1.. SPDX-License-Identifier: GPL-2.0 2 3=================================================================== 4The Definitive KVM (Kernel-based Virtual Machine) API Documentation 5=================================================================== 6 71. General description 8====================== 9 10The kvm API is a set of ioctls that are issued to control various aspects 11of a virtual machine. The ioctls belong to the following classes: 12 13 - System ioctls: These query and set global attributes which affect the 14 whole kvm subsystem. In addition a system ioctl is used to create 15 virtual machines. 16 17 - VM ioctls: These query and set attributes that affect an entire virtual 18 machine, for example memory layout. In addition a VM ioctl is used to 19 create virtual cpus (vcpus) and devices. 20 21 VM ioctls must be issued from the same process (address space) that was 22 used to create the VM. 23 24 - vcpu ioctls: These query and set attributes that control the operation 25 of a single virtual cpu. 26 27 vcpu ioctls should be issued from the same thread that was used to create 28 the vcpu, except for asynchronous vcpu ioctl that are marked as such in 29 the documentation. Otherwise, the first ioctl after switching threads 30 could see a performance impact. 31 32 - device ioctls: These query and set attributes that control the operation 33 of a single device. 34 35 device ioctls must be issued from the same process (address space) that 36 was used to create the VM. 37 382. File descriptors 39=================== 40 41The kvm API is centered around file descriptors. An initial 42open("/dev/kvm") obtains a handle to the kvm subsystem; this handle 43can be used to issue system ioctls. A KVM_CREATE_VM ioctl on this 44handle will create a VM file descriptor which can be used to issue VM 45ioctls. A KVM_CREATE_VCPU or KVM_CREATE_DEVICE ioctl on a VM fd will 46create a virtual cpu or device and return a file descriptor pointing to 47the new resource. Finally, ioctls on a vcpu or device fd can be used 48to control the vcpu or device. For vcpus, this includes the important 49task of actually running guest code. 50 51In general file descriptors can be migrated among processes by means 52of fork() and the SCM_RIGHTS facility of unix domain socket. These 53kinds of tricks are explicitly not supported by kvm. While they will 54not cause harm to the host, their actual behavior is not guaranteed by 55the API. See "General description" for details on the ioctl usage 56model that is supported by KVM. 57 58It is important to note that although VM ioctls may only be issued from 59the process that created the VM, a VM's lifecycle is associated with its 60file descriptor, not its creator (process). In other words, the VM and 61its resources, *including the associated address space*, are not freed 62until the last reference to the VM's file descriptor has been released. 63For example, if fork() is issued after ioctl(KVM_CREATE_VM), the VM will 64not be freed until both the parent (original) process and its child have 65put their references to the VM's file descriptor. 66 67Because a VM's resources are not freed until the last reference to its 68file descriptor is released, creating additional references to a VM 69via fork(), dup(), etc... without careful consideration is strongly 70discouraged and may have unwanted side effects, e.g. memory allocated 71by and on behalf of the VM's process may not be freed/unaccounted when 72the VM is shut down. 73 74 753. Extensions 76============= 77 78As of Linux 2.6.22, the KVM ABI has been stabilized: no backward 79incompatible change are allowed. However, there is an extension 80facility that allows backward-compatible extensions to the API to be 81queried and used. 82 83The extension mechanism is not based on the Linux version number. 84Instead, kvm defines extension identifiers and a facility to query 85whether a particular extension identifier is available. If it is, a 86set of ioctls is available for application use. 87 88 894. API description 90================== 91 92This section describes ioctls that can be used to control kvm guests. 93For each ioctl, the following information is provided along with a 94description: 95 96 Capability: 97 which KVM extension provides this ioctl. Can be 'basic', 98 which means that is will be provided by any kernel that supports 99 API version 12 (see section 4.1), a KVM_CAP_xyz constant, which 100 means availability needs to be checked with KVM_CHECK_EXTENSION 101 (see section 4.4), or 'none' which means that while not all kernels 102 support this ioctl, there's no capability bit to check its 103 availability: for kernels that don't support the ioctl, 104 the ioctl returns -ENOTTY. 105 106 Architectures: 107 which instruction set architectures provide this ioctl. 108 x86 includes both i386 and x86_64. 109 110 Type: 111 system, vm, or vcpu. 112 113 Parameters: 114 what parameters are accepted by the ioctl. 115 116 Returns: 117 the return value. General error numbers (EBADF, ENOMEM, EINVAL) 118 are not detailed, but errors with specific meanings are. 119 120 1214.1 KVM_GET_API_VERSION 122----------------------- 123 124:Capability: basic 125:Architectures: all 126:Type: system ioctl 127:Parameters: none 128:Returns: the constant KVM_API_VERSION (=12) 129 130This identifies the API version as the stable kvm API. It is not 131expected that this number will change. However, Linux 2.6.20 and 1322.6.21 report earlier versions; these are not documented and not 133supported. Applications should refuse to run if KVM_GET_API_VERSION 134returns a value other than 12. If this check passes, all ioctls 135described as 'basic' will be available. 136 137 1384.2 KVM_CREATE_VM 139----------------- 140 141:Capability: basic 142:Architectures: all 143:Type: system ioctl 144:Parameters: machine type identifier (KVM_VM_*) 145:Returns: a VM fd that can be used to control the new virtual machine. 146 147The new VM has no virtual cpus and no memory. 148You probably want to use 0 as machine type. 149 150X86: 151^^^^ 152 153Supported X86 VM types can be queried via KVM_CAP_VM_TYPES. 154 155S390: 156^^^^^ 157 158In order to create user controlled virtual machines on S390, check 159KVM_CAP_S390_UCONTROL and use the flag KVM_VM_S390_UCONTROL as 160privileged user (CAP_SYS_ADMIN). 161 162MIPS: 163^^^^^ 164 165To use hardware assisted virtualization on MIPS (VZ ASE) rather than 166the default trap & emulate implementation (which changes the virtual 167memory layout to fit in user mode), check KVM_CAP_MIPS_VZ and use the 168flag KVM_VM_MIPS_VZ. 169 170ARM64: 171^^^^^^ 172 173On arm64, the physical address size for a VM (IPA Size limit) is limited 174to 40bits by default. The limit can be configured if the host supports the 175extension KVM_CAP_ARM_VM_IPA_SIZE. When supported, use 176KVM_VM_TYPE_ARM_IPA_SIZE(IPA_Bits) to set the size in the machine type 177identifier, where IPA_Bits is the maximum width of any physical 178address used by the VM. The IPA_Bits is encoded in bits[7-0] of the 179machine type identifier. 180 181e.g, to configure a guest to use 48bit physical address size:: 182 183 vm_fd = ioctl(dev_fd, KVM_CREATE_VM, KVM_VM_TYPE_ARM_IPA_SIZE(48)); 184 185The requested size (IPA_Bits) must be: 186 187 == ========================================================= 188 0 Implies default size, 40bits (for backward compatibility) 189 N Implies N bits, where N is a positive integer such that, 190 32 <= N <= Host_IPA_Limit 191 == ========================================================= 192 193Host_IPA_Limit is the maximum possible value for IPA_Bits on the host and 194is dependent on the CPU capability and the kernel configuration. The limit can 195be retrieved using KVM_CAP_ARM_VM_IPA_SIZE of the KVM_CHECK_EXTENSION 196ioctl() at run-time. 197 198Creation of the VM will fail if the requested IPA size (whether it is 199implicit or explicit) is unsupported on the host. 200 201Please note that configuring the IPA size does not affect the capability 202exposed by the guest CPUs in ID_AA64MMFR0_EL1[PARange]. It only affects 203size of the address translated by the stage2 level (guest physical to 204host physical address translations). 205 206 2074.3 KVM_GET_MSR_INDEX_LIST, KVM_GET_MSR_FEATURE_INDEX_LIST 208---------------------------------------------------------- 209 210:Capability: basic, KVM_CAP_GET_MSR_FEATURES for KVM_GET_MSR_FEATURE_INDEX_LIST 211:Architectures: x86 212:Type: system ioctl 213:Parameters: struct kvm_msr_list (in/out) 214:Returns: 0 on success; -1 on error 215 216Errors: 217 218 ====== ============================================================ 219 EFAULT the msr index list cannot be read from or written to 220 E2BIG the msr index list is too big to fit in the array specified by 221 the user. 222 ====== ============================================================ 223 224:: 225 226 struct kvm_msr_list { 227 __u32 nmsrs; /* number of msrs in entries */ 228 __u32 indices[0]; 229 }; 230 231The user fills in the size of the indices array in nmsrs, and in return 232kvm adjusts nmsrs to reflect the actual number of msrs and fills in the 233indices array with their numbers. 234 235KVM_GET_MSR_INDEX_LIST returns the guest msrs that are supported. The list 236varies by kvm version and host processor, but does not change otherwise. 237 238Note: if kvm indicates supports MCE (KVM_CAP_MCE), then the MCE bank MSRs are 239not returned in the MSR list, as different vcpus can have a different number 240of banks, as set via the KVM_X86_SETUP_MCE ioctl. 241 242KVM_GET_MSR_FEATURE_INDEX_LIST returns the list of MSRs that can be passed 243to the KVM_GET_MSRS system ioctl. This lets userspace probe host capabilities 244and processor features that are exposed via MSRs (e.g., VMX capabilities). 245This list also varies by kvm version and host processor, but does not change 246otherwise. 247 248 2494.4 KVM_CHECK_EXTENSION 250----------------------- 251 252:Capability: basic, KVM_CAP_CHECK_EXTENSION_VM for vm ioctl 253:Architectures: all 254:Type: system ioctl, vm ioctl 255:Parameters: extension identifier (KVM_CAP_*) 256:Returns: 0 if unsupported; 1 (or some other positive integer) if supported 257 258The API allows the application to query about extensions to the core 259kvm API. Userspace passes an extension identifier (an integer) and 260receives an integer that describes the extension availability. 261Generally 0 means no and 1 means yes, but some extensions may report 262additional information in the integer return value. 263 264Based on their initialization different VMs may have different capabilities. 265It is thus encouraged to use the vm ioctl to query for capabilities (available 266with KVM_CAP_CHECK_EXTENSION_VM on the vm fd) 267 2684.5 KVM_GET_VCPU_MMAP_SIZE 269-------------------------- 270 271:Capability: basic 272:Architectures: all 273:Type: system ioctl 274:Parameters: none 275:Returns: size of vcpu mmap area, in bytes 276 277The KVM_RUN ioctl (cf.) communicates with userspace via a shared 278memory region. This ioctl returns the size of that region. See the 279KVM_RUN documentation for details. 280 281Besides the size of the KVM_RUN communication region, other areas of 282the VCPU file descriptor can be mmap-ed, including: 283 284- if KVM_CAP_COALESCED_MMIO is available, a page at 285 KVM_COALESCED_MMIO_PAGE_OFFSET * PAGE_SIZE; for historical reasons, 286 this page is included in the result of KVM_GET_VCPU_MMAP_SIZE. 287 KVM_CAP_COALESCED_MMIO is not documented yet. 288 289- if KVM_CAP_DIRTY_LOG_RING is available, a number of pages at 290 KVM_DIRTY_LOG_PAGE_OFFSET * PAGE_SIZE. For more information on 291 KVM_CAP_DIRTY_LOG_RING, see section 8.3. 292 293 2944.7 KVM_CREATE_VCPU 295------------------- 296 297:Capability: basic 298:Architectures: all 299:Type: vm ioctl 300:Parameters: vcpu id (apic id on x86) 301:Returns: vcpu fd on success, -1 on error 302 303This API adds a vcpu to a virtual machine. No more than max_vcpus may be added. 304The vcpu id is an integer in the range [0, max_vcpu_id). 305 306The recommended max_vcpus value can be retrieved using the KVM_CAP_NR_VCPUS of 307the KVM_CHECK_EXTENSION ioctl() at run-time. 308The maximum possible value for max_vcpus can be retrieved using the 309KVM_CAP_MAX_VCPUS of the KVM_CHECK_EXTENSION ioctl() at run-time. 310 311If the KVM_CAP_NR_VCPUS does not exist, you should assume that max_vcpus is 4 312cpus max. 313If the KVM_CAP_MAX_VCPUS does not exist, you should assume that max_vcpus is 314same as the value returned from KVM_CAP_NR_VCPUS. 315 316The maximum possible value for max_vcpu_id can be retrieved using the 317KVM_CAP_MAX_VCPU_ID of the KVM_CHECK_EXTENSION ioctl() at run-time. 318 319If the KVM_CAP_MAX_VCPU_ID does not exist, you should assume that max_vcpu_id 320is the same as the value returned from KVM_CAP_MAX_VCPUS. 321 322On powerpc using book3s_hv mode, the vcpus are mapped onto virtual 323threads in one or more virtual CPU cores. (This is because the 324hardware requires all the hardware threads in a CPU core to be in the 325same partition.) The KVM_CAP_PPC_SMT capability indicates the number 326of vcpus per virtual core (vcore). The vcore id is obtained by 327dividing the vcpu id by the number of vcpus per vcore. The vcpus in a 328given vcore will always be in the same physical core as each other 329(though that might be a different physical core from time to time). 330Userspace can control the threading (SMT) mode of the guest by its 331allocation of vcpu ids. For example, if userspace wants 332single-threaded guest vcpus, it should make all vcpu ids be a multiple 333of the number of vcpus per vcore. 334 335For virtual cpus that have been created with S390 user controlled virtual 336machines, the resulting vcpu fd can be memory mapped at page offset 337KVM_S390_SIE_PAGE_OFFSET in order to obtain a memory map of the virtual 338cpu's hardware control block. 339 340 3414.8 KVM_GET_DIRTY_LOG (vm ioctl) 342-------------------------------- 343 344:Capability: basic 345:Architectures: all 346:Type: vm ioctl 347:Parameters: struct kvm_dirty_log (in/out) 348:Returns: 0 on success, -1 on error 349 350:: 351 352 /* for KVM_GET_DIRTY_LOG */ 353 struct kvm_dirty_log { 354 __u32 slot; 355 __u32 padding; 356 union { 357 void __user *dirty_bitmap; /* one bit per page */ 358 __u64 padding; 359 }; 360 }; 361 362Given a memory slot, return a bitmap containing any pages dirtied 363since the last call to this ioctl. Bit 0 is the first page in the 364memory slot. Ensure the entire structure is cleared to avoid padding 365issues. 366 367If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies 368the address space for which you want to return the dirty bitmap. See 369KVM_SET_USER_MEMORY_REGION for details on the usage of slot field. 370 371The bits in the dirty bitmap are cleared before the ioctl returns, unless 372KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is enabled. For more information, 373see the description of the capability. 374 375Note that the Xen shared_info page, if configured, shall always be assumed 376to be dirty. KVM will not explicitly mark it such. 377 378 3794.10 KVM_RUN 380------------ 381 382:Capability: basic 383:Architectures: all 384:Type: vcpu ioctl 385:Parameters: none 386:Returns: 0 on success, -1 on error 387 388Errors: 389 390 ======= ============================================================== 391 EINTR an unmasked signal is pending 392 ENOEXEC the vcpu hasn't been initialized or the guest tried to execute 393 instructions from device memory (arm64) 394 ENOSYS data abort outside memslots with no syndrome info and 395 KVM_CAP_ARM_NISV_TO_USER not enabled (arm64) 396 EPERM SVE feature set but not finalized (arm64) 397 ======= ============================================================== 398 399This ioctl is used to run a guest virtual cpu. While there are no 400explicit parameters, there is an implicit parameter block that can be 401obtained by mmap()ing the vcpu fd at offset 0, with the size given by 402KVM_GET_VCPU_MMAP_SIZE. The parameter block is formatted as a 'struct 403kvm_run' (see below). 404 405 4064.11 KVM_GET_REGS 407----------------- 408 409:Capability: basic 410:Architectures: all except arm64 411:Type: vcpu ioctl 412:Parameters: struct kvm_regs (out) 413:Returns: 0 on success, -1 on error 414 415Reads the general purpose registers from the vcpu. 416 417:: 418 419 /* x86 */ 420 struct kvm_regs { 421 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */ 422 __u64 rax, rbx, rcx, rdx; 423 __u64 rsi, rdi, rsp, rbp; 424 __u64 r8, r9, r10, r11; 425 __u64 r12, r13, r14, r15; 426 __u64 rip, rflags; 427 }; 428 429 /* mips */ 430 struct kvm_regs { 431 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */ 432 __u64 gpr[32]; 433 __u64 hi; 434 __u64 lo; 435 __u64 pc; 436 }; 437 438 /* LoongArch */ 439 struct kvm_regs { 440 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */ 441 unsigned long gpr[32]; 442 unsigned long pc; 443 }; 444 445 4464.12 KVM_SET_REGS 447----------------- 448 449:Capability: basic 450:Architectures: all except arm64 451:Type: vcpu ioctl 452:Parameters: struct kvm_regs (in) 453:Returns: 0 on success, -1 on error 454 455Writes the general purpose registers into the vcpu. 456 457See KVM_GET_REGS for the data structure. 458 459 4604.13 KVM_GET_SREGS 461------------------ 462 463:Capability: basic 464:Architectures: x86, ppc 465:Type: vcpu ioctl 466:Parameters: struct kvm_sregs (out) 467:Returns: 0 on success, -1 on error 468 469Reads special registers from the vcpu. 470 471:: 472 473 /* x86 */ 474 struct kvm_sregs { 475 struct kvm_segment cs, ds, es, fs, gs, ss; 476 struct kvm_segment tr, ldt; 477 struct kvm_dtable gdt, idt; 478 __u64 cr0, cr2, cr3, cr4, cr8; 479 __u64 efer; 480 __u64 apic_base; 481 __u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64]; 482 }; 483 484 /* ppc -- see arch/powerpc/include/uapi/asm/kvm.h */ 485 486interrupt_bitmap is a bitmap of pending external interrupts. At most 487one bit may be set. This interrupt has been acknowledged by the APIC 488but not yet injected into the cpu core. 489 490 4914.14 KVM_SET_SREGS 492------------------ 493 494:Capability: basic 495:Architectures: x86, ppc 496:Type: vcpu ioctl 497:Parameters: struct kvm_sregs (in) 498:Returns: 0 on success, -1 on error 499 500Writes special registers into the vcpu. See KVM_GET_SREGS for the 501data structures. 502 503 5044.15 KVM_TRANSLATE 505------------------ 506 507:Capability: basic 508:Architectures: x86 509:Type: vcpu ioctl 510:Parameters: struct kvm_translation (in/out) 511:Returns: 0 on success, -1 on error 512 513Translates a virtual address according to the vcpu's current address 514translation mode. 515 516:: 517 518 struct kvm_translation { 519 /* in */ 520 __u64 linear_address; 521 522 /* out */ 523 __u64 physical_address; 524 __u8 valid; 525 __u8 writeable; 526 __u8 usermode; 527 __u8 pad[5]; 528 }; 529 530 5314.16 KVM_INTERRUPT 532------------------ 533 534:Capability: basic 535:Architectures: x86, ppc, mips, riscv, loongarch 536:Type: vcpu ioctl 537:Parameters: struct kvm_interrupt (in) 538:Returns: 0 on success, negative on failure. 539 540Queues a hardware interrupt vector to be injected. 541 542:: 543 544 /* for KVM_INTERRUPT */ 545 struct kvm_interrupt { 546 /* in */ 547 __u32 irq; 548 }; 549 550X86: 551^^^^ 552 553:Returns: 554 555 ========= =================================== 556 0 on success, 557 -EEXIST if an interrupt is already enqueued 558 -EINVAL the irq number is invalid 559 -ENXIO if the PIC is in the kernel 560 -EFAULT if the pointer is invalid 561 ========= =================================== 562 563Note 'irq' is an interrupt vector, not an interrupt pin or line. This 564ioctl is useful if the in-kernel PIC is not used. 565 566PPC: 567^^^^ 568 569Queues an external interrupt to be injected. This ioctl is overloaded 570with 3 different irq values: 571 572a) KVM_INTERRUPT_SET 573 574 This injects an edge type external interrupt into the guest once it's ready 575 to receive interrupts. When injected, the interrupt is done. 576 577b) KVM_INTERRUPT_UNSET 578 579 This unsets any pending interrupt. 580 581 Only available with KVM_CAP_PPC_UNSET_IRQ. 582 583c) KVM_INTERRUPT_SET_LEVEL 584 585 This injects a level type external interrupt into the guest context. The 586 interrupt stays pending until a specific ioctl with KVM_INTERRUPT_UNSET 587 is triggered. 588 589 Only available with KVM_CAP_PPC_IRQ_LEVEL. 590 591Note that any value for 'irq' other than the ones stated above is invalid 592and incurs unexpected behavior. 593 594This is an asynchronous vcpu ioctl and can be invoked from any thread. 595 596MIPS: 597^^^^^ 598 599Queues an external interrupt to be injected into the virtual CPU. A negative 600interrupt number dequeues the interrupt. 601 602This is an asynchronous vcpu ioctl and can be invoked from any thread. 603 604RISC-V: 605^^^^^^^ 606 607Queues an external interrupt to be injected into the virtual CPU. This ioctl 608is overloaded with 2 different irq values: 609 610a) KVM_INTERRUPT_SET 611 612 This sets external interrupt for a virtual CPU and it will receive 613 once it is ready. 614 615b) KVM_INTERRUPT_UNSET 616 617 This clears pending external interrupt for a virtual CPU. 618 619This is an asynchronous vcpu ioctl and can be invoked from any thread. 620 621LOONGARCH: 622^^^^^^^^^^ 623 624Queues an external interrupt to be injected into the virtual CPU. A negative 625interrupt number dequeues the interrupt. 626 627This is an asynchronous vcpu ioctl and can be invoked from any thread. 628 629 6304.18 KVM_GET_MSRS 631----------------- 632 633:Capability: basic (vcpu), KVM_CAP_GET_MSR_FEATURES (system) 634:Architectures: x86 635:Type: system ioctl, vcpu ioctl 636:Parameters: struct kvm_msrs (in/out) 637:Returns: number of msrs successfully returned; 638 -1 on error 639 640When used as a system ioctl: 641Reads the values of MSR-based features that are available for the VM. This 642is similar to KVM_GET_SUPPORTED_CPUID, but it returns MSR indices and values. 643The list of msr-based features can be obtained using KVM_GET_MSR_FEATURE_INDEX_LIST 644in a system ioctl. 645 646When used as a vcpu ioctl: 647Reads model-specific registers from the vcpu. Supported msr indices can 648be obtained using KVM_GET_MSR_INDEX_LIST in a system ioctl. 649 650:: 651 652 struct kvm_msrs { 653 __u32 nmsrs; /* number of msrs in entries */ 654 __u32 pad; 655 656 struct kvm_msr_entry entries[0]; 657 }; 658 659 struct kvm_msr_entry { 660 __u32 index; 661 __u32 reserved; 662 __u64 data; 663 }; 664 665Application code should set the 'nmsrs' member (which indicates the 666size of the entries array) and the 'index' member of each array entry. 667kvm will fill in the 'data' member. 668 669 6704.19 KVM_SET_MSRS 671----------------- 672 673:Capability: basic 674:Architectures: x86 675:Type: vcpu ioctl 676:Parameters: struct kvm_msrs (in) 677:Returns: number of msrs successfully set (see below), -1 on error 678 679Writes model-specific registers to the vcpu. See KVM_GET_MSRS for the 680data structures. 681 682Application code should set the 'nmsrs' member (which indicates the 683size of the entries array), and the 'index' and 'data' members of each 684array entry. 685 686It tries to set the MSRs in array entries[] one by one. If setting an MSR 687fails, e.g., due to setting reserved bits, the MSR isn't supported/emulated 688by KVM, etc..., it stops processing the MSR list and returns the number of 689MSRs that have been set successfully. 690 691 6924.20 KVM_SET_CPUID 693------------------ 694 695:Capability: basic 696:Architectures: x86 697:Type: vcpu ioctl 698:Parameters: struct kvm_cpuid (in) 699:Returns: 0 on success, -1 on error 700 701Defines the vcpu responses to the cpuid instruction. Applications 702should use the KVM_SET_CPUID2 ioctl if available. 703 704Caveat emptor: 705 - If this IOCTL fails, KVM gives no guarantees that previous valid CPUID 706 configuration (if there is) is not corrupted. Userspace can get a copy 707 of the resulting CPUID configuration through KVM_GET_CPUID2 in case. 708 - Using KVM_SET_CPUID{,2} after KVM_RUN, i.e. changing the guest vCPU model 709 after running the guest, may cause guest instability. 710 - Using heterogeneous CPUID configurations, modulo APIC IDs, topology, etc... 711 may cause guest instability. 712 713:: 714 715 struct kvm_cpuid_entry { 716 __u32 function; 717 __u32 eax; 718 __u32 ebx; 719 __u32 ecx; 720 __u32 edx; 721 __u32 padding; 722 }; 723 724 /* for KVM_SET_CPUID */ 725 struct kvm_cpuid { 726 __u32 nent; 727 __u32 padding; 728 struct kvm_cpuid_entry entries[0]; 729 }; 730 731 7324.21 KVM_SET_SIGNAL_MASK 733------------------------ 734 735:Capability: basic 736:Architectures: all 737:Type: vcpu ioctl 738:Parameters: struct kvm_signal_mask (in) 739:Returns: 0 on success, -1 on error 740 741Defines which signals are blocked during execution of KVM_RUN. This 742signal mask temporarily overrides the threads signal mask. Any 743unblocked signal received (except SIGKILL and SIGSTOP, which retain 744their traditional behaviour) will cause KVM_RUN to return with -EINTR. 745 746Note the signal will only be delivered if not blocked by the original 747signal mask. 748 749:: 750 751 /* for KVM_SET_SIGNAL_MASK */ 752 struct kvm_signal_mask { 753 __u32 len; 754 __u8 sigset[0]; 755 }; 756 757 7584.22 KVM_GET_FPU 759---------------- 760 761:Capability: basic 762:Architectures: x86, loongarch 763:Type: vcpu ioctl 764:Parameters: struct kvm_fpu (out) 765:Returns: 0 on success, -1 on error 766 767Reads the floating point state from the vcpu. 768 769:: 770 771 /* x86: for KVM_GET_FPU and KVM_SET_FPU */ 772 struct kvm_fpu { 773 __u8 fpr[8][16]; 774 __u16 fcw; 775 __u16 fsw; 776 __u8 ftwx; /* in fxsave format */ 777 __u8 pad1; 778 __u16 last_opcode; 779 __u64 last_ip; 780 __u64 last_dp; 781 __u8 xmm[16][16]; 782 __u32 mxcsr; 783 __u32 pad2; 784 }; 785 786 /* LoongArch: for KVM_GET_FPU and KVM_SET_FPU */ 787 struct kvm_fpu { 788 __u32 fcsr; 789 __u64 fcc; 790 struct kvm_fpureg { 791 __u64 val64[4]; 792 }fpr[32]; 793 }; 794 795 7964.23 KVM_SET_FPU 797---------------- 798 799:Capability: basic 800:Architectures: x86, loongarch 801:Type: vcpu ioctl 802:Parameters: struct kvm_fpu (in) 803:Returns: 0 on success, -1 on error 804 805Writes the floating point state to the vcpu. 806 807:: 808 809 /* x86: for KVM_GET_FPU and KVM_SET_FPU */ 810 struct kvm_fpu { 811 __u8 fpr[8][16]; 812 __u16 fcw; 813 __u16 fsw; 814 __u8 ftwx; /* in fxsave format */ 815 __u8 pad1; 816 __u16 last_opcode; 817 __u64 last_ip; 818 __u64 last_dp; 819 __u8 xmm[16][16]; 820 __u32 mxcsr; 821 __u32 pad2; 822 }; 823 824 /* LoongArch: for KVM_GET_FPU and KVM_SET_FPU */ 825 struct kvm_fpu { 826 __u32 fcsr; 827 __u64 fcc; 828 struct kvm_fpureg { 829 __u64 val64[4]; 830 }fpr[32]; 831 }; 832 833 8344.24 KVM_CREATE_IRQCHIP 835----------------------- 836 837:Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQCHIP (s390) 838:Architectures: x86, arm64, s390 839:Type: vm ioctl 840:Parameters: none 841:Returns: 0 on success, -1 on error 842 843Creates an interrupt controller model in the kernel. 844On x86, creates a virtual ioapic, a virtual PIC (two PICs, nested), and sets up 845future vcpus to have a local APIC. IRQ routing for GSIs 0-15 is set to both 846PIC and IOAPIC; GSI 16-23 only go to the IOAPIC. 847On arm64, a GICv2 is created. Any other GIC versions require the usage of 848KVM_CREATE_DEVICE, which also supports creating a GICv2. Using 849KVM_CREATE_DEVICE is preferred over KVM_CREATE_IRQCHIP for GICv2. 850On s390, a dummy irq routing table is created. 851 852Note that on s390 the KVM_CAP_S390_IRQCHIP vm capability needs to be enabled 853before KVM_CREATE_IRQCHIP can be used. 854 855 8564.25 KVM_IRQ_LINE 857----------------- 858 859:Capability: KVM_CAP_IRQCHIP 860:Architectures: x86, arm64 861:Type: vm ioctl 862:Parameters: struct kvm_irq_level 863:Returns: 0 on success, -1 on error 864 865Sets the level of a GSI input to the interrupt controller model in the kernel. 866On some architectures it is required that an interrupt controller model has 867been previously created with KVM_CREATE_IRQCHIP. Note that edge-triggered 868interrupts require the level to be set to 1 and then back to 0. 869 870On real hardware, interrupt pins can be active-low or active-high. This 871does not matter for the level field of struct kvm_irq_level: 1 always 872means active (asserted), 0 means inactive (deasserted). 873 874x86 allows the operating system to program the interrupt polarity 875(active-low/active-high) for level-triggered interrupts, and KVM used 876to consider the polarity. However, due to bitrot in the handling of 877active-low interrupts, the above convention is now valid on x86 too. 878This is signaled by KVM_CAP_X86_IOAPIC_POLARITY_IGNORED. Userspace 879should not present interrupts to the guest as active-low unless this 880capability is present (or unless it is not using the in-kernel irqchip, 881of course). 882 883 884arm64 can signal an interrupt either at the CPU level, or at the 885in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to 886use PPIs designated for specific cpus. The irq field is interpreted 887like this:: 888 889 bits: | 31 ... 28 | 27 ... 24 | 23 ... 16 | 15 ... 0 | 890 field: | vcpu2_index | irq_type | vcpu_index | irq_id | 891 892The irq_type field has the following values: 893 894- KVM_ARM_IRQ_TYPE_CPU: 895 out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ 896- KVM_ARM_IRQ_TYPE_SPI: 897 in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.) 898 (the vcpu_index field is ignored) 899- KVM_ARM_IRQ_TYPE_PPI: 900 in-kernel GIC: PPI, irq_id between 16 and 31 (incl.) 901 902(The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs) 903 904In both cases, level is used to assert/deassert the line. 905 906When KVM_CAP_ARM_IRQ_LINE_LAYOUT_2 is supported, the target vcpu is 907identified as (256 * vcpu2_index + vcpu_index). Otherwise, vcpu2_index 908must be zero. 909 910Note that on arm64, the KVM_CAP_IRQCHIP capability only conditions 911injection of interrupts for the in-kernel irqchip. KVM_IRQ_LINE can always 912be used for a userspace interrupt controller. 913 914:: 915 916 struct kvm_irq_level { 917 union { 918 __u32 irq; /* GSI */ 919 __s32 status; /* not used for KVM_IRQ_LEVEL */ 920 }; 921 __u32 level; /* 0 or 1 */ 922 }; 923 924 9254.26 KVM_GET_IRQCHIP 926-------------------- 927 928:Capability: KVM_CAP_IRQCHIP 929:Architectures: x86 930:Type: vm ioctl 931:Parameters: struct kvm_irqchip (in/out) 932:Returns: 0 on success, -1 on error 933 934Reads the state of a kernel interrupt controller created with 935KVM_CREATE_IRQCHIP into a buffer provided by the caller. 936 937:: 938 939 struct kvm_irqchip { 940 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */ 941 __u32 pad; 942 union { 943 char dummy[512]; /* reserving space */ 944 struct kvm_pic_state pic; 945 struct kvm_ioapic_state ioapic; 946 } chip; 947 }; 948 949 9504.27 KVM_SET_IRQCHIP 951-------------------- 952 953:Capability: KVM_CAP_IRQCHIP 954:Architectures: x86 955:Type: vm ioctl 956:Parameters: struct kvm_irqchip (in) 957:Returns: 0 on success, -1 on error 958 959Sets the state of a kernel interrupt controller created with 960KVM_CREATE_IRQCHIP from a buffer provided by the caller. 961 962:: 963 964 struct kvm_irqchip { 965 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */ 966 __u32 pad; 967 union { 968 char dummy[512]; /* reserving space */ 969 struct kvm_pic_state pic; 970 struct kvm_ioapic_state ioapic; 971 } chip; 972 }; 973 974 9754.28 KVM_XEN_HVM_CONFIG 976----------------------- 977 978:Capability: KVM_CAP_XEN_HVM 979:Architectures: x86 980:Type: vm ioctl 981:Parameters: struct kvm_xen_hvm_config (in) 982:Returns: 0 on success, -1 on error 983 984Sets the MSR that the Xen HVM guest uses to initialize its hypercall 985page, and provides the starting address and size of the hypercall 986blobs in userspace. When the guest writes the MSR, kvm copies one 987page of a blob (32- or 64-bit, depending on the vcpu mode) to guest 988memory. 989 990:: 991 992 struct kvm_xen_hvm_config { 993 __u32 flags; 994 __u32 msr; 995 __u64 blob_addr_32; 996 __u64 blob_addr_64; 997 __u8 blob_size_32; 998 __u8 blob_size_64; 999 __u8 pad2[30]; 1000 }; 1001 1002If certain flags are returned from the KVM_CAP_XEN_HVM check, they may 1003be set in the flags field of this ioctl: 1004 1005The KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL flag requests KVM to generate 1006the contents of the hypercall page automatically; hypercalls will be 1007intercepted and passed to userspace through KVM_EXIT_XEN. In this 1008case, all of the blob size and address fields must be zero. 1009 1010The KVM_XEN_HVM_CONFIG_EVTCHN_SEND flag indicates to KVM that userspace 1011will always use the KVM_XEN_HVM_EVTCHN_SEND ioctl to deliver event 1012channel interrupts rather than manipulating the guest's shared_info 1013structures directly. This, in turn, may allow KVM to enable features 1014such as intercepting the SCHEDOP_poll hypercall to accelerate PV 1015spinlock operation for the guest. Userspace may still use the ioctl 1016to deliver events if it was advertised, even if userspace does not 1017send this indication that it will always do so 1018 1019No other flags are currently valid in the struct kvm_xen_hvm_config. 1020 10214.29 KVM_GET_CLOCK 1022------------------ 1023 1024:Capability: KVM_CAP_ADJUST_CLOCK 1025:Architectures: x86 1026:Type: vm ioctl 1027:Parameters: struct kvm_clock_data (out) 1028:Returns: 0 on success, -1 on error 1029 1030Gets the current timestamp of kvmclock as seen by the current guest. In 1031conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios 1032such as migration. 1033 1034When KVM_CAP_ADJUST_CLOCK is passed to KVM_CHECK_EXTENSION, it returns the 1035set of bits that KVM can return in struct kvm_clock_data's flag member. 1036 1037The following flags are defined: 1038 1039KVM_CLOCK_TSC_STABLE 1040 If set, the returned value is the exact kvmclock 1041 value seen by all VCPUs at the instant when KVM_GET_CLOCK was called. 1042 If clear, the returned value is simply CLOCK_MONOTONIC plus a constant 1043 offset; the offset can be modified with KVM_SET_CLOCK. KVM will try 1044 to make all VCPUs follow this clock, but the exact value read by each 1045 VCPU could differ, because the host TSC is not stable. 1046 1047KVM_CLOCK_REALTIME 1048 If set, the `realtime` field in the kvm_clock_data 1049 structure is populated with the value of the host's real time 1050 clocksource at the instant when KVM_GET_CLOCK was called. If clear, 1051 the `realtime` field does not contain a value. 1052 1053KVM_CLOCK_HOST_TSC 1054 If set, the `host_tsc` field in the kvm_clock_data 1055 structure is populated with the value of the host's timestamp counter (TSC) 1056 at the instant when KVM_GET_CLOCK was called. If clear, the `host_tsc` field 1057 does not contain a value. 1058 1059:: 1060 1061 struct kvm_clock_data { 1062 __u64 clock; /* kvmclock current value */ 1063 __u32 flags; 1064 __u32 pad0; 1065 __u64 realtime; 1066 __u64 host_tsc; 1067 __u32 pad[4]; 1068 }; 1069 1070 10714.30 KVM_SET_CLOCK 1072------------------ 1073 1074:Capability: KVM_CAP_ADJUST_CLOCK 1075:Architectures: x86 1076:Type: vm ioctl 1077:Parameters: struct kvm_clock_data (in) 1078:Returns: 0 on success, -1 on error 1079 1080Sets the current timestamp of kvmclock to the value specified in its parameter. 1081In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios 1082such as migration. 1083 1084The following flags can be passed: 1085 1086KVM_CLOCK_REALTIME 1087 If set, KVM will compare the value of the `realtime` field 1088 with the value of the host's real time clocksource at the instant when 1089 KVM_SET_CLOCK was called. The difference in elapsed time is added to the final 1090 kvmclock value that will be provided to guests. 1091 1092Other flags returned by ``KVM_GET_CLOCK`` are accepted but ignored. 1093 1094:: 1095 1096 struct kvm_clock_data { 1097 __u64 clock; /* kvmclock current value */ 1098 __u32 flags; 1099 __u32 pad0; 1100 __u64 realtime; 1101 __u64 host_tsc; 1102 __u32 pad[4]; 1103 }; 1104 1105 11064.31 KVM_GET_VCPU_EVENTS 1107------------------------ 1108 1109:Capability: KVM_CAP_VCPU_EVENTS 1110:Extended by: KVM_CAP_INTR_SHADOW 1111:Architectures: x86, arm64 1112:Type: vcpu ioctl 1113:Parameters: struct kvm_vcpu_events (out) 1114:Returns: 0 on success, -1 on error 1115 1116X86: 1117^^^^ 1118 1119Gets currently pending exceptions, interrupts, and NMIs as well as related 1120states of the vcpu. 1121 1122:: 1123 1124 struct kvm_vcpu_events { 1125 struct { 1126 __u8 injected; 1127 __u8 nr; 1128 __u8 has_error_code; 1129 __u8 pending; 1130 __u32 error_code; 1131 } exception; 1132 struct { 1133 __u8 injected; 1134 __u8 nr; 1135 __u8 soft; 1136 __u8 shadow; 1137 } interrupt; 1138 struct { 1139 __u8 injected; 1140 __u8 pending; 1141 __u8 masked; 1142 __u8 pad; 1143 } nmi; 1144 __u32 sipi_vector; 1145 __u32 flags; 1146 struct { 1147 __u8 smm; 1148 __u8 pending; 1149 __u8 smm_inside_nmi; 1150 __u8 latched_init; 1151 } smi; 1152 __u8 reserved[27]; 1153 __u8 exception_has_payload; 1154 __u64 exception_payload; 1155 }; 1156 1157The following bits are defined in the flags field: 1158 1159- KVM_VCPUEVENT_VALID_SHADOW may be set to signal that 1160 interrupt.shadow contains a valid state. 1161 1162- KVM_VCPUEVENT_VALID_SMM may be set to signal that smi contains a 1163 valid state. 1164 1165- KVM_VCPUEVENT_VALID_PAYLOAD may be set to signal that the 1166 exception_has_payload, exception_payload, and exception.pending 1167 fields contain a valid state. This bit will be set whenever 1168 KVM_CAP_EXCEPTION_PAYLOAD is enabled. 1169 1170- KVM_VCPUEVENT_VALID_TRIPLE_FAULT may be set to signal that the 1171 triple_fault_pending field contains a valid state. This bit will 1172 be set whenever KVM_CAP_X86_TRIPLE_FAULT_EVENT is enabled. 1173 1174ARM64: 1175^^^^^^ 1176 1177If the guest accesses a device that is being emulated by the host kernel in 1178such a way that a real device would generate a physical SError, KVM may make 1179a virtual SError pending for that VCPU. This system error interrupt remains 1180pending until the guest takes the exception by unmasking PSTATE.A. 1181 1182Running the VCPU may cause it to take a pending SError, or make an access that 1183causes an SError to become pending. The event's description is only valid while 1184the VPCU is not running. 1185 1186This API provides a way to read and write the pending 'event' state that is not 1187visible to the guest. To save, restore or migrate a VCPU the struct representing 1188the state can be read then written using this GET/SET API, along with the other 1189guest-visible registers. It is not possible to 'cancel' an SError that has been 1190made pending. 1191 1192A device being emulated in user-space may also wish to generate an SError. To do 1193this the events structure can be populated by user-space. The current state 1194should be read first, to ensure no existing SError is pending. If an existing 1195SError is pending, the architecture's 'Multiple SError interrupts' rules should 1196be followed. (2.5.3 of DDI0587.a "ARM Reliability, Availability, and 1197Serviceability (RAS) Specification"). 1198 1199SError exceptions always have an ESR value. Some CPUs have the ability to 1200specify what the virtual SError's ESR value should be. These systems will 1201advertise KVM_CAP_ARM_INJECT_SERROR_ESR. In this case exception.has_esr will 1202always have a non-zero value when read, and the agent making an SError pending 1203should specify the ISS field in the lower 24 bits of exception.serror_esr. If 1204the system supports KVM_CAP_ARM_INJECT_SERROR_ESR, but user-space sets the events 1205with exception.has_esr as zero, KVM will choose an ESR. 1206 1207Specifying exception.has_esr on a system that does not support it will return 1208-EINVAL. Setting anything other than the lower 24bits of exception.serror_esr 1209will return -EINVAL. 1210 1211It is not possible to read back a pending external abort (injected via 1212KVM_SET_VCPU_EVENTS or otherwise) because such an exception is always delivered 1213directly to the virtual CPU). 1214 1215:: 1216 1217 struct kvm_vcpu_events { 1218 struct { 1219 __u8 serror_pending; 1220 __u8 serror_has_esr; 1221 __u8 ext_dabt_pending; 1222 /* Align it to 8 bytes */ 1223 __u8 pad[5]; 1224 __u64 serror_esr; 1225 } exception; 1226 __u32 reserved[12]; 1227 }; 1228 12294.32 KVM_SET_VCPU_EVENTS 1230------------------------ 1231 1232:Capability: KVM_CAP_VCPU_EVENTS 1233:Extended by: KVM_CAP_INTR_SHADOW 1234:Architectures: x86, arm64 1235:Type: vcpu ioctl 1236:Parameters: struct kvm_vcpu_events (in) 1237:Returns: 0 on success, -1 on error 1238 1239X86: 1240^^^^ 1241 1242Set pending exceptions, interrupts, and NMIs as well as related states of the 1243vcpu. 1244 1245See KVM_GET_VCPU_EVENTS for the data structure. 1246 1247Fields that may be modified asynchronously by running VCPUs can be excluded 1248from the update. These fields are nmi.pending, sipi_vector, smi.smm, 1249smi.pending. Keep the corresponding bits in the flags field cleared to 1250suppress overwriting the current in-kernel state. The bits are: 1251 1252=============================== ================================== 1253KVM_VCPUEVENT_VALID_NMI_PENDING transfer nmi.pending to the kernel 1254KVM_VCPUEVENT_VALID_SIPI_VECTOR transfer sipi_vector 1255KVM_VCPUEVENT_VALID_SMM transfer the smi sub-struct. 1256=============================== ================================== 1257 1258If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in 1259the flags field to signal that interrupt.shadow contains a valid state and 1260shall be written into the VCPU. 1261 1262KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available. 1263 1264If KVM_CAP_EXCEPTION_PAYLOAD is enabled, KVM_VCPUEVENT_VALID_PAYLOAD 1265can be set in the flags field to signal that the 1266exception_has_payload, exception_payload, and exception.pending fields 1267contain a valid state and shall be written into the VCPU. 1268 1269If KVM_CAP_X86_TRIPLE_FAULT_EVENT is enabled, KVM_VCPUEVENT_VALID_TRIPLE_FAULT 1270can be set in flags field to signal that the triple_fault field contains 1271a valid state and shall be written into the VCPU. 1272 1273ARM64: 1274^^^^^^ 1275 1276User space may need to inject several types of events to the guest. 1277 1278Set the pending SError exception state for this VCPU. It is not possible to 1279'cancel' an Serror that has been made pending. 1280 1281If the guest performed an access to I/O memory which could not be handled by 1282userspace, for example because of missing instruction syndrome decode 1283information or because there is no device mapped at the accessed IPA, then 1284userspace can ask the kernel to inject an external abort using the address 1285from the exiting fault on the VCPU. It is a programming error to set 1286ext_dabt_pending after an exit which was not either KVM_EXIT_MMIO or 1287KVM_EXIT_ARM_NISV. This feature is only available if the system supports 1288KVM_CAP_ARM_INJECT_EXT_DABT. This is a helper which provides commonality in 1289how userspace reports accesses for the above cases to guests, across different 1290userspace implementations. Nevertheless, userspace can still emulate all Arm 1291exceptions by manipulating individual registers using the KVM_SET_ONE_REG API. 1292 1293See KVM_GET_VCPU_EVENTS for the data structure. 