCONTROL */ 6837 struct { 6838 __u64 trans_exc_code; 6839 __u32 pgm_code; 6840 } s390_ucontrol; 6841 6842s390 specific. A page fault has occurred for a user controlled virtual 6843machine (KVM_VM_S390_UNCONTROL) on its host page table that cannot be 6844resolved by the kernel. 6845The program code and the translation exception code that were placed 6846in the cpu's lowcore are presented here as defined by the z Architecture 6847Principles of Operation Book in the Chapter for Dynamic Address Translation 6848(DAT) 6849 6850:: 6851 6852 /* KVM_EXIT_DCR */ 6853 struct { 6854 __u32 dcrn; 6855 __u32 data; 6856 __u8 is_write; 6857 } dcr; 6858 6859Deprecated - was used for 440 KVM. 6860 6861:: 6862 6863 /* KVM_EXIT_OSI */ 6864 struct { 6865 __u64 gprs[32]; 6866 } osi; 6867 6868MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch 6869hypercalls and exit with this exit struct that contains all the guest gprs. 6870 6871If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall. 6872Userspace can now handle the hypercall and when it's done modify the gprs as 6873necessary. Upon guest entry all guest GPRs will then be replaced by the values 6874in this struct. 6875 6876:: 6877 6878 /* KVM_EXIT_PAPR_HCALL */ 6879 struct { 6880 __u64 nr; 6881 __u64 ret; 6882 __u64 args[9]; 6883 } papr_hcall; 6884 6885This is used on 64-bit PowerPC when emulating a pSeries partition, 6886e.g. with the 'pseries' machine type in qemu. It occurs when the 6887guest does a hypercall using the 'sc 1' instruction. The 'nr' field 6888contains the hypercall number (from the guest R3), and 'args' contains 6889the arguments (from the guest R4 - R12). Userspace should put the 6890return code in 'ret' and any extra returned values in args[]. 6891The possible hypercalls are defined in the Power Architecture Platform 6892Requirements (PAPR) document available from www.power.org (free 6893developer registration required to access it). 6894 6895:: 6896 6897 /* KVM_EXIT_S390_TSCH */ 6898 struct { 6899 __u16 subchannel_id; 6900 __u16 subchannel_nr; 6901 __u32 io_int_parm; 6902 __u32 io_int_word; 6903 __u32 ipb; 6904 __u8 dequeued; 6905 } s390_tsch; 6906 6907s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled 6908and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O 6909interrupt for the target subchannel has been dequeued and subchannel_id, 6910subchannel_nr, io_int_parm and io_int_word contain the parameters for that 6911interrupt. ipb is needed for instruction parameter decoding. 6912 6913:: 6914 6915 /* KVM_EXIT_EPR */ 6916 struct { 6917 __u32 epr; 6918 } epr; 6919 6920On FSL BookE PowerPC chips, the interrupt controller has a fast patch 6921interrupt acknowledge path to the core. When the core successfully 6922delivers an interrupt, it automatically populates the EPR register with 6923the interrupt vector number and acknowledges the interrupt inside 6924the interrupt controller. 6925 6926In case the interrupt controller lives in user space, we need to do 6927the interrupt acknowledge cycle through it to fetch the next to be 6928delivered interrupt vector using this exit. 6929 6930It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an 6931external interrupt has just been delivered into the guest. User space 6932should put the acknowledged interrupt vector into the 'epr' field. 6933 6934:: 6935 6936 /* KVM_EXIT_SYSTEM_EVENT */ 6937 struct { 6938 #define KVM_SYSTEM_EVENT_SHUTDOWN 1 6939 #define KVM_SYSTEM_EVENT_RESET 2 6940 #define KVM_SYSTEM_EVENT_CRASH 3 6941 #define KVM_SYSTEM_EVENT_WAKEUP 4 6942 #define KVM_SYSTEM_EVENT_SUSPEND 5 6943 #define KVM_SYSTEM_EVENT_SEV_TERM 6 6944 #define KVM_SYSTEM_EVENT_TDX_FATAL 7 6945 __u32 type; 6946 __u32 ndata; 6947 __u64 data[16]; 6948 } system_event; 6949 6950If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered 6951a system-level event using some architecture specific mechanism (hypercall 6952or some special instruction). In case of ARM64, this is triggered using 6953HVC instruction based PSCI call from the vcpu. 6954 6955The 'type' field describes the system-level event type. 6956Valid values for 'type' are: 6957 6958 - KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the 6959 VM. Userspace is not obliged to honour this, and if it does honour 6960 this does not need to destroy the VM synchronously (ie it may call 6961 KVM_RUN again before shutdown finally occurs). 6962 - KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM. 6963 As with SHUTDOWN, userspace can choose to ignore the request, or 6964 to schedule the reset to occur in the future and may call KVM_RUN again. 6965 - KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest 6966 has requested a crash condition maintenance. Userspace can choose 6967 to ignore the request, or to gather VM memory core dump and/or 6968 reset/shutdown of the VM. 6969 - KVM_SYSTEM_EVENT_SEV_TERM -- an AMD SEV guest requested termination. 6970 The guest physical address of the guest's GHCB is stored in `data[0]`. 6971 - KVM_SYSTEM_EVENT_TDX_FATAL -- a TDX guest reported a fatal error state. 6972 KVM doesn't do any parsing or conversion, it just dumps 16 general-purpose 6973 registers to userspace, in ascending order of the 4-bit indices for x86-64 6974 general-purpose registers in instruction encoding, as defined in the Intel 6975 SDM. 6976 - KVM_SYSTEM_EVENT_WAKEUP -- the exiting vCPU is in a suspended state and 6977 KVM has recognized a wakeup event. Userspace may honor this event by 6978 marking the exiting vCPU as runnable, or deny it and call KVM_RUN again. 6979 - KVM_SYSTEM_EVENT_SUSPEND -- the guest has requested a suspension of 6980 the VM. 6981 6982If KVM_CAP_SYSTEM_EVENT_DATA is present, the 'data' field can contain 6983architecture specific information for the system-level event. Only 6984the first `ndata` items (possibly zero) of the data array are valid. 6985 6986 - for arm64, data[0] is set to KVM_SYSTEM_EVENT_RESET_FLAG_PSCI_RESET2 if 6987 the guest issued a SYSTEM_RESET2 call according to v1.1 of the PSCI 6988 specification. 6989 6990 - for arm64, data[0] is set to KVM_SYSTEM_EVENT_SHUTDOWN_FLAG_PSCI_OFF2 6991 if the guest issued a SYSTEM_OFF2 call according to v1.3 of the PSCI 6992 specification. 6993 6994 - for RISC-V, data[0] is set to the value of the second argument of the 6995 ``sbi_system_reset`` call. 6996 6997Previous versions of Linux defined a `flags` member in this struct. The 6998field is now aliased to `data[0]`. Userspace can assume that it is only 6999written if ndata is greater than 0. 7000 7001For arm/arm64: 7002-------------- 7003 7004KVM_SYSTEM_EVENT_SUSPEND exits are enabled with the 7005KVM_CAP_ARM_SYSTEM_SUSPEND VM capability. If a guest invokes the PSCI 7006SYSTEM_SUSPEND function, KVM will exit to userspace with this event 7007type. 7008 7009It is the sole responsibility of userspace to implement the PSCI 7010SYSTEM_SUSPEND call according to ARM DEN0022D.b 5.19 "SYSTEM_SUSPEND". 7011KVM does not change the vCPU's state before exiting to userspace, so 7012the call parameters are left in-place in the vCPU registers. 7013 7014Userspace is _required_ to take action for such an exit. It must 7015either: 7016 7017 - Honor the guest request to suspend the VM. Userspace can request 7018 in-kernel emulation of suspension by setting the calling vCPU's 7019 state to KVM_MP_STATE_SUSPENDED. Userspace must configure the vCPU's 7020 state according to the parameters passed to the PSCI function when 7021 the calling vCPU is resumed. See ARM DEN0022D.b 5.19.1 "Intended use" 7022 for details on the function parameters. 7023 7024 - Deny the guest request to suspend the VM. See ARM DEN0022D.b 5.19.2 7025 "Caller responsibilities" for possible return values. 7026 7027Hibernation using the PSCI SYSTEM_OFF2 call is enabled when PSCI v1.3 7028is enabled. If a guest invokes the PSCI SYSTEM_OFF2 function, KVM will 7029exit to userspace with the KVM_SYSTEM_EVENT_SHUTDOWN event type and with 7030data[0] set to KVM_SYSTEM_EVENT_SHUTDOWN_FLAG_PSCI_OFF2. The only 7031supported hibernate type for the SYSTEM_OFF2 function is HIBERNATE_OFF. 7032 7033:: 7034 7035 /* KVM_EXIT_IOAPIC_EOI */ 7036 struct { 7037 __u8 vector; 7038 } eoi; 7039 7040Indicates that the VCPU's in-kernel local APIC received an EOI for a 7041level-triggered IOAPIC interrupt. This exit only triggers when the 7042IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled); 7043the userspace IOAPIC should process the EOI and retrigger the interrupt if 7044it is still asserted. Vector is the LAPIC interrupt vector for which the 7045EOI was received. 7046 7047:: 7048 7049 struct kvm_hyperv_exit { 7050 #define KVM_EXIT_HYPERV_SYNIC 1 7051 #define KVM_EXIT_HYPERV_HCALL 2 7052 #define KVM_EXIT_HYPERV_SYNDBG 3 7053 __u32 type; 7054 __u32 pad1; 7055 union { 7056 struct { 7057 __u32 msr; 7058 __u32 pad2; 7059 __u64 control; 7060 __u64 evt_page; 7061 __u64 msg_page; 7062 } synic; 7063 struct { 7064 __u64 input; 7065 __u64 result; 7066 __u64 params[2]; 7067 } hcall; 7068 struct { 7069 __u32 msr; 7070 __u32 pad2; 7071 __u64 control; 7072 __u64 status; 7073 __u64 send_page; 7074 __u64 recv_page; 7075 __u64 pending_page; 7076 } syndbg; 7077 } u; 7078 }; 7079 /* KVM_EXIT_HYPERV */ 7080 struct kvm_hyperv_exit hyperv; 7081 7082Indicates that the VCPU exits into userspace to process some tasks 7083related to Hyper-V emulation. 7084 7085Valid values for 'type' are: 7086 7087 - KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about 7088 7089Hyper-V SynIC state change. Notification is used to remap SynIC 7090event/message pages and to enable/disable SynIC messages/events processing 7091in userspace. 7092 7093 - KVM_EXIT_HYPERV_SYNDBG -- synchronously notify user-space about 7094 7095Hyper-V Synthetic debugger state change. Notification is used to either update 7096the pending_page location or to send a control command (send the buffer located 7097in send_page or recv a buffer to recv_page). 7098 7099:: 7100 7101 /* KVM_EXIT_ARM_NISV / KVM_EXIT_ARM_LDST64B */ 7102 struct { 7103 __u64 esr_iss; 7104 __u64 fault_ipa; 7105 } arm_nisv; 7106 7107- KVM_EXIT_ARM_NISV: 7108 7109Used on arm64 systems. If a guest accesses memory not in a memslot, 7110KVM will typically return to userspace and ask it to do MMIO emulation on its 7111behalf. However, for certain classes of instructions, no instruction decode 7112(direction, length of memory access) is provided, and fetching and decoding 7113the instruction from the VM is overly complicated to live in the kernel. 7114 7115Historically, when this situation occurred, KVM would print a warning and kill 7116the VM. KVM assumed that if the guest accessed non-memslot memory, it was 7117trying to do I/O, which just couldn't be emulated, and the warning message was 7118phrased accordingly. However, what happened more often was that a guest bug 7119caused access outside the guest memory areas which should lead to a more 7120meaningful warning message and an external abort in the guest, if the access 7121did not fall within an I/O window. 7122 7123Userspace implementations can query for KVM_CAP_ARM_NISV_TO_USER, and enable 7124this capability at VM creation. Once this is done, these types of errors will 7125instead return to userspace with KVM_EXIT_ARM_NISV, with the valid bits from 7126the ESR_EL2 in the esr_iss field, and the faulting IPA in the fault_ipa field. 7127Userspace can either fix up the access if it's actually an I/O access by 7128decoding the instruction from guest memory (if it's very brave) and continue 7129executing the guest, or it can decide to suspend, dump, or restart the guest. 7130 7131Note that KVM does not skip the faulting instruction as it does for 7132KVM_EXIT_MMIO, but userspace has to emulate any change to the processing state 7133if it decides to decode and emulate the instruction. 7134 7135This feature isn't available to protected VMs, as userspace does not 7136have access to the state that is required to perform the emulation. 7137Instead, a data abort exception is directly injected in the guest. 7138Note that although KVM_CAP_ARM_NISV_TO_USER will be reported if 7139queried outside of a protected VM context, the feature will not be 7140exposed if queried on a protected VM file descriptor. 7141 7142- KVM_EXIT_ARM_LDST64B: 7143 7144Used on arm64 systems. When a guest using a LD64B, ST64B, ST64BV, ST64BV0, 7145outside of a memslot, KVM will return to userspace with KVM_EXIT_ARM_LDST64B, 7146exposing the relevant ESR_EL2 information and faulting IPA, similarly to 7147KVM_EXIT_ARM_NISV. 7148 7149Userspace is supposed to fully emulate the instructions, which includes: 7150 7151 - fetch of the operands for a store, including ACCDATA_EL1 in the case 7152 of a ST64BV0 instruction 7153 - deal with the endianness if the guest is big-endian 7154 - emulate the access, including the delivery of an exception if the 7155 access didn't succeed 7156 - provide a return value in the case of ST64BV/ST64BV0 7157 - return the data in the case of a load 7158 - increment PC if the instruction was successfully executed 7159 7160Note that there is no expectation of performance for this emulation, as it 7161involves a large number of interaction with the guest state. It is, however, 7162expected that the instruction's semantics are preserved, specially the 7163single-copy atomicity property of the 64 byte access. 7164 7165This exit reason must be handled if userspace sets ID_AA64ISAR1_EL1.LS64 to a 7166non-zero value, indicating that FEAT_LS64* is enabled. 7167 7168:: 7169 7170 /* KVM_EXIT_X86_RDMSR / KVM_EXIT_X86_WRMSR */ 7171 struct { 7172 __u8 error; /* user -> kernel */ 7173 __u8 pad[7]; 7174 __u32 reason; /* kernel -> user */ 7175 __u32 index; /* kernel -> user */ 7176 __u64 data; /* kernel <-> user */ 7177 } msr; 7178 7179Used on x86 systems. When the VM capability KVM_CAP_X86_USER_SPACE_MSR is 7180enabled, MSR accesses to registers that would invoke a #GP by KVM kernel code 7181may instead trigger a KVM_EXIT_X86_RDMSR exit for reads and KVM_EXIT_X86_WRMSR 7182exit for writes. 