1.. SPDX-License-Identifier: GPL-2.0 2 3=================================================================== 4The Definitive KVM (Kernel-based Virtual Machine) API Documentation 5=================================================================== 6 71. General description 8====================== 9 10The kvm API is centered around different kinds of file descriptors 11and ioctls that can be issued to these file descriptors. An initial 12open("/dev/kvm") obtains a handle to the kvm subsystem; this handle 13can be used to issue system ioctls. A KVM_CREATE_VM ioctl on this 14handle will create a VM file descriptor which can be used to issue VM 15ioctls. A KVM_CREATE_VCPU or KVM_CREATE_DEVICE ioctl on a VM fd will 16create a virtual cpu or device and return a file descriptor pointing to 17the new resource. 18 19In other words, the kvm API is a set of ioctls that are issued to 20different kinds of file descriptor in order to control various aspects of 21a virtual machine. Depending on the file descriptor that accepts them, 22ioctls belong to the following classes: 23 24 - System ioctls: These query and set global attributes which affect the 25 whole kvm subsystem. In addition a system ioctl is used to create 26 virtual machines. 27 28 - VM ioctls: These query and set attributes that affect an entire virtual 29 machine, for example memory layout. In addition a VM ioctl is used to 30 create virtual cpus (vcpus) and devices. 31 32 VM ioctls must be issued from the same process (address space) that was 33 used to create the VM. 34 35 - vcpu ioctls: These query and set attributes that control the operation 36 of a single virtual cpu. 37 38 vcpu ioctls should be issued from the same thread that was used to create 39 the vcpu, except for asynchronous vcpu ioctl that are marked as such in 40 the documentation. Otherwise, the first ioctl after switching threads 41 could see a performance impact. 42 43 - device ioctls: These query and set attributes that control the operation 44 of a single device. 45 46 device ioctls must be issued from the same process (address space) that 47 was used to create the VM. 48 49While most ioctls are specific to one kind of file descriptor, in some 50cases the same ioctl can belong to more than one class. 51 52The KVM API grew over time. For this reason, KVM defines many constants 53of the form ``KVM_CAP_*``, each corresponding to a set of functionality 54provided by one or more ioctls. Availability of these "capabilities" can 55be checked with :ref:`KVM_CHECK_EXTENSION <KVM_CHECK_EXTENSION>`. Some 56capabilities also need to be enabled for VMs or VCPUs where their 57functionality is desired (see :ref:`cap_enable` and :ref:`cap_enable_vm`). 58 59 602. Restrictions 61=============== 62 63In general file descriptors can be migrated among processes by means 64of fork() and the SCM_RIGHTS facility of unix domain socket. These 65kinds of tricks are explicitly not supported by kvm. While they will 66not cause harm to the host, their actual behavior is not guaranteed by 67the API. See "General description" for details on the ioctl usage 68model that is supported by KVM. 69 70It is important to note that although VM ioctls may only be issued from 71the process that created the VM, a VM's lifecycle is associated with its 72file descriptor, not its creator (process). In other words, the VM and 73its resources, *including the associated address space*, are not freed 74until the last reference to the VM's file descriptor has been released. 75For example, if fork() is issued after ioctl(KVM_CREATE_VM), the VM will 76not be freed until both the parent (original) process and its child have 77put their references to the VM's file descriptor. 78 79Because a VM's resources are not freed until the last reference to its 80file descriptor is released, creating additional references to a VM 81via fork(), dup(), etc... without careful consideration is strongly 82discouraged and may have unwanted side effects, e.g. memory allocated 83by and on behalf of the VM's process may not be freed/unaccounted when 84the VM is shut down. 85 86 873. Extensions 88============= 89 90As of Linux 2.6.22, the KVM ABI has been stabilized: no backward 91incompatible change are allowed. However, there is an extension 92facility that allows backward-compatible extensions to the API to be 93queried and used. 94 95The extension mechanism is not based on the Linux version number. 96Instead, kvm defines extension identifiers and a facility to query 97whether a particular extension identifier is available. If it is, a 98set of ioctls is available for application use. 99 100 1014. API description 102================== 103 104This section describes ioctls that can be used to control kvm guests. 105For each ioctl, the following information is provided along with a 106description: 107 108 Capability: 109 which KVM extension provides this ioctl. Can be 'basic', 110 which means that is will be provided by any kernel that supports 111 API version 12 (see :ref:`KVM_GET_API_VERSION <KVM_GET_API_VERSION>`), 112 or a KVM_CAP_xyz constant that can be checked with 113 :ref:`KVM_CHECK_EXTENSION <KVM_CHECK_EXTENSION>`. 114 115 Architectures: 116 which instruction set architectures provide this ioctl. 117 x86 includes both i386 and x86_64. 118 119 Type: 120 system, vm, or vcpu. 121 122 Parameters: 123 what parameters are accepted by the ioctl. 124 125 Returns: 126 the return value. General error numbers (EBADF, ENOMEM, EINVAL) 127 are not detailed, but errors with specific meanings are. 128 129 130.. _KVM_GET_API_VERSION: 131 1324.1 KVM_GET_API_VERSION 133----------------------- 134 135:Capability: basic 136:Architectures: all 137:Type: system ioctl 138:Parameters: none 139:Returns: the constant KVM_API_VERSION (=12) 140 141This identifies the API version as the stable kvm API. It is not 142expected that this number will change. However, Linux 2.6.20 and 1432.6.21 report earlier versions; these are not documented and not 144supported. Applications should refuse to run if KVM_GET_API_VERSION 145returns a value other than 12. If this check passes, all ioctls 146described as 'basic' will be available. 147 148 1494.2 KVM_CREATE_VM 150----------------- 151 152:Capability: basic 153:Architectures: all 154:Type: system ioctl 155:Parameters: machine type identifier (KVM_VM_*) 156:Returns: a VM fd that can be used to control the new virtual machine. 157 158The new VM has no virtual cpus and no memory. 159You probably want to use 0 as machine type. 160 161X86: 162^^^^ 163 164Supported X86 VM types can be queried via KVM_CAP_VM_TYPES. 165 166S390: 167^^^^^ 168 169In order to create user controlled virtual machines on S390, check 170KVM_CAP_S390_UCONTROL and use the flag KVM_VM_S390_UCONTROL as 171privileged user (CAP_SYS_ADMIN). 172 173MIPS: 174^^^^^ 175 176To use hardware assisted virtualization on MIPS (VZ ASE) rather than 177the default trap & emulate implementation (which changes the virtual 178memory layout to fit in user mode), check KVM_CAP_MIPS_VZ and use the 179flag KVM_VM_MIPS_VZ. 180 181ARM64: 182^^^^^^ 183 184On arm64, the physical address size for a VM (IPA Size limit) is limited 185to 40bits by default. The limit can be configured if the host supports the 186extension KVM_CAP_ARM_VM_IPA_SIZE. When supported, use 187KVM_VM_TYPE_ARM_IPA_SIZE(IPA_Bits) to set the size in the machine type 188identifier, where IPA_Bits is the maximum width of any physical 189address used by the VM. The IPA_Bits is encoded in bits[7-0] of the 190machine type identifier. 191 192e.g, to configure a guest to use 48bit physical address size:: 193 194 vm_fd = ioctl(dev_fd, KVM_CREATE_VM, KVM_VM_TYPE_ARM_IPA_SIZE(48)); 195 196The requested size (IPA_Bits) must be: 197 198 == ========================================================= 199 0 Implies default size, 40bits (for backward compatibility) 200 N Implies N bits, where N is a positive integer such that, 201 32 <= N <= Host_IPA_Limit 202 == ========================================================= 203 204Host_IPA_Limit is the maximum possible value for IPA_Bits on the host and 205is dependent on the CPU capability and the kernel configuration. The limit can 206be retrieved using KVM_CAP_ARM_VM_IPA_SIZE of the KVM_CHECK_EXTENSION 207ioctl() at run-time. 208 209Creation of the VM will fail if the requested IPA size (whether it is 210implicit or explicit) is unsupported on the host. 211 212Please note that configuring the IPA size does not affect the capability 213exposed by the guest CPUs in ID_AA64MMFR0_EL1[PARange]. It only affects 214size of the address translated by the stage2 level (guest physical to 215host physical address translations). 216 217 2184.3 KVM_GET_MSR_INDEX_LIST, KVM_GET_MSR_FEATURE_INDEX_LIST 219---------------------------------------------------------- 220 221:Capability: basic, KVM_CAP_GET_MSR_FEATURES for KVM_GET_MSR_FEATURE_INDEX_LIST 222:Architectures: x86 223:Type: system ioctl 224:Parameters: struct kvm_msr_list (in/out) 225:Returns: 0 on success; -1 on error 226 227Errors: 228 229 ====== ============================================================ 230 EFAULT the msr index list cannot be read from or written to 231 E2BIG the msr index list is too big to fit in the array specified by 232 the user. 233 ====== ============================================================ 234 235:: 236 237 struct kvm_msr_list { 238 __u32 nmsrs; /* number of msrs in entries */ 239 __u32 indices[0]; 240 }; 241 242The user fills in the size of the indices array in nmsrs, and in return 243kvm adjusts nmsrs to reflect the actual number of msrs and fills in the 244indices array with their numbers. 245 246KVM_GET_MSR_INDEX_LIST returns the guest msrs that are supported. The list 247varies by kvm version and host processor, but does not change otherwise. 248 249Note: if kvm indicates supports MCE (KVM_CAP_MCE), then the MCE bank MSRs are 250not returned in the MSR list, as different vcpus can have a different number 251of banks, as set via the KVM_X86_SETUP_MCE ioctl. 252 253KVM_GET_MSR_FEATURE_INDEX_LIST returns the list of MSRs that can be passed 254to the KVM_GET_MSRS system ioctl. This lets userspace probe host capabilities 255and processor features that are exposed via MSRs (e.g., VMX capabilities). 256This list also varies by kvm version and host processor, but does not change 257otherwise. 258 259 260.. _KVM_CHECK_EXTENSION: 261 2624.4 KVM_CHECK_EXTENSION 263----------------------- 264 265:Capability: basic, KVM_CAP_CHECK_EXTENSION_VM for vm ioctl 266:Architectures: all 267:Type: system ioctl, vm ioctl 268:Parameters: extension identifier (KVM_CAP_*) 269:Returns: 0 if unsupported; 1 (or some other positive integer) if supported 270 271The API allows the application to query about extensions to the core 272kvm API. Userspace passes an extension identifier (an integer) and 273receives an integer that describes the extension availability. 274Generally 0 means no and 1 means yes, but some extensions may report 275additional information in the integer return value. 276 277Based on their initialization different VMs may have different capabilities. 278It is thus encouraged to use the vm ioctl to query for capabilities (available 279with KVM_CAP_CHECK_EXTENSION_VM on the vm fd) 280 2814.5 KVM_GET_VCPU_MMAP_SIZE 282-------------------------- 283 284:Capability: basic 285:Architectures: all 286:Type: system ioctl 287:Parameters: none 288:Returns: size of vcpu mmap area, in bytes 289 290The KVM_RUN ioctl (cf.) communicates with userspace via a shared 291memory region. This ioctl returns the size of that region. See the 292KVM_RUN documentation for details. 293 294Besides the size of the KVM_RUN communication region, other areas of 295the VCPU file descriptor can be mmap-ed, including: 296 297- if KVM_CAP_COALESCED_MMIO is available, a page at 298 KVM_COALESCED_MMIO_PAGE_OFFSET * PAGE_SIZE; for historical reasons, 299 this page is included in the result of KVM_GET_VCPU_MMAP_SIZE. 300 KVM_CAP_COALESCED_MMIO is not documented yet. 301 302- if KVM_CAP_DIRTY_LOG_RING is available, a number of pages at 303 KVM_DIRTY_LOG_PAGE_OFFSET * PAGE_SIZE. For more information on 304 KVM_CAP_DIRTY_LOG_RING, see :ref:`KVM_CAP_DIRTY_LOG_RING`. 305 306 3074.7 KVM_CREATE_VCPU 308------------------- 309 310:Capability: basic 311:Architectures: all 312:Type: vm ioctl 313:Parameters: vcpu id (apic id on x86) 314:Returns: vcpu fd on success, -1 on error 315 316This API adds a vcpu to a virtual machine. No more than max_vcpus may be added. 317The vcpu id is an integer in the range [0, max_vcpu_id). 318 319The recommended max_vcpus value can be retrieved using the KVM_CAP_NR_VCPUS of 320the KVM_CHECK_EXTENSION ioctl() at run-time. 321The maximum possible value for max_vcpus can be retrieved using the 322KVM_CAP_MAX_VCPUS of the KVM_CHECK_EXTENSION ioctl() at run-time. 323 324If the KVM_CAP_NR_VCPUS does not exist, you should assume that max_vcpus is 4 325cpus max. 326If the KVM_CAP_MAX_VCPUS does not exist, you should assume that max_vcpus is 327same as the value returned from KVM_CAP_NR_VCPUS. 328 329The maximum possible value for max_vcpu_id can be retrieved using the 330KVM_CAP_MAX_VCPU_ID of the KVM_CHECK_EXTENSION ioctl() at run-time. 331 332If the KVM_CAP_MAX_VCPU_ID does not exist, you should assume that max_vcpu_id 333is the same as the value returned from KVM_CAP_MAX_VCPUS. 334 335On powerpc using book3s_hv mode, the vcpus are mapped onto virtual 336threads in one or more virtual CPU cores. (This is because the 337hardware requires all the hardware threads in a CPU core to be in the 338same partition.) The KVM_CAP_PPC_SMT capability indicates the number 339of vcpus per virtual core (vcore). The vcore id is obtained by 340dividing the vcpu id by the number of vcpus per vcore. The vcpus in a 341given vcore will always be in the same physical core as each other 342(though that might be a different physical core from time to time). 343Userspace can control the threading (SMT) mode of the guest by its 344allocation of vcpu ids. For example, if userspace wants 345single-threaded guest vcpus, it should make all vcpu ids be a multiple 346of the number of vcpus per vcore. 347 348For virtual cpus that have been created with S390 user controlled virtual 349machines, the resulting vcpu fd can be memory mapped at page offset 350KVM_S390_SIE_PAGE_OFFSET in order to obtain a memory map of the virtual 351cpu's hardware control block. 352 353 3544.8 KVM_GET_DIRTY_LOG 355--------------------- 356 357:Capability: basic 358:Architectures: all 359:Type: vm ioctl 360:Parameters: struct kvm_dirty_log (in/out) 361:Returns: 0 on success, -1 on error 362 363:: 364 365 /* for KVM_GET_DIRTY_LOG */ 366 struct kvm_dirty_log { 367 __u32 slot; 368 __u32 padding; 369 union { 370 void __user *dirty_bitmap; /* one bit per page */ 371 __u64 padding; 372 }; 373 }; 374 375Given a memory slot, return a bitmap containing any pages dirtied 376since the last call to this ioctl. Bit 0 is the first page in the 377memory slot. Ensure the entire structure is cleared to avoid padding 378issues. 379 380If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies 381the address space for which you want to return the dirty bitmap. See 382KVM_SET_USER_MEMORY_REGION for details on the usage of slot field. 383 384The bits in the dirty bitmap are cleared before the ioctl returns, unless 385KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is enabled. For more information, 386see the description of the capability. 387 388Note that the Xen shared_info page, if configured, shall always be assumed 389to be dirty. KVM will not explicitly mark it such. 390 391 3924.10 KVM_RUN 393------------ 394 395:Capability: basic 396:Architectures: all 397:Type: vcpu ioctl 398:Parameters: none 399:Returns: 0 on success, -1 on error 400 401Errors: 402 403 ======= ============================================================== 404 EINTR an unmasked signal is pending 405 ENOEXEC the vcpu hasn't been initialized or the guest tried to execute 406 instructions from device memory (arm64) 407 ENOSYS data abort outside memslots with no syndrome info and 408 KVM_CAP_ARM_NISV_TO_USER not enabled (arm64) 409 EPERM SVE feature set but not finalized (arm64) 410 ======= ============================================================== 411 412This ioctl is used to run a guest virtual cpu. While there are no 413explicit parameters, there is an implicit parameter block that can be 414obtained by mmap()ing the vcpu fd at offset 0, with the size given by 415KVM_GET_VCPU_MMAP_SIZE. The parameter block is formatted as a 'struct 416kvm_run' (see below). 417 418 4194.11 KVM_GET_REGS 420----------------- 421 422:Capability: basic 423:Architectures: all except arm64 424:Type: vcpu ioctl 425:Parameters: struct kvm_regs (out) 426:Returns: 0 on success, -1 on error 427 428Reads the general purpose registers from the vcpu. 429 430:: 431 432 /* x86 */ 433 struct kvm_regs { 434 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */ 435 __u64 rax, rbx, rcx, rdx; 436 __u64 rsi, rdi, rsp, rbp; 437 __u64 r8, r9, r10, r11; 438 __u64 r12, r13, r14, r15; 439 __u64 rip, rflags; 440 }; 441 442 /* mips */ 443 struct kvm_regs { 444 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */ 445 __u64 gpr[32]; 446 __u64 hi; 447 __u64 lo; 448 __u64 pc; 449 }; 450 451 /* LoongArch */ 452 struct kvm_regs { 453 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */ 454 unsigned long gpr[32]; 455 unsigned long pc; 456 }; 457 458 4594.12 KVM_SET_REGS 460----------------- 461 462:Capability: basic 463:Architectures: all except arm64 464:Type: vcpu ioctl 465:Parameters: struct kvm_regs (in) 466:Returns: 0 on success, -1 on error 467 468Writes the general purpose registers into the vcpu. 469 470See KVM_GET_REGS for the data structure. 471 472 4734.13 KVM_GET_SREGS 474------------------ 475 476:Capability: basic 477:Architectures: x86, ppc 478:Type: vcpu ioctl 479:Parameters: struct kvm_sregs (out) 480:Returns: 0 on success, -1 on error 481 482Reads special registers from the vcpu. 483 484:: 485 486 /* x86 */ 487 struct kvm_sregs { 488 struct kvm_segment cs, ds, es, fs, gs, ss; 489 struct kvm_segment tr, ldt; 490 struct kvm_dtable gdt, idt; 491 __u64 cr0, cr2, cr3, cr4, cr8; 492 __u64 efer; 493 __u64 apic_base; 494 __u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64]; 495 }; 496 497 /* ppc -- see arch/powerpc/include/uapi/asm/kvm.h */ 498 499interrupt_bitmap is a bitmap of pending external interrupts. At most 500one bit may be set. This interrupt has been acknowledged by the APIC 501but not yet injected into the cpu core. 502 503 5044.14 KVM_SET_SREGS 505------------------ 506 507:Capability: basic 508:Architectures: x86, ppc 509:Type: vcpu ioctl 510:Parameters: struct kvm_sregs (in) 511:Returns: 0 on success, -1 on error 512 513Writes special registers into the vcpu. See KVM_GET_SREGS for the 514data structures. 515 516 5174.15 KVM_TRANSLATE 518------------------ 519 520:Capability: basic 521:Architectures: x86 522:Type: vcpu ioctl 523:Parameters: struct kvm_translation (in/out) 524:Returns: 0 on success, -1 on error 525 526Translates a virtual address according to the vcpu's current address 527translation mode. 528 529:: 530 531 struct kvm_translation { 532 /* in */ 533 __u64 linear_address; 534 535 /* out */ 536 __u64 physical_address; 537 __u8 valid; 538 __u8 writeable; 539 __u8 usermode; 540 __u8 pad[5]; 541 }; 542 543 5444.16 KVM_INTERRUPT 545------------------ 546 547:Capability: basic 548:Architectures: x86, ppc, mips, riscv, loongarch 549:Type: vcpu ioctl 550:Parameters: struct kvm_interrupt (in) 551:Returns: 0 on success, negative on failure. 552 553Queues a hardware interrupt vector to be injected. 554 555:: 556 557 /* for KVM_INTERRUPT */ 558 struct kvm_interrupt { 559 /* in */ 560 __u32 irq; 561 }; 562 563X86: 564^^^^ 565 566:Returns: 567 568 ========= =================================== 569 0 on success, 570 -EEXIST if an interrupt is already enqueued 571 -EINVAL the irq number is invalid 572 -ENXIO if the PIC is in the kernel 573 -EFAULT if the pointer is invalid 574 ========= =================================== 575 576Note 'irq' is an interrupt vector, not an interrupt pin or line. This 577ioctl is useful if the in-kernel PIC is not used. 578 579PPC: 580^^^^ 581 582Queues an external interrupt to be injected. This ioctl is overloaded 583with 3 different irq values: 584 585a) KVM_INTERRUPT_SET 586 587 This injects an edge type external interrupt into the guest once it's ready 588 to receive interrupts. When injected, the interrupt is done. 589 590b) KVM_INTERRUPT_UNSET 591 592 This unsets any pending interrupt. 593 594 Only available with KVM_CAP_PPC_UNSET_IRQ. 595 596c) KVM_INTERRUPT_SET_LEVEL 597 598 This injects a level type external interrupt into the guest context. The 599 interrupt stays pending until a specific ioctl with KVM_INTERRUPT_UNSET 600 is triggered. 601 602 Only available with KVM_CAP_PPC_IRQ_LEVEL. 603 604Note that any value for 'irq' other than the ones stated above is invalid 605and incurs unexpected behavior. 606 607This is an asynchronous vcpu ioctl and can be invoked from any thread. 608 609MIPS: 610^^^^^ 611 612Queues an external interrupt to be injected into the virtual CPU. A negative 613interrupt number dequeues the interrupt. 614 615This is an asynchronous vcpu ioctl and can be invoked from any thread. 616 617RISC-V: 618^^^^^^^ 619 620Queues an external interrupt to be injected into the virtual CPU. This ioctl 621is overloaded with 2 different irq values: 622 623a) KVM_INTERRUPT_SET 624 625 This sets external interrupt for a virtual CPU and it will receive 626 once it is ready. 627 628b) KVM_INTERRUPT_UNSET 629 630 This clears pending external interrupt for a virtual CPU. 631 632This is an asynchronous vcpu ioctl and can be invoked from any thread. 633 634LOONGARCH: 635^^^^^^^^^^ 636 637Queues an external interrupt to be injected into the virtual CPU. A negative 638interrupt number dequeues the interrupt. 639 640This is an asynchronous vcpu ioctl and can be invoked from any thread. 641 642 6434.18 KVM_GET_MSRS 644----------------- 645 646:Capability: basic (vcpu), KVM_CAP_GET_MSR_FEATURES (system) 647:Architectures: x86 648:Type: system ioctl, vcpu ioctl 649:Parameters: struct kvm_msrs (in/out) 650:Returns: number of msrs successfully returned; 651 -1 on error 652 653When used as a system ioctl: 654Reads the values of MSR-based features that are available for the VM. This 655is similar to KVM_GET_SUPPORTED_CPUID, but it returns MSR indices and values. 656The list of msr-based features can be obtained using KVM_GET_MSR_FEATURE_INDEX_LIST 657in a system ioctl. 658 659When used as a vcpu ioctl: 660Reads model-specific registers from the vcpu. Supported msr indices can 661be obtained using KVM_GET_MSR_INDEX_LIST in a system ioctl. 662 663:: 664 665 struct kvm_msrs { 666 __u32 nmsrs; /* number of msrs in entries */ 667 __u32 pad; 668 669 struct kvm_msr_entry entries[0]; 670 }; 671 672 struct kvm_msr_entry { 673 __u32 index; 674 __u32 reserved; 675 __u64 data; 676 }; 677 678Application code should set the 'nmsrs' member (which indicates the 679size of the entries array) and the 'index' member of each array entry. 680kvm will fill in the 'data' member. 681 682 6834.19 KVM_SET_MSRS 684----------------- 685 686:Capability: basic 687:Architectures: x86 688:Type: vcpu ioctl 689:Parameters: struct kvm_msrs (in) 690:Returns: number of msrs successfully set (see below), -1 on error 691 692Writes model-specific registers to the vcpu. See KVM_GET_MSRS for the 693data structures. 694 695Application code should set the 'nmsrs' member (which indicates the 696size of the entries array), and the 'index' and 'data' members of each 697array entry. 698 699It tries to set the MSRs in array entries[] one by one. If setting an MSR 700fails, e.g., due to setting reserved bits, the MSR isn't supported/emulated 701by KVM, etc..., it stops processing the MSR list and returns the number of 702MSRs that have been set successfully. 703 704 7054.20 KVM_SET_CPUID 706------------------ 707 708:Capability: basic 709:Architectures: x86 710:Type: vcpu ioctl 711:Parameters: struct kvm_cpuid (in) 712:Returns: 0 on success, -1 on error 713 714Defines the vcpu responses to the cpuid instruction. Applications 715should use the KVM_SET_CPUID2 ioctl if available. 716 717Caveat emptor: 718 - If this IOCTL fails, KVM gives no guarantees that previous valid CPUID 719 configuration (if there is) is not corrupted. Userspace can get a copy 720 of the resulting CPUID configuration through KVM_GET_CPUID2 in case. 721 - Using KVM_SET_CPUID{,2} after KVM_RUN, i.e. changing the guest vCPU model 722 after running the guest, may cause guest instability. 723 - Using heterogeneous CPUID configurations, modulo APIC IDs, topology, etc... 724 may cause guest instability. 725 726:: 727 728 struct kvm_cpuid_entry { 729 __u32 function; 730 __u32 eax; 731 __u32 ebx; 732 __u32 ecx; 733 __u32 edx; 734 __u32 padding; 735 }; 736 737 /* for KVM_SET_CPUID */ 738 struct kvm_cpuid { 739 __u32 nent; 740 __u32 padding; 741 struct kvm_cpuid_entry entries[0]; 742 }; 743 744 7454.21 KVM_SET_SIGNAL_MASK 746------------------------ 747 748:Capability: basic 749:Architectures: all 750:Type: vcpu ioctl 751:Parameters: struct kvm_signal_mask (in) 752:Returns: 0 on success, -1 on error 753 754Defines which signals are blocked during execution of KVM_RUN. This 755signal mask temporarily overrides the threads signal mask. Any 756unblocked signal received (except SIGKILL and SIGSTOP, which retain 757their traditional behaviour) will cause KVM_RUN to return with -EINTR. 758 759Note the signal will only be delivered if not blocked by the original 760signal mask. 761 762:: 763 764 /* for KVM_SET_SIGNAL_MASK */ 765 struct kvm_signal_mask { 766 __u32 len; 767 __u8 sigset[0]; 768 }; 769 770 7714.22 KVM_GET_FPU 772---------------- 773 774:Capability: basic 775:Architectures: x86, loongarch 776:Type: vcpu ioctl 777:Parameters: struct kvm_fpu (out) 778:Returns: 0 on success, -1 on error 779 780Reads the floating point state from the vcpu. 781 782:: 783 784 /* x86: for KVM_GET_FPU and KVM_SET_FPU */ 785 struct kvm_fpu { 786 __u8 fpr[8][16]; 787 __u16 fcw; 788 __u16 fsw; 789 __u8 ftwx; /* in fxsave format */ 790 __u8 pad1; 791 __u16 last_opcode; 792 __u64 last_ip; 793 __u64 last_dp; 794 __u8 xmm[16][16]; 795 __u32 mxcsr; 796 __u32 pad2; 797 }; 798 799 /* LoongArch: for KVM_GET_FPU and KVM_SET_FPU */ 800 struct kvm_fpu { 801 __u32 fcsr; 802 __u64 fcc; 803 struct kvm_fpureg { 804 __u64 val64[4]; 805 }fpr[32]; 806 }; 807 808 8094.23 KVM_SET_FPU 810---------------- 811 812:Capability: basic 813:Architectures: x86, loongarch 814:Type: vcpu ioctl 815:Parameters: struct kvm_fpu (in) 816:Returns: 0 on success, -1 on error 817 818Writes the floating point state to the vcpu. 819 820:: 821 822 /* x86: for KVM_GET_FPU and KVM_SET_FPU */ 823 struct kvm_fpu { 824 __u8 fpr[8][16]; 825 __u16 fcw; 826 __u16 fsw; 827 __u8 ftwx; /* in fxsave format */ 828 __u8 pad1; 829 __u16 last_opcode; 830 __u64 last_ip; 831 __u64 last_dp; 832 __u8 xmm[16][16]; 833 __u32 mxcsr; 834 __u32 pad2; 835 }; 836 837 /* LoongArch: for KVM_GET_FPU and KVM_SET_FPU */ 838 struct kvm_fpu { 839 __u32 fcsr; 840 __u64 fcc; 841 struct kvm_fpureg { 842 __u64 val64[4]; 843 }fpr[32]; 844 }; 845 846 8474.24 KVM_CREATE_IRQCHIP 848----------------------- 849 850:Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQCHIP (s390) 851:Architectures: x86, arm64, s390 852:Type: vm ioctl 853:Parameters: none 854:Returns: 0 on success, -1 on error 855 856Creates an interrupt controller model in the kernel. 857On x86, creates a virtual ioapic, a virtual PIC (two PICs, nested), and sets up 858future vcpus to have a local APIC. IRQ routing for GSIs 0-15 is set to both 859PIC and IOAPIC; GSI 16-23 only go to the IOAPIC. 860On arm64, a GICv2 is created. Any other GIC versions require the usage of 861KVM_CREATE_DEVICE, which also supports creating a GICv2. Using 862KVM_CREATE_DEVICE is preferred over KVM_CREATE_IRQCHIP for GICv2. 863On s390, a dummy irq routing table is created. 864 865Note that on s390 the KVM_CAP_S390_IRQCHIP vm capability needs to be enabled 866before KVM_CREATE_IRQCHIP can be used. 867 868 8694.25 KVM_IRQ_LINE 870----------------- 871 872:Capability: KVM_CAP_IRQCHIP 873:Architectures: x86, arm64 874:Type: vm ioctl 875:Parameters: struct kvm_irq_level 876:Returns: 0 on success, -1 on error 877 878Sets the level of a GSI input to the interrupt controller model in the kernel. 879On some architectures it is required that an interrupt controller model has 880been previously created with KVM_CREATE_IRQCHIP. Note that edge-triggered 881interrupts require the level to be set to 1 and then back to 0. 882 883On real hardware, interrupt pins can be active-low or active-high. This 884does not matter for the level field of struct kvm_irq_level: 1 always 885means active (asserted), 0 means inactive (deasserted). 886 887x86 allows the operating system to program the interrupt polarity 888(active-low/active-high) for level-triggered interrupts, and KVM used 889to consider the polarity. However, due to bitrot in the handling of 890active-low interrupts, the above convention is now valid on x86 too. 891This is signaled by KVM_CAP_X86_IOAPIC_POLARITY_IGNORED. Userspace 892should not present interrupts to the guest as active-low unless this 893capability is present (or unless it is not using the in-kernel irqchip, 894of course). 895 896 897arm64 can signal an interrupt either at the CPU level, or at the 898in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to 899use PPIs designated for specific cpus. The irq field is interpreted 900like this:: 901 902 bits: | 31 ... 28 | 27 ... 24 | 23 ... 16 | 15 ... 0 | 903 field: | vcpu2_index | irq_type | vcpu_index | irq_id | 904 905The irq_type field has the following values: 906 907- KVM_ARM_IRQ_TYPE_CPU: 908 out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ 909- KVM_ARM_IRQ_TYPE_SPI: 910 in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.) 911 (the vcpu_index field is ignored) 912- KVM_ARM_IRQ_TYPE_PPI: 913 in-kernel GIC: PPI, irq_id between 16 and 31 (incl.) 914 915(The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs) 916 917In both cases, level is used to assert/deassert the line. 918 919When KVM_CAP_ARM_IRQ_LINE_LAYOUT_2 is supported, the target vcpu is 920identified as (256 * vcpu2_index + vcpu_index). Otherwise, vcpu2_index 921must be zero. 922 923Note that on arm64, the KVM_CAP_IRQCHIP capability only conditions 924injection of interrupts for the in-kernel irqchip. KVM_IRQ_LINE can always 925be used for a userspace interrupt controller. 926 927:: 928 929 struct kvm_irq_level { 930 union { 931 __u32 irq; /* GSI */ 932 __s32 status; /* not used for KVM_IRQ_LEVEL */ 933 }; 934 __u32 level; /* 0 or 1 */ 935 }; 936 937 9384.26 KVM_GET_IRQCHIP 939-------------------- 940 941:Capability: KVM_CAP_IRQCHIP 942:Architectures: x86 943:Type: vm ioctl 944:Parameters: struct kvm_irqchip (in/out) 945:Returns: 0 on success, -1 on error 946 947Reads the state of a kernel interrupt controller created with 948KVM_CREATE_IRQCHIP into a buffer provided by the caller. 949 950:: 951 952 struct kvm_irqchip { 953 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */ 954 __u32 pad; 955 union { 956 char dummy[512]; /* reserving space */ 957 struct kvm_pic_state pic; 958 struct kvm_ioapic_state ioapic; 959 } chip; 960 }; 961 962 9634.27 KVM_SET_IRQCHIP 964-------------------- 965 966:Capability: KVM_CAP_IRQCHIP 967:Architectures: x86 968:Type: vm ioctl 969:Parameters: struct kvm_irqchip (in) 970:Returns: 0 on success, -1 on error 971 972Sets the state of a kernel interrupt controller created with 973KVM_CREATE_IRQCHIP from a buffer provided by the caller. 974 975:: 976 977 struct kvm_irqchip { 978 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */ 979 __u32 pad; 980 union { 981 char dummy[512]; /* reserving space */ 982 struct kvm_pic_state pic; 983 struct kvm_ioapic_state ioapic; 984 } chip; 985 }; 986 987 9884.28 KVM_XEN_HVM_CONFIG 989----------------------- 990 991:Capability: KVM_CAP_XEN_HVM 992:Architectures: x86 993:Type: vm ioctl 994:Parameters: struct kvm_xen_hvm_config (in) 995:Returns: 0 on success, -1 on error 996 997Sets the MSR that the Xen HVM guest uses to initialize its hypercall 998page, and provides the starting address and size of the hypercall 999blobs in userspace. When the guest writes the MSR, kvm copies one 1000page of a blob (32- or 64-bit, depending on the vcpu mode) to guest 1001memory. 1002 1003The MSR index must be in the range [0x40000000, 0x4fffffff], i.e. must reside 1004in the range that is unofficially reserved for use by hypervisors. The min/max 1005values are enumerated via KVM_XEN_MSR_MIN_INDEX and KVM_XEN_MSR_MAX_INDEX. 1006 1007:: 1008 1009 struct kvm_xen_hvm_config { 1010 __u32 flags; 1011 __u32 msr; 1012 __u64 blob_addr_32; 1013 __u64 blob_addr_64; 1014 __u8 blob_size_32; 1015 __u8 blob_size_64; 1016 __u8 pad2[30]; 1017 }; 1018 1019If certain flags are returned from the KVM_CAP_XEN_HVM check, they may 1020be set in the flags field of this ioctl: 1021 1022The KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL flag requests KVM to generate 1023the contents of the hypercall page automatically; hypercalls will be 1024intercepted and passed to userspace through KVM_EXIT_XEN. In this 1025case, all of the blob size and address fields must be zero. 1026 1027The KVM_XEN_HVM_CONFIG_EVTCHN_SEND flag indicates to KVM that userspace 1028will always use the KVM_XEN_HVM_EVTCHN_SEND ioctl to deliver event 1029channel interrupts rather than manipulating the guest's shared_info 1030structures directly. This, in turn, may allow KVM to enable features 1031such as intercepting the SCHEDOP_poll hypercall to accelerate PV 1032spinlock operation for the guest. Userspace may still use the ioctl 1033to deliver events if it was advertised, even if userspace does not 1034send this indication that it will always do so 1035 1036No other flags are currently valid in the struct kvm_xen_hvm_config. 1037 10384.29 KVM_GET_CLOCK 1039------------------ 1040 1041:Capability: KVM_CAP_ADJUST_CLOCK 1042:Architectures: x86 1043:Type: vm ioctl 1044:Parameters: struct kvm_clock_data (out) 1045:Returns: 0 on success, -1 on error 1046 1047Gets the current timestamp of kvmclock as seen by the current guest. In 1048conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios 1049such as migration. 1050 1051When KVM_CAP_ADJUST_CLOCK is passed to KVM_CHECK_EXTENSION, it returns the 1052set of bits that KVM can return in struct kvm_clock_data's flag member. 1053 1054The following flags are defined: 1055 1056KVM_CLOCK_TSC_STABLE 1057 If set, the returned value is the exact kvmclock 1058 value seen by all VCPUs at the instant when KVM_GET_CLOCK was called. 1059 If clear, the returned value is simply CLOCK_MONOTONIC plus a constant 1060 offset; the offset can be modified with KVM_SET_CLOCK. KVM will try 1061 to make all VCPUs follow this clock, but the exact value read by each 1062 VCPU could differ, because the host TSC is not stable. 1063 1064KVM_CLOCK_REALTIME 1065 If set, the `realtime` field in the kvm_clock_data 1066 structure is populated with the value of the host's real time 1067 clocksource at the instant when KVM_GET_CLOCK was called. If clear, 1068 the `realtime` field does not contain a value. 1069 1070KVM_CLOCK_HOST_TSC 1071 If set, the `host_tsc` field in the kvm_clock_data 1072 structure is populated with the value of the host's timestamp counter (TSC) 1073 at the instant when KVM_GET_CLOCK was called. If clear, the `host_tsc` field 1074 does not contain a value. 1075 1076:: 1077 1078 struct kvm_clock_data { 1079 __u64 clock; /* kvmclock current value */ 1080 __u32 flags; 1081 __u32 pad0; 1082 __u64 realtime; 1083 __u64 host_tsc; 1084 __u32 pad[4]; 1085 }; 1086 1087 10884.30 KVM_SET_CLOCK 1089------------------ 1090 1091:Capability: KVM_CAP_ADJUST_CLOCK 1092:Architectures: x86 1093:Type: vm ioctl 1094:Parameters: struct kvm_clock_data (in) 1095:Returns: 0 on success, -1 on error 1096 1097Sets the current timestamp of kvmclock to the value specified in its parameter. 1098In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios 1099such as migration. 1100 1101The following flags can be passed: 1102 1103KVM_CLOCK_REALTIME 1104 If set, KVM will compare the value of the `realtime` field 1105 with the value of the host's real time clocksource at the instant when 1106 KVM_SET_CLOCK was called. The difference in elapsed time is added to the final 1107 kvmclock value that will be provided to guests. 1108 1109Other flags returned by ``KVM_GET_CLOCK`` are accepted but ignored. 1110 1111:: 1112 1113 struct kvm_clock_data { 1114 __u64 clock; /* kvmclock current value */ 1115 __u32 flags; 1116 __u32 pad0; 1117 __u64 realtime; 1118 __u64 host_tsc; 1119 __u32 pad[4]; 1120 }; 1121 1122 11234.31 KVM_GET_VCPU_EVENTS 1124------------------------ 1125 1126:Capability: KVM_CAP_VCPU_EVENTS 1127:Extended by: KVM_CAP_INTR_SHADOW 1128:Architectures: x86, arm64 1129:Type: vcpu ioctl 1130:Parameters: struct kvm_vcpu_events (out) 1131:Returns: 0 on success, -1 on error 1132 1133X86: 1134^^^^ 1135 1136Gets currently pending exceptions, interrupts, and NMIs as well as related 1137states of the vcpu. 1138 1139:: 1140 1141 struct kvm_vcpu_events { 1142 struct { 1143 __u8 injected; 1144 __u8 nr; 1145 __u8 has_error_code; 1146 __u8 pending; 1147 __u32 error_code; 1148 } exception; 1149 struct { 1150 __u8 injected; 1151 __u8 nr; 1152 __u8 soft; 1153 __u8 shadow; 1154 } interrupt; 1155 struct { 1156 __u8 injected; 1157 __u8 pending; 1158 __u8 masked; 1159 __u8 pad; 1160 } nmi; 1161 __u32 sipi_vector; 1162 __u32 flags; 1163 struct { 1164 __u8 smm; 1165 __u8 pending; 1166 __u8 smm_inside_nmi; 1167 __u8 latched_init; 1168 } smi; 1169 __u8 reserved[27]; 1170 __u8 exception_has_payload; 1171 __u64 exception_payload; 1172 }; 1173 1174The following bits are defined in the flags field: 1175 1176- KVM_VCPUEVENT_VALID_SHADOW may be set to signal that 1177 interrupt.shadow contains a valid state. 1178 1179- KVM_VCPUEVENT_VALID_SMM may be set to signal that smi contains a 1180 valid state. 1181 1182- KVM_VCPUEVENT_VALID_PAYLOAD may be set to signal that the 1183 exception_has_payload, exception_payload, and exception.pending 1184 fields contain a valid state. This bit will be set whenever 1185 KVM_CAP_EXCEPTION_PAYLOAD is enabled. 1186 1187- KVM_VCPUEVENT_VALID_TRIPLE_FAULT may be set to signal that the 1188 triple_fault_pending field contains a valid state. This bit will 1189 be set whenever KVM_CAP_X86_TRIPLE_FAULT_EVENT is enabled. 1190 1191ARM64: 1192^^^^^^ 1193 1194If the guest accesses a device that is being emulated by the host kernel in 1195such a way that a real device would generate a physical SError, KVM may make 1196a virtual SError pending for that VCPU. This system error interrupt remains 1197pending until the guest takes the exception by unmasking PSTATE.A. 1198 1199Running the VCPU may cause it to take a pending SError, or make an access that 1200causes an SError to become pending. The event's description is only valid while 1201the VPCU is not running. 1202 1203This API provides a way to read and write the pending 'event' state that is not 1204visible to the guest. To save, restore or migrate a VCPU the struct representing 1205the state can be read then written using this GET/SET API, along with the other 1206guest-visible registers. It is not possible to 'cancel' an SError that has been 1207made pending. 1208 1209A device being emulated in user-space may also wish to generate an SError. To do 1210this the events structure can be populated by user-space. The current state 1211should be read first, to ensure no existing SError is pending. If an existing 1212SError is pending, the architecture's 'Multiple SError interrupts' rules should 1213be followed. (2.5.3 of DDI0587.a "ARM Reliability, Availability, and 1214Serviceability (RAS) Specification"). 1215 1216SError exceptions always have an ESR value. Some CPUs have the ability to 1217specify what the virtual SError's ESR value should be. These systems will 1218advertise KVM_CAP_ARM_INJECT_SERROR_ESR. In this case exception.has_esr will 1219always have a non-zero value when read, and the agent making an SError pending 1220should specify the ISS field in the lower 24 bits of exception.serror_esr. If 1221the system supports KVM_CAP_ARM_INJECT_SERROR_ESR, but user-space sets the events 1222with exception.has_esr as zero, KVM will choose an ESR. 1223 1224Specifying exception.has_esr on a system that does not support it will return 1225-EINVAL. Setting anything other than the lower 24bits of exception.serror_esr 1226will return -EINVAL. 1227 1228It is not possible to read back a pending external abort (injected via 1229KVM_SET_VCPU_EVENTS or otherwise) because such an exception is always delivered 1230directly to the virtual CPU). 1231 1232:: 1233 1234 struct kvm_vcpu_events { 1235 struct { 1236 __u8 serror_pending; 1237 __u8 serror_has_esr; 1238 __u8 ext_dabt_pending; 1239 /* Align it to 8 bytes */ 1240 __u8 pad[5]; 1241 __u64 serror_esr; 1242 } exception; 1243 __u32 reserved[12]; 1244 }; 1245 12464.32 KVM_SET_VCPU_EVENTS 1247------------------------ 1248 1249:Capability: KVM_CAP_VCPU_EVENTS 1250:Extended by: KVM_CAP_INTR_SHADOW 1251:Architectures: x86, arm64 1252:Type: vcpu ioctl 1253:Parameters: struct kvm_vcpu_events (in) 1254:Returns: 0 on success, -1 on error 1255 1256X86: 1257^^^^ 1258 1259Set pending exceptions, interrupts, and NMIs as well as related states of the 1260vcpu. 1261 1262See KVM_GET_VCPU_EVENTS for the data structure. 1263 1264Fields that may be modified asynchronously by running VCPUs can be excluded 1265from the update. These fields are nmi.pending, sipi_vector, smi.smm, 1266smi.pending. Keep the corresponding bits in the flags field cleared to 1267suppress overwriting the current in-kernel state. The bits are: 1268 1269=============================== ================================== 1270KVM_VCPUEVENT_VALID_NMI_PENDING transfer nmi.pending to the kernel 1271KVM_VCPUEVENT_VALID_SIPI_VECTOR transfer sipi_vector 1272KVM_VCPUEVENT_VALID_SMM transfer the smi sub-struct. 1273=============================== ================================== 1274 1275If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in 1276the flags field to signal that interrupt.shadow contains a valid state and 1277shall be written into the VCPU. 1278 1279KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available. 1280 1281If KVM_CAP_EXCEPTION_PAYLOAD is enabled, KVM_VCPUEVENT_VALID_PAYLOAD 1282can be set in the flags field to signal that the 1283exception_has_payload, exception_payload, and exception.pending fields 1284contain a valid state and shall be written into the VCPU. 1285 1286If KVM_CAP_X86_TRIPLE_FAULT_EVENT is enabled, KVM_VCPUEVENT_VALID_TRIPLE_FAULT 1287can be set in flags field to signal that the triple_fault field contains 1288a valid state and shall be written into the VCPU. 1289 1290ARM64: 1291^^^^^^ 1292 1293User space may need to inject several types of events to the guest. 1294 1295Set the pending SError exception state for this VCPU. It is not possible to 1296'cancel' an Serror that has been made pending. 