1294 1295 12964.33 KVM_GET_DEBUGREGS 1297---------------------- 1298 1299:Capability: KVM_CAP_DEBUGREGS 1300:Architectures: x86 1301:Type: vm ioctl 1302:Parameters: struct kvm_debugregs (out) 1303:Returns: 0 on success, -1 on error 1304 1305Reads debug registers from the vcpu. 1306 1307:: 1308 1309 struct kvm_debugregs { 1310 __u64 db[4]; 1311 __u64 dr6; 1312 __u64 dr7; 1313 __u64 flags; 1314 __u64 reserved[9]; 1315 }; 1316 1317 13184.34 KVM_SET_DEBUGREGS 1319---------------------- 1320 1321:Capability: KVM_CAP_DEBUGREGS 1322:Architectures: x86 1323:Type: vm ioctl 1324:Parameters: struct kvm_debugregs (in) 1325:Returns: 0 on success, -1 on error 1326 1327Writes debug registers into the vcpu. 1328 1329See KVM_GET_DEBUGREGS for the data structure. The flags field is unused 1330yet and must be cleared on entry. 1331 1332 13334.35 KVM_SET_USER_MEMORY_REGION 1334------------------------------- 1335 1336:Capability: KVM_CAP_USER_MEMORY 1337:Architectures: all 1338:Type: vm ioctl 1339:Parameters: struct kvm_userspace_memory_region (in) 1340:Returns: 0 on success, -1 on error 1341 1342:: 1343 1344 struct kvm_userspace_memory_region { 1345 __u32 slot; 1346 __u32 flags; 1347 __u64 guest_phys_addr; 1348 __u64 memory_size; /* bytes */ 1349 __u64 userspace_addr; /* start of the userspace allocated memory */ 1350 }; 1351 1352 /* for kvm_userspace_memory_region::flags */ 1353 #define KVM_MEM_LOG_DIRTY_PAGES (1UL << 0) 1354 #define KVM_MEM_READONLY (1UL << 1) 1355 1356This ioctl allows the user to create, modify or delete a guest physical 1357memory slot. Bits 0-15 of "slot" specify the slot id and this value 1358should be less than the maximum number of user memory slots supported per 1359VM. The maximum allowed slots can be queried using KVM_CAP_NR_MEMSLOTS. 1360Slots may not overlap in guest physical address space. 1361 1362If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot" 1363specifies the address space which is being modified. They must be 1364less than the value that KVM_CHECK_EXTENSION returns for the 1365KVM_CAP_MULTI_ADDRESS_SPACE capability. Slots in separate address spaces 1366are unrelated; the restriction on overlapping slots only applies within 1367each address space. 1368 1369Deleting a slot is done by passing zero for memory_size. When changing 1370an existing slot, it may be moved in the guest physical memory space, 1371or its flags may be modified, but it may not be resized. 1372 1373Memory for the region is taken starting at the address denoted by the 1374field userspace_addr, which must point at user addressable memory for 1375the entire memory slot size. Any object may back this memory, including 1376anonymous memory, ordinary files, and hugetlbfs. 1377 1378On architectures that support a form of address tagging, userspace_addr must 1379be an untagged address. 1380 1381It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr 1382be identical. This allows large pages in the guest to be backed by large 1383pages in the host. 1384 1385The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and 1386KVM_MEM_READONLY. The former can be set to instruct KVM to keep track of 1387writes to memory within the slot. See KVM_GET_DIRTY_LOG ioctl to know how to 1388use it. The latter can be set, if KVM_CAP_READONLY_MEM capability allows it, 1389to make a new slot read-only. In this case, writes to this memory will be 1390posted to userspace as KVM_EXIT_MMIO exits. 1391 1392When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of 1393the memory region are automatically reflected into the guest. For example, an 1394mmap() that affects the region will be made visible immediately. Another 1395example is madvise(MADV_DROP). 1396 1397Note: On arm64, a write generated by the page-table walker (to update 1398the Access and Dirty flags, for example) never results in a 1399KVM_EXIT_MMIO exit when the slot has the KVM_MEM_READONLY flag. This 1400is because KVM cannot provide the data that would be written by the 1401page-table walker, making it impossible to emulate the access. 1402Instead, an abort (data abort if the cause of the page-table update 1403was a load or a store, instruction abort if it was an instruction 1404fetch) is injected in the guest. 1405 1406S390: 1407^^^^^ 1408 1409Returns -EINVAL if the VM has the KVM_VM_S390_UCONTROL flag set. 1410Returns -EINVAL if called on a protected VM. 1411 14124.36 KVM_SET_TSS_ADDR 1413--------------------- 1414 1415:Capability: KVM_CAP_SET_TSS_ADDR 1416:Architectures: x86 1417:Type: vm ioctl 1418:Parameters: unsigned long tss_address (in) 1419:Returns: 0 on success, -1 on error 1420 1421This ioctl defines the physical address of a three-page region in the guest 1422physical address space. The region must be within the first 4GB of the 1423guest physical address space and must not conflict with any memory slot 1424or any mmio address. The guest may malfunction if it accesses this memory 1425region. 1426 1427This ioctl is required on Intel-based hosts. This is needed on Intel hardware 1428because of a quirk in the virtualization implementation (see the internals 1429documentation when it pops into existence). 1430 1431 14324.37 KVM_ENABLE_CAP 1433------------------- 1434 1435:Capability: KVM_CAP_ENABLE_CAP 1436:Architectures: mips, ppc, s390, x86, loongarch 1437:Type: vcpu ioctl 1438:Parameters: struct kvm_enable_cap (in) 1439:Returns: 0 on success; -1 on error 1440 1441:Capability: KVM_CAP_ENABLE_CAP_VM 1442:Architectures: all 1443:Type: vm ioctl 1444:Parameters: struct kvm_enable_cap (in) 1445:Returns: 0 on success; -1 on error 1446 1447.. note:: 1448 1449 Not all extensions are enabled by default. Using this ioctl the application 1450 can enable an extension, making it available to the guest. 1451 1452On systems that do not support this ioctl, it always fails. On systems that 1453do support it, it only works for extensions that are supported for enablement. 1454 1455To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should 1456be used. 1457 1458:: 1459 1460 struct kvm_enable_cap { 1461 /* in */ 1462 __u32 cap; 1463 1464The capability that is supposed to get enabled. 1465 1466:: 1467 1468 __u32 flags; 1469 1470A bitfield indicating future enhancements. Has to be 0 for now. 1471 1472:: 1473 1474 __u64 args[4]; 1475 1476Arguments for enabling a feature. If a feature needs initial values to 1477function properly, this is the place to put them. 1478 1479:: 1480 1481 __u8 pad[64]; 1482 }; 1483 1484The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl 1485for vm-wide capabilities. 1486 14874.38 KVM_GET_MP_STATE 1488--------------------- 1489 1490:Capability: KVM_CAP_MP_STATE 1491:Architectures: x86, s390, arm64, riscv, loongarch 1492:Type: vcpu ioctl 1493:Parameters: struct kvm_mp_state (out) 1494:Returns: 0 on success; -1 on error 1495 1496:: 1497 1498 struct kvm_mp_state { 1499 __u32 mp_state; 1500 }; 1501 1502Returns the vcpu's current "multiprocessing state" (though also valid on 1503uniprocessor guests). 1504 1505Possible values are: 1506 1507 ========================== =============================================== 1508 KVM_MP_STATE_RUNNABLE the vcpu is currently running 1509 [x86,arm64,riscv,loongarch] 1510 KVM_MP_STATE_UNINITIALIZED the vcpu is an application processor (AP) 1511 which has not yet received an INIT signal [x86] 1512 KVM_MP_STATE_INIT_RECEIVED the vcpu has received an INIT signal, and is 1513 now ready for a SIPI [x86] 1514 KVM_MP_STATE_HALTED the vcpu has executed a HLT instruction and 1515 is waiting for an interrupt [x86] 1516 KVM_MP_STATE_SIPI_RECEIVED the vcpu has just received a SIPI (vector 1517 accessible via KVM_GET_VCPU_EVENTS) [x86] 1518 KVM_MP_STATE_STOPPED the vcpu is stopped [s390,arm64,riscv] 1519 KVM_MP_STATE_CHECK_STOP the vcpu is in a special error state [s390] 1520 KVM_MP_STATE_OPERATING the vcpu is operating (running or halted) 1521 [s390] 1522 KVM_MP_STATE_LOAD the vcpu is in a special load/startup state 1523 [s390] 1524 KVM_MP_STATE_SUSPENDED the vcpu is in a suspend state and is waiting 1525 for a wakeup event [arm64] 1526 ========================== =============================================== 1527 1528On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an 1529in-kernel irqchip, the multiprocessing state must be maintained by userspace on 1530these architectures. 1531 1532For arm64: 1533^^^^^^^^^^ 1534 1535If a vCPU is in the KVM_MP_STATE_SUSPENDED state, KVM will emulate the 1536architectural execution of a WFI instruction. 1537 1538If a wakeup event is recognized, KVM will exit to userspace with a 1539KVM_SYSTEM_EVENT exit, where the event type is KVM_SYSTEM_EVENT_WAKEUP. If 1540userspace wants to honor the wakeup, it must set the vCPU's MP state to 1541KVM_MP_STATE_RUNNABLE. If it does not, KVM will continue to await a wakeup 1542event in subsequent calls to KVM_RUN. 1543 1544.. warning:: 1545 1546 If userspace intends to keep the vCPU in a SUSPENDED state, it is 1547 strongly recommended that userspace take action to suppress the 1548 wakeup event (such as masking an interrupt). Otherwise, subsequent 1549 calls to KVM_RUN will immediately exit with a KVM_SYSTEM_EVENT_WAKEUP 1550 event and inadvertently waste CPU cycles. 1551 1552 Additionally, if userspace takes action to suppress a wakeup event, 1553 it is strongly recommended that it also restores the vCPU to its 1554 original state when the vCPU is made RUNNABLE again. For example, 1555 if userspace masked a pending interrupt to suppress the wakeup, 1556 the interrupt should be unmasked before returning control to the 1557 guest. 1558 1559For riscv: 1560^^^^^^^^^^ 1561 1562The only states that are valid are KVM_MP_STATE_STOPPED and 1563KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not. 1564 1565On LoongArch, only the KVM_MP_STATE_RUNNABLE state is used to reflect 1566whether the vcpu is runnable. 1567 15684.39 KVM_SET_MP_STATE 1569--------------------- 1570 1571:Capability: KVM_CAP_MP_STATE 1572:Architectures: x86, s390, arm64, riscv, loongarch 1573:Type: vcpu ioctl 1574:Parameters: struct kvm_mp_state (in) 1575:Returns: 0 on success; -1 on error 1576 1577Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for 1578arguments. 1579 1580On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an 1581in-kernel irqchip, the multiprocessing state must be maintained by userspace on 1582these architectures. 1583 1584For arm64/riscv: 1585^^^^^^^^^^^^^^^^ 1586 1587The only states that are valid are KVM_MP_STATE_STOPPED and 1588KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not. 1589 1590On LoongArch, only the KVM_MP_STATE_RUNNABLE state is used to reflect 1591whether the vcpu is runnable. 1592 15934.40 KVM_SET_IDENTITY_MAP_ADDR 1594------------------------------ 1595 1596:Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR 1597:Architectures: x86 1598:Type: vm ioctl 1599:Parameters: unsigned long identity (in) 1600:Returns: 0 on success, -1 on error 1601 1602This ioctl defines the physical address of a one-page region in the guest 1603physical address space. The region must be within the first 4GB of the 1604guest physical address space and must not conflict with any memory slot 1605or any mmio address. The guest may malfunction if it accesses this memory 1606region. 1607 1608Setting the address to 0 will result in resetting the address to its default 1609(0xfffbc000). 1610 1611This ioctl is required on Intel-based hosts. This is needed on Intel hardware 1612because of a quirk in the virtualization implementation (see the internals 1613documentation when it pops into existence). 1614 1615Fails if any VCPU has already been created. 1616 16174.41 KVM_SET_BOOT_CPU_ID 1618------------------------ 1619 1620:Capability: KVM_CAP_SET_BOOT_CPU_ID 1621:Architectures: x86 1622:Type: vm ioctl 1623:Parameters: unsigned long vcpu_id 1624:Returns: 0 on success, -1 on error 1625 1626Define which vcpu is the Bootstrap Processor (BSP). Values are the same 1627as the vcpu id in KVM_CREATE_VCPU. If this ioctl is not called, the default 1628is vcpu 0. This ioctl has to be called before vcpu creation, 1629otherwise it will return EBUSY error. 1630 1631 16324.42 KVM_GET_XSAVE 1633------------------ 1634 1635:Capability: KVM_CAP_XSAVE 1636:Architectures: x86 1637:Type: vcpu ioctl 1638:Parameters: struct kvm_xsave (out) 1639:Returns: 0 on success, -1 on error 1640 1641 1642:: 1643 1644 struct kvm_xsave { 1645 __u32 region[1024]; 1646 __u32 extra[0]; 1647 }; 1648 1649This ioctl would copy current vcpu's xsave struct to the userspace. 1650 1651 16524.43 KVM_SET_XSAVE 1653------------------ 1654 1655:Capability: KVM_CAP_XSAVE and KVM_CAP_XSAVE2 1656:Architectures: x86 1657:Type: vcpu ioctl 1658:Parameters: struct kvm_xsave (in) 1659:Returns: 0 on success, -1 on error 1660 1661:: 1662 1663 1664 struct kvm_xsave { 1665 __u32 region[1024]; 1666 __u32 extra[0]; 1667 }; 1668 1669This ioctl would copy userspace's xsave struct to the kernel. It copies 1670as many bytes as are returned by KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2), 1671when invoked on the vm file descriptor. The size value returned by 1672KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) will always be at least 4096. 1673Currently, it is only greater than 4096 if a dynamic feature has been 1674enabled with ``arch_prctl()``, but this may change in the future. 1675 1676The offsets of the state save areas in struct kvm_xsave follow the 1677contents of CPUID leaf 0xD on the host. 1678 1679 16804.44 KVM_GET_XCRS 1681----------------- 1682 1683:Capability: KVM_CAP_XCRS 1684:Architectures: x86 1685:Type: vcpu ioctl 1686:Parameters: struct kvm_xcrs (out) 1687:Returns: 0 on success, -1 on error 1688 1689:: 1690 1691 struct kvm_xcr { 1692 __u32 xcr; 1693 __u32 reserved; 1694 __u64 value; 1695 }; 1696 1697 struct kvm_xcrs { 1698 __u32 nr_xcrs; 1699 __u32 flags; 1700 struct kvm_xcr xcrs[KVM_MAX_XCRS]; 1701 __u64 padding[16]; 1702 }; 1703 1704This ioctl would copy current vcpu's xcrs to the userspace. 1705 1706 17074.45 KVM_SET_XCRS 1708----------------- 1709 1710:Capability: KVM_CAP_XCRS 1711:Architectures: x86 1712:Type: vcpu ioctl 1713:Parameters: struct kvm_xcrs (in) 1714:Returns: 0 on success, -1 on error 1715 1716:: 1717 1718 struct kvm_xcr { 1719 __u32 xcr; 1720 __u32 reserved; 1721 __u64 value; 1722 }; 1723 1724 struct kvm_xcrs { 1725 __u32 nr_xcrs; 1726 __u32 flags; 1727 struct kvm_xcr xcrs[KVM_MAX_XCRS]; 1728 __u64 padding[16]; 1729 }; 1730 1731This ioctl would set vcpu's xcr to the value userspace specified. 1732 1733 17344.46 KVM_GET_SUPPORTED_CPUID 1735---------------------------- 1736 1737:Capability: KVM_CAP_EXT_CPUID 1738:Architectures: x86 1739:Type: system ioctl 1740:Parameters: struct kvm_cpuid2 (in/out) 1741:Returns: 0 on success, -1 on error 1742 1743:: 1744 1745 struct kvm_cpuid2 { 1746 __u32 nent; 1747 __u32 padding; 1748 struct kvm_cpuid_entry2 entries[0]; 1749 }; 1750 1751 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0) 1752 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */ 1753 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */ 1754 1755 struct kvm_cpuid_entry2 { 1756 __u32 function; 1757 __u32 index; 1758 __u32 flags; 1759 __u32 eax; 1760 __u32 ebx; 1761 __u32 ecx; 1762 __u32 edx; 1763 __u32 padding[3]; 1764 }; 1765 1766This ioctl returns x86 cpuid features which are supported by both the 1767hardware and kvm in its default configuration. Userspace can use the 1768information returned by this ioctl to construct cpuid information (for 1769KVM_SET_CPUID2) that is consistent with hardware, kernel, and 1770userspace capabilities, and with user requirements (for example, the 1771user may wish to constrain cpuid to emulate older hardware, or for 1772feature consistency across a cluster). 1773 1774Dynamically-enabled feature bits need to be requested with 1775``arch_prctl()`` before calling this ioctl. Feature bits that have not 1776been requested are excluded from the result. 1777 1778Note that certain capabilities, such as KVM_CAP_X86_DISABLE_EXITS, may 1779expose cpuid features (e.g. MONITOR) which are not supported by kvm in 1780its default configuration. If userspace enables such capabilities, it 1781is responsible for modifying the results of this ioctl appropriately. 1782 1783Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure 1784with the 'nent' field indicating the number of entries in the variable-size 1785array 'entries'. If the number of entries is too low to describe the cpu 1786capabilities, an error (E2BIG) is returned. If the number is too high, 1787the 'nent' field is adjusted and an error (ENOMEM) is returned. If the 1788number is just right, the 'nent' field is adjusted to the number of valid 1789entries in the 'entries' array, which is then filled. 1790 1791The entries returned are the host cpuid as returned by the cpuid instruction, 1792with unknown or unsupported features masked out. Some features (for example, 1793x2apic), may not be present in the host cpu, but are exposed by kvm if it can 1794emulate them efficiently. The fields in each entry are defined as follows: 1795 1796 function: 1797 the eax value used to obtain the entry 1798 1799 index: 1800 the ecx value used to obtain the entry (for entries that are 1801 affected by ecx) 1802 1803 flags: 1804 an OR of zero or more of the following: 1805 1806 KVM_CPUID_FLAG_SIGNIFCANT_INDEX: 1807 if the index field is valid 1808 1809 eax, ebx, ecx, edx: 1810 the values returned by the cpuid instruction for 1811 this function/index combination 1812 1813The TSC deadline timer feature (CPUID leaf 1, ecx[24]) is always returned 1814as false, since the feature depends on KVM_CREATE_IRQCHIP for local APIC 1815support. Instead it is reported via:: 1816 1817 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER) 1818 1819if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the 1820feature in userspace, then you can enable the feature for KVM_SET_CPUID2. 1821 1822 18234.47 KVM_PPC_GET_PVINFO 1824----------------------- 1825 1826:Capability: KVM_CAP_PPC_GET_PVINFO 1827:Architectures: ppc 1828:Type: vm ioctl 1829:Parameters: struct kvm_ppc_pvinfo (out) 1830:Returns: 0 on success, !0 on error 1831 1832:: 1833 1834 struct kvm_ppc_pvinfo { 1835 __u32 flags; 1836 __u32 hcall[4]; 1837 __u8 pad[108]; 1838 }; 1839 1840This ioctl fetches PV specific information that need to be passed to the guest 1841using the device tree or other means from vm context. 1842 1843The hcall array defines 4 instructions that make up a hypercall. 1844 1845If any additional field gets added to this structure later on, a bit for that 1846additional piece of information will be set in the flags bitmap. 1847 1848The flags bitmap is defined as:: 1849 1850 /* the host supports the ePAPR idle hcall 1851 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1<<0) 1852 18534.52 KVM_SET_GSI_ROUTING 1854------------------------ 1855 1856:Capability: KVM_CAP_IRQ_ROUTING 1857:Architectures: x86 s390 arm64 1858:Type: vm ioctl 1859:Parameters: struct kvm_irq_routing (in) 1860:Returns: 0 on success, -1 on error 1861 1862Sets the GSI routing table entries, overwriting any previously set entries. 1863 1864On arm64, GSI routing has the following limitation: 1865 1866- GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD. 1867 1868:: 1869 1870 struct kvm_irq_routing { 1871 __u32 nr; 1872 __u32 flags; 1873 struct kvm_irq_routing_entry entries[0]; 1874 }; 1875 1876No flags are specified so far, the corresponding field must be set to zero. 1877 1878:: 1879 1880 struct kvm_irq_routing_entry { 1881 __u32 gsi; 1882 __u32 type; 1883 __u32 flags; 1884 __u32 pad; 1885 union { 1886 struct kvm_irq_routing_irqchip irqchip; 1887 struct kvm_irq_routing_msi msi; 1888 struct kvm_irq_routing_s390_adapter adapter; 1889 struct kvm_irq_routing_hv_sint hv_sint; 1890 struct kvm_irq_routing_xen_evtchn xen_evtchn; 1891 __u32 pad[8]; 1892 } u; 1893 }; 1894 1895 /* gsi routing entry types */ 1896 #define KVM_IRQ_ROUTING_IRQCHIP 1 1897 #define KVM_IRQ_ROUTING_MSI 2 1898 #define KVM_IRQ_ROUTING_S390_ADAPTER 3 1899 #define KVM_IRQ_ROUTING_HV_SINT 4 1900 #define KVM_IRQ_ROUTING_XEN_EVTCHN 5 1901 1902flags: 1903 1904- KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry 1905 type, specifies that the devid field contains a valid value. The per-VM 1906 KVM_CAP_MSI_DEVID capability advertises the requirement to provide 1907 the device ID. If this capability is not available, userspace should 1908 never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail. 1909- zero otherwise 1910 1911:: 1912 1913 struct kvm_irq_routing_irqchip { 1914 __u32 irqchip; 1915 __u32 pin; 1916 }; 1917 1918 struct kvm_irq_routing_msi { 1919 __u32 address_lo; 1920 __u32 address_hi; 1921 __u32 data; 1922 union { 1923 __u32 pad; 1924 __u32 devid; 1925 }; 1926 }; 1927 1928If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier 1929for the device that wrote the MSI message. For PCI, this is usually a 1930BDF identifier in the lower 16 bits. 1931 1932On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS 1933feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled, 1934address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of 1935address_hi must be zero. 1936 1937:: 1938 1939 struct kvm_irq_routing_s390_adapter { 1940 __u64 ind_addr; 1941 __u64 summary_addr; 1942 __u64 ind_offset; 1943 __u32 summary_offset; 1944 __u32 adapter_id; 1945 }; 1946 1947 struct kvm_irq_routing_hv_sint { 1948 __u32 vcpu; 1949 __u32 sint; 1950 }; 1951 1952 struct kvm_irq_routing_xen_evtchn { 1953 __u32 port; 1954 __u32 vcpu; 1955 __u32 priority; 1956 }; 1957 1958 1959When KVM_CAP_XEN_HVM includes the KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL bit 1960in its indication of supported features, routing to Xen event channels 1961is supported. Although the priority field is present, only the value 1962KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL is supported, which means delivery by 19632 level event channels. FIFO event channel support may be added in 1964the future. 1965 1966 19674.55 KVM_SET_TSC_KHZ 1968-------------------- 1969 1970:Capability: KVM_CAP_TSC_CONTROL / KVM_CAP_VM_TSC_CONTROL 1971:Architectures: x86 1972:Type: vcpu ioctl / vm ioctl 1973:Parameters: virtual tsc_khz 1974:Returns: 0 on success, -1 on error 1975 1976Specifies the tsc frequency for the virtual machine. The unit of the 1977frequency is KHz. 1978 1979If the KVM_CAP_VM_TSC_CONTROL capability is advertised, this can also 1980be used as a vm ioctl to set the initial tsc frequency of subsequently 1981created vCPUs. 1982 19834.56 KVM_GET_TSC_KHZ 1984-------------------- 1985 1986:Capability: KVM_CAP_GET_TSC_KHZ / KVM_CAP_VM_TSC_CONTROL 1987:Architectures: x86 1988:Type: vcpu ioctl / vm ioctl 1989:Parameters: none 1990:Returns: virtual tsc-khz on success, negative value on error 1991 1992Returns the tsc frequency of the guest. The unit of the return value is 1993KHz. If the host has unstable tsc this ioctl returns -EIO instead as an 1994error. 1995 1996 19974.57 KVM_GET_LAPIC 1998------------------ 1999 2000:Capability: KVM_CAP_IRQCHIP 2001:Architectures: x86 2002:Type: vcpu ioctl 2003:Parameters: struct kvm_lapic_state (out) 2004:Returns: 0 on success, -1 on error 2005 2006:: 2007 2008 #define KVM_APIC_REG_SIZE 0x400 2009 struct kvm_lapic_state { 2010 char regs[KVM_APIC_REG_SIZE]; 2011 }; 2012 2013Reads the Local APIC registers and copies them into the input argument. The 2014data format and layout are the same as documented in the architecture manual. 2015 2016If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is 2017enabled, then the format of APIC_ID register depends on the APIC mode 2018(reported by MSR_IA32_APICBASE) of its VCPU. x2APIC stores APIC ID in 2019the APIC_ID register (bytes 32-35). xAPIC only allows an 8-bit APIC ID 2020which is stored in bits 31-24 of the APIC register, or equivalently in 2021byte 35 of struct kvm_lapic_state's regs field. KVM_GET_LAPIC must then 2022be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR. 2023 2024If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state 2025always uses xAPIC format. 2026 2027 20284.58 KVM_SET_LAPIC 2029------------------ 2030 2031:Capability: KVM_CAP_IRQCHIP 2032:Architectures: x86 2033:Type: vcpu ioctl 2034:Parameters: struct kvm_lapic_state (in) 2035:Returns: 0 on success, -1 on error 2036 2037:: 2038 2039 #define KVM_APIC_REG_SIZE 0x400 2040 struct kvm_lapic_state { 2041 char regs[KVM_APIC_REG_SIZE]; 2042 }; 2043 2044Copies the input argument into the Local APIC registers. The data format 2045and layout are the same as documented in the architecture manual. 2046 2047The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's 2048regs field) depends on the state of the KVM_CAP_X2APIC_API capability. 2049See the note in KVM_GET_LAPIC. 2050 2051 20524.59 KVM_IOEVENTFD 2053------------------ 2054 2055:Capability: KVM_CAP_IOEVENTFD 2056:Architectures: all 2057:Type: vm ioctl 2058:Parameters: struct kvm_ioeventfd (in) 2059:Returns: 0 on success, !0 on error 2060 2061This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address 2062within the guest. A guest write in the registered address will signal the 2063provided event instead of triggering an exit. 2064 2065:: 2066 2067 struct kvm_ioeventfd { 2068 __u64 datamatch; 2069 __u64 addr; /* legal pio/mmio address */ 2070 __u32 len; /* 0, 1, 2, 4, or 8 bytes */ 2071 __s32 fd; 2072 __u32 flags; 2073 __u8 pad[36]; 2074 }; 2075 2076For the special case of virtio-ccw devices on s390, the ioevent is matched 2077to a subchannel/virtqueue tuple instead. 2078 2079The following flags are defined:: 2080 2081 #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch) 2082 #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio) 2083 #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign) 2084 #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \ 2085 (1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify) 2086 2087If datamatch flag is set, the event will be signaled only if the written value 2088to the registered address is equal to datamatch in struct kvm_ioeventfd. 2089 2090For virtio-ccw devices, addr contains the subchannel id and datamatch the 2091virtqueue index. 2092 2093With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and 2094the kernel will ignore the length of guest write and may get a faster vmexit. 2095The speedup may only apply to specific architectures, but the ioeventfd will 2096work anyway. 2097 20984.60 KVM_DIRTY_TLB 2099------------------ 2100 2101:Capability: KVM_CAP_SW_TLB 2102:Architectures: ppc 2103:Type: vcpu ioctl 2104:Parameters: struct kvm_dirty_tlb (in) 2105:Returns: 0 on success, -1 on error 2106 2107:: 2108 2109 struct kvm_dirty_tlb { 2110 __u64 bitmap; 2111 __u32 num_dirty; 2112 }; 2113 2114This must be called whenever userspace has changed an entry in the shared 2115TLB, prior to calling KVM_RUN on the associated vcpu. 2116 2117The "bitmap" field is the userspace address of an array. This array 2118consists of a number of bits, equal to the total number of TLB entries as 2119determined by the last successful call to KVM_CONFIG_TLB, rounded up to the 2120nearest multiple of 64. 2121 2122Each bit corresponds to one TLB entry, ordered the same as in the shared TLB 2123array. 2124 2125The array is little-endian: the bit 0 is the least significant bit of the 2126first byte, bit 8 is the least significant bit of the second byte, etc. 2127This avoids any complications with differing word sizes. 2128 2129The "num_dirty" field is a performance hint for KVM to determine whether it 2130should skip processing the bitmap and just invalidate everything. It must 2131be set to the number of set bits in the bitmap. 2132 2133 21344.62 KVM_CREATE_SPAPR_TCE 2135------------------------- 2136 2137:Capability: KVM_CAP_SPAPR_TCE 2138:Architectures: powerpc 2139:Type: vm ioctl 2140:Parameters: struct kvm_create_spapr_tce (in) 2141:Returns: file descriptor for manipulating the created TCE table 2142 2143This creates a virtual TCE (translation control entry) table, which 2144is an IOMMU for PAPR-style virtual I/O. It is used to translate 2145logical addresses used in virtual I/O into guest physical addresses, 2146and provides a scatter/gather capability for PAPR virtual I/O. 2147 2148:: 2149 2150 /* for KVM_CAP_SPAPR_TCE */ 2151 struct kvm_create_spapr_tce { 2152 __u64 liobn; 2153 __u32 window_size; 2154 }; 2155 2156The liobn field gives the logical IO bus number for which to create a 2157TCE table. The window_size field specifies the size of the DMA window 2158which this TCE table will translate - the table will contain one 64 2159bit TCE entry for every 4kiB of the DMA window. 2160 2161When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE 2162table has been created using this ioctl(), the kernel will handle it 2163in real mode, updating the TCE table. H_PUT_TCE calls for other 2164liobns will cause a vm exit and must be handled by userspace. 2165 2166The return value is a file descriptor which can be passed to mmap(2) 2167to map the created TCE table into userspace. This lets userspace read 2168the entries written by kernel-handled H_PUT_TCE calls, and also lets 2169userspace update the TCE table directly which is useful in some 2170circumstances. 2171 2172 21734.63 KVM_ALLOCATE_RMA 2174--------------------- 2175 2176:Capability: KVM_CAP_PPC_RMA 2177:Architectures: powerpc 2178:Type: vm ioctl 2179:Parameters: struct kvm_allocate_rma (out) 2180:Returns: file descriptor for mapping the allocated RMA 2181 2182This allocates a Real Mode Area (RMA) from the pool allocated at boot 2183time by the kernel. An RMA is a physically-contiguous, aligned region 2184of memory used on older POWER processors to provide the memory which 2185will be accessed by real-mode (MMU off) accesses in a KVM guest. 2186POWER processors support a set of sizes for the RMA that usually 2187includes 64MB, 128MB, 256MB and some larger powers of two. 2188 2189:: 2190 2191 /* for KVM_ALLOCATE_RMA */ 2192 struct kvm_allocate_rma { 2193 __u64 rma_size; 2194 }; 2195 2196The return value is a file descriptor which can be passed to mmap(2) 2197to map the allocated RMA into userspace. The mapped area can then be 2198passed to the KVM_SET_USER_MEMORY_REGION ioctl to establish it as the 2199RMA for a virtual machine. The size of the RMA in bytes (which is 2200fixed at host kernel boot time) is returned in the rma_size field of 2201the argument structure. 2202 2203The KVM_CAP_PPC_RMA capability is 1 or 2 if the KVM_ALLOCATE_RMA ioctl 2204is supported; 2 if the processor requires all virtual machines to have 2205an RMA, or 1 if the processor can use an RMA but doesn't require it, 2206because it supports the Virtual RMA (VRMA) facility. 2207 2208 22094.64 KVM_NMI 2210------------ 2211 2212:Capability: KVM_CAP_USER_NMI 2213:Architectures: x86 2214:Type: vcpu ioctl 2215:Parameters: none 2216:Returns: 0 on success, -1 on error 2217 2218Queues an NMI on the thread's vcpu. Note this is well defined only 2219when KVM_CREATE_IRQCHIP has not been called, since this is an interface 2220between the virtual cpu core and virtual local APIC. After KVM_CREATE_IRQCHIP 2221has been called, this interface is completely emulated within the kernel. 2222 2223To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the 2224following algorithm: 2225 2226 - pause the vcpu 2227 - read the local APIC's state (KVM_GET_LAPIC) 2228 - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1) 2229 - if so, issue KVM_NMI 2230 - resume the vcpu 2231 2232Some guests configure the LINT1 NMI input to cause a panic, aiding in 2233debugging. 2234 2235 22364.65 KVM_S390_UCAS_MAP 2237---------------------- 2238 2239:Capability: KVM_CAP_S390_UCONTROL 2240:Architectures: s390 2241:Type: vcpu ioctl 2242:Parameters: struct kvm_s390_ucas_mapping (in) 2243:Returns: 0 in case of success 2244 2245The parameter is defined like this:: 2246 2247 struct kvm_s390_ucas_mapping { 2248 __u64 user_addr; 2249 __u64 vcpu_addr; 2250 __u64 length; 2251 }; 2252 2253This ioctl maps the memory at "user_addr" with the length "length" to 2254the vcpu's address space starting at "vcpu_addr". All parameters need to 2255be aligned by 1 megabyte. 2256 2257 22584.66 KVM_S390_UCAS_UNMAP 2259------------------------ 2260 2261:Capability: KVM_CAP_S390_UCONTROL 2262:Architectures: s390 2263:Type: vcpu ioctl 2264:Parameters: struct kvm_s390_ucas_mapping (in) 2265:Returns: 0 in case of success 2266 2267The parameter is defined like this:: 2268 2269 struct kvm_s390_ucas_mapping { 2270 __u64 user_addr; 2271 __u64 vcpu_addr; 2272 __u64 length; 2273 }; 2274 2275This ioctl unmaps the memory in the vcpu's address space starting at 2276"vcpu_addr" with the length "length". The field "user_addr" is ignored. 2277All parameters need to be aligned by 1 megabyte. 2278 2279 22804.67 KVM_S390_VCPU_FAULT 2281------------------------ 2282 2283:Capability: KVM_CAP_S390_UCONTROL 2284:Architectures: s390 2285:Type: vcpu ioctl 2286:Parameters: vcpu absolute address (in) 2287:Returns: 0 in case of success 2288 2289This call creates a page table entry on the virtual cpu's address space 2290(for user controlled virtual machines) or the virtual machine's address 2291space (for regular virtual machines). This only works for minor faults, 2292thus it's recommended to access subject memory page via the user page 2293table upfront. This is useful to handle validity intercepts for user 2294controlled virtual machines to fault in the virtual cpu's lowcore pages 2295prior to calling the KVM_RUN ioctl. 2296 2297 22984.68 KVM_SET_ONE_REG 2299-------------------- 2300 2301:Capability: KVM_CAP_ONE_REG 2302:Architectures: all 2303:Type: vcpu ioctl 2304:Parameters: struct kvm_one_reg (in) 2305:Returns: 0 on success, negative value on failure 2306 2307Errors: 2308 2309 ====== ============================================================ 2310 ENOENT no such register 2311 EINVAL invalid register ID, or no such register or used with VMs in 2312 protected virtualization mode on s390 2313 EPERM (arm64) register access not allowed before vcpu finalization 2314 EBUSY (riscv) changing register value not allowed after the vcpu 2315 has run at least once 2316 ====== ============================================================ 2317 2318(These error codes are indicative only: do not rely on a specific error 2319code being returned in a specific situation.) 2320 2321:: 2322 2323 struct kvm_one_reg { 2324 __u64 id; 2325 __u64 addr; 2326 }; 2327 2328Using this ioctl, a single vcpu register can be set to a specific value 2329defined by user space with the passed in struct kvm_one_reg, where id 2330refers to the register identifier as described below and addr is a pointer 2331to a variable with the respective size. There can be architecture agnostic 2332and architecture specific registers. Each have their own range of operation 2333and their own constants and width. To keep track of the implemented 2334registers, find a list below: 2335 2336 ======= =============================== ============ 2337 Arch Register Width (bits) 2338 ======= =============================== ============ 2339 PPC KVM_REG_PPC_HIOR 64 2340 PPC KVM_REG_PPC_IAC1 64 2341 PPC KVM_REG_PPC_IAC2 64 2342 PPC KVM_REG_PPC_IAC3 64 2343 PPC KVM_REG_PPC_IAC4 64 2344 PPC KVM_REG_PPC_DAC1 64 2345 PPC KVM_REG_PPC_DAC2 64 2346 PPC KVM_REG_PPC_DABR 64 2347 PPC KVM_REG_PPC_DSCR 64 2348 PPC KVM_REG_PPC_PURR 64 2349 PPC KVM_REG_PPC_SPURR 64 2350 PPC KVM_REG_PPC_DAR 64 2351 PPC KVM_REG_PPC_DSISR 32 2352 PPC KVM_REG_PPC_AMR 64 2353 PPC KVM_REG_PPC_UAMOR 64 2354 PPC KVM_REG_PPC_MMCR0 64 2355 PPC KVM_REG_PPC_MMCR1 64 2356 PPC KVM_REG_PPC_MMCRA 64 2357 PPC KVM_REG_PPC_MMCR2 64 2358 PPC KVM_REG_PPC_MMCRS 64 2359 PPC KVM_REG_PPC_MMCR3 64 2360 PPC KVM_REG_PPC_SIAR 64 2361 PPC KVM_REG_PPC_SDAR 64 2362 PPC KVM_REG_PPC_SIER 64 2363 PPC KVM_REG_PPC_SIER2 64 2364 PPC KVM_REG_PPC_SIER3 64 2365 PPC KVM_REG_PPC_PMC1 32 2366 PPC KVM_REG_PPC_PMC2 32 2367 PPC KVM_REG_PPC_PMC3 32 2368 PPC KVM_REG_PPC_PMC4 32 2369 PPC KVM_REG_PPC_PMC5 32 2370 PPC KVM_REG_PPC_PMC6 32 2371 PPC KVM_REG_PPC_PMC7 32 2372 PPC KVM_REG_PPC_PMC8 32 2373 PPC KVM_REG_PPC_FPR0 64 2374 ... 2375 PPC KVM_REG_PPC_FPR31 64 2376 PPC KVM_REG_PPC_VR0 128 2377 ... 2378 PPC KVM_REG_PPC_VR31 128 2379 PPC KVM_REG_PPC_VSR0 128 2380 ... 2381 PPC KVM_REG_PPC_VSR31 128 2382 PPC KVM_REG_PPC_FPSCR 64 2383 PPC KVM_REG_PPC_VSCR 32 2384 PPC KVM_REG_PPC_VPA_ADDR 64 2385 PPC KVM_REG_PPC_VPA_SLB 128 2386 PPC KVM_REG_PPC_VPA_DTL 128 2387 PPC KVM_REG_PPC_EPCR 32 2388 PPC KVM_REG_PPC_EPR 32 2389 PPC KVM_REG_PPC_TCR 32 2390 PPC KVM_REG_PPC_TSR 32 2391 PPC KVM_REG_PPC_OR_TSR 32 2392 PPC KVM_REG_PPC_CLEAR_TSR 32 2393 PPC KVM_REG_PPC_MAS0 32 2394 PPC KVM_REG_PPC_MAS1 32 2395 PPC KVM_REG_PPC_MAS2 64 2396 PPC KVM_REG_PPC_MAS7_3 64 2397 PPC KVM_REG_PPC_MAS4 32 2398 PPC KVM_REG_PPC_MAS6 32 2399 PPC KVM_REG_PPC_MMUCFG 32 2400 PPC KVM_REG_PPC_TLB0CFG 32 2401 PPC KVM_REG_PPC_TLB1CFG 32 2402 PPC KVM_REG_PPC_TLB2CFG 32 2403 PPC KVM_REG_PPC_TLB3CFG 32 2404 PPC KVM_REG_PPC_TLB0PS 32 2405 PPC KVM_REG_PPC_TLB1PS 32 2406 PPC KVM_REG_PPC_TLB2PS 32 2407 PPC KVM_REG_PPC_TLB3PS 32 2408 PPC KVM_REG_PPC_EPTCFG 32 2409 PPC KVM_REG_PPC_ICP_STATE 64 2410 PPC KVM_REG_PPC_VP_STATE 128 2411 PPC KVM_REG_PPC_TB_OFFSET 64 2412 PPC KVM_REG_PPC_SPMC1 32 2413 PPC KVM_REG_PPC_SPMC2 32 2414 PPC KVM_REG_PPC_IAMR 64 2415 PPC KVM_REG_PPC_TFHAR 64 2416 PPC KVM_REG_PPC_TFIAR 64 2417 PPC KVM_REG_PPC_TEXASR 64 2418 PPC KVM_REG_PPC_FSCR 64 2419 PPC KVM_REG_PPC_PSPB 32 2420 PPC KVM_REG_PPC_EBBHR 64 2421 PPC KVM_REG_PPC_EBBRR 64 2422 PPC KVM_REG_PPC_BESCR 64 2423 PPC KVM_REG_PPC_TAR 64 2424 PPC KVM_REG_PPC_DPDES 64 2425 PPC KVM_REG_PPC_DAWR 64 2426 PPC KVM_REG_PPC_DAWRX 64 2427 PPC KVM_REG_PPC_CIABR 64 2428 PPC KVM_REG_PPC_IC 64 2429 PPC KVM_REG_PPC_VTB 64 2430 PPC KVM_REG_PPC_CSIGR 64 2431 PPC KVM_REG_PPC_TACR 64 2432 PPC KVM_REG_PPC_TCSCR 64 2433 PPC KVM_REG_PPC_PID 64 2434 PPC KVM_REG_PPC_ACOP 64 2435 PPC KVM_REG_PPC_VRSAVE 32 2436 PPC KVM_REG_PPC_LPCR 32 2437 PPC KVM_REG_PPC_LPCR_64 64 2438 PPC KVM_REG_PPC_PPR 64 2439 PPC KVM_REG_PPC_ARCH_COMPAT 32 2440 PPC KVM_REG_PPC_DABRX 32 2441 PPC KVM_REG_PPC_WORT 64 2442 PPC KVM_REG_PPC_SPRG9 64 2443 PPC KVM_REG_PPC_DBSR 32 2444 PPC KVM_REG_PPC_TIDR 64 2445 PPC KVM_REG_PPC_PSSCR 64 2446 PPC KVM_REG_PPC_DEC_EXPIRY 64 2447 PPC KVM_REG_PPC_PTCR 64 2448 PPC KVM_REG_PPC_HASHKEYR 64 2449 PPC KVM_REG_PPC_HASHPKEYR 64 2450 PPC KVM_REG_PPC_DAWR1 64 2451 PPC KVM_REG_PPC_DAWRX1 64 2452 PPC KVM_REG_PPC_DEXCR 64 2453 PPC KVM_REG_PPC_TM_GPR0 64 2454 ... 2455 PPC KVM_REG_PPC_TM_GPR31 64 2456 PPC KVM_REG_PPC_TM_VSR0 128 2457 ... 2458 PPC KVM_REG_PPC_TM_VSR63 128 2459 PPC KVM_REG_PPC_TM_CR 64 2460 PPC KVM_REG_PPC_TM_LR 64 2461 PPC KVM_REG_PPC_TM_CTR 64 2462 PPC KVM_REG_PPC_TM_FPSCR 64 2463 PPC KVM_REG_PPC_TM_AMR 64 2464 PPC KVM_REG_PPC_TM_PPR 64 2465 PPC KVM_REG_PPC_TM_VRSAVE 64 2466 PPC KVM_REG_PPC_TM_VSCR 32 2467 PPC KVM_REG_PPC_TM_DSCR 64 2468 PPC KVM_REG_PPC_TM_TAR 64 2469 PPC KVM_REG_PPC_TM_XER 64 2470 2471 MIPS KVM_REG_MIPS_R0 64 2472 ... 2473 MIPS KVM_REG_MIPS_R31 64 2474 MIPS KVM_REG_MIPS_HI 64 2475 MIPS KVM_REG_MIPS_LO 64 2476 MIPS KVM_REG_MIPS_PC 64 2477 MIPS KVM_REG_MIPS_CP0_INDEX 32 2478 MIPS KVM_REG_MIPS_CP0_ENTRYLO0 64 2479 MIPS KVM_REG_MIPS_CP0_ENTRYLO1 64 2480 MIPS KVM_REG_MIPS_CP0_CONTEXT 64 2481 MIPS KVM_REG_MIPS_CP0_CONTEXTCONFIG 32 2482 MIPS KVM_REG_MIPS_CP0_USERLOCAL 64 2483 MIPS KVM_REG_MIPS_CP0_XCONTEXTCONFIG 64 2484 MIPS KVM_REG_MIPS_CP0_PAGEMASK 32 2485 MIPS KVM_REG_MIPS_CP0_PAGEGRAIN 32 2486 MIPS KVM_REG_MIPS_CP0_SEGCTL0 64 2487 MIPS KVM_REG_MIPS_CP0_SEGCTL1 64 2488 MIPS KVM_REG_MIPS_CP0_SEGCTL2 64 2489 MIPS KVM_REG_MIPS_CP0_PWBASE 64 2490 MIPS KVM_REG_MIPS_CP0_PWFIELD 64 2491 MIPS KVM_REG_MIPS_CP0_PWSIZE 64 2492 MIPS KVM_REG_MIPS_CP0_WIRED 32 2493 MIPS KVM_REG_MIPS_CP0_PWCTL 32 2494 MIPS KVM_REG_MIPS_CP0_HWRENA 32 2495 MIPS KVM_REG_MIPS_CP0_BADVADDR 64 2496 MIPS KVM_REG_MIPS_CP0_BADINSTR 32 2497 MIPS KVM_REG_MIPS_CP0_BADINSTRP 32 2498 MIPS KVM_REG_MIPS_CP0_COUNT 32 2499 MIPS KVM_REG_MIPS_CP0_ENTRYHI 64 2500 MIPS KVM_REG_MIPS_CP0_COMPARE 32 2501 MIPS KVM_REG_MIPS_CP0_STATUS 32 2502 MIPS KVM_REG_MIPS_CP0_INTCTL 32 2503 MIPS KVM_REG_MIPS_CP0_CAUSE 32 2504 MIPS KVM_REG_MIPS_CP0_EPC 64 2505 MIPS KVM_REG_MIPS_CP0_PRID 32 2506 MIPS KVM_REG_MIPS_CP0_EBASE 64 2507 MIPS KVM_REG_MIPS_CP0_CONFIG 32 2508 MIPS KVM_REG_MIPS_CP0_CONFIG1 32 2509 MIPS KVM_REG_MIPS_CP0_CONFIG2 32 2510 MIPS KVM_REG_MIPS_CP0_CONFIG3 32 2511 MIPS KVM_REG_MIPS_CP0_CONFIG4 32 2512 MIPS KVM_REG_MIPS_CP0_CONFIG5 32 2513 MIPS KVM_REG_MIPS_CP0_CONFIG7 32 2514 MIPS KVM_REG_MIPS_CP0_XCONTEXT 64 2515 MIPS KVM_REG_MIPS_CP0_ERROREPC 64 2516 MIPS KVM_REG_MIPS_CP0_KSCRATCH1 64 2517 MIPS KVM_REG_MIPS_CP0_KSCRATCH2 64 2518 MIPS KVM_REG_MIPS_CP0_KSCRATCH3 64 2519 MIPS KVM_REG_MIPS_CP0_KSCRATCH4 64 2520 MIPS KVM_REG_MIPS_CP0_KSCRATCH5 64 2521 MIPS KVM_REG_MIPS_CP0_KSCRATCH6 64 2522 MIPS KVM_REG_MIPS_CP0_MAAR(0..63) 64 2523 MIPS KVM_REG_MIPS_COUNT_CTL 64 2524 MIPS KVM_REG_MIPS_COUNT_RESUME 64 2525 MIPS KVM_REG_MIPS_COUNT_HZ 64 2526 MIPS KVM_REG_MIPS_FPR_32(0..31) 32 2527 MIPS KVM_REG_MIPS_FPR_64(0..31) 64 2528 MIPS KVM_REG_MIPS_VEC_128(0..31) 128 2529 MIPS KVM_REG_MIPS_FCR_IR 32 2530 MIPS KVM_REG_MIPS_FCR_CSR 32 2531 MIPS KVM_REG_MIPS_MSA_IR 32 2532 MIPS KVM_REG_MIPS_MSA_CSR 32 2533 ======= =============================== ============ 2534 2535ARM registers are mapped using the lower 32 bits. The upper 16 of that 2536is the register group type, or coprocessor number: 2537 2538ARM core registers have the following id bit patterns:: 2539 2540 0x4020 0000 0010 <index into the kvm_regs struct:16> 2541 2542ARM 32-bit CP15 registers have the following id bit patterns:: 2543 2544 0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3> 2545 2546ARM 64-bit CP15 registers have the following id bit patterns:: 2547 2548 0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3> 2549 2550ARM CCSIDR registers are demultiplexed by CSSELR value:: 2551 2552 0x4020 0000 0011 00 <csselr:8> 2553 2554ARM 32-bit VFP control registers have the following id bit patterns:: 2555 2556 0x4020 0000 0012 1 <regno:12> 2557 2558ARM 64-bit FP registers have the following id bit patterns:: 2559 2560 0x4030 0000 0012 0 <regno:12> 2561 2562ARM firmware pseudo-registers have the following bit pattern:: 2563 2564 0x4030 0000 0014 <regno:16> 2565 2566 2567arm64 registers are mapped using the lower 32 bits. The upper 16 of 2568that is the register group type, or coprocessor number: 2569 2570arm64 core/FP-SIMD registers have the following id bit patterns. Note 2571that the size of the access is variable, as the kvm_regs structure 2572contains elements ranging from 32 to 128 bits. The index is a 32bit 2573value in the kvm_regs structure seen as a 32bit array:: 2574 2575 0x60x0 0000 0010 <index into the kvm_regs struct:16> 2576 2577Specifically: 2578 2579======================= ========= ===== ======================================= 2580 Encoding Register Bits kvm_regs member 2581======================= ========= ===== ======================================= 2582 0x6030 0000 0010 0000 X0 64 regs.regs[0] 2583 0x6030 0000 0010 0002 X1 64 regs.regs[1] 2584 ... 