7183 7184The "reason" field specifies why the MSR interception occurred. Userspace will 7185only receive MSR exits when a particular reason was requested during through 7186ENABLE_CAP. Currently valid exit reasons are: 7187 7188============================ ======================================== 7189 KVM_MSR_EXIT_REASON_UNKNOWN access to MSR that is unknown to KVM 7190 KVM_MSR_EXIT_REASON_INVAL access to invalid MSRs or reserved bits 7191 KVM_MSR_EXIT_REASON_FILTER access blocked by KVM_X86_SET_MSR_FILTER 7192============================ ======================================== 7193 7194For KVM_EXIT_X86_RDMSR, the "index" field tells userspace which MSR the guest 7195wants to read. To respond to this request with a successful read, userspace 7196writes the respective data into the "data" field and must continue guest 7197execution to ensure the read data is transferred into guest register state. 7198 7199If the RDMSR request was unsuccessful, userspace indicates that with a "1" in 7200the "error" field. This will inject a #GP into the guest when the VCPU is 7201executed again. 7202 7203For KVM_EXIT_X86_WRMSR, the "index" field tells userspace which MSR the guest 7204wants to write. Once finished processing the event, userspace must continue 7205vCPU execution. If the MSR write was unsuccessful, userspace also sets the 7206"error" field to "1". 7207 7208See KVM_X86_SET_MSR_FILTER for details on the interaction with MSR filtering. 7209 7210:: 7211 7212 7213 struct kvm_xen_exit { 7214 #define KVM_EXIT_XEN_HCALL 1 7215 __u32 type; 7216 union { 7217 struct { 7218 __u32 longmode; 7219 __u32 cpl; 7220 __u64 input; 7221 __u64 result; 7222 __u64 params[6]; 7223 } hcall; 7224 } u; 7225 }; 7226 /* KVM_EXIT_XEN */ 7227 struct kvm_hyperv_exit xen; 7228 7229Indicates that the VCPU exits into userspace to process some tasks 7230related to Xen emulation. 7231 7232Valid values for 'type' are: 7233 7234 - KVM_EXIT_XEN_HCALL -- synchronously notify user-space about Xen hypercall. 7235 Userspace is expected to place the hypercall result into the appropriate 7236 field before invoking KVM_RUN again. 7237 7238:: 7239 7240 /* KVM_EXIT_RISCV_SBI */ 7241 struct { 7242 unsigned long extension_id; 7243 unsigned long function_id; 7244 unsigned long args[6]; 7245 unsigned long ret[2]; 7246 } riscv_sbi; 7247 7248If exit reason is KVM_EXIT_RISCV_SBI then it indicates that the VCPU has 7249done a SBI call which is not handled by KVM RISC-V kernel module. The details 7250of the SBI call are available in 'riscv_sbi' member of kvm_run structure. The 7251'extension_id' field of 'riscv_sbi' represents SBI extension ID whereas the 7252'function_id' field represents function ID of given SBI extension. The 'args' 7253array field of 'riscv_sbi' represents parameters for the SBI call and 'ret' 7254array field represents return values. The userspace should update the return 7255values of SBI call before resuming the VCPU. For more details on RISC-V SBI 7256spec refer, https://github.com/riscv/riscv-sbi-doc. 7257 7258:: 7259 7260 /* KVM_EXIT_MEMORY_FAULT */ 7261 struct { 7262 #define KVM_MEMORY_EXIT_FLAG_PRIVATE (1ULL << 3) 7263 __u64 flags; 7264 __u64 gpa; 7265 __u64 size; 7266 } memory_fault; 7267 7268KVM_EXIT_MEMORY_FAULT indicates the vCPU has encountered a memory fault that 7269could not be resolved by KVM. The 'gpa' and 'size' (in bytes) describe the 7270guest physical address range [gpa, gpa + size) of the fault. The 'flags' field 7271describes properties of the faulting access that are likely pertinent: 7272 7273 - KVM_MEMORY_EXIT_FLAG_PRIVATE - When set, indicates the memory fault occurred 7274 on a private memory access. When clear, indicates the fault occurred on a 7275 shared access. 7276 7277Note! KVM_EXIT_MEMORY_FAULT is unique among all KVM exit reasons in that it 7278accompanies a return code of '-1', not '0'! errno will always be set to EFAULT 7279or EHWPOISON when KVM exits with KVM_EXIT_MEMORY_FAULT, userspace should assume 7280kvm_run.exit_reason is stale/undefined for all other error numbers. 7281 7282:: 7283 7284 /* KVM_EXIT_NOTIFY */ 7285 struct { 7286 #define KVM_NOTIFY_CONTEXT_INVALID (1 << 0) 7287 __u32 flags; 7288 } notify; 7289 7290Used on x86 systems. When the VM capability KVM_CAP_X86_NOTIFY_VMEXIT is 7291enabled, a VM exit generated if no event window occurs in VM non-root mode 7292for a specified amount of time. Once KVM_X86_NOTIFY_VMEXIT_USER is set when 7293enabling the cap, it would exit to userspace with the exit reason 7294KVM_EXIT_NOTIFY for further handling. The "flags" field contains more 7295detailed info. 7296 7297The valid value for 'flags' is: 7298 7299 - KVM_NOTIFY_CONTEXT_INVALID -- the VM context is corrupted and not valid 7300 in VMCS. It would run into unknown result if resume the target VM. 7301 7302:: 7303 7304 /* KVM_EXIT_TDX */ 7305 struct { 7306 __u64 flags; 7307 __u64 nr; 7308 union { 7309 struct { 7310 u64 ret; 7311 u64 data[5]; 7312 } unknown; 7313 struct { 7314 u64 ret; 7315 u64 gpa; 7316 u64 size; 7317 } get_quote; 7318 struct { 7319 u64 ret; 7320 u64 leaf; 7321 u64 r11, r12, r13, r14; 7322 } get_tdvmcall_info; 7323 struct { 7324 u64 ret; 7325 u64 vector; 7326 } setup_event_notify; 7327 }; 7328 } tdx; 7329 7330Process a TDVMCALL from the guest. KVM forwards select TDVMCALL based 7331on the Guest-Hypervisor Communication Interface (GHCI) specification; 7332KVM bridges these requests to the userspace VMM with minimal changes, 7333placing the inputs in the union and copying them back to the guest 7334on re-entry. 7335 7336Flags are currently always zero, whereas ``nr`` contains the TDVMCALL 7337number from register R11. The remaining field of the union provide the 7338inputs and outputs of the TDVMCALL. Currently the following values of 7339``nr`` are defined: 7340 7341 * ``TDVMCALL_GET_QUOTE``: the guest has requested to generate a TD-Quote 7342 signed by a service hosting TD-Quoting Enclave operating on the host. 7343 Parameters and return value are in the ``get_quote`` field of the union. 7344 The ``gpa`` field and ``size`` specify the guest physical address 7345 (without the shared bit set) and the size of a shared-memory buffer, in 7346 which the TDX guest passes a TD Report. The ``ret`` field represents 7347 the return value of the GetQuote request. When the request has been 7348 queued successfully, the TDX guest can poll the status field in the 7349 shared-memory area to check whether the Quote generation is completed or 7350 not. When completed, the generated Quote is returned via the same buffer. 7351 7352 * ``TDVMCALL_GET_TD_VM_CALL_INFO``: the guest has requested the support 7353 status of TDVMCALLs. The output values for the given leaf should be 7354 placed in fields from ``r11`` to ``r14`` of the ``get_tdvmcall_info`` 7355 field of the union. 7356 7357 * ``TDVMCALL_SETUP_EVENT_NOTIFY_INTERRUPT``: the guest has requested to 7358 set up a notification interrupt for vector ``vector``. 7359 7360KVM may add support for more values in the future that may cause a userspace 7361exit, even without calls to ``KVM_ENABLE_CAP`` or similar. In this case, 7362it will enter with output fields already valid; in the common case, the 7363``unknown.ret`` field of the union will be ``TDVMCALL_STATUS_SUBFUNC_UNSUPPORTED``. 7364Userspace need not do anything if it does not wish to support a TDVMCALL. 7365 7366:: 7367 7368 /* KVM_EXIT_ARM_SEA */ 7369 struct { 7370 #define KVM_EXIT_ARM_SEA_FLAG_GPA_VALID (1ULL << 0) 7371 __u64 flags; 7372 __u64 esr; 7373 __u64 gva; 7374 __u64 gpa; 7375 } arm_sea; 7376 7377Used on arm64 systems. When the VM capability ``KVM_CAP_ARM_SEA_TO_USER`` is 7378enabled, a KVM exits to userspace if a guest access causes a synchronous 7379external abort (SEA) and the host APEI fails to handle the SEA. 7380 7381``esr`` is set to a sanitized value of ESR_EL2 from the exception taken to KVM, 7382consisting of the following fields: 7383 7384 - ``ESR_EL2.EC`` 7385 - ``ESR_EL2.IL`` 7386 - ``ESR_EL2.FnV`` 7387 - ``ESR_EL2.EA`` 7388 - ``ESR_EL2.CM`` 7389 - ``ESR_EL2.WNR`` 7390 - ``ESR_EL2.FSC`` 7391 - ``ESR_EL2.SET`` (when FEAT_RAS is implemented for the VM) 7392 7393``gva`` is set to the value of FAR_EL2 from the exception taken to KVM when 7394``ESR_EL2.FnV == 0``. Otherwise, the value of ``gva`` is unknown. 7395 7396``gpa`` is set to the faulting IPA from the exception taken to KVM when 7397the ``KVM_EXIT_ARM_SEA_FLAG_GPA_VALID`` flag is set. Otherwise, the value of 7398``gpa`` is unknown. 7399 7400:: 7401 7402 /* Fix the size of the union. */ 7403 char padding[256]; 7404 }; 7405 7406 /* 7407 * shared registers between kvm and userspace. 7408 * kvm_valid_regs specifies the register classes set by the host 7409 * kvm_dirty_regs specified the register classes dirtied by userspace 7410 * struct kvm_sync_regs is architecture specific, as well as the 7411 * bits for kvm_valid_regs and kvm_dirty_regs 7412 */ 7413 __u64 kvm_valid_regs; 7414 __u64 kvm_dirty_regs; 7415 union { 7416 struct kvm_sync_regs regs; 7417 char padding[SYNC_REGS_SIZE_BYTES]; 7418 } s; 7419 7420If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access 7421certain guest registers without having to call SET/GET_*REGS. Thus we can 7422avoid some system call overhead if userspace has to handle the exit. 7423Userspace can query the validity of the structure by checking 7424kvm_valid_regs for specific bits. These bits are architecture specific 7425and usually define the validity of a groups of registers. (e.g. one bit 7426for general purpose registers) 7427 7428Please note that the kernel is allowed to use the kvm_run structure as the 7429primary storage for certain register types. Therefore, the kernel may use the 7430values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set. 7431 7432:: 7433 7434 /* KVM_EXIT_SNP_REQ_CERTS */ 7435 struct kvm_exit_snp_req_certs { 7436 __u64 gpa; 7437 __u64 npages; 7438 __u64 ret; 7439 }; 7440 7441KVM_EXIT_SNP_REQ_CERTS indicates an SEV-SNP guest with certificate-fetching 7442enabled (see KVM_SEV_SNP_ENABLE_REQ_CERTS) has generated an Extended Guest 7443Request NAE #VMGEXIT (SNP_GUEST_REQUEST) with message type MSG_REPORT_REQ, 7444i.e. has requested an attestation report from firmware, and would like the 7445certificate data corresponding to the attestation report signature to be 7446provided by the hypervisor as part of the request. 7447 7448To allow for userspace to provide the certificate, the 'gpa' and 'npages' 7449are forwarded verbatim from the guest request (the RAX and RBX GHCB fields 7450respectively). 'ret' is not an "output" from KVM, and is always '0' on 7451exit. KVM verifies the 'gpa' is 4KiB aligned prior to exiting to userspace, 7452but otherwise the information from the guest isn't validated. 7453 7454Upon the next KVM_RUN, e.g. after userspace has serviced the request (or not), 7455KVM will complete the #VMGEXIT, using the 'ret' field to determine whether to 7456signal success or failure to the guest, and on failure, what reason code will 7457be communicated via SW_EXITINFO2. If 'ret' is set to an unsupported value (see 7458the table below), KVM_RUN will fail with -EINVAL. For a 'ret' of 'ENOSPC', KVM 7459also consumes the 'npages' field, i.e. userspace can use the field to inform 7460the guest of the number of pages needed to hold all the certificate data. 7461 7462The supported 'ret' values and their respective SW_EXITINFO2 encodings: 7463 7464 ====== ============================================================= 7465 0 0x0, i.e. success. KVM will emit an SNP_GUEST_REQUEST command 7466 to SNP firmware. 7467 ENOSPC 0x0000000100000000, i.e. not enough guest pages to hold the 7468 certificate table and certificate data. KVM will also set the 7469 RBX field in the GHBC to 'npages'. 7470 EAGAIN 0x0000000200000000, i.e. the host is busy and the guest should 7471 retry the request. 7472 EIO 0xffffffff00000000, for all other errors (this return code is 7473 a KVM-defined hypervisor value, as allowed by the GHCB) 7474 ====== ============================================================= 7475 7476 7477.. _cap_enable: 7478 74796. Capabilities that can be enabled on vCPUs 7480============================================ 7481 7482There are certain capabilities that change the behavior of the virtual CPU or 7483the virtual machine when enabled. To enable them, please see 7484:ref:`KVM_ENABLE_CAP`. 7485 7486Below you can find a list of capabilities and what their effect on the vCPU or 7487the virtual machine is when enabling them. 7488 7489The following information is provided along with the description: 7490 7491 Architectures: 7492 which instruction set architectures provide this ioctl. 7493 x86 includes both i386 and x86_64. 7494 7495 Target: 7496 whether this is a per-vcpu or per-vm capability. 7497 7498 Parameters: 7499 what parameters are accepted by the capability. 7500 7501 Returns: 7502 the return value. General error numbers (EBADF, ENOMEM, EINVAL) 7503 are not detailed, but errors with specific meanings are. 7504 7505 75066.1 KVM_CAP_PPC_OSI 7507------------------- 7508 7509:Architectures: ppc 7510:Target: vcpu 7511:Parameters: none 7512:Returns: 0 on success; -1 on error 7513 7514This capability enables interception of OSI hypercalls that otherwise would 7515be treated as normal system calls to be injected into the guest. OSI hypercalls 7516were invented by Mac-on-Linux to have a standardized communication mechanism 7517between the guest and the host. 7518 7519When this capability is enabled, KVM_EXIT_OSI can occur. 7520 7521 75226.2 KVM_CAP_PPC_PAPR 7523-------------------- 7524 7525:Architectures: ppc 7526:Target: vcpu 7527:Parameters: none 7528:Returns: 0 on success; -1 on error 7529 7530This capability enables interception of PAPR hypercalls. PAPR hypercalls are 7531done using the hypercall instruction "sc 1". 7532 7533It also sets the guest privilege level to "supervisor" mode. Usually the guest 7534runs in "hypervisor" privilege mode with a few missing features. 7535 7536In addition to the above, it changes the semantics of SDR1. In this mode, the 7537HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the 7538HTAB invisible to the guest. 