1297 1298If the guest performed an access to I/O memory which could not be handled by 1299userspace, for example because of missing instruction syndrome decode 1300information or because there is no device mapped at the accessed IPA, then 1301userspace can ask the kernel to inject an external abort using the address 1302from the exiting fault on the VCPU. It is a programming error to set 1303ext_dabt_pending after an exit which was not either KVM_EXIT_MMIO or 1304KVM_EXIT_ARM_NISV. This feature is only available if the system supports 1305KVM_CAP_ARM_INJECT_EXT_DABT. This is a helper which provides commonality in 1306how userspace reports accesses for the above cases to guests, across different 1307userspace implementations. Nevertheless, userspace can still emulate all Arm 1308exceptions by manipulating individual registers using the KVM_SET_ONE_REG API. 1309 1310See KVM_GET_VCPU_EVENTS for the data structure. 1311 1312 13134.33 KVM_GET_DEBUGREGS 1314---------------------- 1315 1316:Capability: KVM_CAP_DEBUGREGS 1317:Architectures: x86 1318:Type: vcpu ioctl 1319:Parameters: struct kvm_debugregs (out) 1320:Returns: 0 on success, -1 on error 1321 1322Reads debug registers from the vcpu. 1323 1324:: 1325 1326 struct kvm_debugregs { 1327 __u64 db[4]; 1328 __u64 dr6; 1329 __u64 dr7; 1330 __u64 flags; 1331 __u64 reserved[9]; 1332 }; 1333 1334 13354.34 KVM_SET_DEBUGREGS 1336---------------------- 1337 1338:Capability: KVM_CAP_DEBUGREGS 1339:Architectures: x86 1340:Type: vcpu ioctl 1341:Parameters: struct kvm_debugregs (in) 1342:Returns: 0 on success, -1 on error 1343 1344Writes debug registers into the vcpu. 1345 1346See KVM_GET_DEBUGREGS for the data structure. The flags field is unused 1347yet and must be cleared on entry. 1348 1349 13504.35 KVM_SET_USER_MEMORY_REGION 1351------------------------------- 1352 1353:Capability: KVM_CAP_USER_MEMORY 1354:Architectures: all 1355:Type: vm ioctl 1356:Parameters: struct kvm_userspace_memory_region (in) 1357:Returns: 0 on success, -1 on error 1358 1359:: 1360 1361 struct kvm_userspace_memory_region { 1362 __u32 slot; 1363 __u32 flags; 1364 __u64 guest_phys_addr; 1365 __u64 memory_size; /* bytes */ 1366 __u64 userspace_addr; /* start of the userspace allocated memory */ 1367 }; 1368 1369 /* for kvm_userspace_memory_region::flags */ 1370 #define KVM_MEM_LOG_DIRTY_PAGES (1UL << 0) 1371 #define KVM_MEM_READONLY (1UL << 1) 1372 1373This ioctl allows the user to create, modify or delete a guest physical 1374memory slot. Bits 0-15 of "slot" specify the slot id and this value 1375should be less than the maximum number of user memory slots supported per 1376VM. The maximum allowed slots can be queried using KVM_CAP_NR_MEMSLOTS. 1377Slots may not overlap in guest physical address space. 1378 1379If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot" 1380specifies the address space which is being modified. They must be 1381less than the value that KVM_CHECK_EXTENSION returns for the 1382KVM_CAP_MULTI_ADDRESS_SPACE capability. Slots in separate address spaces 1383are unrelated; the restriction on overlapping slots only applies within 1384each address space. 1385 1386Deleting a slot is done by passing zero for memory_size. When changing 1387an existing slot, it may be moved in the guest physical memory space, 1388or its flags may be modified, but it may not be resized. 1389 1390Memory for the region is taken starting at the address denoted by the 1391field userspace_addr, which must point at user addressable memory for 1392the entire memory slot size. Any object may back this memory, including 1393anonymous memory, ordinary files, and hugetlbfs. 1394 1395On architectures that support a form of address tagging, userspace_addr must 1396be an untagged address. 1397 1398It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr 1399be identical. This allows large pages in the guest to be backed by large 1400pages in the host. 1401 1402The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and 1403KVM_MEM_READONLY. The former can be set to instruct KVM to keep track of 1404writes to memory within the slot. See KVM_GET_DIRTY_LOG ioctl to know how to 1405use it. The latter can be set, if KVM_CAP_READONLY_MEM capability allows it, 1406to make a new slot read-only. In this case, writes to this memory will be 1407posted to userspace as KVM_EXIT_MMIO exits. 1408 1409When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of 1410the memory region are automatically reflected into the guest. For example, an 1411mmap() that affects the region will be made visible immediately. Another 1412example is madvise(MADV_DROP). 1413 1414For TDX guest, deleting/moving memory region loses guest memory contents. 1415Read only region isn't supported. Only as-id 0 is supported. 1416 1417Note: On arm64, a write generated by the page-table walker (to update 1418the Access and Dirty flags, for example) never results in a 1419KVM_EXIT_MMIO exit when the slot has the KVM_MEM_READONLY flag. This 1420is because KVM cannot provide the data that would be written by the 1421page-table walker, making it impossible to emulate the access. 1422Instead, an abort (data abort if the cause of the page-table update 1423was a load or a store, instruction abort if it was an instruction 1424fetch) is injected in the guest. 1425 1426S390: 1427^^^^^ 1428 1429Returns -EINVAL or -EEXIST if the VM has the KVM_VM_S390_UCONTROL flag set. 1430Returns -EINVAL if called on a protected VM. 1431 14324.36 KVM_SET_TSS_ADDR 1433--------------------- 1434 1435:Capability: KVM_CAP_SET_TSS_ADDR 1436:Architectures: x86 1437:Type: vm ioctl 1438:Parameters: unsigned long tss_address (in) 1439:Returns: 0 on success, -1 on error 1440 1441This ioctl defines the physical address of a three-page region in the guest 1442physical address space. The region must be within the first 4GB of the 1443guest physical address space and must not conflict with any memory slot 1444or any mmio address. The guest may malfunction if it accesses this memory 1445region. 1446 1447This ioctl is required on Intel-based hosts. This is needed on Intel hardware 1448because of a quirk in the virtualization implementation (see the internals 1449documentation when it pops into existence). 1450 1451 1452.. _KVM_ENABLE_CAP: 1453 14544.37 KVM_ENABLE_CAP 1455------------------- 1456 1457:Capability: KVM_CAP_ENABLE_CAP 1458:Architectures: mips, ppc, s390, x86, loongarch 1459:Type: vcpu ioctl 1460:Parameters: struct kvm_enable_cap (in) 1461:Returns: 0 on success; -1 on error 1462 1463:Capability: KVM_CAP_ENABLE_CAP_VM 1464:Architectures: all 1465:Type: vm ioctl 1466:Parameters: struct kvm_enable_cap (in) 1467:Returns: 0 on success; -1 on error 1468 1469.. note:: 1470 1471 Not all extensions are enabled by default. Using this ioctl the application 1472 can enable an extension, making it available to the guest. 1473 1474On systems that do not support this ioctl, it always fails. On systems that 1475do support it, it only works for extensions that are supported for enablement. 1476 1477To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should 1478be used. 1479 1480:: 1481 1482 struct kvm_enable_cap { 1483 /* in */ 1484 __u32 cap; 1485 1486The capability that is supposed to get enabled. 1487 1488:: 1489 1490 __u32 flags; 1491 1492A bitfield indicating future enhancements. Has to be 0 for now. 1493 1494:: 1495 1496 __u64 args[4]; 1497 1498Arguments for enabling a feature. If a feature needs initial values to 1499function properly, this is the place to put them. 1500 1501:: 1502 1503 __u8 pad[64]; 1504 }; 1505 1506The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl 1507for vm-wide capabilities. 1508 15094.38 KVM_GET_MP_STATE 1510--------------------- 1511 1512:Capability: KVM_CAP_MP_STATE 1513:Architectures: x86, s390, arm64, riscv, loongarch 1514:Type: vcpu ioctl 1515:Parameters: struct kvm_mp_state (out) 1516:Returns: 0 on success; -1 on error 1517 1518:: 1519 1520 struct kvm_mp_state { 1521 __u32 mp_state; 1522 }; 1523 1524Returns the vcpu's current "multiprocessing state" (though also valid on 1525uniprocessor guests). 1526 1527Possible values are: 1528 1529 ========================== =============================================== 1530 KVM_MP_STATE_RUNNABLE the vcpu is currently running 1531 [x86,arm64,riscv,loongarch] 1532 KVM_MP_STATE_UNINITIALIZED the vcpu is an application processor (AP) 1533 which has not yet received an INIT signal [x86] 1534 KVM_MP_STATE_INIT_RECEIVED the vcpu has received an INIT signal, and is 1535 now ready for a SIPI [x86] 1536 KVM_MP_STATE_HALTED the vcpu has executed a HLT instruction and 1537 is waiting for an interrupt [x86] 1538 KVM_MP_STATE_SIPI_RECEIVED the vcpu has just received a SIPI (vector 1539 accessible via KVM_GET_VCPU_EVENTS) [x86] 1540 KVM_MP_STATE_STOPPED the vcpu is stopped [s390,arm64,riscv] 1541 KVM_MP_STATE_CHECK_STOP the vcpu is in a special error state [s390] 1542 KVM_MP_STATE_OPERATING the vcpu is operating (running or halted) 1543 [s390] 1544 KVM_MP_STATE_LOAD the vcpu is in a special load/startup state 1545 [s390] 1546 KVM_MP_STATE_SUSPENDED the vcpu is in a suspend state and is waiting 1547 for a wakeup event [arm64] 1548 ========================== =============================================== 1549 1550On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an 1551in-kernel irqchip, the multiprocessing state must be maintained by userspace on 1552these architectures. 1553 1554For arm64: 1555^^^^^^^^^^ 1556 1557If a vCPU is in the KVM_MP_STATE_SUSPENDED state, KVM will emulate the 1558architectural execution of a WFI instruction. 1559 1560If a wakeup event is recognized, KVM will exit to userspace with a 1561KVM_SYSTEM_EVENT exit, where the event type is KVM_SYSTEM_EVENT_WAKEUP. If 1562userspace wants to honor the wakeup, it must set the vCPU's MP state to 1563KVM_MP_STATE_RUNNABLE. If it does not, KVM will continue to await a wakeup 1564event in subsequent calls to KVM_RUN. 1565 1566.. warning:: 1567 1568 If userspace intends to keep the vCPU in a SUSPENDED state, it is 1569 strongly recommended that userspace take action to suppress the 1570 wakeup event (such as masking an interrupt). Otherwise, subsequent 1571 calls to KVM_RUN will immediately exit with a KVM_SYSTEM_EVENT_WAKEUP 1572 event and inadvertently waste CPU cycles. 1573 1574 Additionally, if userspace takes action to suppress a wakeup event, 1575 it is strongly recommended that it also restores the vCPU to its 1576 original state when the vCPU is made RUNNABLE again. For example, 1577 if userspace masked a pending interrupt to suppress the wakeup, 1578 the interrupt should be unmasked before returning control to the 1579 guest. 1580 1581For riscv: 1582^^^^^^^^^^ 1583 1584The only states that are valid are KVM_MP_STATE_STOPPED and 1585KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not. 1586 1587On LoongArch, only the KVM_MP_STATE_RUNNABLE state is used to reflect 1588whether the vcpu is runnable. 1589 15904.39 KVM_SET_MP_STATE 1591--------------------- 1592 1593:Capability: KVM_CAP_MP_STATE 1594:Architectures: x86, s390, arm64, riscv, loongarch 1595:Type: vcpu ioctl 1596:Parameters: struct kvm_mp_state (in) 1597:Returns: 0 on success; -1 on error 1598 1599Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for 1600arguments. 1601 1602On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an 1603in-kernel irqchip, the multiprocessing state must be maintained by userspace on 1604these architectures. 1605 1606For arm64/riscv: 1607^^^^^^^^^^^^^^^^ 1608 1609The only states that are valid are KVM_MP_STATE_STOPPED and 1610KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not. 1611 1612On LoongArch, only the KVM_MP_STATE_RUNNABLE state is used to reflect 1613whether the vcpu is runnable. 1614 16154.40 KVM_SET_IDENTITY_MAP_ADDR 1616------------------------------ 1617 1618:Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR 1619:Architectures: x86 1620:Type: vm ioctl 1621:Parameters: unsigned long identity (in) 1622:Returns: 0 on success, -1 on error 1623 1624This ioctl defines the physical address of a one-page region in the guest 1625physical address space. The region must be within the first 4GB of the 1626guest physical address space and must not conflict with any memory slot 1627or any mmio address. The guest may malfunction if it accesses this memory 1628region. 1629 1630Setting the address to 0 will result in resetting the address to its default 1631(0xfffbc000). 1632 1633This ioctl is required on Intel-based hosts. This is needed on Intel hardware 1634because of a quirk in the virtualization implementation (see the internals 1635documentation when it pops into existence). 1636 1637Fails if any VCPU has already been created. 1638 16394.41 KVM_SET_BOOT_CPU_ID 1640------------------------ 1641 1642:Capability: KVM_CAP_SET_BOOT_CPU_ID 1643:Architectures: x86 1644:Type: vm ioctl 1645:Parameters: unsigned long vcpu_id 1646:Returns: 0 on success, -1 on error 1647 1648Define which vcpu is the Bootstrap Processor (BSP). Values are the same 1649as the vcpu id in KVM_CREATE_VCPU. If this ioctl is not called, the default 1650is vcpu 0. This ioctl has to be called before vcpu creation, 1651otherwise it will return EBUSY error. 1652 1653 16544.42 KVM_GET_XSAVE 1655------------------ 1656 1657:Capability: KVM_CAP_XSAVE 1658:Architectures: x86 1659:Type: vcpu ioctl 1660:Parameters: struct kvm_xsave (out) 1661:Returns: 0 on success, -1 on error 1662 1663 1664:: 1665 1666 struct kvm_xsave { 1667 __u32 region[1024]; 1668 __u32 extra[0]; 1669 }; 1670 1671This ioctl would copy current vcpu's xsave struct to the userspace. 1672 1673 16744.43 KVM_SET_XSAVE 1675------------------ 1676 1677:Capability: KVM_CAP_XSAVE and KVM_CAP_XSAVE2 1678:Architectures: x86 1679:Type: vcpu ioctl 1680:Parameters: struct kvm_xsave (in) 1681:Returns: 0 on success, -1 on error 1682 1683:: 1684 1685 1686 struct kvm_xsave { 1687 __u32 region[1024]; 1688 __u32 extra[0]; 1689 }; 1690 1691This ioctl would copy userspace's xsave struct to the kernel. It copies 1692as many bytes as are returned by KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2), 1693when invoked on the vm file descriptor. The size value returned by 1694KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) will always be at least 4096. 1695Currently, it is only greater than 4096 if a dynamic feature has been 1696enabled with ``arch_prctl()``, but this may change in the future. 1697 1698The offsets of the state save areas in struct kvm_xsave follow the 1699contents of CPUID leaf 0xD on the host. 1700 1701 17024.44 KVM_GET_XCRS 1703----------------- 1704 1705:Capability: KVM_CAP_XCRS 1706:Architectures: x86 1707:Type: vcpu ioctl 1708:Parameters: struct kvm_xcrs (out) 1709:Returns: 0 on success, -1 on error 1710 1711:: 1712 1713 struct kvm_xcr { 1714 __u32 xcr; 1715 __u32 reserved; 1716 __u64 value; 1717 }; 1718 1719 struct kvm_xcrs { 1720 __u32 nr_xcrs; 1721 __u32 flags; 1722 struct kvm_xcr xcrs[KVM_MAX_XCRS]; 1723 __u64 padding[16]; 1724 }; 1725 1726This ioctl would copy current vcpu's xcrs to the userspace. 1727 1728 17294.45 KVM_SET_XCRS 1730----------------- 1731 1732:Capability: KVM_CAP_XCRS 1733:Architectures: x86 1734:Type: vcpu ioctl 1735:Parameters: struct kvm_xcrs (in) 1736:Returns: 0 on success, -1 on error 1737 1738:: 1739 1740 struct kvm_xcr { 1741 __u32 xcr; 1742 __u32 reserved; 1743 __u64 value; 1744 }; 1745 1746 struct kvm_xcrs { 1747 __u32 nr_xcrs; 1748 __u32 flags; 1749 struct kvm_xcr xcrs[KVM_MAX_XCRS]; 1750 __u64 padding[16]; 1751 }; 1752 1753This ioctl would set vcpu's xcr to the value userspace specified. 1754 1755 17564.46 KVM_GET_SUPPORTED_CPUID 1757---------------------------- 1758 1759:Capability: KVM_CAP_EXT_CPUID 1760:Architectures: x86 1761:Type: system ioctl 1762:Parameters: struct kvm_cpuid2 (in/out) 1763:Returns: 0 on success, -1 on error 1764 1765:: 1766 1767 struct kvm_cpuid2 { 1768 __u32 nent; 1769 __u32 padding; 1770 struct kvm_cpuid_entry2 entries[0]; 1771 }; 1772 1773 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0) 1774 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */ 1775 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */ 1776 1777 struct kvm_cpuid_entry2 { 1778 __u32 function; 1779 __u32 index; 1780 __u32 flags; 1781 __u32 eax; 1782 __u32 ebx; 1783 __u32 ecx; 1784 __u32 edx; 1785 __u32 padding[3]; 1786 }; 1787 1788This ioctl returns x86 cpuid features which are supported by both the 1789hardware and kvm in its default configuration. Userspace can use the 1790information returned by this ioctl to construct cpuid information (for 1791KVM_SET_CPUID2) that is consistent with hardware, kernel, and 1792userspace capabilities, and with user requirements (for example, the 1793user may wish to constrain cpuid to emulate older hardware, or for 1794feature consistency across a cluster). 1795 1796Dynamically-enabled feature bits need to be requested with 1797``arch_prctl()`` before calling this ioctl. Feature bits that have not 1798been requested are excluded from the result. 1799 1800Note that certain capabilities, such as KVM_CAP_X86_DISABLE_EXITS, may 1801expose cpuid features (e.g. MONITOR) which are not supported by kvm in 1802its default configuration. If userspace enables such capabilities, it 1803is responsible for modifying the results of this ioctl appropriately. 1804 1805Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure 1806with the 'nent' field indicating the number of entries in the variable-size 1807array 'entries'. If the number of entries is too low to describe the cpu 1808capabilities, an error (E2BIG) is returned. If the number is too high, 1809the 'nent' field is adjusted and an error (ENOMEM) is returned. If the 1810number is just right, the 'nent' field is adjusted to the number of valid 1811entries in the 'entries' array, which is then filled. 1812 1813The entries returned are the host cpuid as returned by the cpuid instruction, 1814with unknown or unsupported features masked out. Some features (for example, 1815x2apic), may not be present in the host cpu, but are exposed by kvm if it can 1816emulate them efficiently. The fields in each entry are defined as follows: 1817 1818 function: 1819 the eax value used to obtain the entry 1820 1821 index: 1822 the ecx value used to obtain the entry (for entries that are 1823 affected by ecx) 1824 1825 flags: 1826 an OR of zero or more of the following: 1827 1828 KVM_CPUID_FLAG_SIGNIFCANT_INDEX: 1829 if the index field is valid 1830 1831 eax, ebx, ecx, edx: 1832 the values returned by the cpuid instruction for 1833 this function/index combination 1834 1835x2APIC (CPUID leaf 1, ecx[21) and TSC deadline timer (CPUID leaf 1, ecx[24]) 1836may be returned as true, but they depend on KVM_CREATE_IRQCHIP for in-kernel 1837emulation of the local APIC. TSC deadline timer support is also reported via:: 1838 1839 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER) 1840 1841if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the 1842feature in userspace, then you can enable the feature for KVM_SET_CPUID2. 1843 1844Enabling x2APIC in KVM_SET_CPUID2 requires KVM_CREATE_IRQCHIP as KVM doesn't 1845support forwarding x2APIC MSR accesses to userspace, i.e. KVM does not support 1846emulating x2APIC in userspace. 1847 18484.47 KVM_PPC_GET_PVINFO 1849----------------------- 1850 1851:Capability: KVM_CAP_PPC_GET_PVINFO 1852:Architectures: ppc 1853:Type: vm ioctl 1854:Parameters: struct kvm_ppc_pvinfo (out) 1855:Returns: 0 on success, !0 on error 1856 1857:: 1858 1859 struct kvm_ppc_pvinfo { 1860 __u32 flags; 1861 __u32 hcall[4]; 1862 __u8 pad[108]; 1863 }; 1864 1865This ioctl fetches PV specific information that need to be passed to the guest 1866using the device tree or other means from vm context. 1867 1868The hcall array defines 4 instructions that make up a hypercall. 1869 1870If any additional field gets added to this structure later on, a bit for that 1871additional piece of information will be set in the flags bitmap. 1872 1873The flags bitmap is defined as:: 1874 1875 /* the host supports the ePAPR idle hcall 1876 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1<<0) 1877 18784.52 KVM_SET_GSI_ROUTING 1879------------------------ 1880 1881:Capability: KVM_CAP_IRQ_ROUTING 1882:Architectures: x86 s390 arm64 1883:Type: vm ioctl 1884:Parameters: struct kvm_irq_routing (in) 1885:Returns: 0 on success, -1 on error 1886 1887Sets the GSI routing table entries, overwriting any previously set entries. 1888 1889On arm64, GSI routing has the following limitation: 1890 1891- GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD. 1892 1893:: 1894 1895 struct kvm_irq_routing { 1896 __u32 nr; 1897 __u32 flags; 1898 struct kvm_irq_routing_entry entries[0]; 1899 }; 1900 1901No flags are specified so far, the corresponding field must be set to zero. 1902 1903:: 1904 1905 struct kvm_irq_routing_entry { 1906 __u32 gsi; 1907 __u32 type; 1908 __u32 flags; 1909 __u32 pad; 1910 union { 1911 struct kvm_irq_routing_irqchip irqchip; 1912 struct kvm_irq_routing_msi msi; 1913 struct kvm_irq_routing_s390_adapter adapter; 1914 struct kvm_irq_routing_hv_sint hv_sint; 1915 struct kvm_irq_routing_xen_evtchn xen_evtchn; 1916 __u32 pad[8]; 1917 } u; 1918 }; 1919 1920 /* gsi routing entry types */ 1921 #define KVM_IRQ_ROUTING_IRQCHIP 1 1922 #define KVM_IRQ_ROUTING_MSI 2 1923 #define KVM_IRQ_ROUTING_S390_ADAPTER 3 1924 #define KVM_IRQ_ROUTING_HV_SINT 4 1925 #define KVM_IRQ_ROUTING_XEN_EVTCHN 5 1926 1927On s390, adding a KVM_IRQ_ROUTING_S390_ADAPTER is rejected on ucontrol VMs with 1928error -EINVAL. 1929 1930flags: 1931 1932- KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry 1933 type, specifies that the devid field contains a valid value. The per-VM 1934 KVM_CAP_MSI_DEVID capability advertises the requirement to provide 1935 the device ID. If this capability is not available, userspace should 1936 never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail. 1937- zero otherwise 1938 1939:: 1940 1941 struct kvm_irq_routing_irqchip { 1942 __u32 irqchip; 1943 __u32 pin; 1944 }; 1945 1946 struct kvm_irq_routing_msi { 1947 __u32 address_lo; 1948 __u32 address_hi; 1949 __u32 data; 1950 union { 1951 __u32 pad; 1952 __u32 devid; 1953 }; 1954 }; 1955 1956If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier 1957for the device that wrote the MSI message. For PCI, this is usually a 1958BDF identifier in the lower 16 bits. 1959 1960On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS 1961feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled, 1962address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of 1963address_hi must be zero. 1964 1965:: 1966 1967 struct kvm_irq_routing_s390_adapter { 1968 __u64 ind_addr; 1969 __u64 summary_addr; 1970 __u64 ind_offset; 1971 __u32 summary_offset; 1972 __u32 adapter_id; 1973 }; 1974 1975 struct kvm_irq_routing_hv_sint { 1976 __u32 vcpu; 1977 __u32 sint; 1978 }; 1979 1980 struct kvm_irq_routing_xen_evtchn { 1981 __u32 port; 1982 __u32 vcpu; 1983 __u32 priority; 1984 }; 1985 1986 1987When KVM_CAP_XEN_HVM includes the KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL bit 1988in its indication of supported features, routing to Xen event channels 1989is supported. Although the priority field is present, only the value 1990KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL is supported, which means delivery by 19912 level event channels. FIFO event channel support may be added in 1992the future. 1993 1994 19954.55 KVM_SET_TSC_KHZ 1996-------------------- 1997 1998:Capability: KVM_CAP_TSC_CONTROL / KVM_CAP_VM_TSC_CONTROL 1999:Architectures: x86 2000:Type: vcpu ioctl / vm ioctl 2001:Parameters: virtual tsc_khz 2002:Returns: 0 on success, -1 on error 2003 2004Specifies the tsc frequency for the virtual machine. The unit of the 2005frequency is KHz. 2006 2007If the KVM_CAP_VM_TSC_CONTROL capability is advertised, this can also 2008be used as a vm ioctl to set the initial tsc frequency of subsequently 2009created vCPUs. 2010 20114.56 KVM_GET_TSC_KHZ 2012-------------------- 2013 2014:Capability: KVM_CAP_GET_TSC_KHZ / KVM_CAP_VM_TSC_CONTROL 2015:Architectures: x86 2016:Type: vcpu ioctl / vm ioctl 2017:Parameters: none 2018:Returns: virtual tsc-khz on success, negative value on error 2019 2020Returns the tsc frequency of the guest. The unit of the return value is 2021KHz. If the host has unstable tsc this ioctl returns -EIO instead as an 2022error. 2023 2024 20254.57 KVM_GET_LAPIC 2026------------------ 2027 2028:Capability: KVM_CAP_IRQCHIP 2029:Architectures: x86 2030:Type: vcpu ioctl 2031:Parameters: struct kvm_lapic_state (out) 2032:Returns: 0 on success, -1 on error 2033 2034:: 2035 2036 #define KVM_APIC_REG_SIZE 0x400 2037 struct kvm_lapic_state { 2038 char regs[KVM_APIC_REG_SIZE]; 2039 }; 2040 2041Reads the Local APIC registers and copies them into the input argument. The 2042data format and layout are the same as documented in the architecture manual. 2043 2044If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is 2045enabled, then the format of APIC_ID register depends on the APIC mode 2046(reported by MSR_IA32_APICBASE) of its VCPU. x2APIC stores APIC ID in 2047the APIC_ID register (bytes 32-35). xAPIC only allows an 8-bit APIC ID 2048which is stored in bits 31-24 of the APIC register, or equivalently in 2049byte 35 of struct kvm_lapic_state's regs field. KVM_GET_LAPIC must then 2050be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR. 2051 2052If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state 2053always uses xAPIC format. 2054 2055 20564.58 KVM_SET_LAPIC 2057------------------ 2058 2059:Capability: KVM_CAP_IRQCHIP 2060:Architectures: x86 2061:Type: vcpu ioctl 2062:Parameters: struct kvm_lapic_state (in) 2063:Returns: 0 on success, -1 on error 2064 2065:: 2066 2067 #define KVM_APIC_REG_SIZE 0x400 2068 struct kvm_lapic_state { 2069 char regs[KVM_APIC_REG_SIZE]; 2070 }; 2071 2072Copies the input argument into the Local APIC registers. The data format 2073and layout are the same as documented in the architecture manual. 2074 2075The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's 2076regs field) depends on the state of the KVM_CAP_X2APIC_API capability. 2077See the note in KVM_GET_LAPIC. 2078 2079 20804.59 KVM_IOEVENTFD 2081------------------ 2082 2083:Capability: KVM_CAP_IOEVENTFD 2084:Architectures: all 2085:Type: vm ioctl 2086:Parameters: struct kvm_ioeventfd (in) 2087:Returns: 0 on success, !0 on error 2088 2089This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address 2090within the guest. A guest write in the registered address will signal the 2091provided event instead of triggering an exit. 2092 2093:: 2094 2095 struct kvm_ioeventfd { 2096 __u64 datamatch; 2097 __u64 addr; /* legal pio/mmio address */ 2098 __u32 len; /* 0, 1, 2, 4, or 8 bytes */ 2099 __s32 fd; 2100 __u32 flags; 2101 __u8 pad[36]; 2102 }; 2103 2104For the special case of virtio-ccw devices on s390, the ioevent is matched 2105to a subchannel/virtqueue tuple instead. 2106 2107The following flags are defined:: 2108 2109 #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch) 2110 #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio) 2111 #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign) 2112 #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \ 2113 (1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify) 2114 2115If datamatch flag is set, the event will be signaled only if the written value 2116to the registered address is equal to datamatch in struct kvm_ioeventfd. 2117 2118For virtio-ccw devices, addr contains the subchannel id and datamatch the 2119virtqueue index. 2120 2121With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and 2122the kernel will ignore the length of guest write and may get a faster vmexit. 2123The speedup may only apply to specific architectures, but the ioeventfd will 2124work anyway. 2125 21264.60 KVM_DIRTY_TLB 2127------------------ 2128 2129:Capability: KVM_CAP_SW_TLB 2130:Architectures: ppc 2131:Type: vcpu ioctl 2132:Parameters: struct kvm_dirty_tlb (in) 2133:Returns: 0 on success, -1 on error 2134 2135:: 2136 2137 struct kvm_dirty_tlb { 2138 __u64 bitmap; 2139 __u32 num_dirty; 2140 }; 2141 2142This must be called whenever userspace has changed an entry in the shared 2143TLB, prior to calling KVM_RUN on the associated vcpu. 2144 2145The "bitmap" field is the userspace address of an array. This array 2146consists of a number of bits, equal to the total number of TLB entries as 2147determined by the last successful call to ``KVM_ENABLE_CAP(KVM_CAP_SW_TLB)``, 2148rounded up to the nearest multiple of 64. 2149 2150Each bit corresponds to one TLB entry, ordered the same as in the shared TLB 2151array. 2152 2153The array is little-endian: the bit 0 is the least significant bit of the 2154first byte, bit 8 is the least significant bit of the second byte, etc. 2155This avoids any complications with differing word sizes. 2156 2157The "num_dirty" field is a performance hint for KVM to determine whether it 2158should skip processing the bitmap and just invalidate everything. It must 2159be set to the number of set bits in the bitmap. 2160 2161 21624.62 KVM_CREATE_SPAPR_TCE 2163------------------------- 2164 2165:Capability: KVM_CAP_SPAPR_TCE 2166:Architectures: powerpc 2167:Type: vm ioctl 2168:Parameters: struct kvm_create_spapr_tce (in) 2169:Returns: file descriptor for manipulating the created TCE table 2170 2171This creates a virtual TCE (translation control entry) table, which 2172is an IOMMU for PAPR-style virtual I/O. It is used to translate 2173logical addresses used in virtual I/O into guest physical addresses, 2174and provides a scatter/gather capability for PAPR virtual I/O. 2175 2176:: 2177 2178 /* for KVM_CAP_SPAPR_TCE */ 2179 struct kvm_create_spapr_tce { 2180 __u64 liobn; 2181 __u32 window_size; 2182 }; 2183 2184The liobn field gives the logical IO bus number for which to create a 2185TCE table. The window_size field specifies the size of the DMA window 2186which this TCE table will translate - the table will contain one 64 2187bit TCE entry for every 4kiB of the DMA window. 2188 2189When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE 2190table has been created using this ioctl(), the kernel will handle it 2191in real mode, updating the TCE table. H_PUT_TCE calls for other 2192liobns will cause a vm exit and must be handled by userspace. 2193 2194The return value is a file descriptor which can be passed to mmap(2) 2195to map the created TCE table into userspace. This lets userspace read 2196the entries written by kernel-handled H_PUT_TCE calls, and also lets 2197userspace update the TCE table directly which is useful in some 2198circumstances. 2199 2200 22014.64 KVM_NMI 2202------------ 2203 2204:Capability: KVM_CAP_USER_NMI 2205:Architectures: x86 2206:Type: vcpu ioctl 2207:Parameters: none 2208:Returns: 0 on success, -1 on error 2209 2210Queues an NMI on the thread's vcpu. Note this is well defined only 2211when KVM_CREATE_IRQCHIP has not been called, since this is an interface 2212between the virtual cpu core and virtual local APIC. After KVM_CREATE_IRQCHIP 2213has been called, this interface is completely emulated within the kernel. 2214 2215To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the 2216following algorithm: 2217 2218 - pause the vcpu 2219 - read the local APIC's state (KVM_GET_LAPIC) 2220 - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1) 2221 - if so, issue KVM_NMI 2222 - resume the vcpu 2223 2224Some guests configure the LINT1 NMI input to cause a panic, aiding in 2225debugging. 2226 2227 22284.65 KVM_S390_UCAS_MAP 2229---------------------- 2230 2231:Capability: KVM_CAP_S390_UCONTROL 2232:Architectures: s390 2233:Type: vcpu ioctl 2234:Parameters: struct kvm_s390_ucas_mapping (in) 2235:Returns: 0 in case of success 2236 2237The parameter is defined like this:: 2238 2239 struct kvm_s390_ucas_mapping { 2240 __u64 user_addr; 2241 __u64 vcpu_addr; 2242 __u64 length; 2243 }; 2244 2245This ioctl maps the memory at "user_addr" with the length "length" to 2246the vcpu's address space starting at "vcpu_addr". All parameters need to 2247be aligned by 1 megabyte. 2248 2249 22504.66 KVM_S390_UCAS_UNMAP 2251------------------------ 2252 2253:Capability: KVM_CAP_S390_UCONTROL 2254:Architectures: s390 2255:Type: vcpu ioctl 2256:Parameters: struct kvm_s390_ucas_mapping (in) 2257:Returns: 0 in case of success 2258 2259The parameter is defined like this:: 2260 2261 struct kvm_s390_ucas_mapping { 2262 __u64 user_addr; 2263 __u64 vcpu_addr; 2264 __u64 length; 2265 }; 2266 2267This ioctl unmaps the memory in the vcpu's address space starting at 2268"vcpu_addr" with the length "length". The field "user_addr" is ignored. 2269All parameters need to be aligned by 1 megabyte. 2270 2271 22724.67 KVM_S390_VCPU_FAULT 2273------------------------ 2274 2275:Capability: KVM_CAP_S390_UCONTROL 2276:Architectures: s390 2277:Type: vcpu ioctl 2278:Parameters: vcpu absolute address (in) 2279:Returns: 0 in case of success 2280 2281This call creates a page table entry on the virtual cpu's address space 2282(for user controlled virtual machines) or the virtual machine's address 2283space (for regular virtual machines). This only works for minor faults, 2284thus it's recommended to access subject memory page via the user page 2285table upfront. This is useful to handle validity intercepts for user 2286controlled virtual machines to fault in the virtual cpu's lowcore pages 2287prior to calling the KVM_RUN ioctl. 2288 2289 22904.68 KVM_SET_ONE_REG 2291-------------------- 2292 2293:Capability: KVM_CAP_ONE_REG 2294:Architectures: all 2295:Type: vcpu ioctl 2296:Parameters: struct kvm_one_reg (in) 2297:Returns: 0 on success, negative value on failure 2298 2299Errors: 2300 2301 ====== ============================================================ 2302 ENOENT no such register 2303 EINVAL invalid register ID, or no such register or used with VMs in 2304 protected virtualization mode on s390 2305 EPERM (arm64) register access not allowed before vcpu finalization 2306 EBUSY (riscv) changing register value not allowed after the vcpu 2307 has run at least once 2308 ====== ============================================================ 2309 2310(These error codes are indicative only: do not rely on a specific error 2311code being returned in a specific situation.) 2312 2313:: 2314 2315 struct kvm_one_reg { 2316 __u64 id; 2317 __u64 addr; 2318 }; 2319 2320Using this ioctl, a single vcpu register can be set to a specific value 2321defined by user space with the passed in struct kvm_one_reg, where id 2322refers to the register identifier as described below and addr is a pointer 2323to a variable with the respective size. There can be architecture agnostic 2324and architecture specific registers. Each have their own range of operation 2325and their own constants and width. To keep track of the implemented 2326registers, find a list below: 2327 2328 ======= =============================== ============ 2329 Arch Register Width (bits) 2330 ======= =============================== ============ 2331 PPC KVM_REG_PPC_HIOR 64 2332 PPC KVM_REG_PPC_IAC1 64 2333 PPC KVM_REG_PPC_IAC2 64 2334 PPC KVM_REG_PPC_IAC3 64 2335 PPC KVM_REG_PPC_IAC4 64 2336 PPC KVM_REG_PPC_DAC1 64 2337 PPC KVM_REG_PPC_DAC2 64 2338 PPC KVM_REG_PPC_DABR 64 2339 PPC KVM_REG_PPC_DSCR 64 2340 PPC KVM_REG_PPC_PURR 64 2341 PPC KVM_REG_PPC_SPURR 64 2342 PPC KVM_REG_PPC_DAR 64 2343 PPC KVM_REG_PPC_DSISR 32 2344 PPC KVM_REG_PPC_AMR 64 2345 PPC KVM_REG_PPC_UAMOR 64 2346 PPC KVM_REG_PPC_MMCR0 64 2347 PPC KVM_REG_PPC_MMCR1 64 2348 PPC KVM_REG_PPC_MMCRA 64 2349 PPC KVM_REG_PPC_MMCR2 64 2350 PPC KVM_REG_PPC_MMCRS 64 2351 PPC KVM_REG_PPC_MMCR3 64 2352 PPC KVM_REG_PPC_SIAR 64 2353 PPC KVM_REG_PPC_SDAR 64 2354 PPC KVM_REG_PPC_SIER 64 2355 PPC KVM_REG_PPC_SIER2 64 2356 PPC KVM_REG_PPC_SIER3 64 2357 PPC KVM_REG_PPC_PMC1 32 2358 PPC KVM_REG_PPC_PMC2 32 2359 PPC KVM_REG_PPC_PMC3 32 2360 PPC KVM_REG_PPC_PMC4 32 2361 PPC KVM_REG_PPC_PMC5 32 2362 PPC KVM_REG_PPC_PMC6 32 2363 PPC KVM_REG_PPC_PMC7 32 2364 PPC KVM_REG_PPC_PMC8 32 2365 PPC KVM_REG_PPC_FPR0 64 2366 ... 2367 PPC KVM_REG_PPC_FPR31 64 2368 PPC KVM_REG_PPC_VR0 128 2369 ... 2370 PPC KVM_REG_PPC_VR31 128 2371 PPC KVM_REG_PPC_VSR0 128 2372 ... 2373 PPC KVM_REG_PPC_VSR31 128 2374 PPC KVM_REG_PPC_FPSCR 64 2375 PPC KVM_REG_PPC_VSCR 32 2376 PPC KVM_REG_PPC_VPA_ADDR 64 2377 PPC KVM_REG_PPC_VPA_SLB 128 2378 PPC KVM_REG_PPC_VPA_DTL 128 2379 PPC KVM_REG_PPC_EPCR 32 2380 PPC KVM_REG_PPC_EPR 32 2381 PPC KVM_REG_PPC_TCR 32 2382 PPC KVM_REG_PPC_TSR 32 2383 PPC KVM_REG_PPC_OR_TSR 32 2384 PPC KVM_REG_PPC_CLEAR_TSR 32 2385 PPC KVM_REG_PPC_MAS0 32 2386 PPC KVM_REG_PPC_MAS1 32 2387 PPC KVM_REG_PPC_MAS2 64 2388 PPC KVM_REG_PPC_MAS7_3 64 2389 PPC KVM_REG_PPC_MAS4 32 2390 PPC KVM_REG_PPC_MAS6 32 2391 PPC KVM_REG_PPC_MMUCFG 32 2392 PPC KVM_REG_PPC_TLB0CFG 32 2393 PPC KVM_REG_PPC_TLB1CFG 32 2394 PPC KVM_REG_PPC_TLB2CFG 32 2395 PPC KVM_REG_PPC_TLB3CFG 32 2396 PPC KVM_REG_PPC_TLB0PS 32 2397 PPC KVM_REG_PPC_TLB1PS 32 2398 PPC KVM_REG_PPC_TLB2PS 32 2399 PPC KVM_REG_PPC_TLB3PS 32 2400 PPC KVM_REG_PPC_EPTCFG 32 2401 PPC KVM_REG_PPC_ICP_STATE 64 2402 PPC KVM_REG_PPC_VP_STATE 128 2403 PPC KVM_REG_PPC_TB_OFFSET 64 2404 PPC KVM_REG_PPC_SPMC1 32 2405 PPC KVM_REG_PPC_SPMC2 32 2406 PPC KVM_REG_PPC_IAMR 64 2407 PPC KVM_REG_PPC_TFHAR 64 2408 PPC KVM_REG_PPC_TFIAR 64 2409 PPC KVM_REG_PPC_TEXASR 64 2410 PPC KVM_REG_PPC_FSCR 64 2411 PPC KVM_REG_PPC_PSPB 32 2412 PPC KVM_REG_PPC_EBBHR 64 2413 PPC KVM_REG_PPC_EBBRR 64 2414 PPC KVM_REG_PPC_BESCR 64 2415 PPC KVM_REG_PPC_TAR 64 2416 PPC KVM_REG_PPC_DPDES 64 2417 PPC KVM_REG_PPC_DAWR 64 2418 PPC KVM_REG_PPC_DAWRX 64 2419 PPC KVM_REG_PPC_CIABR 64 2420 PPC KVM_REG_PPC_IC 64 2421 PPC KVM_REG_PPC_VTB 64 2422 PPC KVM_REG_PPC_CSIGR 64 2423 PPC KVM_REG_PPC_TACR 64 2424 PPC KVM_REG_PPC_TCSCR 64 2425 PPC KVM_REG_PPC_PID 64 2426 PPC KVM_REG_PPC_ACOP 64 2427 PPC KVM_REG_PPC_VRSAVE 32 2428 PPC KVM_REG_PPC_LPCR 32 2429 PPC KVM_REG_PPC_LPCR_64 64 2430 PPC KVM_REG_PPC_PPR 64 2431 PPC KVM_REG_PPC_ARCH_COMPAT 32 2432 PPC KVM_REG_PPC_DABRX 32 2433 PPC KVM_REG_PPC_WORT 64 2434 PPC KVM_REG_PPC_SPRG9 64 2435 PPC KVM_REG_PPC_DBSR 32 2436 PPC KVM_REG_PPC_TIDR 64 2437 PPC KVM_REG_PPC_PSSCR 64 2438 PPC KVM_REG_PPC_DEC_EXPIRY 64 2439 PPC KVM_REG_PPC_PTCR 64 2440 PPC KVM_REG_PPC_HASHKEYR 64 2441 PPC KVM_REG_PPC_HASHPKEYR 64 2442 PPC KVM_REG_PPC_DAWR1 64 2443 PPC KVM_REG_PPC_DAWRX1 64 2444 PPC KVM_REG_PPC_DEXCR 64 2445 PPC KVM_REG_PPC_TM_GPR0 64 2446 ... 2447 PPC KVM_REG_PPC_TM_GPR31 64 2448 PPC KVM_REG_PPC_TM_VSR0 128 2449 ... 2450 PPC KVM_REG_PPC_TM_VSR63 128 2451 PPC KVM_REG_PPC_TM_CR 64 2452 PPC KVM_REG_PPC_TM_LR 64 2453 PPC KVM_REG_PPC_TM_CTR 64 2454 PPC KVM_REG_PPC_TM_FPSCR 64 2455 PPC KVM_REG_PPC_TM_AMR 64 2456 PPC KVM_REG_PPC_TM_PPR 64 2457 PPC KVM_REG_PPC_TM_VRSAVE 64 2458 PPC KVM_REG_PPC_TM_VSCR 32 2459 PPC KVM_REG_PPC_TM_DSCR 64 2460 PPC KVM_REG_PPC_TM_TAR 64 2461 PPC KVM_REG_PPC_TM_XER 64 2462 2463 MIPS KVM_REG_MIPS_R0 64 2464 ... 2465 MIPS KVM_REG_MIPS_R31 64 2466 MIPS KVM_REG_MIPS_HI 64 2467 MIPS KVM_REG_MIPS_LO 64 2468 MIPS KVM_REG_MIPS_PC 64 2469 MIPS KVM_REG_MIPS_CP0_INDEX 32 2470 MIPS KVM_REG_MIPS_CP0_ENTRYLO0 64 2471 MIPS KVM_REG_MIPS_CP0_ENTRYLO1 64 2472 MIPS KVM_REG_MIPS_CP0_CONTEXT 64 2473 MIPS KVM_REG_MIPS_CP0_CONTEXTCONFIG 32 2474 MIPS KVM_REG_MIPS_CP0_USERLOCAL 64 2475 MIPS KVM_REG_MIPS_CP0_XCONTEXTCONFIG 64 2476 MIPS KVM_REG_MIPS_CP0_PAGEMASK 32 2477 MIPS KVM_REG_MIPS_CP0_PAGEGRAIN 32 2478 MIPS KVM_REG_MIPS_CP0_SEGCTL0 64 2479 MIPS KVM_REG_MIPS_CP0_SEGCTL1 64 2480 MIPS KVM_REG_MIPS_CP0_SEGCTL2 64 2481 MIPS KVM_REG_MIPS_CP0_PWBASE 64 2482 MIPS KVM_REG_MIPS_CP0_PWFIELD 64 2483 MIPS KVM_REG_MIPS_CP0_PWSIZE 64 2484 MIPS KVM_REG_MIPS_CP0_WIRED 32 2485 MIPS KVM_REG_MIPS_CP0_PWCTL 32 2486 MIPS KVM_REG_MIPS_CP0_HWRENA 32 2487 MIPS KVM_REG_MIPS_CP0_BADVADDR 64 2488 MIPS KVM_REG_MIPS_CP0_BADINSTR 32 2489 MIPS KVM_REG_MIPS_CP0_BADINSTRP 32 2490 MIPS KVM_REG_MIPS_CP0_COUNT 32 2491 MIPS KVM_REG_MIPS_CP0_ENTRYHI 64 2492 MIPS KVM_REG_MIPS_CP0_COMPARE 32 2493 MIPS KVM_REG_MIPS_CP0_STATUS 32 2494 MIPS KVM_REG_MIPS_CP0_INTCTL 32 2495 MIPS KVM_REG_MIPS_CP0_CAUSE 32 2496 MIPS KVM_REG_MIPS_CP0_EPC 64 2497 MIPS KVM_REG_MIPS_CP0_PRID 32 2498 MIPS KVM_REG_MIPS_CP0_EBASE 64 2499 MIPS KVM_REG_MIPS_CP0_CONFIG 32 2500 MIPS KVM_REG_MIPS_CP0_CONFIG1 32 2501 MIPS KVM_REG_MIPS_CP0_CONFIG2 32 2502 MIPS KVM_REG_MIPS_CP0_CONFIG3 32 2503 MIPS KVM_REG_MIPS_CP0_CONFIG4 32 2504 MIPS KVM_REG_MIPS_CP0_CONFIG5 32 2505 MIPS KVM_REG_MIPS_CP0_CONFIG7 32 2506 MIPS KVM_REG_MIPS_CP0_XCONTEXT 64 2507 MIPS KVM_REG_MIPS_CP0_ERROREPC 64 2508 MIPS KVM_REG_MIPS_CP0_KSCRATCH1 64 2509 MIPS KVM_REG_MIPS_CP0_KSCRATCH2 64 2510 MIPS KVM_REG_MIPS_CP0_KSCRATCH3 64 2511 MIPS KVM_REG_MIPS_CP0_KSCRATCH4 64 2512 MIPS KVM_REG_MIPS_CP0_KSCRATCH5 64 2513 MIPS KVM_REG_MIPS_CP0_KSCRATCH6 64 2514 MIPS KVM_REG_MIPS_CP0_MAAR(0..63) 64 2515 MIPS KVM_REG_MIPS_COUNT_CTL 64 2516 MIPS KVM_REG_MIPS_COUNT_RESUME 64 2517 MIPS KVM_REG_MIPS_COUNT_HZ 64 2518 MIPS KVM_REG_MIPS_FPR_32(0..31) 32 2519 MIPS KVM_REG_MIPS_FPR_64(0..31) 64 2520 MIPS KVM_REG_MIPS_VEC_128(0..31) 128 2521 MIPS KVM_REG_MIPS_FCR_IR 32 2522 MIPS KVM_REG_MIPS_FCR_CSR 32 2523 MIPS KVM_REG_MIPS_MSA_IR 32 2524 MIPS KVM_REG_MIPS_MSA_CSR 32 2525 ======= =============================== ============ 2526 2527ARM registers are mapped using the lower 32 bits. The upper 16 of that 2528is the register group type, or coprocessor number: 2529 2530ARM core registers have the following id bit patterns:: 2531 2532 0x4020 0000 0010 <index into the kvm_regs struct:16> 2533 2534ARM 32-bit CP15 registers have the following id bit patterns:: 2535 2536 0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3> 2537 2538ARM 64-bit CP15 registers have the following id bit patterns:: 2539 2540 0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3> 2541 2542ARM CCSIDR registers are demultiplexed by CSSELR value:: 2543 2544 0x4020 0000 0011 00 <csselr:8> 2545 2546ARM 32-bit VFP control registers have the following id bit patterns:: 2547 2548 0x4020 0000 0012 1 <regno:12> 2549 2550ARM 64-bit FP registers have the following id bit patterns:: 2551 2552 0x4030 0000 0012 0 <regno:12> 2553 2554ARM firmware pseudo-registers have the following bit pattern:: 2555 2556 0x4030 0000 0014 <regno:16> 2557 2558 2559arm64 registers are mapped using the lower 32 bits. The upper 16 of 2560that is the register group type, or coprocessor number: 2561 2562arm64 core/FP-SIMD registers have the following id bit patterns. Note 2563that the size of the access is variable, as the kvm_regs structure 2564contains elements ranging from 32 to 128 bits. The index is a 32bit 2565value in the kvm_regs structure seen as a 32bit array:: 2566 2567 0x60x0 0000 0010 <index into the kvm_regs struct:16> 2568 2569Specifically: 2570 2571======================= ========= ===== ======================================= 2572 Encoding Register Bits kvm_regs member 2573======================= ========= ===== ======================================= 2574 0x6030 0000 0010 0000 X0 64 regs.regs[0] 2575 0x6030 0000 0010 0002 X1 64 regs.regs[1] 2576 ... 2577 0x6030 0000 0010 003c X30 64 regs.regs[30] 2578 0x6030 0000 0010 003e SP 64 regs.sp 2579 0x6030 0000 0010 0040 PC 64 regs.pc 2580 0x6030 0000 0010 0042 PSTATE 64 regs.pstate 2581 0x6030 0000 0010 0044 SP_EL1 64 sp_el1 2582 0x6030 0000 0010 0046 ELR_EL1 64 elr_el1 2583 0x6030 0000 0010 0048 SPSR_EL1 64 spsr[KVM_SPSR_EL1] (alias SPSR_SVC) 2584 0x6030 0000 0010 004a SPSR_ABT 64 spsr[KVM_SPSR_ABT] 2585 0x6030 0000 0010 004c SPSR_UND 64 spsr[KVM_SPSR_UND] 2586 0x6030 0000 0010 004e SPSR_IRQ 64 spsr[KVM_SPSR_IRQ] 2587 0x6030 0000 0010 0050 SPSR_FIQ 64 spsr[KVM_SPSR_FIQ] 2588 0x6040 0000 0010 0054 V0 128 fp_regs.vregs[0] [1]_ 2589 0x6040 0000 0010 0058 V1 128 fp_regs.vregs[1] [1]_ 2590 ... 