2585 0x6030 0000 0010 003c X30 64 regs.regs[30] 2586 0x6030 0000 0010 003e SP 64 regs.sp 2587 0x6030 0000 0010 0040 PC 64 regs.pc 2588 0x6030 0000 0010 0042 PSTATE 64 regs.pstate 2589 0x6030 0000 0010 0044 SP_EL1 64 sp_el1 2590 0x6030 0000 0010 0046 ELR_EL1 64 elr_el1 2591 0x6030 0000 0010 0048 SPSR_EL1 64 spsr[KVM_SPSR_EL1] (alias SPSR_SVC) 2592 0x6030 0000 0010 004a SPSR_ABT 64 spsr[KVM_SPSR_ABT] 2593 0x6030 0000 0010 004c SPSR_UND 64 spsr[KVM_SPSR_UND] 2594 0x6030 0000 0010 004e SPSR_IRQ 64 spsr[KVM_SPSR_IRQ] 2595 0x6030 0000 0010 0050 SPSR_FIQ 64 spsr[KVM_SPSR_FIQ] 2596 0x6040 0000 0010 0054 V0 128 fp_regs.vregs[0] [1]_ 2597 0x6040 0000 0010 0058 V1 128 fp_regs.vregs[1] [1]_ 2598 ... 2599 0x6040 0000 0010 00d0 V31 128 fp_regs.vregs[31] [1]_ 2600 0x6020 0000 0010 00d4 FPSR 32 fp_regs.fpsr 2601 0x6020 0000 0010 00d5 FPCR 32 fp_regs.fpcr 2602======================= ========= ===== ======================================= 2603 2604.. [1] These encodings are not accepted for SVE-enabled vcpus. See 2605 KVM_ARM_VCPU_INIT. 2606 2607 The equivalent register content can be accessed via bits [127:0] of 2608 the corresponding SVE Zn registers instead for vcpus that have SVE 2609 enabled (see below). 2610 2611arm64 CCSIDR registers are demultiplexed by CSSELR value:: 2612 2613 0x6020 0000 0011 00 <csselr:8> 2614 2615arm64 system registers have the following id bit patterns:: 2616 2617 0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3> 2618 2619.. warning:: 2620 2621 Two system register IDs do not follow the specified pattern. These 2622 are KVM_REG_ARM_TIMER_CVAL and KVM_REG_ARM_TIMER_CNT, which map to 2623 system registers CNTV_CVAL_EL0 and CNTVCT_EL0 respectively. These 2624 two had their values accidentally swapped, which means TIMER_CVAL is 2625 derived from the register encoding for CNTVCT_EL0 and TIMER_CNT is 2626 derived from the register encoding for CNTV_CVAL_EL0. As this is 2627 API, it must remain this way. 2628 2629arm64 firmware pseudo-registers have the following bit pattern:: 2630 2631 0x6030 0000 0014 <regno:16> 2632 2633arm64 SVE registers have the following bit patterns:: 2634 2635 0x6080 0000 0015 00 <n:5> <slice:5> Zn bits[2048*slice + 2047 : 2048*slice] 2636 0x6050 0000 0015 04 <n:4> <slice:5> Pn bits[256*slice + 255 : 256*slice] 2637 0x6050 0000 0015 060 <slice:5> FFR bits[256*slice + 255 : 256*slice] 2638 0x6060 0000 0015 ffff KVM_REG_ARM64_SVE_VLS pseudo-register 2639 2640Access to register IDs where 2048 * slice >= 128 * max_vq will fail with 2641ENOENT. max_vq is the vcpu's maximum supported vector length in 128-bit 2642quadwords: see [2]_ below. 2643 2644These registers are only accessible on vcpus for which SVE is enabled. 2645See KVM_ARM_VCPU_INIT for details. 2646 2647In addition, except for KVM_REG_ARM64_SVE_VLS, these registers are not 2648accessible until the vcpu's SVE configuration has been finalized 2649using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). See KVM_ARM_VCPU_INIT 2650and KVM_ARM_VCPU_FINALIZE for more information about this procedure. 2651 2652KVM_REG_ARM64_SVE_VLS is a pseudo-register that allows the set of vector 2653lengths supported by the vcpu to be discovered and configured by 2654userspace. When transferred to or from user memory via KVM_GET_ONE_REG 2655or KVM_SET_ONE_REG, the value of this register is of type 2656__u64[KVM_ARM64_SVE_VLS_WORDS], and encodes the set of vector lengths as 2657follows:: 2658 2659 __u64 vector_lengths[KVM_ARM64_SVE_VLS_WORDS]; 2660 2661 if (vq >= SVE_VQ_MIN && vq <= SVE_VQ_MAX && 2662 ((vector_lengths[(vq - KVM_ARM64_SVE_VQ_MIN) / 64] >> 2663 ((vq - KVM_ARM64_SVE_VQ_MIN) % 64)) & 1)) 2664 /* Vector length vq * 16 bytes supported */ 2665 else 2666 /* Vector length vq * 16 bytes not supported */ 2667 2668.. [2] The maximum value vq for which the above condition is true is 2669 max_vq. This is the maximum vector length available to the guest on 2670 this vcpu, and determines which register slices are visible through 2671 this ioctl interface. 2672 2673(See Documentation/arch/arm64/sve.rst for an explanation of the "vq" 2674nomenclature.) 2675 2676KVM_REG_ARM64_SVE_VLS is only accessible after KVM_ARM_VCPU_INIT. 2677KVM_ARM_VCPU_INIT initialises it to the best set of vector lengths that 2678the host supports. 2679 2680Userspace may subsequently modify it if desired until the vcpu's SVE 2681configuration is finalized using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). 2682 2683Apart from simply removing all vector lengths from the host set that 2684exceed some value, support for arbitrarily chosen sets of vector lengths 2685is hardware-dependent and may not be available. Attempting to configure 2686an invalid set of vector lengths via KVM_SET_ONE_REG will fail with 2687EINVAL. 2688 2689After the vcpu's SVE configuration is finalized, further attempts to 2690write this register will fail with EPERM. 2691 2692arm64 bitmap feature firmware pseudo-registers have the following bit pattern:: 2693 2694 0x6030 0000 0016 <regno:16> 2695 2696The bitmap feature firmware registers exposes the hypercall services that 2697are available for userspace to configure. The set bits corresponds to the 2698services that are available for the guests to access. By default, KVM 2699sets all the supported bits during VM initialization. The userspace can 2700discover the available services via KVM_GET_ONE_REG, and write back the 2701bitmap corresponding to the features that it wishes guests to see via 2702KVM_SET_ONE_REG. 2703 2704Note: These registers are immutable once any of the vCPUs of the VM has 2705run at least once. A KVM_SET_ONE_REG in such a scenario will return 2706a -EBUSY to userspace. 2707 2708(See Documentation/virt/kvm/arm/hypercalls.rst for more details.) 2709 2710 2711MIPS registers are mapped using the lower 32 bits. The upper 16 of that is 2712the register group type: 2713 2714MIPS core registers (see above) have the following id bit patterns:: 2715 2716 0x7030 0000 0000 <reg:16> 2717 2718MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit 2719patterns depending on whether they're 32-bit or 64-bit registers:: 2720 2721 0x7020 0000 0001 00 <reg:5> <sel:3> (32-bit) 2722 0x7030 0000 0001 00 <reg:5> <sel:3> (64-bit) 2723 2724Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_MIPS_CP0_ENTRYLO1 are the MIPS64 2725versions of the EntryLo registers regardless of the word size of the host 2726hardware, host kernel, guest, and whether XPA is present in the guest, i.e. 2727with the RI and XI bits (if they exist) in bits 63 and 62 respectively, and 2728the PFNX field starting at bit 30. 2729 2730MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) above) have the following id bit 2731patterns:: 2732 2733 0x7030 0000 0001 01 <reg:8> 2734 2735MIPS KVM control registers (see above) have the following id bit patterns:: 2736 2737 0x7030 0000 0002 <reg:16> 2738 2739MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following 2740id bit patterns depending on the size of the register being accessed. They are 2741always accessed according to the current guest FPU mode (Status.FR and 2742Config5.FRE), i.e. as the guest would see them, and they become unpredictable 2743if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector 2744registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they 2745overlap the FPU registers:: 2746 2747 0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers) 2748 0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers) 2749 0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers) 2750 2751MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the 2752following id bit patterns:: 2753 2754 0x7020 0000 0003 01 <0:3> <reg:5> 2755 2756MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the 2757following id bit patterns:: 2758 2759 0x7020 0000 0003 02 <0:3> <reg:5> 2760 2761RISC-V registers are mapped using the lower 32 bits. The upper 8 bits of 2762that is the register group type. 2763 2764RISC-V config registers are meant for configuring a Guest VCPU and it has 2765the following id bit patterns:: 2766 2767 0x8020 0000 01 <index into the kvm_riscv_config struct:24> (32bit Host) 2768 0x8030 0000 01 <index into the kvm_riscv_config struct:24> (64bit Host) 2769 2770Following are the RISC-V config registers: 2771 2772======================= ========= ============================================= 2773 Encoding Register Description 2774======================= ========= ============================================= 2775 0x80x0 0000 0100 0000 isa ISA feature bitmap of Guest VCPU 2776======================= ========= ============================================= 2777 2778The isa config register can be read anytime but can only be written before 2779a Guest VCPU runs. It will have ISA feature bits matching underlying host 2780set by default. 2781 2782RISC-V core registers represent the general execution state of a Guest VCPU 2783and it has the following id bit patterns:: 2784 2785 0x8020 0000 02 <index into the kvm_riscv_core struct:24> (32bit Host) 2786 0x8030 0000 02 <index into the kvm_riscv_core struct:24> (64bit Host) 2787 2788Following are the RISC-V core registers: 2789 2790======================= ========= ============================================= 2791 Encoding Register Description 2792======================= ========= ============================================= 2793 0x80x0 0000 0200 0000 regs.pc Program counter 2794 0x80x0 0000 0200 0001 regs.ra Return address 2795 0x80x0 0000 0200 0002 regs.sp Stack pointer 2796 0x80x0 0000 0200 0003 regs.gp Global pointer 2797 0x80x0 0000 0200 0004 regs.tp Task pointer 2798 0x80x0 0000 0200 0005 regs.t0 Caller saved register 0 2799 0x80x0 0000 0200 0006 regs.t1 Caller saved register 1 2800 0x80x0 0000 0200 0007 regs.t2 Caller saved register 2 2801 0x80x0 0000 0200 0008 regs.s0 Callee saved register 0 2802 0x80x0 0000 0200 0009 regs.s1 Callee saved register 1 2803 0x80x0 0000 0200 000a regs.a0 Function argument (or return value) 0 2804 0x80x0 0000 0200 000b regs.a1 Function argument (or return value) 1 2805 0x80x0 0000 0200 000c regs.a2 Function argument 2 2806 0x80x0 0000 0200 000d regs.a3 Function argument 3 2807 0x80x0 0000 0200 000e regs.a4 Function argument 4 2808 0x80x0 0000 0200 000f regs.a5 Function argument 5 2809 0x80x0 0000 0200 0010 regs.a6 Function argument 6 2810 0x80x0 0000 0200 0011 regs.a7 Function argument 7 2811 0x80x0 0000 0200 0012 regs.s2 Callee saved register 2 2812 0x80x0 0000 0200 0013 regs.s3 Callee saved register 3 2813 0x80x0 0000 0200 0014 regs.s4 Callee saved register 4 2814 0x80x0 0000 0200 0015 regs.s5 Callee saved register 5 2815 0x80x0 0000 0200 0016 regs.s6 Callee saved register 6 2816 0x80x0 0000 0200 0017 regs.s7 Callee saved register 7 2817 0x80x0 0000 0200 0018 regs.s8 Callee saved register 8 2818 0x80x0 0000 0200 0019 regs.s9 Callee saved register 9 2819 0x80x0 0000 0200 001a regs.s10 Callee saved register 10 2820 0x80x0 0000 0200 001b regs.s11 Callee saved register 11 2821 0x80x0 0000 0200 001c regs.t3 Caller saved register 3 2822 0x80x0 0000 0200 001d regs.t4 Caller saved register 4 2823 0x80x0 0000 0200 001e regs.t5 Caller saved register 5 2824 0x80x0 0000 0200 001f regs.t6 Caller saved register 6 2825 0x80x0 0000 0200 0020 mode Privilege mode (1 = S-mode or 0 = U-mode) 2826======================= ========= ============================================= 2827 2828RISC-V csr registers represent the supervisor mode control/status registers 2829of a Guest VCPU and it has the following id bit patterns:: 2830 2831 0x8020 0000 03 <index into the kvm_riscv_csr struct:24> (32bit Host) 2832 0x8030 0000 03 <index into the kvm_riscv_csr struct:24> (64bit Host) 2833 2834Following are the RISC-V csr registers: 2835 2836======================= ========= ============================================= 2837 Encoding Register Description 2838======================= ========= ============================================= 2839 0x80x0 0000 0300 0000 sstatus Supervisor status 2840 0x80x0 0000 0300 0001 sie Supervisor interrupt enable 2841 0x80x0 0000 0300 0002 stvec Supervisor trap vector base 2842 0x80x0 0000 0300 0003 sscratch Supervisor scratch register 2843 0x80x0 0000 0300 0004 sepc Supervisor exception program counter 2844 0x80x0 0000 0300 0005 scause Supervisor trap cause 2845 0x80x0 0000 0300 0006 stval Supervisor bad address or instruction 2846 0x80x0 0000 0300 0007 sip Supervisor interrupt pending 2847 0x80x0 0000 0300 0008 satp Supervisor address translation and protection 2848======================= ========= ============================================= 2849 2850RISC-V timer registers represent the timer state of a Guest VCPU and it has 2851the following id bit patterns:: 2852 2853 0x8030 0000 04 <index into the kvm_riscv_timer struct:24> 2854 2855Following are the RISC-V timer registers: 2856 2857======================= ========= ============================================= 2858 Encoding Register Description 2859======================= ========= ============================================= 2860 0x8030 0000 0400 0000 frequency Time base frequency (read-only) 2861 0x8030 0000 0400 0001 time Time value visible to Guest 2862 0x8030 0000 0400 0002 compare Time compare programmed by Guest 2863 0x8030 0000 0400 0003 state Time compare state (1 = ON or 0 = OFF) 2864======================= ========= ============================================= 2865 2866RISC-V F-extension registers represent the single precision floating point 2867state of a Guest VCPU and it has the following id bit patterns:: 2868 2869 0x8020 0000 05 <index into the __riscv_f_ext_state struct:24> 2870 2871Following are the RISC-V F-extension registers: 2872 2873======================= ========= ============================================= 2874 Encoding Register Description 2875======================= ========= ============================================= 2876 0x8020 0000 0500 0000 f[0] Floating point register 0 2877 ... 2878 0x8020 0000 0500 001f f[31] Floating point register 31 2879 0x8020 0000 0500 0020 fcsr Floating point control and status register 2880======================= ========= ============================================= 2881 2882RISC-V D-extension registers represent the double precision floating point 2883state of a Guest VCPU and it has the following id bit patterns:: 2884 2885 0x8020 0000 06 <index into the __riscv_d_ext_state struct:24> (fcsr) 2886 0x8030 0000 06 <index into the __riscv_d_ext_state struct:24> (non-fcsr) 2887 2888Following are the RISC-V D-extension registers: 2889 2890======================= ========= ============================================= 2891 Encoding Register Description 2892======================= ========= ============================================= 2893 0x8030 0000 0600 0000 f[0] Floating point register 0 2894 ... 2895 0x8030 0000 0600 001f f[31] Floating point register 31 2896 0x8020 0000 0600 0020 fcsr Floating point control and status register 2897======================= ========= ============================================= 2898 2899LoongArch registers are mapped using the lower 32 bits. The upper 16 bits of 2900that is the register group type. 2901 2902LoongArch csr registers are used to control guest cpu or get status of guest 2903cpu, and they have the following id bit patterns:: 2904 2905 0x9030 0000 0001 00 <reg:5> <sel:3> (64-bit) 2906 2907LoongArch KVM control registers are used to implement some new defined functions 2908such as set vcpu counter or reset vcpu, and they have the following id bit patterns:: 2909 2910 0x9030 0000 0002 <reg:16> 2911 2912 29134.69 KVM_GET_ONE_REG 2914-------------------- 2915 2916:Capability: KVM_CAP_ONE_REG 2917:Architectures: all 2918:Type: vcpu ioctl 2919:Parameters: struct kvm_one_reg (in and out) 2920:Returns: 0 on success, negative value on failure 2921 2922Errors include: 2923 2924 ======== ============================================================ 2925 ENOENT no such register 2926 EINVAL invalid register ID, or no such register or used with VMs in 2927 protected virtualization mode on s390 2928 EPERM (arm64) register access not allowed before vcpu finalization 2929 ======== ============================================================ 2930 2931(These error codes are indicative only: do not rely on a specific error 2932code being returned in a specific situation.) 2933 2934This ioctl allows to receive the value of a single register implemented 2935in a vcpu. The register to read is indicated by the "id" field of the 2936kvm_one_reg struct passed in. On success, the register value can be found 2937at the memory location pointed to by "addr". 2938 2939The list of registers accessible using this interface is identical to the 2940list in 4.68. 2941 2942 29434.70 KVM_KVMCLOCK_CTRL 2944---------------------- 2945 2946:Capability: KVM_CAP_KVMCLOCK_CTRL 2947:Architectures: Any that implement pvclocks (currently x86 only) 2948:Type: vcpu ioctl 2949:Parameters: None 2950:Returns: 0 on success, -1 on error 2951 2952This ioctl sets a flag accessible to the guest indicating that the specified 2953vCPU has been paused by the host userspace. 2954 2955The host will set a flag in the pvclock structure that is checked from the 2956soft lockup watchdog. The flag is part of the pvclock structure that is 2957shared between guest and host, specifically the second bit of the flags 2958field of the pvclock_vcpu_time_info structure. It will be set exclusively by 2959the host and read/cleared exclusively by the guest. The guest operation of 2960checking and clearing the flag must be an atomic operation so 2961load-link/store-conditional, or equivalent must be used. There are two cases 2962where the guest will clear the flag: when the soft lockup watchdog timer resets 2963itself or when a soft lockup is detected. This ioctl can be called any time 2964after pausing the vcpu, but before it is resumed. 2965 2966 29674.71 KVM_SIGNAL_MSI 2968------------------- 2969 2970:Capability: KVM_CAP_SIGNAL_MSI 2971:Architectures: x86 arm64 2972:Type: vm ioctl 2973:Parameters: struct kvm_msi (in) 2974:Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error 2975 2976Directly inject a MSI message. Only valid with in-kernel irqchip that handles 2977MSI messages. 2978 2979:: 2980 2981 struct kvm_msi { 2982 __u32 address_lo; 2983 __u32 address_hi; 2984 __u32 data; 2985 __u32 flags; 2986 __u32 devid; 2987 __u8 pad[12]; 2988 }; 2989 2990flags: 2991 KVM_MSI_VALID_DEVID: devid contains a valid value. The per-VM 2992 KVM_CAP_MSI_DEVID capability advertises the requirement to provide 2993 the device ID. If this capability is not available, userspace 2994 should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail. 2995 2996If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier 2997for the device that wrote the MSI message. For PCI, this is usually a 2998BDF identifier in the lower 16 bits. 2999 3000On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS 3001feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled, 3002address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of 3003address_hi must be zero. 3004 3005 30064.71 KVM_CREATE_PIT2 3007-------------------- 3008 3009:Capability: KVM_CAP_PIT2 3010:Architectures: x86 3011:Type: vm ioctl 3012:Parameters: struct kvm_pit_config (in) 3013:Returns: 0 on success, -1 on error 3014 3015Creates an in-kernel device model for the i8254 PIT. This call is only valid 3016after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following 3017parameters have to be passed:: 3018 3019 struct kvm_pit_config { 3020 __u32 flags; 3021 __u32 pad[15]; 3022 }; 3023 3024Valid flags are:: 3025 3026 #define KVM_PIT_SPEAKER_DUMMY 1 /* emulate speaker port stub */ 3027 3028PIT timer interrupts may use a per-VM kernel thread for injection. If it 3029exists, this thread will have a name of the following pattern:: 3030 3031 kvm-pit/<owner-process-pid> 3032 3033When running a guest with elevated priorities, the scheduling parameters of 3034this thread may have to be adjusted accordingly. 3035 3036This IOCTL replaces the obsolete KVM_CREATE_PIT. 3037 3038 30394.72 KVM_GET_PIT2 3040----------------- 3041 3042:Capability: KVM_CAP_PIT_STATE2 3043:Architectures: x86 3044:Type: vm ioctl 3045:Parameters: struct kvm_pit_state2 (out) 3046:Returns: 0 on success, -1 on error 3047 3048Retrieves the state of the in-kernel PIT model. Only valid after 3049KVM_CREATE_PIT2. The state is returned in the following structure:: 3050 3051 struct kvm_pit_state2 { 3052 struct kvm_pit_channel_state channels[3]; 3053 __u32 flags; 3054 __u32 reserved[9]; 3055 }; 3056 3057Valid flags are:: 3058 3059 /* disable PIT in HPET legacy mode */ 3060 #define KVM_PIT_FLAGS_HPET_LEGACY 0x00000001 3061 /* speaker port data bit enabled */ 3062 #define KVM_PIT_FLAGS_SPEAKER_DATA_ON 0x00000002 3063 3064This IOCTL replaces the obsolete KVM_GET_PIT. 3065 3066 30674.73 KVM_SET_PIT2 3068----------------- 3069 3070:Capability: KVM_CAP_PIT_STATE2 3071:Architectures: x86 3072:Type: vm ioctl 3073:Parameters: struct kvm_pit_state2 (in) 3074:Returns: 0 on success, -1 on error 3075 3076Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2. 3077See KVM_GET_PIT2 for details on struct kvm_pit_state2. 3078 3079This IOCTL replaces the obsolete KVM_SET_PIT. 3080 3081 30824.74 KVM_PPC_GET_SMMU_INFO 3083-------------------------- 3084 3085:Capability: KVM_CAP_PPC_GET_SMMU_INFO 3086:Architectures: powerpc 3087:Type: vm ioctl 3088:Parameters: None 3089:Returns: 0 on success, -1 on error 3090 3091This populates and returns a structure describing the features of 3092the "Server" class MMU emulation supported by KVM. 3093This can in turn be used by userspace to generate the appropriate 3094device-tree properties for the guest operating system. 3095 3096The structure contains some global information, followed by an 3097array of supported segment page sizes:: 3098 3099 struct kvm_ppc_smmu_info { 3100 __u64 flags; 3101 __u32 slb_size; 3102 __u32 pad; 3103 struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ]; 3104 }; 3105 3106The supported flags are: 3107 3108 - KVM_PPC_PAGE_SIZES_REAL: 3109 When that flag is set, guest page sizes must "fit" the backing 3110 store page sizes. When not set, any page size in the list can 3111 be used regardless of how they are backed by userspace. 3112 3113 - KVM_PPC_1T_SEGMENTS 3114 The emulated MMU supports 1T segments in addition to the 3115 standard 256M ones. 3116 3117 - KVM_PPC_NO_HASH 3118 This flag indicates that HPT guests are not supported by KVM, 3119 thus all guests must use radix MMU mode. 3120 3121The "slb_size" field indicates how many SLB entries are supported 3122 3123The "sps" array contains 8 entries indicating the supported base 3124page sizes for a segment in increasing order. Each entry is defined 3125as follow:: 3126 3127 struct kvm_ppc_one_seg_page_size { 3128 __u32 page_shift; /* Base page shift of segment (or 0) */ 3129 __u32 slb_enc; /* SLB encoding for BookS */ 3130 struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ]; 3131 }; 3132 3133An entry with a "page_shift" of 0 is unused. Because the array is 3134organized in increasing order, a lookup can stop when encountering 3135such an entry. 3136 3137The "slb_enc" field provides the encoding to use in the SLB for the 3138page size. The bits are in positions such as the value can directly 3139be OR'ed into the "vsid" argument of the slbmte instruction. 3140 3141The "enc" array is a list which for each of those segment base page 3142size provides the list of supported actual page sizes (which can be 3143only larger or equal to the base page size), along with the 3144corresponding encoding in the hash PTE. Similarly, the array is 31458 entries sorted by increasing sizes and an entry with a "0" shift 3146is an empty entry and a terminator:: 3147 3148 struct kvm_ppc_one_page_size { 3149 __u32 page_shift; /* Page shift (or 0) */ 3150 __u32 pte_enc; /* Encoding in the HPTE (>>12) */ 3151 }; 3152 3153The "pte_enc" field provides a value that can OR'ed into the hash 3154PTE's RPN field (ie, it needs to be shifted left by 12 to OR it 3155into the hash PTE second double word). 3156 31574.75 KVM_IRQFD 3158-------------- 3159 3160:Capability: KVM_CAP_IRQFD 3161:Architectures: x86 s390 arm64 3162:Type: vm ioctl 3163:Parameters: struct kvm_irqfd (in) 3164:Returns: 0 on success, -1 on error 3165 3166Allows setting an eventfd to directly trigger a guest interrupt. 3167kvm_irqfd.fd specifies the file descriptor to use as the eventfd and 3168kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When 3169an event is triggered on the eventfd, an interrupt is injected into 3170the guest using the specified gsi pin. The irqfd is removed using 3171the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd 3172and kvm_irqfd.gsi. 3173 3174With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify 3175mechanism allowing emulation of level-triggered, irqfd-based 3176interrupts. When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an 3177additional eventfd in the kvm_irqfd.resamplefd field. When operating 3178in resample mode, posting of an interrupt through kvm_irq.fd asserts 3179the specified gsi in the irqchip. When the irqchip is resampled, such 3180as from an EOI, the gsi is de-asserted and the user is notified via 3181kvm_irqfd.resamplefd. It is the user's responsibility to re-queue 3182the interrupt if the device making use of it still requires service. 3183Note that closing the resamplefd is not sufficient to disable the 3184irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment 3185and need not be specified with KVM_IRQFD_FLAG_DEASSIGN. 3186 3187On arm64, gsi routing being supported, the following can happen: 3188 3189- in case no routing entry is associated to this gsi, injection fails 3190- in case the gsi is associated to an irqchip routing entry, 3191 irqchip.pin + 32 corresponds to the injected SPI ID. 3192- in case the gsi is associated to an MSI routing entry, the MSI 3193 message and device ID are translated into an LPI (support restricted 3194 to GICv3 ITS in-kernel emulation). 3195 31964.76 KVM_PPC_ALLOCATE_HTAB 3197-------------------------- 3198 3199:Capability: KVM_CAP_PPC_ALLOC_HTAB 3200:Architectures: powerpc 3201:Type: vm ioctl 3202:Parameters: Pointer to u32 containing hash table order (in/out) 3203:Returns: 0 on success, -1 on error 3204 3205This requests the host kernel to allocate an MMU hash table for a 3206guest using the PAPR paravirtualization interface. This only does 3207anything if the kernel is configured to use the Book 3S HV style of 3208virtualization. Otherwise the capability doesn't exist and the ioctl 3209returns an ENOTTY error. The rest of this description assumes Book 3S 3210HV. 3211 3212There must be no vcpus running when this ioctl is called; if there 3213are, it will do nothing and return an EBUSY error. 3214 3215The parameter is a pointer to a 32-bit unsigned integer variable 3216containing the order (log base 2) of the desired size of the hash 3217table, which must be between 18 and 46. On successful return from the 3218ioctl, the value will not be changed by the kernel. 3219 3220If no hash table has been allocated when any vcpu is asked to run 3221(with the KVM_RUN ioctl), the host kernel will allocate a 3222default-sized hash table (16 MB). 3223 3224If this ioctl is called when a hash table has already been allocated, 3225with a different order from the existing hash table, the existing hash 3226table will be freed and a new one allocated. If this is ioctl is 3227called when a hash table has already been allocated of the same order 3228as specified, the kernel will clear out the existing hash table (zero 3229all HPTEs). In either case, if the guest is using the virtualized 3230real-mode area (VRMA) facility, the kernel will re-create the VMRA 3231HPTEs on the next KVM_RUN of any vcpu. 3232 32334.77 KVM_S390_INTERRUPT 3234----------------------- 3235 3236:Capability: basic 3237:Architectures: s390 3238:Type: vm ioctl, vcpu ioctl 3239:Parameters: struct kvm_s390_interrupt (in) 3240:Returns: 0 on success, -1 on error 3241 3242Allows to inject an interrupt to the guest. Interrupts can be floating 3243(vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type. 3244 3245Interrupt parameters are passed via kvm_s390_interrupt:: 3246 3247 struct kvm_s390_interrupt { 3248 __u32 type; 3249 __u32 parm; 3250 __u64 parm64; 3251 }; 3252 3253type can be one of the following: 3254 3255KVM_S390_SIGP_STOP (vcpu) 3256 - sigp stop; optional flags in parm 3257KVM_S390_PROGRAM_INT (vcpu) 3258 - program check; code in parm 3259KVM_S390_SIGP_SET_PREFIX (vcpu) 3260 - sigp set prefix; prefix address in parm 3261KVM_S390_RESTART (vcpu) 3262 - restart 3263KVM_S390_INT_CLOCK_COMP (vcpu) 3264 - clock comparator interrupt 3265KVM_S390_INT_CPU_TIMER (vcpu) 3266 - CPU timer interrupt 3267KVM_S390_INT_VIRTIO (vm) 3268 - virtio external interrupt; external interrupt 3269 parameters in parm and parm64 3270KVM_S390_INT_SERVICE (vm) 3271 - sclp external interrupt; sclp parameter in parm 3272KVM_S390_INT_EMERGENCY (vcpu) 3273 - sigp emergency; source cpu in parm 3274KVM_S390_INT_EXTERNAL_CALL (vcpu) 3275 - sigp external call; source cpu in parm 3276KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm) 3277 - compound value to indicate an 3278 I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel); 3279 I/O interruption parameters in parm (subchannel) and parm64 (intparm, 3280 interruption subclass) 3281KVM_S390_MCHK (vm, vcpu) 3282 - machine check interrupt; cr 14 bits in parm, machine check interrupt 3283 code in parm64 (note that machine checks needing further payload are not 3284 supported by this ioctl) 3285 3286This is an asynchronous vcpu ioctl and can be invoked from any thread. 3287 32884.78 KVM_PPC_GET_HTAB_FD 3289------------------------ 3290 3291:Capability: KVM_CAP_PPC_HTAB_FD 3292:Architectures: powerpc 3293:Type: vm ioctl 3294:Parameters: Pointer to struct kvm_get_htab_fd (in) 3295:Returns: file descriptor number (>= 0) on success, -1 on error 3296 3297This returns a file descriptor that can be used either to read out the 3298entries in the guest's hashed page table (HPT), or to write entries to 3299initialize the HPT. The returned fd can only be written to if the 3300KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and 3301can only be read if that bit is clear. The argument struct looks like 3302this:: 3303 3304 /* For KVM_PPC_GET_HTAB_FD */ 3305 struct kvm_get_htab_fd { 3306 __u64 flags; 3307 __u64 start_index; 3308 __u64 reserved[2]; 3309 }; 3310 3311 /* Values for kvm_get_htab_fd.flags */ 3312 #define KVM_GET_HTAB_BOLTED_ONLY ((__u64)0x1) 3313 #define KVM_GET_HTAB_WRITE ((__u64)0x2) 3314 3315The 'start_index' field gives the index in the HPT of the entry at 3316which to start reading. It is ignored when writing. 3317 3318Reads on the fd will initially supply information about all 3319"interesting" HPT entries. Interesting entries are those with the 3320bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise 3321all entries. When the end of the HPT is reached, the read() will 3322return. If read() is called again on the fd, it will start again from 3323the beginning of the HPT, but will only return HPT entries that have 3324changed since they were last read. 3325 3326Data read or written is structured as a header (8 bytes) followed by a 3327series of valid HPT entries (16 bytes) each. The header indicates how 3328many valid HPT entries there are and how many invalid entries follow 3329the valid entries. The invalid entries are not represented explicitly 3330in the stream. The header format is:: 3331 3332 struct kvm_get_htab_header { 3333 __u32 index; 3334 __u16 n_valid; 3335 __u16 n_invalid; 3336 }; 3337 3338Writes to the fd create HPT entries starting at the index given in the 3339header; first 'n_valid' valid entries with contents from the data 3340written, then 'n_invalid' invalid entries, invalidating any previously 3341valid entries found. 3342 33434.79 KVM_CREATE_DEVICE 3344---------------------- 3345 3346:Capability: KVM_CAP_DEVICE_CTRL 3347:Architectures: all 3348:Type: vm ioctl 3349:Parameters: struct kvm_create_device (in/out) 3350:Returns: 0 on success, -1 on error 3351 3352Errors: 3353 3354 ====== ======================================================= 3355 ENODEV The device type is unknown or unsupported 3356 EEXIST Device already created, and this type of device may not 3357 be instantiated multiple times 3358 ====== ======================================================= 3359 3360 Other error conditions may be defined by individual device types or 3361 have their standard meanings. 3362 3363Creates an emulated device in the kernel. The file descriptor returned 3364in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR. 3365 3366If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the 3367device type is supported (not necessarily whether it can be created 3368in the current vm). 3369 3370Individual devices should not define flags. Attributes should be used 3371for specifying any behavior that is not implied by the device type 3372number. 3373 3374:: 3375 3376 struct kvm_create_device { 3377 __u32 type; /* in: KVM_DEV_TYPE_xxx */ 3378 __u32 fd; /* out: device handle */ 3379 __u32 flags; /* in: KVM_CREATE_DEVICE_xxx */ 3380 }; 3381 33824.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR 3383-------------------------------------------- 3384 3385:Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device, 3386 KVM_CAP_VCPU_ATTRIBUTES for vcpu device 3387 KVM_CAP_SYS_ATTRIBUTES for system (/dev/kvm) device (no set) 3388:Architectures: x86, arm64, s390 3389:Type: device ioctl, vm ioctl, vcpu ioctl 3390:Parameters: struct kvm_device_attr 3391:Returns: 0 on success, -1 on error 3392 3393Errors: 3394 3395 ===== ============================================================= 3396 ENXIO The group or attribute is unknown/unsupported for this device 3397 or hardware support is missing. 3398 EPERM The attribute cannot (currently) be accessed this way 3399 (e.g. read-only attribute, or attribute that only makes 3400 sense when the device is in a different state) 3401 ===== ============================================================= 3402 3403 Other error conditions may be defined by individual device types. 3404 3405Gets/sets a specified piece of device configuration and/or state. The 3406semantics are device-specific. See individual device documentation in 3407the "devices" directory. As with ONE_REG, the size of the data 3408transferred is defined by the particular attribute. 3409 3410:: 3411 3412 struct kvm_device_attr { 3413 __u32 flags; /* no flags currently defined */ 3414 __u32 group; /* device-defined */ 3415 __u64 attr; /* group-defined */ 3416 __u64 addr; /* userspace address of attr data */ 3417 }; 3418 34194.81 KVM_HAS_DEVICE_ATTR 3420------------------------ 3421 3422:Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device, 3423 KVM_CAP_VCPU_ATTRIBUTES for vcpu device 3424 KVM_CAP_SYS_ATTRIBUTES for system (/dev/kvm) device 3425:Type: device ioctl, vm ioctl, vcpu ioctl 3426:Parameters: struct kvm_device_attr 3427:Returns: 0 on success, -1 on error 3428 3429Errors: 3430 3431 ===== ============================================================= 3432 ENXIO The group or attribute is unknown/unsupported for this device 3433 or hardware support is missing. 3434 ===== ============================================================= 3435 3436Tests whether a device supports a particular attribute. A successful 3437return indicates the attribute is implemented. It does not necessarily 3438indicate that the attribute can be read or written in the device's 3439current state. "addr" is ignored. 3440 3441.. _KVM_ARM_VCPU_INIT: 3442 34434.82 KVM_ARM_VCPU_INIT 3444---------------------- 3445 3446:Capability: basic 3447:Architectures: arm64 3448:Type: vcpu ioctl 3449:Parameters: struct kvm_vcpu_init (in) 3450:Returns: 0 on success; -1 on error 3451 3452Errors: 3453 3454 ====== ================================================================= 3455 EINVAL the target is unknown, or the combination of features is invalid. 3456 ENOENT a features bit specified is unknown. 3457 ====== ================================================================= 3458 3459This tells KVM what type of CPU to present to the guest, and what 3460optional features it should have. This will cause a reset of the cpu 3461registers to their initial values. If this is not called, KVM_RUN will 3462return ENOEXEC for that vcpu. 3463 3464The initial values are defined as: 3465 - Processor state: 3466 * AArch64: EL1h, D, A, I and F bits set. All other bits 3467 are cleared. 3468 * AArch32: SVC, A, I and F bits set. All other bits are 3469 cleared. 3470 - General Purpose registers, including PC and SP: set to 0 3471 - FPSIMD/NEON registers: set to 0 3472 - SVE registers: set to 0 3473 - System registers: Reset to their architecturally defined 3474 values as for a warm reset to EL1 (resp. SVC) 3475 3476Note that because some registers reflect machine topology, all vcpus 3477should be created before this ioctl is invoked. 3478 3479Userspace can call this function multiple times for a given vcpu, including 3480after the vcpu has been run. This will reset the vcpu to its initial 3481state. All calls to this function after the initial call must use the same 3482target and same set of feature flags, otherwise EINVAL will be returned. 3483 3484Possible features: 3485 3486 - KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state. 3487 Depends on KVM_CAP_ARM_PSCI. If not set, the CPU will be powered on 3488 and execute guest code when KVM_RUN is called. 3489 - KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode. 3490 Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only). 3491 - KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 (or a future revision 3492 backward compatible with v0.2) for the CPU. 3493 Depends on KVM_CAP_ARM_PSCI_0_2. 3494 - KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU. 3495 Depends on KVM_CAP_ARM_PMU_V3. 3496 3497 - KVM_ARM_VCPU_PTRAUTH_ADDRESS: Enables Address Pointer authentication 3498 for arm64 only. 3499 Depends on KVM_CAP_ARM_PTRAUTH_ADDRESS. 3500 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are 3501 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and 3502 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be 3503 requested. 3504 3505 - KVM_ARM_VCPU_PTRAUTH_GENERIC: Enables Generic Pointer authentication 3506 for arm64 only. 3507 Depends on KVM_CAP_ARM_PTRAUTH_GENERIC. 3508 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are 3509 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and 3510 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be 3511 requested. 3512 3513 - KVM_ARM_VCPU_SVE: Enables SVE for the CPU (arm64 only). 3514 Depends on KVM_CAP_ARM_SVE. 3515 Requires KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3516 3517 * After KVM_ARM_VCPU_INIT: 3518 3519 - KVM_REG_ARM64_SVE_VLS may be read using KVM_GET_ONE_REG: the 3520 initial value of this pseudo-register indicates the best set of 3521 vector lengths possible for a vcpu on this host. 3522 3523 * Before KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3524 3525 - KVM_RUN and KVM_GET_REG_LIST are not available; 3526 3527 - KVM_GET_ONE_REG and KVM_SET_ONE_REG cannot be used to access 3528 the scalable architectural SVE registers 3529 KVM_REG_ARM64_SVE_ZREG(), KVM_REG_ARM64_SVE_PREG() or 3530 KVM_REG_ARM64_SVE_FFR; 3531 3532 - KVM_REG_ARM64_SVE_VLS may optionally be written using 3533 KVM_SET_ONE_REG, to modify the set of vector lengths available 3534 for the vcpu. 3535 3536 * After KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3537 3538 - the KVM_REG_ARM64_SVE_VLS pseudo-register is immutable, and can 3539 no longer be written using KVM_SET_ONE_REG. 3540 35414.83 KVM_ARM_PREFERRED_TARGET 3542----------------------------- 3543 3544:Capability: basic 3545:Architectures: arm64 3546:Type: vm ioctl 3547:Parameters: struct kvm_vcpu_init (out) 3548:Returns: 0 on success; -1 on error 3549 3550Errors: 3551 3552 ====== ========================================== 3553 ENODEV no preferred target available for the host 3554 ====== ========================================== 3555 3556This queries KVM for preferred CPU target type which can be emulated 3557by KVM on underlying host. 3558 3559The ioctl returns struct kvm_vcpu_init instance containing information 3560about preferred CPU target type and recommended features for it. The 3561kvm_vcpu_init->features bitmap returned will have feature bits set if 3562the preferred target recommends setting these features, but this is 3563not mandatory. 3564 3565The information returned by this ioctl can be used to prepare an instance 3566of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in 3567VCPU matching underlying host. 3568 3569 35704.84 KVM_GET_REG_LIST 3571--------------------- 3572 3573:Capability: basic 3574:Architectures: arm64, mips, riscv 3575:Type: vcpu ioctl 3576:Parameters: struct kvm_reg_list (in/out) 3577:Returns: 0 on success; -1 on error 3578 3579Errors: 3580 3581 ===== ============================================================== 3582 E2BIG the reg index list is too big to fit in the array specified by 3583 the user (the number required will be written into n). 3584 ===== ============================================================== 3585 3586:: 3587 3588 struct kvm_reg_list { 3589 __u64 n; /* number of registers in reg[] */ 3590 __u64 reg[0]; 3591 }; 3592 3593This ioctl returns the guest registers that are supported for the 3594KVM_GET_ONE_REG/KVM_SET_ONE_REG calls. 3595 3596 35974.85 KVM_ARM_SET_DEVICE_ADDR (deprecated) 3598----------------------------------------- 3599 3600:Capability: KVM_CAP_ARM_SET_DEVICE_ADDR 3601:Architectures: arm64 3602:Type: vm ioctl 3603:Parameters: struct kvm_arm_device_address (in) 3604:Returns: 0 on success, -1 on error 3605 3606Errors: 3607 3608 ====== ============================================ 3609 ENODEV The device id is unknown 3610 ENXIO Device not supported on current system 3611 EEXIST Address already set 3612 E2BIG Address outside guest physical address space 3613 EBUSY Address overlaps with other device range 3614 ====== ============================================ 3615 3616:: 3617 3618 struct kvm_arm_device_addr { 3619 __u64 id; 3620 __u64 addr; 3621 }; 3622 3623Specify a device address in the guest's physical address space where guests 3624can access emulated or directly exposed devices, which the host kernel needs 3625to know about. The id field is an architecture specific identifier for a 3626specific device. 