7539 7540When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur. 7541 7542 75436.3 KVM_CAP_SW_TLB 7544------------------ 7545 7546:Architectures: ppc 7547:Target: vcpu 7548:Parameters: args[0] is the address of a struct kvm_config_tlb 7549:Returns: 0 on success; -1 on error 7550 7551:: 7552 7553 struct kvm_config_tlb { 7554 __u64 params; 7555 __u64 array; 7556 __u32 mmu_type; 7557 __u32 array_len; 7558 }; 7559 7560Configures the virtual CPU's TLB array, establishing a shared memory area 7561between userspace and KVM. The "params" and "array" fields are userspace 7562addresses of mmu-type-specific data structures. The "array_len" field is an 7563safety mechanism, and should be set to the size in bytes of the memory that 7564userspace has reserved for the array. It must be at least the size dictated 7565by "mmu_type" and "params". 7566 7567While KVM_RUN is active, the shared region is under control of KVM. Its 7568contents are undefined, and any modification by userspace results in 7569boundedly undefined behavior. 7570 7571On return from KVM_RUN, the shared region will reflect the current state of 7572the guest's TLB. If userspace makes any changes, it must call KVM_DIRTY_TLB 7573to tell KVM which entries have been changed, prior to calling KVM_RUN again 7574on this vcpu. 7575 7576For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV: 7577 7578 - The "params" field is of type "struct kvm_book3e_206_tlb_params". 7579 - The "array" field points to an array of type "struct 7580 kvm_book3e_206_tlb_entry". 7581 - The array consists of all entries in the first TLB, followed by all 7582 entries in the second TLB. 7583 - Within a TLB, entries are ordered first by increasing set number. Within a 7584 set, entries are ordered by way (increasing ESEL). 7585 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1) 7586 where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value. 7587 - The tsize field of mas1 shall be set to 4K on TLB0, even though the 7588 hardware ignores this value for TLB0. 7589 75906.4 KVM_CAP_S390_CSS_SUPPORT 7591---------------------------- 7592 7593:Architectures: s390 7594:Target: vcpu 7595:Parameters: none 7596:Returns: 0 on success; -1 on error 7597 7598This capability enables support for handling of channel I/O instructions. 7599 7600TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are 7601handled in-kernel, while the other I/O instructions are passed to userspace. 7602 7603When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST 7604SUBCHANNEL intercepts. 7605 7606Note that even though this capability is enabled per-vcpu, the complete 7607virtual machine is affected. 7608 76096.5 KVM_CAP_PPC_EPR 7610------------------- 7611 7612:Architectures: ppc 7613:Target: vcpu 7614:Parameters: args[0] defines whether the proxy facility is active 7615:Returns: 0 on success; -1 on error 7616 7617This capability enables or disables the delivery of interrupts through the 7618external proxy facility. 7619 7620When enabled (args[0] != 0), every time the guest gets an external interrupt 7621delivered, it automatically exits into user space with a KVM_EXIT_EPR exit 7622to receive the topmost interrupt vector. 7623 7624When disabled (args[0] == 0), behavior is as if this facility is unsupported. 7625 7626When this capability is enabled, KVM_EXIT_EPR can occur. 7627 76286.6 KVM_CAP_IRQ_MPIC 7629-------------------- 7630 7631:Architectures: ppc 7632:Parameters: args[0] is the MPIC device fd; 7633 args[1] is the MPIC CPU number for this vcpu 7634 7635This capability connects the vcpu to an in-kernel MPIC device. 7636 76376.7 KVM_CAP_IRQ_XICS 7638-------------------- 7639 7640:Architectures: ppc 7641:Target: vcpu 7642:Parameters: args[0] is the XICS device fd; 7643 args[1] is the XICS CPU number (server ID) for this vcpu 7644 7645This capability connects the vcpu to an in-kernel XICS device. 7646 76476.8 KVM_CAP_S390_IRQCHIP 7648------------------------ 7649 7650:Architectures: s390 7651:Target: vm 7652:Parameters: none 7653 7654This capability enables the in-kernel irqchip for s390. Please refer to 7655"4.24 KVM_CREATE_IRQCHIP" for details. 7656 76576.9 KVM_CAP_MIPS_FPU 7658-------------------- 7659 7660:Architectures: mips 7661:Target: vcpu 7662:Parameters: args[0] is reserved for future use (should be 0). 7663 7664This capability allows the use of the host Floating Point Unit by the guest. It 7665allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is 7666done the ``KVM_REG_MIPS_FPR_*`` and ``KVM_REG_MIPS_FCR_*`` registers can be 7667accessed (depending on the current guest FPU register mode), and the Status.FR, 7668Config5.FRE bits are accessible via the KVM API and also from the guest, 7669depending on them being supported by the FPU. 7670 76716.10 KVM_CAP_MIPS_MSA 7672--------------------- 7673 7674:Architectures: mips 7675:Target: vcpu 7676:Parameters: args[0] is reserved for future use (should be 0). 7677 7678This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest. 7679It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest. 7680Once this is done the ``KVM_REG_MIPS_VEC_*`` and ``KVM_REG_MIPS_MSA_*`` 7681registers can be accessed, and the Config5.MSAEn bit is accessible via the 7682KVM API and also from the guest. 7683 76846.74 KVM_CAP_SYNC_REGS 7685---------------------- 7686 7687:Architectures: s390, x86 7688:Target: s390: always enabled, x86: vcpu 7689:Parameters: none 7690:Returns: x86: KVM_CHECK_EXTENSION returns a bit-array indicating which register 7691 sets are supported 7692 (bitfields defined in arch/x86/include/uapi/asm/kvm.h). 7693 7694As described above in the kvm_sync_regs struct info in section :ref:`kvm_run`, 7695KVM_CAP_SYNC_REGS "allow[s] userspace to access certain guest registers 7696without having to call SET/GET_*REGS". This reduces overhead by eliminating 7697repeated ioctl calls for setting and/or getting register values. This is 7698particularly important when userspace is making synchronous guest state 7699modifications, e.g. when emulating and/or intercepting instructions in 7700userspace. 7701 7702For s390 specifics, please refer to the source code. 7703 7704For x86: 7705 7706- the register sets to be copied out to kvm_run are selectable 7707 by userspace (rather that all sets being copied out for every exit). 7708- vcpu_events are available in addition to regs and sregs. 7709 7710For x86, the 'kvm_valid_regs' field of struct kvm_run is overloaded to 7711function as an input bit-array field set by userspace to indicate the 7712specific register sets to be copied out on the next exit. 7713 7714To indicate when userspace has modified values that should be copied into 7715the vCPU, the all architecture bitarray field, 'kvm_dirty_regs' must be set. 7716This is done using the same bitflags as for the 'kvm_valid_regs' field. 7717If the dirty bit is not set, then the register set values will not be copied 7718into the vCPU even if they've been modified. 7719 7720Unused bitfields in the bitarrays must be set to zero. 7721 7722:: 7723 7724 struct kvm_sync_regs { 7725 struct kvm_regs regs; 7726 struct kvm_sregs sregs; 7727 struct kvm_vcpu_events events; 7728 }; 7729 77306.75 KVM_CAP_PPC_IRQ_XIVE 7731------------------------- 7732 7733:Architectures: ppc 7734:Target: vcpu 7735:Parameters: args[0] is the XIVE device fd; 7736 args[1] is the XIVE CPU number (server ID) for this vcpu 7737 7738This capability connects the vcpu to an in-kernel XIVE device. 7739 77406.76 KVM_CAP_HYPERV_SYNIC 7741------------------------- 7742 7743:Architectures: x86 7744:Target: vcpu 7745 7746This capability, if KVM_CHECK_EXTENSION indicates that it is 7747available, means that the kernel has an implementation of the 7748Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is 7749used to support Windows Hyper-V based guest paravirt drivers(VMBus). 7750 7751In order to use SynIC, it has to be activated by setting this 7752capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this 7753will disable the use of APIC hardware virtualization even if supported 7754by the CPU, as it's incompatible with SynIC auto-EOI behavior. 7755 77566.77 KVM_CAP_HYPERV_SYNIC2 7757-------------------------- 7758 7759:Architectures: x86 7760:Target: vcpu 7761 7762This capability enables a newer version of Hyper-V Synthetic interrupt 7763controller (SynIC). The only difference with KVM_CAP_HYPERV_SYNIC is that KVM 7764doesn't clear SynIC message and event flags pages when they are enabled by 7765writing to the respective MSRs. 7766 77676.78 KVM_CAP_HYPERV_DIRECT_TLBFLUSH 7768----------------------------------- 7769 7770:Architectures: x86 7771:Target: vcpu 7772 7773This capability indicates that KVM running on top of Hyper-V hypervisor 7774enables Direct TLB flush for its guests meaning that TLB flush 7775hypercalls are handled by Level 0 hypervisor (Hyper-V) bypassing KVM. 7776Due to the different ABI for hypercall parameters between Hyper-V and 7777KVM, enabling this capability effectively disables all hypercall 7778handling by KVM (as some KVM hypercall may be mistakenly treated as TLB 7779flush hypercalls by Hyper-V) so userspace should disable KVM identification 7780in CPUID and only exposes Hyper-V identification. In this case, guest 7781thinks it's running on Hyper-V and only use Hyper-V hypercalls. 7782 77836.79 KVM_CAP_HYPERV_ENFORCE_CPUID 7784--------------------------------- 7785 7786:Architectures: x86 7787:Target: vcpu 7788 7789When enabled, KVM will disable emulated Hyper-V features provided to the 7790guest according to the bits Hyper-V CPUID feature leaves. Otherwise, all 7791currently implemented Hyper-V features are provided unconditionally when 7792Hyper-V identification is set in the HYPERV_CPUID_INTERFACE (0x40000001) 7793leaf. 7794 77956.80 KVM_CAP_ENFORCE_PV_FEATURE_CPUID 7796------------------------------------- 7797 7798:Architectures: x86 7799:Target: vcpu 7800 7801When enabled, KVM will disable paravirtual features provided to the 7802guest according to the bits in the KVM_CPUID_FEATURES CPUID leaf 7803(0x40000001). Otherwise, a guest may use the paravirtual features 7804regardless of what has actually been exposed through the CPUID leaf. 7805 7806.. _KVM_CAP_DIRTY_LOG_RING: 7807 7808 7809.. _cap_enable_vm: 7810 78117. Capabilities that can be enabled on VMs 7812========================================== 7813 7814There are certain capabilities that change the behavior of the virtual 7815machine when enabled. To enable them, please see section 7816:ref:`KVM_ENABLE_CAP`. Below you can find a list of capabilities and 7817what their effect on the VM is when enabling them. 7818 7819The following information is provided along with the description: 7820 7821 Architectures: 7822 which instruction set architectures provide this ioctl. 7823 x86 includes both i386 and x86_64. 7824 7825 Parameters: 7826 what parameters are accepted by the capability. 7827 7828 Returns: 7829 the return value. General error numbers (EBADF, ENOMEM, EINVAL) 7830 are not detailed, but errors with specific meanings are. 7831 7832 78337.1 KVM_CAP_PPC_ENABLE_HCALL 7834---------------------------- 7835 7836:Architectures: ppc 7837:Parameters: args[0] is the sPAPR hcall number; 7838 args[1] is 0 to disable, 1 to enable in-kernel handling 7839 7840This capability controls whether individual sPAPR hypercalls (hcalls) 7841get handled by the kernel or not. Enabling or disabling in-kernel 7842handling of an hcall is effective across the VM. On creation, an 7843initial set of hcalls are enabled for in-kernel handling, which 7844consists of those hcalls for which in-kernel handlers were implemented 7845before this capability was implemented. If disabled, the kernel will 7846not to attempt to handle the hcall, but will always exit to userspace 7847to handle it. Note that it may not make sense to enable some and 7848disable others of a group of related hcalls, but KVM does not prevent 7849userspace from doing that. 7850 7851If the hcall number specified is not one that has an in-kernel 7852implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL 7853error. 7854 78557.2 KVM_CAP_S390_USER_SIGP 7856-------------------------- 7857 7858:Architectures: s390 7859:Parameters: none 7860 7861This capability controls which SIGP orders will be handled completely in user 7862space. With this capability enabled, all fast orders will be handled completely 7863in the kernel: 7864 7865- SENSE 7866- SENSE RUNNING 7867- EXTERNAL CALL 7868- EMERGENCY SIGNAL 7869- CONDITIONAL EMERGENCY SIGNAL 7870 7871All other orders will be handled completely in user space. 7872 7873Only privileged operation exceptions will be checked for in the kernel (or even 7874in the hardware prior to interception). If this capability is not enabled, the 7875old way of handling SIGP orders is used (partially in kernel and user space). 7876 78777.3 KVM_CAP_S390_VECTOR_REGISTERS 7878--------------------------------- 7879 7880:Architectures: s390 7881:Parameters: none 7882:Returns: 0 on success, negative value on error 7883 7884Allows use of the vector registers introduced with z13 processor, and 7885provides for the synchronization between host and user space. Will 7886return -EINVAL if the machine does not support vectors. 7887 78887.4 KVM_CAP_S390_USER_STSI 7889-------------------------- 7890 7891:Architectures: s390 7892:Parameters: none 7893 7894This capability allows post-handlers for the STSI instruction. After 7895initial handling in the kernel, KVM exits to user space with 7896KVM_EXIT_S390_STSI to allow user space to insert further data. 7897 7898Before exiting to userspace, kvm handlers should fill in s390_stsi field of 7899vcpu->run:: 7900 7901 struct { 7902 __u64 addr; 7903 __u8 ar; 7904 __u8 reserved; 7905 __u8 fc; 7906 __u8 sel1; 7907 __u16 sel2; 7908 } s390_stsi; 7909 7910 @addr - guest address of STSI SYSIB 7911 @fc - function code 7912 @sel1 - selector 1 7913 @sel2 - selector 2 7914 @ar - access register number 7915 7916KVM handlers should exit to userspace with rc = -EREMOTE. 7917 79187.5 KVM_CAP_SPLIT_IRQCHIP 7919------------------------- 7920 7921:Architectures: x86 7922:Parameters: args[0] - number of routes reserved for userspace IOAPICs 7923:Returns: 0 on success, -1 on error 7924 7925Create a local apic for each processor in the kernel. This can be used 7926instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the 7927IOAPIC and PIC (and also the PIT, even though this has to be enabled 7928separately). 