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- irq_type[0]: 895 out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ 896- irq_type[1]: 897 in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.) 898 (the vcpu_index field is ignored) 899- irq_type[2]: 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 14064.36 KVM_SET_TSS_ADDR 1407--------------------- 1408 1409:Capability: KVM_CAP_SET_TSS_ADDR 1410:Architectures: x86 1411:Type: vm ioctl 1412:Parameters: unsigned long tss_address (in) 1413:Returns: 0 on success, -1 on error 1414 1415This ioctl defines the physical address of a three-page region in the guest 1416physical address space. The region must be within the first 4GB of the 1417guest physical address space and must not conflict with any memory slot 1418or any mmio address. The guest may malfunction if it accesses this memory 1419region. 1420 1421This ioctl is required on Intel-based hosts. This is needed on Intel hardware 1422because of a quirk in the virtualization implementation (see the internals 1423documentation when it pops into existence). 1424 1425 14264.37 KVM_ENABLE_CAP 1427------------------- 1428 1429:Capability: KVM_CAP_ENABLE_CAP 1430:Architectures: mips, ppc, s390, x86, loongarch 1431:Type: vcpu ioctl 1432:Parameters: struct kvm_enable_cap (in) 1433:Returns: 0 on success; -1 on error 1434 1435:Capability: KVM_CAP_ENABLE_CAP_VM 1436:Architectures: all 1437:Type: vm ioctl 1438:Parameters: struct kvm_enable_cap (in) 1439:Returns: 0 on success; -1 on error 1440 1441.. note:: 1442 1443 Not all extensions are enabled by default. Using this ioctl the application 1444 can enable an extension, making it available to the guest. 1445 1446On systems that do not support this ioctl, it always fails. On systems that 1447do support it, it only works for extensions that are supported for enablement. 1448 1449To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should 1450be used. 1451 1452:: 1453 1454 struct kvm_enable_cap { 1455 /* in */ 1456 __u32 cap; 1457 1458The capability that is supposed to get enabled. 1459 1460:: 1461 1462 __u32 flags; 1463 1464A bitfield indicating future enhancements. Has to be 0 for now. 1465 1466:: 1467 1468 __u64 args[4]; 1469 1470Arguments for enabling a feature. If a feature needs initial values to 1471function properly, this is the place to put them. 1472 1473:: 1474 1475 __u8 pad[64]; 1476 }; 1477 1478The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl 1479for vm-wide capabilities. 1480 14814.38 KVM_GET_MP_STATE 1482--------------------- 1483 1484:Capability: KVM_CAP_MP_STATE 1485:Architectures: x86, s390, arm64, riscv, loongarch 1486:Type: vcpu ioctl 1487:Parameters: struct kvm_mp_state (out) 1488:Returns: 0 on success; -1 on error 1489 1490:: 1491 1492 struct kvm_mp_state { 1493 __u32 mp_state; 1494 }; 1495 1496Returns the vcpu's current "multiprocessing state" (though also valid on 1497uniprocessor guests). 1498 1499Possible values are: 1500 1501 ========================== =============================================== 1502 KVM_MP_STATE_RUNNABLE the vcpu is currently running 1503 [x86,arm64,riscv,loongarch] 1504 KVM_MP_STATE_UNINITIALIZED the vcpu is an application processor (AP) 1505 which has not yet received an INIT signal [x86] 1506 KVM_MP_STATE_INIT_RECEIVED the vcpu has received an INIT signal, and is 1507 now ready for a SIPI [x86] 1508 KVM_MP_STATE_HALTED the vcpu has executed a HLT instruction and 1509 is waiting for an interrupt [x86] 1510 KVM_MP_STATE_SIPI_RECEIVED the vcpu has just received a SIPI (vector 1511 accessible via KVM_GET_VCPU_EVENTS) [x86] 1512 KVM_MP_STATE_STOPPED the vcpu is stopped [s390,arm64,riscv] 1513 KVM_MP_STATE_CHECK_STOP the vcpu is in a special error state [s390] 1514 KVM_MP_STATE_OPERATING the vcpu is operating (running or halted) 1515 [s390] 1516 KVM_MP_STATE_LOAD the vcpu is in a special load/startup state 1517 [s390] 1518 KVM_MP_STATE_SUSPENDED the vcpu is in a suspend state and is waiting 1519 for a wakeup event [arm64] 1520 ========================== =============================================== 1521 1522On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an 1523in-kernel irqchip, the multiprocessing state must be maintained by userspace on 1524these architectures. 1525 1526For arm64: 1527^^^^^^^^^^ 1528 1529If a vCPU is in the KVM_MP_STATE_SUSPENDED state, KVM will emulate the 1530architectural execution of a WFI instruction. 1531 1532If a wakeup event is recognized, KVM will exit to userspace with a 1533KVM_SYSTEM_EVENT exit, where the event type is KVM_SYSTEM_EVENT_WAKEUP. If 1534userspace wants to honor the wakeup, it must set the vCPU's MP state to 1535KVM_MP_STATE_RUNNABLE. If it does not, KVM will continue to await a wakeup 1536event in subsequent calls to KVM_RUN. 1537 1538.. warning:: 1539 1540 If userspace intends to keep the vCPU in a SUSPENDED state, it is 1541 strongly recommended that userspace take action to suppress the 1542 wakeup event (such as masking an interrupt). Otherwise, subsequent 1543 calls to KVM_RUN will immediately exit with a KVM_SYSTEM_EVENT_WAKEUP 1544 event and inadvertently waste CPU cycles. 1545 1546 Additionally, if userspace takes action to suppress a wakeup event, 1547 it is strongly recommended that it also restores the vCPU to its 1548 original state when the vCPU is made RUNNABLE again. For example, 1549 if userspace masked a pending interrupt to suppress the wakeup, 1550 the interrupt should be unmasked before returning control to the 1551 guest. 1552 1553For riscv: 1554^^^^^^^^^^ 1555 1556The only states that are valid are KVM_MP_STATE_STOPPED and 1557KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not. 1558 1559On LoongArch, only the KVM_MP_STATE_RUNNABLE state is used to reflect 1560whether the vcpu is runnable. 1561 15624.39 KVM_SET_MP_STATE 1563--------------------- 1564 1565:Capability: KVM_CAP_MP_STATE 1566:Architectures: x86, s390, arm64, riscv, loongarch 1567:Type: vcpu ioctl 1568:Parameters: struct kvm_mp_state (in) 1569:Returns: 0 on success; -1 on error 1570 1571Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for 1572arguments. 1573 1574On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an 1575in-kernel irqchip, the multiprocessing state must be maintained by userspace on 1576these architectures. 1577 1578For arm64/riscv: 1579^^^^^^^^^^^^^^^^ 1580 1581The only states that are valid are KVM_MP_STATE_STOPPED and 1582KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not. 1583 1584On LoongArch, only the KVM_MP_STATE_RUNNABLE state is used to reflect 1585whether the vcpu is runnable. 1586 15874.40 KVM_SET_IDENTITY_MAP_ADDR 1588------------------------------ 1589 1590:Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR 1591:Architectures: x86 1592:Type: vm ioctl 1593:Parameters: unsigned long identity (in) 1594:Returns: 0 on success, -1 on error 1595 1596This ioctl defines the physical address of a one-page region in the guest 1597physical address space. The region must be within the first 4GB of the 1598guest physical address space and must not conflict with any memory slot 1599or any mmio address. The guest may malfunction if it accesses this memory 1600region. 1601 1602Setting the address to 0 will result in resetting the address to its default 1603(0xfffbc000). 1604 1605This ioctl is required on Intel-based hosts. This is needed on Intel hardware 1606because of a quirk in the virtualization implementation (see the internals 1607documentation when it pops into existence). 