3627 3628arm64 divides the id field into two parts, a device id and an 3629address type id specific to the individual device:: 3630 3631 bits: | 63 ... 32 | 31 ... 16 | 15 ... 0 | 3632 field: | 0x00000000 | device id | addr type id | 3633 3634arm64 currently only require this when using the in-kernel GIC 3635support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2 3636as the device id. When setting the base address for the guest's 3637mapping of the VGIC virtual CPU and distributor interface, the ioctl 3638must be called after calling KVM_CREATE_IRQCHIP, but before calling 3639KVM_RUN on any of the VCPUs. Calling this ioctl twice for any of the 3640base addresses will return -EEXIST. 3641 3642Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API 3643should be used instead. 3644 3645 36464.86 KVM_PPC_RTAS_DEFINE_TOKEN 3647------------------------------ 3648 3649:Capability: KVM_CAP_PPC_RTAS 3650:Architectures: ppc 3651:Type: vm ioctl 3652:Parameters: struct kvm_rtas_token_args 3653:Returns: 0 on success, -1 on error 3654 3655Defines a token value for a RTAS (Run Time Abstraction Services) 3656service in order to allow it to be handled in the kernel. The 3657argument struct gives the name of the service, which must be the name 3658of a service that has a kernel-side implementation. If the token 3659value is non-zero, it will be associated with that service, and 3660subsequent RTAS calls by the guest specifying that token will be 3661handled by the kernel. If the token value is 0, then any token 3662associated with the service will be forgotten, and subsequent RTAS 3663calls by the guest for that service will be passed to userspace to be 3664handled. 3665 36664.87 KVM_SET_GUEST_DEBUG 3667------------------------ 3668 3669:Capability: KVM_CAP_SET_GUEST_DEBUG 3670:Architectures: x86, s390, ppc, arm64 3671:Type: vcpu ioctl 3672:Parameters: struct kvm_guest_debug (in) 3673:Returns: 0 on success; -1 on error 3674 3675:: 3676 3677 struct kvm_guest_debug { 3678 __u32 control; 3679 __u32 pad; 3680 struct kvm_guest_debug_arch arch; 3681 }; 3682 3683Set up the processor specific debug registers and configure vcpu for 3684handling guest debug events. There are two parts to the structure, the 3685first a control bitfield indicates the type of debug events to handle 3686when running. Common control bits are: 3687 3688 - KVM_GUESTDBG_ENABLE: guest debugging is enabled 3689 - KVM_GUESTDBG_SINGLESTEP: the next run should single-step 3690 3691The top 16 bits of the control field are architecture specific control 3692flags which can include the following: 3693 3694 - KVM_GUESTDBG_USE_SW_BP: using software breakpoints [x86, arm64] 3695 - KVM_GUESTDBG_USE_HW_BP: using hardware breakpoints [x86, s390] 3696 - KVM_GUESTDBG_USE_HW: using hardware debug events [arm64] 3697 - KVM_GUESTDBG_INJECT_DB: inject DB type exception [x86] 3698 - KVM_GUESTDBG_INJECT_BP: inject BP type exception [x86] 3699 - KVM_GUESTDBG_EXIT_PENDING: trigger an immediate guest exit [s390] 3700 - KVM_GUESTDBG_BLOCKIRQ: avoid injecting interrupts/NMI/SMI [x86] 3701 3702For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints 3703are enabled in memory so we need to ensure breakpoint exceptions are 3704correctly trapped and the KVM run loop exits at the breakpoint and not 3705running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP 3706we need to ensure the guest vCPUs architecture specific registers are 3707updated to the correct (supplied) values. 3708 3709The second part of the structure is architecture specific and 3710typically contains a set of debug registers. 3711 3712For arm64 the number of debug registers is implementation defined and 3713can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and 3714KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number 3715indicating the number of supported registers. 3716 3717For ppc, the KVM_CAP_PPC_GUEST_DEBUG_SSTEP capability indicates whether 3718the single-step debug event (KVM_GUESTDBG_SINGLESTEP) is supported. 3719 3720Also when supported, KVM_CAP_SET_GUEST_DEBUG2 capability indicates the 3721supported KVM_GUESTDBG_* bits in the control field. 3722 3723When debug events exit the main run loop with the reason 3724KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run 3725structure containing architecture specific debug information. 3726 37274.88 KVM_GET_EMULATED_CPUID 3728--------------------------- 3729 3730:Capability: KVM_CAP_EXT_EMUL_CPUID 3731:Architectures: x86 3732:Type: system ioctl 3733:Parameters: struct kvm_cpuid2 (in/out) 3734:Returns: 0 on success, -1 on error 3735 3736:: 3737 3738 struct kvm_cpuid2 { 3739 __u32 nent; 3740 __u32 flags; 3741 struct kvm_cpuid_entry2 entries[0]; 3742 }; 3743 3744The member 'flags' is used for passing flags from userspace. 3745 3746:: 3747 3748 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0) 3749 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */ 3750 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */ 3751 3752 struct kvm_cpuid_entry2 { 3753 __u32 function; 3754 __u32 index; 3755 __u32 flags; 3756 __u32 eax; 3757 __u32 ebx; 3758 __u32 ecx; 3759 __u32 edx; 3760 __u32 padding[3]; 3761 }; 3762 3763This ioctl returns x86 cpuid features which are emulated by 3764kvm.Userspace can use the information returned by this ioctl to query 3765which features are emulated by kvm instead of being present natively. 3766 3767Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2 3768structure with the 'nent' field indicating the number of entries in 3769the variable-size array 'entries'. If the number of entries is too low 3770to describe the cpu capabilities, an error (E2BIG) is returned. If the 3771number is too high, the 'nent' field is adjusted and an error (ENOMEM) 3772is returned. If the number is just right, the 'nent' field is adjusted 3773to the number of valid entries in the 'entries' array, which is then 3774filled. 3775 3776The entries returned are the set CPUID bits of the respective features 3777which kvm emulates, as returned by the CPUID instruction, with unknown 3778or unsupported feature bits cleared. 3779 3780Features like x2apic, for example, may not be present in the host cpu 3781but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be 3782emulated efficiently and thus not included here. 3783 3784The fields in each entry are defined as follows: 3785 3786 function: 3787 the eax value used to obtain the entry 3788 index: 3789 the ecx value used to obtain the entry (for entries that are 3790 affected by ecx) 3791 flags: 3792 an OR of zero or more of the following: 3793 3794 KVM_CPUID_FLAG_SIGNIFCANT_INDEX: 3795 if the index field is valid 3796 3797 eax, ebx, ecx, edx: 3798 3799 the values returned by the cpuid instruction for 3800 this function/index combination 3801 38024.89 KVM_S390_MEM_OP 3803-------------------- 3804 3805:Capability: KVM_CAP_S390_MEM_OP, KVM_CAP_S390_PROTECTED, KVM_CAP_S390_MEM_OP_EXTENSION 3806:Architectures: s390 3807:Type: vm ioctl, vcpu ioctl 3808:Parameters: struct kvm_s390_mem_op (in) 3809:Returns: = 0 on success, 3810 < 0 on generic error (e.g. -EFAULT or -ENOMEM), 3811 16 bit program exception code if the access causes such an exception 3812 3813Read or write data from/to the VM's memory. 3814The KVM_CAP_S390_MEM_OP_EXTENSION capability specifies what functionality is 3815supported. 3816 3817Parameters are specified via the following structure:: 3818 3819 struct kvm_s390_mem_op { 3820 __u64 gaddr; /* the guest address */ 3821 __u64 flags; /* flags */ 3822 __u32 size; /* amount of bytes */ 3823 __u32 op; /* type of operation */ 3824 __u64 buf; /* buffer in userspace */ 3825 union { 3826 struct { 3827 __u8 ar; /* the access register number */ 3828 __u8 key; /* access key, ignored if flag unset */ 3829 __u8 pad1[6]; /* ignored */ 3830 __u64 old_addr; /* ignored if flag unset */ 3831 }; 3832 __u32 sida_offset; /* offset into the sida */ 3833 __u8 reserved[32]; /* ignored */ 3834 }; 3835 }; 3836 3837The start address of the memory region has to be specified in the "gaddr" 3838field, and the length of the region in the "size" field (which must not 3839be 0). The maximum value for "size" can be obtained by checking the 3840KVM_CAP_S390_MEM_OP capability. "buf" is the buffer supplied by the 3841userspace application where the read data should be written to for 3842a read access, or where the data that should be written is stored for 3843a write access. The "reserved" field is meant for future extensions. 3844Reserved and unused values are ignored. Future extension that add members must 3845introduce new flags. 3846 3847The type of operation is specified in the "op" field. Flags modifying 3848their behavior can be set in the "flags" field. Undefined flag bits must 3849be set to 0. 3850 3851Possible operations are: 3852 * ``KVM_S390_MEMOP_LOGICAL_READ`` 3853 * ``KVM_S390_MEMOP_LOGICAL_WRITE`` 3854 * ``KVM_S390_MEMOP_ABSOLUTE_READ`` 3855 * ``KVM_S390_MEMOP_ABSOLUTE_WRITE`` 3856 * ``KVM_S390_MEMOP_SIDA_READ`` 3857 * ``KVM_S390_MEMOP_SIDA_WRITE`` 3858 * ``KVM_S390_MEMOP_ABSOLUTE_CMPXCHG`` 3859 3860Logical read/write: 3861^^^^^^^^^^^^^^^^^^^ 3862 3863Access logical memory, i.e. translate the given guest address to an absolute 3864address given the state of the VCPU and use the absolute address as target of 3865the access. "ar" designates the access register number to be used; the valid 3866range is 0..15. 3867Logical accesses are permitted for the VCPU ioctl only. 3868Logical accesses are permitted for non-protected guests only. 3869 3870Supported flags: 3871 * ``KVM_S390_MEMOP_F_CHECK_ONLY`` 3872 * ``KVM_S390_MEMOP_F_INJECT_EXCEPTION`` 3873 * ``KVM_S390_MEMOP_F_SKEY_PROTECTION`` 3874 3875The KVM_S390_MEMOP_F_CHECK_ONLY flag can be set to check whether the 3876corresponding memory access would cause an access exception; however, 3877no actual access to the data in memory at the destination is performed. 3878In this case, "buf" is unused and can be NULL. 3879 3880In case an access exception occurred during the access (or would occur 3881in case of KVM_S390_MEMOP_F_CHECK_ONLY), the ioctl returns a positive 3882error number indicating the type of exception. This exception is also 3883raised directly at the corresponding VCPU if the flag 3884KVM_S390_MEMOP_F_INJECT_EXCEPTION is set. 3885On protection exceptions, unless specified otherwise, the injected 3886translation-exception identifier (TEID) indicates suppression. 3887 3888If the KVM_S390_MEMOP_F_SKEY_PROTECTION flag is set, storage key 3889protection is also in effect and may cause exceptions if accesses are 3890prohibited given the access key designated by "key"; the valid range is 0..15. 3891KVM_S390_MEMOP_F_SKEY_PROTECTION is available if KVM_CAP_S390_MEM_OP_EXTENSION 3892is > 0. 3893Since the accessed memory may span multiple pages and those pages might have 3894different storage keys, it is possible that a protection exception occurs 3895after memory has been modified. In this case, if the exception is injected, 3896the TEID does not indicate suppression. 3897 3898Absolute read/write: 3899^^^^^^^^^^^^^^^^^^^^ 3900 3901Access absolute memory. This operation is intended to be used with the 3902KVM_S390_MEMOP_F_SKEY_PROTECTION flag, to allow accessing memory and performing 3903the checks required for storage key protection as one operation (as opposed to 3904user space getting the storage keys, performing the checks, and accessing 3905memory thereafter, which could lead to a delay between check and access). 3906Absolute accesses are permitted for the VM ioctl if KVM_CAP_S390_MEM_OP_EXTENSION 3907has the KVM_S390_MEMOP_EXTENSION_CAP_BASE bit set. 3908Currently absolute accesses are not permitted for VCPU ioctls. 3909Absolute accesses are permitted for non-protected guests only. 3910 3911Supported flags: 3912 * ``KVM_S390_MEMOP_F_CHECK_ONLY`` 3913 * ``KVM_S390_MEMOP_F_SKEY_PROTECTION`` 3914 3915The semantics of the flags common with logical accesses are as for logical 3916accesses. 3917 3918Absolute cmpxchg: 3919^^^^^^^^^^^^^^^^^ 3920 3921Perform cmpxchg on absolute guest memory. Intended for use with the 3922KVM_S390_MEMOP_F_SKEY_PROTECTION flag. 3923Instead of doing an unconditional write, the access occurs only if the target 3924location contains the value pointed to by "old_addr". 3925This is performed as an atomic cmpxchg with the length specified by the "size" 3926parameter. "size" must be a power of two up to and including 16. 3927If the exchange did not take place because the target value doesn't match the 3928old value, the value "old_addr" points to is replaced by the target value. 3929User space can tell if an exchange took place by checking if this replacement 3930occurred. The cmpxchg op is permitted for the VM ioctl if 3931KVM_CAP_S390_MEM_OP_EXTENSION has flag KVM_S390_MEMOP_EXTENSION_CAP_CMPXCHG set. 3932 3933Supported flags: 3934 * ``KVM_S390_MEMOP_F_SKEY_PROTECTION`` 3935 3936SIDA read/write: 3937^^^^^^^^^^^^^^^^ 3938 3939Access the secure instruction data area which contains memory operands necessary 3940for instruction emulation for protected guests. 3941SIDA accesses are available if the KVM_CAP_S390_PROTECTED capability is available. 3942SIDA accesses are permitted for the VCPU ioctl only. 3943SIDA accesses are permitted for protected guests only. 3944 3945No flags are supported. 3946 39474.90 KVM_S390_GET_SKEYS 3948----------------------- 3949 3950:Capability: KVM_CAP_S390_SKEYS 3951:Architectures: s390 3952:Type: vm ioctl 3953:Parameters: struct kvm_s390_skeys 3954:Returns: 0 on success, KVM_S390_GET_SKEYS_NONE if guest is not using storage 3955 keys, negative value on error 3956 3957This ioctl is used to get guest storage key values on the s390 3958architecture. The ioctl takes parameters via the kvm_s390_skeys struct:: 3959 3960 struct kvm_s390_skeys { 3961 __u64 start_gfn; 3962 __u64 count; 3963 __u64 skeydata_addr; 3964 __u32 flags; 3965 __u32 reserved[9]; 3966 }; 3967 3968The start_gfn field is the number of the first guest frame whose storage keys 3969you want to get. 3970 3971The count field is the number of consecutive frames (starting from start_gfn) 3972whose storage keys to get. The count field must be at least 1 and the maximum 3973allowed value is defined as KVM_S390_SKEYS_MAX. Values outside this range 3974will cause the ioctl to return -EINVAL. 3975 3976The skeydata_addr field is the address to a buffer large enough to hold count 3977bytes. This buffer will be filled with storage key data by the ioctl. 3978 39794.91 KVM_S390_SET_SKEYS 3980----------------------- 3981 3982:Capability: KVM_CAP_S390_SKEYS 3983:Architectures: s390 3984:Type: vm ioctl 3985:Parameters: struct kvm_s390_skeys 3986:Returns: 0 on success, negative value on error 3987 3988This ioctl is used to set guest storage key values on the s390 3989architecture. The ioctl takes parameters via the kvm_s390_skeys struct. 3990See section on KVM_S390_GET_SKEYS for struct definition. 3991 3992The start_gfn field is the number of the first guest frame whose storage keys 3993you want to set. 3994 3995The count field is the number of consecutive frames (starting from start_gfn) 3996whose storage keys to get. The count field must be at least 1 and the maximum 3997allowed value is defined as KVM_S390_SKEYS_MAX. Values outside this range 3998will cause the ioctl to return -EINVAL. 3999 4000The skeydata_addr field is the address to a buffer containing count bytes of 4001storage keys. Each byte in the buffer will be set as the storage key for a 4002single frame starting at start_gfn for count frames. 4003 4004Note: If any architecturally invalid key value is found in the given data then 4005the ioctl will return -EINVAL. 4006 40074.92 KVM_S390_IRQ 4008----------------- 4009 4010:Capability: KVM_CAP_S390_INJECT_IRQ 4011:Architectures: s390 4012:Type: vcpu ioctl 4013:Parameters: struct kvm_s390_irq (in) 4014:Returns: 0 on success, -1 on error 4015 4016Errors: 4017 4018 4019 ====== ================================================================= 4020 EINVAL interrupt type is invalid 4021 type is KVM_S390_SIGP_STOP and flag parameter is invalid value, 4022 type is KVM_S390_INT_EXTERNAL_CALL and code is bigger 4023 than the maximum of VCPUs 4024 EBUSY type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped, 4025 type is KVM_S390_SIGP_STOP and a stop irq is already pending, 4026 type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt 4027 is already pending 4028 ====== ================================================================= 4029 4030Allows to inject an interrupt to the guest. 4031 4032Using struct kvm_s390_irq as a parameter allows 4033to inject additional payload which is not 4034possible via KVM_S390_INTERRUPT. 4035 4036Interrupt parameters are passed via kvm_s390_irq:: 4037 4038 struct kvm_s390_irq { 4039 __u64 type; 4040 union { 4041 struct kvm_s390_io_info io; 4042 struct kvm_s390_ext_info ext; 4043 struct kvm_s390_pgm_info pgm; 4044 struct kvm_s390_emerg_info emerg; 4045 struct kvm_s390_extcall_info extcall; 4046 struct kvm_s390_prefix_info prefix; 4047 struct kvm_s390_stop_info stop; 4048 struct kvm_s390_mchk_info mchk; 4049 char reserved[64]; 4050 } u; 4051 }; 4052 4053type can be one of the following: 4054 4055- KVM_S390_SIGP_STOP - sigp stop; parameter in .stop 4056- KVM_S390_PROGRAM_INT - program check; parameters in .pgm 4057- KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix 4058- KVM_S390_RESTART - restart; no parameters 4059- KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters 4060- KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters 4061- KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg 4062- KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall 4063- KVM_S390_MCHK - machine check interrupt; parameters in .mchk 4064 4065This is an asynchronous vcpu ioctl and can be invoked from any thread. 4066 40674.94 KVM_S390_GET_IRQ_STATE 4068--------------------------- 4069 4070:Capability: KVM_CAP_S390_IRQ_STATE 4071:Architectures: s390 4072:Type: vcpu ioctl 4073:Parameters: struct kvm_s390_irq_state (out) 4074:Returns: >= number of bytes copied into buffer, 4075 -EINVAL if buffer size is 0, 4076 -ENOBUFS if buffer size is too small to fit all pending interrupts, 4077 -EFAULT if the buffer address was invalid 4078 4079This ioctl allows userspace to retrieve the complete state of all currently 4080pending interrupts in a single buffer. Use cases include migration 4081and introspection. The parameter structure contains the address of a 4082userspace buffer and its length:: 4083 4084 struct kvm_s390_irq_state { 4085 __u64 buf; 4086 __u32 flags; /* will stay unused for compatibility reasons */ 4087 __u32 len; 4088 __u32 reserved[4]; /* will stay unused for compatibility reasons */ 4089 }; 4090 4091Userspace passes in the above struct and for each pending interrupt a 4092struct kvm_s390_irq is copied to the provided buffer. 4093 4094The structure contains a flags and a reserved field for future extensions. As 4095the kernel never checked for flags == 0 and QEMU never pre-zeroed flags and 4096reserved, these fields can not be used in the future without breaking 4097compatibility. 4098 4099If -ENOBUFS is returned the buffer provided was too small and userspace 4100may retry with a bigger buffer. 4101 41024.95 KVM_S390_SET_IRQ_STATE 4103--------------------------- 4104 4105:Capability: KVM_CAP_S390_IRQ_STATE 4106:Architectures: s390 4107:Type: vcpu ioctl 4108:Parameters: struct kvm_s390_irq_state (in) 4109:Returns: 0 on success, 4110 -EFAULT if the buffer address was invalid, 4111 -EINVAL for an invalid buffer length (see below), 4112 -EBUSY if there were already interrupts pending, 4113 errors occurring when actually injecting the 4114 interrupt. See KVM_S390_IRQ. 4115 4116This ioctl allows userspace to set the complete state of all cpu-local 4117interrupts currently pending for the vcpu. It is intended for restoring 4118interrupt state after a migration. The input parameter is a userspace buffer 4119containing a struct kvm_s390_irq_state:: 4120 4121 struct kvm_s390_irq_state { 4122 __u64 buf; 4123 __u32 flags; /* will stay unused for compatibility reasons */ 4124 __u32 len; 4125 __u32 reserved[4]; /* will stay unused for compatibility reasons */ 4126 }; 4127 4128The restrictions for flags and reserved apply as well. 4129(see KVM_S390_GET_IRQ_STATE) 4130 4131The userspace memory referenced by buf contains a struct kvm_s390_irq 4132for each interrupt to be injected into the guest. 4133If one of the interrupts could not be injected for some reason the 4134ioctl aborts. 4135 4136len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0 4137and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq), 4138which is the maximum number of possibly pending cpu-local interrupts. 4139 41404.96 KVM_SMI 4141------------ 4142 4143:Capability: KVM_CAP_X86_SMM 4144:Architectures: x86 4145:Type: vcpu ioctl 4146:Parameters: none 4147:Returns: 0 on success, -1 on error 4148 4149Queues an SMI on the thread's vcpu. 4150 41514.97 KVM_X86_SET_MSR_FILTER 4152---------------------------- 4153 4154:Capability: KVM_CAP_X86_MSR_FILTER 4155:Architectures: x86 4156:Type: vm ioctl 4157:Parameters: struct kvm_msr_filter 4158:Returns: 0 on success, < 0 on error 4159 4160:: 4161 4162 struct kvm_msr_filter_range { 4163 #define KVM_MSR_FILTER_READ (1 << 0) 4164 #define KVM_MSR_FILTER_WRITE (1 << 1) 4165 __u32 flags; 4166 __u32 nmsrs; /* number of msrs in bitmap */ 4167 __u32 base; /* MSR index the bitmap starts at */ 4168 __u8 *bitmap; /* a 1 bit allows the operations in flags, 0 denies */ 4169 }; 4170 4171 #define KVM_MSR_FILTER_MAX_RANGES 16 4172 struct kvm_msr_filter { 4173 #define KVM_MSR_FILTER_DEFAULT_ALLOW (0 << 0) 4174 #define KVM_MSR_FILTER_DEFAULT_DENY (1 << 0) 4175 __u32 flags; 4176 struct kvm_msr_filter_range ranges[KVM_MSR_FILTER_MAX_RANGES]; 4177 }; 4178 4179flags values for ``struct kvm_msr_filter_range``: 4180 4181``KVM_MSR_FILTER_READ`` 4182 4183 Filter read accesses to MSRs using the given bitmap. A 0 in the bitmap 4184 indicates that read accesses should be denied, while a 1 indicates that 4185 a read for a particular MSR should be allowed regardless of the default 4186 filter action. 4187 4188``KVM_MSR_FILTER_WRITE`` 4189 4190 Filter write accesses to MSRs using the given bitmap. A 0 in the bitmap 4191 indicates that write accesses should be denied, while a 1 indicates that 4192 a write for a particular MSR should be allowed regardless of the default 4193 filter action. 4194 4195flags values for ``struct kvm_msr_filter``: 4196 4197``KVM_MSR_FILTER_DEFAULT_ALLOW`` 4198 4199 If no filter range matches an MSR index that is getting accessed, KVM will 4200 allow accesses to all MSRs by default. 4201 4202``KVM_MSR_FILTER_DEFAULT_DENY`` 4203 4204 If no filter range matches an MSR index that is getting accessed, KVM will 4205 deny accesses to all MSRs by default. 4206 4207This ioctl allows userspace to define up to 16 bitmaps of MSR ranges to deny 4208guest MSR accesses that would normally be allowed by KVM. If an MSR is not 4209covered by a specific range, the "default" filtering behavior applies. Each 4210bitmap range covers MSRs from [base .. base+nmsrs). 4211 4212If an MSR access is denied by userspace, the resulting KVM behavior depends on 4213whether or not KVM_CAP_X86_USER_SPACE_MSR's KVM_MSR_EXIT_REASON_FILTER is 4214enabled. If KVM_MSR_EXIT_REASON_FILTER is enabled, KVM will exit to userspace 4215on denied accesses, i.e. userspace effectively intercepts the MSR access. If 4216KVM_MSR_EXIT_REASON_FILTER is not enabled, KVM will inject a #GP into the guest 4217on denied accesses. Note, if an MSR access is denied during emulation of MSR 4218load/stores during VMX transitions, KVM ignores KVM_MSR_EXIT_REASON_FILTER. 4219See the below warning for full details. 4220 4221If an MSR access is allowed by userspace, KVM will emulate and/or virtualize 4222the access in accordance with the vCPU model. Note, KVM may still ultimately 4223inject a #GP if an access is allowed by userspace, e.g. if KVM doesn't support 4224the MSR, or to follow architectural behavior for the MSR. 4225 4226By default, KVM operates in KVM_MSR_FILTER_DEFAULT_ALLOW mode with no MSR range 4227filters. 4228 4229Calling this ioctl with an empty set of ranges (all nmsrs == 0) disables MSR 4230filtering. In that mode, ``KVM_MSR_FILTER_DEFAULT_DENY`` is invalid and causes 4231an error. 4232 4233.. warning:: 4234 MSR accesses that are side effects of instruction execution (emulated or 4235 native) are not filtered as hardware does not honor MSR bitmaps outside of 4236 RDMSR and WRMSR, and KVM mimics that behavior when emulating instructions 4237 to avoid pointless divergence from hardware. E.g. RDPID reads MSR_TSC_AUX, 4238 SYSENTER reads the SYSENTER MSRs, etc. 4239 4240 MSRs that are loaded/stored via dedicated VMCS fields are not filtered as 4241 part of VM-Enter/VM-Exit emulation. 4242 4243 MSRs that are loaded/store via VMX's load/store lists _are_ filtered as part 4244 of VM-Enter/VM-Exit emulation. If an MSR access is denied on VM-Enter, KVM 4245 synthesizes a consistency check VM-Exit(EXIT_REASON_MSR_LOAD_FAIL). If an 4246 MSR access is denied on VM-Exit, KVM synthesizes a VM-Abort. In short, KVM 4247 extends Intel's architectural list of MSRs that cannot be loaded/saved via 4248 the VM-Enter/VM-Exit MSR list. It is platform owner's responsibility to 4249 to communicate any such restrictions to their end users. 4250 4251 x2APIC MSR accesses cannot be filtered (KVM silently ignores filters that 4252 cover any x2APIC MSRs). 4253 4254Note, invoking this ioctl while a vCPU is running is inherently racy. However, 4255KVM does guarantee that vCPUs will see either the previous filter or the new 4256filter, e.g. MSRs with identical settings in both the old and new filter will 4257have deterministic behavior. 4258 4259Similarly, if userspace wishes to intercept on denied accesses, 4260KVM_MSR_EXIT_REASON_FILTER must be enabled before activating any filters, and 4261left enabled until after all filters are deactivated. Failure to do so may 4262result in KVM injecting a #GP instead of exiting to userspace. 4263 42644.98 KVM_CREATE_SPAPR_TCE_64 4265---------------------------- 4266 4267:Capability: KVM_CAP_SPAPR_TCE_64 4268:Architectures: powerpc 4269:Type: vm ioctl 4270:Parameters: struct kvm_create_spapr_tce_64 (in) 4271:Returns: file descriptor for manipulating the created TCE table 4272 4273This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit 4274windows, described in 4.62 KVM_CREATE_SPAPR_TCE 4275 4276This capability uses extended struct in ioctl interface:: 4277 4278 /* for KVM_CAP_SPAPR_TCE_64 */ 4279 struct kvm_create_spapr_tce_64 { 4280 __u64 liobn; 4281 __u32 page_shift; 4282 __u32 flags; 4283 __u64 offset; /* in pages */ 4284 __u64 size; /* in pages */ 4285 }; 4286 4287The aim of extension is to support an additional bigger DMA window with 4288a variable page size. 4289KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and 4290a bus offset of the corresponding DMA window, @size and @offset are numbers 4291of IOMMU pages. 4292 4293@flags are not used at the moment. 4294 4295The rest of functionality is identical to KVM_CREATE_SPAPR_TCE. 4296 42974.99 KVM_REINJECT_CONTROL 4298------------------------- 4299 4300:Capability: KVM_CAP_REINJECT_CONTROL 4301:Architectures: x86 4302:Type: vm ioctl 4303:Parameters: struct kvm_reinject_control (in) 4304:Returns: 0 on success, 4305 -EFAULT if struct kvm_reinject_control cannot be read, 4306 -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier. 4307 4308i8254 (PIT) has two modes, reinject and !reinject. The default is reinject, 4309where KVM queues elapsed i8254 ticks and monitors completion of interrupt from 4310vector(s) that i8254 injects. Reinject mode dequeues a tick and injects its 4311interrupt whenever there isn't a pending interrupt from i8254. 4312!reinject mode injects an interrupt as soon as a tick arrives. 4313 4314:: 4315 4316 struct kvm_reinject_control { 4317 __u8 pit_reinject; 4318 __u8 reserved[31]; 4319 }; 4320 4321pit_reinject = 0 (!reinject mode) is recommended, unless running an old 4322operating system that uses the PIT for timing (e.g. Linux 2.4.x). 4323 43244.100 KVM_PPC_CONFIGURE_V3_MMU 4325------------------------------ 4326 4327:Capability: KVM_CAP_PPC_MMU_RADIX or KVM_CAP_PPC_MMU_HASH_V3 4328:Architectures: ppc 4329:Type: vm ioctl 4330:Parameters: struct kvm_ppc_mmuv3_cfg (in) 4331:Returns: 0 on success, 4332 -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read, 4333 -EINVAL if the configuration is invalid 4334 4335This ioctl controls whether the guest will use radix or HPT (hashed 4336page table) translation, and sets the pointer to the process table for 4337the guest. 4338 4339:: 4340 4341 struct kvm_ppc_mmuv3_cfg { 4342 __u64 flags; 4343 __u64 process_table; 4344 }; 4345 4346There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and 4347KVM_PPC_MMUV3_GTSE. KVM_PPC_MMUV3_RADIX, if set, configures the guest 4348to use radix tree translation, and if clear, to use HPT translation. 4349KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest 4350to be able to use the global TLB and SLB invalidation instructions; 4351if clear, the guest may not use these instructions. 4352 4353The process_table field specifies the address and size of the guest 4354process table, which is in the guest's space. This field is formatted 4355as the second doubleword of the partition table entry, as defined in 4356the Power ISA V3.00, Book III section 5.7.6.1. 4357 43584.101 KVM_PPC_GET_RMMU_INFO 4359--------------------------- 4360 4361:Capability: KVM_CAP_PPC_MMU_RADIX 4362:Architectures: ppc 4363:Type: vm ioctl 4364:Parameters: struct kvm_ppc_rmmu_info (out) 4365:Returns: 0 on success, 4366 -EFAULT if struct kvm_ppc_rmmu_info cannot be written, 4367 -EINVAL if no useful information can be returned 4368 4369This ioctl returns a structure containing two things: (a) a list 4370containing supported radix tree geometries, and (b) a list that maps 4371page sizes to put in the "AP" (actual page size) field for the tlbie 4372(TLB invalidate entry) instruction. 4373 4374:: 4375 4376 struct kvm_ppc_rmmu_info { 4377 struct kvm_ppc_radix_geom { 4378 __u8 page_shift; 4379 __u8 level_bits[4]; 4380 __u8 pad[3]; 4381 } geometries[8]; 4382 __u32 ap_encodings[8]; 4383 }; 4384 4385The geometries[] field gives up to 8 supported geometries for the 4386radix page table, in terms of the log base 2 of the smallest page 4387size, and the number of bits indexed at each level of the tree, from 4388the PTE level up to the PGD level in that order. Any unused entries 4389will have 0 in the page_shift field. 4390 4391The ap_encodings gives the supported page sizes and their AP field 4392encodings, encoded with the AP value in the top 3 bits and the log 4393base 2 of the page size in the bottom 6 bits. 4394 43954.102 KVM_PPC_RESIZE_HPT_PREPARE 4396-------------------------------- 4397 4398:Capability: KVM_CAP_SPAPR_RESIZE_HPT 4399:Architectures: powerpc 4400:Type: vm ioctl 4401:Parameters: struct kvm_ppc_resize_hpt (in) 4402:Returns: 0 on successful completion, 4403 >0 if a new HPT is being prepared, the value is an estimated 4404 number of milliseconds until preparation is complete, 4405 -EFAULT if struct kvm_reinject_control cannot be read, 4406 -EINVAL if the supplied shift or flags are invalid, 4407 -ENOMEM if unable to allocate the new HPT, 4408 4409Used to implement the PAPR extension for runtime resizing of a guest's 4410Hashed Page Table (HPT). Specifically this starts, stops or monitors 4411the preparation of a new potential HPT for the guest, essentially 4412implementing the H_RESIZE_HPT_PREPARE hypercall. 4413 4414:: 4415 4416 struct kvm_ppc_resize_hpt { 4417 __u64 flags; 4418 __u32 shift; 4419 __u32 pad; 4420 }; 4421 4422If called with shift > 0 when there is no pending HPT for the guest, 4423this begins preparation of a new pending HPT of size 2^(shift) bytes. 4424It then returns a positive integer with the estimated number of 4425milliseconds until preparation is complete. 4426 4427If called when there is a pending HPT whose size does not match that 4428requested in the parameters, discards the existing pending HPT and 4429creates a new one as above. 4430 4431If called when there is a pending HPT of the size requested, will: 4432 4433 * If preparation of the pending HPT is already complete, return 0 4434 * If preparation of the pending HPT has failed, return an error 4435 code, then discard the pending HPT. 4436 * If preparation of the pending HPT is still in progress, return an 4437 estimated number of milliseconds until preparation is complete. 4438 4439If called with shift == 0, discards any currently pending HPT and 4440returns 0 (i.e. cancels any in-progress preparation). 4441 4442flags is reserved for future expansion, currently setting any bits in 4443flags will result in an -EINVAL. 4444 4445Normally this will be called repeatedly with the same parameters until 4446it returns <= 0. The first call will initiate preparation, subsequent 4447ones will monitor preparation until it completes or fails. 4448 44494.103 KVM_PPC_RESIZE_HPT_COMMIT 4450------------------------------- 4451 4452:Capability: KVM_CAP_SPAPR_RESIZE_HPT 4453:Architectures: powerpc 4454:Type: vm ioctl 4455:Parameters: struct kvm_ppc_resize_hpt (in) 4456:Returns: 0 on successful completion, 4457 -EFAULT if struct kvm_reinject_control cannot be read, 4458 -EINVAL if the supplied shift or flags are invalid, 4459 -ENXIO is there is no pending HPT, or the pending HPT doesn't 4460 have the requested size, 4461 -EBUSY if the pending HPT is not fully prepared, 4462 -ENOSPC if there was a hash collision when moving existing 4463 HPT entries to the new HPT, 4464 -EIO on other error conditions 4465 4466Used to implement the PAPR extension for runtime resizing of a guest's 4467Hashed Page Table (HPT). Specifically this requests that the guest be 4468transferred to working with the new HPT, essentially implementing the 4469H_RESIZE_HPT_COMMIT hypercall. 4470 4471:: 4472 4473 struct kvm_ppc_resize_hpt { 4474 __u64 flags; 4475 __u32 shift; 4476 __u32 pad; 4477 }; 4478 4479This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has 4480returned 0 with the same parameters. In other cases 4481KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or 4482-EBUSY, though others may be possible if the preparation was started, 4483but failed). 4484 4485This will have undefined effects on the guest if it has not already 4486placed itself in a quiescent state where no vcpu will make MMU enabled 4487memory accesses. 4488 4489On successful completion, the pending HPT will become the guest's active 4490HPT and the previous HPT will be discarded. 4491 4492On failure, the guest will still be operating on its previous HPT. 4493 44944.104 KVM_X86_GET_MCE_CAP_SUPPORTED 4495----------------------------------- 4496 4497:Capability: KVM_CAP_MCE 4498:Architectures: x86 4499:Type: system ioctl 4500:Parameters: u64 mce_cap (out) 4501:Returns: 0 on success, -1 on error 4502 4503Returns supported MCE capabilities. The u64 mce_cap parameter 4504has the same format as the MSR_IA32_MCG_CAP register. Supported 4505capabilities will have the corresponding bits set. 4506 45074.105 KVM_X86_SETUP_MCE 4508----------------------- 4509 4510:Capability: KVM_CAP_MCE 4511:Architectures: x86 4512:Type: vcpu ioctl 4513:Parameters: u64 mcg_cap (in) 4514:Returns: 0 on success, 4515 -EFAULT if u64 mcg_cap cannot be read, 4516 -EINVAL if the requested number of banks is invalid, 4517 -EINVAL if requested MCE capability is not supported. 4518 4519Initializes MCE support for use. The u64 mcg_cap parameter 4520has the same format as the MSR_IA32_MCG_CAP register and 4521specifies which capabilities should be enabled. The maximum 4522supported number of error-reporting banks can be retrieved when 4523checking for KVM_CAP_MCE. The supported capabilities can be 4524retrieved with KVM_X86_GET_MCE_CAP_SUPPORTED. 4525 45264.106 KVM_X86_SET_MCE 4527--------------------- 4528 4529:Capability: KVM_CAP_MCE 4530:Architectures: x86 4531:Type: vcpu ioctl 4532:Parameters: struct kvm_x86_mce (in) 4533:Returns: 0 on success, 4534 -EFAULT if struct kvm_x86_mce cannot be read, 4535 -EINVAL if the bank number is invalid, 4536 -EINVAL if VAL bit is not set in status field. 4537 4538Inject a machine check error (MCE) into the guest. The input 4539parameter is:: 4540 4541 struct kvm_x86_mce { 4542 __u64 status; 4543 __u64 addr; 4544 __u64 misc; 4545 __u64 mcg_status; 4546 __u8 bank; 4547 __u8 pad1[7]; 4548 __u64 pad2[3]; 4549 }; 4550 4551If the MCE being reported is an uncorrected error, KVM will 4552inject it as an MCE exception into the guest. If the guest 4553MCG_STATUS register reports that an MCE is in progress, KVM 4554causes an KVM_EXIT_SHUTDOWN vmexit. 4555 4556Otherwise, if the MCE is a corrected error, KVM will just 4557store it in the corresponding bank (provided this bank is 4558not holding a previously reported uncorrected error). 4559 45604.107 KVM_S390_GET_CMMA_BITS 4561---------------------------- 4562 4563:Capability: KVM_CAP_S390_CMMA_MIGRATION 4564:Architectures: s390 4565:Type: vm ioctl 4566:Parameters: struct kvm_s390_cmma_log (in, out) 4567:Returns: 0 on success, a negative value on error 4568 4569Errors: 4570 4571 ====== ============================================================= 4572 ENOMEM not enough memory can be allocated to complete the task 4573 ENXIO if CMMA is not enabled 4574 EINVAL if KVM_S390_CMMA_PEEK is not set but migration mode was not enabled 4575 EINVAL if KVM_S390_CMMA_PEEK is not set but dirty tracking has been 4576 disabled (and thus migration mode was automatically disabled) 4577 EFAULT if the userspace address is invalid or if no page table is 4578 present for the addresses (e.g. when using hugepages). 4579 ====== ============================================================= 4580 4581This ioctl is used to get the values of the CMMA bits on the s390 4582architecture. It is meant to be used in two scenarios: 4583 4584- During live migration to save the CMMA values. Live migration needs 4585 to be enabled via the KVM_REQ_START_MIGRATION VM property. 4586- To non-destructively peek at the CMMA values, with the flag 4587 KVM_S390_CMMA_PEEK set. 4588 4589The ioctl takes parameters via the kvm_s390_cmma_log struct. The desired 4590values are written to a buffer whose location is indicated via the "values" 4591member in the kvm_s390_cmma_log struct. The values in the input struct are 4592also updated as needed. 4593 4594Each CMMA value takes up one byte. 4595 4596:: 4597 4598 struct kvm_s390_cmma_log { 4599 __u64 start_gfn; 4600 __u32 count; 4601 __u32 flags; 4602 union { 4603 __u64 remaining; 4604 __u64 mask; 4605 }; 4606 __u64 values; 4607 }; 4608 4609start_gfn is the number of the first guest frame whose CMMA values are 4610to be retrieved, 4611 4612count is the length of the buffer in bytes, 4613 4614values points to the buffer where the result will be written to. 4615 4616If count is greater than KVM_S390_SKEYS_MAX, then it is considered to be 4617KVM_S390_SKEYS_MAX. KVM_S390_SKEYS_MAX is re-used for consistency with 4618other ioctls. 4619 4620The result is written in the buffer pointed to by the field values, and 4621the values of the input parameter are updated as follows. 4622 4623Depending on the flags, different actions are performed. The only 4624supported flag so far is KVM_S390_CMMA_PEEK. 4625 4626The default behaviour if KVM_S390_CMMA_PEEK is not set is: 4627start_gfn will indicate the first page frame whose CMMA bits were dirty. 4628It is not necessarily the same as the one passed as input, as clean pages 4629are skipped. 4630 4631count will indicate the number of bytes actually written in the buffer. 4632It can (and very often will) be smaller than the input value, since the 4633buffer is only filled until 16 bytes of clean values are found (which 4634are then not copied in the buffer). Since a CMMA migration block needs 4635the base address and the length, for a total of 16 bytes, we will send 4636back some clean data if there is some dirty data afterwards, as long as 4637the size of the clean data does not exceed the size of the header. This 4638allows to minimize the amount of data to be saved or transferred over 4639the network at the expense of more roundtrips to userspace. The next 4640invocation of the ioctl will skip over all the clean values, saving 4641potentially more than just the 16 bytes we found. 4642 4643If KVM_S390_CMMA_PEEK is set: 4644the existing storage attributes are read even when not in migration 4645mode, and no other action is performed; 4646 4647the output start_gfn will be equal to the input start_gfn, 4648 4649the output count will be equal to the input count, except if the end of 4650memory has been reached. 4651 4652In both cases: 4653the field "remaining" will indicate the total number of dirty CMMA values 4654still remaining, or 0 if KVM_S390_CMMA_PEEK is set and migration mode is 4655not enabled. 4656 4657mask is unused. 4658 4659values points to the userspace buffer where the result will be stored. 4660 46614.108 KVM_S390_SET_CMMA_BITS 4662---------------------------- 4663 4664:Capability: KVM_CAP_S390_CMMA_MIGRATION 4665:Architectures: s390 4666:Type: vm ioctl 4667:Parameters: struct kvm_s390_cmma_log (in) 4668:Returns: 0 on success, a negative value on error 4669 4670This ioctl is used to set the values of the CMMA bits on the s390 4671architecture. It is meant to be used during live migration to restore 4672the CMMA values, but there are no restrictions on its use. 4673The ioctl takes parameters via the kvm_s390_cmma_values struct. 4674Each CMMA value takes up one byte. 4675 4676:: 4677 4678 struct kvm_s390_cmma_log { 4679 __u64 start_gfn; 4680 __u32 count; 4681 __u32 flags; 4682 union { 4683 __u64 remaining; 4684 __u64 mask; 4685 }; 4686 __u64 values; 4687 }; 4688 4689start_gfn indicates the starting guest frame number, 4690 4691count indicates how many values are to be considered in the buffer, 4692 4693flags is not used and must be 0. 4694 4695mask indicates which PGSTE bits are to be considered. 4696 4697remaining is not used. 4698 4699values points to the buffer in userspace where to store the values. 4700 4701This ioctl can fail with -ENOMEM if not enough memory can be allocated to 4702complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if 4703the count field is too large (e.g. more than KVM_S390_CMMA_SIZE_MAX) or 4704if the flags field was not 0, with -EFAULT if the userspace address is 4705invalid, if invalid pages are written to (e.g. after the end of memory) 4706or if no page table is present for the addresses (e.g. when using 4707hugepages). 