7929 7930This capability also enables in kernel routing of interrupt requests; 7931when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are 7932used in the IRQ routing table. The first args[0] MSI routes are reserved 7933for the IOAPIC pins. Whenever the LAPIC receives an EOI for these routes, 7934a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace. 7935 7936Fails if VCPU has already been created, or if the irqchip is already in the 7937kernel (i.e. KVM_CREATE_IRQCHIP has already been called). 7938 79397.6 KVM_CAP_S390_RI 7940------------------- 7941 7942:Architectures: s390 7943:Parameters: none 7944 7945Allows use of runtime-instrumentation introduced with zEC12 processor. 7946Will return -EINVAL if the machine does not support runtime-instrumentation. 7947Will return -EBUSY if a VCPU has already been created. 7948 79497.7 KVM_CAP_X2APIC_API 7950---------------------- 7951 7952:Architectures: x86 7953:Parameters: args[0] - features that should be enabled 7954:Returns: 0 on success, -EINVAL when args[0] contains invalid features 7955 7956Valid feature flags in args[0] are:: 7957 7958 #define KVM_X2APIC_API_USE_32BIT_IDS (1ULL << 0) 7959 #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK (1ULL << 1) 7960 #define KVM_X2APIC_ENABLE_SUPPRESS_EOI_BROADCAST (1ULL << 2) 7961 #define KVM_X2APIC_DISABLE_SUPPRESS_EOI_BROADCAST (1ULL << 3) 7962 7963Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of 7964KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC, 7965allowing the use of 32-bit APIC IDs. See KVM_CAP_X2APIC_API in their 7966respective sections. 7967 7968KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work 7969in logical mode or with more than 255 VCPUs. Otherwise, KVM treats 0xff 7970as a broadcast even in x2APIC mode in order to support physical x2APIC 7971without interrupt remapping. This is undesirable in logical mode, 7972where 0xff represents CPUs 0-7 in cluster 0. 7973 7974Setting KVM_X2APIC_ENABLE_SUPPRESS_EOI_BROADCAST instructs KVM to enable 7975Suppress EOI Broadcasts. KVM will advertise support for Suppress EOI 7976Broadcast to the guest and suppress LAPIC EOI broadcasts when the guest 7977sets the Suppress EOI Broadcast bit in the SPIV register. This flag is 7978supported only when using a split IRQCHIP. 7979 7980Setting KVM_X2APIC_DISABLE_SUPPRESS_EOI_BROADCAST disables support for 7981Suppress EOI Broadcasts entirely, i.e. instructs KVM to NOT advertise 7982support to the guest. 7983 7984Modern VMMs should either enable KVM_X2APIC_ENABLE_SUPPRESS_EOI_BROADCAST 7985or KVM_X2APIC_DISABLE_SUPPRESS_EOI_BROADCAST. If not, legacy quirky 7986behavior will be used by KVM: in split IRQCHIP mode, KVM will advertise 7987support for Suppress EOI Broadcasts but not actually suppress EOI 7988broadcasts; for in-kernel IRQCHIP mode, KVM will not advertise support for 7989Suppress EOI Broadcasts. 7990 7991Setting both KVM_X2APIC_ENABLE_SUPPRESS_EOI_BROADCAST and 7992KVM_X2APIC_DISABLE_SUPPRESS_EOI_BROADCAST will fail with an EINVAL error, 7993as will setting KVM_X2APIC_ENABLE_SUPPRESS_EOI_BROADCAST without a split 7994IRCHIP. 7995 79967.8 KVM_CAP_S390_USER_INSTR0 7997---------------------------- 7998 7999:Architectures: s390 8000:Parameters: none 8001 8002With this capability enabled, the illegal instruction 0x0000 (2 bytes) will 8003be intercepted and forwarded to user space. User space can use this 8004mechanism e.g. to realize 2-byte software breakpoints. The kernel will 8005not inject an operating exception for these instructions, user space has 8006to take care of that. 8007 8008This capability can be enabled dynamically even if VCPUs were already 8009created and are running. 8010 80117.9 KVM_CAP_S390_GS 8012------------------- 8013 8014:Architectures: s390 8015:Parameters: none 8016:Returns: 0 on success; -EINVAL if the machine does not support 8017 guarded storage; -EBUSY if a VCPU has already been created. 8018 8019Allows use of guarded storage for the KVM guest. 8020 80217.10 KVM_CAP_S390_AIS 8022--------------------- 8023 8024:Architectures: s390 8025:Parameters: none 8026 8027Allow use of adapter-interruption suppression. 8028:Returns: 0 on success; -EBUSY if a VCPU has already been created. 8029 80307.11 KVM_CAP_PPC_SMT 8031-------------------- 8032 8033:Architectures: ppc 8034:Parameters: vsmt_mode, flags 8035 8036Enabling this capability on a VM provides userspace with a way to set 8037the desired virtual SMT mode (i.e. the number of virtual CPUs per 8038virtual core). The virtual SMT mode, vsmt_mode, must be a power of 2 8039between 1 and 8. On POWER8, vsmt_mode must also be no greater than 8040the number of threads per subcore for the host. Currently flags must 8041be 0. A successful call to enable this capability will result in 8042vsmt_mode being returned when the KVM_CAP_PPC_SMT capability is 8043subsequently queried for the VM. This capability is only supported by 8044HV KVM, and can only be set before any VCPUs have been created. 8045The KVM_CAP_PPC_SMT_POSSIBLE capability indicates which virtual SMT 8046modes are available. 8047 80487.12 KVM_CAP_PPC_FWNMI 8049---------------------- 8050 8051:Architectures: ppc 8052:Parameters: none 8053 8054With this capability a machine check exception in the guest address 8055space will cause KVM to exit the guest with NMI exit reason. This 8056enables QEMU to build error log and branch to guest kernel registered 8057machine check handling routine. Without this capability KVM will 8058branch to guests' 0x200 interrupt vector. 8059 80607.13 KVM_CAP_X86_DISABLE_EXITS 8061------------------------------ 8062 8063:Architectures: x86 8064:Parameters: args[0] defines which exits are disabled 8065:Returns: 0 on success, -EINVAL when args[0] contains invalid exits 8066 or if any vCPUs have already been created 8067 8068Valid bits in args[0] are:: 8069 8070 #define KVM_X86_DISABLE_EXITS_MWAIT (1 << 0) 8071 #define KVM_X86_DISABLE_EXITS_HLT (1 << 1) 8072 #define KVM_X86_DISABLE_EXITS_PAUSE (1 << 2) 8073 #define KVM_X86_DISABLE_EXITS_CSTATE (1 << 3) 8074 #define KVM_X86_DISABLE_EXITS_APERFMPERF (1 << 4) 8075 8076Enabling this capability on a VM provides userspace with a way to no 8077longer intercept some instructions for improved latency in some 8078workloads, and is suggested when vCPUs are associated to dedicated 8079physical CPUs. More bits can be added in the future; userspace can 8080just pass the KVM_CHECK_EXTENSION result to KVM_ENABLE_CAP to disable 8081all such vmexits. 8082 8083Do not enable KVM_FEATURE_PV_UNHALT if you disable HLT exits. 8084 8085Virtualizing the ``IA32_APERF`` and ``IA32_MPERF`` MSRs requires more 8086than just disabling APERF/MPERF exits. While both Intel and AMD 8087document strict usage conditions for these MSRs--emphasizing that only 8088the ratio of their deltas over a time interval (T0 to T1) is 8089architecturally defined--simply passing through the MSRs can still 8090produce an incorrect ratio. 8091 8092This erroneous ratio can occur if, between T0 and T1: 8093 80941. The vCPU thread migrates between logical processors. 80952. Live migration or suspend/resume operations take place. 80963. Another task shares the vCPU's logical processor. 80974. C-states lower than C0 are emulated (e.g., via HLT interception). 80985. The guest TSC frequency doesn't match the host TSC frequency. 8099 8100Due to these complexities, KVM does not automatically associate this 8101passthrough capability with the guest CPUID bit, 8102``CPUID.6:ECX.APERFMPERF[bit 0]``. Userspace VMMs that deem this 8103mechanism adequate for virtualizing the ``IA32_APERF`` and 8104``IA32_MPERF`` MSRs must set the guest CPUID bit explicitly. 8105 8106 81077.14 KVM_CAP_S390_HPAGE_1M 8108-------------------------- 8109 8110:Architectures: s390 8111:Parameters: none 8112:Returns: 0 on success, -EINVAL if hpage module parameter was not set 8113 or cmma is enabled, or the VM has the KVM_VM_S390_UCONTROL 8114 flag set 8115 8116With this capability the KVM support for memory backing with 1m pages 8117through hugetlbfs can be enabled for a VM. After the capability is 8118enabled, cmma can't be enabled anymore and pfmfi and the storage key 8119interpretation are disabled. If cmma has already been enabled or the 8120hpage module parameter is not set to 1, -EINVAL is returned. 8121 8122While it is generally possible to create a huge page backed VM without 8123this capability, the VM will not be able to run. 8124 81257.15 KVM_CAP_MSR_PLATFORM_INFO 8126------------------------------ 8127 8128:Architectures: x86 8129:Parameters: args[0] whether feature should be enabled or not 8130 8131With this capability, a guest may read the MSR_PLATFORM_INFO MSR. Otherwise, 8132a #GP would be raised when the guest tries to access. Currently, this 8133capability does not enable write permissions of this MSR for the guest. 8134 81357.16 KVM_CAP_PPC_NESTED_HV 8136-------------------------- 8137 8138:Architectures: ppc 8139:Parameters: none 8140:Returns: 0 on success, -EINVAL when the implementation doesn't support 8141 nested-HV virtualization. 8142 8143HV-KVM on POWER9 and later systems allows for "nested-HV" 8144virtualization, which provides a way for a guest VM to run guests that 8145can run using the CPU's supervisor mode (privileged non-hypervisor 8146state). Enabling this capability on a VM depends on the CPU having 8147the necessary functionality and on the facility being enabled with a 8148kvm-hv module parameter. 8149 81507.17 KVM_CAP_EXCEPTION_PAYLOAD 8151------------------------------ 8152 8153:Architectures: x86 8154:Parameters: args[0] whether feature should be enabled or not 8155 8156With this capability enabled, CR2 will not be modified prior to the 8157emulated VM-exit when L1 intercepts a #PF exception that occurs in 8158L2. Similarly, for kvm-intel only, DR6 will not be modified prior to 8159the emulated VM-exit when L1 intercepts a #DB exception that occurs in 8160L2. As a result, when KVM_GET_VCPU_EVENTS reports a pending #PF (or 8161#DB) exception for L2, exception.has_payload will be set and the 8162faulting address (or the new DR6 bits*) will be reported in the 8163exception_payload field. Similarly, when userspace injects a #PF (or 8164#DB) into L2 using KVM_SET_VCPU_EVENTS, it is expected to set 8165exception.has_payload and to put the faulting address - or the new DR6 8166bits\ [#]_ - in the exception_payload field. 8167 8168This capability also enables exception.pending in struct 8169kvm_vcpu_events, which allows userspace to distinguish between pending 8170and injected exceptions. 8171 8172 8173.. [#] For the new DR6 bits, note that bit 16 is set iff the #DB exception 8174 will clear DR6.RTM. 8175 81767.18 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 8177-------------------------------------- 8178 8179:Architectures: x86, arm64, mips 8180:Parameters: args[0] whether feature should be enabled or not 8181 8182Valid flags are:: 8183 8184 #define KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (1 << 0) 8185 #define KVM_DIRTY_LOG_INITIALLY_SET (1 << 1) 8186 8187With KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE is set, KVM_GET_DIRTY_LOG will not 8188automatically clear and write-protect all pages that are returned as dirty. 8189Rather, userspace will have to do this operation separately using 8190KVM_CLEAR_DIRTY_LOG. 8191 8192At the cost of a slightly more complicated operation, this provides better 8193scalability and responsiveness for two reasons. First, 8194KVM_CLEAR_DIRTY_LOG ioctl can operate on a 64-page granularity rather 8195than requiring to sync a full memslot; this ensures that KVM does not 8196take spinlocks for an extended period of time. Second, in some cases a 8197large amount of time can pass between a call to KVM_GET_DIRTY_LOG and 8198userspace actually using the data in the page. Pages can be modified 8199during this time, which is inefficient for both the guest and userspace: 8200the guest will incur a higher penalty due to write protection faults, 8201while userspace can see false reports of dirty pages. Manual reprotection 8202helps reducing this time, improving guest performance and reducing the 8203number of dirty log false positives. 8204 8205With KVM_DIRTY_LOG_INITIALLY_SET set, all the bits of the dirty bitmap 8206will be initialized to 1 when created. This also improves performance because 8207dirty logging can be enabled gradually in small chunks on the first call 8208to KVM_CLEAR_DIRTY_LOG. KVM_DIRTY_LOG_INITIALLY_SET depends on 8209KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (it is also only available on 8210x86, arm64 and riscv for now). 8211 8212KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 was previously available under the name 8213KVM_CAP_MANUAL_DIRTY_LOG_PROTECT, but the implementation had bugs that make 8214it hard or impossible to use it correctly. The availability of 8215KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 signals that those bugs are fixed. 8216Userspace should not try to use KVM_CAP_MANUAL_DIRTY_LOG_PROTECT. 8217 82187.19 KVM_CAP_PPC_SECURE_GUEST 8219------------------------------ 8220 8221:Architectures: ppc 8222 8223This capability indicates that KVM is running on a host that has 8224ultravisor firmware and thus can support a secure guest. On such a 8225system, a guest can ask the ultravisor to make it a secure guest, 8226one whose memory is inaccessible to the host except for pages which 8227are explicitly requested to be shared with the host. The ultravisor 8228notifies KVM when a guest requests to become a secure guest, and KVM 8229has the opportunity to veto the transition. 8230 8231If present, this capability can be enabled for a VM, meaning that KVM 8232will allow the transition to secure guest mode. Otherwise KVM will 8233veto the transition. 8234 82357.20 KVM_CAP_HALT_POLL 8236---------------------- 8237 8238:Architectures: all 8239:Target: VM 8240:Parameters: args[0] is the maximum poll time in nanoseconds 8241:Returns: 0 on success; -1 on error 8242 8243KVM_CAP_HALT_POLL overrides the kvm.halt_poll_ns module parameter to set the 8244maximum halt-polling time for all vCPUs in the target VM. This capability can 8245be invoked at any time and any number of times to dynamically change the 8246maximum halt-polling time. 8247 8248See Documentation/virt/kvm/halt-polling.rst for more information on halt 8249polling. 