2591 0x6040 0000 0010 00d0 V31 128 fp_regs.vregs[31] [1]_ 2592 0x6020 0000 0010 00d4 FPSR 32 fp_regs.fpsr 2593 0x6020 0000 0010 00d5 FPCR 32 fp_regs.fpcr 2594======================= ========= ===== ======================================= 2595 2596.. [1] These encodings are not accepted for SVE-enabled vcpus. See 2597 :ref:`KVM_ARM_VCPU_INIT`. 2598 2599 The equivalent register content can be accessed via bits [127:0] of 2600 the corresponding SVE Zn registers instead for vcpus that have SVE 2601 enabled (see below). 2602 2603arm64 CCSIDR registers are demultiplexed by CSSELR value:: 2604 2605 0x6020 0000 0011 00 <csselr:8> 2606 2607arm64 system registers have the following id bit patterns:: 2608 2609 0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3> 2610 2611.. warning:: 2612 2613 Two system register IDs do not follow the specified pattern. These 2614 are KVM_REG_ARM_TIMER_CVAL and KVM_REG_ARM_TIMER_CNT, which map to 2615 system registers CNTV_CVAL_EL0 and CNTVCT_EL0 respectively. These 2616 two had their values accidentally swapped, which means TIMER_CVAL is 2617 derived from the register encoding for CNTVCT_EL0 and TIMER_CNT is 2618 derived from the register encoding for CNTV_CVAL_EL0. As this is 2619 API, it must remain this way. 2620 2621arm64 firmware pseudo-registers have the following bit pattern:: 2622 2623 0x6030 0000 0014 <regno:16> 2624 2625arm64 SVE registers have the following bit patterns:: 2626 2627 0x6080 0000 0015 00 <n:5> <slice:5> Zn bits[2048*slice + 2047 : 2048*slice] 2628 0x6050 0000 0015 04 <n:4> <slice:5> Pn bits[256*slice + 255 : 256*slice] 2629 0x6050 0000 0015 060 <slice:5> FFR bits[256*slice + 255 : 256*slice] 2630 0x6060 0000 0015 ffff KVM_REG_ARM64_SVE_VLS pseudo-register 2631 2632Access to register IDs where 2048 * slice >= 128 * max_vq will fail with 2633ENOENT. max_vq is the vcpu's maximum supported vector length in 128-bit 2634quadwords: see [2]_ below. 2635 2636These registers are only accessible on vcpus for which SVE is enabled. 2637See KVM_ARM_VCPU_INIT for details. 2638 2639In addition, except for KVM_REG_ARM64_SVE_VLS, these registers are not 2640accessible until the vcpu's SVE configuration has been finalized 2641using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). See KVM_ARM_VCPU_INIT 2642and KVM_ARM_VCPU_FINALIZE for more information about this procedure. 2643 2644KVM_REG_ARM64_SVE_VLS is a pseudo-register that allows the set of vector 2645lengths supported by the vcpu to be discovered and configured by 2646userspace. When transferred to or from user memory via KVM_GET_ONE_REG 2647or KVM_SET_ONE_REG, the value of this register is of type 2648__u64[KVM_ARM64_SVE_VLS_WORDS], and encodes the set of vector lengths as 2649follows:: 2650 2651 __u64 vector_lengths[KVM_ARM64_SVE_VLS_WORDS]; 2652 2653 if (vq >= SVE_VQ_MIN && vq <= SVE_VQ_MAX && 2654 ((vector_lengths[(vq - KVM_ARM64_SVE_VQ_MIN) / 64] >> 2655 ((vq - KVM_ARM64_SVE_VQ_MIN) % 64)) & 1)) 2656 /* Vector length vq * 16 bytes supported */ 2657 else 2658 /* Vector length vq * 16 bytes not supported */ 2659 2660.. [2] The maximum value vq for which the above condition is true is 2661 max_vq. This is the maximum vector length available to the guest on 2662 this vcpu, and determines which register slices are visible through 2663 this ioctl interface. 2664 2665(See Documentation/arch/arm64/sve.rst for an explanation of the "vq" 2666nomenclature.) 2667 2668KVM_REG_ARM64_SVE_VLS is only accessible after KVM_ARM_VCPU_INIT. 2669KVM_ARM_VCPU_INIT initialises it to the best set of vector lengths that 2670the host supports. 2671 2672Userspace may subsequently modify it if desired until the vcpu's SVE 2673configuration is finalized using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). 2674 2675Apart from simply removing all vector lengths from the host set that 2676exceed some value, support for arbitrarily chosen sets of vector lengths 2677is hardware-dependent and may not be available. Attempting to configure 2678an invalid set of vector lengths via KVM_SET_ONE_REG will fail with 2679EINVAL. 2680 2681After the vcpu's SVE configuration is finalized, further attempts to 2682write this register will fail with EPERM. 2683 2684arm64 bitmap feature firmware pseudo-registers have the following bit pattern:: 2685 2686 0x6030 0000 0016 <regno:16> 2687 2688The bitmap feature firmware registers exposes the hypercall services that 2689are available for userspace to configure. The set bits corresponds to the 2690services that are available for the guests to access. By default, KVM 2691sets all the supported bits during VM initialization. The userspace can 2692discover the available services via KVM_GET_ONE_REG, and write back the 2693bitmap corresponding to the features that it wishes guests to see via 2694KVM_SET_ONE_REG. 2695 2696Note: These registers are immutable once any of the vCPUs of the VM has 2697run at least once. A KVM_SET_ONE_REG in such a scenario will return 2698a -EBUSY to userspace. 2699 2700(See Documentation/virt/kvm/arm/hypercalls.rst for more details.) 2701 2702 2703MIPS registers are mapped using the lower 32 bits. The upper 16 of that is 2704the register group type: 2705 2706MIPS core registers (see above) have the following id bit patterns:: 2707 2708 0x7030 0000 0000 <reg:16> 2709 2710MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit 2711patterns depending on whether they're 32-bit or 64-bit registers:: 2712 2713 0x7020 0000 0001 00 <reg:5> <sel:3> (32-bit) 2714 0x7030 0000 0001 00 <reg:5> <sel:3> (64-bit) 2715 2716Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_MIPS_CP0_ENTRYLO1 are the MIPS64 2717versions of the EntryLo registers regardless of the word size of the host 2718hardware, host kernel, guest, and whether XPA is present in the guest, i.e. 2719with the RI and XI bits (if they exist) in bits 63 and 62 respectively, and 2720the PFNX field starting at bit 30. 2721 2722MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) above) have the following id bit 2723patterns:: 2724 2725 0x7030 0000 0001 01 <reg:8> 2726 2727MIPS KVM control registers (see above) have the following id bit patterns:: 2728 2729 0x7030 0000 0002 <reg:16> 2730 2731MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following 2732id bit patterns depending on the size of the register being accessed. They are 2733always accessed according to the current guest FPU mode (Status.FR and 2734Config5.FRE), i.e. as the guest would see them, and they become unpredictable 2735if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector 2736registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they 2737overlap the FPU registers:: 2738 2739 0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers) 2740 0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers) 2741 0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers) 2742 2743MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the 2744following id bit patterns:: 2745 2746 0x7020 0000 0003 01 <0:3> <reg:5> 2747 2748MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the 2749following id bit patterns:: 2750 2751 0x7020 0000 0003 02 <0:3> <reg:5> 2752 2753RISC-V registers are mapped using the lower 32 bits. The upper 8 bits of 2754that is the register group type. 2755 2756RISC-V config registers are meant for configuring a Guest VCPU and it has 2757the following id bit patterns:: 2758 2759 0x8020 0000 01 <index into the kvm_riscv_config struct:24> (32bit Host) 2760 0x8030 0000 01 <index into the kvm_riscv_config struct:24> (64bit Host) 2761 2762Following are the RISC-V config registers: 2763 2764======================= ========= ============================================= 2765 Encoding Register Description 2766======================= ========= ============================================= 2767 0x80x0 0000 0100 0000 isa ISA feature bitmap of Guest VCPU 2768======================= ========= ============================================= 2769 2770The isa config register can be read anytime but can only be written before 2771a Guest VCPU runs. It will have ISA feature bits matching underlying host 2772set by default. 2773 2774RISC-V core registers represent the general execution state of a Guest VCPU 2775and it has the following id bit patterns:: 2776 2777 0x8020 0000 02 <index into the kvm_riscv_core struct:24> (32bit Host) 2778 0x8030 0000 02 <index into the kvm_riscv_core struct:24> (64bit Host) 2779 2780Following are the RISC-V core registers: 2781 2782======================= ========= ============================================= 2783 Encoding Register Description 2784======================= ========= ============================================= 2785 0x80x0 0000 0200 0000 regs.pc Program counter 2786 0x80x0 0000 0200 0001 regs.ra Return address 2787 0x80x0 0000 0200 0002 regs.sp Stack pointer 2788 0x80x0 0000 0200 0003 regs.gp Global pointer 2789 0x80x0 0000 0200 0004 regs.tp Task pointer 2790 0x80x0 0000 0200 0005 regs.t0 Caller saved register 0 2791 0x80x0 0000 0200 0006 regs.t1 Caller saved register 1 2792 0x80x0 0000 0200 0007 regs.t2 Caller saved register 2 2793 0x80x0 0000 0200 0008 regs.s0 Callee saved register 0 2794 0x80x0 0000 0200 0009 regs.s1 Callee saved register 1 2795 0x80x0 0000 0200 000a regs.a0 Function argument (or return value) 0 2796 0x80x0 0000 0200 000b regs.a1 Function argument (or return value) 1 2797 0x80x0 0000 0200 000c regs.a2 Function argument 2 2798 0x80x0 0000 0200 000d regs.a3 Function argument 3 2799 0x80x0 0000 0200 000e regs.a4 Function argument 4 2800 0x80x0 0000 0200 000f regs.a5 Function argument 5 2801 0x80x0 0000 0200 0010 regs.a6 Function argument 6 2802 0x80x0 0000 0200 0011 regs.a7 Function argument 7 2803 0x80x0 0000 0200 0012 regs.s2 Callee saved register 2 2804 0x80x0 0000 0200 0013 regs.s3 Callee saved register 3 2805 0x80x0 0000 0200 0014 regs.s4 Callee saved register 4 2806 0x80x0 0000 0200 0015 regs.s5 Callee saved register 5 2807 0x80x0 0000 0200 0016 regs.s6 Callee saved register 6 2808 0x80x0 0000 0200 0017 regs.s7 Callee saved register 7 2809 0x80x0 0000 0200 0018 regs.s8 Callee saved register 8 2810 0x80x0 0000 0200 0019 regs.s9 Callee saved register 9 2811 0x80x0 0000 0200 001a regs.s10 Callee saved register 10 2812 0x80x0 0000 0200 001b regs.s11 Callee saved register 11 2813 0x80x0 0000 0200 001c regs.t3 Caller saved register 3 2814 0x80x0 0000 0200 001d regs.t4 Caller saved register 4 2815 0x80x0 0000 0200 001e regs.t5 Caller saved register 5 2816 0x80x0 0000 0200 001f regs.t6 Caller saved register 6 2817 0x80x0 0000 0200 0020 mode Privilege mode (1 = S-mode or 0 = U-mode) 2818======================= ========= ============================================= 2819 2820RISC-V csr registers represent the supervisor mode control/status registers 2821of a Guest VCPU and it has the following id bit patterns:: 2822 2823 0x8020 0000 03 <index into the kvm_riscv_csr struct:24> (32bit Host) 2824 0x8030 0000 03 <index into the kvm_riscv_csr struct:24> (64bit Host) 2825 2826Following are the RISC-V csr registers: 2827 2828======================= ========= ============================================= 2829 Encoding Register Description 2830======================= ========= ============================================= 2831 0x80x0 0000 0300 0000 sstatus Supervisor status 2832 0x80x0 0000 0300 0001 sie Supervisor interrupt enable 2833 0x80x0 0000 0300 0002 stvec Supervisor trap vector base 2834 0x80x0 0000 0300 0003 sscratch Supervisor scratch register 2835 0x80x0 0000 0300 0004 sepc Supervisor exception program counter 2836 0x80x0 0000 0300 0005 scause Supervisor trap cause 2837 0x80x0 0000 0300 0006 stval Supervisor bad address or instruction 2838 0x80x0 0000 0300 0007 sip Supervisor interrupt pending 2839 0x80x0 0000 0300 0008 satp Supervisor address translation and protection 2840======================= ========= ============================================= 2841 2842RISC-V timer registers represent the timer state of a Guest VCPU and it has 2843the following id bit patterns:: 2844 2845 0x8030 0000 04 <index into the kvm_riscv_timer struct:24> 2846 2847Following are the RISC-V timer registers: 2848 2849======================= ========= ============================================= 2850 Encoding Register Description 2851======================= ========= ============================================= 2852 0x8030 0000 0400 0000 frequency Time base frequency (read-only) 2853 0x8030 0000 0400 0001 time Time value visible to Guest 2854 0x8030 0000 0400 0002 compare Time compare programmed by Guest 2855 0x8030 0000 0400 0003 state Time compare state (1 = ON or 0 = OFF) 2856======================= ========= ============================================= 2857 2858RISC-V F-extension registers represent the single precision floating point 2859state of a Guest VCPU and it has the following id bit patterns:: 2860 2861 0x8020 0000 05 <index into the __riscv_f_ext_state struct:24> 2862 2863Following are the RISC-V F-extension registers: 2864 2865======================= ========= ============================================= 2866 Encoding Register Description 2867======================= ========= ============================================= 2868 0x8020 0000 0500 0000 f[0] Floating point register 0 2869 ... 2870 0x8020 0000 0500 001f f[31] Floating point register 31 2871 0x8020 0000 0500 0020 fcsr Floating point control and status register 2872======================= ========= ============================================= 2873 2874RISC-V D-extension registers represent the double precision floating point 2875state of a Guest VCPU and it has the following id bit patterns:: 2876 2877 0x8020 0000 06 <index into the __riscv_d_ext_state struct:24> (fcsr) 2878 0x8030 0000 06 <index into the __riscv_d_ext_state struct:24> (non-fcsr) 2879 2880Following are the RISC-V D-extension registers: 2881 2882======================= ========= ============================================= 2883 Encoding Register Description 2884======================= ========= ============================================= 2885 0x8030 0000 0600 0000 f[0] Floating point register 0 2886 ... 2887 0x8030 0000 0600 001f f[31] Floating point register 31 2888 0x8020 0000 0600 0020 fcsr Floating point control and status register 2889======================= ========= ============================================= 2890 2891LoongArch registers are mapped using the lower 32 bits. The upper 16 bits of 2892that is the register group type. 2893 2894LoongArch csr registers are used to control guest cpu or get status of guest 2895cpu, and they have the following id bit patterns:: 2896 2897 0x9030 0000 0001 00 <reg:5> <sel:3> (64-bit) 2898 2899LoongArch KVM control registers are used to implement some new defined functions 2900such as set vcpu counter or reset vcpu, and they have the following id bit patterns:: 2901 2902 0x9030 0000 0002 <reg:16> 2903 2904 29054.69 KVM_GET_ONE_REG 2906-------------------- 2907 2908:Capability: KVM_CAP_ONE_REG 2909:Architectures: all 2910:Type: vcpu ioctl 2911:Parameters: struct kvm_one_reg (in and out) 2912:Returns: 0 on success, negative value on failure 2913 2914Errors include: 2915 2916 ======== ============================================================ 2917 ENOENT no such register 2918 EINVAL invalid register ID, or no such register or used with VMs in 2919 protected virtualization mode on s390 2920 EPERM (arm64) register access not allowed before vcpu finalization 2921 ======== ============================================================ 2922 2923(These error codes are indicative only: do not rely on a specific error 2924code being returned in a specific situation.) 2925 2926This ioctl allows to receive the value of a single register implemented 2927in a vcpu. The register to read is indicated by the "id" field of the 2928kvm_one_reg struct passed in. On success, the register value can be found 2929at the memory location pointed to by "addr". 2930 2931The list of registers accessible using this interface is identical to the 2932list in 4.68. 2933 2934 29354.70 KVM_KVMCLOCK_CTRL 2936---------------------- 2937 2938:Capability: KVM_CAP_KVMCLOCK_CTRL 2939:Architectures: Any that implement pvclocks (currently x86 only) 2940:Type: vcpu ioctl 2941:Parameters: None 2942:Returns: 0 on success, -1 on error 2943 2944This ioctl sets a flag accessible to the guest indicating that the specified 2945vCPU has been paused by the host userspace. 2946 2947The host will set a flag in the pvclock structure that is checked from the 2948soft lockup watchdog. The flag is part of the pvclock structure that is 2949shared between guest and host, specifically the second bit of the flags 2950field of the pvclock_vcpu_time_info structure. It will be set exclusively by 2951the host and read/cleared exclusively by the guest. The guest operation of 2952checking and clearing the flag must be an atomic operation so 2953load-link/store-conditional, or equivalent must be used. There are two cases 2954where the guest will clear the flag: when the soft lockup watchdog timer resets 2955itself or when a soft lockup is detected. This ioctl can be called any time 2956after pausing the vcpu, but before it is resumed. 2957 2958 29594.71 KVM_SIGNAL_MSI 2960------------------- 2961 2962:Capability: KVM_CAP_SIGNAL_MSI 2963:Architectures: x86 arm64 2964:Type: vm ioctl 2965:Parameters: struct kvm_msi (in) 2966:Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error 2967 2968Directly inject a MSI message. Only valid with in-kernel irqchip that handles 2969MSI messages. 2970 2971:: 2972 2973 struct kvm_msi { 2974 __u32 address_lo; 2975 __u32 address_hi; 2976 __u32 data; 2977 __u32 flags; 2978 __u32 devid; 2979 __u8 pad[12]; 2980 }; 2981 2982flags: 2983 KVM_MSI_VALID_DEVID: devid contains a valid value. The per-VM 2984 KVM_CAP_MSI_DEVID capability advertises the requirement to provide 2985 the device ID. If this capability is not available, userspace 2986 should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail. 2987 2988If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier 2989for the device that wrote the MSI message. For PCI, this is usually a 2990BDF identifier in the lower 16 bits. 2991 2992On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS 2993feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled, 2994address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of 2995address_hi must be zero. 2996 2997 29984.71 KVM_CREATE_PIT2 2999-------------------- 3000 3001:Capability: KVM_CAP_PIT2 3002:Architectures: x86 3003:Type: vm ioctl 3004:Parameters: struct kvm_pit_config (in) 3005:Returns: 0 on success, -1 on error 3006 3007Creates an in-kernel device model for the i8254 PIT. This call is only valid 3008after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following 3009parameters have to be passed:: 3010 3011 struct kvm_pit_config { 3012 __u32 flags; 3013 __u32 pad[15]; 3014 }; 3015 3016Valid flags are:: 3017 3018 #define KVM_PIT_SPEAKER_DUMMY 1 /* emulate speaker port stub */ 3019 3020PIT timer interrupts may use a per-VM kernel thread for injection. If it 3021exists, this thread will have a name of the following pattern:: 3022 3023 kvm-pit/<owner-process-pid> 3024 3025When running a guest with elevated priorities, the scheduling parameters of 3026this thread may have to be adjusted accordingly. 3027 3028This IOCTL replaces the obsolete KVM_CREATE_PIT. 3029 3030 30314.72 KVM_GET_PIT2 3032----------------- 3033 3034:Capability: KVM_CAP_PIT_STATE2 3035:Architectures: x86 3036:Type: vm ioctl 3037:Parameters: struct kvm_pit_state2 (out) 3038:Returns: 0 on success, -1 on error 3039 3040Retrieves the state of the in-kernel PIT model. Only valid after 3041KVM_CREATE_PIT2. The state is returned in the following structure:: 3042 3043 struct kvm_pit_state2 { 3044 struct kvm_pit_channel_state channels[3]; 3045 __u32 flags; 3046 __u32 reserved[9]; 3047 }; 3048 3049Valid flags are:: 3050 3051 /* disable PIT in HPET legacy mode */ 3052 #define KVM_PIT_FLAGS_HPET_LEGACY 0x00000001 3053 /* speaker port data bit enabled */ 3054 #define KVM_PIT_FLAGS_SPEAKER_DATA_ON 0x00000002 3055 3056This IOCTL replaces the obsolete KVM_GET_PIT. 3057 3058 30594.73 KVM_SET_PIT2 3060----------------- 3061 3062:Capability: KVM_CAP_PIT_STATE2 3063:Architectures: x86 3064:Type: vm ioctl 3065:Parameters: struct kvm_pit_state2 (in) 3066:Returns: 0 on success, -1 on error 3067 3068Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2. 3069See KVM_GET_PIT2 for details on struct kvm_pit_state2. 3070 3071This IOCTL replaces the obsolete KVM_SET_PIT. 3072 3073 30744.74 KVM_PPC_GET_SMMU_INFO 3075-------------------------- 3076 3077:Capability: KVM_CAP_PPC_GET_SMMU_INFO 3078:Architectures: powerpc 3079:Type: vm ioctl 3080:Parameters: None 3081:Returns: 0 on success, -1 on error 3082 3083This populates and returns a structure describing the features of 3084the "Server" class MMU emulation supported by KVM. 3085This can in turn be used by userspace to generate the appropriate 3086device-tree properties for the guest operating system. 3087 3088The structure contains some global information, followed by an 3089array of supported segment page sizes:: 3090 3091 struct kvm_ppc_smmu_info { 3092 __u64 flags; 3093 __u32 slb_size; 3094 __u32 pad; 3095 struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ]; 3096 }; 3097 3098The supported flags are: 3099 3100 - KVM_PPC_PAGE_SIZES_REAL: 3101 When that flag is set, guest page sizes must "fit" the backing 3102 store page sizes. When not set, any page size in the list can 3103 be used regardless of how they are backed by userspace. 3104 3105 - KVM_PPC_1T_SEGMENTS 3106 The emulated MMU supports 1T segments in addition to the 3107 standard 256M ones. 3108 3109 - KVM_PPC_NO_HASH 3110 This flag indicates that HPT guests are not supported by KVM, 3111 thus all guests must use radix MMU mode. 3112 3113The "slb_size" field indicates how many SLB entries are supported 3114 3115The "sps" array contains 8 entries indicating the supported base 3116page sizes for a segment in increasing order. Each entry is defined 3117as follow:: 3118 3119 struct kvm_ppc_one_seg_page_size { 3120 __u32 page_shift; /* Base page shift of segment (or 0) */ 3121 __u32 slb_enc; /* SLB encoding for BookS */ 3122 struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ]; 3123 }; 3124 3125An entry with a "page_shift" of 0 is unused. Because the array is 3126organized in increasing order, a lookup can stop when encountering 3127such an entry. 3128 3129The "slb_enc" field provides the encoding to use in the SLB for the 3130page size. The bits are in positions such as the value can directly 3131be OR'ed into the "vsid" argument of the slbmte instruction. 3132 3133The "enc" array is a list which for each of those segment base page 3134size provides the list of supported actual page sizes (which can be 3135only larger or equal to the base page size), along with the 3136corresponding encoding in the hash PTE. Similarly, the array is 31378 entries sorted by increasing sizes and an entry with a "0" shift 3138is an empty entry and a terminator:: 3139 3140 struct kvm_ppc_one_page_size { 3141 __u32 page_shift; /* Page shift (or 0) */ 3142 __u32 pte_enc; /* Encoding in the HPTE (>>12) */ 3143 }; 3144 3145The "pte_enc" field provides a value that can OR'ed into the hash 3146PTE's RPN field (ie, it needs to be shifted left by 12 to OR it 3147into the hash PTE second double word). 3148 31494.75 KVM_IRQFD 3150-------------- 3151 3152:Capability: KVM_CAP_IRQFD 3153:Architectures: x86 s390 arm64 3154:Type: vm ioctl 3155:Parameters: struct kvm_irqfd (in) 3156:Returns: 0 on success, -1 on error 3157 3158Allows setting an eventfd to directly trigger a guest interrupt. 3159kvm_irqfd.fd specifies the file descriptor to use as the eventfd and 3160kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When 3161an event is triggered on the eventfd, an interrupt is injected into 3162the guest using the specified gsi pin. The irqfd is removed using 3163the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd 3164and kvm_irqfd.gsi. 3165 3166With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify 3167mechanism allowing emulation of level-triggered, irqfd-based 3168interrupts. When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an 3169additional eventfd in the kvm_irqfd.resamplefd field. When operating 3170in resample mode, posting of an interrupt through kvm_irq.fd asserts 3171the specified gsi in the irqchip. When the irqchip is resampled, such 3172as from an EOI, the gsi is de-asserted and the user is notified via 3173kvm_irqfd.resamplefd. It is the user's responsibility to re-queue 3174the interrupt if the device making use of it still requires service. 3175Note that closing the resamplefd is not sufficient to disable the 3176irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment 3177and need not be specified with KVM_IRQFD_FLAG_DEASSIGN. 3178 3179On arm64, gsi routing being supported, the following can happen: 3180 3181- in case no routing entry is associated to this gsi, injection fails 3182- in case the gsi is associated to an irqchip routing entry, 3183 irqchip.pin + 32 corresponds to the injected SPI ID. 3184- in case the gsi is associated to an MSI routing entry, the MSI 3185 message and device ID are translated into an LPI (support restricted 3186 to GICv3 ITS in-kernel emulation). 3187 31884.76 KVM_PPC_ALLOCATE_HTAB 3189-------------------------- 3190 3191:Capability: KVM_CAP_PPC_ALLOC_HTAB 3192:Architectures: powerpc 3193:Type: vm ioctl 3194:Parameters: Pointer to u32 containing hash table order (in/out) 3195:Returns: 0 on success, -1 on error 3196 3197This requests the host kernel to allocate an MMU hash table for a 3198guest using the PAPR paravirtualization interface. This only does 3199anything if the kernel is configured to use the Book 3S HV style of 3200virtualization. Otherwise the capability doesn't exist and the ioctl 3201returns an ENOTTY error. The rest of this description assumes Book 3S 3202HV. 3203 3204There must be no vcpus running when this ioctl is called; if there 3205are, it will do nothing and return an EBUSY error. 3206 3207The parameter is a pointer to a 32-bit unsigned integer variable 3208containing the order (log base 2) of the desired size of the hash 3209table, which must be between 18 and 46. On successful return from the 3210ioctl, the value will not be changed by the kernel. 3211 3212If no hash table has been allocated when any vcpu is asked to run 3213(with the KVM_RUN ioctl), the host kernel will allocate a 3214default-sized hash table (16 MB). 3215 3216If this ioctl is called when a hash table has already been allocated, 3217with a different order from the existing hash table, the existing hash 3218table will be freed and a new one allocated. If this is ioctl is 3219called when a hash table has already been allocated of the same order 3220as specified, the kernel will clear out the existing hash table (zero 3221all HPTEs). In either case, if the guest is using the virtualized 3222real-mode area (VRMA) facility, the kernel will re-create the VMRA 3223HPTEs on the next KVM_RUN of any vcpu. 3224 32254.77 KVM_S390_INTERRUPT 3226----------------------- 3227 3228:Capability: basic 3229:Architectures: s390 3230:Type: vm ioctl, vcpu ioctl 3231:Parameters: struct kvm_s390_interrupt (in) 3232:Returns: 0 on success, -1 on error 3233 3234Allows to inject an interrupt to the guest. Interrupts can be floating 3235(vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type. 3236 3237Interrupt parameters are passed via kvm_s390_interrupt:: 3238 3239 struct kvm_s390_interrupt { 3240 __u32 type; 3241 __u32 parm; 3242 __u64 parm64; 3243 }; 3244 3245type can be one of the following: 3246 3247KVM_S390_SIGP_STOP (vcpu) 3248 - sigp stop; optional flags in parm 3249KVM_S390_PROGRAM_INT (vcpu) 3250 - program check; code in parm 3251KVM_S390_SIGP_SET_PREFIX (vcpu) 3252 - sigp set prefix; prefix address in parm 3253KVM_S390_RESTART (vcpu) 3254 - restart 3255KVM_S390_INT_CLOCK_COMP (vcpu) 3256 - clock comparator interrupt 3257KVM_S390_INT_CPU_TIMER (vcpu) 3258 - CPU timer interrupt 3259KVM_S390_INT_VIRTIO (vm) 3260 - virtio external interrupt; external interrupt 3261 parameters in parm and parm64 3262KVM_S390_INT_SERVICE (vm) 3263 - sclp external interrupt; sclp parameter in parm 3264KVM_S390_INT_EMERGENCY (vcpu) 3265 - sigp emergency; source cpu in parm 3266KVM_S390_INT_EXTERNAL_CALL (vcpu) 3267 - sigp external call; source cpu in parm 3268KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm) 3269 - compound value to indicate an 3270 I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel); 3271 I/O interruption parameters in parm (subchannel) and parm64 (intparm, 3272 interruption subclass) 3273KVM_S390_MCHK (vm, vcpu) 3274 - machine check interrupt; cr 14 bits in parm, machine check interrupt 3275 code in parm64 (note that machine checks needing further payload are not 3276 supported by this ioctl) 3277 3278This is an asynchronous vcpu ioctl and can be invoked from any thread. 3279 32804.78 KVM_PPC_GET_HTAB_FD 3281------------------------ 3282 3283:Capability: KVM_CAP_PPC_HTAB_FD 3284:Architectures: powerpc 3285:Type: vm ioctl 3286:Parameters: Pointer to struct kvm_get_htab_fd (in) 3287:Returns: file descriptor number (>= 0) on success, -1 on error 3288 3289This returns a file descriptor that can be used either to read out the 3290entries in the guest's hashed page table (HPT), or to write entries to 3291initialize the HPT. The returned fd can only be written to if the 3292KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and 3293can only be read if that bit is clear. The argument struct looks like 3294this:: 3295 3296 /* For KVM_PPC_GET_HTAB_FD */ 3297 struct kvm_get_htab_fd { 3298 __u64 flags; 3299 __u64 start_index; 3300 __u64 reserved[2]; 3301 }; 3302 3303 /* Values for kvm_get_htab_fd.flags */ 3304 #define KVM_GET_HTAB_BOLTED_ONLY ((__u64)0x1) 3305 #define KVM_GET_HTAB_WRITE ((__u64)0x2) 3306 3307The 'start_index' field gives the index in the HPT of the entry at 3308which to start reading. It is ignored when writing. 3309 3310Reads on the fd will initially supply information about all 3311"interesting" HPT entries. Interesting entries are those with the 3312bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise 3313all entries. When the end of the HPT is reached, the read() will 3314return. If read() is called again on the fd, it will start again from 3315the beginning of the HPT, but will only return HPT entries that have 3316changed since they were last read. 3317 3318Data read or written is structured as a header (8 bytes) followed by a 3319series of valid HPT entries (16 bytes) each. The header indicates how 3320many valid HPT entries there are and how many invalid entries follow 3321the valid entries. The invalid entries are not represented explicitly 3322in the stream. The header format is:: 3323 3324 struct kvm_get_htab_header { 3325 __u32 index; 3326 __u16 n_valid; 3327 __u16 n_invalid; 3328 }; 3329 3330Writes to the fd create HPT entries starting at the index given in the 3331header; first 'n_valid' valid entries with contents from the data 3332written, then 'n_invalid' invalid entries, invalidating any previously 3333valid entries found. 3334 33354.79 KVM_CREATE_DEVICE 3336---------------------- 3337 3338:Capability: KVM_CAP_DEVICE_CTRL 3339:Architectures: all 3340:Type: vm ioctl 3341:Parameters: struct kvm_create_device (in/out) 3342:Returns: 0 on success, -1 on error 3343 3344Errors: 3345 3346 ====== ======================================================= 3347 ENODEV The device type is unknown or unsupported 3348 EEXIST Device already created, and this type of device may not 3349 be instantiated multiple times 3350 ====== ======================================================= 3351 3352 Other error conditions may be defined by individual device types or 3353 have their standard meanings. 3354 3355Creates an emulated device in the kernel. The file descriptor returned 3356in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR. 3357 3358If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the 3359device type is supported (not necessarily whether it can be created 3360in the current vm). 3361 3362Individual devices should not define flags. Attributes should be used 3363for specifying any behavior that is not implied by the device type 3364number. 3365 3366:: 3367 3368 struct kvm_create_device { 3369 __u32 type; /* in: KVM_DEV_TYPE_xxx */ 3370 __u32 fd; /* out: device handle */ 3371 __u32 flags; /* in: KVM_CREATE_DEVICE_xxx */ 3372 }; 3373 33744.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR 3375-------------------------------------------- 3376 3377:Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device, 3378 KVM_CAP_VCPU_ATTRIBUTES for vcpu device 3379 KVM_CAP_SYS_ATTRIBUTES for system (/dev/kvm) device (no set) 3380:Architectures: x86, arm64, s390 3381:Type: device ioctl, vm ioctl, vcpu ioctl 3382:Parameters: struct kvm_device_attr 3383:Returns: 0 on success, -1 on error 3384 3385Errors: 3386 3387 ===== ============================================================= 3388 ENXIO The group or attribute is unknown/unsupported for this device 3389 or hardware support is missing. 3390 EPERM The attribute cannot (currently) be accessed this way 3391 (e.g. read-only attribute, or attribute that only makes 3392 sense when the device is in a different state) 3393 ===== ============================================================= 3394 3395 Other error conditions may be defined by individual device types. 3396 3397Gets/sets a specified piece of device configuration and/or state. The 3398semantics are device-specific. See individual device documentation in 3399the "devices" directory. As with ONE_REG, the size of the data 3400transferred is defined by the particular attribute. 3401 3402:: 3403 3404 struct kvm_device_attr { 3405 __u32 flags; /* no flags currently defined */ 3406 __u32 group; /* device-defined */ 3407 __u64 attr; /* group-defined */ 3408 __u64 addr; /* userspace address of attr data */ 3409 }; 3410 34114.81 KVM_HAS_DEVICE_ATTR 3412------------------------ 3413 3414:Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device, 3415 KVM_CAP_VCPU_ATTRIBUTES for vcpu device 3416 KVM_CAP_SYS_ATTRIBUTES for system (/dev/kvm) device 3417:Type: device ioctl, vm ioctl, vcpu ioctl 3418:Parameters: struct kvm_device_attr 3419:Returns: 0 on success, -1 on error 3420 3421Errors: 3422 3423 ===== ============================================================= 3424 ENXIO The group or attribute is unknown/unsupported for this device 3425 or hardware support is missing. 3426 ===== ============================================================= 3427 3428Tests whether a device supports a particular attribute. A successful 3429return indicates the attribute is implemented. It does not necessarily 3430indicate that the attribute can be read or written in the device's 3431current state. "addr" is ignored. 3432 3433.. _KVM_ARM_VCPU_INIT: 3434 34354.82 KVM_ARM_VCPU_INIT 3436---------------------- 3437 3438:Capability: basic 3439:Architectures: arm64 3440:Type: vcpu ioctl 3441:Parameters: struct kvm_vcpu_init (in) 3442:Returns: 0 on success; -1 on error 3443 3444Errors: 3445 3446 ====== ================================================================= 3447 EINVAL the target is unknown, or the combination of features is invalid. 3448 ENOENT a features bit specified is unknown. 3449 ====== ================================================================= 3450 3451This tells KVM what type of CPU to present to the guest, and what 3452optional features it should have. This will cause a reset of the cpu 3453registers to their initial values. If this is not called, KVM_RUN will 3454return ENOEXEC for that vcpu. 3455 3456The initial values are defined as: 3457 - Processor state: 3458 * AArch64: EL1h, D, A, I and F bits set. All other bits 3459 are cleared. 3460 * AArch32: SVC, A, I and F bits set. All other bits are 3461 cleared. 3462 - General Purpose registers, including PC and SP: set to 0 3463 - FPSIMD/NEON registers: set to 0 3464 - SVE registers: set to 0 3465 - System registers: Reset to their architecturally defined 3466 values as for a warm reset to EL1 (resp. SVC) or EL2 (in the 3467 case of EL2 being enabled). 3468 3469Note that because some registers reflect machine topology, all vcpus 3470should be created before this ioctl is invoked. 3471 3472Userspace can call this function multiple times for a given vcpu, including 3473after the vcpu has been run. This will reset the vcpu to its initial 3474state. All calls to this function after the initial call must use the same 3475target and same set of feature flags, otherwise EINVAL will be returned. 3476 3477Possible features: 3478 3479 - KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state. 3480 Depends on KVM_CAP_ARM_PSCI. If not set, the CPU will be powered on 3481 and execute guest code when KVM_RUN is called. 3482 - KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode. 3483 Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only). 3484 - KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 (or a future revision 3485 backward compatible with v0.2) for the CPU. 3486 Depends on KVM_CAP_ARM_PSCI_0_2. 3487 - KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU. 3488 Depends on KVM_CAP_ARM_PMU_V3. 3489 3490 - KVM_ARM_VCPU_PTRAUTH_ADDRESS: Enables Address Pointer authentication 3491 for arm64 only. 3492 Depends on KVM_CAP_ARM_PTRAUTH_ADDRESS. 3493 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are 3494 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and 3495 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be 3496 requested. 3497 3498 - KVM_ARM_VCPU_PTRAUTH_GENERIC: Enables Generic Pointer authentication 3499 for arm64 only. 3500 Depends on KVM_CAP_ARM_PTRAUTH_GENERIC. 3501 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are 3502 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and 3503 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be 3504 requested. 3505 3506 - KVM_ARM_VCPU_SVE: Enables SVE for the CPU (arm64 only). 3507 Depends on KVM_CAP_ARM_SVE. 3508 Requires KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3509 3510 * After KVM_ARM_VCPU_INIT: 3511 3512 - KVM_REG_ARM64_SVE_VLS may be read using KVM_GET_ONE_REG: the 3513 initial value of this pseudo-register indicates the best set of 3514 vector lengths possible for a vcpu on this host. 3515 3516 * Before KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3517 3518 - KVM_RUN and KVM_GET_REG_LIST are not available; 3519 3520 - KVM_GET_ONE_REG and KVM_SET_ONE_REG cannot be used to access 3521 the scalable architectural SVE registers 3522 KVM_REG_ARM64_SVE_ZREG(), KVM_REG_ARM64_SVE_PREG() or 3523 KVM_REG_ARM64_SVE_FFR; 3524 3525 - KVM_REG_ARM64_SVE_VLS may optionally be written using 3526 KVM_SET_ONE_REG, to modify the set of vector lengths available 3527 for the vcpu. 3528 3529 * After KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3530 3531 - the KVM_REG_ARM64_SVE_VLS pseudo-register is immutable, and can 3532 no longer be written using KVM_SET_ONE_REG. 3533 3534 - KVM_ARM_VCPU_HAS_EL2: Enable Nested Virtualisation support, 3535 booting the guest from EL2 instead of EL1. 3536 Depends on KVM_CAP_ARM_EL2. 3537 The VM is running with HCR_EL2.E2H being RES1 (VHE) unless 3538 KVM_ARM_VCPU_HAS_EL2_E2H0 is also set. 3539 3540 - KVM_ARM_VCPU_HAS_EL2_E2H0: Restrict Nested Virtualisation 3541 support to HCR_EL2.E2H being RES0 (non-VHE). 3542 Depends on KVM_CAP_ARM_EL2_E2H0. 3543 KVM_ARM_VCPU_HAS_EL2 must also be set. 3544 35454.83 KVM_ARM_PREFERRED_TARGET 3546----------------------------- 3547 3548:Capability: basic 3549:Architectures: arm64 3550:Type: vm ioctl 3551:Parameters: struct kvm_vcpu_init (out) 3552:Returns: 0 on success; -1 on error 3553 3554Errors: 3555 3556 ====== ========================================== 3557 ENODEV no preferred target available for the host 3558 ====== ========================================== 3559 3560This queries KVM for preferred CPU target type which can be emulated 3561by KVM on underlying host. 3562 3563The ioctl returns struct kvm_vcpu_init instance containing information 3564about preferred CPU target type and recommended features for it. The 3565kvm_vcpu_init->features bitmap returned will have feature bits set if 3566the preferred target recommends setting these features, but this is 3567not mandatory. 3568 3569The information returned by this ioctl can be used to prepare an instance 3570of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in 3571VCPU matching underlying host. 3572 3573 35744.84 KVM_GET_REG_LIST 3575--------------------- 3576 3577:Capability: basic 3578:Architectures: arm64, mips, riscv 3579:Type: vcpu ioctl 3580:Parameters: struct kvm_reg_list (in/out) 3581:Returns: 0 on success; -1 on error 3582 3583Errors: 3584 3585 ===== ============================================================== 3586 E2BIG the reg index list is too big to fit in the array specified by 3587 the user (the number required will be written into n). 3588 ===== ============================================================== 3589 3590:: 3591 3592 struct kvm_reg_list { 3593 __u64 n; /* number of registers in reg[] */ 3594 __u64 reg[0]; 3595 }; 3596 3597This ioctl returns the guest registers that are supported for the 3598KVM_GET_ONE_REG/KVM_SET_ONE_REG calls. 3599 3600Note that s390 does not support KVM_GET_REG_LIST for historical reasons 3601(read: nobody cared). The set of registers in kernels 4.x and newer is: 3602 3603- KVM_REG_S390_TODPR 3604 3605- KVM_REG_S390_EPOCHDIFF 3606 3607- KVM_REG_S390_CPU_TIMER 3608 3609- KVM_REG_S390_CLOCK_COMP 3610 3611- KVM_REG_S390_PFTOKEN 3612 3613- KVM_REG_S390_PFCOMPARE 3614 3615- KVM_REG_S390_PFSELECT 3616 3617- KVM_REG_S390_PP 3618 3619- KVM_REG_S390_GBEA 3620 3621 36224.85 KVM_ARM_SET_DEVICE_ADDR (deprecated) 3623----------------------------------------- 3624 3625:Capability: KVM_CAP_ARM_SET_DEVICE_ADDR 3626:Architectures: arm64 3627:Type: vm ioctl 3628:Parameters: struct kvm_arm_device_address (in) 3629:Returns: 0 on success, -1 on error 3630 3631Errors: 3632 3633 ====== ============================================ 3634 ENODEV The device id is unknown 3635 ENXIO Device not supported on current system 3636 EEXIST Address already set 3637 E2BIG Address outside guest physical address space 3638 EBUSY Address overlaps with other device range 3639 ====== ============================================ 3640 3641:: 3642 3643 struct kvm_arm_device_addr { 3644 __u64 id; 3645 __u64 addr; 3646 }; 3647 3648Specify a device address in the guest's physical address space where guests 3649can access emulated or directly exposed devices, which the host kernel needs 3650to know about. The id field is an architecture specific identifier for a 3651specific device. 3652 3653arm64 divides the id field into two parts, a device id and an 3654address type id specific to the individual device:: 3655 3656 bits: | 63 ... 32 | 31 ... 16 | 15 ... 0 | 3657 field: | 0x00000000 | device id | addr type id | 3658 3659arm64 currently only require this when using the in-kernel GIC 3660support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2 3661as the device id. When setting the base address for the guest's 3662mapping of the VGIC virtual CPU and distributor interface, the ioctl 3663must be called after calling KVM_CREATE_IRQCHIP, but before calling 3664KVM_RUN on any of the VCPUs. Calling this ioctl twice for any of the 3665base addresses will return -EEXIST. 3666 3667Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API 3668should be used instead. 