4708 47094.109 KVM_PPC_GET_CPU_CHAR 4710-------------------------- 4711 4712:Capability: KVM_CAP_PPC_GET_CPU_CHAR 4713:Architectures: powerpc 4714:Type: vm ioctl 4715:Parameters: struct kvm_ppc_cpu_char (out) 4716:Returns: 0 on successful completion, 4717 -EFAULT if struct kvm_ppc_cpu_char cannot be written 4718 4719This ioctl gives userspace information about certain characteristics 4720of the CPU relating to speculative execution of instructions and 4721possible information leakage resulting from speculative execution (see 4722CVE-2017-5715, CVE-2017-5753 and CVE-2017-5754). The information is 4723returned in struct kvm_ppc_cpu_char, which looks like this:: 4724 4725 struct kvm_ppc_cpu_char { 4726 __u64 character; /* characteristics of the CPU */ 4727 __u64 behaviour; /* recommended software behaviour */ 4728 __u64 character_mask; /* valid bits in character */ 4729 __u64 behaviour_mask; /* valid bits in behaviour */ 4730 }; 4731 4732For extensibility, the character_mask and behaviour_mask fields 4733indicate which bits of character and behaviour have been filled in by 4734the kernel. If the set of defined bits is extended in future then 4735userspace will be able to tell whether it is running on a kernel that 4736knows about the new bits. 4737 4738The character field describes attributes of the CPU which can help 4739with preventing inadvertent information disclosure - specifically, 4740whether there is an instruction to flash-invalidate the L1 data cache 4741(ori 30,30,0 or mtspr SPRN_TRIG2,rN), whether the L1 data cache is set 4742to a mode where entries can only be used by the thread that created 4743them, whether the bcctr[l] instruction prevents speculation, and 4744whether a speculation barrier instruction (ori 31,31,0) is provided. 4745 4746The behaviour field describes actions that software should take to 4747prevent inadvertent information disclosure, and thus describes which 4748vulnerabilities the hardware is subject to; specifically whether the 4749L1 data cache should be flushed when returning to user mode from the 4750kernel, and whether a speculation barrier should be placed between an 4751array bounds check and the array access. 4752 4753These fields use the same bit definitions as the new 4754H_GET_CPU_CHARACTERISTICS hypercall. 4755 47564.110 KVM_MEMORY_ENCRYPT_OP 4757--------------------------- 4758 4759:Capability: basic 4760:Architectures: x86 4761:Type: vm 4762:Parameters: an opaque platform specific structure (in/out) 4763:Returns: 0 on success; -1 on error 4764 4765If the platform supports creating encrypted VMs then this ioctl can be used 4766for issuing platform-specific memory encryption commands to manage those 4767encrypted VMs. 4768 4769Currently, this ioctl is used for issuing Secure Encrypted Virtualization 4770(SEV) commands on AMD Processors. The SEV commands are defined in 4771Documentation/virt/kvm/x86/amd-memory-encryption.rst. 4772 47734.111 KVM_MEMORY_ENCRYPT_REG_REGION 4774----------------------------------- 4775 4776:Capability: basic 4777:Architectures: x86 4778:Type: system 4779:Parameters: struct kvm_enc_region (in) 4780:Returns: 0 on success; -1 on error 4781 4782This ioctl can be used to register a guest memory region which may 4783contain encrypted data (e.g. guest RAM, SMRAM etc). 4784 4785It is used in the SEV-enabled guest. When encryption is enabled, a guest 4786memory region may contain encrypted data. The SEV memory encryption 4787engine uses a tweak such that two identical plaintext pages, each at 4788different locations will have differing ciphertexts. So swapping or 4789moving ciphertext of those pages will not result in plaintext being 4790swapped. So relocating (or migrating) physical backing pages for the SEV 4791guest will require some additional steps. 4792 4793Note: The current SEV key management spec does not provide commands to 4794swap or migrate (move) ciphertext pages. Hence, for now we pin the guest 4795memory region registered with the ioctl. 4796 47974.112 KVM_MEMORY_ENCRYPT_UNREG_REGION 4798------------------------------------- 4799 4800:Capability: basic 4801:Architectures: x86 4802:Type: system 4803:Parameters: struct kvm_enc_region (in) 4804:Returns: 0 on success; -1 on error 4805 4806This ioctl can be used to unregister the guest memory region registered 4807with KVM_MEMORY_ENCRYPT_REG_REGION ioctl above. 4808 48094.113 KVM_HYPERV_EVENTFD 4810------------------------ 4811 4812:Capability: KVM_CAP_HYPERV_EVENTFD 4813:Architectures: x86 4814:Type: vm ioctl 4815:Parameters: struct kvm_hyperv_eventfd (in) 4816 4817This ioctl (un)registers an eventfd to receive notifications from the guest on 4818the specified Hyper-V connection id through the SIGNAL_EVENT hypercall, without 4819causing a user exit. SIGNAL_EVENT hypercall with non-zero event flag number 4820(bits 24-31) still triggers a KVM_EXIT_HYPERV_HCALL user exit. 4821 4822:: 4823 4824 struct kvm_hyperv_eventfd { 4825 __u32 conn_id; 4826 __s32 fd; 4827 __u32 flags; 4828 __u32 padding[3]; 4829 }; 4830 4831The conn_id field should fit within 24 bits:: 4832 4833 #define KVM_HYPERV_CONN_ID_MASK 0x00ffffff 4834 4835The acceptable values for the flags field are:: 4836 4837 #define KVM_HYPERV_EVENTFD_DEASSIGN (1 << 0) 4838 4839:Returns: 0 on success, 4840 -EINVAL if conn_id or flags is outside the allowed range, 4841 -ENOENT on deassign if the conn_id isn't registered, 4842 -EEXIST on assign if the conn_id is already registered 4843 48444.114 KVM_GET_NESTED_STATE 4845-------------------------- 4846 4847:Capability: KVM_CAP_NESTED_STATE 4848:Architectures: x86 4849:Type: vcpu ioctl 4850:Parameters: struct kvm_nested_state (in/out) 4851:Returns: 0 on success, -1 on error 4852 4853Errors: 4854 4855 ===== ============================================================= 4856 E2BIG the total state size exceeds the value of 'size' specified by 4857 the user; the size required will be written into size. 4858 ===== ============================================================= 4859 4860:: 4861 4862 struct kvm_nested_state { 4863 __u16 flags; 4864 __u16 format; 4865 __u32 size; 4866 4867 union { 4868 struct kvm_vmx_nested_state_hdr vmx; 4869 struct kvm_svm_nested_state_hdr svm; 4870 4871 /* Pad the header to 128 bytes. */ 4872 __u8 pad[120]; 4873 } hdr; 4874 4875 union { 4876 struct kvm_vmx_nested_state_data vmx[0]; 4877 struct kvm_svm_nested_state_data svm[0]; 4878 } data; 4879 }; 4880 4881 #define KVM_STATE_NESTED_GUEST_MODE 0x00000001 4882 #define KVM_STATE_NESTED_RUN_PENDING 0x00000002 4883 #define KVM_STATE_NESTED_EVMCS 0x00000004 4884 4885 #define KVM_STATE_NESTED_FORMAT_VMX 0 4886 #define KVM_STATE_NESTED_FORMAT_SVM 1 4887 4888 #define KVM_STATE_NESTED_VMX_VMCS_SIZE 0x1000 4889 4890 #define KVM_STATE_NESTED_VMX_SMM_GUEST_MODE 0x00000001 4891 #define KVM_STATE_NESTED_VMX_SMM_VMXON 0x00000002 4892 4893 #define KVM_STATE_VMX_PREEMPTION_TIMER_DEADLINE 0x00000001 4894 4895 struct kvm_vmx_nested_state_hdr { 4896 __u64 vmxon_pa; 4897 __u64 vmcs12_pa; 4898 4899 struct { 4900 __u16 flags; 4901 } smm; 4902 4903 __u32 flags; 4904 __u64 preemption_timer_deadline; 4905 }; 4906 4907 struct kvm_vmx_nested_state_data { 4908 __u8 vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE]; 4909 __u8 shadow_vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE]; 4910 }; 4911 4912This ioctl copies the vcpu's nested virtualization state from the kernel to 4913userspace. 4914 4915The maximum size of the state can be retrieved by passing KVM_CAP_NESTED_STATE 4916to the KVM_CHECK_EXTENSION ioctl(). 4917 49184.115 KVM_SET_NESTED_STATE 4919-------------------------- 4920 4921:Capability: KVM_CAP_NESTED_STATE 4922:Architectures: x86 4923:Type: vcpu ioctl 4924:Parameters: struct kvm_nested_state (in) 4925:Returns: 0 on success, -1 on error 4926 4927This copies the vcpu's kvm_nested_state struct from userspace to the kernel. 4928For the definition of struct kvm_nested_state, see KVM_GET_NESTED_STATE. 4929 49304.116 KVM_(UN)REGISTER_COALESCED_MMIO 4931------------------------------------- 4932 4933:Capability: KVM_CAP_COALESCED_MMIO (for coalesced mmio) 4934 KVM_CAP_COALESCED_PIO (for coalesced pio) 4935:Architectures: all 4936:Type: vm ioctl 4937:Parameters: struct kvm_coalesced_mmio_zone 4938:Returns: 0 on success, < 0 on error 4939 4940Coalesced I/O is a performance optimization that defers hardware 4941register write emulation so that userspace exits are avoided. It is 4942typically used to reduce the overhead of emulating frequently accessed 4943hardware registers. 4944 4945When a hardware register is configured for coalesced I/O, write accesses 4946do not exit to userspace and their value is recorded in a ring buffer 4947that is shared between kernel and userspace. 4948 4949Coalesced I/O is used if one or more write accesses to a hardware 4950register can be deferred until a read or a write to another hardware 4951register on the same device. This last access will cause a vmexit and 4952userspace will process accesses from the ring buffer before emulating 4953it. That will avoid exiting to userspace on repeated writes. 4954 4955Coalesced pio is based on coalesced mmio. There is little difference 4956between coalesced mmio and pio except that coalesced pio records accesses 4957to I/O ports. 4958 49594.117 KVM_CLEAR_DIRTY_LOG (vm ioctl) 4960------------------------------------ 4961 4962:Capability: KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 4963:Architectures: x86, arm64, mips 4964:Type: vm ioctl 4965:Parameters: struct kvm_clear_dirty_log (in) 4966:Returns: 0 on success, -1 on error 4967 4968:: 4969 4970 /* for KVM_CLEAR_DIRTY_LOG */ 4971 struct kvm_clear_dirty_log { 4972 __u32 slot; 4973 __u32 num_pages; 4974 __u64 first_page; 4975 union { 4976 void __user *dirty_bitmap; /* one bit per page */ 4977 __u64 padding; 4978 }; 4979 }; 4980 4981The ioctl clears the dirty status of pages in a memory slot, according to 4982the bitmap that is passed in struct kvm_clear_dirty_log's dirty_bitmap 4983field. Bit 0 of the bitmap corresponds to page "first_page" in the 4984memory slot, and num_pages is the size in bits of the input bitmap. 4985first_page must be a multiple of 64; num_pages must also be a multiple of 498664 unless first_page + num_pages is the size of the memory slot. For each 4987bit that is set in the input bitmap, the corresponding page is marked "clean" 4988in KVM's dirty bitmap, and dirty tracking is re-enabled for that page 4989(for example via write-protection, or by clearing the dirty bit in 4990a page table entry). 4991 4992If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies 4993the address space for which you want to clear the dirty status. See 4994KVM_SET_USER_MEMORY_REGION for details on the usage of slot field. 4995 4996This ioctl is mostly useful when KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 4997is enabled; for more information, see the description of the capability. 4998However, it can always be used as long as KVM_CHECK_EXTENSION confirms 4999that KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is present. 5000 50014.118 KVM_GET_SUPPORTED_HV_CPUID 5002-------------------------------- 5003 5004:Capability: KVM_CAP_HYPERV_CPUID (vcpu), KVM_CAP_SYS_HYPERV_CPUID (system) 5005:Architectures: x86 5006:Type: system ioctl, vcpu ioctl 5007:Parameters: struct kvm_cpuid2 (in/out) 5008:Returns: 0 on success, -1 on error 5009 5010:: 5011 5012 struct kvm_cpuid2 { 5013 __u32 nent; 5014 __u32 padding; 5015 struct kvm_cpuid_entry2 entries[0]; 5016 }; 5017 5018 struct kvm_cpuid_entry2 { 5019 __u32 function; 5020 __u32 index; 5021 __u32 flags; 5022 __u32 eax; 5023 __u32 ebx; 5024 __u32 ecx; 5025 __u32 edx; 5026 __u32 padding[3]; 5027 }; 5028 5029This ioctl returns x86 cpuid features leaves related to Hyper-V emulation in 5030KVM. Userspace can use the information returned by this ioctl to construct 5031cpuid information presented to guests consuming Hyper-V enlightenments (e.g. 5032Windows or Hyper-V guests). 5033 5034CPUID feature leaves returned by this ioctl are defined by Hyper-V Top Level 5035Functional Specification (TLFS). These leaves can't be obtained with 5036KVM_GET_SUPPORTED_CPUID ioctl because some of them intersect with KVM feature 5037leaves (0x40000000, 0x40000001). 5038 5039Currently, the following list of CPUID leaves are returned: 5040 5041 - HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS 5042 - HYPERV_CPUID_INTERFACE 5043 - HYPERV_CPUID_VERSION 5044 - HYPERV_CPUID_FEATURES 5045 - HYPERV_CPUID_ENLIGHTMENT_INFO 5046 - HYPERV_CPUID_IMPLEMENT_LIMITS 5047 - HYPERV_CPUID_NESTED_FEATURES 5048 - HYPERV_CPUID_SYNDBG_VENDOR_AND_MAX_FUNCTIONS 5049 - HYPERV_CPUID_SYNDBG_INTERFACE 5050 - HYPERV_CPUID_SYNDBG_PLATFORM_CAPABILITIES 5051 5052Userspace invokes KVM_GET_SUPPORTED_HV_CPUID by passing a kvm_cpuid2 structure 5053with the 'nent' field indicating the number of entries in the variable-size 5054array 'entries'. If the number of entries is too low to describe all Hyper-V 5055feature leaves, an error (E2BIG) is returned. If the number is more or equal 5056to the number of Hyper-V feature leaves, the 'nent' field is adjusted to the 5057number of valid entries in the 'entries' array, which is then filled. 5058 5059'index' and 'flags' fields in 'struct kvm_cpuid_entry2' are currently reserved, 5060userspace should not expect to get any particular value there. 5061 5062Note, vcpu version of KVM_GET_SUPPORTED_HV_CPUID is currently deprecated. Unlike 5063system ioctl which exposes all supported feature bits unconditionally, vcpu 5064version has the following quirks: 5065 5066- HYPERV_CPUID_NESTED_FEATURES leaf and HV_X64_ENLIGHTENED_VMCS_RECOMMENDED 5067 feature bit are only exposed when Enlightened VMCS was previously enabled 5068 on the corresponding vCPU (KVM_CAP_HYPERV_ENLIGHTENED_VMCS). 5069- HV_STIMER_DIRECT_MODE_AVAILABLE bit is only exposed with in-kernel LAPIC. 5070 (presumes KVM_CREATE_IRQCHIP has already been called). 5071 50724.119 KVM_ARM_VCPU_FINALIZE 5073--------------------------- 5074 5075:Architectures: arm64 5076:Type: vcpu ioctl 5077:Parameters: int feature (in) 5078:Returns: 0 on success, -1 on error 5079 5080Errors: 5081 5082 ====== ============================================================== 5083 EPERM feature not enabled, needs configuration, or already finalized 5084 EINVAL feature unknown or not present 5085 ====== ============================================================== 5086 5087Recognised values for feature: 5088 5089 ===== =========================================== 5090 arm64 KVM_ARM_VCPU_SVE (requires KVM_CAP_ARM_SVE) 5091 ===== =========================================== 5092 5093Finalizes the configuration of the specified vcpu feature. 5094 5095The vcpu must already have been initialised, enabling the affected feature, by 5096means of a successful KVM_ARM_VCPU_INIT call with the appropriate flag set in 5097features[]. 5098 5099For affected vcpu features, this is a mandatory step that must be performed 5100before the vcpu is fully usable. 5101 5102Between KVM_ARM_VCPU_INIT and KVM_ARM_VCPU_FINALIZE, the feature may be 5103configured by use of ioctls such as KVM_SET_ONE_REG. The exact configuration 5104that should be performed and how to do it are feature-dependent. 5105 5106Other calls that depend on a particular feature being finalized, such as 5107KVM_RUN, KVM_GET_REG_LIST, KVM_GET_ONE_REG and KVM_SET_ONE_REG, will fail with 5108-EPERM unless the feature has already been finalized by means of a 5109KVM_ARM_VCPU_FINALIZE call. 5110 5111See KVM_ARM_VCPU_INIT for details of vcpu features that require finalization 5112using this ioctl. 5113 51144.120 KVM_SET_PMU_EVENT_FILTER 5115------------------------------ 5116 5117:Capability: KVM_CAP_PMU_EVENT_FILTER 5118:Architectures: x86 5119:Type: vm ioctl 5120:Parameters: struct kvm_pmu_event_filter (in) 5121:Returns: 0 on success, -1 on error 5122 5123Errors: 5124 5125 ====== ============================================================ 5126 EFAULT args[0] cannot be accessed 5127 EINVAL args[0] contains invalid data in the filter or filter events 5128 E2BIG nevents is too large 5129 EBUSY not enough memory to allocate the filter 5130 ====== ============================================================ 5131 5132:: 5133 5134 struct kvm_pmu_event_filter { 5135 __u32 action; 5136 __u32 nevents; 5137 __u32 fixed_counter_bitmap; 5138 __u32 flags; 5139 __u32 pad[4]; 5140 __u64 events[0]; 5141 }; 5142 5143This ioctl restricts the set of PMU events the guest can program by limiting 5144which event select and unit mask combinations are permitted. 5145 5146The argument holds a list of filter events which will be allowed or denied. 5147 5148Filter events only control general purpose counters; fixed purpose counters 5149are controlled by the fixed_counter_bitmap. 5150 5151Valid values for 'flags':: 5152 5153``0`` 5154 5155To use this mode, clear the 'flags' field. 5156 5157In this mode each event will contain an event select + unit mask. 5158 5159When the guest attempts to program the PMU the guest's event select + 5160unit mask is compared against the filter events to determine whether the 5161guest should have access. 5162 5163``KVM_PMU_EVENT_FLAG_MASKED_EVENTS`` 5164:Capability: KVM_CAP_PMU_EVENT_MASKED_EVENTS 5165 5166In this mode each filter event will contain an event select, mask, match, and 5167exclude value. To encode a masked event use:: 5168 5169 KVM_PMU_ENCODE_MASKED_ENTRY() 5170 5171An encoded event will follow this layout:: 5172 5173 Bits Description 5174 ---- ----------- 5175 7:0 event select (low bits) 5176 15:8 umask match 5177 31:16 unused 5178 35:32 event select (high bits) 5179 36:54 unused 5180 55 exclude bit 5181 63:56 umask mask 5182 5183When the guest attempts to program the PMU, these steps are followed in 5184determining if the guest should have access: 5185 5186 1. Match the event select from the guest against the filter events. 5187 2. If a match is found, match the guest's unit mask to the mask and match 5188 values of the included filter events. 5189 I.e. (unit mask & mask) == match && !exclude. 5190 3. If a match is found, match the guest's unit mask to the mask and match 5191 values of the excluded filter events. 5192 I.e. (unit mask & mask) == match && exclude. 5193 4. 5194 a. If an included match is found and an excluded match is not found, filter 5195 the event. 5196 b. For everything else, do not filter the event. 5197 5. 5198 a. If the event is filtered and it's an allow list, allow the guest to 5199 program the event. 5200 b. If the event is filtered and it's a deny list, do not allow the guest to 5201 program the event. 5202 5203When setting a new pmu event filter, -EINVAL will be returned if any of the 5204unused fields are set or if any of the high bits (35:32) in the event 5205select are set when called on Intel. 5206 5207Valid values for 'action':: 5208 5209 #define KVM_PMU_EVENT_ALLOW 0 5210 #define KVM_PMU_EVENT_DENY 1 5211 5212Via this API, KVM userspace can also control the behavior of the VM's fixed 5213counters (if any) by configuring the "action" and "fixed_counter_bitmap" fields. 5214 5215Specifically, KVM follows the following pseudo-code when determining whether to 5216allow the guest FixCtr[i] to count its pre-defined fixed event:: 5217 5218 FixCtr[i]_is_allowed = (action == ALLOW) && (bitmap & BIT(i)) || 5219 (action == DENY) && !(bitmap & BIT(i)); 5220 FixCtr[i]_is_denied = !FixCtr[i]_is_allowed; 5221 5222KVM always consumes fixed_counter_bitmap, it's userspace's responsibility to 5223ensure fixed_counter_bitmap is set correctly, e.g. if userspace wants to define 5224a filter that only affects general purpose counters. 5225 5226Note, the "events" field also applies to fixed counters' hardcoded event_select 5227and unit_mask values. "fixed_counter_bitmap" has higher priority than "events" 5228if there is a contradiction between the two. 5229 52304.121 KVM_PPC_SVM_OFF 5231--------------------- 5232 5233:Capability: basic 5234:Architectures: powerpc 5235:Type: vm ioctl 5236:Parameters: none 5237:Returns: 0 on successful completion, 5238 5239Errors: 5240 5241 ====== ================================================================ 5242 EINVAL if ultravisor failed to terminate the secure guest 5243 ENOMEM if hypervisor failed to allocate new radix page tables for guest 5244 ====== ================================================================ 5245 5246This ioctl is used to turn off the secure mode of the guest or transition 5247the guest from secure mode to normal mode. This is invoked when the guest 5248is reset. This has no effect if called for a normal guest. 5249 5250This ioctl issues an ultravisor call to terminate the secure guest, 5251unpins the VPA pages and releases all the device pages that are used to 5252track the secure pages by hypervisor. 5253 52544.122 KVM_S390_NORMAL_RESET 5255--------------------------- 5256 5257:Capability: KVM_CAP_S390_VCPU_RESETS 5258:Architectures: s390 5259:Type: vcpu ioctl 5260:Parameters: none 5261:Returns: 0 5262 5263This ioctl resets VCPU registers and control structures according to 5264the cpu reset definition in the POP (Principles Of Operation). 5265 52664.123 KVM_S390_INITIAL_RESET 5267---------------------------- 5268 5269:Capability: none 5270:Architectures: s390 5271:Type: vcpu ioctl 5272:Parameters: none 5273:Returns: 0 5274 5275This ioctl resets VCPU registers and control structures according to 5276the initial cpu reset definition in the POP. However, the cpu is not 5277put into ESA mode. This reset is a superset of the normal reset. 5278 52794.124 KVM_S390_CLEAR_RESET 5280-------------------------- 5281 5282:Capability: KVM_CAP_S390_VCPU_RESETS 5283:Architectures: s390 5284:Type: vcpu ioctl 5285:Parameters: none 5286:Returns: 0 5287 5288This ioctl resets VCPU registers and control structures according to 5289the clear cpu reset definition in the POP. However, the cpu is not put 5290into ESA mode. This reset is a superset of the initial reset. 5291 5292 52934.125 KVM_S390_PV_COMMAND 5294------------------------- 5295 5296:Capability: KVM_CAP_S390_PROTECTED 5297:Architectures: s390 5298:Type: vm ioctl 5299:Parameters: struct kvm_pv_cmd 5300:Returns: 0 on success, < 0 on error 5301 5302:: 5303 5304 struct kvm_pv_cmd { 5305 __u32 cmd; /* Command to be executed */ 5306 __u16 rc; /* Ultravisor return code */ 5307 __u16 rrc; /* Ultravisor return reason code */ 5308 __u64 data; /* Data or address */ 5309 __u32 flags; /* flags for future extensions. Must be 0 for now */ 5310 __u32 reserved[3]; 5311 }; 5312 5313**Ultravisor return codes** 5314The Ultravisor return (reason) codes are provided by the kernel if a 5315Ultravisor call has been executed to achieve the results expected by 5316the command. Therefore they are independent of the IOCTL return 5317code. If KVM changes `rc`, its value will always be greater than 0 5318hence setting it to 0 before issuing a PV command is advised to be 5319able to detect a change of `rc`. 5320 5321**cmd values:** 5322 5323KVM_PV_ENABLE 5324 Allocate memory and register the VM with the Ultravisor, thereby 5325 donating memory to the Ultravisor that will become inaccessible to 5326 KVM. All existing CPUs are converted to protected ones. After this 5327 command has succeeded, any CPU added via hotplug will become 5328 protected during its creation as well. 5329 5330 Errors: 5331 5332 ===== ============================= 5333 EINTR an unmasked signal is pending 5334 ===== ============================= 5335 5336KVM_PV_DISABLE 5337 Deregister the VM from the Ultravisor and reclaim the memory that had 5338 been donated to the Ultravisor, making it usable by the kernel again. 5339 All registered VCPUs are converted back to non-protected ones. If a 5340 previous protected VM had been prepared for asynchronous teardown with 5341 KVM_PV_ASYNC_CLEANUP_PREPARE and not subsequently torn down with 5342 KVM_PV_ASYNC_CLEANUP_PERFORM, it will be torn down in this call 5343 together with the current protected VM. 5344 5345KVM_PV_VM_SET_SEC_PARMS 5346 Pass the image header from VM memory to the Ultravisor in 5347 preparation of image unpacking and verification. 5348 5349KVM_PV_VM_UNPACK 5350 Unpack (protect and decrypt) a page of the encrypted boot image. 5351 5352KVM_PV_VM_VERIFY 5353 Verify the integrity of the unpacked image. Only if this succeeds, 5354 KVM is allowed to start protected VCPUs. 5355 5356KVM_PV_INFO 5357 :Capability: KVM_CAP_S390_PROTECTED_DUMP 5358 5359 Presents an API that provides Ultravisor related data to userspace 5360 via subcommands. len_max is the size of the user space buffer, 5361 len_written is KVM's indication of how much bytes of that buffer 5362 were actually written to. len_written can be used to determine the 5363 valid fields if more response fields are added in the future. 5364 5365 :: 5366 5367 enum pv_cmd_info_id { 5368 KVM_PV_INFO_VM, 5369 KVM_PV_INFO_DUMP, 5370 }; 5371 5372 struct kvm_s390_pv_info_header { 5373 __u32 id; 5374 __u32 len_max; 5375 __u32 len_written; 5376 __u32 reserved; 5377 }; 5378 5379 struct kvm_s390_pv_info { 5380 struct kvm_s390_pv_info_header header; 5381 struct kvm_s390_pv_info_dump dump; 5382 struct kvm_s390_pv_info_vm vm; 5383 }; 5384 5385**subcommands:** 5386 5387 KVM_PV_INFO_VM 5388 This subcommand provides basic Ultravisor information for PV 5389 hosts. These values are likely also exported as files in the sysfs 5390 firmware UV query interface but they are more easily available to 5391 programs in this API. 5392 5393 The installed calls and feature_indication members provide the 5394 installed UV calls and the UV's other feature indications. 5395 5396 The max_* members provide information about the maximum number of PV 5397 vcpus, PV guests and PV guest memory size. 5398 5399 :: 5400 5401 struct kvm_s390_pv_info_vm { 5402 __u64 inst_calls_list[4]; 5403 __u64 max_cpus; 5404 __u64 max_guests; 5405 __u64 max_guest_addr; 5406 __u64 feature_indication; 5407 }; 5408 5409 5410 KVM_PV_INFO_DUMP 5411 This subcommand provides information related to dumping PV guests. 5412 5413 :: 5414 5415 struct kvm_s390_pv_info_dump { 5416 __u64 dump_cpu_buffer_len; 5417 __u64 dump_config_mem_buffer_per_1m; 5418 __u64 dump_config_finalize_len; 5419 }; 5420 5421KVM_PV_DUMP 5422 :Capability: KVM_CAP_S390_PROTECTED_DUMP 5423 5424 Presents an API that provides calls which facilitate dumping a 5425 protected VM. 5426 5427 :: 5428 5429 struct kvm_s390_pv_dmp { 5430 __u64 subcmd; 5431 __u64 buff_addr; 5432 __u64 buff_len; 5433 __u64 gaddr; /* For dump storage state */ 5434 }; 5435 5436 **subcommands:** 5437 5438 KVM_PV_DUMP_INIT 5439 Initializes the dump process of a protected VM. If this call does 5440 not succeed all other subcommands will fail with -EINVAL. This 5441 subcommand will return -EINVAL if a dump process has not yet been 5442 completed. 5443 5444 Not all PV vms can be dumped, the owner needs to set `dump 5445 allowed` PCF bit 34 in the SE header to allow dumping. 5446 5447 KVM_PV_DUMP_CONFIG_STOR_STATE 5448 Stores `buff_len` bytes of tweak component values starting with 5449 the 1MB block specified by the absolute guest address 5450 (`gaddr`). `buff_len` needs to be `conf_dump_storage_state_len` 5451 aligned and at least >= the `conf_dump_storage_state_len` value 5452 provided by the dump uv_info data. buff_user might be written to 5453 even if an error rc is returned. For instance if we encounter a 5454 fault after writing the first page of data. 5455 5456 KVM_PV_DUMP_COMPLETE 5457 If the subcommand succeeds it completes the dump process and lets 5458 KVM_PV_DUMP_INIT be called again. 5459 5460 On success `conf_dump_finalize_len` bytes of completion data will be 5461 stored to the `buff_addr`. The completion data contains a key 5462 derivation seed, IV, tweak nonce and encryption keys as well as an 5463 authentication tag all of which are needed to decrypt the dump at a 5464 later time. 5465 5466KVM_PV_ASYNC_CLEANUP_PREPARE 5467 :Capability: KVM_CAP_S390_PROTECTED_ASYNC_DISABLE 5468 5469 Prepare the current protected VM for asynchronous teardown. Most 5470 resources used by the current protected VM will be set aside for a 5471 subsequent asynchronous teardown. The current protected VM will then 5472 resume execution immediately as non-protected. There can be at most 5473 one protected VM prepared for asynchronous teardown at any time. If 5474 a protected VM had already been prepared for teardown without 5475 subsequently calling KVM_PV_ASYNC_CLEANUP_PERFORM, this call will 5476 fail. In that case, the userspace process should issue a normal 5477 KVM_PV_DISABLE. The resources set aside with this call will need to 5478 be cleaned up with a subsequent call to KVM_PV_ASYNC_CLEANUP_PERFORM 5479 or KVM_PV_DISABLE, otherwise they will be cleaned up when KVM 5480 terminates. KVM_PV_ASYNC_CLEANUP_PREPARE can be called again as soon 5481 as cleanup starts, i.e. before KVM_PV_ASYNC_CLEANUP_PERFORM finishes. 5482 5483KVM_PV_ASYNC_CLEANUP_PERFORM 5484 :Capability: KVM_CAP_S390_PROTECTED_ASYNC_DISABLE 5485 5486 Tear down the protected VM previously prepared for teardown with 5487 KVM_PV_ASYNC_CLEANUP_PREPARE. The resources that had been set aside 5488 will be freed during the execution of this command. This PV command 5489 should ideally be issued by userspace from a separate thread. If a 5490 fatal signal is received (or the process terminates naturally), the 5491 command will terminate immediately without completing, and the normal 5492 KVM shutdown procedure will take care of cleaning up all remaining 5493 protected VMs, including the ones whose teardown was interrupted by 5494 process termination. 5495 54964.126 KVM_XEN_HVM_SET_ATTR 5497-------------------------- 5498 5499:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 5500:Architectures: x86 5501:Type: vm ioctl 5502:Parameters: struct kvm_xen_hvm_attr 5503:Returns: 0 on success, < 0 on error 5504 5505:: 5506 5507 struct kvm_xen_hvm_attr { 5508 __u16 type; 5509 __u16 pad[3]; 5510 union { 5511 __u8 long_mode; 5512 __u8 vector; 5513 __u8 runstate_update_flag; 5514 union { 5515 __u64 gfn; 5516 __u64 hva; 5517 } shared_info; 5518 struct { 5519 __u32 send_port; 5520 __u32 type; /* EVTCHNSTAT_ipi / EVTCHNSTAT_interdomain */ 5521 __u32 flags; 5522 union { 5523 struct { 5524 __u32 port; 5525 __u32 vcpu; 5526 __u32 priority; 5527 } port; 5528 struct { 5529 __u32 port; /* Zero for eventfd */ 5530 __s32 fd; 5531 } eventfd; 5532 __u32 padding[4]; 5533 } deliver; 5534 } evtchn; 5535 __u32 xen_version; 5536 __u64 pad[8]; 5537 } u; 5538 }; 5539 5540type values: 5541 5542KVM_XEN_ATTR_TYPE_LONG_MODE 5543 Sets the ABI mode of the VM to 32-bit or 64-bit (long mode). This 5544 determines the layout of the shared_info page exposed to the VM. 5545 5546KVM_XEN_ATTR_TYPE_SHARED_INFO 5547 Sets the guest physical frame number at which the Xen shared_info 5548 page resides. Note that although Xen places vcpu_info for the first 5549 32 vCPUs in the shared_info page, KVM does not automatically do so 5550 and instead requires that KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO or 5551 KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO_HVA be used explicitly even when 5552 the vcpu_info for a given vCPU resides at the "default" location 5553 in the shared_info page. This is because KVM may not be aware of 5554 the Xen CPU id which is used as the index into the vcpu_info[] 5555 array, so may know the correct default location. 5556 5557 Note that the shared_info page may be constantly written to by KVM; 5558 it contains the event channel bitmap used to deliver interrupts to 5559 a Xen guest, amongst other things. It is exempt from dirty tracking 5560 mechanisms — KVM will not explicitly mark the page as dirty each 5561 time an event channel interrupt is delivered to the guest! Thus, 5562 userspace should always assume that the designated GFN is dirty if 5563 any vCPU has been running or any event channel interrupts can be 5564 routed to the guest. 5565 5566 Setting the gfn to KVM_XEN_INVALID_GFN will disable the shared_info 5567 page. 5568 5569KVM_XEN_ATTR_TYPE_SHARED_INFO_HVA 5570 If the KVM_XEN_HVM_CONFIG_SHARED_INFO_HVA flag is also set in the 5571 Xen capabilities, then this attribute may be used to set the 5572 userspace address at which the shared_info page resides, which 5573 will always be fixed in the VMM regardless of where it is mapped 5574 in guest physical address space. This attribute should be used in 5575 preference to KVM_XEN_ATTR_TYPE_SHARED_INFO as it avoids 5576 unnecessary invalidation of an internal cache when the page is 5577 re-mapped in guest physcial address space. 5578 5579 Setting the hva to zero will disable the shared_info page. 5580 5581KVM_XEN_ATTR_TYPE_UPCALL_VECTOR 5582 Sets the exception vector used to deliver Xen event channel upcalls. 5583 This is the HVM-wide vector injected directly by the hypervisor 5584 (not through the local APIC), typically configured by a guest via 5585 HVM_PARAM_CALLBACK_IRQ. This can be disabled again (e.g. for guest 5586 SHUTDOWN_soft_reset) by setting it to zero. 5587 5588KVM_XEN_ATTR_TYPE_EVTCHN 5589 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5590 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It configures 5591 an outbound port number for interception of EVTCHNOP_send requests 5592 from the guest. A given sending port number may be directed back to 5593 a specified vCPU (by APIC ID) / port / priority on the guest, or to 5594 trigger events on an eventfd. The vCPU and priority can be changed 5595 by setting KVM_XEN_EVTCHN_UPDATE in a subsequent call, but other 5596 fields cannot change for a given sending port. A port mapping is 5597 removed by using KVM_XEN_EVTCHN_DEASSIGN in the flags field. Passing 5598 KVM_XEN_EVTCHN_RESET in the flags field removes all interception of 5599 outbound event channels. The values of the flags field are mutually 5600 exclusive and cannot be combined as a bitmask. 5601 5602KVM_XEN_ATTR_TYPE_XEN_VERSION 5603 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5604 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It configures 5605 the 32-bit version code returned to the guest when it invokes the 5606 XENVER_version call; typically (XEN_MAJOR << 16 | XEN_MINOR). PV 5607 Xen guests will often use this to as a dummy hypercall to trigger 5608 event channel delivery, so responding within the kernel without 5609 exiting to userspace is beneficial. 5610 5611KVM_XEN_ATTR_TYPE_RUNSTATE_UPDATE_FLAG 5612 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5613 support for KVM_XEN_HVM_CONFIG_RUNSTATE_UPDATE_FLAG. It enables the 5614 XEN_RUNSTATE_UPDATE flag which allows guest vCPUs to safely read 5615 other vCPUs' vcpu_runstate_info. Xen guests enable this feature via 5616 the VMASST_TYPE_runstate_update_flag of the HYPERVISOR_vm_assist 5617 hypercall. 5618 56194.127 KVM_XEN_HVM_GET_ATTR 5620-------------------------- 5621 5622:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 5623:Architectures: x86 5624:Type: vm ioctl 5625:Parameters: struct kvm_xen_hvm_attr 5626:Returns: 0 on success, < 0 on error 5627 5628Allows Xen VM attributes to be read. For the structure and types, 5629see KVM_XEN_HVM_SET_ATTR above. The KVM_XEN_ATTR_TYPE_EVTCHN 5630attribute cannot be read. 5631 56324.128 KVM_XEN_VCPU_SET_ATTR 5633--------------------------- 5634 5635:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 5636:Architectures: x86 5637:Type: vcpu ioctl 5638:Parameters: struct kvm_xen_vcpu_attr 5639:Returns: 0 on success, < 0 on error 5640 5641:: 5642 5643 struct kvm_xen_vcpu_attr { 5644 __u16 type; 5645 __u16 pad[3]; 5646 union { 5647 __u64 gpa; 5648 __u64 pad[4]; 5649 struct { 5650 __u64 state; 5651 __u64 state_entry_time; 5652 __u64 time_running; 5653 __u64 time_runnable; 5654 __u64 time_blocked; 5655 __u64 time_offline; 5656 } runstate; 5657 __u32 vcpu_id; 5658 struct { 5659 __u32 port; 5660 __u32 priority; 5661 __u64 expires_ns; 5662 } timer; 5663 __u8 vector; 5664 } u; 5665 }; 5666 5667type values: 5668 5669KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO 5670 Sets the guest physical address of the vcpu_info for a given vCPU. 5671 As with the shared_info page for the VM, the corresponding page may be 5672 dirtied at any time if event channel interrupt delivery is enabled, so 5673 userspace should always assume that the page is dirty without relying 5674 on dirty logging. Setting the gpa to KVM_XEN_INVALID_GPA will disable 5675 the vcpu_info. 5676 5677KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO_HVA 5678 If the KVM_XEN_HVM_CONFIG_SHARED_INFO_HVA flag is also set in the 5679 Xen capabilities, then this attribute may be used to set the 5680 userspace address of the vcpu_info for a given vCPU. It should 5681 only be used when the vcpu_info resides at the "default" location 5682 in the shared_info page. In this case it is safe to assume the 5683 userspace address will not change, because the shared_info page is 5684 an overlay on guest memory and remains at a fixed host address 5685 regardless of where it is mapped in guest physical address space 5686 and hence unnecessary invalidation of an internal cache may be 5687 avoided if the guest memory layout is modified. 5688 If the vcpu_info does not reside at the "default" location then 5689 it is not guaranteed to remain at the same host address and 5690 hence the aforementioned cache invalidation is required. 5691 5692KVM_XEN_VCPU_ATTR_TYPE_VCPU_TIME_INFO 5693 Sets the guest physical address of an additional pvclock structure 5694 for a given vCPU. This is typically used for guest vsyscall support. 5695 Setting the gpa to KVM_XEN_INVALID_GPA will disable the structure. 5696 5697KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR 5698 Sets the guest physical address of the vcpu_runstate_info for a given 5699 vCPU. This is how a Xen guest tracks CPU state such as steal time. 5700 Setting the gpa to KVM_XEN_INVALID_GPA will disable the runstate area. 5701 5702KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_CURRENT 5703 Sets the runstate (RUNSTATE_running/_runnable/_blocked/_offline) of 5704 the given vCPU from the .u.runstate.state member of the structure. 5705 KVM automatically accounts running and runnable time but blocked 5706 and offline states are only entered explicitly. 5707 5708KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_DATA 5709 Sets all fields of the vCPU runstate data from the .u.runstate member 5710 of the structure, including the current runstate. The state_entry_time 5711 must equal the sum of the other four times. 5712 5713KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST 5714 This *adds* the contents of the .u.runstate members of the structure 5715 to the corresponding members of the given vCPU's runstate data, thus 5716 permitting atomic adjustments to the runstate times. The adjustment 5717 to the state_entry_time must equal the sum of the adjustments to the 5718 other four times. The state field must be set to -1, or to a valid 5719 runstate value (RUNSTATE_running, RUNSTATE_runnable, RUNSTATE_blocked 5720 or RUNSTATE_offline) to set the current accounted state as of the 5721 adjusted state_entry_time. 5722 5723KVM_XEN_VCPU_ATTR_TYPE_VCPU_ID 5724 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5725 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It sets the Xen 5726 vCPU ID of the given vCPU, to allow timer-related VCPU operations to 5727 be intercepted by KVM. 5728 5729KVM_XEN_VCPU_ATTR_TYPE_TIMER 5730 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5731 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It sets the 5732 event channel port/priority for the VIRQ_TIMER of the vCPU, as well 5733 as allowing a pending timer to be saved/restored. Setting the timer 5734 port to zero disables kernel handling of the singleshot timer. 5735 5736KVM_XEN_VCPU_ATTR_TYPE_UPCALL_VECTOR 5737 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5738 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It sets the 5739 per-vCPU local APIC upcall vector, configured by a Xen guest with 5740 the HVMOP_set_evtchn_upcall_vector hypercall. This is typically 5741 used by Windows guests, and is distinct from the HVM-wide upcall 5742 vector configured with HVM_PARAM_CALLBACK_IRQ. It is disabled by 5743 setting the vector to zero. 5744 5745 57464.129 KVM_XEN_VCPU_GET_ATTR 5747--------------------------- 5748 5749:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 5750:Architectures: x86 5751:Type: vcpu ioctl 5752:Parameters: struct kvm_xen_vcpu_attr 5753:Returns: 0 on success, < 0 on error 5754 5755Allows Xen vCPU attributes to be read. For the structure and types, 5756see KVM_XEN_VCPU_SET_ATTR above. 5757 5758The KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST type may not be used 5759with the KVM_XEN_VCPU_GET_ATTR ioctl. 5760 57614.130 KVM_ARM_MTE_COPY_TAGS 5762--------------------------- 5763 5764:Capability: KVM_CAP_ARM_MTE 5765:Architectures: arm64 5766:Type: vm ioctl 5767:Parameters: struct kvm_arm_copy_mte_tags 5768:Returns: number of bytes copied, < 0 on error (-EINVAL for incorrect 5769 arguments, -EFAULT if memory cannot be accessed). 5770 5771:: 5772 5773 struct kvm_arm_copy_mte_tags { 5774 __u64 guest_ipa; 5775 __u64 length; 5776 void __user *addr; 5777 __u64 flags; 5778 __u64 reserved[2]; 5779 }; 5780 5781Copies Memory Tagging Extension (MTE) tags to/from guest tag memory. The 5782``guest_ipa`` and ``length`` fields must be ``PAGE_SIZE`` aligned. 5783``length`` must not be bigger than 2^31 - PAGE_SIZE bytes. The ``addr`` 5784field must point to a buffer which the tags will be copied to or from. 5785 5786``flags`` specifies the direction of copy, either ``KVM_ARM_TAGS_TO_GUEST`` or 5787``KVM_ARM_TAGS_FROM_GUEST``. 5788 5789The size of the buffer to store the tags is ``(length / 16)`` bytes 5790(granules in MTE are 16 bytes long). Each byte contains a single tag 5791value. This matches the format of ``PTRACE_PEEKMTETAGS`` and 5792``PTRACE_POKEMTETAGS``. 5793 5794If an error occurs before any data is copied then a negative error code is 5795returned. If some tags have been copied before an error occurs then the number 5796of bytes successfully copied is returned. If the call completes successfully 5797then ``length`` is returned. 5798 57994.131 KVM_GET_SREGS2 5800-------------------- 5801 5802:Capability: KVM_CAP_SREGS2 5803:Architectures: x86 5804:Type: vcpu ioctl 5805:Parameters: struct kvm_sregs2 (out) 5806:Returns: 0 on success, -1 on error 5807 5808Reads special registers from the vcpu. 5809This ioctl (when supported) replaces the KVM_GET_SREGS. 5810 5811:: 5812 5813 struct kvm_sregs2 { 5814 /* out (KVM_GET_SREGS2) / in (KVM_SET_SREGS2) */ 5815 struct kvm_segment cs, ds, es, fs, gs, ss; 5816 struct kvm_segment tr, ldt; 5817 struct kvm_dtable gdt, idt; 5818 __u64 cr0, cr2, cr3, cr4, cr8; 5819 __u64 efer; 5820 __u64 apic_base; 5821 __u64 flags; 5822 __u64 pdptrs[4]; 5823 }; 5824 5825flags values for ``kvm_sregs2``: 5826 5827``KVM_SREGS2_FLAGS_PDPTRS_VALID`` 5828 5829 Indicates that the struct contains valid PDPTR values. 5830 5831 58324.132 KVM_SET_SREGS2 5833-------------------- 5834 5835:Capability: KVM_CAP_SREGS2 5836:Architectures: x86 5837:Type: vcpu ioctl 5838:Parameters: struct kvm_sregs2 (in) 5839:Returns: 0 on success, -1 on error 5840 5841Writes special registers into the vcpu. 5842See KVM_GET_SREGS2 for the data structures. 5843This ioctl (when supported) replaces the KVM_SET_SREGS. 