8250 82517.21 KVM_CAP_X86_USER_SPACE_MSR 8252------------------------------- 8253 8254:Architectures: x86 8255:Target: VM 8256:Parameters: args[0] contains the mask of KVM_MSR_EXIT_REASON_* events to report 8257:Returns: 0 on success; -1 on error 8258 8259This capability allows userspace to intercept RDMSR and WRMSR instructions if 8260access to an MSR is denied. By default, KVM injects #GP on denied accesses. 8261 8262When a guest requests to read or write an MSR, KVM may not implement all MSRs 8263that are relevant to a respective system. It also does not differentiate by 8264CPU type. 8265 8266To allow more fine grained control over MSR handling, userspace may enable 8267this capability. With it enabled, MSR accesses that match the mask specified in 8268args[0] and would trigger a #GP inside the guest will instead trigger 8269KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR exit notifications. Userspace 8270can then implement model specific MSR handling and/or user notifications 8271to inform a user that an MSR was not emulated/virtualized by KVM. 8272 8273The valid mask flags are: 8274 8275============================ =============================================== 8276 KVM_MSR_EXIT_REASON_UNKNOWN intercept accesses to unknown (to KVM) MSRs 8277 KVM_MSR_EXIT_REASON_INVAL intercept accesses that are architecturally 8278 invalid according to the vCPU model and/or mode 8279 KVM_MSR_EXIT_REASON_FILTER intercept accesses that are denied by userspace 8280 via KVM_X86_SET_MSR_FILTER 8281============================ =============================================== 8282 82837.22 KVM_CAP_X86_BUS_LOCK_EXIT 8284------------------------------- 8285 8286:Architectures: x86 8287:Target: VM 8288:Parameters: args[0] defines the policy used when bus locks detected in guest 8289:Returns: 0 on success, -EINVAL when args[0] contains invalid bits 8290 8291Valid bits in args[0] are:: 8292 8293 #define KVM_BUS_LOCK_DETECTION_OFF (1 << 0) 8294 #define KVM_BUS_LOCK_DETECTION_EXIT (1 << 1) 8295 8296Enabling this capability on a VM provides userspace with a way to select a 8297policy to handle the bus locks detected in guest. Userspace can obtain the 8298supported modes from the result of KVM_CHECK_EXTENSION and define it through 8299the KVM_ENABLE_CAP. The supported modes are mutually-exclusive. 8300 8301This capability allows userspace to force VM exits on bus locks detected in the 8302guest, irrespective whether or not the host has enabled split-lock detection 8303(which triggers an #AC exception that KVM intercepts). This capability is 8304intended to mitigate attacks where a malicious/buggy guest can exploit bus 8305locks to degrade the performance of the whole system. 8306 8307If KVM_BUS_LOCK_DETECTION_OFF is set, KVM doesn't force guest bus locks to VM 8308exit, although the host kernel's split-lock #AC detection still applies, if 8309enabled. 8310 8311If KVM_BUS_LOCK_DETECTION_EXIT is set, KVM enables a CPU feature that ensures 8312bus locks in the guest trigger a VM exit, and KVM exits to userspace for all 8313such VM exits, e.g. to allow userspace to throttle the offending guest and/or 8314apply some other policy-based mitigation. When exiting to userspace, KVM sets 8315KVM_RUN_X86_BUS_LOCK in vcpu-run->flags, and conditionally sets the exit_reason 8316to KVM_EXIT_X86_BUS_LOCK. 8317 8318Due to differences in the underlying hardware implementation, the vCPU's RIP at 8319the time of exit diverges between Intel and AMD. On Intel hosts, RIP points at 8320the next instruction, i.e. the exit is trap-like. On AMD hosts, RIP points at 8321the offending instruction, i.e. the exit is fault-like. 8322 8323Note! Detected bus locks may be coincident with other exits to userspace, i.e. 8324KVM_RUN_X86_BUS_LOCK should be checked regardless of the primary exit reason if 8325userspace wants to take action on all detected bus locks. 8326 83277.23 KVM_CAP_PPC_DAWR1 8328---------------------- 8329 8330:Architectures: ppc 8331:Parameters: none 8332:Returns: 0 on success, -EINVAL when CPU doesn't support 2nd DAWR 8333 8334This capability can be used to check / enable 2nd DAWR feature provided 8335by POWER10 processor. 8336 8337 83387.24 KVM_CAP_VM_COPY_ENC_CONTEXT_FROM 8339------------------------------------- 8340 8341:Architectures: x86 SEV enabled 8342:Type: vm 8343:Parameters: args[0] is the fd of the source vm 8344:Returns: 0 on success; ENOTTY on error 8345 8346This capability enables userspace to copy encryption context from the vm 8347indicated by the fd to the vm this is called on. 8348 8349This is intended to support in-guest workloads scheduled by the host. This 8350allows the in-guest workload to maintain its own NPTs and keeps the two vms 8351from accidentally clobbering each other with interrupts and the like (separate 8352APIC/MSRs/etc). 8353 83547.25 KVM_CAP_SGX_ATTRIBUTE 8355-------------------------- 8356 8357:Architectures: x86 8358:Target: VM 8359:Parameters: args[0] is a file handle of a SGX attribute file in securityfs 8360:Returns: 0 on success, -EINVAL if the file handle is invalid or if a requested 8361 attribute is not supported by KVM. 8362 8363KVM_CAP_SGX_ATTRIBUTE enables a userspace VMM to grant a VM access to one or 8364more privileged enclave attributes. args[0] must hold a file handle to a valid 8365SGX attribute file corresponding to an attribute that is supported/restricted 8366by KVM (currently only PROVISIONKEY). 8367 8368The SGX subsystem restricts access to a subset of enclave attributes to provide 8369additional security for an uncompromised kernel, e.g. use of the PROVISIONKEY 8370is restricted to deter malware from using the PROVISIONKEY to obtain a stable 8371system fingerprint. To prevent userspace from circumventing such restrictions 8372by running an enclave in a VM, KVM prevents access to privileged attributes by 8373default. 8374 8375See Documentation/arch/x86/sgx.rst for more details. 8376 83777.27 KVM_CAP_EXIT_ON_EMULATION_FAILURE 8378-------------------------------------- 8379 8380:Architectures: x86 8381:Parameters: args[0] whether the feature should be enabled or not 8382 8383When this capability is enabled, an emulation failure will result in an exit 8384to userspace with KVM_INTERNAL_ERROR (except when the emulator was invoked 8385to handle a VMware backdoor instruction). Furthermore, KVM will now provide up 8386to 15 instruction bytes for any exit to userspace resulting from an emulation 8387failure. When these exits to userspace occur use the emulation_failure struct 8388instead of the internal struct. They both have the same layout, but the 8389emulation_failure struct matches the content better. It also explicitly 8390defines the 'flags' field which is used to describe the fields in the struct 8391that are valid (ie: if KVM_INTERNAL_ERROR_EMULATION_FLAG_INSTRUCTION_BYTES is 8392set in the 'flags' field then both 'insn_size' and 'insn_bytes' have valid data 8393in them.) 8394 83957.28 KVM_CAP_ARM_MTE 8396-------------------- 8397 8398:Architectures: arm64 8399:Parameters: none 8400 8401This capability indicates that KVM (and the hardware) supports exposing the 8402Memory Tagging Extensions (MTE) to the guest. It must also be enabled by the 8403VMM before creating any VCPUs to allow the guest access. Note that MTE is only 8404available to a guest running in AArch64 mode and enabling this capability will 8405cause attempts to create AArch32 VCPUs to fail. 8406 8407When enabled the guest is able to access tags associated with any memory given 8408to the guest. KVM will ensure that the tags are maintained during swap or 8409hibernation of the host; however the VMM needs to manually save/restore the 8410tags as appropriate if the VM is migrated. 8411 8412When this capability is enabled all memory in memslots must be mapped as 8413``MAP_ANONYMOUS`` or with a RAM-based file mapping (``tmpfs``, ``memfd``), 8414attempts to create a memslot with an invalid mmap will result in an 8415-EINVAL return. 8416 8417When enabled the VMM may make use of the ``KVM_ARM_MTE_COPY_TAGS`` ioctl to 8418perform a bulk copy of tags to/from the guest. 8419 84207.29 KVM_CAP_VM_MOVE_ENC_CONTEXT_FROM 8421------------------------------------- 8422 8423:Architectures: x86 SEV enabled 8424:Type: vm 8425:Parameters: args[0] is the fd of the source vm 8426:Returns: 0 on success 8427 8428This capability enables userspace to migrate the encryption context from the VM 8429indicated by the fd to the VM this is called on. 8430 8431This is intended to support intra-host migration of VMs between userspace VMMs, 8432upgrading the VMM process without interrupting the guest. 8433 84347.31 KVM_CAP_DISABLE_QUIRKS2 8435---------------------------- 8436 8437:Parameters: args[0] - set of KVM quirks to disable 8438:Architectures: x86 8439:Type: vm 8440 8441This capability, if enabled, will cause KVM to disable some behavior 8442quirks. 8443 8444Calling KVM_CHECK_EXTENSION for this capability returns a bitmask of 8445quirks that can be disabled in KVM. 8446 8447The argument to KVM_ENABLE_CAP for this capability is a bitmask of 8448quirks to disable, and must be a subset of the bitmask returned by 8449KVM_CHECK_EXTENSION. 8450 8451The valid bits in cap.args[0] are: 8452 8453======================================== ================================================ 8454KVM_X86_QUIRK_LINT0_REENABLED By default, the reset value for the LVT 8455 LINT0 register is 0x700 (APIC_MODE_EXTINT). 8456 When this quirk is disabled, the reset value 8457 is 0x10000 (APIC_LVT_MASKED). 8458 8459KVM_X86_QUIRK_CD_NW_CLEARED By default, KVM clears CR0.CD and CR0.NW on 8460 AMD CPUs to workaround buggy guest firmware 8461 that runs in perpetuity with CR0.CD, i.e. 8462 with caches in "no fill" mode. 8463 8464 When this quirk is disabled, KVM does not 8465 change the value of CR0.CD and CR0.NW. 8466 8467KVM_X86_QUIRK_LAPIC_MMIO_HOLE By default, the MMIO LAPIC interface is 8468 available even when configured for x2APIC 8469 mode. When this quirk is disabled, KVM 8470 disables the MMIO LAPIC interface if the 8471 LAPIC is in x2APIC mode. 8472 8473KVM_X86_QUIRK_OUT_7E_INC_RIP By default, KVM pre-increments %rip before 8474 exiting to userspace for an OUT instruction 8475 to port 0x7e. When this quirk is disabled, 8476 KVM does not pre-increment %rip before 8477 exiting to userspace. 8478 8479KVM_X86_QUIRK_MISC_ENABLE_NO_MWAIT When this quirk is disabled, KVM sets 8480 CPUID.01H:ECX[bit 3] (MONITOR/MWAIT) if 8481 IA32_MISC_ENABLE[bit 18] (MWAIT) is set. 8482 Additionally, when this quirk is disabled, 8483 KVM clears CPUID.01H:ECX[bit 3] if 8484 IA32_MISC_ENABLE[bit 18] is cleared. 8485 8486KVM_X86_QUIRK_FIX_HYPERCALL_INSN By default, KVM rewrites guest 8487 VMMCALL/VMCALL instructions to match the 8488 vendor's hypercall instruction for the 8489 system. When this quirk is disabled, KVM 8490 will no longer rewrite invalid guest 8491 hypercall instructions. Executing the 8492 incorrect hypercall instruction will 8493 generate a #UD within the guest. 8494 8495KVM_X86_QUIRK_MWAIT_NEVER_UD_FAULTS By default, KVM emulates MONITOR/MWAIT (if 8496 they are intercepted) as NOPs regardless of 8497 whether or not MONITOR/MWAIT are supported 8498 according to guest CPUID. When this quirk 8499 is disabled and KVM_X86_DISABLE_EXITS_MWAIT 8500 is not set (MONITOR/MWAIT are intercepted), 8501 KVM will inject a #UD on MONITOR/MWAIT if 8502 they're unsupported per guest CPUID. Note, 8503 KVM will modify MONITOR/MWAIT support in 8504 guest CPUID on writes to MISC_ENABLE if 8505 KVM_X86_QUIRK_MISC_ENABLE_NO_MWAIT is 8506 disabled. 8507 8508KVM_X86_QUIRK_SLOT_ZAP_ALL By default, for KVM_X86_DEFAULT_VM VMs, KVM 8509 invalidates all SPTEs in all memslots and 8510 address spaces when a memslot is deleted or 8511 moved. When this quirk is disabled (or the 8512 VM type isn't KVM_X86_DEFAULT_VM), KVM only 8513 ensures the backing memory of the deleted 8514 or moved memslot isn't reachable, i.e KVM 8515 _may_ invalidate only SPTEs related to the 8516 memslot. 8517 8518KVM_X86_QUIRK_STUFF_FEATURE_MSRS By default, at vCPU creation, KVM sets the 8519 vCPU's MSR_IA32_PERF_CAPABILITIES (0x345), 8520 MSR_IA32_ARCH_CAPABILITIES (0x10a), 8521 MSR_PLATFORM_INFO (0xce), and all VMX MSRs 8522 (0x480..0x492) to the maximal capabilities 8523 supported by KVM. KVM also sets 8524 MSR_IA32_UCODE_REV (0x8b) to an arbitrary 8525 value (which is different for Intel vs. 8526 AMD). Lastly, when guest CPUID is set (by 8527 userspace), KVM modifies select VMX MSR 8528 fields to force consistency between guest 8529 CPUID and L2's effective ISA. When this 8530 quirk is disabled, KVM zeroes the vCPU's MSR 8531 values (with two exceptions, see below), 8532 i.e. treats the feature MSRs like CPUID 8533 leaves and gives userspace full control of 8534 the vCPU model definition. This quirk does 8535 not affect VMX MSRs CR0/CR4_FIXED1 (0x487 8536 and 0x489), as KVM does now allow them to 8537 be set by userspace (KVM sets them based on 8538 guest CPUID, for safety purposes). 8539 8540KVM_X86_QUIRK_IGNORE_GUEST_PAT By default, on Intel platforms, KVM ignores 8541 guest PAT and forces the effective memory 8542 type to WB in EPT. The quirk is not available 8543 on Intel platforms which are incapable of 8544 safely honoring guest PAT (i.e., without CPU 8545 self-snoop, KVM always ignores guest PAT and 8546 forces effective memory type to WB). It is 8547 also ignored on AMD platforms or, on Intel, 8548 when a VM has non-coherent DMA devices 8549 assigned; KVM always honors guest PAT in 8550 such case. The quirk is needed to avoid 8551 slowdowns on certain Intel Xeon platforms 8552 (e.g. ICX, SPR) where self-snoop feature is 8553 supported but UC is slow enough to cause 8554 issues with some older guests that use 8555 UC instead of WC to map the video RAM. 8556 Userspace can disable the quirk to honor 8557 guest PAT if it knows that there is no such 8558 guest software, for example if it does not 8559 expose a bochs graphics device (which is 8560 known to have had a buggy driver). 