3669 3670 36714.86 KVM_PPC_RTAS_DEFINE_TOKEN 3672------------------------------ 3673 3674:Capability: KVM_CAP_PPC_RTAS 3675:Architectures: ppc 3676:Type: vm ioctl 3677:Parameters: struct kvm_rtas_token_args 3678:Returns: 0 on success, -1 on error 3679 3680Defines a token value for a RTAS (Run Time Abstraction Services) 3681service in order to allow it to be handled in the kernel. The 3682argument struct gives the name of the service, which must be the name 3683of a service that has a kernel-side implementation. If the token 3684value is non-zero, it will be associated with that service, and 3685subsequent RTAS calls by the guest specifying that token will be 3686handled by the kernel. If the token value is 0, then any token 3687associated with the service will be forgotten, and subsequent RTAS 3688calls by the guest for that service will be passed to userspace to be 3689handled. 3690 36914.87 KVM_SET_GUEST_DEBUG 3692------------------------ 3693 3694:Capability: KVM_CAP_SET_GUEST_DEBUG 3695:Architectures: x86, s390, ppc, arm64 3696:Type: vcpu ioctl 3697:Parameters: struct kvm_guest_debug (in) 3698:Returns: 0 on success; -1 on error 3699 3700:: 3701 3702 struct kvm_guest_debug { 3703 __u32 control; 3704 __u32 pad; 3705 struct kvm_guest_debug_arch arch; 3706 }; 3707 3708Set up the processor specific debug registers and configure vcpu for 3709handling guest debug events. There are two parts to the structure, the 3710first a control bitfield indicates the type of debug events to handle 3711when running. Common control bits are: 3712 3713 - KVM_GUESTDBG_ENABLE: guest debugging is enabled 3714 - KVM_GUESTDBG_SINGLESTEP: the next run should single-step 3715 3716The top 16 bits of the control field are architecture specific control 3717flags which can include the following: 3718 3719 - KVM_GUESTDBG_USE_SW_BP: using software breakpoints [x86, arm64] 3720 - KVM_GUESTDBG_USE_HW_BP: using hardware breakpoints [x86, s390] 3721 - KVM_GUESTDBG_USE_HW: using hardware debug events [arm64] 3722 - KVM_GUESTDBG_INJECT_DB: inject DB type exception [x86] 3723 - KVM_GUESTDBG_INJECT_BP: inject BP type exception [x86] 3724 - KVM_GUESTDBG_EXIT_PENDING: trigger an immediate guest exit [s390] 3725 - KVM_GUESTDBG_BLOCKIRQ: avoid injecting interrupts/NMI/SMI [x86] 3726 3727For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints 3728are enabled in memory so we need to ensure breakpoint exceptions are 3729correctly trapped and the KVM run loop exits at the breakpoint and not 3730running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP 3731we need to ensure the guest vCPUs architecture specific registers are 3732updated to the correct (supplied) values. 3733 3734The second part of the structure is architecture specific and 3735typically contains a set of debug registers. 3736 3737For arm64 the number of debug registers is implementation defined and 3738can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and 3739KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number 3740indicating the number of supported registers. 3741 3742For ppc, the KVM_CAP_PPC_GUEST_DEBUG_SSTEP capability indicates whether 3743the single-step debug event (KVM_GUESTDBG_SINGLESTEP) is supported. 3744 3745Also when supported, KVM_CAP_SET_GUEST_DEBUG2 capability indicates the 3746supported KVM_GUESTDBG_* bits in the control field. 3747 3748When debug events exit the main run loop with the reason 3749KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run 3750structure containing architecture specific debug information. 3751 37524.88 KVM_GET_EMULATED_CPUID 3753--------------------------- 3754 3755:Capability: KVM_CAP_EXT_EMUL_CPUID 3756:Architectures: x86 3757:Type: system ioctl 3758:Parameters: struct kvm_cpuid2 (in/out) 3759:Returns: 0 on success, -1 on error 3760 3761:: 3762 3763 struct kvm_cpuid2 { 3764 __u32 nent; 3765 __u32 flags; 3766 struct kvm_cpuid_entry2 entries[0]; 3767 }; 3768 3769The member 'flags' is used for passing flags from userspace. 3770 3771:: 3772 3773 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0) 3774 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */ 3775 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */ 3776 3777 struct kvm_cpuid_entry2 { 3778 __u32 function; 3779 __u32 index; 3780 __u32 flags; 3781 __u32 eax; 3782 __u32 ebx; 3783 __u32 ecx; 3784 __u32 edx; 3785 __u32 padding[3]; 3786 }; 3787 3788This ioctl returns x86 cpuid features which are emulated by 3789kvm.Userspace can use the information returned by this ioctl to query 3790which features are emulated by kvm instead of being present natively. 3791 3792Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2 3793structure with the 'nent' field indicating the number of entries in 3794the variable-size array 'entries'. If the number of entries is too low 3795to describe the cpu capabilities, an error (E2BIG) is returned. If the 3796number is too high, the 'nent' field is adjusted and an error (ENOMEM) 3797is returned. If the number is just right, the 'nent' field is adjusted 3798to the number of valid entries in the 'entries' array, which is then 3799filled. 3800 3801The entries returned are the set CPUID bits of the respective features 3802which kvm emulates, as returned by the CPUID instruction, with unknown 3803or unsupported feature bits cleared. 3804 3805Features like x2apic, for example, may not be present in the host cpu 3806but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be 3807emulated efficiently and thus not included here. 3808 3809The fields in each entry are defined as follows: 3810 3811 function: 3812 the eax value used to obtain the entry 3813 index: 3814 the ecx value used to obtain the entry (for entries that are 3815 affected by ecx) 3816 flags: 3817 an OR of zero or more of the following: 3818 3819 KVM_CPUID_FLAG_SIGNIFCANT_INDEX: 3820 if the index field is valid 3821 3822 eax, ebx, ecx, edx: 3823 3824 the values returned by the cpuid instruction for 3825 this function/index combination 3826 38274.89 KVM_S390_MEM_OP 3828-------------------- 3829 3830:Capability: KVM_CAP_S390_MEM_OP, KVM_CAP_S390_PROTECTED, KVM_CAP_S390_MEM_OP_EXTENSION 3831:Architectures: s390 3832:Type: vm ioctl, vcpu ioctl 3833:Parameters: struct kvm_s390_mem_op (in) 3834:Returns: = 0 on success, 3835 < 0 on generic error (e.g. -EFAULT or -ENOMEM), 3836 16 bit program exception code if the access causes such an exception 3837 3838Read or write data from/to the VM's memory. 3839The KVM_CAP_S390_MEM_OP_EXTENSION capability specifies what functionality is 3840supported. 3841 3842Parameters are specified via the following structure:: 3843 3844 struct kvm_s390_mem_op { 3845 __u64 gaddr; /* the guest address */ 3846 __u64 flags; /* flags */ 3847 __u32 size; /* amount of bytes */ 3848 __u32 op; /* type of operation */ 3849 __u64 buf; /* buffer in userspace */ 3850 union { 3851 struct { 3852 __u8 ar; /* the access register number */ 3853 __u8 key; /* access key, ignored if flag unset */ 3854 __u8 pad1[6]; /* ignored */ 3855 __u64 old_addr; /* ignored if flag unset */ 3856 }; 3857 __u32 sida_offset; /* offset into the sida */ 3858 __u8 reserved[32]; /* ignored */ 3859 }; 3860 }; 3861 3862The start address of the memory region has to be specified in the "gaddr" 3863field, and the length of the region in the "size" field (which must not 3864be 0). The maximum value for "size" can be obtained by checking the 3865KVM_CAP_S390_MEM_OP capability. "buf" is the buffer supplied by the 3866userspace application where the read data should be written to for 3867a read access, or where the data that should be written is stored for 3868a write access. The "reserved" field is meant for future extensions. 3869Reserved and unused values are ignored. Future extension that add members must 3870introduce new flags. 3871 3872The type of operation is specified in the "op" field. Flags modifying 3873their behavior can be set in the "flags" field. Undefined flag bits must 3874be set to 0. 3875 3876Possible operations are: 3877 * ``KVM_S390_MEMOP_LOGICAL_READ`` 3878 * ``KVM_S390_MEMOP_LOGICAL_WRITE`` 3879 * ``KVM_S390_MEMOP_ABSOLUTE_READ`` 3880 * ``KVM_S390_MEMOP_ABSOLUTE_WRITE`` 3881 * ``KVM_S390_MEMOP_SIDA_READ`` 3882 * ``KVM_S390_MEMOP_SIDA_WRITE`` 3883 * ``KVM_S390_MEMOP_ABSOLUTE_CMPXCHG`` 3884 3885Logical read/write: 3886^^^^^^^^^^^^^^^^^^^ 3887 3888Access logical memory, i.e. translate the given guest address to an absolute 3889address given the state of the VCPU and use the absolute address as target of 3890the access. "ar" designates the access register number to be used; the valid 3891range is 0..15. 3892Logical accesses are permitted for the VCPU ioctl only. 3893Logical accesses are permitted for non-protected guests only. 3894 3895Supported flags: 3896 * ``KVM_S390_MEMOP_F_CHECK_ONLY`` 3897 * ``KVM_S390_MEMOP_F_INJECT_EXCEPTION`` 3898 * ``KVM_S390_MEMOP_F_SKEY_PROTECTION`` 3899 3900The KVM_S390_MEMOP_F_CHECK_ONLY flag can be set to check whether the 3901corresponding memory access would cause an access exception; however, 3902no actual access to the data in memory at the destination is performed. 3903In this case, "buf" is unused and can be NULL. 3904 3905In case an access exception occurred during the access (or would occur 3906in case of KVM_S390_MEMOP_F_CHECK_ONLY), the ioctl returns a positive 3907error number indicating the type of exception. This exception is also 3908raised directly at the corresponding VCPU if the flag 3909KVM_S390_MEMOP_F_INJECT_EXCEPTION is set. 3910On protection exceptions, unless specified otherwise, the injected 3911translation-exception identifier (TEID) indicates suppression. 3912 3913If the KVM_S390_MEMOP_F_SKEY_PROTECTION flag is set, storage key 3914protection is also in effect and may cause exceptions if accesses are 3915prohibited given the access key designated by "key"; the valid range is 0..15. 3916KVM_S390_MEMOP_F_SKEY_PROTECTION is available if KVM_CAP_S390_MEM_OP_EXTENSION 3917is > 0. 3918Since the accessed memory may span multiple pages and those pages might have 3919different storage keys, it is possible that a protection exception occurs 3920after memory has been modified. In this case, if the exception is injected, 3921the TEID does not indicate suppression. 3922 3923Absolute read/write: 3924^^^^^^^^^^^^^^^^^^^^ 3925 3926Access absolute memory. This operation is intended to be used with the 3927KVM_S390_MEMOP_F_SKEY_PROTECTION flag, to allow accessing memory and performing 3928the checks required for storage key protection as one operation (as opposed to 3929user space getting the storage keys, performing the checks, and accessing 3930memory thereafter, which could lead to a delay between check and access). 3931Absolute accesses are permitted for the VM ioctl if KVM_CAP_S390_MEM_OP_EXTENSION 3932has the KVM_S390_MEMOP_EXTENSION_CAP_BASE bit set. 3933Currently absolute accesses are not permitted for VCPU ioctls. 3934Absolute accesses are permitted for non-protected guests only. 3935 3936Supported flags: 3937 * ``KVM_S390_MEMOP_F_CHECK_ONLY`` 3938 * ``KVM_S390_MEMOP_F_SKEY_PROTECTION`` 3939 3940The semantics of the flags common with logical accesses are as for logical 3941accesses. 3942 3943Absolute cmpxchg: 3944^^^^^^^^^^^^^^^^^ 3945 3946Perform cmpxchg on absolute guest memory. Intended for use with the 3947KVM_S390_MEMOP_F_SKEY_PROTECTION flag. 3948Instead of doing an unconditional write, the access occurs only if the target 3949location contains the value pointed to by "old_addr". 3950This is performed as an atomic cmpxchg with the length specified by the "size" 3951parameter. "size" must be a power of two up to and including 16. 3952If the exchange did not take place because the target value doesn't match the 3953old value, the value "old_addr" points to is replaced by the target value. 3954User space can tell if an exchange took place by checking if this replacement 3955occurred. The cmpxchg op is permitted for the VM ioctl if 3956KVM_CAP_S390_MEM_OP_EXTENSION has flag KVM_S390_MEMOP_EXTENSION_CAP_CMPXCHG set. 3957 3958Supported flags: 3959 * ``KVM_S390_MEMOP_F_SKEY_PROTECTION`` 3960 3961SIDA read/write: 3962^^^^^^^^^^^^^^^^ 3963 3964Access the secure instruction data area which contains memory operands necessary 3965for instruction emulation for protected guests. 3966SIDA accesses are available if the KVM_CAP_S390_PROTECTED capability is available. 3967SIDA accesses are permitted for the VCPU ioctl only. 3968SIDA accesses are permitted for protected guests only. 3969 3970No flags are supported. 3971 39724.90 KVM_S390_GET_SKEYS 3973----------------------- 3974 3975:Capability: KVM_CAP_S390_SKEYS 3976:Architectures: s390 3977:Type: vm ioctl 3978:Parameters: struct kvm_s390_skeys 3979:Returns: 0 on success, KVM_S390_GET_SKEYS_NONE if guest is not using storage 3980 keys, negative value on error 3981 3982This ioctl is used to get guest storage key values on the s390 3983architecture. The ioctl takes parameters via the kvm_s390_skeys struct:: 3984 3985 struct kvm_s390_skeys { 3986 __u64 start_gfn; 3987 __u64 count; 3988 __u64 skeydata_addr; 3989 __u32 flags; 3990 __u32 reserved[9]; 3991 }; 3992 3993The start_gfn field is the number of the first guest frame whose storage keys 3994you want to get. 3995 3996The count field is the number of consecutive frames (starting from start_gfn) 3997whose storage keys to get. The count field must be at least 1 and the maximum 3998allowed value is defined as KVM_S390_SKEYS_MAX. Values outside this range 3999will cause the ioctl to return -EINVAL. 4000 4001The skeydata_addr field is the address to a buffer large enough to hold count 4002bytes. This buffer will be filled with storage key data by the ioctl. 4003 40044.91 KVM_S390_SET_SKEYS 4005----------------------- 4006 4007:Capability: KVM_CAP_S390_SKEYS 4008:Architectures: s390 4009:Type: vm ioctl 4010:Parameters: struct kvm_s390_skeys 4011:Returns: 0 on success, negative value on error 4012 4013This ioctl is used to set guest storage key values on the s390 4014architecture. The ioctl takes parameters via the kvm_s390_skeys struct. 4015See section on KVM_S390_GET_SKEYS for struct definition. 4016 4017The start_gfn field is the number of the first guest frame whose storage keys 4018you want to set. 4019 4020The count field is the number of consecutive frames (starting from start_gfn) 4021whose storage keys to get. The count field must be at least 1 and the maximum 4022allowed value is defined as KVM_S390_SKEYS_MAX. Values outside this range 4023will cause the ioctl to return -EINVAL. 4024 4025The skeydata_addr field is the address to a buffer containing count bytes of 4026storage keys. Each byte in the buffer will be set as the storage key for a 4027single frame starting at start_gfn for count frames. 4028 4029Note: If any architecturally invalid key value is found in the given data then 4030the ioctl will return -EINVAL. 4031 40324.92 KVM_S390_IRQ 4033----------------- 4034 4035:Capability: KVM_CAP_S390_INJECT_IRQ 4036:Architectures: s390 4037:Type: vcpu ioctl 4038:Parameters: struct kvm_s390_irq (in) 4039:Returns: 0 on success, -1 on error 4040 4041Errors: 4042 4043 4044 ====== ================================================================= 4045 EINVAL interrupt type is invalid 4046 type is KVM_S390_SIGP_STOP and flag parameter is invalid value, 4047 type is KVM_S390_INT_EXTERNAL_CALL and code is bigger 4048 than the maximum of VCPUs 4049 EBUSY type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped, 4050 type is KVM_S390_SIGP_STOP and a stop irq is already pending, 4051 type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt 4052 is already pending 4053 ====== ================================================================= 4054 4055Allows to inject an interrupt to the guest. 4056 4057Using struct kvm_s390_irq as a parameter allows 4058to inject additional payload which is not 4059possible via KVM_S390_INTERRUPT. 4060 4061Interrupt parameters are passed via kvm_s390_irq:: 4062 4063 struct kvm_s390_irq { 4064 __u64 type; 4065 union { 4066 struct kvm_s390_io_info io; 4067 struct kvm_s390_ext_info ext; 4068 struct kvm_s390_pgm_info pgm; 4069 struct kvm_s390_emerg_info emerg; 4070 struct kvm_s390_extcall_info extcall; 4071 struct kvm_s390_prefix_info prefix; 4072 struct kvm_s390_stop_info stop; 4073 struct kvm_s390_mchk_info mchk; 4074 char reserved[64]; 4075 } u; 4076 }; 4077 4078type can be one of the following: 4079 4080- KVM_S390_SIGP_STOP - sigp stop; parameter in .stop 4081- KVM_S390_PROGRAM_INT - program check; parameters in .pgm 4082- KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix 4083- KVM_S390_RESTART - restart; no parameters 4084- KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters 4085- KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters 4086- KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg 4087- KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall 4088- KVM_S390_MCHK - machine check interrupt; parameters in .mchk 4089 4090This is an asynchronous vcpu ioctl and can be invoked from any thread. 4091 40924.94 KVM_S390_GET_IRQ_STATE 4093--------------------------- 4094 4095:Capability: KVM_CAP_S390_IRQ_STATE 4096:Architectures: s390 4097:Type: vcpu ioctl 4098:Parameters: struct kvm_s390_irq_state (out) 4099:Returns: >= number of bytes copied into buffer, 4100 -EINVAL if buffer size is 0, 4101 -ENOBUFS if buffer size is too small to fit all pending interrupts, 4102 -EFAULT if the buffer address was invalid 4103 4104This ioctl allows userspace to retrieve the complete state of all currently 4105pending interrupts in a single buffer. Use cases include migration 4106and introspection. The parameter structure contains the address of a 4107userspace buffer and its length:: 4108 4109 struct kvm_s390_irq_state { 4110 __u64 buf; 4111 __u32 flags; /* will stay unused for compatibility reasons */ 4112 __u32 len; 4113 __u32 reserved[4]; /* will stay unused for compatibility reasons */ 4114 }; 4115 4116Userspace passes in the above struct and for each pending interrupt a 4117struct kvm_s390_irq is copied to the provided buffer. 4118 4119The structure contains a flags and a reserved field for future extensions. As 4120the kernel never checked for flags == 0 and QEMU never pre-zeroed flags and 4121reserved, these fields can not be used in the future without breaking 4122compatibility. 4123 4124If -ENOBUFS is returned the buffer provided was too small and userspace 4125may retry with a bigger buffer. 4126 41274.95 KVM_S390_SET_IRQ_STATE 4128--------------------------- 4129 4130:Capability: KVM_CAP_S390_IRQ_STATE 4131:Architectures: s390 4132:Type: vcpu ioctl 4133:Parameters: struct kvm_s390_irq_state (in) 4134:Returns: 0 on success, 4135 -EFAULT if the buffer address was invalid, 4136 -EINVAL for an invalid buffer length (see below), 4137 -EBUSY if there were already interrupts pending, 4138 errors occurring when actually injecting the 4139 interrupt. See KVM_S390_IRQ. 4140 4141This ioctl allows userspace to set the complete state of all cpu-local 4142interrupts currently pending for the vcpu. It is intended for restoring 4143interrupt state after a migration. The input parameter is a userspace buffer 4144containing a struct kvm_s390_irq_state:: 4145 4146 struct kvm_s390_irq_state { 4147 __u64 buf; 4148 __u32 flags; /* will stay unused for compatibility reasons */ 4149 __u32 len; 4150 __u32 reserved[4]; /* will stay unused for compatibility reasons */ 4151 }; 4152 4153The restrictions for flags and reserved apply as well. 4154(see KVM_S390_GET_IRQ_STATE) 4155 4156The userspace memory referenced by buf contains a struct kvm_s390_irq 4157for each interrupt to be injected into the guest. 4158If one of the interrupts could not be injected for some reason the 4159ioctl aborts. 4160 4161len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0 4162and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq), 4163which is the maximum number of possibly pending cpu-local interrupts. 4164 41654.96 KVM_SMI 4166------------ 4167 4168:Capability: KVM_CAP_X86_SMM 4169:Architectures: x86 4170:Type: vcpu ioctl 4171:Parameters: none 4172:Returns: 0 on success, -1 on error 4173 4174Queues an SMI on the thread's vcpu. 4175 41764.97 KVM_X86_SET_MSR_FILTER 4177---------------------------- 4178 4179:Capability: KVM_CAP_X86_MSR_FILTER 4180:Architectures: x86 4181:Type: vm ioctl 4182:Parameters: struct kvm_msr_filter 4183:Returns: 0 on success, < 0 on error 4184 4185:: 4186 4187 struct kvm_msr_filter_range { 4188 #define KVM_MSR_FILTER_READ (1 << 0) 4189 #define KVM_MSR_FILTER_WRITE (1 << 1) 4190 __u32 flags; 4191 __u32 nmsrs; /* number of msrs in bitmap */ 4192 __u32 base; /* MSR index the bitmap starts at */ 4193 __u8 *bitmap; /* a 1 bit allows the operations in flags, 0 denies */ 4194 }; 4195 4196 #define KVM_MSR_FILTER_MAX_RANGES 16 4197 struct kvm_msr_filter { 4198 #define KVM_MSR_FILTER_DEFAULT_ALLOW (0 << 0) 4199 #define KVM_MSR_FILTER_DEFAULT_DENY (1 << 0) 4200 __u32 flags; 4201 struct kvm_msr_filter_range ranges[KVM_MSR_FILTER_MAX_RANGES]; 4202 }; 4203 4204flags values for ``struct kvm_msr_filter_range``: 4205 4206``KVM_MSR_FILTER_READ`` 4207 4208 Filter read accesses to MSRs using the given bitmap. A 0 in the bitmap 4209 indicates that read accesses should be denied, while a 1 indicates that 4210 a read for a particular MSR should be allowed regardless of the default 4211 filter action. 4212 4213``KVM_MSR_FILTER_WRITE`` 4214 4215 Filter write accesses to MSRs using the given bitmap. A 0 in the bitmap 4216 indicates that write accesses should be denied, while a 1 indicates that 4217 a write for a particular MSR should be allowed regardless of the default 4218 filter action. 4219 4220flags values for ``struct kvm_msr_filter``: 4221 4222``KVM_MSR_FILTER_DEFAULT_ALLOW`` 4223 4224 If no filter range matches an MSR index that is getting accessed, KVM will 4225 allow accesses to all MSRs by default. 4226 4227``KVM_MSR_FILTER_DEFAULT_DENY`` 4228 4229 If no filter range matches an MSR index that is getting accessed, KVM will 4230 deny accesses to all MSRs by default. 4231 4232This ioctl allows userspace to define up to 16 bitmaps of MSR ranges to deny 4233guest MSR accesses that would normally be allowed by KVM. If an MSR is not 4234covered by a specific range, the "default" filtering behavior applies. Each 4235bitmap range covers MSRs from [base .. base+nmsrs). 4236 4237If an MSR access is denied by userspace, the resulting KVM behavior depends on 4238whether or not KVM_CAP_X86_USER_SPACE_MSR's KVM_MSR_EXIT_REASON_FILTER is 4239enabled. If KVM_MSR_EXIT_REASON_FILTER is enabled, KVM will exit to userspace 4240on denied accesses, i.e. userspace effectively intercepts the MSR access. If 4241KVM_MSR_EXIT_REASON_FILTER is not enabled, KVM will inject a #GP into the guest 4242on denied accesses. Note, if an MSR access is denied during emulation of MSR 4243load/stores during VMX transitions, KVM ignores KVM_MSR_EXIT_REASON_FILTER. 4244See the below warning for full details. 4245 4246If an MSR access is allowed by userspace, KVM will emulate and/or virtualize 4247the access in accordance with the vCPU model. Note, KVM may still ultimately 4248inject a #GP if an access is allowed by userspace, e.g. if KVM doesn't support 4249the MSR, or to follow architectural behavior for the MSR. 4250 4251By default, KVM operates in KVM_MSR_FILTER_DEFAULT_ALLOW mode with no MSR range 4252filters. 4253 4254Calling this ioctl with an empty set of ranges (all nmsrs == 0) disables MSR 4255filtering. In that mode, ``KVM_MSR_FILTER_DEFAULT_DENY`` is invalid and causes 4256an error. 4257 4258.. warning:: 4259 MSR accesses that are side effects of instruction execution (emulated or 4260 native) are not filtered as hardware does not honor MSR bitmaps outside of 4261 RDMSR and WRMSR, and KVM mimics that behavior when emulating instructions 4262 to avoid pointless divergence from hardware. E.g. RDPID reads MSR_TSC_AUX, 4263 SYSENTER reads the SYSENTER MSRs, etc. 4264 4265 MSRs that are loaded/stored via dedicated VMCS fields are not filtered as 4266 part of VM-Enter/VM-Exit emulation. 4267 4268 MSRs that are loaded/store via VMX's load/store lists _are_ filtered as part 4269 of VM-Enter/VM-Exit emulation. If an MSR access is denied on VM-Enter, KVM 4270 synthesizes a consistency check VM-Exit(EXIT_REASON_MSR_LOAD_FAIL). If an 4271 MSR access is denied on VM-Exit, KVM synthesizes a VM-Abort. In short, KVM 4272 extends Intel's architectural list of MSRs that cannot be loaded/saved via 4273 the VM-Enter/VM-Exit MSR list. It is platform owner's responsibility to 4274 to communicate any such restrictions to their end users. 4275 4276 x2APIC MSR accesses cannot be filtered (KVM silently ignores filters that 4277 cover any x2APIC MSRs). 4278 4279Note, invoking this ioctl while a vCPU is running is inherently racy. However, 4280KVM does guarantee that vCPUs will see either the previous filter or the new 4281filter, e.g. MSRs with identical settings in both the old and new filter will 4282have deterministic behavior. 4283 4284Similarly, if userspace wishes to intercept on denied accesses, 4285KVM_MSR_EXIT_REASON_FILTER must be enabled before activating any filters, and 4286left enabled until after all filters are deactivated. Failure to do so may 4287result in KVM injecting a #GP instead of exiting to userspace. 4288 42894.98 KVM_CREATE_SPAPR_TCE_64 4290---------------------------- 4291 4292:Capability: KVM_CAP_SPAPR_TCE_64 4293:Architectures: powerpc 4294:Type: vm ioctl 4295:Parameters: struct kvm_create_spapr_tce_64 (in) 4296:Returns: file descriptor for manipulating the created TCE table 4297 4298This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit 4299windows, described in 4.62 KVM_CREATE_SPAPR_TCE 4300 4301This capability uses extended struct in ioctl interface:: 4302 4303 /* for KVM_CAP_SPAPR_TCE_64 */ 4304 struct kvm_create_spapr_tce_64 { 4305 __u64 liobn; 4306 __u32 page_shift; 4307 __u32 flags; 4308 __u64 offset; /* in pages */ 4309 __u64 size; /* in pages */ 4310 }; 4311 4312The aim of extension is to support an additional bigger DMA window with 4313a variable page size. 4314KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and 4315a bus offset of the corresponding DMA window, @size and @offset are numbers 4316of IOMMU pages. 4317 4318@flags are not used at the moment. 4319 4320The rest of functionality is identical to KVM_CREATE_SPAPR_TCE. 4321 43224.99 KVM_REINJECT_CONTROL 4323------------------------- 4324 4325:Capability: KVM_CAP_REINJECT_CONTROL 4326:Architectures: x86 4327:Type: vm ioctl 4328:Parameters: struct kvm_reinject_control (in) 4329:Returns: 0 on success, 4330 -EFAULT if struct kvm_reinject_control cannot be read, 4331 -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier. 4332 4333i8254 (PIT) has two modes, reinject and !reinject. The default is reinject, 4334where KVM queues elapsed i8254 ticks and monitors completion of interrupt from 4335vector(s) that i8254 injects. Reinject mode dequeues a tick and injects its 4336interrupt whenever there isn't a pending interrupt from i8254. 4337!reinject mode injects an interrupt as soon as a tick arrives. 4338 4339:: 4340 4341 struct kvm_reinject_control { 4342 __u8 pit_reinject; 4343 __u8 reserved[31]; 4344 }; 4345 4346pit_reinject = 0 (!reinject mode) is recommended, unless running an old 4347operating system that uses the PIT for timing (e.g. Linux 2.4.x). 4348 43494.100 KVM_PPC_CONFIGURE_V3_MMU 4350------------------------------ 4351 4352:Capability: KVM_CAP_PPC_MMU_RADIX or KVM_CAP_PPC_MMU_HASH_V3 4353:Architectures: ppc 4354:Type: vm ioctl 4355:Parameters: struct kvm_ppc_mmuv3_cfg (in) 4356:Returns: 0 on success, 4357 -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read, 4358 -EINVAL if the configuration is invalid 4359 4360This ioctl controls whether the guest will use radix or HPT (hashed 4361page table) translation, and sets the pointer to the process table for 4362the guest. 4363 4364:: 4365 4366 struct kvm_ppc_mmuv3_cfg { 4367 __u64 flags; 4368 __u64 process_table; 4369 }; 4370 4371There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and 4372KVM_PPC_MMUV3_GTSE. KVM_PPC_MMUV3_RADIX, if set, configures the guest 4373to use radix tree translation, and if clear, to use HPT translation. 4374KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest 4375to be able to use the global TLB and SLB invalidation instructions; 4376if clear, the guest may not use these instructions. 4377 4378The process_table field specifies the address and size of the guest 4379process table, which is in the guest's space. This field is formatted 4380as the second doubleword of the partition table entry, as defined in 4381the Power ISA V3.00, Book III section 5.7.6.1. 4382 43834.101 KVM_PPC_GET_RMMU_INFO 4384--------------------------- 4385 4386:Capability: KVM_CAP_PPC_MMU_RADIX 4387:Architectures: ppc 4388:Type: vm ioctl 4389:Parameters: struct kvm_ppc_rmmu_info (out) 4390:Returns: 0 on success, 4391 -EFAULT if struct kvm_ppc_rmmu_info cannot be written, 4392 -EINVAL if no useful information can be returned 4393 4394This ioctl returns a structure containing two things: (a) a list 4395containing supported radix tree geometries, and (b) a list that maps 4396page sizes to put in the "AP" (actual page size) field for the tlbie 4397(TLB invalidate entry) instruction. 4398 4399:: 4400 4401 struct kvm_ppc_rmmu_info { 4402 struct kvm_ppc_radix_geom { 4403 __u8 page_shift; 4404 __u8 level_bits[4]; 4405 __u8 pad[3]; 4406 } geometries[8]; 4407 __u32 ap_encodings[8]; 4408 }; 4409 4410The geometries[] field gives up to 8 supported geometries for the 4411radix page table, in terms of the log base 2 of the smallest page 4412size, and the number of bits indexed at each level of the tree, from 4413the PTE level up to the PGD level in that order. Any unused entries 4414will have 0 in the page_shift field. 4415 4416The ap_encodings gives the supported page sizes and their AP field 4417encodings, encoded with the AP value in the top 3 bits and the log 4418base 2 of the page size in the bottom 6 bits. 4419 44204.102 KVM_PPC_RESIZE_HPT_PREPARE 4421-------------------------------- 4422 4423:Capability: KVM_CAP_SPAPR_RESIZE_HPT 4424:Architectures: powerpc 4425:Type: vm ioctl 4426:Parameters: struct kvm_ppc_resize_hpt (in) 4427:Returns: 0 on successful completion, 4428 >0 if a new HPT is being prepared, the value is an estimated 4429 number of milliseconds until preparation is complete, 4430 -EFAULT if struct kvm_reinject_control cannot be read, 4431 -EINVAL if the supplied shift or flags are invalid, 4432 -ENOMEM if unable to allocate the new HPT, 4433 4434Used to implement the PAPR extension for runtime resizing of a guest's 4435Hashed Page Table (HPT). Specifically this starts, stops or monitors 4436the preparation of a new potential HPT for the guest, essentially 4437implementing the H_RESIZE_HPT_PREPARE hypercall. 4438 4439:: 4440 4441 struct kvm_ppc_resize_hpt { 4442 __u64 flags; 4443 __u32 shift; 4444 __u32 pad; 4445 }; 4446 4447If called with shift > 0 when there is no pending HPT for the guest, 4448this begins preparation of a new pending HPT of size 2^(shift) bytes. 4449It then returns a positive integer with the estimated number of 4450milliseconds until preparation is complete. 4451 4452If called when there is a pending HPT whose size does not match that 4453requested in the parameters, discards the existing pending HPT and 4454creates a new one as above. 4455 4456If called when there is a pending HPT of the size requested, will: 4457 4458 * If preparation of the pending HPT is already complete, return 0 4459 * If preparation of the pending HPT has failed, return an error 4460 code, then discard the pending HPT. 4461 * If preparation of the pending HPT is still in progress, return an 4462 estimated number of milliseconds until preparation is complete. 4463 4464If called with shift == 0, discards any currently pending HPT and 4465returns 0 (i.e. cancels any in-progress preparation). 4466 4467flags is reserved for future expansion, currently setting any bits in 4468flags will result in an -EINVAL. 4469 4470Normally this will be called repeatedly with the same parameters until 4471it returns <= 0. The first call will initiate preparation, subsequent 4472ones will monitor preparation until it completes or fails. 4473 44744.103 KVM_PPC_RESIZE_HPT_COMMIT 4475------------------------------- 4476 4477:Capability: KVM_CAP_SPAPR_RESIZE_HPT 4478:Architectures: powerpc 4479:Type: vm ioctl 4480:Parameters: struct kvm_ppc_resize_hpt (in) 4481:Returns: 0 on successful completion, 4482 -EFAULT if struct kvm_reinject_control cannot be read, 4483 -EINVAL if the supplied shift or flags are invalid, 4484 -ENXIO is there is no pending HPT, or the pending HPT doesn't 4485 have the requested size, 4486 -EBUSY if the pending HPT is not fully prepared, 4487 -ENOSPC if there was a hash collision when moving existing 4488 HPT entries to the new HPT, 4489 -EIO on other error conditions 4490 4491Used to implement the PAPR extension for runtime resizing of a guest's 4492Hashed Page Table (HPT). Specifically this requests that the guest be 4493transferred to working with the new HPT, essentially implementing the 4494H_RESIZE_HPT_COMMIT hypercall. 4495 4496:: 4497 4498 struct kvm_ppc_resize_hpt { 4499 __u64 flags; 4500 __u32 shift; 4501 __u32 pad; 4502 }; 4503 4504This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has 4505returned 0 with the same parameters. In other cases 4506KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or 4507-EBUSY, though others may be possible if the preparation was started, 4508but failed). 4509 4510This will have undefined effects on the guest if it has not already 4511placed itself in a quiescent state where no vcpu will make MMU enabled 4512memory accesses. 4513 4514On successful completion, the pending HPT will become the guest's active 4515HPT and the previous HPT will be discarded. 4516 4517On failure, the guest will still be operating on its previous HPT. 4518 45194.104 KVM_X86_GET_MCE_CAP_SUPPORTED 4520----------------------------------- 4521 4522:Capability: KVM_CAP_MCE 4523:Architectures: x86 4524:Type: system ioctl 4525:Parameters: u64 mce_cap (out) 4526:Returns: 0 on success, -1 on error 4527 4528Returns supported MCE capabilities. The u64 mce_cap parameter 4529has the same format as the MSR_IA32_MCG_CAP register. Supported 4530capabilities will have the corresponding bits set. 4531 45324.105 KVM_X86_SETUP_MCE 4533----------------------- 4534 4535:Capability: KVM_CAP_MCE 4536:Architectures: x86 4537:Type: vcpu ioctl 4538:Parameters: u64 mcg_cap (in) 4539:Returns: 0 on success, 4540 -EFAULT if u64 mcg_cap cannot be read, 4541 -EINVAL if the requested number of banks is invalid, 4542 -EINVAL if requested MCE capability is not supported. 4543 4544Initializes MCE support for use. The u64 mcg_cap parameter 4545has the same format as the MSR_IA32_MCG_CAP register and 4546specifies which capabilities should be enabled. The maximum 4547supported number of error-reporting banks can be retrieved when 4548checking for KVM_CAP_MCE. The supported capabilities can be 4549retrieved with KVM_X86_GET_MCE_CAP_SUPPORTED. 4550 45514.106 KVM_X86_SET_MCE 4552--------------------- 4553 4554:Capability: KVM_CAP_MCE 4555:Architectures: x86 4556:Type: vcpu ioctl 4557:Parameters: struct kvm_x86_mce (in) 4558:Returns: 0 on success, 4559 -EFAULT if struct kvm_x86_mce cannot be read, 4560 -EINVAL if the bank number is invalid, 4561 -EINVAL if VAL bit is not set in status field. 4562 4563Inject a machine check error (MCE) into the guest. The input 4564parameter is:: 4565 4566 struct kvm_x86_mce { 4567 __u64 status; 4568 __u64 addr; 4569 __u64 misc; 4570 __u64 mcg_status; 4571 __u8 bank; 4572 __u8 pad1[7]; 4573 __u64 pad2[3]; 4574 }; 4575 4576If the MCE being reported is an uncorrected error, KVM will 4577inject it as an MCE exception into the guest. If the guest 4578MCG_STATUS register reports that an MCE is in progress, KVM 4579causes an KVM_EXIT_SHUTDOWN vmexit. 4580 4581Otherwise, if the MCE is a corrected error, KVM will just 4582store it in the corresponding bank (provided this bank is 4583not holding a previously reported uncorrected error). 4584 45854.107 KVM_S390_GET_CMMA_BITS 4586---------------------------- 4587 4588:Capability: KVM_CAP_S390_CMMA_MIGRATION 4589:Architectures: s390 4590:Type: vm ioctl 4591:Parameters: struct kvm_s390_cmma_log (in, out) 4592:Returns: 0 on success, a negative value on error 4593 4594Errors: 4595 4596 ====== ============================================================= 4597 ENOMEM not enough memory can be allocated to complete the task 4598 ENXIO if CMMA is not enabled 4599 EINVAL if KVM_S390_CMMA_PEEK is not set but migration mode was not enabled 4600 EINVAL if KVM_S390_CMMA_PEEK is not set but dirty tracking has been 4601 disabled (and thus migration mode was automatically disabled) 4602 EFAULT if the userspace address is invalid or if no page table is 4603 present for the addresses (e.g. when using hugepages). 4604 ====== ============================================================= 4605 4606This ioctl is used to get the values of the CMMA bits on the s390 4607architecture. It is meant to be used in two scenarios: 4608 4609- During live migration to save the CMMA values. Live migration needs 4610 to be enabled via the KVM_REQ_START_MIGRATION VM property. 4611- To non-destructively peek at the CMMA values, with the flag 4612 KVM_S390_CMMA_PEEK set. 4613 4614The ioctl takes parameters via the kvm_s390_cmma_log struct. The desired 4615values are written to a buffer whose location is indicated via the "values" 4616member in the kvm_s390_cmma_log struct. The values in the input struct are 4617also updated as needed. 4618 4619Each CMMA value takes up one byte. 4620 4621:: 4622 4623 struct kvm_s390_cmma_log { 4624 __u64 start_gfn; 4625 __u32 count; 4626 __u32 flags; 4627 union { 4628 __u64 remaining; 4629 __u64 mask; 4630 }; 4631 __u64 values; 4632 }; 4633 4634start_gfn is the number of the first guest frame whose CMMA values are 4635to be retrieved, 4636 4637count is the length of the buffer in bytes, 4638 4639values points to the buffer where the result will be written to. 4640 4641If count is greater than KVM_S390_SKEYS_MAX, then it is considered to be 4642KVM_S390_SKEYS_MAX. KVM_S390_SKEYS_MAX is re-used for consistency with 4643other ioctls. 4644 4645The result is written in the buffer pointed to by the field values, and 4646the values of the input parameter are updated as follows. 4647 4648Depending on the flags, different actions are performed. The only 4649supported flag so far is KVM_S390_CMMA_PEEK. 4650 4651The default behaviour if KVM_S390_CMMA_PEEK is not set is: 4652start_gfn will indicate the first page frame whose CMMA bits were dirty. 4653It is not necessarily the same as the one passed as input, as clean pages 4654are skipped. 4655 4656count will indicate the number of bytes actually written in the buffer. 4657It can (and very often will) be smaller than the input value, since the 4658buffer is only filled until 16 bytes of clean values are found (which 4659are then not copied in the buffer). Since a CMMA migration block needs 4660the base address and the length, for a total of 16 bytes, we will send 4661back some clean data if there is some dirty data afterwards, as long as 4662the size of the clean data does not exceed the size of the header. This 4663allows to minimize the amount of data to be saved or transferred over 4664the network at the expense of more roundtrips to userspace. The next 4665invocation of the ioctl will skip over all the clean values, saving 4666potentially more than just the 16 bytes we found. 4667 4668If KVM_S390_CMMA_PEEK is set: 4669the existing storage attributes are read even when not in migration 4670mode, and no other action is performed; 4671 4672the output start_gfn will be equal to the input start_gfn, 4673 4674the output count will be equal to the input count, except if the end of 4675memory has been reached. 4676 4677In both cases: 4678the field "remaining" will indicate the total number of dirty CMMA values 4679still remaining, or 0 if KVM_S390_CMMA_PEEK is set and migration mode is 4680not enabled. 4681 4682mask is unused. 4683 4684values points to the userspace buffer where the result will be stored. 4685 46864.108 KVM_S390_SET_CMMA_BITS 4687---------------------------- 4688 4689:Capability: KVM_CAP_S390_CMMA_MIGRATION 4690:Architectures: s390 4691:Type: vm ioctl 4692:Parameters: struct kvm_s390_cmma_log (in) 4693:Returns: 0 on success, a negative value on error 4694 4695This ioctl is used to set the values of the CMMA bits on the s390 4696architecture. It is meant to be used during live migration to restore 4697the CMMA values, but there are no restrictions on its use. 4698The ioctl takes parameters via the kvm_s390_cmma_values struct. 4699Each CMMA value takes up one byte. 4700 4701:: 4702 4703 struct kvm_s390_cmma_log { 4704 __u64 start_gfn; 4705 __u32 count; 4706 __u32 flags; 4707 union { 4708 __u64 remaining; 4709 __u64 mask; 4710 }; 4711 __u64 values; 4712 }; 4713 4714start_gfn indicates the starting guest frame number, 4715 4716count indicates how many values are to be considered in the buffer, 4717 4718flags is not used and must be 0. 4719 4720mask indicates which PGSTE bits are to be considered. 4721 4722remaining is not used. 4723 4724values points to the buffer in userspace where to store the values. 4725 4726This ioctl can fail with -ENOMEM if not enough memory can be allocated to 4727complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if 4728the count field is too large (e.g. more than KVM_S390_CMMA_SIZE_MAX) or 4729if the flags field was not 0, with -EFAULT if the userspace address is 4730invalid, if invalid pages are written to (e.g. after the end of memory) 4731or if no page table is present for the addresses (e.g. when using 4732hugepages). 4733 47344.109 KVM_PPC_GET_CPU_CHAR 4735-------------------------- 4736 4737:Capability: KVM_CAP_PPC_GET_CPU_CHAR 4738:Architectures: powerpc 4739:Type: vm ioctl 4740:Parameters: struct kvm_ppc_cpu_char (out) 4741:Returns: 0 on successful completion, 4742 -EFAULT if struct kvm_ppc_cpu_char cannot be written 4743 4744This ioctl gives userspace information about certain characteristics 4745of the CPU relating to speculative execution of instructions and 4746possible information leakage resulting from speculative execution (see 4747CVE-2017-5715, CVE-2017-5753 and CVE-2017-5754). The information is 4748returned in struct kvm_ppc_cpu_char, which looks like this:: 4749 4750 struct kvm_ppc_cpu_char { 4751 __u64 character; /* characteristics of the CPU */ 4752 __u64 behaviour; /* recommended software behaviour */ 4753 __u64 character_mask; /* valid bits in character */ 4754 __u64 behaviour_mask; /* valid bits in behaviour */ 4755 }; 4756 4757For extensibility, the character_mask and behaviour_mask fields 4758indicate which bits of character and behaviour have been filled in by 4759the kernel. If the set of defined bits is extended in future then 4760userspace will be able to tell whether it is running on a kernel that 4761knows about the new bits. 4762 4763The character field describes attributes of the CPU which can help 4764with preventing inadvertent information disclosure - specifically, 4765whether there is an instruction to flash-invalidate the L1 data cache 4766(ori 30,30,0 or mtspr SPRN_TRIG2,rN), whether the L1 data cache is set 4767to a mode where entries can only be used by the thread that created 4768them, whether the bcctr[l] instruction prevents speculation, and 4769whether a speculation barrier instruction (ori 31,31,0) is provided. 