1608 1609Fails if any VCPU has already been created. 1610 16114.41 KVM_SET_BOOT_CPU_ID 1612------------------------ 1613 1614:Capability: KVM_CAP_SET_BOOT_CPU_ID 1615:Architectures: x86 1616:Type: vm ioctl 1617:Parameters: unsigned long vcpu_id 1618:Returns: 0 on success, -1 on error 1619 1620Define which vcpu is the Bootstrap Processor (BSP). Values are the same 1621as the vcpu id in KVM_CREATE_VCPU. If this ioctl is not called, the default 1622is vcpu 0. This ioctl has to be called before vcpu creation, 1623otherwise it will return EBUSY error. 1624 1625 16264.42 KVM_GET_XSAVE 1627------------------ 1628 1629:Capability: KVM_CAP_XSAVE 1630:Architectures: x86 1631:Type: vcpu ioctl 1632:Parameters: struct kvm_xsave (out) 1633:Returns: 0 on success, -1 on error 1634 1635 1636:: 1637 1638 struct kvm_xsave { 1639 __u32 region[1024]; 1640 __u32 extra[0]; 1641 }; 1642 1643This ioctl would copy current vcpu's xsave struct to the userspace. 1644 1645 16464.43 KVM_SET_XSAVE 1647------------------ 1648 1649:Capability: KVM_CAP_XSAVE and KVM_CAP_XSAVE2 1650:Architectures: x86 1651:Type: vcpu ioctl 1652:Parameters: struct kvm_xsave (in) 1653:Returns: 0 on success, -1 on error 1654 1655:: 1656 1657 1658 struct kvm_xsave { 1659 __u32 region[1024]; 1660 __u32 extra[0]; 1661 }; 1662 1663This ioctl would copy userspace's xsave struct to the kernel. It copies 1664as many bytes as are returned by KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2), 1665when invoked on the vm file descriptor. The size value returned by 1666KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) will always be at least 4096. 1667Currently, it is only greater than 4096 if a dynamic feature has been 1668enabled with ``arch_prctl()``, but this may change in the future. 1669 1670The offsets of the state save areas in struct kvm_xsave follow the 1671contents of CPUID leaf 0xD on the host. 1672 1673 16744.44 KVM_GET_XCRS 1675----------------- 1676 1677:Capability: KVM_CAP_XCRS 1678:Architectures: x86 1679:Type: vcpu ioctl 1680:Parameters: struct kvm_xcrs (out) 1681:Returns: 0 on success, -1 on error 1682 1683:: 1684 1685 struct kvm_xcr { 1686 __u32 xcr; 1687 __u32 reserved; 1688 __u64 value; 1689 }; 1690 1691 struct kvm_xcrs { 1692 __u32 nr_xcrs; 1693 __u32 flags; 1694 struct kvm_xcr xcrs[KVM_MAX_XCRS]; 1695 __u64 padding[16]; 1696 }; 1697 1698This ioctl would copy current vcpu's xcrs to the userspace. 1699 1700 17014.45 KVM_SET_XCRS 1702----------------- 1703 1704:Capability: KVM_CAP_XCRS 1705:Architectures: x86 1706:Type: vcpu ioctl 1707:Parameters: struct kvm_xcrs (in) 1708:Returns: 0 on success, -1 on error 1709 1710:: 1711 1712 struct kvm_xcr { 1713 __u32 xcr; 1714 __u32 reserved; 1715 __u64 value; 1716 }; 1717 1718 struct kvm_xcrs { 1719 __u32 nr_xcrs; 1720 __u32 flags; 1721 struct kvm_xcr xcrs[KVM_MAX_XCRS]; 1722 __u64 padding[16]; 1723 }; 1724 1725This ioctl would set vcpu's xcr to the value userspace specified. 1726 1727 17284.46 KVM_GET_SUPPORTED_CPUID 1729---------------------------- 1730 1731:Capability: KVM_CAP_EXT_CPUID 1732:Architectures: x86 1733:Type: system ioctl 1734:Parameters: struct kvm_cpuid2 (in/out) 1735:Returns: 0 on success, -1 on error 1736 1737:: 1738 1739 struct kvm_cpuid2 { 1740 __u32 nent; 1741 __u32 padding; 1742 struct kvm_cpuid_entry2 entries[0]; 1743 }; 1744 1745 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0) 1746 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */ 1747 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */ 1748 1749 struct kvm_cpuid_entry2 { 1750 __u32 function; 1751 __u32 index; 1752 __u32 flags; 1753 __u32 eax; 1754 __u32 ebx; 1755 __u32 ecx; 1756 __u32 edx; 1757 __u32 padding[3]; 1758 }; 1759 1760This ioctl returns x86 cpuid features which are supported by both the 1761hardware and kvm in its default configuration. Userspace can use the 1762information returned by this ioctl to construct cpuid information (for 1763KVM_SET_CPUID2) that is consistent with hardware, kernel, and 1764userspace capabilities, and with user requirements (for example, the 1765user may wish to constrain cpuid to emulate older hardware, or for 1766feature consistency across a cluster). 1767 1768Dynamically-enabled feature bits need to be requested with 1769``arch_prctl()`` before calling this ioctl. Feature bits that have not 1770been requested are excluded from the result. 1771 1772Note that certain capabilities, such as KVM_CAP_X86_DISABLE_EXITS, may 1773expose cpuid features (e.g. MONITOR) which are not supported by kvm in 1774its default configuration. If userspace enables such capabilities, it 1775is responsible for modifying the results of this ioctl appropriately. 1776 1777Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure 1778with the 'nent' field indicating the number of entries in the variable-size 1779array 'entries'. If the number of entries is too low to describe the cpu 1780capabilities, an error (E2BIG) is returned. If the number is too high, 1781the 'nent' field is adjusted and an error (ENOMEM) is returned. If the 1782number is just right, the 'nent' field is adjusted to the number of valid 1783entries in the 'entries' array, which is then filled. 1784 1785The entries returned are the host cpuid as returned by the cpuid instruction, 1786with unknown or unsupported features masked out. Some features (for example, 1787x2apic), may not be present in the host cpu, but are exposed by kvm if it can 1788emulate them efficiently. The fields in each entry are defined as follows: 1789 1790 function: 1791 the eax value used to obtain the entry 1792 1793 index: 1794 the ecx value used to obtain the entry (for entries that are 1795 affected by ecx) 1796 1797 flags: 1798 an OR of zero or more of the following: 1799 1800 KVM_CPUID_FLAG_SIGNIFCANT_INDEX: 1801 if the index field is valid 1802 1803 eax, ebx, ecx, edx: 1804 the values returned by the cpuid instruction for 1805 this function/index combination 1806 1807The TSC deadline timer feature (CPUID leaf 1, ecx[24]) is always returned 1808as false, since the feature depends on KVM_CREATE_IRQCHIP for local APIC 1809support. Instead it is reported via:: 1810 1811 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER) 1812 1813if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the 1814feature in userspace, then you can enable the feature for KVM_SET_CPUID2. 1815 1816 18174.47 KVM_PPC_GET_PVINFO 1818----------------------- 1819 1820:Capability: KVM_CAP_PPC_GET_PVINFO 1821:Architectures: ppc 1822:Type: vm ioctl 1823:Parameters: struct kvm_ppc_pvinfo (out) 1824:Returns: 0 on success, !0 on error 1825 1826:: 1827 1828 struct kvm_ppc_pvinfo { 1829 __u32 flags; 1830 __u32 hcall[4]; 1831 __u8 pad[108]; 1832 }; 1833 1834This ioctl fetches PV specific information that need to be passed to the guest 1835using the device tree or other means from vm context. 1836 1837The hcall array defines 4 instructions that make up a hypercall. 1838 1839If any additional field gets added to this structure later on, a bit for that 1840additional piece of information will be set in the flags bitmap. 1841 1842The flags bitmap is defined as:: 1843 1844 /* the host supports the ePAPR idle hcall 1845 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1<<0) 1846 18474.52 KVM_SET_GSI_ROUTING 1848------------------------ 1849 1850:Capability: KVM_CAP_IRQ_ROUTING 1851:Architectures: x86 s390 arm64 1852:Type: vm ioctl 1853:Parameters: struct kvm_irq_routing (in) 1854:Returns: 0 on success, -1 on error 1855 1856Sets the GSI routing table entries, overwriting any previously set entries. 