8561 8562KVM_X86_QUIRK_VMCS12_ALLOW_FREEZE_IN_SMM By default, KVM relaxes the consistency 8563 check for GUEST_IA32_DEBUGCTL in vmcs12 8564 to allow FREEZE_IN_SMM to be set. When 8565 this quirk is disabled, KVM requires this 8566 bit to be cleared. Note that the vmcs02 8567 bit is still completely controlled by the 8568 host, regardless of the quirk setting. 8569 8570KVM_X86_QUIRK_NESTED_SVM_SHARED_PAT By default, KVM for nested SVM guests 8571 shares the IA32_PAT MSR between L1 and 8572 L2. This is legacy behavior and does 8573 not match the AMD architecture 8574 specification. When this quirk is 8575 disabled and nested paging (NPT) is 8576 enabled for L2, KVM correctly 8577 virtualizes a separate guest PAT 8578 register for L2, using the g_pat 8579 field in the VMCB. When NPT is 8580 disabled for L2, L1 and L2 continue 8581 to share the IA32_PAT MSR regardless 8582 of the quirk setting. 8583======================================== ================================================ 8584 85857.32 KVM_CAP_MAX_VCPU_ID 8586------------------------ 8587 8588:Architectures: x86 8589:Target: VM 8590:Parameters: args[0] - maximum APIC ID value set for current VM 8591:Returns: 0 on success, -EINVAL if args[0] is beyond KVM_MAX_VCPU_IDS 8592 supported in KVM or if it has been set. 8593 8594This capability allows userspace to specify maximum possible APIC ID 8595assigned for current VM session prior to the creation of vCPUs, saving 8596memory for data structures indexed by the APIC ID. Userspace is able 8597to calculate the limit to APIC ID values from designated 8598CPU topology. 8599 8600The value can be changed only until KVM_ENABLE_CAP is set to a nonzero 8601value or until a vCPU is created. Upon creation of the first vCPU, 8602if the value was set to zero or KVM_ENABLE_CAP was not invoked, KVM 8603uses the return value of KVM_CHECK_EXTENSION(KVM_CAP_MAX_VCPU_ID) as 8604the maximum APIC ID. 8605 86067.33 KVM_CAP_X86_NOTIFY_VMEXIT 8607------------------------------ 8608 8609:Architectures: x86 8610:Target: VM 8611:Parameters: args[0] is the value of notify window as well as some flags 8612:Returns: 0 on success, -EINVAL if args[0] contains invalid flags or notify 8613 VM exit is unsupported. 8614 8615Bits 63:32 of args[0] are used for notify window. 8616Bits 31:0 of args[0] are for some flags. Valid bits are:: 8617 8618 #define KVM_X86_NOTIFY_VMEXIT_ENABLED (1 << 0) 8619 #define KVM_X86_NOTIFY_VMEXIT_USER (1 << 1) 8620 8621This capability allows userspace to configure the notify VM exit on/off 8622in per-VM scope during VM creation. Notify VM exit is disabled by default. 8623When userspace sets KVM_X86_NOTIFY_VMEXIT_ENABLED bit in args[0], VMM will 8624enable this feature with the notify window provided, which will generate 8625a VM exit if no event window occurs in VM non-root mode for a specified of 8626time (notify window). 8627 8628If KVM_X86_NOTIFY_VMEXIT_USER is set in args[0], upon notify VM exits happen, 8629KVM would exit to userspace for handling. 8630 8631This capability is aimed to mitigate the threat that malicious VMs can 8632cause CPU stuck (due to event windows don't open up) and make the CPU 8633unavailable to host or other VMs. 8634 86357.35 KVM_CAP_X86_APIC_BUS_CYCLES_NS 8636----------------------------------- 8637 8638:Architectures: x86 8639:Target: VM 8640:Parameters: args[0] is the desired APIC bus clock rate, in nanoseconds 8641:Returns: 0 on success, -EINVAL if args[0] contains an invalid value for the 8642 frequency or if any vCPUs have been created, -ENXIO if a virtual 8643 local APIC has not been created using KVM_CREATE_IRQCHIP. 8644 8645This capability sets the VM's APIC bus clock frequency, used by KVM's in-kernel 8646virtual APIC when emulating APIC timers. KVM's default value can be retrieved 8647by KVM_CHECK_EXTENSION. 8648 8649Note: Userspace is responsible for correctly configuring CPUID 0x15, a.k.a. the 8650core crystal clock frequency, if a non-zero CPUID 0x15 is exposed to the guest. 8651 86527.36 KVM_CAP_DIRTY_LOG_RING/KVM_CAP_DIRTY_LOG_RING_ACQ_REL 8653---------------------------------------------------------- 8654 8655:Architectures: x86, arm64, riscv 8656:Type: vm 8657:Parameters: args[0] - size of the dirty log ring 8658 8659KVM is capable of tracking dirty memory using ring buffers that are 8660mmapped into userspace; there is one dirty ring per vcpu. 8661 8662The dirty ring is available to userspace as an array of 8663``struct kvm_dirty_gfn``. Each dirty entry is defined as:: 8664 8665 struct kvm_dirty_gfn { 8666 __u32 flags; 8667 __u32 slot; /* as_id | slot_id */ 8668 __u64 offset; 8669 }; 8670 8671The following values are defined for the flags field to define the 8672current state of the entry:: 8673 8674 #define KVM_DIRTY_GFN_F_DIRTY BIT(0) 8675 #define KVM_DIRTY_GFN_F_RESET BIT(1) 8676 #define KVM_DIRTY_GFN_F_MASK 0x3 8677 8678Userspace should call KVM_ENABLE_CAP ioctl right after KVM_CREATE_VM 8679ioctl to enable this capability for the new guest and set the size of 8680the rings. Enabling the capability is only allowed before creating any 8681vCPU, and the size of the ring must be a power of two. The larger the 8682ring buffer, the less likely the ring is full and the VM is forced to 8683exit to userspace. The optimal size depends on the workload, but it is 8684recommended that it be at least 64 KiB (4096 entries). 8685 8686Just like for dirty page bitmaps, the buffer tracks writes to 8687all user memory regions for which the KVM_MEM_LOG_DIRTY_PAGES flag was 8688set in KVM_SET_USER_MEMORY_REGION. Once a memory region is registered 8689with the flag set, userspace can start harvesting dirty pages from the 8690ring buffer. 8691 8692An entry in the ring buffer can be unused (flag bits ``00``), 8693dirty (flag bits ``01``) or harvested (flag bits ``1X``). The 8694state machine for the entry is as follows:: 8695 8696 dirtied harvested reset 8697 00 -----------> 01 -------------> 1X -------+ 8698 ^ | 8699 | | 8700 +------------------------------------------+ 8701 8702To harvest the dirty pages, userspace accesses the mmapped ring buffer 8703to read the dirty GFNs. If the flags has the DIRTY bit set (at this stage 8704the RESET bit must be cleared), then it means this GFN is a dirty GFN. 8705The userspace should harvest this GFN and mark the flags from state 8706``01b`` to ``1Xb`` (bit 0 will be ignored by KVM, but bit 1 must be set 8707to show that this GFN is harvested and waiting for a reset), and move 8708on to the next GFN. The userspace should continue to do this until the 8709flags of a GFN have the DIRTY bit cleared, meaning that it has harvested 8710all the dirty GFNs that were available. 8711 8712Note that on weakly ordered architectures, userspace accesses to the 8713ring buffer (and more specifically the 'flags' field) must be ordered, 8714using load-acquire/store-release accessors when available, or any 8715other memory barrier that will ensure this ordering. 8716 8717It's not necessary for userspace to harvest the all dirty GFNs at once. 8718However it must collect the dirty GFNs in sequence, i.e., the userspace 8719program cannot skip one dirty GFN to collect the one next to it. 8720 8721After processing one or more entries in the ring buffer, userspace 8722calls the VM ioctl KVM_RESET_DIRTY_RINGS to notify the kernel about 8723it, so that the kernel will reprotect those collected GFNs. 8724Therefore, the ioctl must be called *before* reading the content of 8725the dirty pages. 8726 8727The dirty ring can get full. When it happens, the KVM_RUN of the 8728vcpu will return with exit reason KVM_EXIT_DIRTY_RING_FULL. 8729 8730The dirty ring interface has a major difference comparing to the 8731KVM_GET_DIRTY_LOG interface in that, when reading the dirty ring from 8732userspace, it's still possible that the kernel has not yet flushed the 8733processor's dirty page buffers into the kernel buffer (with dirty bitmaps, the 8734flushing is done by the KVM_GET_DIRTY_LOG ioctl). To achieve that, one 8735needs to kick the vcpu out of KVM_RUN using a signal. The resulting 8736vmexit ensures that all dirty GFNs are flushed to the dirty rings. 8737 8738NOTE: KVM_CAP_DIRTY_LOG_RING_ACQ_REL is the only capability that 8739should be exposed by weakly ordered architecture, in order to indicate 8740the additional memory ordering requirements imposed on userspace when 8741reading the state of an entry and mutating it from DIRTY to HARVESTED. 8742Architecture with TSO-like ordering (such as x86) are allowed to 8743expose both KVM_CAP_DIRTY_LOG_RING and KVM_CAP_DIRTY_LOG_RING_ACQ_REL 8744to userspace. 8745 8746After enabling the dirty rings, the userspace needs to detect the 8747capability of KVM_CAP_DIRTY_LOG_RING_WITH_BITMAP to see whether the 8748ring structures can be backed by per-slot bitmaps. With this capability 8749advertised, it means the architecture can dirty guest pages without 8750vcpu/ring context, so that some of the dirty information will still be 8751maintained in the bitmap structure. KVM_CAP_DIRTY_LOG_RING_WITH_BITMAP 8752can't be enabled if the capability of KVM_CAP_DIRTY_LOG_RING_ACQ_REL 8753hasn't been enabled, or any memslot has been existing. 8754 8755Note that the bitmap here is only a backup of the ring structure. The 8756use of the ring and bitmap combination is only beneficial if there is 8757only a very small amount of memory that is dirtied out of vcpu/ring 8758context. Otherwise, the stand-alone per-slot bitmap mechanism needs to 8759be considered. 8760 8761To collect dirty bits in the backup bitmap, userspace can use the same 8762KVM_GET_DIRTY_LOG ioctl. KVM_CLEAR_DIRTY_LOG isn't needed as long as all 8763the generation of the dirty bits is done in a single pass. Collecting 8764the dirty bitmap should be the very last thing that the VMM does before 8765considering the state as complete. VMM needs to ensure that the dirty 8766state is final and avoid missing dirty pages from another ioctl ordered 8767after the bitmap collection. 8768 8769NOTE: Multiple examples of using the backup bitmap: (1) save vgic/its 8770tables through command KVM_DEV_ARM_{VGIC_GRP_CTRL, ITS_SAVE_TABLES} on 8771KVM device "kvm-arm-vgic-its". (2) restore vgic/its tables through 8772command KVM_DEV_ARM_{VGIC_GRP_CTRL, ITS_RESTORE_TABLES} on KVM device 8773"kvm-arm-vgic-its". VGICv3 LPI pending status is restored. (3) save 8774vgic3 pending table through KVM_DEV_ARM_VGIC_{GRP_CTRL, SAVE_PENDING_TABLES} 8775command on KVM device "kvm-arm-vgic-v3". 8776 87777.37 KVM_CAP_PMU_CAPABILITY 8778--------------------------- 8779 8780:Architectures: x86 8781:Type: vm 8782:Parameters: arg[0] is bitmask of PMU virtualization capabilities. 8783:Returns: 0 on success, -EINVAL when arg[0] contains invalid bits 8784 8785This capability alters PMU virtualization in KVM. 8786 8787Calling KVM_CHECK_EXTENSION for this capability returns a bitmask of 8788PMU virtualization capabilities that can be adjusted on a VM. 8789 8790The argument to KVM_ENABLE_CAP is also a bitmask and selects specific 8791PMU virtualization capabilities to be applied to the VM. This can 8792only be invoked on a VM prior to the creation of VCPUs. 8793 8794At this time, KVM_PMU_CAP_DISABLE is the only capability. Setting 8795this capability will disable PMU virtualization for that VM. Usermode 8796should adjust CPUID leaf 0xA to reflect that the PMU is disabled. 8797 87987.38 KVM_CAP_VM_DISABLE_NX_HUGE_PAGES 8799------------------------------------- 8800 8801:Architectures: x86 8802:Type: vm 8803:Parameters: arg[0] must be 0. 8804:Returns: 0 on success, -EPERM if the userspace process does not 8805 have CAP_SYS_BOOT, -EINVAL if args[0] is not 0 or any vCPUs have been 8806 created. 8807 8808This capability disables the NX huge pages mitigation for iTLB MULTIHIT. 8809 8810The capability has no effect if the nx_huge_pages module parameter is not set. 8811 8812This capability may only be set before any vCPUs are created. 8813 88147.39 KVM_CAP_ARM_EAGER_SPLIT_CHUNK_SIZE 8815--------------------------------------- 8816 8817:Architectures: arm64 8818:Type: vm 8819:Parameters: arg[0] is the new split chunk size. 8820:Returns: 0 on success, -EINVAL if any memslot was already created. 8821 8822This capability sets the chunk size used in Eager Page Splitting. 8823 8824Eager Page Splitting improves the performance of dirty-logging (used 8825in live migrations) when guest memory is backed by huge-pages. It 8826avoids splitting huge-pages (into PAGE_SIZE pages) on fault, by doing 8827it eagerly when enabling dirty logging (with the 8828KVM_MEM_LOG_DIRTY_PAGES flag for a memory region), or when using 8829KVM_CLEAR_DIRTY_LOG. 8830 8831The chunk size specifies how many pages to break at a time, using a 8832single allocation for each chunk. Bigger the chunk size, more pages 8833need to be allocated ahead of time. 8834 8835The chunk size needs to be a valid block size. The list of acceptable 8836block sizes is exposed in KVM_CAP_ARM_SUPPORTED_BLOCK_SIZES as a 883764-bit bitmap (each bit describing a block size). The default value is 88380, to disable the eager page splitting. 8839 88407.40 KVM_CAP_EXIT_HYPERCALL 8841--------------------------- 8842 8843:Architectures: x86 8844:Type: vm 8845 8846This capability, if enabled, will cause KVM to exit to userspace 8847with KVM_EXIT_HYPERCALL exit reason to process some hypercalls. 8848Userspace may fail the hypercall by setting hypercall.ret to EINVAL 8849or may request the hypercall to be retried the next time the guest run 8850by setting hypercall.ret to EAGAIN. 8851 8852Calling KVM_CHECK_EXTENSION for this capability will return a bitmask 8853of hypercalls that can be configured to exit to userspace. 8854Right now, the only such hypercall is KVM_HC_MAP_GPA_RANGE. 8855 8856The argument to KVM_ENABLE_CAP is also a bitmask, and must be a subset 8857of the result of KVM_CHECK_EXTENSION. KVM will forward to userspace 8858the hypercalls whose corresponding bit is in the argument, and return 8859ENOSYS for the others. 8860 88617.41 KVM_CAP_ARM_SYSTEM_SUSPEND 8862------------------------------- 8863 8864:Architectures: arm64 8865:Type: vm 8866 8867When enabled, KVM will exit to userspace with KVM_EXIT_SYSTEM_EVENT of 8868type KVM_SYSTEM_EVENT_SUSPEND to process the guest suspend request. 