4770 4771The behaviour field describes actions that software should take to 4772prevent inadvertent information disclosure, and thus describes which 4773vulnerabilities the hardware is subject to; specifically whether the 4774L1 data cache should be flushed when returning to user mode from the 4775kernel, and whether a speculation barrier should be placed between an 4776array bounds check and the array access. 4777 4778These fields use the same bit definitions as the new 4779H_GET_CPU_CHARACTERISTICS hypercall. 4780 47814.110 KVM_MEMORY_ENCRYPT_OP 4782--------------------------- 4783 4784:Capability: basic 4785:Architectures: x86 4786:Type: vm ioctl, vcpu ioctl 4787:Parameters: an opaque platform specific structure (in/out) 4788:Returns: 0 on success; -1 on error 4789 4790If the platform supports creating encrypted VMs then this ioctl can be used 4791for issuing platform-specific memory encryption commands to manage those 4792encrypted VMs. 4793 4794Currently, this ioctl is used for issuing both Secure Encrypted Virtualization 4795(SEV) commands on AMD Processors and Trusted Domain Extensions (TDX) commands 4796on Intel Processors. The detailed commands are defined in 4797Documentation/virt/kvm/x86/amd-memory-encryption.rst and 4798Documentation/virt/kvm/x86/intel-tdx.rst. 4799 48004.111 KVM_MEMORY_ENCRYPT_REG_REGION 4801----------------------------------- 4802 4803:Capability: basic 4804:Architectures: x86 4805:Type: system 4806:Parameters: struct kvm_enc_region (in) 4807:Returns: 0 on success; -1 on error 4808 4809This ioctl can be used to register a guest memory region which may 4810contain encrypted data (e.g. guest RAM, SMRAM etc). 4811 4812It is used in the SEV-enabled guest. When encryption is enabled, a guest 4813memory region may contain encrypted data. The SEV memory encryption 4814engine uses a tweak such that two identical plaintext pages, each at 4815different locations will have differing ciphertexts. So swapping or 4816moving ciphertext of those pages will not result in plaintext being 4817swapped. So relocating (or migrating) physical backing pages for the SEV 4818guest will require some additional steps. 4819 4820Note: The current SEV key management spec does not provide commands to 4821swap or migrate (move) ciphertext pages. Hence, for now we pin the guest 4822memory region registered with the ioctl. 4823 48244.112 KVM_MEMORY_ENCRYPT_UNREG_REGION 4825------------------------------------- 4826 4827:Capability: basic 4828:Architectures: x86 4829:Type: system 4830:Parameters: struct kvm_enc_region (in) 4831:Returns: 0 on success; -1 on error 4832 4833This ioctl can be used to unregister the guest memory region registered 4834with KVM_MEMORY_ENCRYPT_REG_REGION ioctl above. 4835 48364.113 KVM_HYPERV_EVENTFD 4837------------------------ 4838 4839:Capability: KVM_CAP_HYPERV_EVENTFD 4840:Architectures: x86 4841:Type: vm ioctl 4842:Parameters: struct kvm_hyperv_eventfd (in) 4843 4844This ioctl (un)registers an eventfd to receive notifications from the guest on 4845the specified Hyper-V connection id through the SIGNAL_EVENT hypercall, without 4846causing a user exit. SIGNAL_EVENT hypercall with non-zero event flag number 4847(bits 24-31) still triggers a KVM_EXIT_HYPERV_HCALL user exit. 4848 4849:: 4850 4851 struct kvm_hyperv_eventfd { 4852 __u32 conn_id; 4853 __s32 fd; 4854 __u32 flags; 4855 __u32 padding[3]; 4856 }; 4857 4858The conn_id field should fit within 24 bits:: 4859 4860 #define KVM_HYPERV_CONN_ID_MASK 0x00ffffff 4861 4862The acceptable values for the flags field are:: 4863 4864 #define KVM_HYPERV_EVENTFD_DEASSIGN (1 << 0) 4865 4866:Returns: 0 on success, 4867 -EINVAL if conn_id or flags is outside the allowed range, 4868 -ENOENT on deassign if the conn_id isn't registered, 4869 -EEXIST on assign if the conn_id is already registered 4870 48714.114 KVM_GET_NESTED_STATE 4872-------------------------- 4873 4874:Capability: KVM_CAP_NESTED_STATE 4875:Architectures: x86 4876:Type: vcpu ioctl 4877:Parameters: struct kvm_nested_state (in/out) 4878:Returns: 0 on success, -1 on error 4879 4880Errors: 4881 4882 ===== ============================================================= 4883 E2BIG the total state size exceeds the value of 'size' specified by 4884 the user; the size required will be written into size. 4885 ===== ============================================================= 4886 4887:: 4888 4889 struct kvm_nested_state { 4890 __u16 flags; 4891 __u16 format; 4892 __u32 size; 4893 4894 union { 4895 struct kvm_vmx_nested_state_hdr vmx; 4896 struct kvm_svm_nested_state_hdr svm; 4897 4898 /* Pad the header to 128 bytes. */ 4899 __u8 pad[120]; 4900 } hdr; 4901 4902 union { 4903 struct kvm_vmx_nested_state_data vmx[0]; 4904 struct kvm_svm_nested_state_data svm[0]; 4905 } data; 4906 }; 4907 4908 #define KVM_STATE_NESTED_GUEST_MODE 0x00000001 4909 #define KVM_STATE_NESTED_RUN_PENDING 0x00000002 4910 #define KVM_STATE_NESTED_EVMCS 0x00000004 4911 4912 #define KVM_STATE_NESTED_FORMAT_VMX 0 4913 #define KVM_STATE_NESTED_FORMAT_SVM 1 4914 4915 #define KVM_STATE_NESTED_VMX_VMCS_SIZE 0x1000 4916 4917 #define KVM_STATE_NESTED_VMX_SMM_GUEST_MODE 0x00000001 4918 #define KVM_STATE_NESTED_VMX_SMM_VMXON 0x00000002 4919 4920 #define KVM_STATE_VMX_PREEMPTION_TIMER_DEADLINE 0x00000001 4921 4922 struct kvm_vmx_nested_state_hdr { 4923 __u64 vmxon_pa; 4924 __u64 vmcs12_pa; 4925 4926 struct { 4927 __u16 flags; 4928 } smm; 4929 4930 __u32 flags; 4931 __u64 preemption_timer_deadline; 4932 }; 4933 4934 struct kvm_vmx_nested_state_data { 4935 __u8 vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE]; 4936 __u8 shadow_vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE]; 4937 }; 4938 4939This ioctl copies the vcpu's nested virtualization state from the kernel to 4940userspace. 4941 4942The maximum size of the state can be retrieved by passing KVM_CAP_NESTED_STATE 4943to the KVM_CHECK_EXTENSION ioctl(). 4944 49454.115 KVM_SET_NESTED_STATE 4946-------------------------- 4947 4948:Capability: KVM_CAP_NESTED_STATE 4949:Architectures: x86 4950:Type: vcpu ioctl 4951:Parameters: struct kvm_nested_state (in) 4952:Returns: 0 on success, -1 on error 4953 4954This copies the vcpu's kvm_nested_state struct from userspace to the kernel. 4955For the definition of struct kvm_nested_state, see KVM_GET_NESTED_STATE. 4956 49574.116 KVM_(UN)REGISTER_COALESCED_MMIO 4958------------------------------------- 4959 4960:Capability: KVM_CAP_COALESCED_MMIO (for coalesced mmio) 4961 KVM_CAP_COALESCED_PIO (for coalesced pio) 4962:Architectures: all 4963:Type: vm ioctl 4964:Parameters: struct kvm_coalesced_mmio_zone 4965:Returns: 0 on success, < 0 on error 4966 4967Coalesced I/O is a performance optimization that defers hardware 4968register write emulation so that userspace exits are avoided. It is 4969typically used to reduce the overhead of emulating frequently accessed 4970hardware registers. 4971 4972When a hardware register is configured for coalesced I/O, write accesses 4973do not exit to userspace and their value is recorded in a ring buffer 4974that is shared between kernel and userspace. 4975 4976Coalesced I/O is used if one or more write accesses to a hardware 4977register can be deferred until a read or a write to another hardware 4978register on the same device. This last access will cause a vmexit and 4979userspace will process accesses from the ring buffer before emulating 4980it. That will avoid exiting to userspace on repeated writes. 4981 4982Coalesced pio is based on coalesced mmio. There is little difference 4983between coalesced mmio and pio except that coalesced pio records accesses 4984to I/O ports. 4985 49864.117 KVM_CLEAR_DIRTY_LOG 4987------------------------- 4988 4989:Capability: KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 4990:Architectures: x86, arm64, mips 4991:Type: vm ioctl 4992:Parameters: struct kvm_clear_dirty_log (in) 4993:Returns: 0 on success, -1 on error 4994 4995:: 4996 4997 /* for KVM_CLEAR_DIRTY_LOG */ 4998 struct kvm_clear_dirty_log { 4999 __u32 slot; 5000 __u32 num_pages; 5001 __u64 first_page; 5002 union { 5003 void __user *dirty_bitmap; /* one bit per page */ 5004 __u64 padding; 5005 }; 5006 }; 5007 5008The ioctl clears the dirty status of pages in a memory slot, according to 5009the bitmap that is passed in struct kvm_clear_dirty_log's dirty_bitmap 5010field. Bit 0 of the bitmap corresponds to page "first_page" in the 5011memory slot, and num_pages is the size in bits of the input bitmap. 5012first_page must be a multiple of 64; num_pages must also be a multiple of 501364 unless first_page + num_pages is the size of the memory slot. For each 5014bit that is set in the input bitmap, the corresponding page is marked "clean" 5015in KVM's dirty bitmap, and dirty tracking is re-enabled for that page 5016(for example via write-protection, or by clearing the dirty bit in 5017a page table entry). 5018 5019If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies 5020the address space for which you want to clear the dirty status. See 5021KVM_SET_USER_MEMORY_REGION for details on the usage of slot field. 5022 5023This ioctl is mostly useful when KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 5024is enabled; for more information, see the description of the capability. 5025However, it can always be used as long as KVM_CHECK_EXTENSION confirms 5026that KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is present. 5027 50284.118 KVM_GET_SUPPORTED_HV_CPUID 5029-------------------------------- 5030 5031:Capability: KVM_CAP_HYPERV_CPUID (vcpu), KVM_CAP_SYS_HYPERV_CPUID (system) 5032:Architectures: x86 5033:Type: system ioctl, vcpu ioctl 5034:Parameters: struct kvm_cpuid2 (in/out) 5035:Returns: 0 on success, -1 on error 5036 5037:: 5038 5039 struct kvm_cpuid2 { 5040 __u32 nent; 5041 __u32 padding; 5042 struct kvm_cpuid_entry2 entries[0]; 5043 }; 5044 5045 struct kvm_cpuid_entry2 { 5046 __u32 function; 5047 __u32 index; 5048 __u32 flags; 5049 __u32 eax; 5050 __u32 ebx; 5051 __u32 ecx; 5052 __u32 edx; 5053 __u32 padding[3]; 5054 }; 5055 5056This ioctl returns x86 cpuid features leaves related to Hyper-V emulation in 5057KVM. Userspace can use the information returned by this ioctl to construct 5058cpuid information presented to guests consuming Hyper-V enlightenments (e.g. 5059Windows or Hyper-V guests). 5060 5061CPUID feature leaves returned by this ioctl are defined by Hyper-V Top Level 5062Functional Specification (TLFS). These leaves can't be obtained with 5063KVM_GET_SUPPORTED_CPUID ioctl because some of them intersect with KVM feature 5064leaves (0x40000000, 0x40000001). 5065 5066Currently, the following list of CPUID leaves are returned: 5067 5068 - HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS 5069 - HYPERV_CPUID_INTERFACE 5070 - HYPERV_CPUID_VERSION 5071 - HYPERV_CPUID_FEATURES 5072 - HYPERV_CPUID_ENLIGHTMENT_INFO 5073 - HYPERV_CPUID_IMPLEMENT_LIMITS 5074 - HYPERV_CPUID_NESTED_FEATURES 5075 - HYPERV_CPUID_SYNDBG_VENDOR_AND_MAX_FUNCTIONS 5076 - HYPERV_CPUID_SYNDBG_INTERFACE 5077 - HYPERV_CPUID_SYNDBG_PLATFORM_CAPABILITIES 5078 5079Userspace invokes KVM_GET_SUPPORTED_HV_CPUID by passing a kvm_cpuid2 structure 5080with the 'nent' field indicating the number of entries in the variable-size 5081array 'entries'. If the number of entries is too low to describe all Hyper-V 5082feature leaves, an error (E2BIG) is returned. If the number is more or equal 5083to the number of Hyper-V feature leaves, the 'nent' field is adjusted to the 5084number of valid entries in the 'entries' array, which is then filled. 5085 5086'index' and 'flags' fields in 'struct kvm_cpuid_entry2' are currently reserved, 5087userspace should not expect to get any particular value there. 5088 5089Note, vcpu version of KVM_GET_SUPPORTED_HV_CPUID is currently deprecated. Unlike 5090system ioctl which exposes all supported feature bits unconditionally, vcpu 5091version has the following quirks: 5092 5093- HYPERV_CPUID_NESTED_FEATURES leaf and HV_X64_ENLIGHTENED_VMCS_RECOMMENDED 5094 feature bit are only exposed when Enlightened VMCS was previously enabled 5095 on the corresponding vCPU (KVM_CAP_HYPERV_ENLIGHTENED_VMCS). 5096- HV_STIMER_DIRECT_MODE_AVAILABLE bit is only exposed with in-kernel LAPIC. 5097 (presumes KVM_CREATE_IRQCHIP has already been called). 5098 50994.119 KVM_ARM_VCPU_FINALIZE 5100--------------------------- 5101 5102:Architectures: arm64 5103:Type: vcpu ioctl 5104:Parameters: int feature (in) 5105:Returns: 0 on success, -1 on error 5106 5107Errors: 5108 5109 ====== ============================================================== 5110 EPERM feature not enabled, needs configuration, or already finalized 5111 EINVAL feature unknown or not present 5112 ====== ============================================================== 5113 5114Recognised values for feature: 5115 5116 ===== =========================================== 5117 arm64 KVM_ARM_VCPU_SVE (requires KVM_CAP_ARM_SVE) 5118 ===== =========================================== 5119 5120Finalizes the configuration of the specified vcpu feature. 5121 5122The vcpu must already have been initialised, enabling the affected feature, by 5123means of a successful :ref:`KVM_ARM_VCPU_INIT <KVM_ARM_VCPU_INIT>` call with the 5124appropriate flag set in features[]. 5125 5126For affected vcpu features, this is a mandatory step that must be performed 5127before the vcpu is fully usable. 5128 5129Between KVM_ARM_VCPU_INIT and KVM_ARM_VCPU_FINALIZE, the feature may be 5130configured by use of ioctls such as KVM_SET_ONE_REG. The exact configuration 5131that should be performed and how to do it are feature-dependent. 5132 5133Other calls that depend on a particular feature being finalized, such as 5134KVM_RUN, KVM_GET_REG_LIST, KVM_GET_ONE_REG and KVM_SET_ONE_REG, will fail with 5135-EPERM unless the feature has already been finalized by means of a 5136KVM_ARM_VCPU_FINALIZE call. 5137 5138See KVM_ARM_VCPU_INIT for details of vcpu features that require finalization 5139using this ioctl. 5140 51414.120 KVM_SET_PMU_EVENT_FILTER 5142------------------------------ 5143 5144:Capability: KVM_CAP_PMU_EVENT_FILTER 5145:Architectures: x86 5146:Type: vm ioctl 5147:Parameters: struct kvm_pmu_event_filter (in) 5148:Returns: 0 on success, -1 on error 5149 5150Errors: 5151 5152 ====== ============================================================ 5153 EFAULT args[0] cannot be accessed 5154 EINVAL args[0] contains invalid data in the filter or filter events 5155 E2BIG nevents is too large 5156 EBUSY not enough memory to allocate the filter 5157 ====== ============================================================ 5158 5159:: 5160 5161 struct kvm_pmu_event_filter { 5162 __u32 action; 5163 __u32 nevents; 5164 __u32 fixed_counter_bitmap; 5165 __u32 flags; 5166 __u32 pad[4]; 5167 __u64 events[0]; 5168 }; 5169 5170This ioctl restricts the set of PMU events the guest can program by limiting 5171which event select and unit mask combinations are permitted. 5172 5173The argument holds a list of filter events which will be allowed or denied. 5174 5175Filter events only control general purpose counters; fixed purpose counters 5176are controlled by the fixed_counter_bitmap. 5177 5178Valid values for 'flags':: 5179 5180``0`` 5181 5182To use this mode, clear the 'flags' field. 5183 5184In this mode each event will contain an event select + unit mask. 5185 5186When the guest attempts to program the PMU the guest's event select + 5187unit mask is compared against the filter events to determine whether the 5188guest should have access. 5189 5190``KVM_PMU_EVENT_FLAG_MASKED_EVENTS`` 5191:Capability: KVM_CAP_PMU_EVENT_MASKED_EVENTS 5192 5193In this mode each filter event will contain an event select, mask, match, and 5194exclude value. To encode a masked event use:: 5195 5196 KVM_PMU_ENCODE_MASKED_ENTRY() 5197 5198An encoded event will follow this layout:: 5199 5200 Bits Description 5201 ---- ----------- 5202 7:0 event select (low bits) 5203 15:8 umask match 5204 31:16 unused 5205 35:32 event select (high bits) 5206 36:54 unused 5207 55 exclude bit 5208 63:56 umask mask 5209 5210When the guest attempts to program the PMU, these steps are followed in 5211determining if the guest should have access: 5212 5213 1. Match the event select from the guest against the filter events. 5214 2. If a match is found, match the guest's unit mask to the mask and match 5215 values of the included filter events. 5216 I.e. (unit mask & mask) == match && !exclude. 5217 3. If a match is found, match the guest's unit mask to the mask and match 5218 values of the excluded filter events. 5219 I.e. (unit mask & mask) == match && exclude. 5220 4. 5221 a. If an included match is found and an excluded match is not found, filter 5222 the event. 5223 b. For everything else, do not filter the event. 5224 5. 5225 a. If the event is filtered and it's an allow list, allow the guest to 5226 program the event. 5227 b. If the event is filtered and it's a deny list, do not allow the guest to 5228 program the event. 5229 5230When setting a new pmu event filter, -EINVAL will be returned if any of the 5231unused fields are set or if any of the high bits (35:32) in the event 5232select are set when called on Intel. 5233 5234Valid values for 'action':: 5235 5236 #define KVM_PMU_EVENT_ALLOW 0 5237 #define KVM_PMU_EVENT_DENY 1 5238 5239Via this API, KVM userspace can also control the behavior of the VM's fixed 5240counters (if any) by configuring the "action" and "fixed_counter_bitmap" fields. 5241 5242Specifically, KVM follows the following pseudo-code when determining whether to 5243allow the guest FixCtr[i] to count its pre-defined fixed event:: 5244 5245 FixCtr[i]_is_allowed = (action == ALLOW) && (bitmap & BIT(i)) || 5246 (action == DENY) && !(bitmap & BIT(i)); 5247 FixCtr[i]_is_denied = !FixCtr[i]_is_allowed; 5248 5249KVM always consumes fixed_counter_bitmap, it's userspace's responsibility to 5250ensure fixed_counter_bitmap is set correctly, e.g. if userspace wants to define 5251a filter that only affects general purpose counters. 5252 5253Note, the "events" field also applies to fixed counters' hardcoded event_select 5254and unit_mask values. "fixed_counter_bitmap" has higher priority than "events" 5255if there is a contradiction between the two. 5256 52574.121 KVM_PPC_SVM_OFF 5258--------------------- 5259 5260:Capability: basic 5261:Architectures: powerpc 5262:Type: vm ioctl 5263:Parameters: none 5264:Returns: 0 on successful completion, 5265 5266Errors: 5267 5268 ====== ================================================================ 5269 EINVAL if ultravisor failed to terminate the secure guest 5270 ENOMEM if hypervisor failed to allocate new radix page tables for guest 5271 ====== ================================================================ 5272 5273This ioctl is used to turn off the secure mode of the guest or transition 5274the guest from secure mode to normal mode. This is invoked when the guest 5275is reset. This has no effect if called for a normal guest. 5276 5277This ioctl issues an ultravisor call to terminate the secure guest, 5278unpins the VPA pages and releases all the device pages that are used to 5279track the secure pages by hypervisor. 5280 52814.122 KVM_S390_NORMAL_RESET 5282--------------------------- 5283 5284:Capability: KVM_CAP_S390_VCPU_RESETS 5285:Architectures: s390 5286:Type: vcpu ioctl 5287:Parameters: none 5288:Returns: 0 5289 5290This ioctl resets VCPU registers and control structures according to 5291the cpu reset definition in the POP (Principles Of Operation). 5292 52934.123 KVM_S390_INITIAL_RESET 5294---------------------------- 5295 5296:Capability: basic 5297:Architectures: s390 5298:Type: vcpu ioctl 5299:Parameters: none 5300:Returns: 0 5301 5302This ioctl resets VCPU registers and control structures according to 5303the initial cpu reset definition in the POP. However, the cpu is not 5304put into ESA mode. This reset is a superset of the normal reset. 5305 53064.124 KVM_S390_CLEAR_RESET 5307-------------------------- 5308 5309:Capability: KVM_CAP_S390_VCPU_RESETS 5310:Architectures: s390 5311:Type: vcpu ioctl 5312:Parameters: none 5313:Returns: 0 5314 5315This ioctl resets VCPU registers and control structures according to 5316the clear cpu reset definition in the POP. However, the cpu is not put 5317into ESA mode. This reset is a superset of the initial reset. 5318 5319 53204.125 KVM_S390_PV_COMMAND 5321------------------------- 5322 5323:Capability: KVM_CAP_S390_PROTECTED 5324:Architectures: s390 5325:Type: vm ioctl 5326:Parameters: struct kvm_pv_cmd 5327:Returns: 0 on success, < 0 on error 5328 5329:: 5330 5331 struct kvm_pv_cmd { 5332 __u32 cmd; /* Command to be executed */ 5333 __u16 rc; /* Ultravisor return code */ 5334 __u16 rrc; /* Ultravisor return reason code */ 5335 __u64 data; /* Data or address */ 5336 __u32 flags; /* flags for future extensions. Must be 0 for now */ 5337 __u32 reserved[3]; 5338 }; 5339 5340**Ultravisor return codes** 5341The Ultravisor return (reason) codes are provided by the kernel if a 5342Ultravisor call has been executed to achieve the results expected by 5343the command. Therefore they are independent of the IOCTL return 5344code. If KVM changes `rc`, its value will always be greater than 0 5345hence setting it to 0 before issuing a PV command is advised to be 5346able to detect a change of `rc`. 5347 5348**cmd values:** 5349 5350KVM_PV_ENABLE 5351 Allocate memory and register the VM with the Ultravisor, thereby 5352 donating memory to the Ultravisor that will become inaccessible to 5353 KVM. All existing CPUs are converted to protected ones. After this 5354 command has succeeded, any CPU added via hotplug will become 5355 protected during its creation as well. 5356 5357 Errors: 5358 5359 ===== ============================= 5360 EINTR an unmasked signal is pending 5361 ===== ============================= 5362 5363KVM_PV_DISABLE 5364 Deregister the VM from the Ultravisor and reclaim the memory that had 5365 been donated to the Ultravisor, making it usable by the kernel again. 5366 All registered VCPUs are converted back to non-protected ones. If a 5367 previous protected VM had been prepared for asynchronous teardown with 5368 KVM_PV_ASYNC_CLEANUP_PREPARE and not subsequently torn down with 5369 KVM_PV_ASYNC_CLEANUP_PERFORM, it will be torn down in this call 5370 together with the current protected VM. 5371 5372KVM_PV_VM_SET_SEC_PARMS 5373 Pass the image header from VM memory to the Ultravisor in 5374 preparation of image unpacking and verification. 5375 5376KVM_PV_VM_UNPACK 5377 Unpack (protect and decrypt) a page of the encrypted boot image. 5378 5379KVM_PV_VM_VERIFY 5380 Verify the integrity of the unpacked image. Only if this succeeds, 5381 KVM is allowed to start protected VCPUs. 5382 5383KVM_PV_INFO 5384 :Capability: KVM_CAP_S390_PROTECTED_DUMP 5385 5386 Presents an API that provides Ultravisor related data to userspace 5387 via subcommands. len_max is the size of the user space buffer, 5388 len_written is KVM's indication of how much bytes of that buffer 5389 were actually written to. len_written can be used to determine the 5390 valid fields if more response fields are added in the future. 5391 5392 :: 5393 5394 enum pv_cmd_info_id { 5395 KVM_PV_INFO_VM, 5396 KVM_PV_INFO_DUMP, 5397 }; 5398 5399 struct kvm_s390_pv_info_header { 5400 __u32 id; 5401 __u32 len_max; 5402 __u32 len_written; 5403 __u32 reserved; 5404 }; 5405 5406 struct kvm_s390_pv_info { 5407 struct kvm_s390_pv_info_header header; 5408 struct kvm_s390_pv_info_dump dump; 5409 struct kvm_s390_pv_info_vm vm; 5410 }; 5411 5412**subcommands:** 5413 5414 KVM_PV_INFO_VM 5415 This subcommand provides basic Ultravisor information for PV 5416 hosts. These values are likely also exported as files in the sysfs 5417 firmware UV query interface but they are more easily available to 5418 programs in this API. 5419 5420 The installed calls and feature_indication members provide the 5421 installed UV calls and the UV's other feature indications. 5422 5423 The max_* members provide information about the maximum number of PV 5424 vcpus, PV guests and PV guest memory size. 5425 5426 :: 5427 5428 struct kvm_s390_pv_info_vm { 5429 __u64 inst_calls_list[4]; 5430 __u64 max_cpus; 5431 __u64 max_guests; 5432 __u64 max_guest_addr; 5433 __u64 feature_indication; 5434 }; 5435 5436 5437 KVM_PV_INFO_DUMP 5438 This subcommand provides information related to dumping PV guests. 5439 5440 :: 5441 5442 struct kvm_s390_pv_info_dump { 5443 __u64 dump_cpu_buffer_len; 5444 __u64 dump_config_mem_buffer_per_1m; 5445 __u64 dump_config_finalize_len; 5446 }; 5447 5448KVM_PV_DUMP 5449 :Capability: KVM_CAP_S390_PROTECTED_DUMP 5450 5451 Presents an API that provides calls which facilitate dumping a 5452 protected VM. 5453 5454 :: 5455 5456 struct kvm_s390_pv_dmp { 5457 __u64 subcmd; 5458 __u64 buff_addr; 5459 __u64 buff_len; 5460 __u64 gaddr; /* For dump storage state */ 5461 }; 5462 5463 **subcommands:** 5464 5465 KVM_PV_DUMP_INIT 5466 Initializes the dump process of a protected VM. If this call does 5467 not succeed all other subcommands will fail with -EINVAL. This 5468 subcommand will return -EINVAL if a dump process has not yet been 5469 completed. 5470 5471 Not all PV vms can be dumped, the owner needs to set `dump 5472 allowed` PCF bit 34 in the SE header to allow dumping. 5473 5474 KVM_PV_DUMP_CONFIG_STOR_STATE 5475 Stores `buff_len` bytes of tweak component values starting with 5476 the 1MB block specified by the absolute guest address 5477 (`gaddr`). `buff_len` needs to be `conf_dump_storage_state_len` 5478 aligned and at least >= the `conf_dump_storage_state_len` value 5479 provided by the dump uv_info data. buff_user might be written to 5480 even if an error rc is returned. For instance if we encounter a 5481 fault after writing the first page of data. 5482 5483 KVM_PV_DUMP_COMPLETE 5484 If the subcommand succeeds it completes the dump process and lets 5485 KVM_PV_DUMP_INIT be called again. 5486 5487 On success `conf_dump_finalize_len` bytes of completion data will be 5488 stored to the `buff_addr`. The completion data contains a key 5489 derivation seed, IV, tweak nonce and encryption keys as well as an 5490 authentication tag all of which are needed to decrypt the dump at a 5491 later time. 5492 5493KVM_PV_ASYNC_CLEANUP_PREPARE 5494 :Capability: KVM_CAP_S390_PROTECTED_ASYNC_DISABLE 5495 5496 Prepare the current protected VM for asynchronous teardown. Most 5497 resources used by the current protected VM will be set aside for a 5498 subsequent asynchronous teardown. The current protected VM will then 5499 resume execution immediately as non-protected. There can be at most 5500 one protected VM prepared for asynchronous teardown at any time. If 5501 a protected VM had already been prepared for teardown without 5502 subsequently calling KVM_PV_ASYNC_CLEANUP_PERFORM, this call will 5503 fail. In that case, the userspace process should issue a normal 5504 KVM_PV_DISABLE. The resources set aside with this call will need to 5505 be cleaned up with a subsequent call to KVM_PV_ASYNC_CLEANUP_PERFORM 5506 or KVM_PV_DISABLE, otherwise they will be cleaned up when KVM 5507 terminates. KVM_PV_ASYNC_CLEANUP_PREPARE can be called again as soon 5508 as cleanup starts, i.e. before KVM_PV_ASYNC_CLEANUP_PERFORM finishes. 5509 5510KVM_PV_ASYNC_CLEANUP_PERFORM 5511 :Capability: KVM_CAP_S390_PROTECTED_ASYNC_DISABLE 5512 5513 Tear down the protected VM previously prepared for teardown with 5514 KVM_PV_ASYNC_CLEANUP_PREPARE. The resources that had been set aside 5515 will be freed during the execution of this command. This PV command 5516 should ideally be issued by userspace from a separate thread. If a 5517 fatal signal is received (or the process terminates naturally), the 5518 command will terminate immediately without completing, and the normal 5519 KVM shutdown procedure will take care of cleaning up all remaining 5520 protected VMs, including the ones whose teardown was interrupted by 5521 process termination. 5522 55234.126 KVM_XEN_HVM_SET_ATTR 5524-------------------------- 5525 5526:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 5527:Architectures: x86 5528:Type: vm ioctl 5529:Parameters: struct kvm_xen_hvm_attr 5530:Returns: 0 on success, < 0 on error 5531 5532:: 5533 5534 struct kvm_xen_hvm_attr { 5535 __u16 type; 5536 __u16 pad[3]; 5537 union { 5538 __u8 long_mode; 5539 __u8 vector; 5540 __u8 runstate_update_flag; 5541 union { 5542 __u64 gfn; 5543 __u64 hva; 5544 } shared_info; 5545 struct { 5546 __u32 send_port; 5547 __u32 type; /* EVTCHNSTAT_ipi / EVTCHNSTAT_interdomain */ 5548 __u32 flags; 5549 union { 5550 struct { 5551 __u32 port; 5552 __u32 vcpu; 5553 __u32 priority; 5554 } port; 5555 struct { 5556 __u32 port; /* Zero for eventfd */ 5557 __s32 fd; 5558 } eventfd; 5559 __u32 padding[4]; 5560 } deliver; 5561 } evtchn; 5562 __u32 xen_version; 5563 __u64 pad[8]; 5564 } u; 5565 }; 5566 5567type values: 5568 5569KVM_XEN_ATTR_TYPE_LONG_MODE 5570 Sets the ABI mode of the VM to 32-bit or 64-bit (long mode). This 5571 determines the layout of the shared_info page exposed to the VM. 5572 5573KVM_XEN_ATTR_TYPE_SHARED_INFO 5574 Sets the guest physical frame number at which the Xen shared_info 5575 page resides. Note that although Xen places vcpu_info for the first 5576 32 vCPUs in the shared_info page, KVM does not automatically do so 5577 and instead requires that KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO or 5578 KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO_HVA be used explicitly even when 5579 the vcpu_info for a given vCPU resides at the "default" location 5580 in the shared_info page. This is because KVM may not be aware of 5581 the Xen CPU id which is used as the index into the vcpu_info[] 5582 array, so may know the correct default location. 5583 5584 Note that the shared_info page may be constantly written to by KVM; 5585 it contains the event channel bitmap used to deliver interrupts to 5586 a Xen guest, amongst other things. It is exempt from dirty tracking 5587 mechanisms — KVM will not explicitly mark the page as dirty each 5588 time an event channel interrupt is delivered to the guest! Thus, 5589 userspace should always assume that the designated GFN is dirty if 5590 any vCPU has been running or any event channel interrupts can be 5591 routed to the guest. 5592 5593 Setting the gfn to KVM_XEN_INVALID_GFN will disable the shared_info 5594 page. 5595 5596KVM_XEN_ATTR_TYPE_SHARED_INFO_HVA 5597 If the KVM_XEN_HVM_CONFIG_SHARED_INFO_HVA flag is also set in the 5598 Xen capabilities, then this attribute may be used to set the 5599 userspace address at which the shared_info page resides, which 5600 will always be fixed in the VMM regardless of where it is mapped 5601 in guest physical address space. This attribute should be used in 5602 preference to KVM_XEN_ATTR_TYPE_SHARED_INFO as it avoids 5603 unnecessary invalidation of an internal cache when the page is 5604 re-mapped in guest physical address space. 5605 5606 Setting the hva to zero will disable the shared_info page. 5607 5608KVM_XEN_ATTR_TYPE_UPCALL_VECTOR 5609 Sets the exception vector used to deliver Xen event channel upcalls. 5610 This is the HVM-wide vector injected directly by the hypervisor 5611 (not through the local APIC), typically configured by a guest via 5612 HVM_PARAM_CALLBACK_IRQ. This can be disabled again (e.g. for guest 5613 SHUTDOWN_soft_reset) by setting it to zero. 5614 5615KVM_XEN_ATTR_TYPE_EVTCHN 5616 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5617 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It configures 5618 an outbound port number for interception of EVTCHNOP_send requests 5619 from the guest. A given sending port number may be directed back to 5620 a specified vCPU (by APIC ID) / port / priority on the guest, or to 5621 trigger events on an eventfd. The vCPU and priority can be changed 5622 by setting KVM_XEN_EVTCHN_UPDATE in a subsequent call, but other 5623 fields cannot change for a given sending port. A port mapping is 5624 removed by using KVM_XEN_EVTCHN_DEASSIGN in the flags field. Passing 5625 KVM_XEN_EVTCHN_RESET in the flags field removes all interception of 5626 outbound event channels. The values of the flags field are mutually 5627 exclusive and cannot be combined as a bitmask. 5628 5629KVM_XEN_ATTR_TYPE_XEN_VERSION 5630 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5631 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It configures 5632 the 32-bit version code returned to the guest when it invokes the 5633 XENVER_version call; typically (XEN_MAJOR << 16 | XEN_MINOR). PV 5634 Xen guests will often use this to as a dummy hypercall to trigger 5635 event channel delivery, so responding within the kernel without 5636 exiting to userspace is beneficial. 5637 5638KVM_XEN_ATTR_TYPE_RUNSTATE_UPDATE_FLAG 5639 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5640 support for KVM_XEN_HVM_CONFIG_RUNSTATE_UPDATE_FLAG. It enables the 5641 XEN_RUNSTATE_UPDATE flag which allows guest vCPUs to safely read 5642 other vCPUs' vcpu_runstate_info. Xen guests enable this feature via 5643 the VMASST_TYPE_runstate_update_flag of the HYPERVISOR_vm_assist 5644 hypercall. 5645 56464.127 KVM_XEN_HVM_GET_ATTR 5647-------------------------- 5648 5649:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 5650:Architectures: x86 5651:Type: vm ioctl 5652:Parameters: struct kvm_xen_hvm_attr 5653:Returns: 0 on success, < 0 on error 5654 5655Allows Xen VM attributes to be read. For the structure and types, 5656see KVM_XEN_HVM_SET_ATTR above. The KVM_XEN_ATTR_TYPE_EVTCHN 5657attribute cannot be read. 5658 56594.128 KVM_XEN_VCPU_SET_ATTR 5660--------------------------- 5661 5662:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 5663:Architectures: x86 5664:Type: vcpu ioctl 5665:Parameters: struct kvm_xen_vcpu_attr 5666:Returns: 0 on success, < 0 on error 5667 5668:: 5669 5670 struct kvm_xen_vcpu_attr { 5671 __u16 type; 5672 __u16 pad[3]; 5673 union { 5674 __u64 gpa; 5675 __u64 pad[4]; 5676 struct { 5677 __u64 state; 5678 __u64 state_entry_time; 5679 __u64 time_running; 5680 __u64 time_runnable; 5681 __u64 time_blocked; 5682 __u64 time_offline; 5683 } runstate; 5684 __u32 vcpu_id; 5685 struct { 5686 __u32 port; 5687 __u32 priority; 5688 __u64 expires_ns; 5689 } timer; 5690 __u8 vector; 5691 } u; 5692 }; 5693 5694type values: 5695 5696KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO 5697 Sets the guest physical address of the vcpu_info for a given vCPU. 5698 As with the shared_info page for the VM, the corresponding page may be 5699 dirtied at any time if event channel interrupt delivery is enabled, so 5700 userspace should always assume that the page is dirty without relying 5701 on dirty logging. Setting the gpa to KVM_XEN_INVALID_GPA will disable 5702 the vcpu_info. 5703 5704KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO_HVA 5705 If the KVM_XEN_HVM_CONFIG_SHARED_INFO_HVA flag is also set in the 5706 Xen capabilities, then this attribute may be used to set the 5707 userspace address of the vcpu_info for a given vCPU. It should 5708 only be used when the vcpu_info resides at the "default" location 5709 in the shared_info page. In this case it is safe to assume the 5710 userspace address will not change, because the shared_info page is 5711 an overlay on guest memory and remains at a fixed host address 5712 regardless of where it is mapped in guest physical address space 5713 and hence unnecessary invalidation of an internal cache may be 5714 avoided if the guest memory layout is modified. 5715 If the vcpu_info does not reside at the "default" location then 5716 it is not guaranteed to remain at the same host address and 5717 hence the aforementioned cache invalidation is required. 5718 5719KVM_XEN_VCPU_ATTR_TYPE_VCPU_TIME_INFO 5720 Sets the guest physical address of an additional pvclock structure 5721 for a given vCPU. This is typically used for guest vsyscall support. 5722 Setting the gpa to KVM_XEN_INVALID_GPA will disable the structure. 5723 5724KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR 5725 Sets the guest physical address of the vcpu_runstate_info for a given 5726 vCPU. This is how a Xen guest tracks CPU state such as steal time. 5727 Setting the gpa to KVM_XEN_INVALID_GPA will disable the runstate area. 5728 5729KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_CURRENT 5730 Sets the runstate (RUNSTATE_running/_runnable/_blocked/_offline) of 5731 the given vCPU from the .u.runstate.state member of the structure. 5732 KVM automatically accounts running and runnable time but blocked 5733 and offline states are only entered explicitly. 5734 5735KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_DATA 5736 Sets all fields of the vCPU runstate data from the .u.runstate member 5737 of the structure, including the current runstate. The state_entry_time 5738 must equal the sum of the other four times. 5739 5740KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST 5741 This *adds* the contents of the .u.runstate members of the structure 5742 to the corresponding members of the given vCPU's runstate data, thus 5743 permitting atomic adjustments to the runstate times. The adjustment 5744 to the state_entry_time must equal the sum of the adjustments to the 5745 other four times. The state field must be set to -1, or to a valid 5746 runstate value (RUNSTATE_running, RUNSTATE_runnable, RUNSTATE_blocked 5747 or RUNSTATE_offline) to set the current accounted state as of the 5748 adjusted state_entry_time. 5749 5750KVM_XEN_VCPU_ATTR_TYPE_VCPU_ID 5751 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5752 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It sets the Xen 5753 vCPU ID of the given vCPU, to allow timer-related VCPU operations to 5754 be intercepted by KVM. 5755 5756KVM_XEN_VCPU_ATTR_TYPE_TIMER 5757 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5758 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It sets the 5759 event channel port/priority for the VIRQ_TIMER of the vCPU, as well 5760 as allowing a pending timer to be saved/restored. Setting the timer 5761 port to zero disables kernel handling of the singleshot timer. 5762 5763KVM_XEN_VCPU_ATTR_TYPE_UPCALL_VECTOR 5764 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5765 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It sets the 5766 per-vCPU local APIC upcall vector, configured by a Xen guest with 5767 the HVMOP_set_evtchn_upcall_vector hypercall. This is typically 5768 used by Windows guests, and is distinct from the HVM-wide upcall 5769 vector configured with HVM_PARAM_CALLBACK_IRQ. It is disabled by 5770 setting the vector to zero. 5771 5772 57734.129 KVM_XEN_VCPU_GET_ATTR 5774--------------------------- 5775 5776:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 5777:Architectures: x86 5778:Type: vcpu ioctl 5779:Parameters: struct kvm_xen_vcpu_attr 5780:Returns: 0 on success, < 0 on error 5781 5782Allows Xen vCPU attributes to be read. For the structure and types, 5783see KVM_XEN_VCPU_SET_ATTR above. 5784 5785The KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST type may not be used 5786with the KVM_XEN_VCPU_GET_ATTR ioctl. 5787 57884.130 KVM_ARM_MTE_COPY_TAGS 5789--------------------------- 5790 5791:Capability: KVM_CAP_ARM_MTE 5792:Architectures: arm64 5793:Type: vm ioctl 5794:Parameters: struct kvm_arm_copy_mte_tags 5795:Returns: number of bytes copied, < 0 on error (-EINVAL for incorrect 5796 arguments, -EFAULT if memory cannot be accessed). 5797 5798:: 5799 5800 struct kvm_arm_copy_mte_tags { 5801 __u64 guest_ipa; 5802 __u64 length; 5803 void __user *addr; 5804 __u64 flags; 5805 __u64 reserved[2]; 5806 }; 5807 5808Copies Memory Tagging Extension (MTE) tags to/from guest tag memory. The 5809``guest_ipa`` and ``length`` fields must be ``PAGE_SIZE`` aligned. 5810``length`` must not be bigger than 2^31 - PAGE_SIZE bytes. The ``addr`` 5811field must point to a buffer which the tags will be copied to or from. 5812 5813``flags`` specifies the direction of copy, either ``KVM_ARM_TAGS_TO_GUEST`` or 5814``KVM_ARM_TAGS_FROM_GUEST``. 5815 5816The size of the buffer to store the tags is ``(length / 16)`` bytes 5817(granules in MTE are 16 bytes long). Each byte contains a single tag 5818value. This matches the format of ``PTRACE_PEEKMTETAGS`` and 5819``PTRACE_POKEMTETAGS``. 5820 5821If an error occurs before any data is copied then a negative error code is 5822returned. If some tags have been copied before an error occurs then the number 5823of bytes successfully copied is returned. If the call completes successfully 5824then ``length`` is returned. 5825 58264.131 KVM_GET_SREGS2 5827-------------------- 5828 5829:Capability: KVM_CAP_SREGS2 5830:Architectures: x86 5831:Type: vcpu ioctl 5832:Parameters: struct kvm_sregs2 (out) 5833:Returns: 0 on success, -1 on error 5834 5835Reads special registers from the vcpu. 5836This ioctl (when supported) replaces the KVM_GET_SREGS. 5837 5838:: 5839 5840 struct kvm_sregs2 { 5841 /* out (KVM_GET_SREGS2) / in (KVM_SET_SREGS2) */ 5842 struct kvm_segment cs, ds, es, fs, gs, ss; 5843 struct kvm_segment tr, ldt; 5844 struct kvm_dtable gdt, idt; 5845 __u64 cr0, cr2, cr3, cr4, cr8; 5846 __u64 efer; 5847 __u64 apic_base; 5848 __u64 flags; 5849 __u64 pdptrs[4]; 5850 }; 5851 5852flags values for ``kvm_sregs2``: 5853 5854``KVM_SREGS2_FLAGS_PDPTRS_VALID`` 5855 5856 Indicates that the struct contains valid PDPTR values. 5857 5858 58594.132 KVM_SET_SREGS2 5860-------------------- 5861 5862:Capability: KVM_CAP_SREGS2 5863:Architectures: x86 5864:Type: vcpu ioctl 5865:Parameters: struct kvm_sregs2 (in) 5866:Returns: 0 on success, -1 on error 5867 5868Writes special registers into the vcpu. 5869See KVM_GET_SREGS2 for the data structures. 5870This ioctl (when supported) replaces the KVM_SET_SREGS. 5871 58724.133 KVM_GET_STATS_FD 5873---------------------- 5874 5875:Capability: KVM_CAP_STATS_BINARY_FD 5876:Architectures: all 5877:Type: vm ioctl, vcpu ioctl 5878:Parameters: none 5879:Returns: statistics file descriptor on success, < 0 on error 5880 5881Errors: 5882 5883 ====== ====================================================== 5884 ENOMEM if the fd could not be created due to lack of memory 5885 EMFILE if the number of opened files exceeds the limit 5886 ====== ====================================================== 5887 5888The returned file descriptor can be used to read VM/vCPU statistics data in 5889binary format. The data in the file descriptor consists of four blocks 5890organized as follows: 5891 5892+-------------+ 5893| Header | 5894+-------------+ 5895| id string | 5896+-------------+ 5897| Descriptors | 5898+-------------+ 5899| Stats Data | 5900+-------------+ 5901 5902Apart from the header starting at offset 0, please be aware that it is 5903not guaranteed that the four blocks are adjacent or in the above order; 5904the offsets of the id, descriptors and data blocks are found in the 5905header. However, all four blocks are aligned to 64 bit offsets in the 5906file and they do not overlap. 5907 5908All blocks except the data block are immutable. Userspace can read them 5909only one time after retrieving the file descriptor, and then use ``pread`` or 5910``lseek`` to read the statistics repeatedly. 5911 5912All data is in system endianness. 5913 5914The format of the header is as follows:: 5915 5916 struct kvm_stats_header { 5917 __u32 flags; 5918 __u32 name_size; 5919 __u32 num_desc; 5920 __u32 id_offset; 5921 __u32 desc_offset; 5922 __u32 data_offset; 5923 }; 5924 5925The ``flags`` field is not used at the moment. It is always read as 0. 