1857 1858On arm64, GSI routing has the following limitation: 1859 1860- GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD. 1861 1862:: 1863 1864 struct kvm_irq_routing { 1865 __u32 nr; 1866 __u32 flags; 1867 struct kvm_irq_routing_entry entries[0]; 1868 }; 1869 1870No flags are specified so far, the corresponding field must be set to zero. 1871 1872:: 1873 1874 struct kvm_irq_routing_entry { 1875 __u32 gsi; 1876 __u32 type; 1877 __u32 flags; 1878 __u32 pad; 1879 union { 1880 struct kvm_irq_routing_irqchip irqchip; 1881 struct kvm_irq_routing_msi msi; 1882 struct kvm_irq_routing_s390_adapter adapter; 1883 struct kvm_irq_routing_hv_sint hv_sint; 1884 struct kvm_irq_routing_xen_evtchn xen_evtchn; 1885 __u32 pad[8]; 1886 } u; 1887 }; 1888 1889 /* gsi routing entry types */ 1890 #define KVM_IRQ_ROUTING_IRQCHIP 1 1891 #define KVM_IRQ_ROUTING_MSI 2 1892 #define KVM_IRQ_ROUTING_S390_ADAPTER 3 1893 #define KVM_IRQ_ROUTING_HV_SINT 4 1894 #define KVM_IRQ_ROUTING_XEN_EVTCHN 5 1895 1896flags: 1897 1898- KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry 1899 type, specifies that the devid field contains a valid value. The per-VM 1900 KVM_CAP_MSI_DEVID capability advertises the requirement to provide 1901 the device ID. If this capability is not available, userspace should 1902 never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail. 1903- zero otherwise 1904 1905:: 1906 1907 struct kvm_irq_routing_irqchip { 1908 __u32 irqchip; 1909 __u32 pin; 1910 }; 1911 1912 struct kvm_irq_routing_msi { 1913 __u32 address_lo; 1914 __u32 address_hi; 1915 __u32 data; 1916 union { 1917 __u32 pad; 1918 __u32 devid; 1919 }; 1920 }; 1921 1922If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier 1923for the device that wrote the MSI message. For PCI, this is usually a 1924BFD identifier in the lower 16 bits. 1925 1926On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS 1927feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled, 1928address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of 1929address_hi must be zero. 1930 1931:: 1932 1933 struct kvm_irq_routing_s390_adapter { 1934 __u64 ind_addr; 1935 __u64 summary_addr; 1936 __u64 ind_offset; 1937 __u32 summary_offset; 1938 __u32 adapter_id; 1939 }; 1940 1941 struct kvm_irq_routing_hv_sint { 1942 __u32 vcpu; 1943 __u32 sint; 1944 }; 1945 1946 struct kvm_irq_routing_xen_evtchn { 1947 __u32 port; 1948 __u32 vcpu; 1949 __u32 priority; 1950 }; 1951 1952 1953When KVM_CAP_XEN_HVM includes the KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL bit 1954in its indication of supported features, routing to Xen event channels 1955is supported. Although the priority field is present, only the value 1956KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL is supported, which means delivery by 19572 level event channels. FIFO event channel support may be added in 1958the future. 1959 1960 19614.55 KVM_SET_TSC_KHZ 1962-------------------- 1963 1964:Capability: KVM_CAP_TSC_CONTROL / KVM_CAP_VM_TSC_CONTROL 1965:Architectures: x86 1966:Type: vcpu ioctl / vm ioctl 1967:Parameters: virtual tsc_khz 1968:Returns: 0 on success, -1 on error 1969 1970Specifies the tsc frequency for the virtual machine. The unit of the 1971frequency is KHz. 1972 1973If the KVM_CAP_VM_TSC_CONTROL capability is advertised, this can also 1974be used as a vm ioctl to set the initial tsc frequency of subsequently 1975created vCPUs. 1976 19774.56 KVM_GET_TSC_KHZ 1978-------------------- 1979 1980:Capability: KVM_CAP_GET_TSC_KHZ / KVM_CAP_VM_TSC_CONTROL 1981:Architectures: x86 1982:Type: vcpu ioctl / vm ioctl 1983:Parameters: none 1984:Returns: virtual tsc-khz on success, negative value on error 1985 1986Returns the tsc frequency of the guest. The unit of the return value is 1987KHz. If the host has unstable tsc this ioctl returns -EIO instead as an 1988error. 1989 1990 19914.57 KVM_GET_LAPIC 1992------------------ 1993 1994:Capability: KVM_CAP_IRQCHIP 1995:Architectures: x86 1996:Type: vcpu ioctl 1997:Parameters: struct kvm_lapic_state (out) 1998:Returns: 0 on success, -1 on error 1999 2000:: 2001 2002 #define KVM_APIC_REG_SIZE 0x400 2003 struct kvm_lapic_state { 2004 char regs[KVM_APIC_REG_SIZE]; 2005 }; 2006 2007Reads the Local APIC registers and copies them into the input argument. The 2008data format and layout are the same as documented in the architecture manual. 2009 2010If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is 2011enabled, then the format of APIC_ID register depends on the APIC mode 2012(reported by MSR_IA32_APICBASE) of its VCPU. x2APIC stores APIC ID in 2013the APIC_ID register (bytes 32-35). xAPIC only allows an 8-bit APIC ID 2014which is stored in bits 31-24 of the APIC register, or equivalently in 2015byte 35 of struct kvm_lapic_state's regs field. KVM_GET_LAPIC must then 2016be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR. 2017 2018If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state 2019always uses xAPIC format. 2020 2021 20224.58 KVM_SET_LAPIC 2023------------------ 2024 2025:Capability: KVM_CAP_IRQCHIP 2026:Architectures: x86 2027:Type: vcpu ioctl 2028:Parameters: struct kvm_lapic_state (in) 2029:Returns: 0 on success, -1 on error 2030 2031:: 2032 2033 #define KVM_APIC_REG_SIZE 0x400 2034 struct kvm_lapic_state { 2035 char regs[KVM_APIC_REG_SIZE]; 2036 }; 2037 2038Copies the input argument into the Local APIC registers. The data format 2039and layout are the same as documented in the architecture manual. 2040 2041The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's 2042regs field) depends on the state of the KVM_CAP_X2APIC_API capability. 2043See the note in KVM_GET_LAPIC. 2044 2045 20464.59 KVM_IOEVENTFD 2047------------------ 2048 2049:Capability: KVM_CAP_IOEVENTFD 2050:Architectures: all 2051:Type: vm ioctl 2052:Parameters: struct kvm_ioeventfd (in) 2053:Returns: 0 on success, !0 on error 2054 2055This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address 2056within the guest. A guest write in the registered address will signal the 2057provided event instead of triggering an exit. 2058 2059:: 2060 2061 struct kvm_ioeventfd { 2062 __u64 datamatch; 2063 __u64 addr; /* legal pio/mmio address */ 2064 __u32 len; /* 0, 1, 2, 4, or 8 bytes */ 2065 __s32 fd; 2066 __u32 flags; 2067 __u8 pad[36]; 2068 }; 2069 2070For the special case of virtio-ccw devices on s390, the ioevent is matched 2071to a subchannel/virtqueue tuple instead. 2072 2073The following flags are defined:: 2074 2075 #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch) 2076 #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio) 2077 #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign) 2078 #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \ 2079 (1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify) 2080 2081If datamatch flag is set, the event will be signaled only if the written value 2082to the registered address is equal to datamatch in struct kvm_ioeventfd. 2083 2084For virtio-ccw devices, addr contains the subchannel id and datamatch the 2085virtqueue index. 2086 2087With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and 2088the kernel will ignore the length of guest write and may get a faster vmexit. 2089The speedup may only apply to specific architectures, but the ioeventfd will 2090work anyway. 