5844 58454.133 KVM_GET_STATS_FD 5846---------------------- 5847 5848:Capability: KVM_CAP_STATS_BINARY_FD 5849:Architectures: all 5850:Type: vm ioctl, vcpu ioctl 5851:Parameters: none 5852:Returns: statistics file descriptor on success, < 0 on error 5853 5854Errors: 5855 5856 ====== ====================================================== 5857 ENOMEM if the fd could not be created due to lack of memory 5858 EMFILE if the number of opened files exceeds the limit 5859 ====== ====================================================== 5860 5861The returned file descriptor can be used to read VM/vCPU statistics data in 5862binary format. The data in the file descriptor consists of four blocks 5863organized as follows: 5864 5865+-------------+ 5866| Header | 5867+-------------+ 5868| id string | 5869+-------------+ 5870| Descriptors | 5871+-------------+ 5872| Stats Data | 5873+-------------+ 5874 5875Apart from the header starting at offset 0, please be aware that it is 5876not guaranteed that the four blocks are adjacent or in the above order; 5877the offsets of the id, descriptors and data blocks are found in the 5878header. However, all four blocks are aligned to 64 bit offsets in the 5879file and they do not overlap. 5880 5881All blocks except the data block are immutable. Userspace can read them 5882only one time after retrieving the file descriptor, and then use ``pread`` or 5883``lseek`` to read the statistics repeatedly. 5884 5885All data is in system endianness. 5886 5887The format of the header is as follows:: 5888 5889 struct kvm_stats_header { 5890 __u32 flags; 5891 __u32 name_size; 5892 __u32 num_desc; 5893 __u32 id_offset; 5894 __u32 desc_offset; 5895 __u32 data_offset; 5896 }; 5897 5898The ``flags`` field is not used at the moment. It is always read as 0. 5899 5900The ``name_size`` field is the size (in byte) of the statistics name string 5901(including trailing '\0') which is contained in the "id string" block and 5902appended at the end of every descriptor. 5903 5904The ``num_desc`` field is the number of descriptors that are included in the 5905descriptor block. (The actual number of values in the data block may be 5906larger, since each descriptor may comprise more than one value). 5907 5908The ``id_offset`` field is the offset of the id string from the start of the 5909file indicated by the file descriptor. It is a multiple of 8. 5910 5911The ``desc_offset`` field is the offset of the Descriptors block from the start 5912of the file indicated by the file descriptor. It is a multiple of 8. 5913 5914The ``data_offset`` field is the offset of the Stats Data block from the start 5915of the file indicated by the file descriptor. It is a multiple of 8. 5916 5917The id string block contains a string which identifies the file descriptor on 5918which KVM_GET_STATS_FD was invoked. The size of the block, including the 5919trailing ``'\0'``, is indicated by the ``name_size`` field in the header. 5920 5921The descriptors block is only needed to be read once for the lifetime of the 5922file descriptor contains a sequence of ``struct kvm_stats_desc``, each followed 5923by a string of size ``name_size``. 5924:: 5925 5926 #define KVM_STATS_TYPE_SHIFT 0 5927 #define KVM_STATS_TYPE_MASK (0xF << KVM_STATS_TYPE_SHIFT) 5928 #define KVM_STATS_TYPE_CUMULATIVE (0x0 << KVM_STATS_TYPE_SHIFT) 5929 #define KVM_STATS_TYPE_INSTANT (0x1 << KVM_STATS_TYPE_SHIFT) 5930 #define KVM_STATS_TYPE_PEAK (0x2 << KVM_STATS_TYPE_SHIFT) 5931 #define KVM_STATS_TYPE_LINEAR_HIST (0x3 << KVM_STATS_TYPE_SHIFT) 5932 #define KVM_STATS_TYPE_LOG_HIST (0x4 << KVM_STATS_TYPE_SHIFT) 5933 #define KVM_STATS_TYPE_MAX KVM_STATS_TYPE_LOG_HIST 5934 5935 #define KVM_STATS_UNIT_SHIFT 4 5936 #define KVM_STATS_UNIT_MASK (0xF << KVM_STATS_UNIT_SHIFT) 5937 #define KVM_STATS_UNIT_NONE (0x0 << KVM_STATS_UNIT_SHIFT) 5938 #define KVM_STATS_UNIT_BYTES (0x1 << KVM_STATS_UNIT_SHIFT) 5939 #define KVM_STATS_UNIT_SECONDS (0x2 << KVM_STATS_UNIT_SHIFT) 5940 #define KVM_STATS_UNIT_CYCLES (0x3 << KVM_STATS_UNIT_SHIFT) 5941 #define KVM_STATS_UNIT_BOOLEAN (0x4 << KVM_STATS_UNIT_SHIFT) 5942 #define KVM_STATS_UNIT_MAX KVM_STATS_UNIT_BOOLEAN 5943 5944 #define KVM_STATS_BASE_SHIFT 8 5945 #define KVM_STATS_BASE_MASK (0xF << KVM_STATS_BASE_SHIFT) 5946 #define KVM_STATS_BASE_POW10 (0x0 << KVM_STATS_BASE_SHIFT) 5947 #define KVM_STATS_BASE_POW2 (0x1 << KVM_STATS_BASE_SHIFT) 5948 #define KVM_STATS_BASE_MAX KVM_STATS_BASE_POW2 5949 5950 struct kvm_stats_desc { 5951 __u32 flags; 5952 __s16 exponent; 5953 __u16 size; 5954 __u32 offset; 5955 __u32 bucket_size; 5956 char name[]; 5957 }; 5958 5959The ``flags`` field contains the type and unit of the statistics data described 5960by this descriptor. Its endianness is CPU native. 5961The following flags are supported: 5962 5963Bits 0-3 of ``flags`` encode the type: 5964 5965 * ``KVM_STATS_TYPE_CUMULATIVE`` 5966 The statistics reports a cumulative count. The value of data can only be increased. 5967 Most of the counters used in KVM are of this type. 5968 The corresponding ``size`` field for this type is always 1. 5969 All cumulative statistics data are read/write. 5970 * ``KVM_STATS_TYPE_INSTANT`` 5971 The statistics reports an instantaneous value. Its value can be increased or 5972 decreased. This type is usually used as a measurement of some resources, 5973 like the number of dirty pages, the number of large pages, etc. 5974 All instant statistics are read only. 5975 The corresponding ``size`` field for this type is always 1. 5976 * ``KVM_STATS_TYPE_PEAK`` 5977 The statistics data reports a peak value, for example the maximum number 5978 of items in a hash table bucket, the longest time waited and so on. 5979 The value of data can only be increased. 5980 The corresponding ``size`` field for this type is always 1. 5981 * ``KVM_STATS_TYPE_LINEAR_HIST`` 5982 The statistic is reported as a linear histogram. The number of 5983 buckets is specified by the ``size`` field. The size of buckets is specified 5984 by the ``hist_param`` field. The range of the Nth bucket (1 <= N < ``size``) 5985 is [``hist_param``*(N-1), ``hist_param``*N), while the range of the last 5986 bucket is [``hist_param``*(``size``-1), +INF). (+INF means positive infinity 5987 value.) 5988 * ``KVM_STATS_TYPE_LOG_HIST`` 5989 The statistic is reported as a logarithmic histogram. The number of 5990 buckets is specified by the ``size`` field. The range of the first bucket is 5991 [0, 1), while the range of the last bucket is [pow(2, ``size``-2), +INF). 5992 Otherwise, The Nth bucket (1 < N < ``size``) covers 5993 [pow(2, N-2), pow(2, N-1)). 5994 5995Bits 4-7 of ``flags`` encode the unit: 5996 5997 * ``KVM_STATS_UNIT_NONE`` 5998 There is no unit for the value of statistics data. This usually means that 5999 the value is a simple counter of an event. 6000 * ``KVM_STATS_UNIT_BYTES`` 6001 It indicates that the statistics data is used to measure memory size, in the 6002 unit of Byte, KiByte, MiByte, GiByte, etc. The unit of the data is 6003 determined by the ``exponent`` field in the descriptor. 6004 * ``KVM_STATS_UNIT_SECONDS`` 6005 It indicates that the statistics data is used to measure time or latency. 6006 * ``KVM_STATS_UNIT_CYCLES`` 6007 It indicates that the statistics data is used to measure CPU clock cycles. 6008 * ``KVM_STATS_UNIT_BOOLEAN`` 6009 It indicates that the statistic will always be either 0 or 1. Boolean 6010 statistics of "peak" type will never go back from 1 to 0. Boolean 6011 statistics can be linear histograms (with two buckets) but not logarithmic 6012 histograms. 6013 6014Note that, in the case of histograms, the unit applies to the bucket 6015ranges, while the bucket value indicates how many samples fell in the 6016bucket's range. 6017 6018Bits 8-11 of ``flags``, together with ``exponent``, encode the scale of the 6019unit: 6020 6021 * ``KVM_STATS_BASE_POW10`` 6022 The scale is based on power of 10. It is used for measurement of time and 6023 CPU clock cycles. For example, an exponent of -9 can be used with 6024 ``KVM_STATS_UNIT_SECONDS`` to express that the unit is nanoseconds. 6025 * ``KVM_STATS_BASE_POW2`` 6026 The scale is based on power of 2. It is used for measurement of memory size. 6027 For example, an exponent of 20 can be used with ``KVM_STATS_UNIT_BYTES`` to 6028 express that the unit is MiB. 6029 6030The ``size`` field is the number of values of this statistics data. Its 6031value is usually 1 for most of simple statistics. 1 means it contains an 6032unsigned 64bit data. 6033 6034The ``offset`` field is the offset from the start of Data Block to the start of 6035the corresponding statistics data. 6036 6037The ``bucket_size`` field is used as a parameter for histogram statistics data. 6038It is only used by linear histogram statistics data, specifying the size of a 6039bucket in the unit expressed by bits 4-11 of ``flags`` together with ``exponent``. 6040 6041The ``name`` field is the name string of the statistics data. The name string 6042starts at the end of ``struct kvm_stats_desc``. The maximum length including 6043the trailing ``'\0'``, is indicated by ``name_size`` in the header. 6044 6045The Stats Data block contains an array of 64-bit values in the same order 6046as the descriptors in Descriptors block. 6047 60484.134 KVM_GET_XSAVE2 6049-------------------- 6050 6051:Capability: KVM_CAP_XSAVE2 6052:Architectures: x86 6053:Type: vcpu ioctl 6054:Parameters: struct kvm_xsave (out) 6055:Returns: 0 on success, -1 on error 6056 6057 6058:: 6059 6060 struct kvm_xsave { 6061 __u32 region[1024]; 6062 __u32 extra[0]; 6063 }; 6064 6065This ioctl would copy current vcpu's xsave struct to the userspace. It 6066copies as many bytes as are returned by KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) 6067when invoked on the vm file descriptor. The size value returned by 6068KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) will always be at least 4096. 6069Currently, it is only greater than 4096 if a dynamic feature has been 6070enabled with ``arch_prctl()``, but this may change in the future. 6071 6072The offsets of the state save areas in struct kvm_xsave follow the contents 6073of CPUID leaf 0xD on the host. 6074 60754.135 KVM_XEN_HVM_EVTCHN_SEND 6076----------------------------- 6077 6078:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_EVTCHN_SEND 6079:Architectures: x86 6080:Type: vm ioctl 6081:Parameters: struct kvm_irq_routing_xen_evtchn 6082:Returns: 0 on success, < 0 on error 6083 6084 6085:: 6086 6087 struct kvm_irq_routing_xen_evtchn { 6088 __u32 port; 6089 __u32 vcpu; 6090 __u32 priority; 6091 }; 6092 6093This ioctl injects an event channel interrupt directly to the guest vCPU. 6094 60954.136 KVM_S390_PV_CPU_COMMAND 6096----------------------------- 6097 6098:Capability: KVM_CAP_S390_PROTECTED_DUMP 6099:Architectures: s390 6100:Type: vcpu ioctl 6101:Parameters: none 6102:Returns: 0 on success, < 0 on error 6103 6104This ioctl closely mirrors `KVM_S390_PV_COMMAND` but handles requests 6105for vcpus. It re-uses the kvm_s390_pv_dmp struct and hence also shares 6106the command ids. 6107 6108**command:** 6109 6110KVM_PV_DUMP 6111 Presents an API that provides calls which facilitate dumping a vcpu 6112 of a protected VM. 6113 6114**subcommand:** 6115 6116KVM_PV_DUMP_CPU 6117 Provides encrypted dump data like register values. 6118 The length of the returned data is provided by uv_info.guest_cpu_stor_len. 6119 61204.137 KVM_S390_ZPCI_OP 6121---------------------- 6122 6123:Capability: KVM_CAP_S390_ZPCI_OP 6124:Architectures: s390 6125:Type: vm ioctl 6126:Parameters: struct kvm_s390_zpci_op (in) 6127:Returns: 0 on success, <0 on error 6128 6129Used to manage hardware-assisted virtualization features for zPCI devices. 6130 6131Parameters are specified via the following structure:: 6132 6133 struct kvm_s390_zpci_op { 6134 /* in */ 6135 __u32 fh; /* target device */ 6136 __u8 op; /* operation to perform */ 6137 __u8 pad[3]; 6138 union { 6139 /* for KVM_S390_ZPCIOP_REG_AEN */ 6140 struct { 6141 __u64 ibv; /* Guest addr of interrupt bit vector */ 6142 __u64 sb; /* Guest addr of summary bit */ 6143 __u32 flags; 6144 __u32 noi; /* Number of interrupts */ 6145 __u8 isc; /* Guest interrupt subclass */ 6146 __u8 sbo; /* Offset of guest summary bit vector */ 6147 __u16 pad; 6148 } reg_aen; 6149 __u64 reserved[8]; 6150 } u; 6151 }; 6152 6153The type of operation is specified in the "op" field. 6154KVM_S390_ZPCIOP_REG_AEN is used to register the VM for adapter event 6155notification interpretation, which will allow firmware delivery of adapter 6156events directly to the vm, with KVM providing a backup delivery mechanism; 6157KVM_S390_ZPCIOP_DEREG_AEN is used to subsequently disable interpretation of 6158adapter event notifications. 6159 6160The target zPCI function must also be specified via the "fh" field. For the 6161KVM_S390_ZPCIOP_REG_AEN operation, additional information to establish firmware 6162delivery must be provided via the "reg_aen" struct. 6163 6164The "pad" and "reserved" fields may be used for future extensions and should be 6165set to 0s by userspace. 6166 61674.138 KVM_ARM_SET_COUNTER_OFFSET 6168-------------------------------- 6169 6170:Capability: KVM_CAP_COUNTER_OFFSET 6171:Architectures: arm64 6172:Type: vm ioctl 6173:Parameters: struct kvm_arm_counter_offset (in) 6174:Returns: 0 on success, < 0 on error 6175 6176This capability indicates that userspace is able to apply a single VM-wide 6177offset to both the virtual and physical counters as viewed by the guest 6178using the KVM_ARM_SET_CNT_OFFSET ioctl and the following data structure: 6179 6180:: 6181 6182 struct kvm_arm_counter_offset { 6183 __u64 counter_offset; 6184 __u64 reserved; 6185 }; 6186 6187The offset describes a number of counter cycles that are subtracted from 6188both virtual and physical counter views (similar to the effects of the 6189CNTVOFF_EL2 and CNTPOFF_EL2 system registers, but only global). The offset 6190always applies to all vcpus (already created or created after this ioctl) 6191for this VM. 6192 6193It is userspace's responsibility to compute the offset based, for example, 6194on previous values of the guest counters. 6195 6196Any value other than 0 for the "reserved" field may result in an error 6197(-EINVAL) being returned. This ioctl can also return -EBUSY if any vcpu 6198ioctl is issued concurrently. 6199 6200Note that using this ioctl results in KVM ignoring subsequent userspace 6201writes to the CNTVCT_EL0 and CNTPCT_EL0 registers using the SET_ONE_REG 6202interface. No error will be returned, but the resulting offset will not be 6203applied. 6204 6205.. _KVM_ARM_GET_REG_WRITABLE_MASKS: 6206 62074.139 KVM_ARM_GET_REG_WRITABLE_MASKS 6208------------------------------------------- 6209 6210:Capability: KVM_CAP_ARM_SUPPORTED_REG_MASK_RANGES 6211:Architectures: arm64 6212:Type: vm ioctl 6213:Parameters: struct reg_mask_range (in/out) 6214:Returns: 0 on success, < 0 on error 6215 6216 6217:: 6218 6219 #define KVM_ARM_FEATURE_ID_RANGE 0 6220 #define KVM_ARM_FEATURE_ID_RANGE_SIZE (3 * 8 * 8) 6221 6222 struct reg_mask_range { 6223 __u64 addr; /* Pointer to mask array */ 6224 __u32 range; /* Requested range */ 6225 __u32 reserved[13]; 6226 }; 6227 6228This ioctl copies the writable masks for a selected range of registers to 6229userspace. 6230 6231The ``addr`` field is a pointer to the destination array where KVM copies 6232the writable masks. 6233 6234The ``range`` field indicates the requested range of registers. 6235``KVM_CHECK_EXTENSION`` for the ``KVM_CAP_ARM_SUPPORTED_REG_MASK_RANGES`` 6236capability returns the supported ranges, expressed as a set of flags. Each 6237flag's bit index represents a possible value for the ``range`` field. 6238All other values are reserved for future use and KVM may return an error. 6239 6240The ``reserved[13]`` array is reserved for future use and should be 0, or 6241KVM may return an error. 6242 6243KVM_ARM_FEATURE_ID_RANGE (0) 6244^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6245 6246The Feature ID range is defined as the AArch64 System register space with 6247op0==3, op1=={0, 1, 3}, CRn==0, CRm=={0-7}, op2=={0-7}. 6248 6249The mask returned array pointed to by ``addr`` is indexed by the macro 6250``ARM64_FEATURE_ID_RANGE_IDX(op0, op1, crn, crm, op2)``, allowing userspace 6251to know what fields can be changed for the system register described by 6252``op0, op1, crn, crm, op2``. KVM rejects ID register values that describe a 6253superset of the features supported by the system. 6254 62554.140 KVM_SET_USER_MEMORY_REGION2 6256--------------------------------- 6257 6258:Capability: KVM_CAP_USER_MEMORY2 6259:Architectures: all 6260:Type: vm ioctl 6261:Parameters: struct kvm_userspace_memory_region2 (in) 6262:Returns: 0 on success, -1 on error 6263 6264KVM_SET_USER_MEMORY_REGION2 is an extension to KVM_SET_USER_MEMORY_REGION that 6265allows mapping guest_memfd memory into a guest. All fields shared with 6266KVM_SET_USER_MEMORY_REGION identically. Userspace can set KVM_MEM_GUEST_MEMFD 6267in flags to have KVM bind the memory region to a given guest_memfd range of 6268[guest_memfd_offset, guest_memfd_offset + memory_size]. The target guest_memfd 6269must point at a file created via KVM_CREATE_GUEST_MEMFD on the current VM, and 6270the target range must not be bound to any other memory region. All standard 6271bounds checks apply (use common sense). 6272 6273:: 6274 6275 struct kvm_userspace_memory_region2 { 6276 __u32 slot; 6277 __u32 flags; 6278 __u64 guest_phys_addr; 6279 __u64 memory_size; /* bytes */ 6280 __u64 userspace_addr; /* start of the userspace allocated memory */ 6281 __u64 guest_memfd_offset; 6282 __u32 guest_memfd; 6283 __u32 pad1; 6284 __u64 pad2[14]; 6285 }; 6286 6287A KVM_MEM_GUEST_MEMFD region _must_ have a valid guest_memfd (private memory) and 6288userspace_addr (shared memory). However, "valid" for userspace_addr simply 6289means that the address itself must be a legal userspace address. The backing 6290mapping for userspace_addr is not required to be valid/populated at the time of 6291KVM_SET_USER_MEMORY_REGION2, e.g. shared memory can be lazily mapped/allocated 6292on-demand. 6293 6294When mapping a gfn into the guest, KVM selects shared vs. private, i.e consumes 6295userspace_addr vs. guest_memfd, based on the gfn's KVM_MEMORY_ATTRIBUTE_PRIVATE 6296state. At VM creation time, all memory is shared, i.e. the PRIVATE attribute 6297is '0' for all gfns. Userspace can control whether memory is shared/private by 6298toggling KVM_MEMORY_ATTRIBUTE_PRIVATE via KVM_SET_MEMORY_ATTRIBUTES as needed. 6299 6300S390: 6301^^^^^ 6302 6303Returns -EINVAL if the VM has the KVM_VM_S390_UCONTROL flag set. 6304Returns -EINVAL if called on a protected VM. 6305 63064.141 KVM_SET_MEMORY_ATTRIBUTES 6307------------------------------- 6308 6309:Capability: KVM_CAP_MEMORY_ATTRIBUTES 6310:Architectures: x86 6311:Type: vm ioctl 6312:Parameters: struct kvm_memory_attributes (in) 6313:Returns: 0 on success, <0 on error 6314 6315KVM_SET_MEMORY_ATTRIBUTES allows userspace to set memory attributes for a range 6316of guest physical memory. 6317 6318:: 6319 6320 struct kvm_memory_attributes { 6321 __u64 address; 6322 __u64 size; 6323 __u64 attributes; 6324 __u64 flags; 6325 }; 6326 6327 #define KVM_MEMORY_ATTRIBUTE_PRIVATE (1ULL << 3) 6328 6329The address and size must be page aligned. The supported attributes can be 6330retrieved via ioctl(KVM_CHECK_EXTENSION) on KVM_CAP_MEMORY_ATTRIBUTES. If 6331executed on a VM, KVM_CAP_MEMORY_ATTRIBUTES precisely returns the attributes 6332supported by that VM. If executed at system scope, KVM_CAP_MEMORY_ATTRIBUTES 6333returns all attributes supported by KVM. The only attribute defined at this 6334time is KVM_MEMORY_ATTRIBUTE_PRIVATE, which marks the associated gfn as being 6335guest private memory. 6336 6337Note, there is no "get" API. Userspace is responsible for explicitly tracking 6338the state of a gfn/page as needed. 6339 6340The "flags" field is reserved for future extensions and must be '0'. 6341 63424.142 KVM_CREATE_GUEST_MEMFD 6343---------------------------- 6344 6345:Capability: KVM_CAP_GUEST_MEMFD 6346:Architectures: none 6347:Type: vm ioctl 6348:Parameters: struct kvm_create_guest_memfd(in) 6349:Returns: A file descriptor on success, <0 on error 6350 6351KVM_CREATE_GUEST_MEMFD creates an anonymous file and returns a file descriptor 6352that refers to it. guest_memfd files are roughly analogous to files created 6353via memfd_create(), e.g. guest_memfd files live in RAM, have volatile storage, 6354and are automatically released when the last reference is dropped. Unlike 6355"regular" memfd_create() files, guest_memfd files are bound to their owning 6356virtual machine (see below), cannot be mapped, read, or written by userspace, 6357and cannot be resized (guest_memfd files do however support PUNCH_HOLE). 6358 6359:: 6360 6361 struct kvm_create_guest_memfd { 6362 __u64 size; 6363 __u64 flags; 6364 __u64 reserved[6]; 6365 }; 6366 6367Conceptually, the inode backing a guest_memfd file represents physical memory, 6368i.e. is coupled to the virtual machine as a thing, not to a "struct kvm". The 6369file itself, which is bound to a "struct kvm", is that instance's view of the 6370underlying memory, e.g. effectively provides the translation of guest addresses 6371to host memory. This allows for use cases where multiple KVM structures are 6372used to manage a single virtual machine, e.g. when performing intrahost 6373migration of a virtual machine. 6374 6375KVM currently only supports mapping guest_memfd via KVM_SET_USER_MEMORY_REGION2, 6376and more specifically via the guest_memfd and guest_memfd_offset fields in 6377"struct kvm_userspace_memory_region2", where guest_memfd_offset is the offset 6378into the guest_memfd instance. For a given guest_memfd file, there can be at 6379most one mapping per page, i.e. binding multiple memory regions to a single 6380guest_memfd range is not allowed (any number of memory regions can be bound to 6381a single guest_memfd file, but the bound ranges must not overlap). 6382 6383See KVM_SET_USER_MEMORY_REGION2 for additional details. 6384 63854.143 KVM_PRE_FAULT_MEMORY 6386--------------------------- 6387 6388:Capability: KVM_CAP_PRE_FAULT_MEMORY 6389:Architectures: none 6390:Type: vcpu ioctl 6391:Parameters: struct kvm_pre_fault_memory (in/out) 6392:Returns: 0 if at least one page is processed, < 0 on error 6393 6394Errors: 6395 6396 ========== =============================================================== 6397 EINVAL The specified `gpa` and `size` were invalid (e.g. not 6398 page aligned, causes an overflow, or size is zero). 6399 ENOENT The specified `gpa` is outside defined memslots. 6400 EINTR An unmasked signal is pending and no page was processed. 6401 EFAULT The parameter address was invalid. 6402 EOPNOTSUPP Mapping memory for a GPA is unsupported by the 6403 hypervisor, and/or for the current vCPU state/mode. 6404 EIO unexpected error conditions (also causes a WARN) 6405 ========== =============================================================== 6406 6407:: 6408 6409 struct kvm_pre_fault_memory { 6410 /* in/out */ 6411 __u64 gpa; 6412 __u64 size; 6413 /* in */ 6414 __u64 flags; 6415 __u64 padding[5]; 6416 }; 6417 6418KVM_PRE_FAULT_MEMORY populates KVM's stage-2 page tables used to map memory 6419for the current vCPU state. KVM maps memory as if the vCPU generated a 6420stage-2 read page fault, e.g. faults in memory as needed, but doesn't break 6421CoW. However, KVM does not mark any newly created stage-2 PTE as Accessed. 6422 6423In the case of confidential VM types where there is an initial set up of 6424private guest memory before the guest is 'finalized'/measured, this ioctl 6425should only be issued after completing all the necessary setup to put the 6426guest into a 'finalized' state so that the above semantics can be reliably 6427ensured. 6428 6429In some cases, multiple vCPUs might share the page tables. In this 6430case, the ioctl can be called in parallel. 6431 6432When the ioctl returns, the input values are updated to point to the 6433remaining range. If `size` > 0 on return, the caller can just issue 6434the ioctl again with the same `struct kvm_map_memory` argument. 6435 6436Shadow page tables cannot support this ioctl because they 6437are indexed by virtual address or nested guest physical address. 6438Calling this ioctl when the guest is using shadow page tables (for 6439example because it is running a nested guest with nested page tables) 6440will fail with `EOPNOTSUPP` even if `KVM_CHECK_EXTENSION` reports 6441the capability to be present. 6442 6443`flags` must currently be zero. 6444 6445 64465. The kvm_run structure 6447======================== 6448 6449Application code obtains a pointer to the kvm_run structure by 6450mmap()ing a vcpu fd. From that point, application code can control 6451execution by changing fields in kvm_run prior to calling the KVM_RUN 6452ioctl, and obtain information about the reason KVM_RUN returned by 6453looking up structure members. 6454 6455:: 6456 6457 struct kvm_run { 6458 /* in */ 6459 __u8 request_interrupt_window; 6460 6461Request that KVM_RUN return when it becomes possible to inject external 6462interrupts into the guest. Useful in conjunction with KVM_INTERRUPT. 6463 6464:: 6465 6466 __u8 immediate_exit; 6467 6468This field is polled once when KVM_RUN starts; if non-zero, KVM_RUN 6469exits immediately, returning -EINTR. In the common scenario where a 6470signal is used to "kick" a VCPU out of KVM_RUN, this field can be used 6471to avoid usage of KVM_SET_SIGNAL_MASK, which has worse scalability. 6472Rather than blocking the signal outside KVM_RUN, userspace can set up 6473a signal handler that sets run->immediate_exit to a non-zero value. 6474 6475This field is ignored if KVM_CAP_IMMEDIATE_EXIT is not available. 6476 6477:: 6478 6479 __u8 padding1[6]; 6480 6481 /* out */ 6482 __u32 exit_reason; 6483 6484When KVM_RUN has returned successfully (return value 0), this informs 6485application code why KVM_RUN has returned. Allowable values for this 6486field are detailed below. 6487 6488:: 6489 6490 __u8 ready_for_interrupt_injection; 6491 6492If request_interrupt_window has been specified, this field indicates 6493an interrupt can be injected now with KVM_INTERRUPT. 6494 6495:: 6496 6497 __u8 if_flag; 6498 6499The value of the current interrupt flag. Only valid if in-kernel 6500local APIC is not used. 6501 6502:: 6503 6504 __u16 flags; 6505 6506More architecture-specific flags detailing state of the VCPU that may 6507affect the device's behavior. Current defined flags:: 6508 6509 /* x86, set if the VCPU is in system management mode */ 6510 #define KVM_RUN_X86_SMM (1 << 0) 6511 /* x86, set if bus lock detected in VM */ 6512 #define KVM_RUN_X86_BUS_LOCK (1 << 1) 6513 /* x86, set if the VCPU is executing a nested (L2) guest */ 6514 #define KVM_RUN_X86_GUEST_MODE (1 << 2) 6515 6516 /* arm64, set for KVM_EXIT_DEBUG */ 6517 #define KVM_DEBUG_ARCH_HSR_HIGH_VALID (1 << 0) 6518 6519:: 6520 6521 /* in (pre_kvm_run), out (post_kvm_run) */ 6522 __u64 cr8; 6523 6524The value of the cr8 register. Only valid if in-kernel local APIC is 6525not used. Both input and output. 6526 6527:: 6528 6529 __u64 apic_base; 6530 6531The value of the APIC BASE msr. Only valid if in-kernel local 6532APIC is not used. Both input and output. 6533 6534:: 6535 6536 union { 6537 /* KVM_EXIT_UNKNOWN */ 6538 struct { 6539 __u64 hardware_exit_reason; 6540 } hw; 6541 6542If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown 6543reasons. Further architecture-specific information is available in 6544hardware_exit_reason. 6545 6546:: 6547 6548 /* KVM_EXIT_FAIL_ENTRY */ 6549 struct { 6550 __u64 hardware_entry_failure_reason; 6551 __u32 cpu; /* if KVM_LAST_CPU */ 6552 } fail_entry; 6553 6554If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due 6555to unknown reasons. Further architecture-specific information is 6556available in hardware_entry_failure_reason. 6557 6558:: 6559 6560 /* KVM_EXIT_EXCEPTION */ 6561 struct { 6562 __u32 exception; 6563 __u32 error_code; 6564 } ex; 6565 6566Unused. 6567 6568:: 6569 6570 /* KVM_EXIT_IO */ 6571 struct { 6572 #define KVM_EXIT_IO_IN 0 6573 #define KVM_EXIT_IO_OUT 1 6574 __u8 direction; 6575 __u8 size; /* bytes */ 6576 __u16 port; 6577 __u32 count; 6578 __u64 data_offset; /* relative to kvm_run start */ 6579 } io; 6580 6581If exit_reason is KVM_EXIT_IO, then the vcpu has 6582executed a port I/O instruction which could not be satisfied by kvm. 6583data_offset describes where the data is located (KVM_EXIT_IO_OUT) or 6584where kvm expects application code to place the data for the next 6585KVM_RUN invocation (KVM_EXIT_IO_IN). Data format is a packed array. 6586 6587:: 6588 6589 /* KVM_EXIT_DEBUG */ 6590 struct { 6591 struct kvm_debug_exit_arch arch; 6592 } debug; 6593 6594If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event 6595for which architecture specific information is returned. 6596 6597:: 6598 6599 /* KVM_EXIT_MMIO */ 6600 struct { 6601 __u64 phys_addr; 6602 __u8 data[8]; 6603 __u32 len; 6604 __u8 is_write; 6605 } mmio; 6606 6607If exit_reason is KVM_EXIT_MMIO, then the vcpu has 6608executed a memory-mapped I/O instruction which could not be satisfied 6609by kvm. The 'data' member contains the written data if 'is_write' is 6610true, and should be filled by application code otherwise. 6611 6612The 'data' member contains, in its first 'len' bytes, the value as it would 6613appear if the VCPU performed a load or store of the appropriate width directly 6614to the byte array. 6615 6616.. note:: 6617 6618 For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR, KVM_EXIT_XEN, 6619 KVM_EXIT_EPR, KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR the corresponding 6620 operations are complete (and guest state is consistent) only after userspace 6621 has re-entered the kernel with KVM_RUN. The kernel side will first finish 6622 incomplete operations and then check for pending signals. 6623 6624 The pending state of the operation is not preserved in state which is 6625 visible to userspace, thus userspace should ensure that the operation is 6626 completed before performing a live migration. Userspace can re-enter the 6627 guest with an unmasked signal pending or with the immediate_exit field set 6628 to complete pending operations without allowing any further instructions 6629 to be executed. 6630 6631:: 6632 6633 /* KVM_EXIT_HYPERCALL */ 6634 struct { 6635 __u64 nr; 6636 __u64 args[6]; 6637 __u64 ret; 6638 __u64 flags; 6639 } hypercall; 6640 6641 6642It is strongly recommended that userspace use ``KVM_EXIT_IO`` (x86) or 6643``KVM_EXIT_MMIO`` (all except s390) to implement functionality that 6644requires a guest to interact with host userspace. 6645 6646.. note:: KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO. 6647 6648For arm64: 6649---------- 6650 6651SMCCC exits can be enabled depending on the configuration of the SMCCC 6652filter. See the Documentation/virt/kvm/devices/vm.rst 6653``KVM_ARM_SMCCC_FILTER`` for more details. 6654 6655``nr`` contains the function ID of the guest's SMCCC call. Userspace is 6656expected to use the ``KVM_GET_ONE_REG`` ioctl to retrieve the call 6657parameters from the vCPU's GPRs. 6658 6659Definition of ``flags``: 6660 - ``KVM_HYPERCALL_EXIT_SMC``: Indicates that the guest used the SMC 6661 conduit to initiate the SMCCC call. If this bit is 0 then the guest 6662 used the HVC conduit for the SMCCC call. 6663 6664 - ``KVM_HYPERCALL_EXIT_16BIT``: Indicates that the guest used a 16bit 6665 instruction to initiate the SMCCC call. If this bit is 0 then the 6666 guest used a 32bit instruction. An AArch64 guest always has this 6667 bit set to 0. 6668 6669At the point of exit, PC points to the instruction immediately following 6670the trapping instruction. 6671 6672:: 6673 6674 /* KVM_EXIT_TPR_ACCESS */ 6675 struct { 6676 __u64 rip; 6677 __u32 is_write; 6678 __u32 pad; 6679 } tpr_access; 6680 6681To be documented (KVM_TPR_ACCESS_REPORTING). 6682 6683:: 6684 6685 /* KVM_EXIT_S390_SIEIC */ 6686 struct { 6687 __u8 icptcode; 6688 __u64 mask; /* psw upper half */ 6689 __u64 addr; /* psw lower half */ 6690 __u16 ipa; 6691 __u32 ipb; 6692 } s390_sieic; 6693 6694s390 specific. 6695 6696:: 6697 6698 /* KVM_EXIT_S390_RESET */ 6699 #define KVM_S390_RESET_POR 1 6700 #define KVM_S390_RESET_CLEAR 2 6701 #define KVM_S390_RESET_SUBSYSTEM 4 6702 #define KVM_S390_RESET_CPU_INIT 8 6703 #define KVM_S390_RESET_IPL 16 6704 __u64 s390_reset_flags; 6705 6706s390 specific. 6707 6708:: 6709 6710 /* KVM_EXIT_S390_UCONTROL */ 6711 struct { 6712 __u64 trans_exc_code; 6713 __u32 pgm_code; 6714 } s390_ucontrol; 6715 6716s390 specific. A page fault has occurred for a user controlled virtual 6717machine (KVM_VM_S390_UNCONTROL) on its host page table that cannot be 6718resolved by the kernel. 6719The program code and the translation exception code that were placed 6720in the cpu's lowcore are presented here as defined by the z Architecture 6721Principles of Operation Book in the Chapter for Dynamic Address Translation 6722(DAT) 6723 6724:: 6725 6726 /* KVM_EXIT_DCR */ 6727 struct { 6728 __u32 dcrn; 6729 __u32 data; 6730 __u8 is_write; 6731 } dcr; 6732 6733Deprecated - was used for 440 KVM. 6734 6735:: 6736 6737 /* KVM_EXIT_OSI */ 6738 struct { 6739 __u64 gprs[32]; 6740 } osi; 6741 6742MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch 6743hypercalls and exit with this exit struct that contains all the guest gprs. 6744 6745If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall. 6746Userspace can now handle the hypercall and when it's done modify the gprs as 6747necessary. Upon guest entry all guest GPRs will then be replaced by the values 6748in this struct. 6749 6750:: 6751 6752 /* KVM_EXIT_PAPR_HCALL */ 6753 struct { 6754 __u64 nr; 6755 __u64 ret; 6756 __u64 args[9]; 6757 } papr_hcall; 6758 6759This is used on 64-bit PowerPC when emulating a pSeries partition, 6760e.g. with the 'pseries' machine type in qemu. It occurs when the 6761guest does a hypercall using the 'sc 1' instruction. The 'nr' field 6762contains the hypercall number (from the guest R3), and 'args' contains 6763the arguments (from the guest R4 - R12). Userspace should put the 6764return code in 'ret' and any extra returned values in args[]. 6765The possible hypercalls are defined in the Power Architecture Platform 6766Requirements (PAPR) document available from www.power.org (free 6767developer registration required to access it). 6768 6769:: 6770 6771 /* KVM_EXIT_S390_TSCH */ 6772 struct { 6773 __u16 subchannel_id; 6774 __u16 subchannel_nr; 6775 __u32 io_int_parm; 6776 __u32 io_int_word; 6777 __u32 ipb; 6778 __u8 dequeued; 6779 } s390_tsch; 6780 6781s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled 6782and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O 6783interrupt for the target subchannel has been dequeued and subchannel_id, 6784subchannel_nr, io_int_parm and io_int_word contain the parameters for that 6785interrupt. ipb is needed for instruction parameter decoding. 6786 6787:: 6788 6789 /* KVM_EXIT_EPR */ 6790 struct { 6791 __u32 epr; 6792 } epr; 6793 6794On FSL BookE PowerPC chips, the interrupt controller has a fast patch 6795interrupt acknowledge path to the core. When the core successfully 6796delivers an interrupt, it automatically populates the EPR register with 6797the interrupt vector number and acknowledges the interrupt inside 6798the interrupt controller. 6799 6800In case the interrupt controller lives in user space, we need to do 6801the interrupt acknowledge cycle through it to fetch the next to be 6802delivered interrupt vector using this exit. 6803 6804It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an 6805external interrupt has just been delivered into the guest. User space 6806should put the acknowledged interrupt vector into the 'epr' field. 6807 6808:: 6809 6810 /* KVM_EXIT_SYSTEM_EVENT */ 6811 struct { 6812 #define KVM_SYSTEM_EVENT_SHUTDOWN 1 6813 #define KVM_SYSTEM_EVENT_RESET 2 6814 #define KVM_SYSTEM_EVENT_CRASH 3 6815 #define KVM_SYSTEM_EVENT_WAKEUP 4 6816 #define KVM_SYSTEM_EVENT_SUSPEND 5 6817 #define KVM_SYSTEM_EVENT_SEV_TERM 6 6818 __u32 type; 6819 __u32 ndata; 6820 __u64 data[16]; 6821 } system_event; 6822 6823If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered 6824a system-level event using some architecture specific mechanism (hypercall 6825or some special instruction). In case of ARM64, this is triggered using 6826HVC instruction based PSCI call from the vcpu. 6827 6828The 'type' field describes the system-level event type. 6829Valid values for 'type' are: 6830 6831 - KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the 6832 VM. Userspace is not obliged to honour this, and if it does honour 6833 this does not need to destroy the VM synchronously (ie it may call 6834 KVM_RUN again before shutdown finally occurs). 6835 - KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM. 6836 As with SHUTDOWN, userspace can choose to ignore the request, or 6837 to schedule the reset to occur in the future and may call KVM_RUN again. 6838 - KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest 6839 has requested a crash condition maintenance. Userspace can choose 6840 to ignore the request, or to gather VM memory core dump and/or 6841 reset/shutdown of the VM. 6842 - KVM_SYSTEM_EVENT_SEV_TERM -- an AMD SEV guest requested termination. 6843 The guest physical address of the guest's GHCB is stored in `data[0]`. 6844 - KVM_SYSTEM_EVENT_WAKEUP -- the exiting vCPU is in a suspended state and 6845 KVM has recognized a wakeup event. Userspace may honor this event by 6846 marking the exiting vCPU as runnable, or deny it and call KVM_RUN again. 6847 - KVM_SYSTEM_EVENT_SUSPEND -- the guest has requested a suspension of 6848 the VM. 6849 6850If KVM_CAP_SYSTEM_EVENT_DATA is present, the 'data' field can contain 6851architecture specific information for the system-level event. Only 6852the first `ndata` items (possibly zero) of the data array are valid. 6853 6854 - for arm64, data[0] is set to KVM_SYSTEM_EVENT_RESET_FLAG_PSCI_RESET2 if 6855 the guest issued a SYSTEM_RESET2 call according to v1.1 of the PSCI 6856 specification. 6857 6858 - for RISC-V, data[0] is set to the value of the second argument of the 6859 ``sbi_system_reset`` call. 6860 6861Previous versions of Linux defined a `flags` member in this struct. The 6862field is now aliased to `data[0]`. Userspace can assume that it is only 6863written if ndata is greater than 0. 6864 6865For arm/arm64: 6866-------------- 6867 6868KVM_SYSTEM_EVENT_SUSPEND exits are enabled with the 6869KVM_CAP_ARM_SYSTEM_SUSPEND VM capability. If a guest invokes the PSCI 6870SYSTEM_SUSPEND function, KVM will exit to userspace with this event 6871type. 6872 6873It is the sole responsibility of userspace to implement the PSCI 6874SYSTEM_SUSPEND call according to ARM DEN0022D.b 5.19 "SYSTEM_SUSPEND". 6875KVM does not change the vCPU's state before exiting to userspace, so 6876the call parameters are left in-place in the vCPU registers. 6877 6878Userspace is _required_ to take action for such an exit. It must 6879either: 6880 6881 - Honor the guest request to suspend the VM. Userspace can request 6882 in-kernel emulation of suspension by setting the calling vCPU's 6883 state to KVM_MP_STATE_SUSPENDED. Userspace must configure the vCPU's 6884 state according to the parameters passed to the PSCI function when 6885 the calling vCPU is resumed. See ARM DEN0022D.b 5.19.1 "Intended use" 6886 for details on the function parameters. 6887 6888 - Deny the guest request to suspend the VM. See ARM DEN0022D.b 5.19.2 6889 "Caller responsibilities" for possible return values. 6890 6891:: 6892 6893 /* KVM_EXIT_IOAPIC_EOI */ 6894 struct { 6895 __u8 vector; 6896 } eoi; 6897 6898Indicates that the VCPU's in-kernel local APIC received an EOI for a 6899level-triggered IOAPIC interrupt. This exit only triggers when the 6900IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled); 6901the userspace IOAPIC should process the EOI and retrigger the interrupt if 6902it is still asserted. Vector is the LAPIC interrupt vector for which the 6903EOI was received. 6904 6905:: 6906 6907 struct kvm_hyperv_exit { 6908 #define KVM_EXIT_HYPERV_SYNIC 1 6909 #define KVM_EXIT_HYPERV_HCALL 2 6910 #define KVM_EXIT_HYPERV_SYNDBG 3 6911 __u32 type; 6912 __u32 pad1; 6913 union { 6914 struct { 6915 __u32 msr; 6916 __u32 pad2; 6917 __u64 control; 6918 __u64 evt_page; 6919 __u64 msg_page; 6920 } synic; 6921 struct { 6922 __u64 input; 6923 __u64 result; 6924 __u64 params[2]; 6925 } hcall; 6926 struct { 6927 __u32 msr; 6928 __u32 pad2; 6929 __u64 control; 6930 __u64 status; 6931 __u64 send_page; 6932 __u64 recv_page; 6933 __u64 pending_page; 6934 } syndbg; 6935 } u; 6936 }; 6937 /* KVM_EXIT_HYPERV */ 6938 struct kvm_hyperv_exit hyperv; 6939 6940Indicates that the VCPU exits into userspace to process some tasks 6941related to Hyper-V emulation. 6942 6943Valid values for 'type' are: 6944 6945 - KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about 6946 6947Hyper-V SynIC state change. Notification is used to remap SynIC 6948event/message pages and to enable/disable SynIC messages/events processing 6949in userspace. 6950 6951 - KVM_EXIT_HYPERV_SYNDBG -- synchronously notify user-space about 6952 6953Hyper-V Synthetic debugger state change. Notification is used to either update 6954the pending_page location or to send a control command (send the buffer located 6955in send_page or recv a buffer to recv_page). 6956 6957:: 6958 6959 /* KVM_EXIT_ARM_NISV */ 6960 struct { 6961 __u64 esr_iss; 6962 __u64 fault_ipa; 6963 } arm_nisv; 6964 6965Used on arm64 systems. If a guest accesses memory not in a memslot, 6966KVM will typically return to userspace and ask it to do MMIO emulation on its 6967behalf. However, for certain classes of instructions, no instruction decode 6968(direction, length of memory access) is provided, and fetching and decoding 6969the instruction from the VM is overly complicated to live in the kernel. 6970 6971Historically, when this situation occurred, KVM would print a warning and kill 6972the VM. KVM assumed that if the guest accessed non-memslot memory, it was 6973trying to do I/O, which just couldn't be emulated, and the warning message was 6974phrased accordingly. However, what happened more often was that a guest bug 6975caused access outside the guest memory areas which should lead to a more 6976meaningful warning message and an external abort in the guest, if the access 6977did not fall within an I/O window. 6978 6979Userspace implementations can query for KVM_CAP_ARM_NISV_TO_USER, and enable 6980this capability at VM creation. Once this is done, these types of errors will 6981instead return to userspace with KVM_EXIT_ARM_NISV, with the valid bits from 6982the ESR_EL2 in the esr_iss field, and the faulting IPA in the fault_ipa field. 6983Userspace can either fix up the access if it's actually an I/O access by 6984decoding the instruction from guest memory (if it's very brave) and continue 6985executing the guest, or it can decide to suspend, dump, or restart the guest. 