8869 88707.42 KVM_CAP_ARM_WRITABLE_IMP_ID_REGS 8871------------------------------------- 8872 8873:Architectures: arm64 8874:Target: VM 8875:Parameters: None 8876:Returns: 0 on success, -EINVAL if vCPUs have been created before enabling this 8877 capability. 8878 8879This capability changes the behavior of the registers that identify a PE 8880implementation of the Arm architecture: MIDR_EL1, REVIDR_EL1, and AIDR_EL1. 8881By default, these registers are visible to userspace but treated as invariant. 8882 8883When this capability is enabled, KVM allows userspace to change the 8884aforementioned registers before the first KVM_RUN. These registers are VM 8885scoped, meaning that the same set of values are presented on all vCPUs in a 8886given VM. 8887 88887.43 KVM_CAP_RISCV_MP_STATE_RESET 8889--------------------------------- 8890 8891:Architectures: riscv 8892:Type: VM 8893:Parameters: None 8894:Returns: 0 on success, -EINVAL if arg[0] is not zero 8895 8896When this capability is enabled, KVM resets the VCPU when setting 8897MP_STATE_INIT_RECEIVED through IOCTL. The original MP_STATE is preserved. 8898 88997.44 KVM_CAP_ARM_CACHEABLE_PFNMAP_SUPPORTED 8900------------------------------------------- 8901 8902:Architectures: arm64 8903:Target: VM 8904:Parameters: None 8905 8906This capability indicate to the userspace whether a PFNMAP memory region 8907can be safely mapped as cacheable. This relies on the presence of 8908force write back (FWB) feature support on the hardware. 8909 89107.45 KVM_CAP_ARM_SEA_TO_USER 8911---------------------------- 8912 8913:Architecture: arm64 8914:Target: VM 8915:Parameters: none 8916:Returns: 0 on success, -EINVAL if unsupported. 8917 8918When this capability is enabled, KVM may exit to userspace for SEAs taken to 8919EL2 resulting from a guest access. See ``KVM_EXIT_ARM_SEA`` for more 8920information. 8921 89227.46 KVM_CAP_S390_USER_OPEREXEC 8923------------------------------- 8924 8925:Architectures: s390 8926:Parameters: none 8927 8928When this capability is enabled KVM forwards all operation exceptions 8929that it doesn't handle itself to user space. This also includes the 89300x0000 instructions managed by KVM_CAP_S390_USER_INSTR0. This is 8931helpful if user space wants to emulate instructions which are not 8932(yet) implemented in hardware. 8933 8934This capability can be enabled dynamically even if VCPUs were already 8935created and are running. 8936 89377.47 KVM_CAP_S390_HPAGE_2G 8938-------------------------- 8939 8940:Architectures: s390 8941:Parameters: none 8942:Returns: 0 on success; -EINVAL if hpage_2g module parameter was not set, 8943 cmma is enabled, or the VM has the KVM_VM_S390_UCONTROL 8944 flag set; -EBUSY if vCPUs were already created for the VM. 8945 8946With this capability the KVM support for memory backing with 2g pages 8947through hugetlbfs can be enabled for a VM. After the capability is 8948enabled, cmma can't be enabled anymore and pfmfi and the storage key 8949interpretation are disabled. If cmma has already been enabled or the 8950hpage_2g module parameter is not set to 1, -EINVAL is returned. 8951 89528. Other capabilities. 8953====================== 8954 8955This section lists capabilities that give information about other 8956features of the KVM implementation. 8957 89588.1 KVM_CAP_PPC_HWRNG 8959--------------------- 8960 8961:Architectures: ppc 8962 8963This capability, if KVM_CHECK_EXTENSION indicates that it is 8964available, means that the kernel has an implementation of the 8965H_RANDOM hypercall backed by a hardware random-number generator. 8966If present, the kernel H_RANDOM handler can be enabled for guest use 8967with the KVM_CAP_PPC_ENABLE_HCALL capability. 8968 89698.3 KVM_CAP_PPC_MMU_RADIX 8970------------------------- 8971 8972:Architectures: ppc 8973 8974This capability, if KVM_CHECK_EXTENSION indicates that it is 8975available, means that the kernel can support guests using the 8976radix MMU defined in Power ISA V3.00 (as implemented in the POWER9 8977processor). 8978 89798.4 KVM_CAP_PPC_MMU_HASH_V3 8980--------------------------- 8981 8982:Architectures: ppc 8983 8984This capability, if KVM_CHECK_EXTENSION indicates that it is 8985available, means that the kernel can support guests using the 8986hashed page table MMU defined in Power ISA V3.00 (as implemented in 8987the POWER9 processor), including in-memory segment tables. 8988 89898.5 KVM_CAP_MIPS_VZ 8990------------------- 8991 8992:Architectures: mips 8993 8994This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that 8995it is available, means that full hardware assisted virtualization capabilities 8996of the hardware are available for use through KVM. An appropriate 8997KVM_VM_MIPS_* type must be passed to KVM_CREATE_VM to create a VM which 8998utilises it. 8999 9000If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is 9001available, it means that the VM is using full hardware assisted virtualization 9002capabilities of the hardware. This is useful to check after creating a VM with 9003KVM_VM_MIPS_DEFAULT. 9004 9005The value returned by KVM_CHECK_EXTENSION should be compared against known 9006values (see below). All other values are reserved. This is to allow for the 9007possibility of other hardware assisted virtualization implementations which 9008may be incompatible with the MIPS VZ ASE. 9009 9010== ========================================================================== 9011 0 The trap & emulate implementation is in use to run guest code in user 9012 mode. Guest virtual memory segments are rearranged to fit the guest in the 9013 user mode address space. 9014 9015 1 The MIPS VZ ASE is in use, providing full hardware assisted 9016 virtualization, including standard guest virtual memory segments. 9017== ========================================================================== 9018 90198.7 KVM_CAP_MIPS_64BIT 9020---------------------- 9021 9022:Architectures: mips 9023 9024This capability indicates the supported architecture type of the guest, i.e. the 9025supported register and address width. 9026 9027The values returned when this capability is checked by KVM_CHECK_EXTENSION on a 9028kvm VM handle correspond roughly to the CP0_Config.AT register field, and should 9029be checked specifically against known values (see below). All other values are 9030reserved. 9031 9032== ======================================================================== 9033 0 MIPS32 or microMIPS32. 9034 Both registers and addresses are 32-bits wide. 9035 It will only be possible to run 32-bit guest code. 9036 9037 1 MIPS64 or microMIPS64 with access only to 32-bit compatibility segments. 9038 Registers are 64-bits wide, but addresses are 32-bits wide. 9039 64-bit guest code may run but cannot access MIPS64 memory segments. 9040 It will also be possible to run 32-bit guest code. 9041 9042 2 MIPS64 or microMIPS64 with access to all address segments. 9043 Both registers and addresses are 64-bits wide. 9044 It will be possible to run 64-bit or 32-bit guest code. 9045== ======================================================================== 9046 90478.9 KVM_CAP_ARM_USER_IRQ 9048------------------------ 9049 9050:Architectures: arm64 9051 9052This capability, if KVM_CHECK_EXTENSION indicates that it is available, means 9053that if userspace creates a VM without an in-kernel interrupt controller, it 9054will be notified of changes to the output level of in-kernel emulated devices, 9055which can generate virtual interrupts, presented to the VM. 9056For such VMs, on every return to userspace, the kernel 9057updates the vcpu's run->s.regs.device_irq_level field to represent the actual 9058output level of the device. 9059 9060Whenever kvm detects a change in the device output level, kvm guarantees at 9061least one return to userspace before running the VM. This exit could either 9062be a KVM_EXIT_INTR or any other exit event, like KVM_EXIT_MMIO. This way, 9063userspace can always sample the device output level and re-compute the state of 9064the userspace interrupt controller. Userspace should always check the state 9065of run->s.regs.device_irq_level on every kvm exit. 9066The value in run->s.regs.device_irq_level can represent both level and edge 9067triggered interrupt signals, depending on the device. Edge triggered interrupt 9068signals will exit to userspace with the bit in run->s.regs.device_irq_level 9069set exactly once per edge signal. 9070 9071The field run->s.regs.device_irq_level is available independent of 9072run->kvm_valid_regs or run->kvm_dirty_regs bits. 9073 9074If KVM_CAP_ARM_USER_IRQ is supported, the KVM_CHECK_EXTENSION ioctl returns a 9075number larger than 0 indicating the version of this capability is implemented 9076and thereby which bits in run->s.regs.device_irq_level can signal values. 9077 9078Currently the following bits are defined for the device_irq_level bitmap:: 9079 9080 KVM_CAP_ARM_USER_IRQ >= 1: 9081 9082 KVM_ARM_DEV_EL1_VTIMER - EL1 virtual timer 9083 KVM_ARM_DEV_EL1_PTIMER - EL1 physical timer 9084 KVM_ARM_DEV_PMU - ARM PMU overflow interrupt signal 9085 9086Future versions of kvm may implement additional events. These will get 9087indicated by returning a higher number from KVM_CHECK_EXTENSION and will be 9088listed above. 9089 90908.10 KVM_CAP_PPC_SMT_POSSIBLE 9091----------------------------- 9092 9093:Architectures: ppc 9094 9095Querying this capability returns a bitmap indicating the possible 9096virtual SMT modes that can be set using KVM_CAP_PPC_SMT. If bit N 9097(counting from the right) is set, then a virtual SMT mode of 2^N is 9098available. 9099 91008.12 KVM_CAP_HYPERV_VP_INDEX 9101---------------------------- 9102 9103:Architectures: x86 9104 9105This capability indicates that userspace can load HV_X64_MSR_VP_INDEX msr. Its 9106value is used to denote the target vcpu for a SynIC interrupt. For 9107compatibility, KVM initializes this msr to KVM's internal vcpu index. When this 9108capability is absent, userspace can still query this msr's value. 9109 91108.13 KVM_CAP_S390_AIS_MIGRATION 9111------------------------------- 9112 9113:Architectures: s390 9114 9115This capability indicates if the flic device will be able to get/set the 9116AIS states for migration via the KVM_DEV_FLIC_AISM_ALL attribute and allows 9117to discover this without having to create a flic device. 9118 91198.14 KVM_CAP_S390_PSW 9120--------------------- 9121 9122:Architectures: s390 9123 9124This capability indicates that the PSW is exposed via the kvm_run structure. 9125 91268.15 KVM_CAP_S390_GMAP 9127---------------------- 9128 9129:Architectures: s390 9130 9131This capability indicates that the user space memory used as guest mapping can 9132be anywhere in the user memory address space, as long as the memory slots are 9133aligned and sized to a segment (1MB) boundary. 9134 91358.16 KVM_CAP_S390_COW 9136--------------------- 9137 9138:Architectures: s390 9139 9140This capability indicates that the user space memory used as guest mapping can 9141use copy-on-write semantics as well as dirty pages tracking via read-only page 9142tables. 9143 91448.17 KVM_CAP_S390_BPB 9145--------------------- 9146 9147:Architectures: s390 9148 9149This capability indicates that kvm will implement the interfaces to handle 9150reset, migration and nested KVM for branch prediction blocking. The stfle 9151facility 82 should not be provided to the guest without this capability. 9152 91538.18 KVM_CAP_HYPERV_TLBFLUSH 9154---------------------------- 9155 9156:Architectures: x86 9157 9158This capability indicates that KVM supports paravirtualized Hyper-V TLB Flush 9159hypercalls: 9160HvFlushVirtualAddressSpace, HvFlushVirtualAddressSpaceEx, 9161HvFlushVirtualAddressList, HvFlushVirtualAddressListEx. 9162 91638.19 KVM_CAP_ARM_INJECT_SERROR_ESR 9164---------------------------------- 9165 9166:Architectures: arm64 9167 9168This capability indicates that userspace can specify (via the 9169KVM_SET_VCPU_EVENTS ioctl) the syndrome value reported to the guest when it 9170takes a virtual SError interrupt exception. 9171If KVM advertises this capability, userspace can only specify the ISS field for 9172the ESR syndrome. Other parts of the ESR, such as the EC are generated by the 9173CPU when the exception is taken. If this virtual SError is taken to EL1 using 9174AArch64, this value will be reported in the ISS field of ESR_ELx. 9175 9176See KVM_CAP_VCPU_EVENTS for more details. 9177 91788.20 KVM_CAP_HYPERV_SEND_IPI 9179---------------------------- 9180 9181:Architectures: x86 9182 9183This capability indicates that KVM supports paravirtualized Hyper-V IPI send 9184hypercalls: 9185HvCallSendSyntheticClusterIpi, HvCallSendSyntheticClusterIpiEx. 9186 91878.22 KVM_CAP_S390_VCPU_RESETS 9188----------------------------- 9189 9190:Architectures: s390 9191 9192This capability indicates that the KVM_S390_NORMAL_RESET and 9193KVM_S390_CLEAR_RESET ioctls are available. 9194 91958.23 KVM_CAP_S390_PROTECTED 9196--------------------------- 9197 9198:Architectures: s390 9199 9200This capability indicates that the Ultravisor has been initialized and 9201KVM can therefore start protected VMs. 9202This capability governs the KVM_S390_PV_COMMAND ioctl and the 9203KVM_MP_STATE_LOAD MP_STATE. KVM_SET_MP_STATE can fail for protected 9204guests when the state change is invalid. 9205 92068.24 KVM_CAP_STEAL_TIME 9207----------------------- 9208 9209:Architectures: arm64, x86 9210 9211This capability indicates that KVM supports steal time accounting. 9212When steal time accounting is supported it may be enabled with 9213architecture-specific interfaces. This capability and the architecture- 9214specific interfaces must be consistent, i.e. if one says the feature 9215is supported, than the other should as well and vice versa. For arm64 9216see Documentation/virt/kvm/devices/vcpu.rst "KVM_ARM_VCPU_PVTIME_CTRL". 9217For x86 see Documentation/virt/kvm/x86/msr.rst "MSR_KVM_STEAL_TIME". 9218 92198.25 KVM_CAP_S390_DIAG318 9220------------------------- 9221 9222:Architectures: s390 9223 9224This capability enables a guest to set information about its control program 9225(i.e. guest kernel type and version). The information is helpful during 9226system/firmware service events, providing additional data about the guest 9227environments running on the machine. 