2091 20924.60 KVM_DIRTY_TLB 2093------------------ 2094 2095:Capability: KVM_CAP_SW_TLB 2096:Architectures: ppc 2097:Type: vcpu ioctl 2098:Parameters: struct kvm_dirty_tlb (in) 2099:Returns: 0 on success, -1 on error 2100 2101:: 2102 2103 struct kvm_dirty_tlb { 2104 __u64 bitmap; 2105 __u32 num_dirty; 2106 }; 2107 2108This must be called whenever userspace has changed an entry in the shared 2109TLB, prior to calling KVM_RUN on the associated vcpu. 2110 2111The "bitmap" field is the userspace address of an array. This array 2112consists of a number of bits, equal to the total number of TLB entries as 2113determined by the last successful call to KVM_CONFIG_TLB, rounded up to the 2114nearest multiple of 64. 2115 2116Each bit corresponds to one TLB entry, ordered the same as in the shared TLB 2117array. 2118 2119The array is little-endian: the bit 0 is the least significant bit of the 2120first byte, bit 8 is the least significant bit of the second byte, etc. 2121This avoids any complications with differing word sizes. 2122 2123The "num_dirty" field is a performance hint for KVM to determine whether it 2124should skip processing the bitmap and just invalidate everything. It must 2125be set to the number of set bits in the bitmap. 2126 2127 21284.62 KVM_CREATE_SPAPR_TCE 2129------------------------- 2130 2131:Capability: KVM_CAP_SPAPR_TCE 2132:Architectures: powerpc 2133:Type: vm ioctl 2134:Parameters: struct kvm_create_spapr_tce (in) 2135:Returns: file descriptor for manipulating the created TCE table 2136 2137This creates a virtual TCE (translation control entry) table, which 2138is an IOMMU for PAPR-style virtual I/O. It is used to translate 2139logical addresses used in virtual I/O into guest physical addresses, 2140and provides a scatter/gather capability for PAPR virtual I/O. 2141 2142:: 2143 2144 /* for KVM_CAP_SPAPR_TCE */ 2145 struct kvm_create_spapr_tce { 2146 __u64 liobn; 2147 __u32 window_size; 2148 }; 2149 2150The liobn field gives the logical IO bus number for which to create a 2151TCE table. The window_size field specifies the size of the DMA window 2152which this TCE table will translate - the table will contain one 64 2153bit TCE entry for every 4kiB of the DMA window. 2154 2155When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE 2156table has been created using this ioctl(), the kernel will handle it 2157in real mode, updating the TCE table. H_PUT_TCE calls for other 2158liobns will cause a vm exit and must be handled by userspace. 2159 2160The return value is a file descriptor which can be passed to mmap(2) 2161to map the created TCE table into userspace. This lets userspace read 2162the entries written by kernel-handled H_PUT_TCE calls, and also lets 2163userspace update the TCE table directly which is useful in some 2164circumstances. 2165 2166 21674.63 KVM_ALLOCATE_RMA 2168--------------------- 2169 2170:Capability: KVM_CAP_PPC_RMA 2171:Architectures: powerpc 2172:Type: vm ioctl 2173:Parameters: struct kvm_allocate_rma (out) 2174:Returns: file descriptor for mapping the allocated RMA 2175 2176This allocates a Real Mode Area (RMA) from the pool allocated at boot 2177time by the kernel. An RMA is a physically-contiguous, aligned region 2178of memory used on older POWER processors to provide the memory which 2179will be accessed by real-mode (MMU off) accesses in a KVM guest. 2180POWER processors support a set of sizes for the RMA that usually 2181includes 64MB, 128MB, 256MB and some larger powers of two. 2182 2183:: 2184 2185 /* for KVM_ALLOCATE_RMA */ 2186 struct kvm_allocate_rma { 2187 __u64 rma_size; 2188 }; 2189 2190The return value is a file descriptor which can be passed to mmap(2) 2191to map the allocated RMA into userspace. The mapped area can then be 2192passed to the KVM_SET_USER_MEMORY_REGION ioctl to establish it as the 2193RMA for a virtual machine. The size of the RMA in bytes (which is 2194fixed at host kernel boot time) is returned in the rma_size field of 2195the argument structure. 2196 2197The KVM_CAP_PPC_RMA capability is 1 or 2 if the KVM_ALLOCATE_RMA ioctl 2198is supported; 2 if the processor requires all virtual machines to have 2199an RMA, or 1 if the processor can use an RMA but doesn't require it, 2200because it supports the Virtual RMA (VRMA) facility. 2201 2202 22034.64 KVM_NMI 2204------------ 2205 2206:Capability: KVM_CAP_USER_NMI 2207:Architectures: x86 2208:Type: vcpu ioctl 2209:Parameters: none 2210:Returns: 0 on success, -1 on error 2211 2212Queues an NMI on the thread's vcpu. Note this is well defined only 2213when KVM_CREATE_IRQCHIP has not been called, since this is an interface 2214between the virtual cpu core and virtual local APIC. After KVM_CREATE_IRQCHIP 2215has been called, this interface is completely emulated within the kernel. 2216 2217To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the 2218following algorithm: 2219 2220 - pause the vcpu 2221 - read the local APIC's state (KVM_GET_LAPIC) 2222 - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1) 2223 - if so, issue KVM_NMI 2224 - resume the vcpu 2225 2226Some guests configure the LINT1 NMI input to cause a panic, aiding in 2227debugging. 2228 2229 22304.65 KVM_S390_UCAS_MAP 2231---------------------- 2232 2233:Capability: KVM_CAP_S390_UCONTROL 2234:Architectures: s390 2235:Type: vcpu ioctl 2236:Parameters: struct kvm_s390_ucas_mapping (in) 2237:Returns: 0 in case of success 2238 2239The parameter is defined like this:: 2240 2241 struct kvm_s390_ucas_mapping { 2242 __u64 user_addr; 2243 __u64 vcpu_addr; 2244 __u64 length; 2245 }; 2246 2247This ioctl maps the memory at "user_addr" with the length "length" to 2248the vcpu's address space starting at "vcpu_addr". All parameters need to 2249be aligned by 1 megabyte. 2250 2251 22524.66 KVM_S390_UCAS_UNMAP 2253------------------------ 2254 2255:Capability: KVM_CAP_S390_UCONTROL 2256:Architectures: s390 2257:Type: vcpu ioctl 2258:Parameters: struct kvm_s390_ucas_mapping (in) 2259:Returns: 0 in case of success 2260 2261The parameter is defined like this:: 2262 2263 struct kvm_s390_ucas_mapping { 2264 __u64 user_addr; 2265 __u64 vcpu_addr; 2266 __u64 length; 2267 }; 2268 2269This ioctl unmaps the memory in the vcpu's address space starting at 2270"vcpu_addr" with the length "length". The field "user_addr" is ignored. 2271All parameters need to be aligned by 1 megabyte. 2272 2273 22744.67 KVM_S390_VCPU_FAULT 2275------------------------ 2276 2277:Capability: KVM_CAP_S390_UCONTROL 2278:Architectures: s390 2279:Type: vcpu ioctl 2280:Parameters: vcpu absolute address (in) 2281:Returns: 0 in case of success 2282 2283This call creates a page table entry on the virtual cpu's address space 2284(for user controlled virtual machines) or the virtual machine's address 2285space (for regular virtual machines). This only works for minor faults, 2286thus it's recommended to access subject memory page via the user page 2287table upfront. This is useful to handle validity intercepts for user 2288controlled virtual machines to fault in the virtual cpu's lowcore pages 2289prior to calling the KVM_RUN ioctl. 