5926 5927The ``name_size`` field is the size (in byte) of the statistics name string 5928(including trailing '\0') which is contained in the "id string" block and 5929appended at the end of every descriptor. 5930 5931The ``num_desc`` field is the number of descriptors that are included in the 5932descriptor block. (The actual number of values in the data block may be 5933larger, since each descriptor may comprise more than one value). 5934 5935The ``id_offset`` field is the offset of the id string from the start of the 5936file indicated by the file descriptor. It is a multiple of 8. 5937 5938The ``desc_offset`` field is the offset of the Descriptors block from the start 5939of the file indicated by the file descriptor. It is a multiple of 8. 5940 5941The ``data_offset`` field is the offset of the Stats Data block from the start 5942of the file indicated by the file descriptor. It is a multiple of 8. 5943 5944The id string block contains a string which identifies the file descriptor on 5945which KVM_GET_STATS_FD was invoked. The size of the block, including the 5946trailing ``'\0'``, is indicated by the ``name_size`` field in the header. 5947 5948The descriptors block is only needed to be read once for the lifetime of the 5949file descriptor contains a sequence of ``struct kvm_stats_desc``, each followed 5950by a string of size ``name_size``. 5951:: 5952 5953 #define KVM_STATS_TYPE_SHIFT 0 5954 #define KVM_STATS_TYPE_MASK (0xF << KVM_STATS_TYPE_SHIFT) 5955 #define KVM_STATS_TYPE_CUMULATIVE (0x0 << KVM_STATS_TYPE_SHIFT) 5956 #define KVM_STATS_TYPE_INSTANT (0x1 << KVM_STATS_TYPE_SHIFT) 5957 #define KVM_STATS_TYPE_PEAK (0x2 << KVM_STATS_TYPE_SHIFT) 5958 #define KVM_STATS_TYPE_LINEAR_HIST (0x3 << KVM_STATS_TYPE_SHIFT) 5959 #define KVM_STATS_TYPE_LOG_HIST (0x4 << KVM_STATS_TYPE_SHIFT) 5960 #define KVM_STATS_TYPE_MAX KVM_STATS_TYPE_LOG_HIST 5961 5962 #define KVM_STATS_UNIT_SHIFT 4 5963 #define KVM_STATS_UNIT_MASK (0xF << KVM_STATS_UNIT_SHIFT) 5964 #define KVM_STATS_UNIT_NONE (0x0 << KVM_STATS_UNIT_SHIFT) 5965 #define KVM_STATS_UNIT_BYTES (0x1 << KVM_STATS_UNIT_SHIFT) 5966 #define KVM_STATS_UNIT_SECONDS (0x2 << KVM_STATS_UNIT_SHIFT) 5967 #define KVM_STATS_UNIT_CYCLES (0x3 << KVM_STATS_UNIT_SHIFT) 5968 #define KVM_STATS_UNIT_BOOLEAN (0x4 << KVM_STATS_UNIT_SHIFT) 5969 #define KVM_STATS_UNIT_MAX KVM_STATS_UNIT_BOOLEAN 5970 5971 #define KVM_STATS_BASE_SHIFT 8 5972 #define KVM_STATS_BASE_MASK (0xF << KVM_STATS_BASE_SHIFT) 5973 #define KVM_STATS_BASE_POW10 (0x0 << KVM_STATS_BASE_SHIFT) 5974 #define KVM_STATS_BASE_POW2 (0x1 << KVM_STATS_BASE_SHIFT) 5975 #define KVM_STATS_BASE_MAX KVM_STATS_BASE_POW2 5976 5977 struct kvm_stats_desc { 5978 __u32 flags; 5979 __s16 exponent; 5980 __u16 size; 5981 __u32 offset; 5982 __u32 bucket_size; 5983 char name[]; 5984 }; 5985 5986The ``flags`` field contains the type and unit of the statistics data described 5987by this descriptor. Its endianness is CPU native. 5988The following flags are supported: 5989 5990Bits 0-3 of ``flags`` encode the type: 5991 5992 * ``KVM_STATS_TYPE_CUMULATIVE`` 5993 The statistics reports a cumulative count. The value of data can only be increased. 5994 Most of the counters used in KVM are of this type. 5995 The corresponding ``size`` field for this type is always 1. 5996 All cumulative statistics data are read/write. 5997 * ``KVM_STATS_TYPE_INSTANT`` 5998 The statistics reports an instantaneous value. Its value can be increased or 5999 decreased. This type is usually used as a measurement of some resources, 6000 like the number of dirty pages, the number of large pages, etc. 6001 All instant statistics are read only. 6002 The corresponding ``size`` field for this type is always 1. 6003 * ``KVM_STATS_TYPE_PEAK`` 6004 The statistics data reports a peak value, for example the maximum number 6005 of items in a hash table bucket, the longest time waited and so on. 6006 The value of data can only be increased. 6007 The corresponding ``size`` field for this type is always 1. 6008 * ``KVM_STATS_TYPE_LINEAR_HIST`` 6009 The statistic is reported as a linear histogram. The number of 6010 buckets is specified by the ``size`` field. The size of buckets is specified 6011 by the ``hist_param`` field. The range of the Nth bucket (1 <= N < ``size``) 6012 is [``hist_param``*(N-1), ``hist_param``*N), while the range of the last 6013 bucket is [``hist_param``*(``size``-1), +INF). (+INF means positive infinity 6014 value.) 6015 * ``KVM_STATS_TYPE_LOG_HIST`` 6016 The statistic is reported as a logarithmic histogram. The number of 6017 buckets is specified by the ``size`` field. The range of the first bucket is 6018 [0, 1), while the range of the last bucket is [pow(2, ``size``-2), +INF). 6019 Otherwise, The Nth bucket (1 < N < ``size``) covers 6020 [pow(2, N-2), pow(2, N-1)). 6021 6022Bits 4-7 of ``flags`` encode the unit: 6023 6024 * ``KVM_STATS_UNIT_NONE`` 6025 There is no unit for the value of statistics data. This usually means that 6026 the value is a simple counter of an event. 6027 * ``KVM_STATS_UNIT_BYTES`` 6028 It indicates that the statistics data is used to measure memory size, in the 6029 unit of Byte, KiByte, MiByte, GiByte, etc. The unit of the data is 6030 determined by the ``exponent`` field in the descriptor. 6031 * ``KVM_STATS_UNIT_SECONDS`` 6032 It indicates that the statistics data is used to measure time or latency. 6033 * ``KVM_STATS_UNIT_CYCLES`` 6034 It indicates that the statistics data is used to measure CPU clock cycles. 6035 * ``KVM_STATS_UNIT_BOOLEAN`` 6036 It indicates that the statistic will always be either 0 or 1. Boolean 6037 statistics of "peak" type will never go back from 1 to 0. Boolean 6038 statistics can be linear histograms (with two buckets) but not logarithmic 6039 histograms. 6040 6041Note that, in the case of histograms, the unit applies to the bucket 6042ranges, while the bucket value indicates how many samples fell in the 6043bucket's range. 6044 6045Bits 8-11 of ``flags``, together with ``exponent``, encode the scale of the 6046unit: 6047 6048 * ``KVM_STATS_BASE_POW10`` 6049 The scale is based on power of 10. It is used for measurement of time and 6050 CPU clock cycles. For example, an exponent of -9 can be used with 6051 ``KVM_STATS_UNIT_SECONDS`` to express that the unit is nanoseconds. 6052 * ``KVM_STATS_BASE_POW2`` 6053 The scale is based on power of 2. It is used for measurement of memory size. 6054 For example, an exponent of 20 can be used with ``KVM_STATS_UNIT_BYTES`` to 6055 express that the unit is MiB. 6056 6057The ``size`` field is the number of values of this statistics data. Its 6058value is usually 1 for most of simple statistics. 1 means it contains an 6059unsigned 64bit data. 6060 6061The ``offset`` field is the offset from the start of Data Block to the start of 6062the corresponding statistics data. 6063 6064The ``bucket_size`` field is used as a parameter for histogram statistics data. 6065It is only used by linear histogram statistics data, specifying the size of a 6066bucket in the unit expressed by bits 4-11 of ``flags`` together with ``exponent``. 6067 6068The ``name`` field is the name string of the statistics data. The name string 6069starts at the end of ``struct kvm_stats_desc``. The maximum length including 6070the trailing ``'\0'``, is indicated by ``name_size`` in the header. 6071 6072The Stats Data block contains an array of 64-bit values in the same order 6073as the descriptors in Descriptors block. 6074 60754.134 KVM_GET_XSAVE2 6076-------------------- 6077 6078:Capability: KVM_CAP_XSAVE2 6079:Architectures: x86 6080:Type: vcpu ioctl 6081:Parameters: struct kvm_xsave (out) 6082:Returns: 0 on success, -1 on error 6083 6084 6085:: 6086 6087 struct kvm_xsave { 6088 __u32 region[1024]; 6089 __u32 extra[0]; 6090 }; 6091 6092This ioctl would copy current vcpu's xsave struct to the userspace. It 6093copies as many bytes as are returned by KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) 6094when invoked on the vm file descriptor. The size value returned by 6095KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) will always be at least 4096. 6096Currently, it is only greater than 4096 if a dynamic feature has been 6097enabled with ``arch_prctl()``, but this may change in the future. 6098 6099The offsets of the state save areas in struct kvm_xsave follow the contents 6100of CPUID leaf 0xD on the host. 6101 61024.135 KVM_XEN_HVM_EVTCHN_SEND 6103----------------------------- 6104 6105:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_EVTCHN_SEND 6106:Architectures: x86 6107:Type: vm ioctl 6108:Parameters: struct kvm_irq_routing_xen_evtchn 6109:Returns: 0 on success, < 0 on error 6110 6111 6112:: 6113 6114 struct kvm_irq_routing_xen_evtchn { 6115 __u32 port; 6116 __u32 vcpu; 6117 __u32 priority; 6118 }; 6119 6120This ioctl injects an event channel interrupt directly to the guest vCPU. 6121 61224.136 KVM_S390_PV_CPU_COMMAND 6123----------------------------- 6124 6125:Capability: KVM_CAP_S390_PROTECTED_DUMP 6126:Architectures: s390 6127:Type: vcpu ioctl 6128:Parameters: none 6129:Returns: 0 on success, < 0 on error 6130 6131This ioctl closely mirrors `KVM_S390_PV_COMMAND` but handles requests 6132for vcpus. It re-uses the kvm_s390_pv_dmp struct and hence also shares 6133the command ids. 6134 6135**command:** 6136 6137KVM_PV_DUMP 6138 Presents an API that provides calls which facilitate dumping a vcpu 6139 of a protected VM. 6140 6141**subcommand:** 6142 6143KVM_PV_DUMP_CPU 6144 Provides encrypted dump data like register values. 6145 The length of the returned data is provided by uv_info.guest_cpu_stor_len. 6146 61474.137 KVM_S390_ZPCI_OP 6148---------------------- 6149 6150:Capability: KVM_CAP_S390_ZPCI_OP 6151:Architectures: s390 6152:Type: vm ioctl 6153:Parameters: struct kvm_s390_zpci_op (in) 6154:Returns: 0 on success, <0 on error 6155 6156Used to manage hardware-assisted virtualization features for zPCI devices. 6157 6158Parameters are specified via the following structure:: 6159 6160 struct kvm_s390_zpci_op { 6161 /* in */ 6162 __u32 fh; /* target device */ 6163 __u8 op; /* operation to perform */ 6164 __u8 pad[3]; 6165 union { 6166 /* for KVM_S390_ZPCIOP_REG_AEN */ 6167 struct { 6168 __u64 ibv; /* Guest addr of interrupt bit vector */ 6169 __u64 sb; /* Guest addr of summary bit */ 6170 __u32 flags; 6171 __u32 noi; /* Number of interrupts */ 6172 __u8 isc; /* Guest interrupt subclass */ 6173 __u8 sbo; /* Offset of guest summary bit vector */ 6174 __u16 pad; 6175 } reg_aen; 6176 __u64 reserved[8]; 6177 } u; 6178 }; 6179 6180The type of operation is specified in the "op" field. 6181KVM_S390_ZPCIOP_REG_AEN is used to register the VM for adapter event 6182notification interpretation, which will allow firmware delivery of adapter 6183events directly to the vm, with KVM providing a backup delivery mechanism; 6184KVM_S390_ZPCIOP_DEREG_AEN is used to subsequently disable interpretation of 6185adapter event notifications. 6186 6187The target zPCI function must also be specified via the "fh" field. For the 6188KVM_S390_ZPCIOP_REG_AEN operation, additional information to establish firmware 6189delivery must be provided via the "reg_aen" struct. 6190 6191The "pad" and "reserved" fields may be used for future extensions and should be 6192set to 0s by userspace. 6193 61944.138 KVM_ARM_SET_COUNTER_OFFSET 6195-------------------------------- 6196 6197:Capability: KVM_CAP_COUNTER_OFFSET 6198:Architectures: arm64 6199:Type: vm ioctl 6200:Parameters: struct kvm_arm_counter_offset (in) 6201:Returns: 0 on success, < 0 on error 6202 6203This capability indicates that userspace is able to apply a single VM-wide 6204offset to both the virtual and physical counters as viewed by the guest 6205using the KVM_ARM_SET_CNT_OFFSET ioctl and the following data structure: 6206 6207:: 6208 6209 struct kvm_arm_counter_offset { 6210 __u64 counter_offset; 6211 __u64 reserved; 6212 }; 6213 6214The offset describes a number of counter cycles that are subtracted from 6215both virtual and physical counter views (similar to the effects of the 6216CNTVOFF_EL2 and CNTPOFF_EL2 system registers, but only global). The offset 6217always applies to all vcpus (already created or created after this ioctl) 6218for this VM. 6219 6220It is userspace's responsibility to compute the offset based, for example, 6221on previous values of the guest counters. 6222 6223Any value other than 0 for the "reserved" field may result in an error 6224(-EINVAL) being returned. This ioctl can also return -EBUSY if any vcpu 6225ioctl is issued concurrently. 6226 6227Note that using this ioctl results in KVM ignoring subsequent userspace 6228writes to the CNTVCT_EL0 and CNTPCT_EL0 registers using the SET_ONE_REG 6229interface. No error will be returned, but the resulting offset will not be 6230applied. 6231 6232.. _KVM_ARM_GET_REG_WRITABLE_MASKS: 6233 62344.139 KVM_ARM_GET_REG_WRITABLE_MASKS 6235------------------------------------ 6236 6237:Capability: KVM_CAP_ARM_SUPPORTED_REG_MASK_RANGES 6238:Architectures: arm64 6239:Type: vm ioctl 6240:Parameters: struct reg_mask_range (in/out) 6241:Returns: 0 on success, < 0 on error 6242 6243 6244:: 6245 6246 #define KVM_ARM_FEATURE_ID_RANGE 0 6247 #define KVM_ARM_FEATURE_ID_RANGE_SIZE (3 * 8 * 8) 6248 6249 struct reg_mask_range { 6250 __u64 addr; /* Pointer to mask array */ 6251 __u32 range; /* Requested range */ 6252 __u32 reserved[13]; 6253 }; 6254 6255This ioctl copies the writable masks for a selected range of registers to 6256userspace. 6257 6258The ``addr`` field is a pointer to the destination array where KVM copies 6259the writable masks. 6260 6261The ``range`` field indicates the requested range of registers. 6262``KVM_CHECK_EXTENSION`` for the ``KVM_CAP_ARM_SUPPORTED_REG_MASK_RANGES`` 6263capability returns the supported ranges, expressed as a set of flags. Each 6264flag's bit index represents a possible value for the ``range`` field. 6265All other values are reserved for future use and KVM may return an error. 6266 6267The ``reserved[13]`` array is reserved for future use and should be 0, or 6268KVM may return an error. 6269 6270KVM_ARM_FEATURE_ID_RANGE (0) 6271^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6272 6273The Feature ID range is defined as the AArch64 System register space with 6274op0==3, op1=={0, 1, 3}, CRn==0, CRm=={0-7}, op2=={0-7}. 6275 6276The mask returned array pointed to by ``addr`` is indexed by the macro 6277``ARM64_FEATURE_ID_RANGE_IDX(op0, op1, crn, crm, op2)``, allowing userspace 6278to know what fields can be changed for the system register described by 6279``op0, op1, crn, crm, op2``. KVM rejects ID register values that describe a 6280superset of the features supported by the system. 6281 62824.140 KVM_SET_USER_MEMORY_REGION2 6283--------------------------------- 6284 6285:Capability: KVM_CAP_USER_MEMORY2 6286:Architectures: all 6287:Type: vm ioctl 6288:Parameters: struct kvm_userspace_memory_region2 (in) 6289:Returns: 0 on success, -1 on error 6290 6291KVM_SET_USER_MEMORY_REGION2 is an extension to KVM_SET_USER_MEMORY_REGION that 6292allows mapping guest_memfd memory into a guest. All fields shared with 6293KVM_SET_USER_MEMORY_REGION identically. Userspace can set KVM_MEM_GUEST_MEMFD 6294in flags to have KVM bind the memory region to a given guest_memfd range of 6295[guest_memfd_offset, guest_memfd_offset + memory_size]. The target guest_memfd 6296must point at a file created via KVM_CREATE_GUEST_MEMFD on the current VM, and 6297the target range must not be bound to any other memory region. All standard 6298bounds checks apply (use common sense). 6299 6300:: 6301 6302 struct kvm_userspace_memory_region2 { 6303 __u32 slot; 6304 __u32 flags; 6305 __u64 guest_phys_addr; 6306 __u64 memory_size; /* bytes */ 6307 __u64 userspace_addr; /* start of the userspace allocated memory */ 6308 __u64 guest_memfd_offset; 6309 __u32 guest_memfd; 6310 __u32 pad1; 6311 __u64 pad2[14]; 6312 }; 6313 6314A KVM_MEM_GUEST_MEMFD region _must_ have a valid guest_memfd (private memory) and 6315userspace_addr (shared memory). However, "valid" for userspace_addr simply 6316means that the address itself must be a legal userspace address. The backing 6317mapping for userspace_addr is not required to be valid/populated at the time of 6318KVM_SET_USER_MEMORY_REGION2, e.g. shared memory can be lazily mapped/allocated 6319on-demand. 6320 6321When mapping a gfn into the guest, KVM selects shared vs. private, i.e consumes 6322userspace_addr vs. guest_memfd, based on the gfn's KVM_MEMORY_ATTRIBUTE_PRIVATE 6323state. At VM creation time, all memory is shared, i.e. the PRIVATE attribute 6324is '0' for all gfns. Userspace can control whether memory is shared/private by 6325toggling KVM_MEMORY_ATTRIBUTE_PRIVATE via KVM_SET_MEMORY_ATTRIBUTES as needed. 6326 6327S390: 6328^^^^^ 6329 6330Returns -EINVAL if the VM has the KVM_VM_S390_UCONTROL flag set. 6331Returns -EINVAL if called on a protected VM. 6332 63334.141 KVM_SET_MEMORY_ATTRIBUTES 6334------------------------------- 6335 6336:Capability: KVM_CAP_MEMORY_ATTRIBUTES 6337:Architectures: x86 6338:Type: vm ioctl 6339:Parameters: struct kvm_memory_attributes (in) 6340:Returns: 0 on success, <0 on error 6341 6342KVM_SET_MEMORY_ATTRIBUTES allows userspace to set memory attributes for a range 6343of guest physical memory. 6344 6345:: 6346 6347 struct kvm_memory_attributes { 6348 __u64 address; 6349 __u64 size; 6350 __u64 attributes; 6351 __u64 flags; 6352 }; 6353 6354 #define KVM_MEMORY_ATTRIBUTE_PRIVATE (1ULL << 3) 6355 6356The address and size must be page aligned. The supported attributes can be 6357retrieved via ioctl(KVM_CHECK_EXTENSION) on KVM_CAP_MEMORY_ATTRIBUTES. If 6358executed on a VM, KVM_CAP_MEMORY_ATTRIBUTES precisely returns the attributes 6359supported by that VM. If executed at system scope, KVM_CAP_MEMORY_ATTRIBUTES 6360returns all attributes supported by KVM. The only attribute defined at this 6361time is KVM_MEMORY_ATTRIBUTE_PRIVATE, which marks the associated gfn as being 6362guest private memory. 6363 6364Note, there is no "get" API. Userspace is responsible for explicitly tracking 6365the state of a gfn/page as needed. 6366 6367The "flags" field is reserved for future extensions and must be '0'. 6368 63694.142 KVM_CREATE_GUEST_MEMFD 6370---------------------------- 6371 6372:Capability: KVM_CAP_GUEST_MEMFD 6373:Architectures: none 6374:Type: vm ioctl 6375:Parameters: struct kvm_create_guest_memfd(in) 6376:Returns: A file descriptor on success, <0 on error 6377 6378KVM_CREATE_GUEST_MEMFD creates an anonymous file and returns a file descriptor 6379that refers to it. guest_memfd files are roughly analogous to files created 6380via memfd_create(), e.g. guest_memfd files live in RAM, have volatile storage, 6381and are automatically released when the last reference is dropped. Unlike 6382"regular" memfd_create() files, guest_memfd files are bound to their owning 6383virtual machine (see below), cannot be mapped, read, or written by userspace, 6384and cannot be resized (guest_memfd files do however support PUNCH_HOLE). 6385 6386:: 6387 6388 struct kvm_create_guest_memfd { 6389 __u64 size; 6390 __u64 flags; 6391 __u64 reserved[6]; 6392 }; 6393 6394Conceptually, the inode backing a guest_memfd file represents physical memory, 6395i.e. is coupled to the virtual machine as a thing, not to a "struct kvm". The 6396file itself, which is bound to a "struct kvm", is that instance's view of the 6397underlying memory, e.g. effectively provides the translation of guest addresses 6398to host memory. This allows for use cases where multiple KVM structures are 6399used to manage a single virtual machine, e.g. when performing intrahost 6400migration of a virtual machine. 6401 6402KVM currently only supports mapping guest_memfd via KVM_SET_USER_MEMORY_REGION2, 6403and more specifically via the guest_memfd and guest_memfd_offset fields in 6404"struct kvm_userspace_memory_region2", where guest_memfd_offset is the offset 6405into the guest_memfd instance. For a given guest_memfd file, there can be at 6406most one mapping per page, i.e. binding multiple memory regions to a single 6407guest_memfd range is not allowed (any number of memory regions can be bound to 6408a single guest_memfd file, but the bound ranges must not overlap). 6409 6410See KVM_SET_USER_MEMORY_REGION2 for additional details. 6411 64124.143 KVM_PRE_FAULT_MEMORY 6413--------------------------- 6414 6415:Capability: KVM_CAP_PRE_FAULT_MEMORY 6416:Architectures: none 6417:Type: vcpu ioctl 6418:Parameters: struct kvm_pre_fault_memory (in/out) 6419:Returns: 0 if at least one page is processed, < 0 on error 6420 6421Errors: 6422 6423 ========== =============================================================== 6424 EINVAL The specified `gpa` and `size` were invalid (e.g. not 6425 page aligned, causes an overflow, or size is zero). 6426 ENOENT The specified `gpa` is outside defined memslots. 6427 EINTR An unmasked signal is pending and no page was processed. 6428 EFAULT The parameter address was invalid. 6429 EOPNOTSUPP Mapping memory for a GPA is unsupported by the 6430 hypervisor, and/or for the current vCPU state/mode. 6431 EIO unexpected error conditions (also causes a WARN) 6432 ========== =============================================================== 6433 6434:: 6435 6436 struct kvm_pre_fault_memory { 6437 /* in/out */ 6438 __u64 gpa; 6439 __u64 size; 6440 /* in */ 6441 __u64 flags; 6442 __u64 padding[5]; 6443 }; 6444 6445KVM_PRE_FAULT_MEMORY populates KVM's stage-2 page tables used to map memory 6446for the current vCPU state. KVM maps memory as if the vCPU generated a 6447stage-2 read page fault, e.g. faults in memory as needed, but doesn't break 6448CoW. However, KVM does not mark any newly created stage-2 PTE as Accessed. 6449 6450In the case of confidential VM types where there is an initial set up of 6451private guest memory before the guest is 'finalized'/measured, this ioctl 6452should only be issued after completing all the necessary setup to put the 6453guest into a 'finalized' state so that the above semantics can be reliably 6454ensured. 6455 6456In some cases, multiple vCPUs might share the page tables. In this 6457case, the ioctl can be called in parallel. 6458 6459When the ioctl returns, the input values are updated to point to the 6460remaining range. If `size` > 0 on return, the caller can just issue 6461the ioctl again with the same `struct kvm_map_memory` argument. 6462 6463Shadow page tables cannot support this ioctl because they 6464are indexed by virtual address or nested guest physical address. 6465Calling this ioctl when the guest is using shadow page tables (for 6466example because it is running a nested guest with nested page tables) 6467will fail with `EOPNOTSUPP` even if `KVM_CHECK_EXTENSION` reports 6468the capability to be present. 6469 6470`flags` must currently be zero. 6471 6472 6473.. _kvm_run: 6474 64755. The kvm_run structure 6476======================== 6477 6478Application code obtains a pointer to the kvm_run structure by 6479mmap()ing a vcpu fd. From that point, application code can control 6480execution by changing fields in kvm_run prior to calling the KVM_RUN 6481ioctl, and obtain information about the reason KVM_RUN returned by 6482looking up structure members. 6483 6484:: 6485 6486 struct kvm_run { 6487 /* in */ 6488 __u8 request_interrupt_window; 6489 6490Request that KVM_RUN return when it becomes possible to inject external 6491interrupts into the guest. Useful in conjunction with KVM_INTERRUPT. 6492 6493:: 6494 6495 __u8 immediate_exit; 6496 6497This field is polled once when KVM_RUN starts; if non-zero, KVM_RUN 6498exits immediately, returning -EINTR. In the common scenario where a 6499signal is used to "kick" a VCPU out of KVM_RUN, this field can be used 6500to avoid usage of KVM_SET_SIGNAL_MASK, which has worse scalability. 6501Rather than blocking the signal outside KVM_RUN, userspace can set up 6502a signal handler that sets run->immediate_exit to a non-zero value. 6503 6504This field is ignored if KVM_CAP_IMMEDIATE_EXIT is not available. 6505 6506:: 6507 6508 __u8 padding1[6]; 6509 6510 /* out */ 6511 __u32 exit_reason; 6512 6513When KVM_RUN has returned successfully (return value 0), this informs 6514application code why KVM_RUN has returned. Allowable values for this 6515field are detailed below. 6516 6517:: 6518 6519 __u8 ready_for_interrupt_injection; 6520 6521If request_interrupt_window has been specified, this field indicates 6522an interrupt can be injected now with KVM_INTERRUPT. 6523 6524:: 6525 6526 __u8 if_flag; 6527 6528The value of the current interrupt flag. Only valid if in-kernel 6529local APIC is not used. 6530 6531:: 6532 6533 __u16 flags; 6534 6535More architecture-specific flags detailing state of the VCPU that may 6536affect the device's behavior. Current defined flags:: 6537 6538 /* x86, set if the VCPU is in system management mode */ 6539 #define KVM_RUN_X86_SMM (1 << 0) 6540 /* x86, set if bus lock detected in VM */ 6541 #define KVM_RUN_X86_BUS_LOCK (1 << 1) 6542 /* x86, set if the VCPU is executing a nested (L2) guest */ 6543 #define KVM_RUN_X86_GUEST_MODE (1 << 2) 6544 6545 /* arm64, set for KVM_EXIT_DEBUG */ 6546 #define KVM_DEBUG_ARCH_HSR_HIGH_VALID (1 << 0) 6547 6548:: 6549 6550 /* in (pre_kvm_run), out (post_kvm_run) */ 6551 __u64 cr8; 6552 6553The value of the cr8 register. Only valid if in-kernel local APIC is 6554not used. Both input and output. 6555 6556:: 6557 6558 __u64 apic_base; 6559 6560The value of the APIC BASE msr. Only valid if in-kernel local 6561APIC is not used. Both input and output. 6562 6563:: 6564 6565 union { 6566 /* KVM_EXIT_UNKNOWN */ 6567 struct { 6568 __u64 hardware_exit_reason; 6569 } hw; 6570 6571If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown 6572reasons. Further architecture-specific information is available in 6573hardware_exit_reason. 6574 6575:: 6576 6577 /* KVM_EXIT_FAIL_ENTRY */ 6578 struct { 6579 __u64 hardware_entry_failure_reason; 6580 __u32 cpu; /* if KVM_LAST_CPU */ 6581 } fail_entry; 6582 6583If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due 6584to unknown reasons. Further architecture-specific information is 6585available in hardware_entry_failure_reason. 6586 6587:: 6588 6589 /* KVM_EXIT_EXCEPTION */ 6590 struct { 6591 __u32 exception; 6592 __u32 error_code; 6593 } ex; 6594 6595Unused. 6596 6597:: 6598 6599 /* KVM_EXIT_IO */ 6600 struct { 6601 #define KVM_EXIT_IO_IN 0 6602 #define KVM_EXIT_IO_OUT 1 6603 __u8 direction; 6604 __u8 size; /* bytes */ 6605 __u16 port; 6606 __u32 count; 6607 __u64 data_offset; /* relative to kvm_run start */ 6608 } io; 6609 6610If exit_reason is KVM_EXIT_IO, then the vcpu has 6611executed a port I/O instruction which could not be satisfied by kvm. 6612data_offset describes where the data is located (KVM_EXIT_IO_OUT) or 6613where kvm expects application code to place the data for the next 6614KVM_RUN invocation (KVM_EXIT_IO_IN). Data format is a packed array. 6615 6616:: 6617 6618 /* KVM_EXIT_DEBUG */ 6619 struct { 6620 struct kvm_debug_exit_arch arch; 6621 } debug; 6622 6623If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event 6624for which architecture specific information is returned. 6625 6626:: 6627 6628 /* KVM_EXIT_MMIO */ 6629 struct { 6630 __u64 phys_addr; 6631 __u8 data[8]; 6632 __u32 len; 6633 __u8 is_write; 6634 } mmio; 6635 6636If exit_reason is KVM_EXIT_MMIO, then the vcpu has 6637executed a memory-mapped I/O instruction which could not be satisfied 6638by kvm. The 'data' member contains the written data if 'is_write' is 6639true, and should be filled by application code otherwise. 6640 6641The 'data' member contains, in its first 'len' bytes, the value as it would 6642appear if the VCPU performed a load or store of the appropriate width directly 6643to the byte array. 6644 6645.. note:: 6646 6647 For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR, KVM_EXIT_XEN, 6648 KVM_EXIT_EPR, KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR the corresponding 6649 operations are complete (and guest state is consistent) only after userspace 6650 has re-entered the kernel with KVM_RUN. The kernel side will first finish 6651 incomplete operations and then check for pending signals. 6652 6653 The pending state of the operation is not preserved in state which is 6654 visible to userspace, thus userspace should ensure that the operation is 6655 completed before performing a live migration. Userspace can re-enter the 6656 guest with an unmasked signal pending or with the immediate_exit field set 6657 to complete pending operations without allowing any further instructions 6658 to be executed. 6659 6660:: 6661 6662 /* KVM_EXIT_HYPERCALL */ 6663 struct { 6664 __u64 nr; 6665 __u64 args[6]; 6666 __u64 ret; 6667 __u64 flags; 6668 } hypercall; 6669 6670 6671It is strongly recommended that userspace use ``KVM_EXIT_IO`` (x86) or 6672``KVM_EXIT_MMIO`` (all except s390) to implement functionality that 6673requires a guest to interact with host userspace. 6674 6675.. note:: KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO. 6676 6677For arm64: 6678---------- 6679 6680SMCCC exits can be enabled depending on the configuration of the SMCCC 6681filter. See the Documentation/virt/kvm/devices/vm.rst 6682``KVM_ARM_SMCCC_FILTER`` for more details. 6683 6684``nr`` contains the function ID of the guest's SMCCC call. Userspace is 6685expected to use the ``KVM_GET_ONE_REG`` ioctl to retrieve the call 6686parameters from the vCPU's GPRs. 6687 6688Definition of ``flags``: 6689 - ``KVM_HYPERCALL_EXIT_SMC``: Indicates that the guest used the SMC 6690 conduit to initiate the SMCCC call. If this bit is 0 then the guest 6691 used the HVC conduit for the SMCCC call. 6692 6693 - ``KVM_HYPERCALL_EXIT_16BIT``: Indicates that the guest used a 16bit 6694 instruction to initiate the SMCCC call. If this bit is 0 then the 6695 guest used a 32bit instruction. An AArch64 guest always has this 6696 bit set to 0. 6697 6698At the point of exit, PC points to the instruction immediately following 6699the trapping instruction. 6700 6701:: 6702 6703 /* KVM_EXIT_TPR_ACCESS */ 6704 struct { 6705 __u64 rip; 6706 __u32 is_write; 6707 __u32 pad; 6708 } tpr_access; 6709 6710To be documented (KVM_TPR_ACCESS_REPORTING). 6711 6712:: 6713 6714 /* KVM_EXIT_S390_SIEIC */ 6715 struct { 6716 __u8 icptcode; 6717 __u64 mask; /* psw upper half */ 6718 __u64 addr; /* psw lower half */ 6719 __u16 ipa; 6720 __u32 ipb; 6721 } s390_sieic; 6722 6723s390 specific. 6724 6725:: 6726 6727 /* KVM_EXIT_S390_RESET */ 6728 #define KVM_S390_RESET_POR 1 6729 #define KVM_S390_RESET_CLEAR 2 6730 #define KVM_S390_RESET_SUBSYSTEM 4 6731 #define KVM_S390_RESET_CPU_INIT 8 6732 #define KVM_S390_RESET_IPL 16 6733 __u64 s390_reset_flags; 6734 6735s390 specific. 6736 6737:: 6738 6739 /* KVM_EXIT_S390_UCONTROL */ 6740 struct { 6741 __u64 trans_exc_code; 6742 __u32 pgm_code; 6743 } s390_ucontrol; 6744 6745s390 specific. A page fault has occurred for a user controlled virtual 6746machine (KVM_VM_S390_UNCONTROL) on its host page table that cannot be 6747resolved by the kernel. 6748The program code and the translation exception code that were placed 6749in the cpu's lowcore are presented here as defined by the z Architecture 6750Principles of Operation Book in the Chapter for Dynamic Address Translation 6751(DAT) 6752 6753:: 6754 6755 /* KVM_EXIT_DCR */ 6756 struct { 6757 __u32 dcrn; 6758 __u32 data; 6759 __u8 is_write; 6760 } dcr; 6761 6762Deprecated - was used for 440 KVM. 6763 6764:: 6765 6766 /* KVM_EXIT_OSI */ 6767 struct { 6768 __u64 gprs[32]; 6769 } osi; 6770 6771MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch 6772hypercalls and exit with this exit struct that contains all the guest gprs. 6773 6774If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall. 6775Userspace can now handle the hypercall and when it's done modify the gprs as 6776necessary. Upon guest entry all guest GPRs will then be replaced by the values 6777in this struct. 6778 6779:: 6780 6781 /* KVM_EXIT_PAPR_HCALL */ 6782 struct { 6783 __u64 nr; 6784 __u64 ret; 6785 __u64 args[9]; 6786 } papr_hcall; 6787 6788This is used on 64-bit PowerPC when emulating a pSeries partition, 6789e.g. with the 'pseries' machine type in qemu. It occurs when the 6790guest does a hypercall using the 'sc 1' instruction. The 'nr' field 6791contains the hypercall number (from the guest R3), and 'args' contains 6792the arguments (from the guest R4 - R12). Userspace should put the 6793return code in 'ret' and any extra returned values in args[]. 6794The possible hypercalls are defined in the Power Architecture Platform 6795Requirements (PAPR) document available from www.power.org (free 6796developer registration required to access it). 6797 6798:: 6799 6800 /* KVM_EXIT_S390_TSCH */ 6801 struct { 6802 __u16 subchannel_id; 6803 __u16 subchannel_nr; 6804 __u32 io_int_parm; 6805 __u32 io_int_word; 6806 __u32 ipb; 6807 __u8 dequeued; 6808 } s390_tsch; 6809 6810s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled 6811and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O 6812interrupt for the target subchannel has been dequeued and subchannel_id, 6813subchannel_nr, io_int_parm and io_int_word contain the parameters for that 6814interrupt. ipb is needed for instruction parameter decoding. 6815 6816:: 6817 6818 /* KVM_EXIT_EPR */ 6819 struct { 6820 __u32 epr; 6821 } epr; 6822 6823On FSL BookE PowerPC chips, the interrupt controller has a fast patch 6824interrupt acknowledge path to the core. When the core successfully 6825delivers an interrupt, it automatically populates the EPR register with 6826the interrupt vector number and acknowledges the interrupt inside 6827the interrupt controller. 6828 6829In case the interrupt controller lives in user space, we need to do 6830the interrupt acknowledge cycle through it to fetch the next to be 6831delivered interrupt vector using this exit. 6832 6833It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an 6834external interrupt has just been delivered into the guest. User space 6835should put the acknowledged interrupt vector into the 'epr' field. 6836 6837:: 6838 6839 /* KVM_EXIT_SYSTEM_EVENT */ 6840 struct { 6841 #define KVM_SYSTEM_EVENT_SHUTDOWN 1 6842 #define KVM_SYSTEM_EVENT_RESET 2 6843 #define KVM_SYSTEM_EVENT_CRASH 3 6844 #define KVM_SYSTEM_EVENT_WAKEUP 4 6845 #define KVM_SYSTEM_EVENT_SUSPEND 5 6846 #define KVM_SYSTEM_EVENT_SEV_TERM 6 6847 #define KVM_SYSTEM_EVENT_TDX_FATAL 7 6848 __u32 type; 6849 __u32 ndata; 6850 __u64 data[16]; 6851 } system_event; 6852 6853If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered 6854a system-level event using some architecture specific mechanism (hypercall 6855or some special instruction). In case of ARM64, this is triggered using 6856HVC instruction based PSCI call from the vcpu. 6857 6858The 'type' field describes the system-level event type. 6859Valid values for 'type' are: 6860 6861 - KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the 6862 VM. Userspace is not obliged to honour this, and if it does honour 6863 this does not need to destroy the VM synchronously (ie it may call 6864 KVM_RUN again before shutdown finally occurs). 6865 - KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM. 6866 As with SHUTDOWN, userspace can choose to ignore the request, or 6867 to schedule the reset to occur in the future and may call KVM_RUN again. 6868 - KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest 6869 has requested a crash condition maintenance. Userspace can choose 6870 to ignore the request, or to gather VM memory core dump and/or 6871 reset/shutdown of the VM. 6872 - KVM_SYSTEM_EVENT_SEV_TERM -- an AMD SEV guest requested termination. 6873 The guest physical address of the guest's GHCB is stored in `data[0]`. 6874 - KVM_SYSTEM_EVENT_TDX_FATAL -- a TDX guest reported a fatal error state. 6875 KVM doesn't do any parsing or conversion, it just dumps 16 general-purpose 6876 registers to userspace, in ascending order of the 4-bit indices for x86-64 6877 general-purpose registers in instruction encoding, as defined in the Intel 6878 SDM. 6879 - KVM_SYSTEM_EVENT_WAKEUP -- the exiting vCPU is in a suspended state and 6880 KVM has recognized a wakeup event. Userspace may honor this event by 6881 marking the exiting vCPU as runnable, or deny it and call KVM_RUN again. 6882 - KVM_SYSTEM_EVENT_SUSPEND -- the guest has requested a suspension of 6883 the VM. 6884 6885If KVM_CAP_SYSTEM_EVENT_DATA is present, the 'data' field can contain 6886architecture specific information for the system-level event. Only 6887the first `ndata` items (possibly zero) of the data array are valid. 6888 6889 - for arm64, data[0] is set to KVM_SYSTEM_EVENT_RESET_FLAG_PSCI_RESET2 if 6890 the guest issued a SYSTEM_RESET2 call according to v1.1 of the PSCI 6891 specification. 6892 6893 - for arm64, data[0] is set to KVM_SYSTEM_EVENT_SHUTDOWN_FLAG_PSCI_OFF2 6894 if the guest issued a SYSTEM_OFF2 call according to v1.3 of the PSCI 6895 specification. 6896 6897 - for RISC-V, data[0] is set to the value of the second argument of the 6898 ``sbi_system_reset`` call. 6899 6900Previous versions of Linux defined a `flags` member in this struct. The 6901field is now aliased to `data[0]`. Userspace can assume that it is only 6902written if ndata is greater than 0. 6903 6904For arm/arm64: 6905-------------- 6906 6907KVM_SYSTEM_EVENT_SUSPEND exits are enabled with the 6908KVM_CAP_ARM_SYSTEM_SUSPEND VM capability. If a guest invokes the PSCI 6909SYSTEM_SUSPEND function, KVM will exit to userspace with this event 6910type. 6911 6912It is the sole responsibility of userspace to implement the PSCI 6913SYSTEM_SUSPEND call according to ARM DEN0022D.b 5.19 "SYSTEM_SUSPEND". 6914KVM does not change the vCPU's state before exiting to userspace, so 6915the call parameters are left in-place in the vCPU registers. 6916 6917Userspace is _required_ to take action for such an exit. It must 6918either: 6919 6920 - Honor the guest request to suspend the VM. Userspace can request 6921 in-kernel emulation of suspension by setting the calling vCPU's 6922 state to KVM_MP_STATE_SUSPENDED. Userspace must configure the vCPU's 6923 state according to the parameters passed to the PSCI function when 6924 the calling vCPU is resumed. See ARM DEN0022D.b 5.19.1 "Intended use" 6925 for details on the function parameters. 6926 6927 - Deny the guest request to suspend the VM. See ARM DEN0022D.b 5.19.2 6928 "Caller responsibilities" for possible return values. 6929 6930Hibernation using the PSCI SYSTEM_OFF2 call is enabled when PSCI v1.3 6931is enabled. If a guest invokes the PSCI SYSTEM_OFF2 function, KVM will 6932exit to userspace with the KVM_SYSTEM_EVENT_SHUTDOWN event type and with 6933data[0] set to KVM_SYSTEM_EVENT_SHUTDOWN_FLAG_PSCI_OFF2. The only 6934supported hibernate type for the SYSTEM_OFF2 function is HIBERNATE_OFF. 6935 6936:: 6937 6938 /* KVM_EXIT_IOAPIC_EOI */ 6939 struct { 6940 __u8 vector; 6941 } eoi; 6942 6943Indicates that the VCPU's in-kernel local APIC received an EOI for a 6944level-triggered IOAPIC interrupt. This exit only triggers when the 6945IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled); 6946the userspace IOAPIC should process the EOI and retrigger the interrupt if 6947it is still asserted. Vector is the LAPIC interrupt vector for which the 6948EOI was received. 6949 6950:: 6951 6952 struct kvm_hyperv_exit { 6953 #define KVM_EXIT_HYPERV_SYNIC 1 6954 #define KVM_EXIT_HYPERV_HCALL 2 6955 #define KVM_EXIT_HYPERV_SYNDBG 3 6956 __u32 type; 6957 __u32 pad1; 6958 union { 6959 struct { 6960 __u32 msr; 6961 __u32 pad2; 6962 __u64 control; 6963 __u64 evt_page; 6964 __u64 msg_page; 6965 } synic; 6966 struct { 6967 __u64 input; 6968 __u64 result; 6969 __u64 params[2]; 6970 } hcall; 6971 struct { 6972 __u32 msr; 6973 __u32 pad2; 6974 __u64 control; 6975 __u64 status; 6976 __u64 send_page; 6977 __u64 recv_page; 6978 __u64 pending_page; 6979 } syndbg; 6980 } u; 6981 }; 6982 /* KVM_EXIT_HYPERV */ 6983 struct kvm_hyperv_exit hyperv; 6984 6985Indicates that the VCPU exits into userspace to process some tasks 6986related to Hyper-V emulation. 6987 6988Valid values for 'type' are: 6989 6990 - KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about 6991 6992Hyper-V SynIC state change. Notification is used to remap SynIC 6993event/message pages and to enable/disable SynIC messages/events processing 6994in userspace. 6995 6996 - KVM_EXIT_HYPERV_SYNDBG -- synchronously notify user-space about 6997 6998Hyper-V Synthetic debugger state change. Notification is used to either update 6999the pending_page location or to send a control command (send the buffer located 7000in send_page or recv a buffer to recv_page). 7001 7002:: 7003 7004 /* KVM_EXIT_ARM_NISV */ 7005 struct { 7006 __u64 esr_iss; 7007 __u64 fault_ipa; 7008 } arm_nisv; 7009 7010Used on arm64 systems. If a guest accesses memory not in a memslot, 7011KVM will typically return to userspace and ask it to do MMIO emulation on its 7012behalf. However, for certain classes of instructions, no instruction decode 7013(direction, length of memory access) is provided, and fetching and decoding 7014the instruction from the VM is overly complicated to live in the kernel. 7015 7016Historically, when this situation occurred, KVM would print a warning and kill 7017the VM. KVM assumed that if the guest accessed non-memslot memory, it was 7018trying to do I/O, which just couldn't be emulated, and the warning message was 7019phrased accordingly. However, what happened more often was that a guest bug 7020caused access outside the guest memory areas which should lead to a more 7021meaningful warning message and an external abort in the guest, if the access 7022did not fall within an I/O window. 7023 7024Userspace implementations can query for KVM_CAP_ARM_NISV_TO_USER, and enable 7025this capability at VM creation. Once this is done, these types of errors will 7026instead return to userspace with KVM_EXIT_ARM_NISV, with the valid bits from 7027the ESR_EL2 in the esr_iss field, and the faulting IPA in the fault_ipa field. 7028Userspace can either fix up the access if it's actually an I/O access by 7029decoding the instruction from guest memory (if it's very brave) and continue 7030executing the guest, or it can decide to suspend, dump, or restart the guest. 7031 7032Note that KVM does not skip the faulting instruction as it does for 7033KVM_EXIT_MMIO, but userspace has to emulate any change to the processing state 7034if it decides to decode and emulate the instruction. 7035 7036This feature isn't available to protected VMs, as userspace does not 7037have access to the state that is required to perform the emulation. 7038Instead, a data abort exception is directly injected in the guest. 7039Note that although KVM_CAP_ARM_NISV_TO_USER will be reported if 7040queried outside of a protected VM context, the feature will not be 7041exposed if queried on a protected VM file descriptor. 7042 7043:: 7044 7045 /* KVM_EXIT_X86_RDMSR / KVM_EXIT_X86_WRMSR */ 7046 struct { 7047 __u8 error; /* user -> kernel */ 7048 __u8 pad[7]; 7049 __u32 reason; /* kernel -> user */ 7050 __u32 index; /* kernel -> user */ 7051 __u64 data; /* kernel <-> user */ 7052 } msr; 7053 7054Used on x86 systems. When the VM capability KVM_CAP_X86_USER_SPACE_MSR is 7055enabled, MSR accesses to registers that would invoke a #GP by KVM kernel code 7056may instead trigger a KVM_EXIT_X86_RDMSR exit for reads and KVM_EXIT_X86_WRMSR 7057exit for writes. 