9228 9229The information is associated with the DIAGNOSE 0x318 instruction, which sets 9230an 8-byte value consisting of a one-byte Control Program Name Code (CPNC) and 9231a 7-byte Control Program Version Code (CPVC). The CPNC determines what 9232environment the control program is running in (e.g. Linux, z/VM...), and the 9233CPVC is used for information specific to OS (e.g. Linux version, Linux 9234distribution...) 9235 9236If this capability is available, then the CPNC and CPVC can be synchronized 9237between KVM and userspace via the sync regs mechanism (KVM_SYNC_DIAG318). 9238 92398.26 KVM_CAP_X86_USER_SPACE_MSR 9240------------------------------- 9241 9242:Architectures: x86 9243 9244This capability indicates that KVM supports deflection of MSR reads and 9245writes to user space. It can be enabled on a VM level. If enabled, MSR 9246accesses that would usually trigger a #GP by KVM into the guest will 9247instead get bounced to user space through the KVM_EXIT_X86_RDMSR and 9248KVM_EXIT_X86_WRMSR exit notifications. 9249 92508.27 KVM_CAP_X86_MSR_FILTER 9251--------------------------- 9252 9253:Architectures: x86 9254 9255This capability indicates that KVM supports that accesses to user defined MSRs 9256may be rejected. With this capability exposed, KVM exports new VM ioctl 9257KVM_X86_SET_MSR_FILTER which user space can call to specify bitmaps of MSR 9258ranges that KVM should deny access to. 9259 9260In combination with KVM_CAP_X86_USER_SPACE_MSR, this allows user space to 9261trap and emulate MSRs that are outside of the scope of KVM as well as 9262limit the attack surface on KVM's MSR emulation code. 9263 92648.30 KVM_CAP_XEN_HVM 9265-------------------- 9266 9267:Architectures: x86 9268 9269This capability indicates the features that Xen supports for hosting Xen 9270PVHVM guests. Valid flags are:: 9271 9272 #define KVM_XEN_HVM_CONFIG_HYPERCALL_MSR (1 << 0) 9273 #define KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL (1 << 1) 9274 #define KVM_XEN_HVM_CONFIG_SHARED_INFO (1 << 2) 9275 #define KVM_XEN_HVM_CONFIG_RUNSTATE (1 << 3) 9276 #define KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL (1 << 4) 9277 #define KVM_XEN_HVM_CONFIG_EVTCHN_SEND (1 << 5) 9278 #define KVM_XEN_HVM_CONFIG_RUNSTATE_UPDATE_FLAG (1 << 6) 9279 #define KVM_XEN_HVM_CONFIG_PVCLOCK_TSC_UNSTABLE (1 << 7) 9280 9281The KVM_XEN_HVM_CONFIG_HYPERCALL_MSR flag indicates that the KVM_XEN_HVM_CONFIG 9282ioctl is available, for the guest to set its hypercall page. 9283 9284If KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL is also set, the same flag may also be 9285provided in the flags to KVM_XEN_HVM_CONFIG, without providing hypercall page 9286contents, to request that KVM generate hypercall page content automatically 9287and also enable interception of guest hypercalls with KVM_EXIT_XEN. 9288 9289The KVM_XEN_HVM_CONFIG_SHARED_INFO flag indicates the availability of the 9290KVM_XEN_HVM_SET_ATTR, KVM_XEN_HVM_GET_ATTR, KVM_XEN_VCPU_SET_ATTR and 9291KVM_XEN_VCPU_GET_ATTR ioctls, as well as the delivery of exception vectors 9292for event channel upcalls when the evtchn_upcall_pending field of a vcpu's 9293vcpu_info is set. 9294 9295The KVM_XEN_HVM_CONFIG_RUNSTATE flag indicates that the runstate-related 9296features KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR/_CURRENT/_DATA/_ADJUST are 9297supported by the KVM_XEN_VCPU_SET_ATTR/KVM_XEN_VCPU_GET_ATTR ioctls. 9298 9299The KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL flag indicates that IRQ routing entries 9300of the type KVM_IRQ_ROUTING_XEN_EVTCHN are supported, with the priority 9301field set to indicate 2 level event channel delivery. 9302 9303The KVM_XEN_HVM_CONFIG_EVTCHN_SEND flag indicates that KVM supports 9304injecting event channel events directly into the guest with the 9305KVM_XEN_HVM_EVTCHN_SEND ioctl. It also indicates support for the 9306KVM_XEN_ATTR_TYPE_EVTCHN/XEN_VERSION HVM attributes and the 9307KVM_XEN_VCPU_ATTR_TYPE_VCPU_ID/TIMER/UPCALL_VECTOR vCPU attributes. 9308related to event channel delivery, timers, and the XENVER_version 9309interception. 9310 9311The KVM_XEN_HVM_CONFIG_RUNSTATE_UPDATE_FLAG flag indicates that KVM supports 9312the KVM_XEN_ATTR_TYPE_RUNSTATE_UPDATE_FLAG attribute in the KVM_XEN_SET_ATTR 9313and KVM_XEN_GET_ATTR ioctls. This controls whether KVM will set the 9314XEN_RUNSTATE_UPDATE flag in guest memory mapped vcpu_runstate_info during 9315updates of the runstate information. Note that versions of KVM which support 9316the RUNSTATE feature above, but not the RUNSTATE_UPDATE_FLAG feature, will 9317always set the XEN_RUNSTATE_UPDATE flag when updating the guest structure, 9318which is perhaps counterintuitive. When this flag is advertised, KVM will 9319behave more correctly, not using the XEN_RUNSTATE_UPDATE flag until/unless 9320specifically enabled (by the guest making the hypercall, causing the VMM 9321to enable the KVM_XEN_ATTR_TYPE_RUNSTATE_UPDATE_FLAG attribute). 9322 9323The KVM_XEN_HVM_CONFIG_PVCLOCK_TSC_UNSTABLE flag indicates that KVM supports 9324clearing the PVCLOCK_TSC_STABLE_BIT flag in Xen pvclock sources. This will be 9325done when the KVM_CAP_XEN_HVM ioctl sets the 9326KVM_XEN_HVM_CONFIG_PVCLOCK_TSC_UNSTABLE flag. 9327 93288.31 KVM_CAP_SPAPR_MULTITCE 9329--------------------------- 9330 9331:Architectures: ppc 9332:Type: vm 9333 9334This capability means the kernel is capable of handling hypercalls 9335H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user 9336space. This significantly accelerates DMA operations for PPC KVM guests. 9337User space should expect that its handlers for these hypercalls 9338are not going to be called if user space previously registered LIOBN 9339in KVM (via KVM_CREATE_SPAPR_TCE or similar calls). 9340 9341In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest, 9342user space might have to advertise it for the guest. For example, 9343IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is 9344present in the "ibm,hypertas-functions" device-tree property. 9345 9346The hypercalls mentioned above may or may not be processed successfully 9347in the kernel based fast path. If they can not be handled by the kernel, 9348they will get passed on to user space. So user space still has to have 9349an implementation for these despite the in kernel acceleration. 9350 9351This capability is always enabled. 9352 93538.32 KVM_CAP_PTP_KVM 9354-------------------- 9355 9356:Architectures: arm64 9357 9358This capability indicates that the KVM virtual PTP service is 9359supported in the host. A VMM can check whether the service is 9360available to the guest on migration. 9361 93628.37 KVM_CAP_S390_PROTECTED_DUMP 9363-------------------------------- 9364 9365:Architectures: s390 9366:Type: vm 9367 9368This capability indicates that KVM and the Ultravisor support dumping 9369PV guests. The `KVM_PV_DUMP` command is available for the 9370`KVM_S390_PV_COMMAND` ioctl and the `KVM_PV_INFO` command provides 9371dump related UV data. Also the vcpu ioctl `KVM_S390_PV_CPU_COMMAND` is 9372available and supports the `KVM_PV_DUMP_CPU` subcommand. 9373 93748.39 KVM_CAP_S390_CPU_TOPOLOGY 9375------------------------------ 9376 9377:Architectures: s390 9378:Type: vm 9379 9380This capability indicates that KVM will provide the S390 CPU Topology 9381facility which consist of the interpretation of the PTF instruction for 9382the function code 2 along with interception and forwarding of both the 9383PTF instruction with function codes 0 or 1 and the STSI(15,1,x) 9384instruction to the userland hypervisor. 9385 9386The stfle facility 11, CPU Topology facility, should not be indicated 9387to the guest without this capability. 9388 9389When this capability is present, KVM provides a new attribute group 9390on vm fd, KVM_S390_VM_CPU_TOPOLOGY. 9391This new attribute allows to get, set or clear the Modified Change 9392Topology Report (MTCR) bit of the SCA through the kvm_device_attr 9393structure. 9394 9395When getting the Modified Change Topology Report value, the attr->addr 9396must point to a byte where the value will be stored or retrieved from. 9397 93988.41 KVM_CAP_VM_TYPES 9399--------------------- 9400 9401:Architectures: x86 9402:Type: system ioctl 9403 9404This capability returns a bitmap of support VM types. The 1-setting of bit @n 9405means the VM type with value @n is supported. Possible values of @n are:: 9406 9407 #define KVM_X86_DEFAULT_VM 0 9408 #define KVM_X86_SW_PROTECTED_VM 1 9409 #define KVM_X86_SEV_VM 2 9410 #define KVM_X86_SEV_ES_VM 3 9411 #define KVM_X86_SNP_VM 4 9412 #define KVM_X86_TDX_VM 5 9413 9414Note, KVM_X86_SW_PROTECTED_VM is currently only for development and testing. 9415Do not use KVM_X86_SW_PROTECTED_VM for "real" VMs, and especially not in 9416production. The behavior and effective ABI for software-protected VMs is 9417unstable. 9418 94198.42 KVM_CAP_PPC_RPT_INVALIDATE 9420------------------------------- 9421 9422:Architectures: ppc 9423 9424This capability indicates that the kernel is capable of handling 9425H_RPT_INVALIDATE hcall. 9426 9427In order to enable the use of H_RPT_INVALIDATE in the guest, 9428user space might have to advertise it for the guest. For example, 9429IBM pSeries (sPAPR) guest starts using it if "hcall-rpt-invalidate" is 9430present in the "ibm,hypertas-functions" device-tree property. 9431 9432This capability is enabled for hypervisors on platforms like POWER9 9433that support radix MMU. 9434 94358.43 KVM_CAP_PPC_AIL_MODE_3 9436--------------------------- 9437 9438:Architectures: ppc 9439 9440This capability indicates that the kernel supports the mode 3 setting for the 9441"Address Translation Mode on Interrupt" aka "Alternate Interrupt Location" 9442resource that is controlled with the H_SET_MODE hypercall. 9443 9444This capability allows a guest kernel to use a better-performance mode for 9445handling interrupts and system calls. 9446 94478.44 KVM_CAP_MEMORY_FAULT_INFO 9448------------------------------ 9449 9450:Architectures: x86 9451 9452The presence of this capability indicates that KVM_RUN will fill 9453kvm_run.memory_fault if KVM cannot resolve a guest page fault VM-Exit, e.g. if 9454there is a valid memslot but no backing VMA for the corresponding host virtual 9455address. 9456 9457The information in kvm_run.memory_fault is valid if and only if KVM_RUN returns 9458an error with errno=EFAULT or errno=EHWPOISON *and* kvm_run.exit_reason is set 9459to KVM_EXIT_MEMORY_FAULT. 9460 9461Note: Userspaces which attempt to resolve memory faults so that they can retry 9462KVM_RUN are encouraged to guard against repeatedly receiving the same 9463error/annotated fault. 9464 9465See KVM_EXIT_MEMORY_FAULT for more information. 9466 94678.45 KVM_CAP_X86_GUEST_MODE 9468--------------------------- 9469 9470:Architectures: x86 9471 9472The presence of this capability indicates that KVM_RUN will update the 9473KVM_RUN_X86_GUEST_MODE bit in kvm_run.flags to indicate whether the 9474vCPU was executing nested guest code when it exited. 9475 94768.46 KVM_CAP_S390_KEYOP 9477----------------------- 9478 9479:Architectures: s390 9480 9481The presence of this capability indicates that the KVM_S390_KEYOP ioctl is 9482available. 9483 9484KVM exits with the register state of either the L1 or L2 guest 9485depending on which executed at the time of an exit. Userspace must 9486take care to differentiate between these cases. 9487 94888.47 KVM_CAP_S390_VSIE_ESAMODE 9489------------------------------ 9490 9491:Architectures: s390 9492 9493The presence of this capability indicates that the nested KVM guest can 9494start in ESA mode. 9495 94969. Known KVM API problems 9497========================= 9498 9499In some cases, KVM's API has some inconsistencies or common pitfalls 9500that userspace need to be aware of. This section details some of 9501these issues. 9502 9503Most of them are architecture specific, so the section is split by 9504architecture. 9505 95069.1. x86 9507-------- 9508 9509``KVM_GET_SUPPORTED_CPUID`` issues 9510^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9511 9512In general, ``KVM_GET_SUPPORTED_CPUID`` is designed so that it is possible 9513to take its result and pass it directly to ``KVM_SET_CPUID2``. This section 9514documents some cases in which that requires some care. 9515 9516Local APIC features 9517~~~~~~~~~~~~~~~~~~~ 9518 9519CPU[EAX=1]:ECX[21] (X2APIC) is reported by ``KVM_GET_SUPPORTED_CPUID``, 9520but it can only be enabled if ``KVM_CREATE_IRQCHIP`` or 9521``KVM_ENABLE_CAP(KVM_CAP_IRQCHIP_SPLIT)`` are used to enable in-kernel emulation of 9522the local APIC. 9523 9524The same is true for the ``KVM_FEATURE_PV_UNHALT`` paravirtualized feature. 9525 9526On older versions of Linux, CPU[EAX=1]:ECX[24] (TSC_DEADLINE) is not reported by 9527``KVM_GET_SUPPORTED_CPUID``, but it can be enabled if ``KVM_CAP_TSC_DEADLINE_TIMER`` 9528is present and the kernel has enabled in-kernel emulation of the local APIC. 9529On newer versions, ``KVM_GET_SUPPORTED_CPUID`` does report the bit as available. 9530 9531CPU topology 9532~~~~~~~~~~~~ 9533 9534Several CPUID values include topology information for the host CPU: 95350x0b and 0x1f for Intel systems, 0x8000001e for AMD systems. Different 9536versions of KVM return different values for this information and userspace 9537should not rely on it. Currently they return all zeroes. 9538 9539If userspace wishes to set up a guest topology, it should be careful that 9540the values of these three leaves differ for each CPU. In particular, 9541the APIC ID is found in EDX for all subleaves of 0x0b and 0x1f, and in EAX 9542for 0x8000001e; the latter also encodes the core id and node id in bits 95437:0 of EBX and ECX respectively. 9544 9545Obsolete ioctls and capabilities 9546^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9547 9548KVM_CAP_DISABLE_QUIRKS does not let userspace know which quirks are actually 9549available. Use ``KVM_CHECK_EXTENSION(KVM_CAP_DISABLE_QUIRKS2)`` instead if 9550available. 9551 9552Ordering of KVM_GET_*/KVM_SET_* ioctls 9553^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9554 9555TBD 9556