2290 2291 22924.68 KVM_SET_ONE_REG 2293-------------------- 2294 2295:Capability: KVM_CAP_ONE_REG 2296:Architectures: all 2297:Type: vcpu ioctl 2298:Parameters: struct kvm_one_reg (in) 2299:Returns: 0 on success, negative value on failure 2300 2301Errors: 2302 2303 ====== ============================================================ 2304 ENOENT no such register 2305 EINVAL invalid register ID, or no such register or used with VMs in 2306 protected virtualization mode on s390 2307 EPERM (arm64) register access not allowed before vcpu finalization 2308 EBUSY (riscv) changing register value not allowed after the vcpu 2309 has run at least once 2310 ====== ============================================================ 2311 2312(These error codes are indicative only: do not rely on a specific error 2313code being returned in a specific situation.) 2314 2315:: 2316 2317 struct kvm_one_reg { 2318 __u64 id; 2319 __u64 addr; 2320 }; 2321 2322Using this ioctl, a single vcpu register can be set to a specific value 2323defined by user space with the passed in struct kvm_one_reg, where id 2324refers to the register identifier as described below and addr is a pointer 2325to a variable with the respective size. There can be architecture agnostic 2326and architecture specific registers. Each have their own range of operation 2327and their own constants and width. To keep track of the implemented 2328registers, find a list below: 2329 2330 ======= =============================== ============ 2331 Arch Register Width (bits) 2332 ======= =============================== ============ 2333 PPC KVM_REG_PPC_HIOR 64 2334 PPC KVM_REG_PPC_IAC1 64 2335 PPC KVM_REG_PPC_IAC2 64 2336 PPC KVM_REG_PPC_IAC3 64 2337 PPC KVM_REG_PPC_IAC4 64 2338 PPC KVM_REG_PPC_DAC1 64 2339 PPC KVM_REG_PPC_DAC2 64 2340 PPC KVM_REG_PPC_DABR 64 2341 PPC KVM_REG_PPC_DSCR 64 2342 PPC KVM_REG_PPC_PURR 64 2343 PPC KVM_REG_PPC_SPURR 64 2344 PPC KVM_REG_PPC_DAR 64 2345 PPC KVM_REG_PPC_DSISR 32 2346 PPC KVM_REG_PPC_AMR 64 2347 PPC KVM_REG_PPC_UAMOR 64 2348 PPC KVM_REG_PPC_MMCR0 64 2349 PPC KVM_REG_PPC_MMCR1 64 2350 PPC KVM_REG_PPC_MMCRA 64 2351 PPC KVM_REG_PPC_MMCR2 64 2352 PPC KVM_REG_PPC_MMCRS 64 2353 PPC KVM_REG_PPC_MMCR3 64 2354 PPC KVM_REG_PPC_SIAR 64 2355 PPC KVM_REG_PPC_SDAR 64 2356 PPC KVM_REG_PPC_SIER 64 2357 PPC KVM_REG_PPC_SIER2 64 2358 PPC KVM_REG_PPC_SIER3 64 2359 PPC KVM_REG_PPC_PMC1 32 2360 PPC KVM_REG_PPC_PMC2 32 2361 PPC KVM_REG_PPC_PMC3 32 2362 PPC KVM_REG_PPC_PMC4 32 2363 PPC KVM_REG_PPC_PMC5 32 2364 PPC KVM_REG_PPC_PMC6 32 2365 PPC KVM_REG_PPC_PMC7 32 2366 PPC KVM_REG_PPC_PMC8 32 2367 PPC KVM_REG_PPC_FPR0 64 2368 ... 2369 PPC KVM_REG_PPC_FPR31 64 2370 PPC KVM_REG_PPC_VR0 128 2371 ... 2372 PPC KVM_REG_PPC_VR31 128 2373 PPC KVM_REG_PPC_VSR0 128 2374 ... 2375 PPC KVM_REG_PPC_VSR31 128 2376 PPC KVM_REG_PPC_FPSCR 64 2377 PPC KVM_REG_PPC_VSCR 32 2378 PPC KVM_REG_PPC_VPA_ADDR 64 2379 PPC KVM_REG_PPC_VPA_SLB 128 2380 PPC KVM_REG_PPC_VPA_DTL 128 2381 PPC KVM_REG_PPC_EPCR 32 2382 PPC KVM_REG_PPC_EPR 32 2383 PPC KVM_REG_PPC_TCR 32 2384 PPC KVM_REG_PPC_TSR 32 2385 PPC KVM_REG_PPC_OR_TSR 32 2386 PPC KVM_REG_PPC_CLEAR_TSR 32 2387 PPC KVM_REG_PPC_MAS0 32 2388 PPC KVM_REG_PPC_MAS1 32 2389 PPC KVM_REG_PPC_MAS2 64 2390 PPC KVM_REG_PPC_MAS7_3 64 2391 PPC KVM_REG_PPC_MAS4 32 2392 PPC KVM_REG_PPC_MAS6 32 2393 PPC KVM_REG_PPC_MMUCFG 32 2394 PPC KVM_REG_PPC_TLB0CFG 32 2395 PPC KVM_REG_PPC_TLB1CFG 32 2396 PPC KVM_REG_PPC_TLB2CFG 32 2397 PPC KVM_REG_PPC_TLB3CFG 32 2398 PPC KVM_REG_PPC_TLB0PS 32 2399 PPC KVM_REG_PPC_TLB1PS 32 2400 PPC KVM_REG_PPC_TLB2PS 32 2401 PPC KVM_REG_PPC_TLB3PS 32 2402 PPC KVM_REG_PPC_EPTCFG 32 2403 PPC KVM_REG_PPC_ICP_STATE 64 2404 PPC KVM_REG_PPC_VP_STATE 128 2405 PPC KVM_REG_PPC_TB_OFFSET 64 2406 PPC KVM_REG_PPC_SPMC1 32 2407 PPC KVM_REG_PPC_SPMC2 32 2408 PPC KVM_REG_PPC_IAMR 64 2409 PPC KVM_REG_PPC_TFHAR 64 2410 PPC KVM_REG_PPC_TFIAR 64 2411 PPC KVM_REG_PPC_TEXASR 64 2412 PPC KVM_REG_PPC_FSCR 64 2413 PPC KVM_REG_PPC_PSPB 32 2414 PPC KVM_REG_PPC_EBBHR 64 2415 PPC KVM_REG_PPC_EBBRR 64 2416 PPC KVM_REG_PPC_BESCR 64 2417 PPC KVM_REG_PPC_TAR 64 2418 PPC KVM_REG_PPC_DPDES 64 2419 PPC KVM_REG_PPC_DAWR 64 2420 PPC KVM_REG_PPC_DAWRX 64 2421 PPC KVM_REG_PPC_CIABR 64 2422 PPC KVM_REG_PPC_IC 64 2423 PPC KVM_REG_PPC_VTB 64 2424 PPC KVM_REG_PPC_CSIGR 64 2425 PPC KVM_REG_PPC_TACR 64 2426 PPC KVM_REG_PPC_TCSCR 64 2427 PPC KVM_REG_PPC_PID 64 2428 PPC KVM_REG_PPC_ACOP 64 2429 PPC KVM_REG_PPC_VRSAVE 32 2430 PPC KVM_REG_PPC_LPCR 32 2431 PPC KVM_REG_PPC_LPCR_64 64 2432 PPC KVM_REG_PPC_PPR 64 2433 PPC KVM_REG_PPC_ARCH_COMPAT 32 2434 PPC KVM_REG_PPC_DABRX 32 2435 PPC KVM_REG_PPC_WORT 64 2436 PPC KVM_REG_PPC_SPRG9 64 2437 PPC KVM_REG_PPC_DBSR 32 2438 PPC KVM_REG_PPC_TIDR 64 2439 PPC KVM_REG_PPC_PSSCR 64 2440 PPC KVM_REG_PPC_DEC_EXPIRY 64 2441 PPC KVM_REG_PPC_PTCR 64 2442 PPC KVM_REG_PPC_DAWR1 64 2443 PPC KVM_REG_PPC_DAWRX1 64 2444 PPC KVM_REG_PPC_TM_GPR0 64 2445 ... 2446 PPC KVM_REG_PPC_TM_GPR31 64 2447 PPC KVM_REG_PPC_TM_VSR0 128 2448 ... 2449 PPC KVM_REG_PPC_TM_VSR63 128 2450 PPC KVM_REG_PPC_TM_CR 64 2451 PPC KVM_REG_PPC_TM_LR 64 2452 PPC KVM_REG_PPC_TM_CTR 64 2453 PPC KVM_REG_PPC_TM_FPSCR 64 2454 PPC KVM_REG_PPC_TM_AMR 64 2455 PPC KVM_REG_PPC_TM_PPR 64 2456 PPC KVM_REG_PPC_TM_VRSAVE 64 2457 PPC KVM_REG_PPC_TM_VSCR 32 2458 PPC KVM_REG_PPC_TM_DSCR 64 2459 PPC KVM_REG_PPC_TM_TAR 64 2460 PPC KVM_REG_PPC_TM_XER 64 2461 2462 MIPS KVM_REG_MIPS_R0 64 2463 ... 2464 MIPS KVM_REG_MIPS_R31 64 2465 MIPS KVM_REG_MIPS_HI 64 2466 MIPS KVM_REG_MIPS_LO 64 2467 MIPS KVM_REG_MIPS_PC 64 2468 MIPS KVM_REG_MIPS_CP0_INDEX 32 2469 MIPS KVM_REG_MIPS_CP0_ENTRYLO0 64 2470 MIPS KVM_REG_MIPS_CP0_ENTRYLO1 64 2471 MIPS KVM_REG_MIPS_CP0_CONTEXT 64 2472 MIPS KVM_REG_MIPS_CP0_CONTEXTCONFIG 32 2473 MIPS KVM_REG_MIPS_CP0_USERLOCAL 64 2474 MIPS KVM_REG_MIPS_CP0_XCONTEXTCONFIG 64 2475 MIPS KVM_REG_MIPS_CP0_PAGEMASK 32 2476 MIPS KVM_REG_MIPS_CP0_PAGEGRAIN 32 2477 MIPS KVM_REG_MIPS_CP0_SEGCTL0 64 2478 MIPS KVM_REG_MIPS_CP0_SEGCTL1 64 2479 MIPS KVM_REG_MIPS_CP0_SEGCTL2 64 2480 MIPS KVM_REG_MIPS_CP0_PWBASE 64 2481 MIPS KVM_REG_MIPS_CP0_PWFIELD 64 2482 MIPS KVM_REG_MIPS_CP0_PWSIZE 64 2483 MIPS KVM_REG_MIPS_CP0_WIRED 32 2484 MIPS KVM_REG_MIPS_CP0_PWCTL 32 2485 MIPS KVM_REG_MIPS_CP0_HWRENA 32 2486 MIPS KVM_REG_MIPS_CP0_BADVADDR 64 2487 MIPS KVM_REG_MIPS_CP0_BADINSTR 32 2488 MIPS KVM_REG_MIPS_CP0_BADINSTRP 32 2489 MIPS KVM_REG_MIPS_CP0_COUNT 32 2490 MIPS KVM_REG_MIPS_CP0_ENTRYHI 64 2491 MIPS KVM_REG_MIPS_CP0_COMPARE 32 2492 MIPS KVM_REG_MIPS_CP0_STATUS 32 2493 MIPS KVM_REG_MIPS_CP0_INTCTL 32 2494 MIPS KVM_REG_MIPS_CP0_CAUSE 32 2495 MIPS KVM_REG_MIPS_CP0_EPC 64 2496 MIPS KVM_REG_MIPS_CP0_PRID 32 2497 MIPS KVM_REG_MIPS_CP0_EBASE 64 2498 MIPS KVM_REG_MIPS_CP0_CONFIG 32 2499 MIPS KVM_REG_MIPS_CP0_CONFIG1 32 2500 MIPS KVM_REG_MIPS_CP0_CONFIG2 32 2501 MIPS KVM_REG_MIPS_CP0_CONFIG3 32 2502 MIPS KVM_REG_MIPS_CP0_CONFIG4 32 2503 MIPS KVM_REG_MIPS_CP0_CONFIG5 32 2504 MIPS KVM_REG_MIPS_CP0_CONFIG7 32 2505 MIPS KVM_REG_MIPS_CP0_XCONTEXT 64 2506 MIPS KVM_REG_MIPS_CP0_ERROREPC 64 2507 MIPS KVM_REG_MIPS_CP0_KSCRATCH1 64 2508 MIPS KVM_REG_MIPS_CP0_KSCRATCH2 64 2509 MIPS KVM_REG_MIPS_CP0_KSCRATCH3 64 2510 MIPS KVM_REG_MIPS_CP0_KSCRATCH4 64 2511 MIPS KVM_REG_MIPS_CP0_KSCRATCH5 64 2512 MIPS KVM_REG_MIPS_CP0_KSCRATCH6 64 2513 MIPS KVM_REG_MIPS_CP0_MAAR(0..63) 64 2514 MIPS KVM_REG_MIPS_COUNT_CTL 64 2515 MIPS KVM_REG_MIPS_COUNT_RESUME 64 2516 MIPS KVM_REG_MIPS_COUNT_HZ 64 2517 MIPS KVM_REG_MIPS_FPR_32(0..31) 32 2518 MIPS KVM_REG_MIPS_FPR_64(0..31) 64 2519 MIPS KVM_REG_MIPS_VEC_128(0..31) 128 2520 MIPS KVM_REG_MIPS_FCR_IR 32 2521 MIPS KVM_REG_MIPS_FCR_CSR 32 2522 MIPS KVM_REG_MIPS_MSA_IR 32 2523 MIPS KVM_REG_MIPS_MSA_CSR 32 2524 ======= =============================== ============ 2525 2526ARM registers are mapped using the lower 32 bits. The upper 16 of that 2527is the register group type, or coprocessor number: 2528 2529ARM core registers have the following