7058 7059The "reason" field specifies why the MSR interception occurred. Userspace will 7060only receive MSR exits when a particular reason was requested during through 7061ENABLE_CAP. Currently valid exit reasons are: 7062 7063============================ ======================================== 7064 KVM_MSR_EXIT_REASON_UNKNOWN access to MSR that is unknown to KVM 7065 KVM_MSR_EXIT_REASON_INVAL access to invalid MSRs or reserved bits 7066 KVM_MSR_EXIT_REASON_FILTER access blocked by KVM_X86_SET_MSR_FILTER 7067============================ ======================================== 7068 7069For KVM_EXIT_X86_RDMSR, the "index" field tells userspace which MSR the guest 7070wants to read. To respond to this request with a successful read, userspace 7071writes the respective data into the "data" field and must continue guest 7072execution to ensure the read data is transferred into guest register state. 7073 7074If the RDMSR request was unsuccessful, userspace indicates that with a "1" in 7075the "error" field. This will inject a #GP into the guest when the VCPU is 7076executed again. 7077 7078For KVM_EXIT_X86_WRMSR, the "index" field tells userspace which MSR the guest 7079wants to write. Once finished processing the event, userspace must continue 7080vCPU execution. If the MSR write was unsuccessful, userspace also sets the 7081"error" field to "1". 7082 7083See KVM_X86_SET_MSR_FILTER for details on the interaction with MSR filtering. 7084 7085:: 7086 7087 7088 struct kvm_xen_exit { 7089 #define KVM_EXIT_XEN_HCALL 1 7090 __u32 type; 7091 union { 7092 struct { 7093 __u32 longmode; 7094 __u32 cpl; 7095 __u64 input; 7096 __u64 result; 7097 __u64 params[6]; 7098 } hcall; 7099 } u; 7100 }; 7101 /* KVM_EXIT_XEN */ 7102 struct kvm_hyperv_exit xen; 7103 7104Indicates that the VCPU exits into userspace to process some tasks 7105related to Xen emulation. 7106 7107Valid values for 'type' are: 7108 7109 - KVM_EXIT_XEN_HCALL -- synchronously notify user-space about Xen hypercall. 7110 Userspace is expected to place the hypercall result into the appropriate 7111 field before invoking KVM_RUN again. 7112 7113:: 7114 7115 /* KVM_EXIT_RISCV_SBI */ 7116 struct { 7117 unsigned long extension_id; 7118 unsigned long function_id; 7119 unsigned long args[6]; 7120 unsigned long ret[2]; 7121 } riscv_sbi; 7122 7123If exit reason is KVM_EXIT_RISCV_SBI then it indicates that the VCPU has 7124done a SBI call which is not handled by KVM RISC-V kernel module. The details 7125of the SBI call are available in 'riscv_sbi' member of kvm_run structure. The 7126'extension_id' field of 'riscv_sbi' represents SBI extension ID whereas the 7127'function_id' field represents function ID of given SBI extension. The 'args' 7128array field of 'riscv_sbi' represents parameters for the SBI call and 'ret' 7129array field represents return values. The userspace should update the return 7130values of SBI call before resuming the VCPU. For more details on RISC-V SBI 7131spec refer, https://github.com/riscv/riscv-sbi-doc. 7132 7133:: 7134 7135 /* KVM_EXIT_MEMORY_FAULT */ 7136 struct { 7137 #define KVM_MEMORY_EXIT_FLAG_PRIVATE (1ULL << 3) 7138 __u64 flags; 7139 __u64 gpa; 7140 __u64 size; 7141 } memory_fault; 7142 7143KVM_EXIT_MEMORY_FAULT indicates the vCPU has encountered a memory fault that 7144could not be resolved by KVM. The 'gpa' and 'size' (in bytes) describe the 7145guest physical address range [gpa, gpa + size) of the fault. The 'flags' field 7146describes properties of the faulting access that are likely pertinent: 7147 7148 - KVM_MEMORY_EXIT_FLAG_PRIVATE - When set, indicates the memory fault occurred 7149 on a private memory access. When clear, indicates the fault occurred on a 7150 shared access. 7151 7152Note! KVM_EXIT_MEMORY_FAULT is unique among all KVM exit reasons in that it 7153accompanies a return code of '-1', not '0'! errno will always be set to EFAULT 7154or EHWPOISON when KVM exits with KVM_EXIT_MEMORY_FAULT, userspace should assume 7155kvm_run.exit_reason is stale/undefined for all other error numbers. 7156 7157:: 7158 7159 /* KVM_EXIT_NOTIFY */ 7160 struct { 7161 #define KVM_NOTIFY_CONTEXT_INVALID (1 << 0) 7162 __u32 flags; 7163 } notify; 7164 7165Used on x86 systems. When the VM capability KVM_CAP_X86_NOTIFY_VMEXIT is 7166enabled, a VM exit generated if no event window occurs in VM non-root mode 7167for a specified amount of time. Once KVM_X86_NOTIFY_VMEXIT_USER is set when 7168enabling the cap, it would exit to userspace with the exit reason 7169KVM_EXIT_NOTIFY for further handling. The "flags" field contains more 7170detailed info. 7171 7172The valid value for 'flags' is: 7173 7174 - KVM_NOTIFY_CONTEXT_INVALID -- the VM context is corrupted and not valid 7175 in VMCS. It would run into unknown result if resume the target VM. 7176 7177:: 7178 7179 /* Fix the size of the union. */ 7180 char padding[256]; 7181 }; 7182 7183 /* 7184 * shared registers between kvm and userspace. 7185 * kvm_valid_regs specifies the register classes set by the host 7186 * kvm_dirty_regs specified the register classes dirtied by userspace 7187 * struct kvm_sync_regs is architecture specific, as well as the 7188 * bits for kvm_valid_regs and kvm_dirty_regs 7189 */ 7190 __u64 kvm_valid_regs; 7191 __u64 kvm_dirty_regs; 7192 union { 7193 struct kvm_sync_regs regs; 7194 char padding[SYNC_REGS_SIZE_BYTES]; 7195 } s; 7196 7197If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access 7198certain guest registers without having to call SET/GET_*REGS. Thus we can 7199avoid some system call overhead if userspace has to handle the exit. 7200Userspace can query the validity of the structure by checking 7201kvm_valid_regs for specific bits. These bits are architecture specific 7202and usually define the validity of a groups of registers. (e.g. one bit 7203for general purpose registers) 7204 7205Please note that the kernel is allowed to use the kvm_run structure as the 7206primary storage for certain register types. Therefore, the kernel may use the 7207values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set. 7208 7209 7210.. _cap_enable: 7211 72126. Capabilities that can be enabled on vCPUs 7213============================================ 7214 7215There are certain capabilities that change the behavior of the virtual CPU or 7216the virtual machine when enabled. To enable them, please see 7217:ref:`KVM_ENABLE_CAP`. 7218 7219Below you can find a list of capabilities and what their effect on the vCPU or 7220the virtual machine is when enabling them. 7221 7222The following information is provided along with the description: 7223 7224 Architectures: 7225 which instruction set architectures provide this ioctl. 7226 x86 includes both i386 and x86_64. 7227 7228 Target: 7229 whether this is a per-vcpu or per-vm capability. 7230 7231 Parameters: 7232 what parameters are accepted by the capability. 7233 7234 Returns: 7235 the return value. General error numbers (EBADF, ENOMEM, EINVAL) 7236 are not detailed, but errors with specific meanings are. 7237 7238 72396.1 KVM_CAP_PPC_OSI 7240------------------- 7241 7242:Architectures: ppc 7243:Target: vcpu 7244:Parameters: none 7245:Returns: 0 on success; -1 on error 7246 7247This capability enables interception of OSI hypercalls that otherwise would 7248be treated as normal system calls to be injected into the guest. OSI hypercalls 7249were invented by Mac-on-Linux to have a standardized communication mechanism 7250between the guest and the host. 7251 7252When this capability is enabled, KVM_EXIT_OSI can occur. 7253 7254 72556.2 KVM_CAP_PPC_PAPR 7256-------------------- 7257 7258:Architectures: ppc 7259:Target: vcpu 7260:Parameters: none 7261:Returns: 0 on success; -1 on error 7262 7263This capability enables interception of PAPR hypercalls. PAPR hypercalls are 7264done using the hypercall instruction "sc 1". 7265 7266It also sets the guest privilege level to "supervisor" mode. Usually the guest 7267runs in "hypervisor" privilege mode with a few missing features. 7268 7269In addition to the above, it changes the semantics of SDR1. In this mode, the 7270HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the 7271HTAB invisible to the guest. 7272 7273When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur. 7274 7275 72766.3 KVM_CAP_SW_TLB 7277------------------ 7278 7279:Architectures: ppc 7280:Target: vcpu 7281:Parameters: args[0] is the address of a struct kvm_config_tlb 7282:Returns: 0 on success; -1 on error 7283 7284:: 7285 7286 struct kvm_config_tlb { 7287 __u64 params; 7288 __u64 array; 7289 __u32 mmu_type; 7290 __u32 array_len; 7291 }; 7292 7293Configures the virtual CPU's TLB array, establishing a shared memory area 7294between userspace and KVM. The "params" and "array" fields are userspace 7295addresses of mmu-type-specific data structures. The "array_len" field is an 7296safety mechanism, and should be set to the size in bytes of the memory that 7297userspace has reserved for the array. It must be at least the size dictated 7298by "mmu_type" and "params". 7299 7300While KVM_RUN is active, the shared region is under control of KVM. Its 7301contents are undefined, and any modification by userspace results in 7302boundedly undefined behavior. 7303 7304On return from KVM_RUN, the shared region will reflect the current state of 7305the guest's TLB. If userspace makes any changes, it must call KVM_DIRTY_TLB 7306to tell KVM which entries have been changed, prior to calling KVM_RUN again 7307on this vcpu. 7308 7309For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV: 7310 7311 - The "params" field is of type "struct kvm_book3e_206_tlb_params". 7312 - The "array" field points to an array of type "struct 7313 kvm_book3e_206_tlb_entry". 7314 - The array consists of all entries in the first TLB, followed by all 7315 entries in the second TLB. 7316 - Within a TLB, entries are ordered first by increasing set number. Within a 7317 set, entries are ordered by way (increasing ESEL). 7318 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1) 7319 where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value. 7320 - The tsize field of mas1 shall be set to 4K on TLB0, even though the 7321 hardware ignores this value for TLB0. 7322 73236.4 KVM_CAP_S390_CSS_SUPPORT 7324---------------------------- 7325 7326:Architectures: s390 7327:Target: vcpu 7328:Parameters: none 7329:Returns: 0 on success; -1 on error 7330 7331This capability enables support for handling of channel I/O instructions. 7332 7333TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are 7334handled in-kernel, while the other I/O instructions are passed to userspace. 7335 7336When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST 7337SUBCHANNEL intercepts. 7338 7339Note that even though this capability is enabled per-vcpu, the complete 7340virtual machine is affected. 7341 73426.5 KVM_CAP_PPC_EPR 7343------------------- 7344 7345:Architectures: ppc 7346:Target: vcpu 7347:Parameters: args[0] defines whether the proxy facility is active 7348:Returns: 0 on success; -1 on error 7349 7350This capability enables or disables the delivery of interrupts through the 7351external proxy facility. 7352 7353When enabled (args[0] != 0), every time the guest gets an external interrupt 7354delivered, it automatically exits into user space with a KVM_EXIT_EPR exit 7355to receive the topmost interrupt vector. 7356 7357When disabled (args[0] == 0), behavior is as if this facility is unsupported. 7358 7359When this capability is enabled, KVM_EXIT_EPR can occur. 7360 73616.6 KVM_CAP_IRQ_MPIC 7362-------------------- 7363 7364:Architectures: ppc 7365:Parameters: args[0] is the MPIC device fd; 7366 args[1] is the MPIC CPU number for this vcpu 7367 7368This capability connects the vcpu to an in-kernel MPIC device. 7369 73706.7 KVM_CAP_IRQ_XICS 7371-------------------- 7372 7373:Architectures: ppc 7374:Target: vcpu 7375:Parameters: args[0] is the XICS device fd; 7376 args[1] is the XICS CPU number (server ID) for this vcpu 7377 7378This capability connects the vcpu to an in-kernel XICS device. 7379 73806.8 KVM_CAP_S390_IRQCHIP 7381------------------------ 7382 7383:Architectures: s390 7384:Target: vm 7385:Parameters: none 7386 7387This capability enables the in-kernel irqchip for s390. Please refer to 7388"4.24 KVM_CREATE_IRQCHIP" for details. 7389 73906.9 KVM_CAP_MIPS_FPU 7391-------------------- 7392 7393:Architectures: mips 7394:Target: vcpu 7395:Parameters: args[0] is reserved for future use (should be 0). 7396 7397This capability allows the use of the host Floating Point Unit by the guest. It 7398allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is 7399done the ``KVM_REG_MIPS_FPR_*`` and ``KVM_REG_MIPS_FCR_*`` registers can be 7400accessed (depending on the current guest FPU register mode), and the Status.FR, 7401Config5.FRE bits are accessible via the KVM API and also from the guest, 7402depending on them being supported by the FPU. 7403 74046.10 KVM_CAP_MIPS_MSA 7405--------------------- 7406 7407:Architectures: mips 7408:Target: vcpu 7409:Parameters: args[0] is reserved for future use (should be 0). 7410 7411This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest. 7412It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest. 7413Once this is done the ``KVM_REG_MIPS_VEC_*`` and ``KVM_REG_MIPS_MSA_*`` 7414registers can be accessed, and the Config5.MSAEn bit is accessible via the 7415KVM API and also from the guest. 7416 74176.74 KVM_CAP_SYNC_REGS 7418---------------------- 7419 7420:Architectures: s390, x86 7421:Target: s390: always enabled, x86: vcpu 7422:Parameters: none 7423:Returns: x86: KVM_CHECK_EXTENSION returns a bit-array indicating which register 7424 sets are supported 7425 (bitfields defined in arch/x86/include/uapi/asm/kvm.h). 7426 7427As described above in the kvm_sync_regs struct info in section :ref:`kvm_run`, 7428KVM_CAP_SYNC_REGS "allow[s] userspace to access certain guest registers 7429without having to call SET/GET_*REGS". This reduces overhead by eliminating 7430repeated ioctl calls for setting and/or getting register values. This is 7431particularly important when userspace is making synchronous guest state 7432modifications, e.g. when emulating and/or intercepting instructions in 7433userspace. 7434 7435For s390 specifics, please refer to the source code. 7436 7437For x86: 7438 7439- the register sets to be copied out to kvm_run are selectable 7440 by userspace (rather that all sets being copied out for every exit). 7441- vcpu_events are available in addition to regs and sregs. 7442 7443For x86, the 'kvm_valid_regs' field of struct kvm_run is overloaded to 7444function as an input bit-array field set by userspace to indicate the 7445specific register sets to be copied out on the next exit. 7446 7447To indicate when userspace has modified values that should be copied into 7448the vCPU, the all architecture bitarray field, 'kvm_dirty_regs' must be set. 7449This is done using the same bitflags as for the 'kvm_valid_regs' field. 7450If the dirty bit is not set, then the register set values will not be copied 7451into the vCPU even if they've been modified. 7452 7453Unused bitfields in the bitarrays must be set to zero. 7454 7455:: 7456 7457 struct kvm_sync_regs { 7458 struct kvm_regs regs; 7459 struct kvm_sregs sregs; 7460 struct kvm_vcpu_events events; 7461 }; 7462 74636.75 KVM_CAP_PPC_IRQ_XIVE 7464------------------------- 7465 7466:Architectures: ppc 7467:Target: vcpu 7468:Parameters: args[0] is the XIVE device fd; 7469 args[1] is the XIVE CPU number (server ID) for this vcpu 7470 7471This capability connects the vcpu to an in-kernel XIVE device. 7472 74736.76 KVM_CAP_HYPERV_SYNIC 7474------------------------- 7475 7476:Architectures: x86 7477:Target: vcpu 7478 7479This capability, if KVM_CHECK_EXTENSION indicates that it is 7480available, means that the kernel has an implementation of the 7481Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is 7482used to support Windows Hyper-V based guest paravirt drivers(VMBus). 7483 7484In order to use SynIC, it has to be activated by setting this 7485capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this 7486will disable the use of APIC hardware virtualization even if supported 7487by the CPU, as it's incompatible with SynIC auto-EOI behavior. 7488 74896.77 KVM_CAP_HYPERV_SYNIC2 7490-------------------------- 7491 7492:Architectures: x86 7493:Target: vcpu 7494 7495This capability enables a newer version of Hyper-V Synthetic interrupt 7496controller (SynIC). The only difference with KVM_CAP_HYPERV_SYNIC is that KVM 7497doesn't clear SynIC message and event flags pages when they are enabled by 7498writing to the respective MSRs. 7499 75006.78 KVM_CAP_HYPERV_DIRECT_TLBFLUSH 7501----------------------------------- 7502 7503:Architectures: x86 7504:Target: vcpu 7505 7506This capability indicates that KVM running on top of Hyper-V hypervisor 7507enables Direct TLB flush for its guests meaning that TLB flush 7508hypercalls are handled by Level 0 hypervisor (Hyper-V) bypassing KVM. 7509Due to the different ABI for hypercall parameters between Hyper-V and 7510KVM, enabling this capability effectively disables all hypercall 7511handling by KVM (as some KVM hypercall may be mistakenly treated as TLB 7512flush hypercalls by Hyper-V) so userspace should disable KVM identification 7513in CPUID and only exposes Hyper-V identification. In this case, guest 7514thinks it's running on Hyper-V and only use Hyper-V hypercalls. 7515 75166.79 KVM_CAP_HYPERV_ENFORCE_CPUID 7517--------------------------------- 7518 7519:Architectures: x86 7520:Target: vcpu 7521 7522When enabled, KVM will disable emulated Hyper-V features provided to the 7523guest according to the bits Hyper-V CPUID feature leaves. Otherwise, all 7524currently implemented Hyper-V features are provided unconditionally when 7525Hyper-V identification is set in the HYPERV_CPUID_INTERFACE (0x40000001) 7526leaf. 7527 75286.80 KVM_CAP_ENFORCE_PV_FEATURE_CPUID 7529------------------------------------- 7530 7531:Architectures: x86 7532:Target: vcpu 7533 7534When enabled, KVM will disable paravirtual features provided to the 7535guest according to the bits in the KVM_CPUID_FEATURES CPUID leaf 7536(0x40000001). Otherwise, a guest may use the paravirtual features 7537regardless of what has actually been exposed through the CPUID leaf. 7538 7539.. _KVM_CAP_DIRTY_LOG_RING: 7540 7541 7542.. _cap_enable_vm: 7543 75447. Capabilities that can be enabled on VMs 7545========================================== 7546 7547There are certain capabilities that change the behavior of the virtual 7548machine when enabled. To enable them, please see section 7549:ref:`KVM_ENABLE_CAP`. Below you can find a list of capabilities and 7550what their effect on the VM is when enabling them. 7551 7552The following information is provided along with the description: 7553 7554 Architectures: 7555 which instruction set architectures provide this ioctl. 7556 x86 includes both i386 and x86_64. 7557 7558 Parameters: 7559 what parameters are accepted by the capability. 7560 7561 Returns: 7562 the return value. General error numbers (EBADF, ENOMEM, EINVAL) 7563 are not detailed, but errors with specific meanings are. 7564 7565 75667.1 KVM_CAP_PPC_ENABLE_HCALL 7567---------------------------- 7568 7569:Architectures: ppc 7570:Parameters: args[0] is the sPAPR hcall number; 7571 args[1] is 0 to disable, 1 to enable in-kernel handling 7572 7573This capability controls whether individual sPAPR hypercalls (hcalls) 7574get handled by the kernel or not. Enabling or disabling in-kernel 7575handling of an hcall is effective across the VM. On creation, an 7576initial set of hcalls are enabled for in-kernel handling, which 7577consists of those hcalls for which in-kernel handlers were implemented 7578before this capability was implemented. If disabled, the kernel will 7579not to attempt to handle the hcall, but will always exit to userspace 7580to handle it. Note that it may not make sense to enable some and 7581disable others of a group of related hcalls, but KVM does not prevent 7582userspace from doing that. 7583 7584If the hcall number specified is not one that has an in-kernel 7585implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL 7586error. 7587 75887.2 KVM_CAP_S390_USER_SIGP 7589-------------------------- 7590 7591:Architectures: s390 7592:Parameters: none 7593 7594This capability controls which SIGP orders will be handled completely in user 7595space. With this capability enabled, all fast orders will be handled completely 7596in the kernel: 7597 7598- SENSE 7599- SENSE RUNNING 7600- EXTERNAL CALL 7601- EMERGENCY SIGNAL 7602- CONDITIONAL EMERGENCY SIGNAL 7603 7604All other orders will be handled completely in user space. 7605 7606Only privileged operation exceptions will be checked for in the kernel (or even 7607in the hardware prior to interception). If this capability is not enabled, the 7608old way of handling SIGP orders is used (partially in kernel and user space). 7609 76107.3 KVM_CAP_S390_VECTOR_REGISTERS 7611--------------------------------- 7612 7613:Architectures: s390 7614:Parameters: none 7615:Returns: 0 on success, negative value on error 7616 7617Allows use of the vector registers introduced with z13 processor, and 7618provides for the synchronization between host and user space. Will 7619return -EINVAL if the machine does not support vectors. 7620 76217.4 KVM_CAP_S390_USER_STSI 7622-------------------------- 7623 7624:Architectures: s390 7625:Parameters: none 7626 7627This capability allows post-handlers for the STSI instruction. After 7628initial handling in the kernel, KVM exits to user space with 7629KVM_EXIT_S390_STSI to allow user space to insert further data. 7630 7631Before exiting to userspace, kvm handlers should fill in s390_stsi field of 7632vcpu->run:: 7633 7634 struct { 7635 __u64 addr; 7636 __u8 ar; 7637 __u8 reserved; 7638 __u8 fc; 7639 __u8 sel1; 7640 __u16 sel2; 7641 } s390_stsi; 7642 7643 @addr - guest address of STSI SYSIB 7644 @fc - function code 7645 @sel1 - selector 1 7646 @sel2 - selector 2 7647 @ar - access register number 7648 7649KVM handlers should exit to userspace with rc = -EREMOTE. 7650 76517.5 KVM_CAP_SPLIT_IRQCHIP 7652------------------------- 7653 7654:Architectures: x86 7655:Parameters: args[0] - number of routes reserved for userspace IOAPICs 7656:Returns: 0 on success, -1 on error 7657 7658Create a local apic for each processor in the kernel. This can be used 7659instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the 7660IOAPIC and PIC (and also the PIT, even though this has to be enabled 7661separately). 7662 7663This capability also enables in kernel routing of interrupt requests; 7664when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are 7665used in the IRQ routing table. The first args[0] MSI routes are reserved 7666for the IOAPIC pins. Whenever the LAPIC receives an EOI for these routes, 7667a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace. 7668 7669Fails if VCPU has already been created, or if the irqchip is already in the 7670kernel (i.e. KVM_CREATE_IRQCHIP has already been called). 7671 76727.6 KVM_CAP_S390_RI 7673------------------- 7674 7675:Architectures: s390 7676:Parameters: none 7677 7678Allows use of runtime-instrumentation introduced with zEC12 processor. 7679Will return -EINVAL if the machine does not support runtime-instrumentation. 7680Will return -EBUSY if a VCPU has already been created. 7681 76827.7 KVM_CAP_X2APIC_API 7683---------------------- 7684 7685:Architectures: x86 7686:Parameters: args[0] - features that should be enabled 7687:Returns: 0 on success, -EINVAL when args[0] contains invalid features 7688 7689Valid feature flags in args[0] are:: 7690 7691 #define KVM_X2APIC_API_USE_32BIT_IDS (1ULL << 0) 7692 #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK (1ULL << 1) 7693 7694Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of 7695KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC, 7696allowing the use of 32-bit APIC IDs. See KVM_CAP_X2APIC_API in their 7697respective sections. 7698 7699KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work 7700in logical mode or with more than 255 VCPUs. Otherwise, KVM treats 0xff 7701as a broadcast even in x2APIC mode in order to support physical x2APIC 7702without interrupt remapping. This is undesirable in logical mode, 7703where 0xff represents CPUs 0-7 in cluster 0. 7704 77057.8 KVM_CAP_S390_USER_INSTR0 7706---------------------------- 7707 7708:Architectures: s390 7709:Parameters: none 7710 7711With this capability enabled, all illegal instructions 0x0000 (2 bytes) will 7712be intercepted and forwarded to user space. User space can use this 7713mechanism e.g. to realize 2-byte software breakpoints. The kernel will 7714not inject an operating exception for these instructions, user space has 7715to take care of that. 7716 7717This capability can be enabled dynamically even if VCPUs were already 7718created and are running. 7719 77207.9 KVM_CAP_S390_GS 7721------------------- 7722 7723:Architectures: s390 7724:Parameters: none 7725:Returns: 0 on success; -EINVAL if the machine does not support 7726 guarded storage; -EBUSY if a VCPU has already been created. 7727 7728Allows use of guarded storage for the KVM guest. 7729 77307.10 KVM_CAP_S390_AIS 7731--------------------- 7732 7733:Architectures: s390 7734:Parameters: none 7735 7736Allow use of adapter-interruption suppression. 7737:Returns: 0 on success; -EBUSY if a VCPU has already been created. 7738 77397.11 KVM_CAP_PPC_SMT 7740-------------------- 7741 7742:Architectures: ppc 7743:Parameters: vsmt_mode, flags 7744 7745Enabling this capability on a VM provides userspace with a way to set 7746the desired virtual SMT mode (i.e. the number of virtual CPUs per 7747virtual core). The virtual SMT mode, vsmt_mode, must be a power of 2 7748between 1 and 8. On POWER8, vsmt_mode must also be no greater than 7749the number of threads per subcore for the host. Currently flags must 7750be 0. A successful call to enable this capability will result in 7751vsmt_mode being returned when the KVM_CAP_PPC_SMT capability is 7752subsequently queried for the VM. This capability is only supported by 7753HV KVM, and can only be set before any VCPUs have been created. 7754The KVM_CAP_PPC_SMT_POSSIBLE capability indicates which virtual SMT 7755modes are available. 7756 77577.12 KVM_CAP_PPC_FWNMI 7758---------------------- 7759 7760:Architectures: ppc 7761:Parameters: none 7762 7763With this capability a machine check exception in the guest address 7764space will cause KVM to exit the guest with NMI exit reason. This 7765enables QEMU to build error log and branch to guest kernel registered 7766machine check handling routine. Without this capability KVM will 7767branch to guests' 0x200 interrupt vector. 7768 77697.13 KVM_CAP_X86_DISABLE_EXITS 7770------------------------------ 7771 7772:Architectures: x86 7773:Parameters: args[0] defines which exits are disabled 7774:Returns: 0 on success, -EINVAL when args[0] contains invalid exits 7775 or if any vCPUs have already been created 7776 7777Valid bits in args[0] are:: 7778 7779 #define KVM_X86_DISABLE_EXITS_MWAIT (1 << 0) 7780 #define KVM_X86_DISABLE_EXITS_HLT (1 << 1) 7781 #define KVM_X86_DISABLE_EXITS_PAUSE (1 << 2) 7782 #define KVM_X86_DISABLE_EXITS_CSTATE (1 << 3) 7783 7784Enabling this capability on a VM provides userspace with a way to no 7785longer intercept some instructions for improved latency in some 7786workloads, and is suggested when vCPUs are associated to dedicated 7787physical CPUs. More bits can be added in the future; userspace can 7788just pass the KVM_CHECK_EXTENSION result to KVM_ENABLE_CAP to disable 7789all such vmexits. 7790 7791Do not enable KVM_FEATURE_PV_UNHALT if you disable HLT exits. 7792 77937.14 KVM_CAP_S390_HPAGE_1M 7794-------------------------- 7795 7796:Architectures: s390 7797:Parameters: none 7798:Returns: 0 on success, -EINVAL if hpage module parameter was not set 7799 or cmma is enabled, or the VM has the KVM_VM_S390_UCONTROL 7800 flag set 7801 7802With this capability the KVM support for memory backing with 1m pages 7803through hugetlbfs can be enabled for a VM. After the capability is 7804enabled, cmma can't be enabled anymore and pfmfi and the storage key 7805interpretation are disabled. If cmma has already been enabled or the 7806hpage module parameter is not set to 1, -EINVAL is returned. 7807 7808While it is generally possible to create a huge page backed VM without 7809this capability, the VM will not be able to run. 7810 78117.15 KVM_CAP_MSR_PLATFORM_INFO 7812------------------------------ 7813 7814:Architectures: x86 7815:Parameters: args[0] whether feature should be enabled or not 7816 7817With this capability, a guest may read the MSR_PLATFORM_INFO MSR. Otherwise, 7818a #GP would be raised when the guest tries to access. Currently, this 7819capability does not enable write permissions of this MSR for the guest. 7820 78217.16 KVM_CAP_PPC_NESTED_HV 7822-------------------------- 7823 7824:Architectures: ppc 7825:Parameters: none 7826:Returns: 0 on success, -EINVAL when the implementation doesn't support 7827 nested-HV virtualization. 7828 7829HV-KVM on POWER9 and later systems allows for "nested-HV" 7830virtualization, which provides a way for a guest VM to run guests that 7831can run using the CPU's supervisor mode (privileged non-hypervisor 7832state). Enabling this capability on a VM depends on the CPU having 7833the necessary functionality and on the facility being enabled with a 7834kvm-hv module parameter. 7835 78367.17 KVM_CAP_EXCEPTION_PAYLOAD 7837------------------------------ 7838 7839:Architectures: x86 7840:Parameters: args[0] whether feature should be enabled or not 7841 7842With this capability enabled, CR2 will not be modified prior to the 7843emulated VM-exit when L1 intercepts a #PF exception that occurs in 7844L2. Similarly, for kvm-intel only, DR6 will not be modified prior to 7845the emulated VM-exit when L1 intercepts a #DB exception that occurs in 7846L2. As a result, when KVM_GET_VCPU_EVENTS reports a pending #PF (or 7847#DB) exception for L2, exception.has_payload will be set and the 7848faulting address (or the new DR6 bits*) will be reported in the 7849exception_payload field. Similarly, when userspace injects a #PF (or 7850#DB) into L2 using KVM_SET_VCPU_EVENTS, it is expected to set 7851exception.has_payload and to put the faulting address - or the new DR6 7852bits\ [#]_ - in the exception_payload field. 7853 7854This capability also enables exception.pending in struct 7855kvm_vcpu_events, which allows userspace to distinguish between pending 7856and injected exceptions. 7857 7858 7859.. [#] For the new DR6 bits, note that bit 16 is set iff the #DB exception 7860 will clear DR6.RTM. 7861 78627.18 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 7863-------------------------------------- 7864 7865:Architectures: x86, arm64, mips 7866:Parameters: args[0] whether feature should be enabled or not 7867 7868Valid flags are:: 7869 7870 #define KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (1 << 0) 7871 #define KVM_DIRTY_LOG_INITIALLY_SET (1 << 1) 7872 7873With KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE is set, KVM_GET_DIRTY_LOG will not 7874automatically clear and write-protect all pages that are returned as dirty. 7875Rather, userspace will have to do this operation separately using 7876KVM_CLEAR_DIRTY_LOG. 7877 7878At the cost of a slightly more complicated operation, this provides better 7879scalability and responsiveness for two reasons. First, 7880KVM_CLEAR_DIRTY_LOG ioctl can operate on a 64-page granularity rather 7881than requiring to sync a full memslot; this ensures that KVM does not 7882take spinlocks for an extended period of time. Second, in some cases a 7883large amount of time can pass between a call to KVM_GET_DIRTY_LOG and 7884userspace actually using the data in the page. Pages can be modified 7885during this time, which is inefficient for both the guest and userspace: 7886the guest will incur a higher penalty due to write protection faults, 7887while userspace can see false reports of dirty pages. Manual reprotection 7888helps reducing this time, improving guest performance and reducing the 7889number of dirty log false positives. 7890 7891With KVM_DIRTY_LOG_INITIALLY_SET set, all the bits of the dirty bitmap 7892will be initialized to 1 when created. This also improves performance because 7893dirty logging can be enabled gradually in small chunks on the first call 7894to KVM_CLEAR_DIRTY_LOG. KVM_DIRTY_LOG_INITIALLY_SET depends on 7895KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (it is also only available on 7896x86 and arm64 for now). 7897 7898KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 was previously available under the name 7899KVM_CAP_MANUAL_DIRTY_LOG_PROTECT, but the implementation had bugs that make 7900it hard or impossible to use it correctly. The availability of 7901KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 signals that those bugs are fixed. 7902Userspace should not try to use KVM_CAP_MANUAL_DIRTY_LOG_PROTECT. 7903 79047.19 KVM_CAP_PPC_SECURE_GUEST 7905------------------------------ 7906 7907:Architectures: ppc 7908 7909This capability indicates that KVM is running on a host that has 7910ultravisor firmware and thus can support a secure guest. On such a 7911system, a guest can ask the ultravisor to make it a secure guest, 7912one whose memory is inaccessible to the host except for pages which 7913are explicitly requested to be shared with the host. The ultravisor 7914notifies KVM when a guest requests to become a secure guest, and KVM 7915has the opportunity to veto the transition. 7916 7917If present, this capability can be enabled for a VM, meaning that KVM 7918will allow the transition to secure guest mode. Otherwise KVM will 7919veto the transition. 7920 79217.20 KVM_CAP_HALT_POLL 7922---------------------- 7923 7924:Architectures: all 7925:Target: VM 7926:Parameters: args[0] is the maximum poll time in nanoseconds 7927:Returns: 0 on success; -1 on error 7928 7929KVM_CAP_HALT_POLL overrides the kvm.halt_poll_ns module parameter to set the 7930maximum halt-polling time for all vCPUs in the target VM. This capability can 7931be invoked at any time and any number of times to dynamically change the 7932maximum halt-polling time. 7933 7934See Documentation/virt/kvm/halt-polling.rst for more information on halt 7935polling. 7936 79377.21 KVM_CAP_X86_USER_SPACE_MSR 7938------------------------------- 7939 7940:Architectures: x86 7941:Target: VM 7942:Parameters: args[0] contains the mask of KVM_MSR_EXIT_REASON_* events to report 7943:Returns: 0 on success; -1 on error 7944 7945This capability allows userspace to intercept RDMSR and WRMSR instructions if 7946access to an MSR is denied. By default, KVM injects #GP on denied accesses. 7947 7948When a guest requests to read or write an MSR, KVM may not implement all MSRs 7949that are relevant to a respective system. It also does not differentiate by 7950CPU type. 7951 7952To allow more fine grained control over MSR handling, userspace may enable 7953this capability. With it enabled, MSR accesses that match the mask specified in 7954args[0] and would trigger a #GP inside the guest will instead trigger 7955KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR exit notifications. Userspace 7956can then implement model specific MSR handling and/or user notifications 7957to inform a user that an MSR was not emulated/virtualized by KVM. 7958 7959The valid mask flags are: 7960 7961============================ =============================================== 7962 KVM_MSR_EXIT_REASON_UNKNOWN intercept accesses to unknown (to KVM) MSRs 7963 KVM_MSR_EXIT_REASON_INVAL intercept accesses that are architecturally 7964 invalid according to the vCPU model and/or mode 7965 KVM_MSR_EXIT_REASON_FILTER intercept accesses that are denied by userspace 7966 via KVM_X86_SET_MSR_FILTER 7967============================ =============================================== 7968 79697.22 KVM_CAP_X86_BUS_LOCK_EXIT 7970------------------------------- 7971 7972:Architectures: x86 7973:Target: VM 7974:Parameters: args[0] defines the policy used when bus locks detected in guest 7975:Returns: 0 on success, -EINVAL when args[0] contains invalid bits 7976 7977Valid bits in args[0] are:: 7978 7979 #define KVM_BUS_LOCK_DETECTION_OFF (1 << 0) 7980 #define KVM_BUS_LOCK_DETECTION_EXIT (1 << 1) 7981 7982Enabling this capability on a VM provides userspace with a way to select a 7983policy to handle the bus locks detected in guest. Userspace can obtain the 7984supported modes from the result of KVM_CHECK_EXTENSION and define it through 7985the KVM_ENABLE_CAP. The supported modes are mutually-exclusive. 7986 7987This capability allows userspace to force VM exits on bus locks detected in the 7988guest, irrespective whether or not the host has enabled split-lock detection 7989(which triggers an #AC exception that KVM intercepts). This capability is 7990intended to mitigate attacks where a malicious/buggy guest can exploit bus 7991locks to degrade the performance of the whole system. 7992 7993If KVM_BUS_LOCK_DETECTION_OFF is set, KVM doesn't force guest bus locks to VM 7994exit, although the host kernel's split-lock #AC detection still applies, if 7995enabled. 7996 7997If KVM_BUS_LOCK_DETECTION_EXIT is set, KVM enables a CPU feature that ensures 7998bus locks in the guest trigger a VM exit, and KVM exits to userspace for all 7999such VM exits, e.g. to allow userspace to throttle the offending guest and/or 8000apply some other policy-based mitigation. When exiting to userspace, KVM sets 8001KVM_RUN_X86_BUS_LOCK in vcpu-run->flags, and conditionally sets the exit_reason 8002to KVM_EXIT_X86_BUS_LOCK. 8003 8004Due to differences in the underlying hardware implementation, the vCPU's RIP at 8005the time of exit diverges between Intel and AMD. On Intel hosts, RIP points at 8006the next instruction, i.e. the exit is trap-like. On AMD hosts, RIP points at 8007the offending instruction, i.e. the exit is fault-like. 8008 8009Note! Detected bus locks may be coincident with other exits to userspace, i.e. 8010KVM_RUN_X86_BUS_LOCK should be checked regardless of the primary exit reason if 8011userspace wants to take action on all detected bus locks. 8012 80137.23 KVM_CAP_PPC_DAWR1 8014---------------------- 8015 8016:Architectures: ppc 8017:Parameters: none 8018:Returns: 0 on success, -EINVAL when CPU doesn't support 2nd DAWR 8019 8020This capability can be used to check / enable 2nd DAWR feature provided 8021by POWER10 processor. 8022 8023 80247.24 KVM_CAP_VM_COPY_ENC_CONTEXT_FROM 8025------------------------------------- 8026 8027:Architectures: x86 SEV enabled 8028:Type: vm 8029:Parameters: args[0] is the fd of the source vm 8030:Returns: 0 on success; ENOTTY on error 8031 8032This capability enables userspace to copy encryption context from the vm 8033indicated by the fd to the vm this is called on. 8034 8035This is intended to support in-guest workloads scheduled by the host. This 8036allows the in-guest workload to maintain its own NPTs and keeps the two vms 8037from accidentally clobbering each other with interrupts and the like (separate 8038APIC/MSRs/etc). 8039 80407.25 KVM_CAP_SGX_ATTRIBUTE 8041-------------------------- 8042 8043:Architectures: x86 8044:Target: VM 8045:Parameters: args[0] is a file handle of a SGX attribute file in securityfs 8046:Returns: 0 on success, -EINVAL if the file handle is invalid or if a requested 8047 attribute is not supported by KVM. 8048 8049KVM_CAP_SGX_ATTRIBUTE enables a userspace VMM to grant a VM access to one or 8050more privileged enclave attributes. args[0] must hold a file handle to a valid 8051SGX attribute file corresponding to an attribute that is supported/restricted 8052by KVM (currently only PROVISIONKEY). 8053 8054The SGX subsystem restricts access to a subset of enclave attributes to provide 8055additional security for an uncompromised kernel, e.g. use of the PROVISIONKEY 8056is restricted to deter malware from using the PROVISIONKEY to obtain a stable 8057system fingerprint. To prevent userspace from circumventing such restrictions 8058by running an enclave in a VM, KVM prevents access to privileged attributes by 8059default. 8060 8061See Documentation/arch/x86/sgx.rst for more details. 8062 80637.27 KVM_CAP_EXIT_ON_EMULATION_FAILURE 8064-------------------------------------- 8065 8066:Architectures: x86 8067:Parameters: args[0] whether the feature should be enabled or not 8068 8069When this capability is enabled, an emulation failure will result in an exit 8070to userspace with KVM_INTERNAL_ERROR (except when the emulator was invoked 8071to handle a VMware backdoor instruction). Furthermore, KVM will now provide up 8072to 15 instruction bytes for any exit to userspace resulting from an emulation 8073failure. When these exits to userspace occur use the emulation_failure struct 8074instead of the internal struct. They both have the same layout, but the 8075emulation_failure struct matches the content better. It also explicitly 8076defines the 'flags' field which is used to describe the fields in the struct 8077that are valid (ie: if KVM_INTERNAL_ERROR_EMULATION_FLAG_INSTRUCTION_BYTES is 8078set in the 'flags' field then both 'insn_size' and 'insn_bytes' have valid data 8079in them.) 8080 80817.28 KVM_CAP_ARM_MTE 8082-------------------- 8083 8084:Architectures: arm64 8085:Parameters: none 8086 8087This capability indicates that KVM (and the hardware) supports exposing the 8088Memory Tagging Extensions (MTE) to the guest. It must also be enabled by the 8089VMM before creating any VCPUs to allow the guest access. Note that MTE is only 8090available to a guest running in AArch64 mode and enabling this capability will 8091cause attempts to create AArch32 VCPUs to fail. 8092 8093When enabled the guest is able to access tags associated with any memory given 8094to the guest. KVM will ensure that the tags are maintained during swap or 8095hibernation of the host; however the VMM needs to manually save/restore the 8096tags as appropriate if the VM is migrated. 8097 8098When this capability is enabled all memory in memslots must be mapped as 8099``MAP_ANONYMOUS`` or with a RAM-based file mapping (``tmpfs``, ``memfd``), 8100attempts to create a memslot with an invalid mmap will result in an 8101-EINVAL return. 8102 8103When enabled the VMM may make use of the ``KVM_ARM_MTE_COPY_TAGS`` ioctl to 8104perform a bulk copy of tags to/from the guest. 8105 81067.29 KVM_CAP_VM_MOVE_ENC_CONTEXT_FROM 8107------------------------------------- 8108 8109:Architectures: x86 SEV enabled 8110:Type: vm 8111:Parameters: args[0] is the fd of the source vm 8112:Returns: 0 on success 8113 8114This capability enables userspace to migrate the encryption context from the VM 8115indicated by the fd to the VM this is called on. 8116 8117This is intended to support intra-host migration of VMs between userspace VMMs, 8118upgrading the VMM process without interrupting the guest. 8119 81207.31 KVM_CAP_DISABLE_QUIRKS2 8121---------------------------- 8122 8123:Parameters: args[0] - set of KVM quirks to disable 8124:Architectures: x86 8125:Type: vm 8126 8127This capability, if enabled, will cause KVM to disable some behavior 8128quirks. 8129 8130Calling KVM_CHECK_EXTENSION for this capability returns a bitmask of 8131quirks that can be disabled in KVM. 8132 8133The argument to KVM_ENABLE_CAP for this capability is a bitmask of 8134quirks to disable, and must be a subset of the bitmask returned by 8135KVM_CHECK_EXTENSION. 8136 8137The valid bits in cap.args[0] are: 8138 8139=================================== ============================================ 8140 KVM_X86_QUIRK_LINT0_REENABLED By default, the reset value for the LVT 8141 LINT0 register is 0x700 (APIC_MODE_EXTINT). 