6986 6987Note that KVM does not skip the faulting instruction as it does for 6988KVM_EXIT_MMIO, but userspace has to emulate any change to the processing state 6989if it decides to decode and emulate the instruction. 6990 6991This feature isn't available to protected VMs, as userspace does not 6992have access to the state that is required to perform the emulation. 6993Instead, a data abort exception is directly injected in the guest. 6994Note that although KVM_CAP_ARM_NISV_TO_USER will be reported if 6995queried outside of a protected VM context, the feature will not be 6996exposed if queried on a protected VM file descriptor. 6997 6998:: 6999 7000 /* KVM_EXIT_X86_RDMSR / KVM_EXIT_X86_WRMSR */ 7001 struct { 7002 __u8 error; /* user -> kernel */ 7003 __u8 pad[7]; 7004 __u32 reason; /* kernel -> user */ 7005 __u32 index; /* kernel -> user */ 7006 __u64 data; /* kernel <-> user */ 7007 } msr; 7008 7009Used on x86 systems. When the VM capability KVM_CAP_X86_USER_SPACE_MSR is 7010enabled, MSR accesses to registers that would invoke a #GP by KVM kernel code 7011may instead trigger a KVM_EXIT_X86_RDMSR exit for reads and KVM_EXIT_X86_WRMSR 7012exit for writes. 7013 7014The "reason" field specifies why the MSR interception occurred. Userspace will 7015only receive MSR exits when a particular reason was requested during through 7016ENABLE_CAP. Currently valid exit reasons are: 7017 7018============================ ======================================== 7019 KVM_MSR_EXIT_REASON_UNKNOWN access to MSR that is unknown to KVM 7020 KVM_MSR_EXIT_REASON_INVAL access to invalid MSRs or reserved bits 7021 KVM_MSR_EXIT_REASON_FILTER access blocked by KVM_X86_SET_MSR_FILTER 7022============================ ======================================== 7023 7024For KVM_EXIT_X86_RDMSR, the "index" field tells userspace which MSR the guest 7025wants to read. To respond to this request with a successful read, userspace 7026writes the respective data into the "data" field and must continue guest 7027execution to ensure the read data is transferred into guest register state. 7028 7029If the RDMSR request was unsuccessful, userspace indicates that with a "1" in 7030the "error" field. This will inject a #GP into the guest when the VCPU is 7031executed again. 7032 7033For KVM_EXIT_X86_WRMSR, the "index" field tells userspace which MSR the guest 7034wants to write. Once finished processing the event, userspace must continue 7035vCPU execution. If the MSR write was unsuccessful, userspace also sets the 7036"error" field to "1". 7037 7038See KVM_X86_SET_MSR_FILTER for details on the interaction with MSR filtering. 7039 7040:: 7041 7042 7043 struct kvm_xen_exit { 7044 #define KVM_EXIT_XEN_HCALL 1 7045 __u32 type; 7046 union { 7047 struct { 7048 __u32 longmode; 7049 __u32 cpl; 7050 __u64 input; 7051 __u64 result; 7052 __u64 params[6]; 7053 } hcall; 7054 } u; 7055 }; 7056 /* KVM_EXIT_XEN */ 7057 struct kvm_hyperv_exit xen; 7058 7059Indicates that the VCPU exits into userspace to process some tasks 7060related to Xen emulation. 7061 7062Valid values for 'type' are: 7063 7064 - KVM_EXIT_XEN_HCALL -- synchronously notify user-space about Xen hypercall. 7065 Userspace is expected to place the hypercall result into the appropriate 7066 field before invoking KVM_RUN again. 7067 7068:: 7069 7070 /* KVM_EXIT_RISCV_SBI */ 7071 struct { 7072 unsigned long extension_id; 7073 unsigned long function_id; 7074 unsigned long args[6]; 7075 unsigned long ret[2]; 7076 } riscv_sbi; 7077 7078If exit reason is KVM_EXIT_RISCV_SBI then it indicates that the VCPU has 7079done a SBI call which is not handled by KVM RISC-V kernel module. The details 7080of the SBI call are available in 'riscv_sbi' member of kvm_run structure. The 7081'extension_id' field of 'riscv_sbi' represents SBI extension ID whereas the 7082'function_id' field represents function ID of given SBI extension. The 'args' 7083array field of 'riscv_sbi' represents parameters for the SBI call and 'ret' 7084array field represents return values. The userspace should update the return 7085values of SBI call before resuming the VCPU. For more details on RISC-V SBI 7086spec refer, https://github.com/riscv/riscv-sbi-doc. 7087 7088:: 7089 7090 /* KVM_EXIT_MEMORY_FAULT */ 7091 struct { 7092 #define KVM_MEMORY_EXIT_FLAG_PRIVATE (1ULL << 3) 7093 __u64 flags; 7094 __u64 gpa; 7095 __u64 size; 7096 } memory_fault; 7097 7098KVM_EXIT_MEMORY_FAULT indicates the vCPU has encountered a memory fault that 7099could not be resolved by KVM. The 'gpa' and 'size' (in bytes) describe the 7100guest physical address range [gpa, gpa + size) of the fault. The 'flags' field 7101describes properties of the faulting access that are likely pertinent: 7102 7103 - KVM_MEMORY_EXIT_FLAG_PRIVATE - When set, indicates the memory fault occurred 7104 on a private memory access. When clear, indicates the fault occurred on a 7105 shared access. 7106 7107Note! KVM_EXIT_MEMORY_FAULT is unique among all KVM exit reasons in that it 7108accompanies a return code of '-1', not '0'! errno will always be set to EFAULT 7109or EHWPOISON when KVM exits with KVM_EXIT_MEMORY_FAULT, userspace should assume 7110kvm_run.exit_reason is stale/undefined for all other error numbers. 7111 7112:: 7113 7114 /* KVM_EXIT_NOTIFY */ 7115 struct { 7116 #define KVM_NOTIFY_CONTEXT_INVALID (1 << 0) 7117 __u32 flags; 7118 } notify; 7119 7120Used on x86 systems. When the VM capability KVM_CAP_X86_NOTIFY_VMEXIT is 7121enabled, a VM exit generated if no event window occurs in VM non-root mode 7122for a specified amount of time. Once KVM_X86_NOTIFY_VMEXIT_USER is set when 7123enabling the cap, it would exit to userspace with the exit reason 7124KVM_EXIT_NOTIFY for further handling. The "flags" field contains more 7125detailed info. 7126 7127The valid value for 'flags' is: 7128 7129 - KVM_NOTIFY_CONTEXT_INVALID -- the VM context is corrupted and not valid 7130 in VMCS. It would run into unknown result if resume the target VM. 7131 7132:: 7133 7134 /* Fix the size of the union. */ 7135 char padding[256]; 7136 }; 7137 7138 /* 7139 * shared registers between kvm and userspace. 7140 * kvm_valid_regs specifies the register classes set by the host 7141 * kvm_dirty_regs specified the register classes dirtied by userspace 7142 * struct kvm_sync_regs is architecture specific, as well as the 7143 * bits for kvm_valid_regs and kvm_dirty_regs 7144 */ 7145 __u64 kvm_valid_regs; 7146 __u64 kvm_dirty_regs; 7147 union { 7148 struct kvm_sync_regs regs; 7149 char padding[SYNC_REGS_SIZE_BYTES]; 7150 } s; 7151 7152If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access 7153certain guest registers without having to call SET/GET_*REGS. Thus we can 7154avoid some system call overhead if userspace has to handle the exit. 7155Userspace can query the validity of the structure by checking 7156kvm_valid_regs for specific bits. These bits are architecture specific 7157and usually define the validity of a groups of registers. (e.g. one bit 7158for general purpose registers) 7159 7160Please note that the kernel is allowed to use the kvm_run structure as the 7161primary storage for certain register types. Therefore, the kernel may use the 7162values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set. 7163 7164 71656. Capabilities that can be enabled on vCPUs 7166============================================ 7167 7168There are certain capabilities that change the behavior of the virtual CPU or 7169the virtual machine when enabled. To enable them, please see section 4.37. 7170Below you can find a list of capabilities and what their effect on the vCPU or 7171the virtual machine is when enabling them. 7172 7173The following information is provided along with the description: 7174 7175 Architectures: 7176 which instruction set architectures provide this ioctl. 7177 x86 includes both i386 and x86_64. 7178 7179 Target: 7180 whether this is a per-vcpu or per-vm capability. 7181 7182 Parameters: 7183 what parameters are accepted by the capability. 7184 7185 Returns: 7186 the return value. General error numbers (EBADF, ENOMEM, EINVAL) 7187 are not detailed, but errors with specific meanings are. 7188 7189 71906.1 KVM_CAP_PPC_OSI 7191------------------- 7192 7193:Architectures: ppc 7194:Target: vcpu 7195:Parameters: none 7196:Returns: 0 on success; -1 on error 7197 7198This capability enables interception of OSI hypercalls that otherwise would 7199be treated as normal system calls to be injected into the guest. OSI hypercalls 7200were invented by Mac-on-Linux to have a standardized communication mechanism 7201between the guest and the host. 7202 7203When this capability is enabled, KVM_EXIT_OSI can occur. 7204 7205 72066.2 KVM_CAP_PPC_PAPR 7207-------------------- 7208 7209:Architectures: ppc 7210:Target: vcpu 7211:Parameters: none 7212:Returns: 0 on success; -1 on error 7213 7214This capability enables interception of PAPR hypercalls. PAPR hypercalls are 7215done using the hypercall instruction "sc 1". 7216 7217It also sets the guest privilege level to "supervisor" mode. Usually the guest 7218runs in "hypervisor" privilege mode with a few missing features. 7219 7220In addition to the above, it changes the semantics of SDR1. In this mode, the 7221HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the 7222HTAB invisible to the guest. 7223 7224When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur. 7225 7226 72276.3 KVM_CAP_SW_TLB 7228------------------ 7229 7230:Architectures: ppc 7231:Target: vcpu 7232:Parameters: args[0] is the address of a struct kvm_config_tlb 7233:Returns: 0 on success; -1 on error 7234 7235:: 7236 7237 struct kvm_config_tlb { 7238 __u64 params; 7239 __u64 array; 7240 __u32 mmu_type; 7241 __u32 array_len; 7242 }; 7243 7244Configures the virtual CPU's TLB array, establishing a shared memory area 7245between userspace and KVM. The "params" and "array" fields are userspace 7246addresses of mmu-type-specific data structures. The "array_len" field is an 7247safety mechanism, and should be set to the size in bytes of the memory that 7248userspace has reserved for the array. It must be at least the size dictated 7249by "mmu_type" and "params". 7250 7251While KVM_RUN is active, the shared region is under control of KVM. Its 7252contents are undefined, and any modification by userspace results in 7253boundedly undefined behavior. 7254 7255On return from KVM_RUN, the shared region will reflect the current state of 7256the guest's TLB. If userspace makes any changes, it must call KVM_DIRTY_TLB 7257to tell KVM which entries have been changed, prior to calling KVM_RUN again 7258on this vcpu. 7259 7260For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV: 7261 7262 - The "params" field is of type "struct kvm_book3e_206_tlb_params". 7263 - The "array" field points to an array of type "struct 7264 kvm_book3e_206_tlb_entry". 7265 - The array consists of all entries in the first TLB, followed by all 7266 entries in the second TLB. 7267 - Within a TLB, entries are ordered first by increasing set number. Within a 7268 set, entries are ordered by way (increasing ESEL). 7269 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1) 7270 where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value. 7271 - The tsize field of mas1 shall be set to 4K on TLB0, even though the 7272 hardware ignores this value for TLB0. 7273 72746.4 KVM_CAP_S390_CSS_SUPPORT 7275---------------------------- 7276 7277:Architectures: s390 7278:Target: vcpu 7279:Parameters: none 7280:Returns: 0 on success; -1 on error 7281 7282This capability enables support for handling of channel I/O instructions. 7283 7284TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are 7285handled in-kernel, while the other I/O instructions are passed to userspace. 7286 7287When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST 7288SUBCHANNEL intercepts. 7289 7290Note that even though this capability is enabled per-vcpu, the complete 7291virtual machine is affected. 7292 72936.5 KVM_CAP_PPC_EPR 7294------------------- 7295 7296:Architectures: ppc 7297:Target: vcpu 7298:Parameters: args[0] defines whether the proxy facility is active 7299:Returns: 0 on success; -1 on error 7300 7301This capability enables or disables the delivery of interrupts through the 7302external proxy facility. 7303 7304When enabled (args[0] != 0), every time the guest gets an external interrupt 7305delivered, it automatically exits into user space with a KVM_EXIT_EPR exit 7306to receive the topmost interrupt vector. 7307 7308When disabled (args[0] == 0), behavior is as if this facility is unsupported. 7309 7310When this capability is enabled, KVM_EXIT_EPR can occur. 7311 73126.6 KVM_CAP_IRQ_MPIC 7313-------------------- 7314 7315:Architectures: ppc 7316:Parameters: args[0] is the MPIC device fd; 7317 args[1] is the MPIC CPU number for this vcpu 7318 7319This capability connects the vcpu to an in-kernel MPIC device. 7320 73216.7 KVM_CAP_IRQ_XICS 7322-------------------- 7323 7324:Architectures: ppc 7325:Target: vcpu 7326:Parameters: args[0] is the XICS device fd; 7327 args[1] is the XICS CPU number (server ID) for this vcpu 7328 7329This capability connects the vcpu to an in-kernel XICS device. 7330 73316.8 KVM_CAP_S390_IRQCHIP 7332------------------------ 7333 7334:Architectures: s390 7335:Target: vm 7336:Parameters: none 7337 7338This capability enables the in-kernel irqchip for s390. Please refer to 7339"4.24 KVM_CREATE_IRQCHIP" for details. 7340 73416.9 KVM_CAP_MIPS_FPU 7342-------------------- 7343 7344:Architectures: mips 7345:Target: vcpu 7346:Parameters: args[0] is reserved for future use (should be 0). 7347 7348This capability allows the use of the host Floating Point Unit by the guest. It 7349allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is 7350done the ``KVM_REG_MIPS_FPR_*`` and ``KVM_REG_MIPS_FCR_*`` registers can be 7351accessed (depending on the current guest FPU register mode), and the Status.FR, 7352Config5.FRE bits are accessible via the KVM API and also from the guest, 7353depending on them being supported by the FPU. 7354 73556.10 KVM_CAP_MIPS_MSA 7356--------------------- 7357 7358:Architectures: mips 7359:Target: vcpu 7360:Parameters: args[0] is reserved for future use (should be 0). 7361 7362This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest. 7363It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest. 7364Once this is done the ``KVM_REG_MIPS_VEC_*`` and ``KVM_REG_MIPS_MSA_*`` 7365registers can be accessed, and the Config5.MSAEn bit is accessible via the 7366KVM API and also from the guest. 7367 73686.74 KVM_CAP_SYNC_REGS 7369---------------------- 7370 7371:Architectures: s390, x86 7372:Target: s390: always enabled, x86: vcpu 7373:Parameters: none 7374:Returns: x86: KVM_CHECK_EXTENSION returns a bit-array indicating which register 7375 sets are supported 7376 (bitfields defined in arch/x86/include/uapi/asm/kvm.h). 7377 7378As described above in the kvm_sync_regs struct info in section 5 (kvm_run): 7379KVM_CAP_SYNC_REGS "allow[s] userspace to access certain guest registers 7380without having to call SET/GET_*REGS". This reduces overhead by eliminating 7381repeated ioctl calls for setting and/or getting register values. This is 7382particularly important when userspace is making synchronous guest state 7383modifications, e.g. when emulating and/or intercepting instructions in 7384userspace. 7385 7386For s390 specifics, please refer to the source code. 7387 7388For x86: 7389 7390- the register sets to be copied out to kvm_run are selectable 7391 by userspace (rather that all sets being copied out for every exit). 7392- vcpu_events are available in addition to regs and sregs. 7393 7394For x86, the 'kvm_valid_regs' field of struct kvm_run is overloaded to 7395function as an input bit-array field set by userspace to indicate the 7396specific register sets to be copied out on the next exit. 7397 7398To indicate when userspace has modified values that should be copied into 7399the vCPU, the all architecture bitarray field, 'kvm_dirty_regs' must be set. 7400This is done using the same bitflags as for the 'kvm_valid_regs' field. 7401If the dirty bit is not set, then the register set values will not be copied 7402into the vCPU even if they've been modified. 7403 7404Unused bitfields in the bitarrays must be set to zero. 7405 7406:: 7407 7408 struct kvm_sync_regs { 7409 struct kvm_regs regs; 7410 struct kvm_sregs sregs; 7411 struct kvm_vcpu_events events; 7412 }; 7413 74146.75 KVM_CAP_PPC_IRQ_XIVE 7415------------------------- 7416 7417:Architectures: ppc 7418:Target: vcpu 7419:Parameters: args[0] is the XIVE device fd; 7420 args[1] is the XIVE CPU number (server ID) for this vcpu 7421 7422This capability connects the vcpu to an in-kernel XIVE device. 7423 74247. Capabilities that can be enabled on VMs 7425========================================== 7426 7427There are certain capabilities that change the behavior of the virtual 7428machine when enabled. To enable them, please see section 4.37. Below 7429you can find a list of capabilities and what their effect on the VM 7430is when enabling them. 7431 7432The following information is provided along with the description: 7433 7434 Architectures: 7435 which instruction set architectures provide this ioctl. 7436 x86 includes both i386 and x86_64. 7437 7438 Parameters: 7439 what parameters are accepted by the capability. 7440 7441 Returns: 7442 the return value. General error numbers (EBADF, ENOMEM, EINVAL) 7443 are not detailed, but errors with specific meanings are. 7444 7445 74467.1 KVM_CAP_PPC_ENABLE_HCALL 7447---------------------------- 7448 7449:Architectures: ppc 7450:Parameters: args[0] is the sPAPR hcall number; 7451 args[1] is 0 to disable, 1 to enable in-kernel handling 7452 7453This capability controls whether individual sPAPR hypercalls (hcalls) 7454get handled by the kernel or not. Enabling or disabling in-kernel 7455handling of an hcall is effective across the VM. On creation, an 7456initial set of hcalls are enabled for in-kernel handling, which 7457consists of those hcalls for which in-kernel handlers were implemented 7458before this capability was implemented. If disabled, the kernel will 7459not to attempt to handle the hcall, but will always exit to userspace 7460to handle it. Note that it may not make sense to enable some and 7461disable others of a group of related hcalls, but KVM does not prevent 7462userspace from doing that. 7463 7464If the hcall number specified is not one that has an in-kernel 7465implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL 7466error. 7467 74687.2 KVM_CAP_S390_USER_SIGP 7469-------------------------- 7470 7471:Architectures: s390 7472:Parameters: none 7473 7474This capability controls which SIGP orders will be handled completely in user 7475space. With this capability enabled, all fast orders will be handled completely 7476in the kernel: 7477 7478- SENSE 7479- SENSE RUNNING 7480- EXTERNAL CALL 7481- EMERGENCY SIGNAL 7482- CONDITIONAL EMERGENCY SIGNAL 7483 7484All other orders will be handled completely in user space. 7485 7486Only privileged operation exceptions will be checked for in the kernel (or even 7487in the hardware prior to interception). If this capability is not enabled, the 7488old way of handling SIGP orders is used (partially in kernel and user space). 7489 74907.3 KVM_CAP_S390_VECTOR_REGISTERS 7491--------------------------------- 7492 7493:Architectures: s390 7494:Parameters: none 7495:Returns: 0 on success, negative value on error 7496 7497Allows use of the vector registers introduced with z13 processor, and 7498provides for the synchronization between host and user space. Will 7499return -EINVAL if the machine does not support vectors. 7500 75017.4 KVM_CAP_S390_USER_STSI 7502-------------------------- 7503 7504:Architectures: s390 7505:Parameters: none 7506 7507This capability allows post-handlers for the STSI instruction. After 7508initial handling in the kernel, KVM exits to user space with 7509KVM_EXIT_S390_STSI to allow user space to insert further data. 7510 7511Before exiting to userspace, kvm handlers should fill in s390_stsi field of 7512vcpu->run:: 7513 7514 struct { 7515 __u64 addr; 7516 __u8 ar; 7517 __u8 reserved; 7518 __u8 fc; 7519 __u8 sel1; 7520 __u16 sel2; 7521 } s390_stsi; 7522 7523 @addr - guest address of STSI SYSIB 7524 @fc - function code 7525 @sel1 - selector 1 7526 @sel2 - selector 2 7527 @ar - access register number 7528 7529KVM handlers should exit to userspace with rc = -EREMOTE. 7530 75317.5 KVM_CAP_SPLIT_IRQCHIP 7532------------------------- 7533 7534:Architectures: x86 7535:Parameters: args[0] - number of routes reserved for userspace IOAPICs 7536:Returns: 0 on success, -1 on error 7537 7538Create a local apic for each processor in the kernel. This can be used 7539instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the 7540IOAPIC and PIC (and also the PIT, even though this has to be enabled 7541separately). 7542 7543This capability also enables in kernel routing of interrupt requests; 7544when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are 7545used in the IRQ routing table. The first args[0] MSI routes are reserved 7546for the IOAPIC pins. Whenever the LAPIC receives an EOI for these routes, 7547a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace. 7548 7549Fails if VCPU has already been created, or if the irqchip is already in the 7550kernel (i.e. KVM_CREATE_IRQCHIP has already been called). 7551 75527.6 KVM_CAP_S390_RI 7553------------------- 7554 7555:Architectures: s390 7556:Parameters: none 7557 7558Allows use of runtime-instrumentation introduced with zEC12 processor. 7559Will return -EINVAL if the machine does not support runtime-instrumentation. 7560Will return -EBUSY if a VCPU has already been created. 7561 75627.7 KVM_CAP_X2APIC_API 7563---------------------- 7564 7565:Architectures: x86 7566:Parameters: args[0] - features that should be enabled 7567:Returns: 0 on success, -EINVAL when args[0] contains invalid features 7568 7569Valid feature flags in args[0] are:: 7570 7571 #define KVM_X2APIC_API_USE_32BIT_IDS (1ULL << 0) 7572 #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK (1ULL << 1) 7573 7574Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of 7575KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC, 7576allowing the use of 32-bit APIC IDs. See KVM_CAP_X2APIC_API in their 7577respective sections. 7578 7579KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work 7580in logical mode or with more than 255 VCPUs. Otherwise, KVM treats 0xff 7581as a broadcast even in x2APIC mode in order to support physical x2APIC 7582without interrupt remapping. This is undesirable in logical mode, 7583where 0xff represents CPUs 0-7 in cluster 0. 7584 75857.8 KVM_CAP_S390_USER_INSTR0 7586---------------------------- 7587 7588:Architectures: s390 7589:Parameters: none 7590 7591With this capability enabled, all illegal instructions 0x0000 (2 bytes) will 7592be intercepted and forwarded to user space. User space can use this 7593mechanism e.g. to realize 2-byte software breakpoints. The kernel will 7594not inject an operating exception for these instructions, user space has 7595to take care of that. 7596 7597This capability can be enabled dynamically even if VCPUs were already 7598created and are running. 7599 76007.9 KVM_CAP_S390_GS 7601------------------- 7602 7603:Architectures: s390 7604:Parameters: none 7605:Returns: 0 on success; -EINVAL if the machine does not support 7606 guarded storage; -EBUSY if a VCPU has already been created. 7607 7608Allows use of guarded storage for the KVM guest. 7609 76107.10 KVM_CAP_S390_AIS 7611--------------------- 7612 7613:Architectures: s390 7614:Parameters: none 7615 7616Allow use of adapter-interruption suppression. 7617:Returns: 0 on success; -EBUSY if a VCPU has already been created. 7618 76197.11 KVM_CAP_PPC_SMT 7620-------------------- 7621 7622:Architectures: ppc 7623:Parameters: vsmt_mode, flags 7624 7625Enabling this capability on a VM provides userspace with a way to set 7626the desired virtual SMT mode (i.e. the number of virtual CPUs per 7627virtual core). The virtual SMT mode, vsmt_mode, must be a power of 2 7628between 1 and 8. On POWER8, vsmt_mode must also be no greater than 7629the number of threads per subcore for the host. Currently flags must 7630be 0. A successful call to enable this capability will result in 7631vsmt_mode being returned when the KVM_CAP_PPC_SMT capability is 7632subsequently queried for the VM. This capability is only supported by 7633HV KVM, and can only be set before any VCPUs have been created. 7634The KVM_CAP_PPC_SMT_POSSIBLE capability indicates which virtual SMT 7635modes are available. 7636 76377.12 KVM_CAP_PPC_FWNMI 7638---------------------- 7639 7640:Architectures: ppc 7641:Parameters: none 7642 7643With this capability a machine check exception in the guest address 7644space will cause KVM to exit the guest with NMI exit reason. This 7645enables QEMU to build error log and branch to guest kernel registered 7646machine check handling routine. Without this capability KVM will 7647branch to guests' 0x200 interrupt vector. 7648 76497.13 KVM_CAP_X86_DISABLE_EXITS 7650------------------------------ 7651 7652:Architectures: x86 7653:Parameters: args[0] defines which exits are disabled 7654:Returns: 0 on success, -EINVAL when args[0] contains invalid exits 7655 7656Valid bits in args[0] are:: 7657 7658 #define KVM_X86_DISABLE_EXITS_MWAIT (1 << 0) 7659 #define KVM_X86_DISABLE_EXITS_HLT (1 << 1) 7660 #define KVM_X86_DISABLE_EXITS_PAUSE (1 << 2) 7661 #define KVM_X86_DISABLE_EXITS_CSTATE (1 << 3) 7662 7663Enabling this capability on a VM provides userspace with a way to no 7664longer intercept some instructions for improved latency in some 7665workloads, and is suggested when vCPUs are associated to dedicated 7666physical CPUs. More bits can be added in the future; userspace can 7667just pass the KVM_CHECK_EXTENSION result to KVM_ENABLE_CAP to disable 7668all such vmexits. 7669 7670Do not enable KVM_FEATURE_PV_UNHALT if you disable HLT exits. 7671 76727.14 KVM_CAP_S390_HPAGE_1M 7673-------------------------- 7674 7675:Architectures: s390 7676:Parameters: none 7677:Returns: 0 on success, -EINVAL if hpage module parameter was not set 7678 or cmma is enabled, or the VM has the KVM_VM_S390_UCONTROL 7679 flag set 7680 7681With this capability the KVM support for memory backing with 1m pages 7682through hugetlbfs can be enabled for a VM. After the capability is 7683enabled, cmma can't be enabled anymore and pfmfi and the storage key 7684interpretation are disabled. If cmma has already been enabled or the 7685hpage module parameter is not set to 1, -EINVAL is returned. 7686 7687While it is generally possible to create a huge page backed VM without 7688this capability, the VM will not be able to run. 7689 76907.15 KVM_CAP_MSR_PLATFORM_INFO 7691------------------------------ 7692 7693:Architectures: x86 7694:Parameters: args[0] whether feature should be enabled or not 7695 7696With this capability, a guest may read the MSR_PLATFORM_INFO MSR. Otherwise, 7697a #GP would be raised when the guest tries to access. Currently, this 7698capability does not enable write permissions of this MSR for the guest. 7699 77007.16 KVM_CAP_PPC_NESTED_HV 7701-------------------------- 7702 7703:Architectures: ppc 7704:Parameters: none 7705:Returns: 0 on success, -EINVAL when the implementation doesn't support 7706 nested-HV virtualization. 7707 7708HV-KVM on POWER9 and later systems allows for "nested-HV" 7709virtualization, which provides a way for a guest VM to run guests that 7710can run using the CPU's supervisor mode (privileged non-hypervisor 7711state). Enabling this capability on a VM depends on the CPU having 7712the necessary functionality and on the facility being enabled with a 7713kvm-hv module parameter. 7714 77157.17 KVM_CAP_EXCEPTION_PAYLOAD 7716------------------------------ 7717 7718:Architectures: x86 7719:Parameters: args[0] whether feature should be enabled or not 7720 7721With this capability enabled, CR2 will not be modified prior to the 7722emulated VM-exit when L1 intercepts a #PF exception that occurs in 7723L2. Similarly, for kvm-intel only, DR6 will not be modified prior to 7724the emulated VM-exit when L1 intercepts a #DB exception that occurs in 7725L2. As a result, when KVM_GET_VCPU_EVENTS reports a pending #PF (or 7726#DB) exception for L2, exception.has_payload will be set and the 7727faulting address (or the new DR6 bits*) will be reported in the 7728exception_payload field. Similarly, when userspace injects a #PF (or 7729#DB) into L2 using KVM_SET_VCPU_EVENTS, it is expected to set 7730exception.has_payload and to put the faulting address - or the new DR6 7731bits\ [#]_ - in the exception_payload field. 7732 7733This capability also enables exception.pending in struct 7734kvm_vcpu_events, which allows userspace to distinguish between pending 7735and injected exceptions. 7736 7737 7738.. [#] For the new DR6 bits, note that bit 16 is set iff the #DB exception 7739 will clear DR6.RTM. 7740 77417.18 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 7742-------------------------------------- 7743 7744:Architectures: x86, arm64, mips 7745:Parameters: args[0] whether feature should be enabled or not 7746 7747Valid flags are:: 7748 7749 #define KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (1 << 0) 7750 #define KVM_DIRTY_LOG_INITIALLY_SET (1 << 1) 7751 7752With KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE is set, KVM_GET_DIRTY_LOG will not 7753automatically clear and write-protect all pages that are returned as dirty. 7754Rather, userspace will have to do this operation separately using 7755KVM_CLEAR_DIRTY_LOG. 7756 7757At the cost of a slightly more complicated operation, this provides better 7758scalability and responsiveness for two reasons. First, 7759KVM_CLEAR_DIRTY_LOG ioctl can operate on a 64-page granularity rather 7760than requiring to sync a full memslot; this ensures that KVM does not 7761take spinlocks for an extended period of time. Second, in some cases a 7762large amount of time can pass between a call to KVM_GET_DIRTY_LOG and 7763userspace actually using the data in the page. Pages can be modified 7764during this time, which is inefficient for both the guest and userspace: 7765the guest will incur a higher penalty due to write protection faults, 7766while userspace can see false reports of dirty pages. Manual reprotection 7767helps reducing this time, improving guest performance and reducing the 7768number of dirty log false positives. 7769 7770With KVM_DIRTY_LOG_INITIALLY_SET set, all the bits of the dirty bitmap 7771will be initialized to 1 when created. This also improves performance because 7772dirty logging can be enabled gradually in small chunks on the first call 7773to KVM_CLEAR_DIRTY_LOG. KVM_DIRTY_LOG_INITIALLY_SET depends on 7774KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (it is also only available on 7775x86 and arm64 for now). 7776 7777KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 was previously available under the name 7778KVM_CAP_MANUAL_DIRTY_LOG_PROTECT, but the implementation had bugs that make 7779it hard or impossible to use it correctly. The availability of 7780KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 signals that those bugs are fixed. 7781Userspace should not try to use KVM_CAP_MANUAL_DIRTY_LOG_PROTECT. 7782 77837.19 KVM_CAP_PPC_SECURE_GUEST 7784------------------------------ 7785 7786:Architectures: ppc 7787 7788This capability indicates that KVM is running on a host that has 7789ultravisor firmware and thus can support a secure guest. On such a 7790system, a guest can ask the ultravisor to make it a secure guest, 7791one whose memory is inaccessible to the host except for pages which 7792are explicitly requested to be shared with the host. The ultravisor 7793notifies KVM when a guest requests to become a secure guest, and KVM 7794has the opportunity to veto the transition. 7795 7796If present, this capability can be enabled for a VM, meaning that KVM 7797will allow the transition to secure guest mode. Otherwise KVM will 7798veto the transition. 7799 78007.20 KVM_CAP_HALT_POLL 7801---------------------- 7802 7803:Architectures: all 7804:Target: VM 7805:Parameters: args[0] is the maximum poll time in nanoseconds 7806:Returns: 0 on success; -1 on error 7807 7808KVM_CAP_HALT_POLL overrides the kvm.halt_poll_ns module parameter to set the 7809maximum halt-polling time for all vCPUs in the target VM. This capability can 7810be invoked at any time and any number of times to dynamically change the 7811maximum halt-polling time. 7812 7813See Documentation/virt/kvm/halt-polling.rst for more information on halt 7814polling. 7815 78167.21 KVM_CAP_X86_USER_SPACE_MSR 7817------------------------------- 7818 7819:Architectures: x86 7820:Target: VM 7821:Parameters: args[0] contains the mask of KVM_MSR_EXIT_REASON_* events to report 7822:Returns: 0 on success; -1 on error 7823 7824This capability allows userspace to intercept RDMSR and WRMSR instructions if 7825access to an MSR is denied. By default, KVM injects #GP on denied accesses. 7826 7827When a guest requests to read or write an MSR, KVM may not implement all MSRs 7828that are relevant to a respective system. It also does not differentiate by 7829CPU type. 7830 7831To allow more fine grained control over MSR handling, userspace may enable 7832this capability. With it enabled, MSR accesses that match the mask specified in 7833args[0] and would trigger a #GP inside the guest will instead trigger 7834KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR exit notifications. Userspace 7835can then implement model specific MSR handling and/or user notifications 7836to inform a user that an MSR was not emulated/virtualized by KVM. 7837 7838The valid mask flags are: 7839 7840============================ =============================================== 7841 KVM_MSR_EXIT_REASON_UNKNOWN intercept accesses to unknown (to KVM) MSRs 7842 KVM_MSR_EXIT_REASON_INVAL intercept accesses that are architecturally 7843 invalid according to the vCPU model and/or mode 7844 KVM_MSR_EXIT_REASON_FILTER intercept accesses that are denied by userspace 7845 via KVM_X86_SET_MSR_FILTER 7846============================ =============================================== 7847 78487.22 KVM_CAP_X86_BUS_LOCK_EXIT 7849------------------------------- 7850 7851:Architectures: x86 7852:Target: VM 7853:Parameters: args[0] defines the policy used when bus locks detected in guest 7854:Returns: 0 on success, -EINVAL when args[0] contains invalid bits 7855 7856Valid bits in args[0] are:: 7857 7858 #define KVM_BUS_LOCK_DETECTION_OFF (1 << 0) 7859 #define KVM_BUS_LOCK_DETECTION_EXIT (1 << 1) 7860 7861Enabling this capability on a VM provides userspace with a way to select a 7862policy to handle the bus locks detected in guest. Userspace can obtain the 7863supported modes from the result of KVM_CHECK_EXTENSION and define it through 7864the KVM_ENABLE_CAP. The supported modes are mutually-exclusive. 7865 7866This capability allows userspace to force VM exits on bus locks detected in the 7867guest, irrespective whether or not the host has enabled split-lock detection 7868(which triggers an #AC exception that KVM intercepts). This capability is 7869intended to mitigate attacks where a malicious/buggy guest can exploit bus 7870locks to degrade the performance of the whole system. 7871 7872If KVM_BUS_LOCK_DETECTION_OFF is set, KVM doesn't force guest bus locks to VM 7873exit, although the host kernel's split-lock #AC detection still applies, if 7874enabled. 7875 7876If KVM_BUS_LOCK_DETECTION_EXIT is set, KVM enables a CPU feature that ensures 7877bus locks in the guest trigger a VM exit, and KVM exits to userspace for all 7878such VM exits, e.g. to allow userspace to throttle the offending guest and/or 7879apply some other policy-based mitigation. When exiting to userspace, KVM sets 7880KVM_RUN_X86_BUS_LOCK in vcpu-run->flags, and conditionally sets the exit_reason 7881to KVM_EXIT_X86_BUS_LOCK. 7882 7883Note! Detected bus locks may be coincident with other exits to userspace, i.e. 7884KVM_RUN_X86_BUS_LOCK should be checked regardless of the primary exit reason if 7885userspace wants to take action on all detected bus locks. 7886 78877.23 KVM_CAP_PPC_DAWR1 7888---------------------- 7889 7890:Architectures: ppc 7891:Parameters: none 7892:Returns: 0 on success, -EINVAL when CPU doesn't support 2nd DAWR 7893 7894This capability can be used to check / enable 2nd DAWR feature provided 7895by POWER10 processor. 7896 7897 78987.24 KVM_CAP_VM_COPY_ENC_CONTEXT_FROM 7899------------------------------------- 7900 7901Architectures: x86 SEV enabled 7902Type: vm 7903Parameters: args[0] is the fd of the source vm 7904Returns: 0 on success; ENOTTY on error 7905 7906This capability enables userspace to copy encryption context from the vm 7907indicated by the fd to the vm this is called on. 7908 7909This is intended to support in-guest workloads scheduled by the host. This 7910allows the in-guest workload to maintain its own NPTs and keeps the two vms 7911from accidentally clobbering each other with interrupts and the like (separate 7912APIC/MSRs/etc). 7913 79147.25 KVM_CAP_SGX_ATTRIBUTE 7915-------------------------- 7916 7917:Architectures: x86 7918:Target: VM 7919:Parameters: args[0] is a file handle of a SGX attribute file in securityfs 7920:Returns: 0 on success, -EINVAL if the file handle is invalid or if a requested 7921 attribute is not supported by KVM. 7922 7923KVM_CAP_SGX_ATTRIBUTE enables a userspace VMM to grant a VM access to one or 7924more privileged enclave attributes. args[0] must hold a file handle to a valid 7925SGX attribute file corresponding to an attribute that is supported/restricted 7926by KVM (currently only PROVISIONKEY). 7927 7928The SGX subsystem restricts access to a subset of enclave attributes to provide 7929additional security for an uncompromised kernel, e.g. use of the PROVISIONKEY 7930is restricted to deter malware from using the PROVISIONKEY to obtain a stable 7931system fingerprint. To prevent userspace from circumventing such restrictions 7932by running an enclave in a VM, KVM prevents access to privileged attributes by 7933default. 7934 7935See Documentation/arch/x86/sgx.rst for more details. 7936 79377.26 KVM_CAP_PPC_RPT_INVALIDATE 7938------------------------------- 7939 7940:Capability: KVM_CAP_PPC_RPT_INVALIDATE 7941:Architectures: ppc 7942:Type: vm 7943 7944This capability indicates that the kernel is capable of handling 7945H_RPT_INVALIDATE hcall. 7946 7947In order to enable the use of H_RPT_INVALIDATE in the guest, 7948user space might have to advertise it for the guest. For example, 7949IBM pSeries (sPAPR) guest starts using it if "hcall-rpt-invalidate" is 7950present in the "ibm,hypertas-functions" device-tree property. 7951 7952This capability is enabled for hypervisors on platforms like POWER9 7953that support radix MMU. 7954 79557.27 KVM_CAP_EXIT_ON_EMULATION_FAILURE 7956-------------------------------------- 7957 7958:Architectures: x86 7959:Parameters: args[0] whether the feature should be enabled or not 7960 7961When this capability is enabled, an emulation failure will result in an exit 7962to userspace with KVM_INTERNAL_ERROR (except when the emulator was invoked 7963to handle a VMware backdoor instruction). Furthermore, KVM will now provide up 7964to 15 instruction bytes for any exit to userspace resulting from an emulation 7965failure. When these exits to userspace occur use the emulation_failure struct 7966instead of the internal struct. They both have the same layout, but the 7967emulation_failure struct matches the content better. It also explicitly 7968defines the 'flags' field which is used to describe the fields in the struct 7969that are valid (ie: if KVM_INTERNAL_ERROR_EMULATION_FLAG_INSTRUCTION_BYTES is 7970set in the 'flags' field then both 'insn_size' and 'insn_bytes' have valid data 7971in them.) 7972 79737.28 KVM_CAP_ARM_MTE 7974-------------------- 7975 7976:Architectures: arm64 7977:Parameters: none 7978 7979This capability indicates that KVM (and the hardware) supports exposing the 7980Memory Tagging Extensions (MTE) to the guest. It must also be enabled by the 7981VMM before creating any VCPUs to allow the guest access. Note that MTE is only 7982available to a guest running in AArch64 mode and enabling this capability will 7983cause attempts to create AArch32 VCPUs to fail. 7984 7985When enabled the guest is able to access tags associated with any memory given 7986to the guest. KVM will ensure that the tags are maintained during swap or 7987hibernation of the host; however the VMM needs to manually save/restore the 7988tags as appropriate if the VM is migrated. 7989 7990When this capability is enabled all memory in memslots must be mapped as 7991``MAP_ANONYMOUS`` or with a RAM-based file mapping (``tmpfs``, ``memfd``), 7992attempts to create a memslot with an invalid mmap will result in an 7993-EINVAL return. 7994 7995When enabled the VMM may make use of the ``KVM_ARM_MTE_COPY_TAGS`` ioctl to 7996perform a bulk copy of tags to/from the guest. 7997 79987.29 KVM_CAP_VM_MOVE_ENC_CONTEXT_FROM 7999------------------------------------- 8000 8001:Architectures: x86 SEV enabled 8002:Type: vm 8003:Parameters: args[0] is the fd of the source vm 8004:Returns: 0 on success 8005 8006This capability enables userspace to migrate the encryption context from the VM 8007indicated by the fd to the VM this is called on. 8008 8009This is intended to support intra-host migration of VMs between userspace VMMs, 8010upgrading the VMM process without interrupting the guest. 8011 80127.30 KVM_CAP_PPC_AIL_MODE_3 8013------------------------------- 8014 8015:Capability: KVM_CAP_PPC_AIL_MODE_3 8016:Architectures: ppc 8017:Type: vm 8018 8019This capability indicates that the kernel supports the mode 3 setting for the 8020"Address Translation Mode on Interrupt" aka "Alternate Interrupt Location" 8021resource that is controlled with the H_SET_MODE hypercall. 8022 8023This capability allows a guest kernel to use a better-performance mode for 8024handling interrupts and system calls. 8025 80267.31 KVM_CAP_DISABLE_QUIRKS2 8027---------------------------- 8028 8029:Capability: KVM_CAP_DISABLE_QUIRKS2 8030:Parameters: args[0] - set of KVM quirks to disable 8031:Architectures: x86 8032:Type: vm 8033 8034This capability, if enabled, will cause KVM to disable some behavior 8035quirks. 8036 8037Calling KVM_CHECK_EXTENSION for this capability returns a bitmask of 8038quirks that can be disabled in KVM. 8039 8040The argument to KVM_ENABLE_CAP for this capability is a bitmask of 8041quirks to disable, and must be a subset of the bitmask returned by 8042KVM_CHECK_EXTENSION. 8043 8044The valid bits in cap.args[0] are: 8045 8046=================================== ============================================ 8047 KVM_X86_QUIRK_LINT0_REENABLED By default, the reset value for the LVT 8048 LINT0 register is 0x700 (APIC_MODE_EXTINT). 