id bit patterns:: 2530 2531 0x4020 0000 0010 <index into the kvm_regs struct:16> 2532 2533ARM 32-bit CP15 registers have the following id bit patterns:: 2534 2535 0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3> 2536 2537ARM 64-bit CP15 registers have the following id bit patterns:: 2538 2539 0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3> 2540 2541ARM CCSIDR registers are demultiplexed by CSSELR value:: 2542 2543 0x4020 0000 0011 00 <csselr:8> 2544 2545ARM 32-bit VFP control registers have the following id bit patterns:: 2546 2547 0x4020 0000 0012 1 <regno:12> 2548 2549ARM 64-bit FP registers have the following id bit patterns:: 2550 2551 0x4030 0000 0012 0 <regno:12> 2552 2553ARM firmware pseudo-registers have the following bit pattern:: 2554 2555 0x4030 0000 0014 <regno:16> 2556 2557 2558arm64 registers are mapped using the lower 32 bits. The upper 16 of 2559that is the register group type, or coprocessor number: 2560 2561arm64 core/FP-SIMD registers have the following id bit patterns. Note 2562that the size of the access is variable, as the kvm_regs structure 2563contains elements ranging from 32 to 128 bits. The index is a 32bit 2564value in the kvm_regs structure seen as a 32bit array:: 2565 2566 0x60x0 0000 0010 <index into the kvm_regs struct:16> 2567 2568Specifically: 2569 2570======================= ========= ===== ======================================= 2571 Encoding Register Bits kvm_regs member 2572======================= ========= ===== ======================================= 2573 0x6030 0000 0010 0000 X0 64 regs.regs[0] 2574 0x6030 0000 0010 0002 X1 64 regs.regs[1] 2575 ... 2576 0x6030 0000 0010 003c X30 64 regs.regs[30] 2577 0x6030 0000 0010 003e SP 64 regs.sp 2578 0x6030 0000 0010 0040 PC 64 regs.pc 2579 0x6030 0000 0010 0042 PSTATE 64 regs.pstate 2580 0x6030 0000 0010 0044 SP_EL1 64 sp_el1 2581 0x6030 0000 0010 0046 ELR_EL1 64 elr_el1 2582 0x6030 0000 0010 0048 SPSR_EL1 64 spsr[KVM_SPSR_EL1] (alias SPSR_SVC) 2583 0x6030 0000 0010 004a SPSR_ABT 64 spsr[KVM_SPSR_ABT] 2584 0x6030 0000 0010 004c SPSR_UND 64 spsr[KVM_SPSR_UND] 2585 0x6030 0000 0010 004e SPSR_IRQ 64 spsr[KVM_SPSR_IRQ] 2586 0x6060 0000 0010 0050 SPSR_FIQ 64 spsr[KVM_SPSR_FIQ] 2587 0x6040 0000 0010 0054 V0 128 fp_regs.vregs[0] [1]_ 2588 0x6040 0000 0010 0058 V1 128 fp_regs.vregs[1] [1]_ 2589 ... 2590 0x6040 0000 0010 00d0 V31 128 fp_regs.vregs[31] [1]_ 2591 0x6020 0000 0010 00d4 FPSR 32 fp_regs.fpsr 2592 0x6020 0000 0010 00d5 FPCR 32 fp_regs.fpcr 2593======================= ========= ===== ======================================= 2594 2595.. [1] These encodings are not accepted for SVE-enabled vcpus. See 2596 KVM_ARM_VCPU_INIT. 2597 2598 The equivalent register content can be accessed via bits [127:0] of 2599 the corresponding SVE Zn registers instead for vcpus that have SVE 2600 enabled (see below). 2601 2602arm64 CCSIDR registers are demultiplexed by CSSELR value:: 2603 2604 0x6020 0000 0011 00 <csselr:8> 2605 2606arm64 system registers have the following id bit patterns:: 2607 2608 0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3> 2609 2610.. warning:: 2611 2612 Two system register IDs do not follow the specified pattern. These 2613 are KVM_REG_ARM_TIMER_CVAL and KVM_REG_ARM_TIMER_CNT, which map to 2614 system registers CNTV_CVAL_EL0 and CNTVCT_EL0 respectively. These 2615 two had their values accidentally swapped, which means TIMER_CVAL is 2616 derived from the register encoding for CNTVCT_EL0 and TIMER_CNT is 2617 derived from the register encoding for CNTV_CVAL_EL0. As this is 2618 API, it must remain this way. 2619 2620arm64 firmware pseudo-registers have the following bit pattern:: 2621 2622 0x6030 0000 0014 <regno:16> 2623 2624arm64 SVE registers have the following bit patterns:: 2625 2626 0x6080 0000 0015 00 <n:5> <slice:5> Zn bits[2048*slice + 2047 : 2048*slice] 2627 0x6050 0000 0015 04 <n:4> <slice:5> Pn bits[256*slice + 255 : 256*slice] 2628 0x6050 0000 0015 060 <slice:5> FFR bits[256*slice + 255 : 256*slice] 2629 0x6060 0000 0015 ffff KVM_REG_ARM64_SVE_VLS pseudo-register 2630 2631Access to register IDs where 2048 * slice >= 128 * max_vq will fail with 2632ENOENT. max_vq is the vcpu's maximum supported vector length in 128-bit 2633quadwords: see [2]_ below. 2634 2635These registers are only accessible on vcpus for which SVE is enabled. 2636See KVM_ARM_VCPU_INIT for details. 2637 2638In addition, except for KVM_REG_ARM64_SVE_VLS, these registers are not 2639accessible until the vcpu's SVE configuration has been finalized 2640using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). See KVM_ARM_VCPU_INIT 2641and KVM_ARM_VCPU_FINALIZE for more information about this procedure. 2642 2643KVM_REG_ARM64_SVE_VLS is a pseudo-register that allows the set of vector 2644lengths supported by the vcpu to be discovered and configured by 2645userspace. When transferred to or from user memory via KVM_GET_ONE_REG 2646or KVM_SET_ONE_REG, the value of this register is of type 2647__u64[KVM_ARM64_SVE_VLS_WORDS], and encodes the set of vector lengths as 2648follows:: 2649 2650 __u64 vector_lengths[KVM_ARM64_SVE_VLS_WORDS]; 2651 2652 if (vq >= SVE_VQ_MIN && vq <= SVE_VQ_MAX && 2653 ((vector_lengths[(vq - KVM_ARM64_SVE_VQ_MIN) / 64] >> 2654 ((vq - KVM_ARM64_SVE_VQ_MIN) % 64)) & 1)) 2655 /* Vector length vq * 16 bytes supported */ 2656 else 2657 /* Vector length vq * 16 bytes not supported */ 2658 2659.. [2] The maximum value vq for which the above condition is true is 2660 max_vq. This is the maximum vector length available to the guest on 2661 this vcpu, and determines which register slices are visible through 2662 this ioctl interface. 2663 2664(See Documentation/arch/arm64/sve.rst for an explanation of the "vq" 2665nomenclature.) 2666 2667KVM_REG_ARM64_SVE_VLS is only accessible after KVM_ARM_VCPU_INIT. 2668KVM_ARM_VCPU_INIT initialises it to the best set of vector lengths that 2669the host supports. 2670 2671Userspace may subsequently modify it if desired until the vcpu's SVE 2672configuration is finalized using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). 2673 2674Apart from simply removing all vector lengths from the host set that 2675exceed some value, support for arbitrarily chosen sets of vector lengths 2676is hardware-dependent and may not be available. Attempting to configure 2677an invalid set of vector lengths via KVM_SET_ONE_REG will fail with 2678EINVAL. 2679 2680After the vcpu's SVE configuration is finalized, further attempts to 2681write this register will fail with EPERM. 2682 2683arm64 bitmap feature firmware pseudo-registers have the following bit pattern:: 2684 2685 0x6030 0000 0016 <regno:16> 2686 2687The bitmap feature firmware registers exposes the hypercall services that 2688are available for userspace to configure. The set bits corresponds to the 2689services that are available for the guests to access. By default, KVM 2690sets all the supported bits during VM initialization. The userspace can 2691discover the available services via KVM_GET_ONE_REG, and write back the 2692bitmap corresponding to the features that it wishes guests to see via 2693KVM_SET_ONE_REG. 2694 2695Note: These registers are immutable once any of the vCPUs of the VM has 2696run at least once. A KVM_SET_ONE_REG in such a scenario will return 2697a -EBUSY to userspace. 2698 2699(See Documentation/virt/kvm/arm/hypercalls.rst for more details.) 