8142 When this quirk is disabled, the reset value 8143 is 0x10000 (APIC_LVT_MASKED). 8144 8145 KVM_X86_QUIRK_CD_NW_CLEARED By default, KVM clears CR0.CD and CR0.NW on 8146 AMD CPUs to workaround buggy guest firmware 8147 that runs in perpetuity with CR0.CD, i.e. 8148 with caches in "no fill" mode. 8149 8150 When this quirk is disabled, KVM does not 8151 change the value of CR0.CD and CR0.NW. 8152 8153 KVM_X86_QUIRK_LAPIC_MMIO_HOLE By default, the MMIO LAPIC interface is 8154 available even when configured for x2APIC 8155 mode. When this quirk is disabled, KVM 8156 disables the MMIO LAPIC interface if the 8157 LAPIC is in x2APIC mode. 8158 8159 KVM_X86_QUIRK_OUT_7E_INC_RIP By default, KVM pre-increments %rip before 8160 exiting to userspace for an OUT instruction 8161 to port 0x7e. When this quirk is disabled, 8162 KVM does not pre-increment %rip before 8163 exiting to userspace. 8164 8165 KVM_X86_QUIRK_MISC_ENABLE_NO_MWAIT When this quirk is disabled, KVM sets 8166 CPUID.01H:ECX[bit 3] (MONITOR/MWAIT) if 8167 IA32_MISC_ENABLE[bit 18] (MWAIT) is set. 8168 Additionally, when this quirk is disabled, 8169 KVM clears CPUID.01H:ECX[bit 3] if 8170 IA32_MISC_ENABLE[bit 18] is cleared. 8171 8172 KVM_X86_QUIRK_FIX_HYPERCALL_INSN By default, KVM rewrites guest 8173 VMMCALL/VMCALL instructions to match the 8174 vendor's hypercall instruction for the 8175 system. When this quirk is disabled, KVM 8176 will no longer rewrite invalid guest 8177 hypercall instructions. Executing the 8178 incorrect hypercall instruction will 8179 generate a #UD within the guest. 8180 8181KVM_X86_QUIRK_MWAIT_NEVER_UD_FAULTS By default, KVM emulates MONITOR/MWAIT (if 8182 they are intercepted) as NOPs regardless of 8183 whether or not MONITOR/MWAIT are supported 8184 according to guest CPUID. When this quirk 8185 is disabled and KVM_X86_DISABLE_EXITS_MWAIT 8186 is not set (MONITOR/MWAIT are intercepted), 8187 KVM will inject a #UD on MONITOR/MWAIT if 8188 they're unsupported per guest CPUID. Note, 8189 KVM will modify MONITOR/MWAIT support in 8190 guest CPUID on writes to MISC_ENABLE if 8191 KVM_X86_QUIRK_MISC_ENABLE_NO_MWAIT is 8192 disabled. 8193 8194KVM_X86_QUIRK_SLOT_ZAP_ALL By default, for KVM_X86_DEFAULT_VM VMs, KVM 8195 invalidates all SPTEs in all memslots and 8196 address spaces when a memslot is deleted or 8197 moved. When this quirk is disabled (or the 8198 VM type isn't KVM_X86_DEFAULT_VM), KVM only 8199 ensures the backing memory of the deleted 8200 or moved memslot isn't reachable, i.e KVM 8201 _may_ invalidate only SPTEs related to the 8202 memslot. 8203 8204KVM_X86_QUIRK_STUFF_FEATURE_MSRS By default, at vCPU creation, KVM sets the 8205 vCPU's MSR_IA32_PERF_CAPABILITIES (0x345), 8206 MSR_IA32_ARCH_CAPABILITIES (0x10a), 8207 MSR_PLATFORM_INFO (0xce), and all VMX MSRs 8208 (0x480..0x492) to the maximal capabilities 8209 supported by KVM. KVM also sets 8210 MSR_IA32_UCODE_REV (0x8b) to an arbitrary 8211 value (which is different for Intel vs. 8212 AMD). Lastly, when guest CPUID is set (by 8213 userspace), KVM modifies select VMX MSR 8214 fields to force consistency between guest 8215 CPUID and L2's effective ISA. When this 8216 quirk is disabled, KVM zeroes the vCPU's MSR 8217 values (with two exceptions, see below), 8218 i.e. treats the feature MSRs like CPUID 8219 leaves and gives userspace full control of 8220 the vCPU model definition. This quirk does 8221 not affect VMX MSRs CR0/CR4_FIXED1 (0x487 8222 and 0x489), as KVM does now allow them to 8223 be set by userspace (KVM sets them based on 8224 guest CPUID, for safety purposes). 8225 8226KVM_X86_QUIRK_IGNORE_GUEST_PAT By default, on Intel platforms, KVM ignores 8227 guest PAT and forces the effective memory 8228 type to WB in EPT. The quirk is not available 8229 on Intel platforms which are incapable of 8230 safely honoring guest PAT (i.e., without CPU 8231 self-snoop, KVM always ignores guest PAT and 8232 forces effective memory type to WB). It is 8233 also ignored on AMD platforms or, on Intel, 8234 when a VM has non-coherent DMA devices 8235 assigned; KVM always honors guest PAT in 8236 such case. The quirk is needed to avoid 8237 slowdowns on certain Intel Xeon platforms 8238 (e.g. ICX, SPR) where self-snoop feature is 8239 supported but UC is slow enough to cause 8240 issues with some older guests that use 8241 UC instead of WC to map the video RAM. 8242 Userspace can disable the quirk to honor 8243 guest PAT if it knows that there is no such 8244 guest software, for example if it does not 8245 expose a bochs graphics device (which is 8246 known to have had a buggy driver). 8247=================================== ============================================ 8248 82497.32 KVM_CAP_MAX_VCPU_ID 8250------------------------ 8251 8252:Architectures: x86 8253:Target: VM 8254:Parameters: args[0] - maximum APIC ID value set for current VM 8255:Returns: 0 on success, -EINVAL if args[0] is beyond KVM_MAX_VCPU_IDS 8256 supported in KVM or if it has been set. 8257 8258This capability allows userspace to specify maximum possible APIC ID 8259assigned for current VM session prior to the creation of vCPUs, saving 8260memory for data structures indexed by the APIC ID. Userspace is able 8261to calculate the limit to APIC ID values from designated 8262CPU topology. 8263 8264The value can be changed only until KVM_ENABLE_CAP is set to a nonzero 8265value or until a vCPU is created. Upon creation of the first vCPU, 8266if the value was set to zero or KVM_ENABLE_CAP was not invoked, KVM 8267uses the return value of KVM_CHECK_EXTENSION(KVM_CAP_MAX_VCPU_ID) as 8268the maximum APIC ID. 8269 82707.33 KVM_CAP_X86_NOTIFY_VMEXIT 8271------------------------------ 8272 8273:Architectures: x86 8274:Target: VM 8275:Parameters: args[0] is the value of notify window as well as some flags 8276:Returns: 0 on success, -EINVAL if args[0] contains invalid flags or notify 8277 VM exit is unsupported. 8278 8279Bits 63:32 of args[0] are used for notify window. 8280Bits 31:0 of args[0] are for some flags. Valid bits are:: 8281 8282 #define KVM_X86_NOTIFY_VMEXIT_ENABLED (1 << 0) 8283 #define KVM_X86_NOTIFY_VMEXIT_USER (1 << 1) 8284 8285This capability allows userspace to configure the notify VM exit on/off 8286in per-VM scope during VM creation. Notify VM exit is disabled by default. 8287When userspace sets KVM_X86_NOTIFY_VMEXIT_ENABLED bit in args[0], VMM will 8288enable this feature with the notify window provided, which will generate 8289a VM exit if no event window occurs in VM non-root mode for a specified of 8290time (notify window). 8291 8292If KVM_X86_NOTIFY_VMEXIT_USER is set in args[0], upon notify VM exits happen, 8293KVM would exit to userspace for handling. 8294 8295This capability is aimed to mitigate the threat that malicious VMs can 8296cause CPU stuck (due to event windows don't open up) and make the CPU 8297unavailable to host or other VMs. 8298 82997.35 KVM_CAP_X86_APIC_BUS_CYCLES_NS 8300----------------------------------- 8301 8302:Architectures: x86 8303:Target: VM 8304:Parameters: args[0] is the desired APIC bus clock rate, in nanoseconds 8305:Returns: 0 on success, -EINVAL if args[0] contains an invalid value for the 8306 frequency or if any vCPUs have been created, -ENXIO if a virtual 8307 local APIC has not been created using KVM_CREATE_IRQCHIP. 8308 8309This capability sets the VM's APIC bus clock frequency, used by KVM's in-kernel 8310virtual APIC when emulating APIC timers. KVM's default value can be retrieved 8311by KVM_CHECK_EXTENSION. 8312 8313Note: Userspace is responsible for correctly configuring CPUID 0x15, a.k.a. the 8314core crystal clock frequency, if a non-zero CPUID 0x15 is exposed to the guest. 8315 83167.36 KVM_CAP_DIRTY_LOG_RING/KVM_CAP_DIRTY_LOG_RING_ACQ_REL 8317---------------------------------------------------------- 8318 8319:Architectures: x86, arm64 8320:Type: vm 8321:Parameters: args[0] - size of the dirty log ring 8322 8323KVM is capable of tracking dirty memory using ring buffers that are 8324mmapped into userspace; there is one dirty ring per vcpu. 8325 8326The dirty ring is available to userspace as an array of 8327``struct kvm_dirty_gfn``. Each dirty entry is defined as:: 8328 8329 struct kvm_dirty_gfn { 8330 __u32 flags; 8331 __u32 slot; /* as_id | slot_id */ 8332 __u64 offset; 8333 }; 8334 8335The following values are defined for the flags field to define the 8336current state of the entry:: 8337 8338 #define KVM_DIRTY_GFN_F_DIRTY BIT(0) 8339 #define KVM_DIRTY_GFN_F_RESET BIT(1) 8340 #define KVM_DIRTY_GFN_F_MASK 0x3 8341 8342Userspace should call KVM_ENABLE_CAP ioctl right after KVM_CREATE_VM 8343ioctl to enable this capability for the new guest and set the size of 8344the rings. Enabling the capability is only allowed before creating any 8345vCPU, and the size of the ring must be a power of two. The larger the 8346ring buffer, the less likely the ring is full and the VM is forced to 8347exit to userspace. The optimal size depends on the workload, but it is 8348recommended that it be at least 64 KiB (4096 entries). 8349 8350Just like for dirty page bitmaps, the buffer tracks writes to 8351all user memory regions for which the KVM_MEM_LOG_DIRTY_PAGES flag was 8352set in KVM_SET_USER_MEMORY_REGION. Once a memory region is registered 8353with the flag set, userspace can start harvesting dirty pages from the 8354ring buffer. 8355 8356An entry in the ring buffer can be unused (flag bits ``00``), 8357dirty (flag bits ``01``) or harvested (flag bits ``1X``). The 8358state machine for the entry is as follows:: 8359 8360 dirtied harvested reset 8361 00 -----------> 01 -------------> 1X -------+ 8362 ^ | 8363 | | 8364 +------------------------------------------+ 8365 8366To harvest the dirty pages, userspace accesses the mmapped ring buffer 8367to read the dirty GFNs. If the flags has the DIRTY bit set (at this stage 8368the RESET bit must be cleared), then it means this GFN is a dirty GFN. 8369The userspace should harvest this GFN and mark the flags from state 8370``01b`` to ``1Xb`` (bit 0 will be ignored by KVM, but bit 1 must be set 8371to show that this GFN is harvested and waiting for a reset), and move 8372on to the next GFN. The userspace should continue to do this until the 8373flags of a GFN have the DIRTY bit cleared, meaning that it has harvested 8374all the dirty GFNs that were available. 8375 8376Note that on weakly ordered architectures, userspace accesses to the 8377ring buffer (and more specifically the 'flags' field) must be ordered, 8378using load-acquire/store-release accessors when available, or any 8379other memory barrier that will ensure this ordering. 8380 8381It's not necessary for userspace to harvest the all dirty GFNs at once. 8382However it must collect the dirty GFNs in sequence, i.e., the userspace 8383program cannot skip one dirty GFN to collect the one next to it. 8384 8385After processing one or more entries in the ring buffer, userspace 8386calls the VM ioctl KVM_RESET_DIRTY_RINGS to notify the kernel about 8387it, so that the kernel will reprotect those collected GFNs. 8388Therefore, the ioctl must be called *before* reading the content of 8389the dirty pages. 8390 8391The dirty ring can get full. When it happens, the KVM_RUN of the 8392vcpu will return with exit reason KVM_EXIT_DIRTY_LOG_FULL. 8393 8394The dirty ring interface has a major difference comparing to the 8395KVM_GET_DIRTY_LOG interface in that, when reading the dirty ring from 8396userspace, it's still possible that the kernel has not yet flushed the 8397processor's dirty page buffers into the kernel buffer (with dirty bitmaps, the 8398flushing is done by the KVM_GET_DIRTY_LOG ioctl). To achieve that, one 8399needs to kick the vcpu out of KVM_RUN using a signal. The resulting 8400vmexit ensures that all dirty GFNs are flushed to the dirty rings. 8401 8402NOTE: KVM_CAP_DIRTY_LOG_RING_ACQ_REL is the only capability that 8403should be exposed by weakly ordered architecture, in order to indicate 8404the additional memory ordering requirements imposed on userspace when 8405reading the state of an entry and mutating it from DIRTY to HARVESTED. 8406Architecture with TSO-like ordering (such as x86) are allowed to 8407expose both KVM_CAP_DIRTY_LOG_RING and KVM_CAP_DIRTY_LOG_RING_ACQ_REL 8408to userspace. 8409 8410After enabling the dirty rings, the userspace needs to detect the 8411capability of KVM_CAP_DIRTY_LOG_RING_WITH_BITMAP to see whether the 8412ring structures can be backed by per-slot bitmaps. With this capability 8413advertised, it means the architecture can dirty guest pages without 8414vcpu/ring context, so that some of the dirty information will still be 8415maintained in the bitmap structure. KVM_CAP_DIRTY_LOG_RING_WITH_BITMAP 8416can't be enabled if the capability of KVM_CAP_DIRTY_LOG_RING_ACQ_REL 8417hasn't been enabled, or any memslot has been existing. 8418 8419Note that the bitmap here is only a backup of the ring structure. The 8420use of the ring and bitmap combination is only beneficial if there is 8421only a very small amount of memory that is dirtied out of vcpu/ring 8422context. Otherwise, the stand-alone per-slot bitmap mechanism needs to 8423be considered. 8424 8425To collect dirty bits in the backup bitmap, userspace can use the same 8426KVM_GET_DIRTY_LOG ioctl. KVM_CLEAR_DIRTY_LOG isn't needed as long as all 8427the generation of the dirty bits is done in a single pass. Collecting 8428the dirty bitmap should be the very last thing that the VMM does before 8429considering the state as complete. VMM needs to ensure that the dirty 8430state is final and avoid missing dirty pages from another ioctl ordered 8431after the bitmap collection. 8432 8433NOTE: Multiple examples of using the backup bitmap: (1) save vgic/its 8434tables through command KVM_DEV_ARM_{VGIC_GRP_CTRL, ITS_SAVE_TABLES} on 8435KVM device "kvm-arm-vgic-its". (2) restore vgic/its tables through 8436command KVM_DEV_ARM_{VGIC_GRP_CTRL, ITS_RESTORE_TABLES} on KVM device 8437"kvm-arm-vgic-its". VGICv3 LPI pending status is restored. (3) save 8438vgic3 pending table through KVM_DEV_ARM_VGIC_{GRP_CTRL, SAVE_PENDING_TABLES} 8439command on KVM device "kvm-arm-vgic-v3". 8440 84417.37 KVM_CAP_PMU_CAPABILITY 8442--------------------------- 8443 8444:Architectures: x86 8445:Type: vm 8446:Parameters: arg[0] is bitmask of PMU virtualization capabilities. 8447:Returns: 0 on success, -EINVAL when arg[0] contains invalid bits 8448 8449This capability alters PMU virtualization in KVM. 8450 8451Calling KVM_CHECK_EXTENSION for this capability returns a bitmask of 8452PMU virtualization capabilities that can be adjusted on a VM. 8453 8454The argument to KVM_ENABLE_CAP is also a bitmask and selects specific 8455PMU virtualization capabilities to be applied to the VM. This can 8456only be invoked on a VM prior to the creation of VCPUs. 8457 8458At this time, KVM_PMU_CAP_DISABLE is the only capability. Setting 8459this capability will disable PMU virtualization for that VM. Usermode 8460should adjust CPUID leaf 0xA to reflect that the PMU is disabled. 8461 84627.38 KVM_CAP_VM_DISABLE_NX_HUGE_PAGES 8463------------------------------------- 8464 8465:Architectures: x86 8466:Type: vm 8467:Parameters: arg[0] must be 0. 8468:Returns: 0 on success, -EPERM if the userspace process does not 8469 have CAP_SYS_BOOT, -EINVAL if args[0] is not 0 or any vCPUs have been 8470 created. 8471 8472This capability disables the NX huge pages mitigation for iTLB MULTIHIT. 8473 8474The capability has no effect if the nx_huge_pages module parameter is not set. 8475 8476This capability may only be set before any vCPUs are created. 8477 84787.39 KVM_CAP_ARM_EAGER_SPLIT_CHUNK_SIZE 8479--------------------------------------- 8480 8481:Architectures: arm64 8482:Type: vm 8483:Parameters: arg[0] is the new split chunk size. 8484:Returns: 0 on success, -EINVAL if any memslot was already created. 8485 8486This capability sets the chunk size used in Eager Page Splitting. 8487 8488Eager Page Splitting improves the performance of dirty-logging (used 8489in live migrations) when guest memory is backed by huge-pages. It 8490avoids splitting huge-pages (into PAGE_SIZE pages) on fault, by doing 8491it eagerly when enabling dirty logging (with the 8492KVM_MEM_LOG_DIRTY_PAGES flag for a memory region), or when using 8493KVM_CLEAR_DIRTY_LOG. 8494 8495The chunk size specifies how many pages to break at a time, using a 8496single allocation for each chunk. Bigger the chunk size, more pages 8497need to be allocated ahead of time. 8498 8499The chunk size needs to be a valid block size. The list of acceptable 8500block sizes is exposed in KVM_CAP_ARM_SUPPORTED_BLOCK_SIZES as a 850164-bit bitmap (each bit describing a block size). The default value is 85020, to disable the eager page splitting. 8503 85047.40 KVM_CAP_EXIT_HYPERCALL 8505--------------------------- 8506 8507:Architectures: x86 8508:Type: vm 8509 8510This capability, if enabled, will cause KVM to exit to userspace 8511with KVM_EXIT_HYPERCALL exit reason to process some hypercalls. 8512 8513Calling KVM_CHECK_EXTENSION for this capability will return a bitmask 8514of hypercalls that can be configured to exit to userspace. 8515Right now, the only such hypercall is KVM_HC_MAP_GPA_RANGE. 8516 8517The argument to KVM_ENABLE_CAP is also a bitmask, and must be a subset 8518of the result of KVM_CHECK_EXTENSION. KVM will forward to userspace 8519the hypercalls whose corresponding bit is in the argument, and return 8520ENOSYS for the others. 8521 85227.41 KVM_CAP_ARM_SYSTEM_SUSPEND 8523------------------------------- 8524 8525:Architectures: arm64 8526:Type: vm 8527 8528When enabled, KVM will exit to userspace with KVM_EXIT_SYSTEM_EVENT of 8529type KVM_SYSTEM_EVENT_SUSPEND to process the guest suspend request. 8530 85317.37 KVM_CAP_ARM_WRITABLE_IMP_ID_REGS 8532------------------------------------- 8533 8534:Architectures: arm64 8535:Target: VM 8536:Parameters: None 8537:Returns: 0 on success, -EINVAL if vCPUs have been created before enabling this 8538 capability. 8539 8540This capability changes the behavior of the registers that identify a PE 8541implementation of the Arm architecture: MIDR_EL1, REVIDR_EL1, and AIDR_EL1. 8542By default, these registers are visible to userspace but treated as invariant. 8543 8544When this capability is enabled, KVM allows userspace to change the 8545aforementioned registers before the first KVM_RUN. These registers are VM 8546scoped, meaning that the same set of values are presented on all vCPUs in a 8547given VM. 8548 85497.43 KVM_CAP_RISCV_MP_STATE_RESET 8550--------------------------------- 8551 8552:Architectures: riscv 8553:Type: VM 8554:Parameters: None 8555:Returns: 0 on success, -EINVAL if arg[0] is not zero 8556 8557When this capability is enabled, KVM resets the VCPU when setting 8558MP_STATE_INIT_RECEIVED through IOCTL. The original MP_STATE is preserved. 8559 85608. Other capabilities. 8561====================== 8562 8563This section lists capabilities that give information about other 8564features of the KVM implementation. 8565 85668.1 KVM_CAP_PPC_HWRNG 8567--------------------- 8568 8569:Architectures: ppc 8570 8571This capability, if KVM_CHECK_EXTENSION indicates that it is 8572available, means that the kernel has an implementation of the 8573H_RANDOM hypercall backed by a hardware random-number generator. 8574If present, the kernel H_RANDOM handler can be enabled for guest use 8575with the KVM_CAP_PPC_ENABLE_HCALL capability. 8576 85778.3 KVM_CAP_PPC_MMU_RADIX 8578------------------------- 8579 8580:Architectures: ppc 8581 8582This capability, if KVM_CHECK_EXTENSION indicates that it is 8583available, means that the kernel can support guests using the 8584radix MMU defined in Power ISA V3.00 (as implemented in the POWER9 8585processor). 8586 85878.4 KVM_CAP_PPC_MMU_HASH_V3 8588--------------------------- 8589 8590:Architectures: ppc 8591 8592This capability, if KVM_CHECK_EXTENSION indicates that it is 8593available, means that the kernel can support guests using the 8594hashed page table MMU defined in Power ISA V3.00 (as implemented in 8595the POWER9 processor), including in-memory segment tables. 8596 85978.5 KVM_CAP_MIPS_VZ 8598------------------- 8599 8600:Architectures: mips 8601 8602This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that 8603it is available, means that full hardware assisted virtualization capabilities 8604of the hardware are available for use through KVM. An appropriate 8605KVM_VM_MIPS_* type must be passed to KVM_CREATE_VM to create a VM which 8606utilises it. 8607 8608If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is 8609available, it means that the VM is using full hardware assisted virtualization 8610capabilities of the hardware. This is useful to check after creating a VM with 8611KVM_VM_MIPS_DEFAULT. 8612 8613The value returned by KVM_CHECK_EXTENSION should be compared against known 8614values (see below). All other values are reserved. This is to allow for the 8615possibility of other hardware assisted virtualization implementations which 8616may be incompatible with the MIPS VZ ASE. 8617 8618== ========================================================================== 8619 0 The trap & emulate implementation is in use to run guest code in user 8620 mode. Guest virtual memory segments are rearranged to fit the guest in the 8621 user mode address space. 8622 8623 1 The MIPS VZ ASE is in use, providing full hardware assisted 8624 virtualization, including standard guest virtual memory segments. 8625== ========================================================================== 8626 86278.7 KVM_CAP_MIPS_64BIT 8628---------------------- 8629 8630:Architectures: mips 8631 8632This capability indicates the supported architecture type of the guest, i.e. the 8633supported register and address width. 8634 8635The values returned when this capability is checked by KVM_CHECK_EXTENSION on a 8636kvm VM handle correspond roughly to the CP0_Config.AT register field, and should 8637be checked specifically against known values (see below). All other values are 8638reserved. 8639 8640== ======================================================================== 8641 0 MIPS32 or microMIPS32. 8642 Both registers and addresses are 32-bits wide. 8643 It will only be possible to run 32-bit guest code. 8644 8645 1 MIPS64 or microMIPS64 with access only to 32-bit compatibility segments. 8646 Registers are 64-bits wide, but addresses are 32-bits wide. 8647 64-bit guest code may run but cannot access MIPS64 memory segments. 8648 It will also be possible to run 32-bit guest code. 8649 8650 2 MIPS64 or microMIPS64 with access to all address segments. 8651 Both registers and addresses are 64-bits wide. 8652 It will be possible to run 64-bit or 32-bit guest code. 8653== ======================================================================== 8654 86558.9 KVM_CAP_ARM_USER_IRQ 8656------------------------ 8657 8658:Architectures: arm64 8659 8660This capability, if KVM_CHECK_EXTENSION indicates that it is available, means 8661that if userspace creates a VM without an in-kernel interrupt controller, it 8662will be notified of changes to the output level of in-kernel emulated devices, 8663which can generate virtual interrupts, presented to the VM. 8664For such VMs, on every return to userspace, the kernel 8665updates the vcpu's run->s.regs.device_irq_level field to represent the actual 8666output level of the device. 8667 8668Whenever kvm detects a change in the device output level, kvm guarantees at 8669least one return to userspace before running the VM. This exit could either 8670be a KVM_EXIT_INTR or any other exit event, like KVM_EXIT_MMIO. This way, 8671userspace can always sample the device output level and re-compute the state of 8672the userspace interrupt controller. Userspace should always check the state 8673of run->s.regs.device_irq_level on every kvm exit. 8674The value in run->s.regs.device_irq_level can represent both level and edge 8675triggered interrupt signals, depending on the device. Edge triggered interrupt 8676signals will exit to userspace with the bit in run->s.regs.device_irq_level 8677set exactly once per edge signal. 8678 8679The field run->s.regs.device_irq_level is available independent of 8680run->kvm_valid_regs or run->kvm_dirty_regs bits. 8681 8682If KVM_CAP_ARM_USER_IRQ is supported, the KVM_CHECK_EXTENSION ioctl returns a 8683number larger than 0 indicating the version of this capability is implemented 8684and thereby which bits in run->s.regs.device_irq_level can signal values. 8685 8686Currently the following bits are defined for the device_irq_level bitmap:: 8687 8688 KVM_CAP_ARM_USER_IRQ >= 1: 8689 8690 KVM_ARM_DEV_EL1_VTIMER - EL1 virtual timer 8691 KVM_ARM_DEV_EL1_PTIMER - EL1 physical timer 8692 KVM_ARM_DEV_PMU - ARM PMU overflow interrupt signal 8693 8694Future versions of kvm may implement additional events. These will get 8695indicated by returning a higher number from KVM_CHECK_EXTENSION and will be 8696listed above. 8697 86988.10 KVM_CAP_PPC_SMT_POSSIBLE 8699----------------------------- 8700 8701:Architectures: ppc 8702 8703Querying this capability returns a bitmap indicating the possible 8704virtual SMT modes that can be set using KVM_CAP_PPC_SMT. If bit N 8705(counting from the right) is set, then a virtual SMT mode of 2^N is 8706available. 8707 87088.12 KVM_CAP_HYPERV_VP_INDEX 8709---------------------------- 8710 8711:Architectures: x86 8712 8713This capability indicates that userspace can load HV_X64_MSR_VP_INDEX msr. Its 8714value is used to denote the target vcpu for a SynIC interrupt. For 8715compatibility, KVM initializes this msr to KVM's internal vcpu index. When this 8716capability is absent, userspace can still query this msr's value. 8717 87188.13 KVM_CAP_S390_AIS_MIGRATION 8719------------------------------- 8720 8721:Architectures: s390 8722 8723This capability indicates if the flic device will be able to get/set the 8724AIS states for migration via the KVM_DEV_FLIC_AISM_ALL attribute and allows 8725to discover this without having to create a flic device. 8726 87278.14 KVM_CAP_S390_PSW 8728--------------------- 8729 8730:Architectures: s390 8731 8732This capability indicates that the PSW is exposed via the kvm_run structure. 8733 87348.15 KVM_CAP_S390_GMAP 8735---------------------- 8736 8737:Architectures: s390 8738 8739This capability indicates that the user space memory used as guest mapping can 8740be anywhere in the user memory address space, as long as the memory slots are 8741aligned and sized to a segment (1MB) boundary. 8742 87438.16 KVM_CAP_S390_COW 8744--------------------- 8745 8746:Architectures: s390 8747 8748This capability indicates that the user space memory used as guest mapping can 8749use copy-on-write semantics as well as dirty pages tracking via read-only page 8750tables. 8751 87528.17 KVM_CAP_S390_BPB 8753--------------------- 8754 8755:Architectures: s390 8756 8757This capability indicates that kvm will implement the interfaces to handle 8758reset, migration and nested KVM for branch prediction blocking. The stfle 8759facility 82 should not be provided to the guest without this capability. 8760 87618.18 KVM_CAP_HYPERV_TLBFLUSH 8762---------------------------- 8763 8764:Architectures: x86 8765 8766This capability indicates that KVM supports paravirtualized Hyper-V TLB Flush 8767hypercalls: 8768HvFlushVirtualAddressSpace, HvFlushVirtualAddressSpaceEx, 8769HvFlushVirtualAddressList, HvFlushVirtualAddressListEx. 8770 87718.19 KVM_CAP_ARM_INJECT_SERROR_ESR 8772---------------------------------- 8773 8774:Architectures: arm64 8775 8776This capability indicates that userspace can specify (via the 8777KVM_SET_VCPU_EVENTS ioctl) the syndrome value reported to the guest when it 8778takes a virtual SError interrupt exception. 8779If KVM advertises this capability, userspace can only specify the ISS field for 8780the ESR syndrome. Other parts of the ESR, such as the EC are generated by the 8781CPU when the exception is taken. If this virtual SError is taken to EL1 using 8782AArch64, this value will be reported in the ISS field of ESR_ELx. 8783 8784See KVM_CAP_VCPU_EVENTS for more details. 8785 87868.20 KVM_CAP_HYPERV_SEND_IPI 8787---------------------------- 8788 8789:Architectures: x86 8790 8791This capability indicates that KVM supports paravirtualized Hyper-V IPI send 8792hypercalls: 8793HvCallSendSyntheticClusterIpi, HvCallSendSyntheticClusterIpiEx. 8794 87958.22 KVM_CAP_S390_VCPU_RESETS 8796----------------------------- 8797 8798:Architectures: s390 8799 8800This capability indicates that the KVM_S390_NORMAL_RESET and 8801KVM_S390_CLEAR_RESET ioctls are available. 8802 88038.23 KVM_CAP_S390_PROTECTED 8804--------------------------- 8805 8806:Architectures: s390 8807 8808This capability indicates that the Ultravisor has been initialized and 8809KVM can therefore start protected VMs. 8810This capability governs the KVM_S390_PV_COMMAND ioctl and the 8811KVM_MP_STATE_LOAD MP_STATE. KVM_SET_MP_STATE can fail for protected 8812guests when the state change is invalid. 8813 88148.24 KVM_CAP_STEAL_TIME 8815----------------------- 8816 8817:Architectures: arm64, x86 8818 8819This capability indicates that KVM supports steal time accounting. 8820When steal time accounting is supported it may be enabled with 8821architecture-specific interfaces. This capability and the architecture- 8822specific interfaces must be consistent, i.e. if one says the feature 8823is supported, than the other should as well and vice versa. For arm64 8824see Documentation/virt/kvm/devices/vcpu.rst "KVM_ARM_VCPU_PVTIME_CTRL". 8825For x86 see Documentation/virt/kvm/x86/msr.rst "MSR_KVM_STEAL_TIME". 8826 88278.25 KVM_CAP_S390_DIAG318 8828------------------------- 8829 8830:Architectures: s390 8831 8832This capability enables a guest to set information about its control program 8833(i.e. guest kernel type and version). The information is helpful during 8834system/firmware service events, providing additional data about the guest 8835environments running on the machine. 8836 8837The information is associated with the DIAGNOSE 0x318 instruction, which sets 8838an 8-byte value consisting of a one-byte Control Program Name Code (CPNC) and 8839a 7-byte Control Program Version Code (CPVC). The CPNC determines what 8840environment the control program is running in (e.g. Linux, z/VM...), and the 8841CPVC is used for information specific to OS (e.g. Linux version, Linux 8842distribution...) 8843 8844If this capability is available, then the CPNC and CPVC can be synchronized 8845between KVM and userspace via the sync regs mechanism (KVM_SYNC_DIAG318). 8846 88478.26 KVM_CAP_X86_USER_SPACE_MSR 8848------------------------------- 8849 8850:Architectures: x86 8851 8852This capability indicates that KVM supports deflection of MSR reads and 8853writes to user space. It can be enabled on a VM level. If enabled, MSR 8854accesses that would usually trigger a #GP by KVM into the guest will 8855instead get bounced to user space through the KVM_EXIT_X86_RDMSR and 8856KVM_EXIT_X86_WRMSR exit notifications. 8857 88588.27 KVM_CAP_X86_MSR_FILTER 8859--------------------------- 8860 8861:Architectures: x86 8862 8863This capability indicates that KVM supports that accesses to user defined MSRs 8864may be rejected. With this capability exposed, KVM exports new VM ioctl 8865KVM_X86_SET_MSR_FILTER which user space can call to specify bitmaps of MSR 8866ranges that KVM should deny access to. 8867 8868In combination with KVM_CAP_X86_USER_SPACE_MSR, this allows user space to 8869trap and emulate MSRs that are outside of the scope of KVM as well as 8870limit the attack surface on KVM's MSR emulation code. 8871 88728.30 KVM_CAP_XEN_HVM 8873-------------------- 8874 8875:Architectures: x86 8876 8877This capability indicates the features that Xen supports for hosting Xen 8878PVHVM guests. Valid flags are:: 8879 8880 #define KVM_XEN_HVM_CONFIG_HYPERCALL_MSR (1 << 0) 8881 #define KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL (1 << 1) 8882 #define KVM_XEN_HVM_CONFIG_SHARED_INFO (1 << 2) 8883 #define KVM_XEN_HVM_CONFIG_RUNSTATE (1 << 3) 8884 #define KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL (1 << 4) 8885 #define KVM_XEN_HVM_CONFIG_EVTCHN_SEND (1 << 5) 8886 #define KVM_XEN_HVM_CONFIG_RUNSTATE_UPDATE_FLAG (1 << 6) 8887 #define KVM_XEN_HVM_CONFIG_PVCLOCK_TSC_UNSTABLE (1 << 7) 8888 8889The KVM_XEN_HVM_CONFIG_HYPERCALL_MSR flag indicates that the KVM_XEN_HVM_CONFIG 8890ioctl is available, for the guest to set its hypercall page. 8891 8892If KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL is also set, the same flag may also be 8893provided in the flags to KVM_XEN_HVM_CONFIG, without providing hypercall page 8894contents, to request that KVM generate hypercall page content automatically 8895and also enable interception of guest hypercalls with KVM_EXIT_XEN. 8896 8897The KVM_XEN_HVM_CONFIG_SHARED_INFO flag indicates the availability of the 8898KVM_XEN_HVM_SET_ATTR, KVM_XEN_HVM_GET_ATTR, KVM_XEN_VCPU_SET_ATTR and 8899KVM_XEN_VCPU_GET_ATTR ioctls, as well as the delivery of exception vectors 8900for event channel upcalls when the evtchn_upcall_pending field of a vcpu's 8901vcpu_info is set. 8902 8903The KVM_XEN_HVM_CONFIG_RUNSTATE flag indicates that the runstate-related 8904features KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR/_CURRENT/_DATA/_ADJUST are 8905supported by the KVM_XEN_VCPU_SET_ATTR/KVM_XEN_VCPU_GET_ATTR ioctls. 8906 8907The KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL flag indicates that IRQ routing entries 8908of the type KVM_IRQ_ROUTING_XEN_EVTCHN are supported, with the priority 8909field set to indicate 2 level event channel delivery. 8910 8911The KVM_XEN_HVM_CONFIG_EVTCHN_SEND flag indicates that KVM supports 8912injecting event channel events directly into the guest with the 8913KVM_XEN_HVM_EVTCHN_SEND ioctl. It also indicates support for the 8914KVM_XEN_ATTR_TYPE_EVTCHN/XEN_VERSION HVM attributes and the 8915KVM_XEN_VCPU_ATTR_TYPE_VCPU_ID/TIMER/UPCALL_VECTOR vCPU attributes. 8916related to event channel delivery, timers, and the XENVER_version 8917interception. 8918 8919The KVM_XEN_HVM_CONFIG_RUNSTATE_UPDATE_FLAG flag indicates that KVM supports 8920the KVM_XEN_ATTR_TYPE_RUNSTATE_UPDATE_FLAG attribute in the KVM_XEN_SET_ATTR 8921and KVM_XEN_GET_ATTR ioctls. This controls whether KVM will set the 8922XEN_RUNSTATE_UPDATE flag in guest memory mapped vcpu_runstate_info during 8923updates of the runstate information. Note that versions of KVM which support 8924the RUNSTATE feature above, but not the RUNSTATE_UPDATE_FLAG feature, will 8925always set the XEN_RUNSTATE_UPDATE flag when updating the guest structure, 8926which is perhaps counterintuitive. When this flag is advertised, KVM will 8927behave more correctly, not using the XEN_RUNSTATE_UPDATE flag until/unless 8928specifically enabled (by the guest making the hypercall, causing the VMM 8929to enable the KVM_XEN_ATTR_TYPE_RUNSTATE_UPDATE_FLAG attribute). 8930 8931The KVM_XEN_HVM_CONFIG_PVCLOCK_TSC_UNSTABLE flag indicates that KVM supports 8932clearing the PVCLOCK_TSC_STABLE_BIT flag in Xen pvclock sources. This will be 8933done when the KVM_CAP_XEN_HVM ioctl sets the 8934KVM_XEN_HVM_CONFIG_PVCLOCK_TSC_UNSTABLE flag. 8935 89368.31 KVM_CAP_SPAPR_MULTITCE 8937--------------------------- 8938 8939:Architectures: ppc 8940:Type: vm 8941 8942This capability means the kernel is capable of handling hypercalls 8943H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user 8944space. This significantly accelerates DMA operations for PPC KVM guests. 8945User space should expect that its handlers for these hypercalls 8946are not going to be called if user space previously registered LIOBN 8947in KVM (via KVM_CREATE_SPAPR_TCE or similar calls). 8948 8949In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest, 8950user space might have to advertise it for the guest. For example, 8951IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is 8952present in the "ibm,hypertas-functions" device-tree property. 8953 8954The hypercalls mentioned above may or may not be processed successfully 8955in the kernel based fast path. If they can not be handled by the kernel, 8956they will get passed on to user space. So user space still has to have 8957an implementation for these despite the in kernel acceleration. 8958 8959This capability is always enabled. 8960 89618.32 KVM_CAP_PTP_KVM 8962-------------------- 8963 8964:Architectures: arm64 8965 8966This capability indicates that the KVM virtual PTP service is 8967supported in the host. A VMM can check whether the service is 8968available to the guest on migration. 8969 89708.37 KVM_CAP_S390_PROTECTED_DUMP 8971-------------------------------- 8972 8973:Architectures: s390 8974:Type: vm 8975 8976This capability indicates that KVM and the Ultravisor support dumping 8977PV guests. The `KVM_PV_DUMP` command is available for the 8978`KVM_S390_PV_COMMAND` ioctl and the `KVM_PV_INFO` command provides 8979dump related UV data. Also the vcpu ioctl `KVM_S390_PV_CPU_COMMAND` is 8980available and supports the `KVM_PV_DUMP_CPU` subcommand. 8981 89828.39 KVM_CAP_S390_CPU_TOPOLOGY 8983------------------------------ 8984 8985:Architectures: s390 8986:Type: vm 8987 8988This capability indicates that KVM will provide the S390 CPU Topology 8989facility which consist of the interpretation of the PTF instruction for 8990the function code 2 along with interception and forwarding of both the 8991PTF instruction with function codes 0 or 1 and the STSI(15,1,x) 8992instruction to the userland hypervisor. 8993 8994The stfle facility 11, CPU Topology facility, should not be indicated 8995to the guest without this capability. 8996 8997When this capability is present, KVM provides a new attribute group 8998on vm fd, KVM_S390_VM_CPU_TOPOLOGY. 8999This new attribute allows to get, set or clear the Modified Change 9000Topology Report (MTCR) bit of the SCA through the kvm_device_attr 9001structure. 9002 9003When getting the Modified Change Topology Report value, the attr->addr 9004must point to a byte where the value will be stored or retrieved from. 9005 90068.41 KVM_CAP_VM_TYPES 9007--------------------- 9008 9009:Architectures: x86 9010:Type: system ioctl 9011 9012This capability returns a bitmap of support VM types. The 1-setting of bit @n 9013means the VM type with value @n is supported. Possible values of @n are:: 9014 9015 #define KVM_X86_DEFAULT_VM 0 9016 #define KVM_X86_SW_PROTECTED_VM 1 9017 #define KVM_X86_SEV_VM 2 9018 #define KVM_X86_SEV_ES_VM 3 9019 9020Note, KVM_X86_SW_PROTECTED_VM is currently only for development and testing. 9021Do not use KVM_X86_SW_PROTECTED_VM for "real" VMs, and especially not in 9022production. The behavior and effective ABI for software-protected VMs is 9023unstable. 9024 90258.42 KVM_CAP_PPC_RPT_INVALIDATE 9026------------------------------- 9027 9028:Architectures: ppc 9029 9030This capability indicates that the kernel is capable of handling 9031H_RPT_INVALIDATE hcall. 9032 9033In order to enable the use of H_RPT_INVALIDATE in the guest, 9034user space might have to advertise it for the guest. For example, 9035IBM pSeries (sPAPR) guest starts using it if "hcall-rpt-invalidate" is 9036present in the "ibm,hypertas-functions" device-tree property. 9037 9038This capability is enabled for hypervisors on platforms like POWER9 9039that support radix MMU. 9040 90418.43 KVM_CAP_PPC_AIL_MODE_3 9042--------------------------- 9043 9044:Architectures: ppc 9045 9046This capability indicates that the kernel supports the mode 3 setting for the 9047"Address Translation Mode on Interrupt" aka "Alternate Interrupt Location" 9048resource that is controlled with the H_SET_MODE hypercall. 9049 9050This capability allows a guest kernel to use a better-performance mode for 9051handling interrupts and system calls. 9052 90538.44 KVM_CAP_MEMORY_FAULT_INFO 9054------------------------------ 9055 9056:Architectures: x86 9057 9058The presence of this capability indicates that KVM_RUN will fill 9059kvm_run.memory_fault if KVM cannot resolve a guest page fault VM-Exit, e.g. if 9060there is a valid memslot but no backing VMA for the corresponding host virtual 9061address. 9062 9063The information in kvm_run.memory_fault is valid if and only if KVM_RUN returns 9064an error with errno=EFAULT or errno=EHWPOISON *and* kvm_run.exit_reason is set 9065to KVM_EXIT_MEMORY_FAULT. 9066 9067Note: Userspaces which attempt to resolve memory faults so that they can retry 9068KVM_RUN are encouraged to guard against repeatedly receiving the same 9069error/annotated fault. 9070 9071See KVM_EXIT_MEMORY_FAULT for more information. 9072 90738.45 KVM_CAP_X86_GUEST_MODE 9074--------------------------- 9075 9076:Architectures: x86 9077 9078The presence of this capability indicates that KVM_RUN will update the 9079KVM_RUN_X86_GUEST_MODE bit in kvm_run.flags to indicate whether the 9080vCPU was executing nested guest code when it exited. 9081 9082KVM exits with the register state of either the L1 or L2 guest 9083depending on which executed at the time of an exit. Userspace must 9084take care to differentiate between these cases. 9085 90869. Known KVM API problems 9087========================= 9088 9089In some cases, KVM's API has some inconsistencies or common pitfalls 9090that userspace need to be aware of. This section details some of 9091these issues. 9092 9093Most of them are architecture specific, so the section is split by 9094architecture. 9095 90969.1. x86 9097-------- 9098 9099``KVM_GET_SUPPORTED_CPUID`` issues 9100^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9101 9102In general, ``KVM_GET_SUPPORTED_CPUID`` is designed so that it is possible 9103to take its result and pass it directly to ``KVM_SET_CPUID2``. This section 9104documents some cases in which that requires some care. 9105 9106Local APIC features 9107~~~~~~~~~~~~~~~~~~~ 9108 9109CPU[EAX=1]:ECX[21] (X2APIC) is reported by ``KVM_GET_SUPPORTED_CPUID``, 9110but it can only be enabled if ``KVM_CREATE_IRQCHIP`` or 9111``KVM_ENABLE_CAP(KVM_CAP_IRQCHIP_SPLIT)`` are used to enable in-kernel emulation of 9112the local APIC. 9113 9114The same is true for the ``KVM_FEATURE_PV_UNHALT`` paravirtualized feature. 9115 9116On older versions of Linux, CPU[EAX=1]:ECX[24] (TSC_DEADLINE) is not reported by 9117``KVM_GET_SUPPORTED_CPUID``, but it can be enabled if ``KVM_CAP_TSC_DEADLINE_TIMER`` 9118is present and the kernel has enabled in-kernel emulation of the local APIC. 9119On newer versions, ``KVM_GET_SUPPORTED_CPUID`` does report the bit as available. 9120 9121CPU topology 9122~~~~~~~~~~~~ 9123 9124Several CPUID values include topology information for the host CPU: 91250x0b and 0x1f for Intel systems, 0x8000001e for AMD systems. Different 9126versions of KVM return different values for this information and userspace 9127should not rely on it. Currently they return all zeroes. 9128 9129If userspace wishes to set up a guest topology, it should be careful that 9130the values of these three leaves differ for each CPU. In particular, 9131the APIC ID is found in EDX for all subleaves of 0x0b and 0x1f, and in EAX 9132for 0x8000001e; the latter also encodes the core id and node id in bits 91337:0 of EBX and ECX respectively. 9134 9135Obsolete ioctls and capabilities 9136^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9137 9138KVM_CAP_DISABLE_QUIRKS does not let userspace know which quirks are actually 9139available. Use ``KVM_CHECK_EXTENSION(KVM_CAP_DISABLE_QUIRKS2)`` instead if 9140available. 9141 9142Ordering of KVM_GET_*/KVM_SET_* ioctls 9143^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9144 9145TBD 9146