8049 When this quirk is disabled, the reset value 8050 is 0x10000 (APIC_LVT_MASKED). 8051 8052 KVM_X86_QUIRK_CD_NW_CLEARED By default, KVM clears CR0.CD and CR0.NW on 8053 AMD CPUs to workaround buggy guest firmware 8054 that runs in perpetuity with CR0.CD, i.e. 8055 with caches in "no fill" mode. 8056 8057 When this quirk is disabled, KVM does not 8058 change the value of CR0.CD and CR0.NW. 8059 8060 KVM_X86_QUIRK_LAPIC_MMIO_HOLE By default, the MMIO LAPIC interface is 8061 available even when configured for x2APIC 8062 mode. When this quirk is disabled, KVM 8063 disables the MMIO LAPIC interface if the 8064 LAPIC is in x2APIC mode. 8065 8066 KVM_X86_QUIRK_OUT_7E_INC_RIP By default, KVM pre-increments %rip before 8067 exiting to userspace for an OUT instruction 8068 to port 0x7e. When this quirk is disabled, 8069 KVM does not pre-increment %rip before 8070 exiting to userspace. 8071 8072 KVM_X86_QUIRK_MISC_ENABLE_NO_MWAIT When this quirk is disabled, KVM sets 8073 CPUID.01H:ECX[bit 3] (MONITOR/MWAIT) if 8074 IA32_MISC_ENABLE[bit 18] (MWAIT) is set. 8075 Additionally, when this quirk is disabled, 8076 KVM clears CPUID.01H:ECX[bit 3] if 8077 IA32_MISC_ENABLE[bit 18] is cleared. 8078 8079 KVM_X86_QUIRK_FIX_HYPERCALL_INSN By default, KVM rewrites guest 8080 VMMCALL/VMCALL instructions to match the 8081 vendor's hypercall instruction for the 8082 system. When this quirk is disabled, KVM 8083 will no longer rewrite invalid guest 8084 hypercall instructions. Executing the 8085 incorrect hypercall instruction will 8086 generate a #UD within the guest. 8087 8088KVM_X86_QUIRK_MWAIT_NEVER_UD_FAULTS By default, KVM emulates MONITOR/MWAIT (if 8089 they are intercepted) as NOPs regardless of 8090 whether or not MONITOR/MWAIT are supported 8091 according to guest CPUID. When this quirk 8092 is disabled and KVM_X86_DISABLE_EXITS_MWAIT 8093 is not set (MONITOR/MWAIT are intercepted), 8094 KVM will inject a #UD on MONITOR/MWAIT if 8095 they're unsupported per guest CPUID. Note, 8096 KVM will modify MONITOR/MWAIT support in 8097 guest CPUID on writes to MISC_ENABLE if 8098 KVM_X86_QUIRK_MISC_ENABLE_NO_MWAIT is 8099 disabled. 8100 8101KVM_X86_QUIRK_SLOT_ZAP_ALL By default, KVM invalidates all SPTEs in 8102 fast way for memslot deletion when VM type 8103 is KVM_X86_DEFAULT_VM. 8104 When this quirk is disabled or when VM type 8105 is other than KVM_X86_DEFAULT_VM, KVM zaps 8106 only leaf SPTEs that are within the range of 8107 the memslot being deleted. 8108=================================== ============================================ 8109 81107.32 KVM_CAP_MAX_VCPU_ID 8111------------------------ 8112 8113:Architectures: x86 8114:Target: VM 8115:Parameters: args[0] - maximum APIC ID value set for current VM 8116:Returns: 0 on success, -EINVAL if args[0] is beyond KVM_MAX_VCPU_IDS 8117 supported in KVM or if it has been set. 8118 8119This capability allows userspace to specify maximum possible APIC ID 8120assigned for current VM session prior to the creation of vCPUs, saving 8121memory for data structures indexed by the APIC ID. Userspace is able 8122to calculate the limit to APIC ID values from designated 8123CPU topology. 8124 8125The value can be changed only until KVM_ENABLE_CAP is set to a nonzero 8126value or until a vCPU is created. Upon creation of the first vCPU, 8127if the value was set to zero or KVM_ENABLE_CAP was not invoked, KVM 8128uses the return value of KVM_CHECK_EXTENSION(KVM_CAP_MAX_VCPU_ID) as 8129the maximum APIC ID. 8130 81317.33 KVM_CAP_X86_NOTIFY_VMEXIT 8132------------------------------ 8133 8134:Architectures: x86 8135:Target: VM 8136:Parameters: args[0] is the value of notify window as well as some flags 8137:Returns: 0 on success, -EINVAL if args[0] contains invalid flags or notify 8138 VM exit is unsupported. 8139 8140Bits 63:32 of args[0] are used for notify window. 8141Bits 31:0 of args[0] are for some flags. Valid bits are:: 8142 8143 #define KVM_X86_NOTIFY_VMEXIT_ENABLED (1 << 0) 8144 #define KVM_X86_NOTIFY_VMEXIT_USER (1 << 1) 8145 8146This capability allows userspace to configure the notify VM exit on/off 8147in per-VM scope during VM creation. Notify VM exit is disabled by default. 8148When userspace sets KVM_X86_NOTIFY_VMEXIT_ENABLED bit in args[0], VMM will 8149enable this feature with the notify window provided, which will generate 8150a VM exit if no event window occurs in VM non-root mode for a specified of 8151time (notify window). 8152 8153If KVM_X86_NOTIFY_VMEXIT_USER is set in args[0], upon notify VM exits happen, 8154KVM would exit to userspace for handling. 8155 8156This capability is aimed to mitigate the threat that malicious VMs can 8157cause CPU stuck (due to event windows don't open up) and make the CPU 8158unavailable to host or other VMs. 8159 81607.34 KVM_CAP_MEMORY_FAULT_INFO 8161------------------------------ 8162 8163:Architectures: x86 8164:Returns: Informational only, -EINVAL on direct KVM_ENABLE_CAP. 8165 8166The presence of this capability indicates that KVM_RUN will fill 8167kvm_run.memory_fault if KVM cannot resolve a guest page fault VM-Exit, e.g. if 8168there is a valid memslot but no backing VMA for the corresponding host virtual 8169address. 8170 8171The information in kvm_run.memory_fault is valid if and only if KVM_RUN returns 8172an error with errno=EFAULT or errno=EHWPOISON *and* kvm_run.exit_reason is set 8173to KVM_EXIT_MEMORY_FAULT. 8174 8175Note: Userspaces which attempt to resolve memory faults so that they can retry 8176KVM_RUN are encouraged to guard against repeatedly receiving the same 8177error/annotated fault. 8178 8179See KVM_EXIT_MEMORY_FAULT for more information. 8180 81817.35 KVM_CAP_X86_APIC_BUS_CYCLES_NS 8182----------------------------------- 8183 8184:Architectures: x86 8185:Target: VM 8186:Parameters: args[0] is the desired APIC bus clock rate, in nanoseconds 8187:Returns: 0 on success, -EINVAL if args[0] contains an invalid value for the 8188 frequency or if any vCPUs have been created, -ENXIO if a virtual 8189 local APIC has not been created using KVM_CREATE_IRQCHIP. 8190 8191This capability sets the VM's APIC bus clock frequency, used by KVM's in-kernel 8192virtual APIC when emulating APIC timers. KVM's default value can be retrieved 8193by KVM_CHECK_EXTENSION. 8194 8195Note: Userspace is responsible for correctly configuring CPUID 0x15, a.k.a. the 8196core crystal clock frequency, if a non-zero CPUID 0x15 is exposed to the guest. 8197 81987.36 KVM_CAP_X86_GUEST_MODE 8199------------------------------ 8200 8201:Architectures: x86 8202:Returns: Informational only, -EINVAL on direct KVM_ENABLE_CAP. 8203 8204The presence of this capability indicates that KVM_RUN will update the 8205KVM_RUN_X86_GUEST_MODE bit in kvm_run.flags to indicate whether the 8206vCPU was executing nested guest code when it exited. 8207 8208KVM exits with the register state of either the L1 or L2 guest 8209depending on which executed at the time of an exit. Userspace must 8210take care to differentiate between these cases. 8211 82128. Other capabilities. 8213====================== 8214 8215This section lists capabilities that give information about other 8216features of the KVM implementation. 8217 82188.1 KVM_CAP_PPC_HWRNG 8219--------------------- 8220 8221:Architectures: ppc 8222 8223This capability, if KVM_CHECK_EXTENSION indicates that it is 8224available, means that the kernel has an implementation of the 8225H_RANDOM hypercall backed by a hardware random-number generator. 8226If present, the kernel H_RANDOM handler can be enabled for guest use 8227with the KVM_CAP_PPC_ENABLE_HCALL capability. 8228 82298.2 KVM_CAP_HYPERV_SYNIC 8230------------------------ 8231 8232:Architectures: x86 8233 8234This capability, if KVM_CHECK_EXTENSION indicates that it is 8235available, means that the kernel has an implementation of the 8236Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is 8237used to support Windows Hyper-V based guest paravirt drivers(VMBus). 8238 8239In order to use SynIC, it has to be activated by setting this 8240capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this 8241will disable the use of APIC hardware virtualization even if supported 8242by the CPU, as it's incompatible with SynIC auto-EOI behavior. 8243 82448.3 KVM_CAP_PPC_MMU_RADIX 8245------------------------- 8246 8247:Architectures: ppc 8248 8249This capability, if KVM_CHECK_EXTENSION indicates that it is 8250available, means that the kernel can support guests using the 8251radix MMU defined in Power ISA V3.00 (as implemented in the POWER9 8252processor). 8253 82548.4 KVM_CAP_PPC_MMU_HASH_V3 8255--------------------------- 8256 8257:Architectures: ppc 8258 8259This capability, if KVM_CHECK_EXTENSION indicates that it is 8260available, means that the kernel can support guests using the 8261hashed page table MMU defined in Power ISA V3.00 (as implemented in 8262the POWER9 processor), including in-memory segment tables. 8263 82648.5 KVM_CAP_MIPS_VZ 8265------------------- 8266 8267:Architectures: mips 8268 8269This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that 8270it is available, means that full hardware assisted virtualization capabilities 8271of the hardware are available for use through KVM. An appropriate 8272KVM_VM_MIPS_* type must be passed to KVM_CREATE_VM to create a VM which 8273utilises it. 8274 8275If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is 8276available, it means that the VM is using full hardware assisted virtualization 8277capabilities of the hardware. This is useful to check after creating a VM with 8278KVM_VM_MIPS_DEFAULT. 8279 8280The value returned by KVM_CHECK_EXTENSION should be compared against known 8281values (see below). All other values are reserved. This is to allow for the 8282possibility of other hardware assisted virtualization implementations which 8283may be incompatible with the MIPS VZ ASE. 8284 8285== ========================================================================== 8286 0 The trap & emulate implementation is in use to run guest code in user 8287 mode. Guest virtual memory segments are rearranged to fit the guest in the 8288 user mode address space. 8289 8290 1 The MIPS VZ ASE is in use, providing full hardware assisted 8291 virtualization, including standard guest virtual memory segments. 8292== ========================================================================== 8293 82948.6 KVM_CAP_MIPS_TE 8295------------------- 8296 8297:Architectures: mips 8298 8299This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that 8300it is available, means that the trap & emulate implementation is available to 8301run guest code in user mode, even if KVM_CAP_MIPS_VZ indicates that hardware 8302assisted virtualisation is also available. KVM_VM_MIPS_TE (0) must be passed 8303to KVM_CREATE_VM to create a VM which utilises it. 8304 8305If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is 8306available, it means that the VM is using trap & emulate. 8307 83088.7 KVM_CAP_MIPS_64BIT 8309---------------------- 8310 8311:Architectures: mips 8312 8313This capability indicates the supported architecture type of the guest, i.e. the 8314supported register and address width. 8315 8316The values returned when this capability is checked by KVM_CHECK_EXTENSION on a 8317kvm VM handle correspond roughly to the CP0_Config.AT register field, and should 8318be checked specifically against known values (see below). All other values are 8319reserved. 8320 8321== ======================================================================== 8322 0 MIPS32 or microMIPS32. 8323 Both registers and addresses are 32-bits wide. 8324 It will only be possible to run 32-bit guest code. 8325 8326 1 MIPS64 or microMIPS64 with access only to 32-bit compatibility segments. 8327 Registers are 64-bits wide, but addresses are 32-bits wide. 8328 64-bit guest code may run but cannot access MIPS64 memory segments. 8329 It will also be possible to run 32-bit guest code. 8330 8331 2 MIPS64 or microMIPS64 with access to all address segments. 8332 Both registers and addresses are 64-bits wide. 8333 It will be possible to run 64-bit or 32-bit guest code. 8334== ======================================================================== 8335 83368.9 KVM_CAP_ARM_USER_IRQ 8337------------------------ 8338 8339:Architectures: arm64 8340 8341This capability, if KVM_CHECK_EXTENSION indicates that it is available, means 8342that if userspace creates a VM without an in-kernel interrupt controller, it 8343will be notified of changes to the output level of in-kernel emulated devices, 8344which can generate virtual interrupts, presented to the VM. 8345For such VMs, on every return to userspace, the kernel 8346updates the vcpu's run->s.regs.device_irq_level field to represent the actual 8347output level of the device. 8348 8349Whenever kvm detects a change in the device output level, kvm guarantees at 8350least one return to userspace before running the VM. This exit could either 8351be a KVM_EXIT_INTR or any other exit event, like KVM_EXIT_MMIO. This way, 8352userspace can always sample the device output level and re-compute the state of 8353the userspace interrupt controller. Userspace should always check the state 8354of run->s.regs.device_irq_level on every kvm exit. 8355The value in run->s.regs.device_irq_level can represent both level and edge 8356triggered interrupt signals, depending on the device. Edge triggered interrupt 8357signals will exit to userspace with the bit in run->s.regs.device_irq_level 8358set exactly once per edge signal. 8359 8360The field run->s.regs.device_irq_level is available independent of 8361run->kvm_valid_regs or run->kvm_dirty_regs bits. 8362 8363If KVM_CAP_ARM_USER_IRQ is supported, the KVM_CHECK_EXTENSION ioctl returns a 8364number larger than 0 indicating the version of this capability is implemented 8365and thereby which bits in run->s.regs.device_irq_level can signal values. 8366 8367Currently the following bits are defined for the device_irq_level bitmap:: 8368 8369 KVM_CAP_ARM_USER_IRQ >= 1: 8370 8371 KVM_ARM_DEV_EL1_VTIMER - EL1 virtual timer 8372 KVM_ARM_DEV_EL1_PTIMER - EL1 physical timer 8373 KVM_ARM_DEV_PMU - ARM PMU overflow interrupt signal 8374 8375Future versions of kvm may implement additional events. These will get 8376indicated by returning a higher number from KVM_CHECK_EXTENSION and will be 8377listed above. 8378 83798.10 KVM_CAP_PPC_SMT_POSSIBLE 8380----------------------------- 8381 8382:Architectures: ppc 8383 8384Querying this capability returns a bitmap indicating the possible 8385virtual SMT modes that can be set using KVM_CAP_PPC_SMT. If bit N 8386(counting from the right) is set, then a virtual SMT mode of 2^N is 8387available. 8388 83898.11 KVM_CAP_HYPERV_SYNIC2 8390-------------------------- 8391 8392:Architectures: x86 8393 8394This capability enables a newer version of Hyper-V Synthetic interrupt 8395controller (SynIC). The only difference with KVM_CAP_HYPERV_SYNIC is that KVM 8396doesn't clear SynIC message and event flags pages when they are enabled by 8397writing to the respective MSRs. 8398 83998.12 KVM_CAP_HYPERV_VP_INDEX 8400---------------------------- 8401 8402:Architectures: x86 8403 8404This capability indicates that userspace can load HV_X64_MSR_VP_INDEX msr. Its 8405value is used to denote the target vcpu for a SynIC interrupt. For 8406compatibility, KVM initializes this msr to KVM's internal vcpu index. When this 8407capability is absent, userspace can still query this msr's value. 8408 84098.13 KVM_CAP_S390_AIS_MIGRATION 8410------------------------------- 8411 8412:Architectures: s390 8413:Parameters: none 8414 8415This capability indicates if the flic device will be able to get/set the 8416AIS states for migration via the KVM_DEV_FLIC_AISM_ALL attribute and allows 8417to discover this without having to create a flic device. 8418 84198.14 KVM_CAP_S390_PSW 8420--------------------- 8421 8422:Architectures: s390 8423 8424This capability indicates that the PSW is exposed via the kvm_run structure. 8425 84268.15 KVM_CAP_S390_GMAP 8427---------------------- 8428 8429:Architectures: s390 8430 8431This capability indicates that the user space memory used as guest mapping can 8432be anywhere in the user memory address space, as long as the memory slots are 8433aligned and sized to a segment (1MB) boundary. 8434 84358.16 KVM_CAP_S390_COW 8436--------------------- 8437 8438:Architectures: s390 8439 8440This capability indicates that the user space memory used as guest mapping can 8441use copy-on-write semantics as well as dirty pages tracking via read-only page 8442tables. 8443 84448.17 KVM_CAP_S390_BPB 8445--------------------- 8446 8447:Architectures: s390 8448 8449This capability indicates that kvm will implement the interfaces to handle 8450reset, migration and nested KVM for branch prediction blocking. The stfle 8451facility 82 should not be provided to the guest without this capability. 8452 84538.18 KVM_CAP_HYPERV_TLBFLUSH 8454---------------------------- 8455 8456:Architectures: x86 8457 8458This capability indicates that KVM supports paravirtualized Hyper-V TLB Flush 8459hypercalls: 8460HvFlushVirtualAddressSpace, HvFlushVirtualAddressSpaceEx, 8461HvFlushVirtualAddressList, HvFlushVirtualAddressListEx. 8462 84638.19 KVM_CAP_ARM_INJECT_SERROR_ESR 8464---------------------------------- 8465 8466:Architectures: arm64 8467 8468This capability indicates that userspace can specify (via the 8469KVM_SET_VCPU_EVENTS ioctl) the syndrome value reported to the guest when it 8470takes a virtual SError interrupt exception. 8471If KVM advertises this capability, userspace can only specify the ISS field for 8472the ESR syndrome. Other parts of the ESR, such as the EC are generated by the 8473CPU when the exception is taken. If this virtual SError is taken to EL1 using 8474AArch64, this value will be reported in the ISS field of ESR_ELx. 8475 8476See KVM_CAP_VCPU_EVENTS for more details. 8477 84788.20 KVM_CAP_HYPERV_SEND_IPI 8479---------------------------- 8480 8481:Architectures: x86 8482 8483This capability indicates that KVM supports paravirtualized Hyper-V IPI send 8484hypercalls: 8485HvCallSendSyntheticClusterIpi, HvCallSendSyntheticClusterIpiEx. 8486 84878.21 KVM_CAP_HYPERV_DIRECT_TLBFLUSH 8488----------------------------------- 8489 8490:Architectures: x86 8491 8492This capability indicates that KVM running on top of Hyper-V hypervisor 8493enables Direct TLB flush for its guests meaning that TLB flush 8494hypercalls are handled by Level 0 hypervisor (Hyper-V) bypassing KVM. 8495Due to the different ABI for hypercall parameters between Hyper-V and 8496KVM, enabling this capability effectively disables all hypercall 8497handling by KVM (as some KVM hypercall may be mistakenly treated as TLB 8498flush hypercalls by Hyper-V) so userspace should disable KVM identification 8499in CPUID and only exposes Hyper-V identification. In this case, guest 8500thinks it's running on Hyper-V and only use Hyper-V hypercalls. 8501 85028.22 KVM_CAP_S390_VCPU_RESETS 8503----------------------------- 8504 8505:Architectures: s390 8506 8507This capability indicates that the KVM_S390_NORMAL_RESET and 8508KVM_S390_CLEAR_RESET ioctls are available. 8509 85108.23 KVM_CAP_S390_PROTECTED 8511--------------------------- 8512 8513:Architectures: s390 8514 8515This capability indicates that the Ultravisor has been initialized and 8516KVM can therefore start protected VMs. 8517This capability governs the KVM_S390_PV_COMMAND ioctl and the 8518KVM_MP_STATE_LOAD MP_STATE. KVM_SET_MP_STATE can fail for protected 8519guests when the state change is invalid. 8520 85218.24 KVM_CAP_STEAL_TIME 8522----------------------- 8523 8524:Architectures: arm64, x86 8525 8526This capability indicates that KVM supports steal time accounting. 8527When steal time accounting is supported it may be enabled with 8528architecture-specific interfaces. This capability and the architecture- 8529specific interfaces must be consistent, i.e. if one says the feature 8530is supported, than the other should as well and vice versa. For arm64 8531see Documentation/virt/kvm/devices/vcpu.rst "KVM_ARM_VCPU_PVTIME_CTRL". 8532For x86 see Documentation/virt/kvm/x86/msr.rst "MSR_KVM_STEAL_TIME". 8533 85348.25 KVM_CAP_S390_DIAG318 8535------------------------- 8536 8537:Architectures: s390 8538 8539This capability enables a guest to set information about its control program 8540(i.e. guest kernel type and version). The information is helpful during 8541system/firmware service events, providing additional data about the guest 8542environments running on the machine. 8543 8544The information is associated with the DIAGNOSE 0x318 instruction, which sets 8545an 8-byte value consisting of a one-byte Control Program Name Code (CPNC) and 8546a 7-byte Control Program Version Code (CPVC). The CPNC determines what 8547environment the control program is running in (e.g. Linux, z/VM...), and the 8548CPVC is used for information specific to OS (e.g. Linux version, Linux 8549distribution...) 8550 8551If this capability is available, then the CPNC and CPVC can be synchronized 8552between KVM and userspace via the sync regs mechanism (KVM_SYNC_DIAG318). 8553 85548.26 KVM_CAP_X86_USER_SPACE_MSR 8555------------------------------- 8556 8557:Architectures: x86 8558 8559This capability indicates that KVM supports deflection of MSR reads and 8560writes to user space. It can be enabled on a VM level. If enabled, MSR 8561accesses that would usually trigger a #GP by KVM into the guest will 8562instead get bounced to user space through the KVM_EXIT_X86_RDMSR and 8563KVM_EXIT_X86_WRMSR exit notifications. 8564 85658.27 KVM_CAP_X86_MSR_FILTER 8566--------------------------- 8567 8568:Architectures: x86 8569 8570This capability indicates that KVM supports that accesses to user defined MSRs 8571may be rejected. With this capability exposed, KVM exports new VM ioctl 8572KVM_X86_SET_MSR_FILTER which user space can call to specify bitmaps of MSR 8573ranges that KVM should deny access to. 8574 8575In combination with KVM_CAP_X86_USER_SPACE_MSR, this allows user space to 8576trap and emulate MSRs that are outside of the scope of KVM as well as 8577limit the attack surface on KVM's MSR emulation code. 8578 85798.28 KVM_CAP_ENFORCE_PV_FEATURE_CPUID 8580------------------------------------- 8581 8582Architectures: x86 8583 8584When enabled, KVM will disable paravirtual features provided to the 8585guest according to the bits in the KVM_CPUID_FEATURES CPUID leaf 8586(0x40000001). Otherwise, a guest may use the paravirtual features 8587regardless of what has actually been exposed through the CPUID leaf. 8588 85898.29 KVM_CAP_DIRTY_LOG_RING/KVM_CAP_DIRTY_LOG_RING_ACQ_REL 8590---------------------------------------------------------- 8591 8592:Architectures: x86, arm64 8593:Parameters: args[0] - size of the dirty log ring 8594 8595KVM is capable of tracking dirty memory using ring buffers that are 8596mmapped into userspace; there is one dirty ring per vcpu. 8597 8598The dirty ring is available to userspace as an array of 8599``struct kvm_dirty_gfn``. Each dirty entry is defined as:: 8600 8601 struct kvm_dirty_gfn { 8602 __u32 flags; 8603 __u32 slot; /* as_id | slot_id */ 8604 __u64 offset; 8605 }; 8606 8607The following values are defined for the flags field to define the 8608current state of the entry:: 8609 8610 #define KVM_DIRTY_GFN_F_DIRTY BIT(0) 8611 #define KVM_DIRTY_GFN_F_RESET BIT(1) 8612 #define KVM_DIRTY_GFN_F_MASK 0x3 8613 8614Userspace should call KVM_ENABLE_CAP ioctl right after KVM_CREATE_VM 8615ioctl to enable this capability for the new guest and set the size of 8616the rings. Enabling the capability is only allowed before creating any 8617vCPU, and the size of the ring must be a power of two. The larger the 8618ring buffer, the less likely the ring is full and the VM is forced to 8619exit to userspace. The optimal size depends on the workload, but it is 8620recommended that it be at least 64 KiB (4096 entries). 8621 8622Just like for dirty page bitmaps, the buffer tracks writes to 8623all user memory regions for which the KVM_MEM_LOG_DIRTY_PAGES flag was 8624set in KVM_SET_USER_MEMORY_REGION. Once a memory region is registered 8625with the flag set, userspace can start harvesting dirty pages from the 8626ring buffer. 8627 8628An entry in the ring buffer can be unused (flag bits ``00``), 8629dirty (flag bits ``01``) or harvested (flag bits ``1X``). The 8630state machine for the entry is as follows:: 8631 8632 dirtied harvested reset 8633 00 -----------> 01 -------------> 1X -------+ 8634 ^ | 8635 | | 8636 +------------------------------------------+ 8637 8638To harvest the dirty pages, userspace accesses the mmapped ring buffer 8639to read the dirty GFNs. If the flags has the DIRTY bit set (at this stage 8640the RESET bit must be cleared), then it means this GFN is a dirty GFN. 8641The userspace should harvest this GFN and mark the flags from state 8642``01b`` to ``1Xb`` (bit 0 will be ignored by KVM, but bit 1 must be set 8643to show that this GFN is harvested and waiting for a reset), and move 8644on to the next GFN. The userspace should continue to do this until the 8645flags of a GFN have the DIRTY bit cleared, meaning that it has harvested 8646all the dirty GFNs that were available. 8647 8648Note that on weakly ordered architectures, userspace accesses to the 8649ring buffer (and more specifically the 'flags' field) must be ordered, 8650using load-acquire/store-release accessors when available, or any 8651other memory barrier that will ensure this ordering. 8652 8653It's not necessary for userspace to harvest the all dirty GFNs at once. 8654However it must collect the dirty GFNs in sequence, i.e., the userspace 8655program cannot skip one dirty GFN to collect the one next to it. 8656 8657After processing one or more entries in the ring buffer, userspace 8658calls the VM ioctl KVM_RESET_DIRTY_RINGS to notify the kernel about 8659it, so that the kernel will reprotect those collected GFNs. 8660Therefore, the ioctl must be called *before* reading the content of 8661the dirty pages. 8662 8663The dirty ring can get full. When it happens, the KVM_RUN of the 8664vcpu will return with exit reason KVM_EXIT_DIRTY_LOG_FULL. 8665 8666The dirty ring interface has a major difference comparing to the 8667KVM_GET_DIRTY_LOG interface in that, when reading the dirty ring from 8668userspace, it's still possible that the kernel has not yet flushed the 8669processor's dirty page buffers into the kernel buffer (with dirty bitmaps, the 8670flushing is done by the KVM_GET_DIRTY_LOG ioctl). To achieve that, one 8671needs to kick the vcpu out of KVM_RUN using a signal. The resulting 8672vmexit ensures that all dirty GFNs are flushed to the dirty rings. 8673 8674NOTE: KVM_CAP_DIRTY_LOG_RING_ACQ_REL is the only capability that 8675should be exposed by weakly ordered architecture, in order to indicate 8676the additional memory ordering requirements imposed on userspace when 8677reading the state of an entry and mutating it from DIRTY to HARVESTED. 8678Architecture with TSO-like ordering (such as x86) are allowed to 8679expose both KVM_CAP_DIRTY_LOG_RING and KVM_CAP_DIRTY_LOG_RING_ACQ_REL 8680to userspace. 8681 8682After enabling the dirty rings, the userspace needs to detect the 8683capability of KVM_CAP_DIRTY_LOG_RING_WITH_BITMAP to see whether the 8684ring structures can be backed by per-slot bitmaps. With this capability 8685advertised, it means the architecture can dirty guest pages without 8686vcpu/ring context, so that some of the dirty information will still be 8687maintained in the bitmap structure. KVM_CAP_DIRTY_LOG_RING_WITH_BITMAP 8688can't be enabled if the capability of KVM_CAP_DIRTY_LOG_RING_ACQ_REL 8689hasn't been enabled, or any memslot has been existing. 8690 8691Note that the bitmap here is only a backup of the ring structure. The 8692use of the ring and bitmap combination is only beneficial if there is 8693only a very small amount of memory that is dirtied out of vcpu/ring 8694context. Otherwise, the stand-alone per-slot bitmap mechanism needs to 8695be considered. 8696 8697To collect dirty bits in the backup bitmap, userspace can use the same 8698KVM_GET_DIRTY_LOG ioctl. KVM_CLEAR_DIRTY_LOG isn't needed as long as all 8699the generation of the dirty bits is done in a single pass. Collecting 8700the dirty bitmap should be the very last thing that the VMM does before 8701considering the state as complete. VMM needs to ensure that the dirty 8702state is final and avoid missing dirty pages from another ioctl ordered 8703after the bitmap collection. 8704 8705NOTE: Multiple examples of using the backup bitmap: (1) save vgic/its 8706tables through command KVM_DEV_ARM_{VGIC_GRP_CTRL, ITS_SAVE_TABLES} on 8707KVM device "kvm-arm-vgic-its". (2) restore vgic/its tables through 8708command KVM_DEV_ARM_{VGIC_GRP_CTRL, ITS_RESTORE_TABLES} on KVM device 8709"kvm-arm-vgic-its". VGICv3 LPI pending status is restored. (3) save 8710vgic3 pending table through KVM_DEV_ARM_VGIC_{GRP_CTRL, SAVE_PENDING_TABLES} 8711command on KVM device "kvm-arm-vgic-v3". 8712 87138.30 KVM_CAP_XEN_HVM 8714-------------------- 8715 8716:Architectures: x86 8717 8718This capability indicates the features that Xen supports for hosting Xen 8719PVHVM guests. Valid flags are:: 8720 8721 #define KVM_XEN_HVM_CONFIG_HYPERCALL_MSR (1 << 0) 8722 #define KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL (1 << 1) 8723 #define KVM_XEN_HVM_CONFIG_SHARED_INFO (1 << 2) 8724 #define KVM_XEN_HVM_CONFIG_RUNSTATE (1 << 3) 8725 #define KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL (1 << 4) 8726 #define KVM_XEN_HVM_CONFIG_EVTCHN_SEND (1 << 5) 8727 #define KVM_XEN_HVM_CONFIG_RUNSTATE_UPDATE_FLAG (1 << 6) 8728 #define KVM_XEN_HVM_CONFIG_PVCLOCK_TSC_UNSTABLE (1 << 7) 8729 8730The KVM_XEN_HVM_CONFIG_HYPERCALL_MSR flag indicates that the KVM_XEN_HVM_CONFIG 8731ioctl is available, for the guest to set its hypercall page. 8732 8733If KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL is also set, the same flag may also be 8734provided in the flags to KVM_XEN_HVM_CONFIG, without providing hypercall page 8735contents, to request that KVM generate hypercall page content automatically 8736and also enable interception of guest hypercalls with KVM_EXIT_XEN. 8737 8738The KVM_XEN_HVM_CONFIG_SHARED_INFO flag indicates the availability of the 8739KVM_XEN_HVM_SET_ATTR, KVM_XEN_HVM_GET_ATTR, KVM_XEN_VCPU_SET_ATTR and 8740KVM_XEN_VCPU_GET_ATTR ioctls, as well as the delivery of exception vectors 8741for event channel upcalls when the evtchn_upcall_pending field of a vcpu's 8742vcpu_info is set. 8743 8744The KVM_XEN_HVM_CONFIG_RUNSTATE flag indicates that the runstate-related 8745features KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR/_CURRENT/_DATA/_ADJUST are 8746supported by the KVM_XEN_VCPU_SET_ATTR/KVM_XEN_VCPU_GET_ATTR ioctls. 8747 8748The KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL flag indicates that IRQ routing entries 8749of the type KVM_IRQ_ROUTING_XEN_EVTCHN are supported, with the priority 8750field set to indicate 2 level event channel delivery. 8751 8752The KVM_XEN_HVM_CONFIG_EVTCHN_SEND flag indicates that KVM supports 8753injecting event channel events directly into the guest with the 8754KVM_XEN_HVM_EVTCHN_SEND ioctl. It also indicates support for the 8755KVM_XEN_ATTR_TYPE_EVTCHN/XEN_VERSION HVM attributes and the 8756KVM_XEN_VCPU_ATTR_TYPE_VCPU_ID/TIMER/UPCALL_VECTOR vCPU attributes. 8757related to event channel delivery, timers, and the XENVER_version 8758interception. 8759 8760The KVM_XEN_HVM_CONFIG_RUNSTATE_UPDATE_FLAG flag indicates that KVM supports 8761the KVM_XEN_ATTR_TYPE_RUNSTATE_UPDATE_FLAG attribute in the KVM_XEN_SET_ATTR 8762and KVM_XEN_GET_ATTR ioctls. This controls whether KVM will set the 8763XEN_RUNSTATE_UPDATE flag in guest memory mapped vcpu_runstate_info during 8764updates of the runstate information. Note that versions of KVM which support 8765the RUNSTATE feature above, but not the RUNSTATE_UPDATE_FLAG feature, will 8766always set the XEN_RUNSTATE_UPDATE flag when updating the guest structure, 8767which is perhaps counterintuitive. When this flag is advertised, KVM will 8768behave more correctly, not using the XEN_RUNSTATE_UPDATE flag until/unless 8769specifically enabled (by the guest making the hypercall, causing the VMM 8770to enable the KVM_XEN_ATTR_TYPE_RUNSTATE_UPDATE_FLAG attribute). 8771 8772The KVM_XEN_HVM_CONFIG_PVCLOCK_TSC_UNSTABLE flag indicates that KVM supports 8773clearing the PVCLOCK_TSC_STABLE_BIT flag in Xen pvclock sources. This will be 8774done when the KVM_CAP_XEN_HVM ioctl sets the 8775KVM_XEN_HVM_CONFIG_PVCLOCK_TSC_UNSTABLE flag. 8776 87778.31 KVM_CAP_PPC_MULTITCE 8778------------------------- 8779 8780:Capability: KVM_CAP_PPC_MULTITCE 8781:Architectures: ppc 8782:Type: vm 8783 8784This capability means the kernel is capable of handling hypercalls 8785H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user 8786space. This significantly accelerates DMA operations for PPC KVM guests. 8787User space should expect that its handlers for these hypercalls 8788are not going to be called if user space previously registered LIOBN 8789in KVM (via KVM_CREATE_SPAPR_TCE or similar calls). 8790 8791In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest, 8792user space might have to advertise it for the guest. For example, 8793IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is 8794present in the "ibm,hypertas-functions" device-tree property. 8795 8796The hypercalls mentioned above may or may not be processed successfully 8797in the kernel based fast path. If they can not be handled by the kernel, 8798they will get passed on to user space. So user space still has to have 8799an implementation for these despite the in kernel acceleration. 8800 8801This capability is always enabled. 8802 88038.32 KVM_CAP_PTP_KVM 8804-------------------- 8805 8806:Architectures: arm64 8807 8808This capability indicates that the KVM virtual PTP service is 8809supported in the host. A VMM can check whether the service is 8810available to the guest on migration. 8811 88128.33 KVM_CAP_HYPERV_ENFORCE_CPUID 8813--------------------------------- 8814 8815Architectures: x86 8816 8817When enabled, KVM will disable emulated Hyper-V features provided to the 8818guest according to the bits Hyper-V CPUID feature leaves. Otherwise, all 8819currently implemented Hyper-V features are provided unconditionally when 8820Hyper-V identification is set in the HYPERV_CPUID_INTERFACE (0x40000001) 8821leaf. 8822 88238.34 KVM_CAP_EXIT_HYPERCALL 8824--------------------------- 8825 8826:Capability: KVM_CAP_EXIT_HYPERCALL 8827:Architectures: x86 8828:Type: vm 8829 8830This capability, if enabled, will cause KVM to exit to userspace 8831with KVM_EXIT_HYPERCALL exit reason to process some hypercalls. 8832 8833Calling KVM_CHECK_EXTENSION for this capability will return a bitmask 8834of hypercalls that can be configured to exit to userspace. 8835Right now, the only such hypercall is KVM_HC_MAP_GPA_RANGE. 8836 8837The argument to KVM_ENABLE_CAP is also a bitmask, and must be a subset 8838of the result of KVM_CHECK_EXTENSION. KVM will forward to userspace 8839the hypercalls whose corresponding bit is in the argument, and return 8840ENOSYS for the others. 8841 88428.35 KVM_CAP_PMU_CAPABILITY 8843--------------------------- 8844 8845:Capability: KVM_CAP_PMU_CAPABILITY 8846:Architectures: x86 8847:Type: vm 8848:Parameters: arg[0] is bitmask of PMU virtualization capabilities. 8849:Returns: 0 on success, -EINVAL when arg[0] contains invalid bits 8850 8851This capability alters PMU virtualization in KVM. 8852 8853Calling KVM_CHECK_EXTENSION for this capability returns a bitmask of 8854PMU virtualization capabilities that can be adjusted on a VM. 8855 8856The argument to KVM_ENABLE_CAP is also a bitmask and selects specific 8857PMU virtualization capabilities to be applied to the VM. This can 8858only be invoked on a VM prior to the creation of VCPUs. 8859 8860At this time, KVM_PMU_CAP_DISABLE is the only capability. Setting 8861this capability will disable PMU virtualization for that VM. Usermode 8862should adjust CPUID leaf 0xA to reflect that the PMU is disabled. 8863 88648.36 KVM_CAP_ARM_SYSTEM_SUSPEND 8865------------------------------- 8866 8867:Capability: KVM_CAP_ARM_SYSTEM_SUSPEND 8868:Architectures: arm64 8869:Type: vm 8870 8871When enabled, KVM will exit to userspace with KVM_EXIT_SYSTEM_EVENT of 8872type KVM_SYSTEM_EVENT_SUSPEND to process the guest suspend request. 8873 88748.37 KVM_CAP_S390_PROTECTED_DUMP 8875-------------------------------- 8876 8877:Capability: KVM_CAP_S390_PROTECTED_DUMP 8878:Architectures: s390 8879:Type: vm 8880 8881This capability indicates that KVM and the Ultravisor support dumping 8882PV guests. The `KVM_PV_DUMP` command is available for the 8883`KVM_S390_PV_COMMAND` ioctl and the `KVM_PV_INFO` command provides 8884dump related UV data. Also the vcpu ioctl `KVM_S390_PV_CPU_COMMAND` is 8885available and supports the `KVM_PV_DUMP_CPU` subcommand. 8886 88878.38 KVM_CAP_VM_DISABLE_NX_HUGE_PAGES 8888------------------------------------- 8889 8890:Capability: KVM_CAP_VM_DISABLE_NX_HUGE_PAGES 8891:Architectures: x86 8892:Type: vm 8893:Parameters: arg[0] must be 0. 8894:Returns: 0 on success, -EPERM if the userspace process does not 8895 have CAP_SYS_BOOT, -EINVAL if args[0] is not 0 or any vCPUs have been 8896 created. 8897 8898This capability disables the NX huge pages mitigation for iTLB MULTIHIT. 8899 8900The capability has no effect if the nx_huge_pages module parameter is not set. 8901 8902This capability may only be set before any vCPUs are created. 8903 89048.39 KVM_CAP_S390_CPU_TOPOLOGY 8905------------------------------ 8906 8907:Capability: KVM_CAP_S390_CPU_TOPOLOGY 8908:Architectures: s390 8909:Type: vm 8910 8911This capability indicates that KVM will provide the S390 CPU Topology 8912facility which consist of the interpretation of the PTF instruction for 8913the function code 2 along with interception and forwarding of both the 8914PTF instruction with function codes 0 or 1 and the STSI(15,1,x) 8915instruction to the userland hypervisor. 8916 8917The stfle facility 11, CPU Topology facility, should not be indicated 8918to the guest without this capability. 8919 8920When this capability is present, KVM provides a new attribute group 8921on vm fd, KVM_S390_VM_CPU_TOPOLOGY. 8922This new attribute allows to get, set or clear the Modified Change 8923Topology Report (MTCR) bit of the SCA through the kvm_device_attr 8924structure. 8925 8926When getting the Modified Change Topology Report value, the attr->addr 8927must point to a byte where the value will be stored or retrieved from. 8928 89298.40 KVM_CAP_ARM_EAGER_SPLIT_CHUNK_SIZE 8930--------------------------------------- 8931 8932:Capability: KVM_CAP_ARM_EAGER_SPLIT_CHUNK_SIZE 8933:Architectures: arm64 8934:Type: vm 8935:Parameters: arg[0] is the new split chunk size. 8936:Returns: 0 on success, -EINVAL if any memslot was already created. 8937 8938This capability sets the chunk size used in Eager Page Splitting. 8939 8940Eager Page Splitting improves the performance of dirty-logging (used 8941in live migrations) when guest memory is backed by huge-pages. It 8942avoids splitting huge-pages (into PAGE_SIZE pages) on fault, by doing 8943it eagerly when enabling dirty logging (with the 8944KVM_MEM_LOG_DIRTY_PAGES flag for a memory region), or when using 8945KVM_CLEAR_DIRTY_LOG. 8946 8947The chunk size specifies how many pages to break at a time, using a 8948single allocation for each chunk. Bigger the chunk size, more pages 8949need to be allocated ahead of time. 8950 8951The chunk size needs to be a valid block size. The list of acceptable 8952block sizes is exposed in KVM_CAP_ARM_SUPPORTED_BLOCK_SIZES as a 895364-bit bitmap (each bit describing a block size). The default value is 89540, to disable the eager page splitting. 8955 89568.41 KVM_CAP_VM_TYPES 8957--------------------- 8958 8959:Capability: KVM_CAP_MEMORY_ATTRIBUTES 8960:Architectures: x86 8961:Type: system ioctl 8962 8963This capability returns a bitmap of support VM types. The 1-setting of bit @n 8964means the VM type with value @n is supported. Possible values of @n are:: 8965 8966 #define KVM_X86_DEFAULT_VM 0 8967 #define KVM_X86_SW_PROTECTED_VM 1 8968 #define KVM_X86_SEV_VM 2 8969 #define KVM_X86_SEV_ES_VM 3 8970 8971Note, KVM_X86_SW_PROTECTED_VM is currently only for development and testing. 8972Do not use KVM_X86_SW_PROTECTED_VM for "real" VMs, and especially not in 8973production. The behavior and effective ABI for software-protected VMs is 8974unstable. 8975 89769. Known KVM API problems 8977========================= 8978 8979In some cases, KVM's API has some inconsistencies or common pitfalls 8980that userspace need to be aware of. This section details some of 8981these issues. 8982 8983Most of them are architecture specific, so the section is split by 8984architecture. 8985 89869.1. x86 8987-------- 8988 8989``KVM_GET_SUPPORTED_CPUID`` issues 8990^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 8991 8992In general, ``KVM_GET_SUPPORTED_CPUID`` is designed so that it is possible 8993to take its result and pass it directly to ``KVM_SET_CPUID2``. This section 8994documents some cases in which that requires some care. 8995 8996Local APIC features 8997~~~~~~~~~~~~~~~~~~~ 8998 8999CPU[EAX=1]:ECX[21] (X2APIC) is reported by ``KVM_GET_SUPPORTED_CPUID``, 9000but it can only be enabled if ``KVM_CREATE_IRQCHIP`` or 9001``KVM_ENABLE_CAP(KVM_CAP_IRQCHIP_SPLIT)`` are used to enable in-kernel emulation of 9002the local APIC. 9003 9004The same is true for the ``KVM_FEATURE_PV_UNHALT`` paravirtualized feature. 9005 9006CPU[EAX=1]:ECX[24] (TSC_DEADLINE) is not reported by ``KVM_GET_SUPPORTED_CPUID``. 9007It can be enabled if ``KVM_CAP_TSC_DEADLINE_TIMER`` is present and the kernel 9008has enabled in-kernel emulation of the local APIC. 9009 9010CPU topology 9011~~~~~~~~~~~~ 9012 9013Several CPUID values include topology information for the host CPU: 90140x0b and 0x1f for Intel systems, 0x8000001e for AMD systems. Different 9015versions of KVM return different values for this information and userspace 9016should not rely on it. Currently they return all zeroes. 9017 9018If userspace wishes to set up a guest topology, it should be careful that 9019the values of these three leaves differ for each CPU. In particular, 9020the APIC ID is found in EDX for all subleaves of 0x0b and 0x1f, and in EAX 9021for 0x8000001e; the latter also encodes the core id and node id in bits 90227:0 of EBX and ECX respectively. 9023 9024Obsolete ioctls and capabilities 9025^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9026 9027KVM_CAP_DISABLE_QUIRKS does not let userspace know which quirks are actually 9028available. Use ``KVM_CHECK_EXTENSION(KVM_CAP_DISABLE_QUIRKS2)`` instead if 9029available. 9030 9031Ordering of KVM_GET_*/KVM_SET_* ioctls 9032^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9033 9034TBD 9035