2700 2701 2702MIPS registers are mapped using the lower 32 bits. The upper 16 of that is 2703the register group type: 2704 2705MIPS core registers (see above) have the following id bit patterns:: 2706 2707 0x7030 0000 0000 <reg:16> 2708 2709MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit 2710patterns depending on whether they're 32-bit or 64-bit registers:: 2711 2712 0x7020 0000 0001 00 <reg:5> <sel:3> (32-bit) 2713 0x7030 0000 0001 00 <reg:5> <sel:3> (64-bit) 2714 2715Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_MIPS_CP0_ENTRYLO1 are the MIPS64 2716versions of the EntryLo registers regardless of the word size of the host 2717hardware, host kernel, guest, and whether XPA is present in the guest, i.e. 2718with the RI and XI bits (if they exist) in bits 63 and 62 respectively, and 2719the PFNX field starting at bit 30. 2720 2721MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) above) have the following id bit 2722patterns:: 2723 2724 0x7030 0000 0001 01 <reg:8> 2725 2726MIPS KVM control registers (see above) have the following id bit patterns:: 2727 2728 0x7030 0000 0002 <reg:16> 2729 2730MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following 2731id bit patterns depending on the size of the register being accessed. They are 2732always accessed according to the current guest FPU mode (Status.FR and 2733Config5.FRE), i.e. as the guest would see them, and they become unpredictable 2734if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector 2735registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they 2736overlap the FPU registers:: 2737 2738 0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers) 2739 0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers) 2740 0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers) 2741 2742MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the 2743following id bit patterns:: 2744 2745 0x7020 0000 0003 01 <0:3> <reg:5> 2746 2747MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the 2748following id bit patterns:: 2749 2750 0x7020 0000 0003 02 <0:3> <reg:5> 2751 2752RISC-V registers are mapped using the lower 32 bits. The upper 8 bits of 2753that is the register group type. 2754 2755RISC-V config registers are meant for configuring a Guest VCPU and it has 2756the following id bit patterns:: 2757 2758 0x8020 0000 01 <index into the kvm_riscv_config struct:24> (32bit Host) 2759 0x8030 0000 01 <index into the kvm_riscv_config struct:24> (64bit Host) 2760 2761Following are the RISC-V config registers: 2762 2763======================= ========= ============================================= 2764 Encoding Register Description 2765======================= ========= ============================================= 2766 0x80x0 0000 0100 0000 isa ISA feature bitmap of Guest VCPU 2767======================= ========= ============================================= 2768 2769The isa config register can be read anytime but can only be written before 2770a Guest VCPU runs. It will have ISA feature bits matching underlying host 2771set by default. 2772 2773RISC-V core registers represent the general execution state of a Guest VCPU 2774and it has the following id bit patterns:: 2775 2776 0x8020 0000 02 <index into the kvm_riscv_core struct:24> (32bit Host) 2777 0x8030 0000 02 <index into the kvm_riscv_core struct:24> (64bit Host) 2778 2779Following are the RISC-V core registers: 2780 2781======================= ========= ============================================= 2782 Encoding Register Description 2783======================= ========= ============================================= 2784 0x80x0 0000 0200 0000 regs.pc Program counter 2785 0x80x0 0000 0200 0001 regs.ra Return address 2786 0x80x0 0000 0200 0002 regs.sp Stack pointer 2787 0x80x0 0000 0200 0003 regs.gp Global pointer 2788 0x80x0 0000 0200 0004 regs.tp Task pointer 2789 0x80x0 0000 0200 0005 regs.t0 Caller saved register 0 2790 0x80x0 0000 0200 0006 regs.t1 Caller saved register 1 2791 0x80x0 0000 0200 0007 regs.t2 Caller saved register 2 2792 0x80x0 0000 0200 0008 regs.s0 Callee saved register 0 2793 0x80x0 0000 0200 0009 regs.s1 Callee saved register 1 2794 0x80x0 0000 0200 000a regs.a0 Function argument (or return value) 0 2795 0x80x0 0000 0200 000b regs.a1 Function argument (or return value) 1 2796 0x80x0 0000 0200 000c regs.a2 Function argument 2 2797 0x80x0 0000 0200 000d regs.a3 Function argument 3 2798 0x80x0 0000 0200 000e regs.a4 Function argument 4 2799 0x80x0 0000 0200 000f regs.a5 Function argument 5 2800 0x80x0 0000 0200 0010 regs.a6 Function argument 6 2801 0x80x0 0000 0200 0011 regs.a7 Function argument 7 2802 0x80x0 0000 0200 0012 regs.s2 Callee saved register 2 2803 0x80x0 0000 0200 0013 regs.s3 Callee saved register 3 2804 0x80x0 0000 0200 0014 regs.s4 Callee saved register 4 2805 0x80x0 0000 0200 0015 regs.s5 Callee saved register 5 2806 0x80x0 0000 0200 0016 regs.s6 Callee saved register 6 2807 0x80x0 0000 0200 0017 regs.s7 Callee saved register 7 2808 0x80x0 0000 0200 0018 regs.s8 Callee saved register 8 2809 0x80x0 0000 0200 0019 regs.s9 Callee saved register 9 2810 0x80x0 0000 0200 001a regs.s10 Callee saved register 10 2811 0x80x0 0000 0200 001b regs.s11 Callee saved register 11 2812 0x80x0 0000 0200 001c regs.t3 Caller saved register 3 2813 0x80x0 0000 0200 001d regs.t4 Caller saved register 4 2814 0x80x0 0000 0200 001e regs.t5 Caller saved register 5 2815 0x80x0 0000 0200 001f regs.t6 Caller saved register 6 2816 0x80x0 0000 0200 0020 mode Privilege mode (1 = S-mode or 0 = U-mode) 2817======================= ========= ============================================= 2818 2819RISC-V csr registers represent the supervisor mode control/status registers 2820of a Guest VCPU and it has the following id bit patterns:: 2821 2822 0x8020 0000 03 <index into the kvm_riscv_csr struct:24> (32bit Host) 2823 0x8030 0000 03 <index into the kvm_riscv_csr struct:24> (64bit Host) 2824 2825Following are the RISC-V csr registers: 2826 2827======================= ========= ============================================= 2828 Encoding Register Description 2829======================= ========= ============================================= 2830 0x80x0 0000 0300 0000 sstatus Supervisor status 2831 0x80x0 0000 0300 0001 sie Supervisor interrupt enable 2832 0x80x0 0000 0300 0002 stvec Supervisor trap vector base 2833 0x80x0 0000 0300 0003 sscratch Supervisor scratch register 2834 0x80x0 0000 0300 0004 sepc Supervisor exception program counter 2835 0x80x0 0000 0300 0005 scause Supervisor trap cause 2836 0x80x0 0000 0300 0006 stval Supervisor bad address or instruction 2837 0x80x0 0000 0300 0007 sip Supervisor interrupt pending 2838 0x80x0 0000 0300 0008 satp Supervisor address translation and protection 2839======================= ========= ============================================= 2840 2841RISC-V timer registers represent the timer state of a Guest VCPU and it has 2842the following id bit patterns:: 2843 2844 0x8030 0000 04 <index into the kvm_riscv_timer struct:24> 2845 2846Following are the RISC-V timer registers: 2847 2848======================= ========= ============================================= 2849 Encoding Register Description 2850======================= ========= ============================================= 2851 0x8030 0000 0400 0000 frequency Time base frequency (read-only) 2852 0x8030 0000 0400 0001 time Time value visible to Guest 2853 0x8030 0000 0400 0002 compare Time compare programmed by Guest 2854 0x8030 0000 0400 0003 state Time compare state (1 = ON or 0 = OFF) 2855======================= ========= ============================================= 2856 2857RISC-V F-extension registers represent the single precision floating point 2858state of a Guest VCPU and it has the following id bit patterns:: 2859 2860 0x8020 0000 05 <index into the __riscv_f_ext_state struct:24> 2861 2862Following are the RISC-V F-extension registers: 2863 2864======================= ========= ============================================= 2865 Encoding Register Description 2866======================= ========= ============================================= 2867 0x8020 0000 0500 0000 f[0] Floating point register 0 2868 ... 2869 0x8020 0000 0500 001f f[31] Floating point register 31 2870 0x8020 0000 0500 0020 fcsr Floating point control and status register 2871======================= ========= ============================================= 2872 2873RISC-V D-extension registers represent the double precision floating point 2874state of a Guest VCPU and it has the following id bit patterns:: 2875 2876 0x8020 0000 06 <index into the __riscv_d_ext_state struct:24> (fcsr) 2877 0x8030 0000 06 <index into the __riscv_d_ext_state struct:24> (non-fcsr) 2878 2879Following are the RISC-V D-extension registers: 2880 2881======================= ========= ============================================= 2882 Encoding Register Description 2883======================= ========= ============================================= 2884 0x8030 0000 0600 0000 f[0] Floating point register 0 2885 ... 2886 0x8030 0000 0600 001f f[31] Floating point register 31 2887 0x8020 0000 0600 0020 fcsr Floating point control and status register 2888======================= ========= ============================================= 2889 2890LoongArch registers are mapped using the lower 32 bits. The upper 16 bits of 2891that is the register group type. 2892 2893LoongArch csr registers are used to control guest cpu or get status of guest 2894cpu, and they have the following id bit patterns:: 2895 2896 0x9030 0000 0001 00 <reg:5> <sel:3> (64-bit) 2897 2898LoongArch KVM control registers are used to implement some new defined functions 2899such as set vcpu counter or reset vcpu, and they have the following id bit patterns:: 2900 2901 0x9030 0000 0002 <reg:16> 2902 2903 29044.69 KVM_GET_ONE_REG 2905-------------------- 2906 2907:Capability: KVM_CAP_ONE_REG 2908:Architectures: all 2909:Type: vcpu ioctl 2910:Parameters: struct kvm_one_reg (in and out) 2911:Returns: 0 on success, negative value on failure 2912 2913Errors include: 2914 2915 ======== ============================================================ 2916 ENOENT no such register 2917 EINVAL invalid register ID, or no such register or used with VMs in 2918 protected virtualization mode on s390 2919 EPERM (arm64) register access not allowed before vcpu finalization 2920 ======== ============================================================ 2921 2922(These error codes are indicative only: do not rely on a specific error 2923code being returned in a specific situation.) 2924 2925This ioctl allows to receive the value of a single register implemented 2926in a vcpu. The register to read is indicated by the "id" field of the 2927kvm_one_reg struct passed in. On success, the register value can be found 2928at the memory location pointed to by "addr". 2929 2930The list of registers accessible using this interface is identical to the 2931list in 4.68. 2932 2933 29344.70 KVM_KVMCLOCK_CTRL 2935---------------------- 2936 2937:Capability: KVM_CAP_KVMCLOCK_CTRL 2938:Architectures: Any that implement pvclocks (currently x86 only) 2939:Type: vcpu ioctl 2940:Parameters: None 2941:Returns: 0 on success, -1 on error 2942 2943This ioctl sets a flag accessible to the guest indicating that the specified 2944vCPU has been paused by the host userspace. 2945 2946The host will set a flag in the pvclock structure that is checked from the 2947soft lockup watchdog. The flag is part of the pvclock structure that is 2948shared between guest and host, specifically the second bit of the flags 2949field of the pvclock_vcpu_time_info structure. It will be set exclusively by 2950the host and read/cleared exclusively by the guest. The guest operation of 2951checking and clearing the flag must be an atomic operation so 2952load-link/store-conditional, or equivalent must be used. There are two cases 2953where the guest will clear the flag: when the soft lockup watchdog timer resets 2954itself or when a soft lockup is detected. This ioctl can be called any time 2955after pausing the vcpu, but before it is resumed. 2956 2957 29584.71 KVM_SIGNAL_MSI 2959------------------- 2960 2961:Capability: KVM_CAP_SIGNAL_MSI 2962:Architectures: x86 arm64 2963:Type: vm ioctl 2964:Parameters: struct kvm_msi (in) 2965:Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error 2966 2967Directly inject a MSI message. Only valid with in-kernel irqchip that handles 2968MSI messages. 2969 2970:: 2971 2972 struct kvm_msi { 2973 __u32 address_lo; 2974 __u32 address_hi; 2975 __u32 data; 2976 __u32 flags; 2977 __u32 devid; 2978 __u8 pad[12]; 2979 }; 2980 2981flags: 2982 KVM_MSI_VALID_DEVID: devid contains a valid value. The per-VM 2983 KVM_CAP_MSI_DEVID capability advertises the requirement to provide 2984 the device ID. If this capability is not available, userspace 2985 should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail. 2986 2987If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier 2988for the device that wrote the MSI message. For PCI, this is usually a 2989BFD identifier in the lower 16 bits. 2990 2991On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS 2992feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled, 2993address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of 2994address_hi must be zero. 2995 2996 29974.71 KVM_CREATE_PIT2 2998-------------------- 2999 3000:Capability: KVM_CAP_PIT2 3001:Architectures: x86 3002:Type: vm ioctl 3003:Parameters: struct kvm_pit_config (in) 3004:Returns: 0 on success, -1 on error 3005 3006Creates an in-kernel device model for the i8254 PIT. This call is only valid 3007after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following 3008parameters have to be passed:: 3009 3010 struct kvm_pit_config { 3011 __u32 flags; 3012 __u32 pad[15]; 3013 }; 3014 3015Valid flags are:: 3016 3017 #define KVM_PIT_SPEAKER_DUMMY 1 /* emulate speaker port stub */ 3018 3019PIT timer interrupts may use a per-VM kernel thread for injection. If it 3020exists, this thread will have a name of the following pattern:: 3021 3022 kvm-pit/<owner-process-pid> 3023 3024When running a guest with elevated priorities, the scheduling parameters of 3025this thread may have to be adjusted accordingly. 3026 3027This IOCTL replaces the obsolete KVM_CREATE_PIT. 3028 3029 30304.72 KVM_GET_PIT2 3031----------------- 3032 3033:Capability: KVM_CAP_PIT_STATE2 3034:Architectures: x86 3035:Type: vm ioctl 3036:Parameters: struct kvm_pit_state2 (out) 3037:Returns: 0 on success, -1 on error 3038 3039Retrieves the state of the in-kernel PIT model. Only valid after 3040KVM_CREATE_PIT2. The state is returned in the following structure:: 3041 3042 struct kvm_pit_state2 { 3043 struct kvm_pit_channel_state channels[3]; 3044 __u32 flags; 3045 __u32 reserved[9]; 3046 }; 3047 3048Valid flags are:: 3049 3050 /* disable PIT in HPET legacy mode */ 3051 #define KVM_PIT_FLAGS_HPET_LEGACY 0x00000001 3052 /* speaker port data bit enabled */ 3053 #define KVM_PIT_FLAGS_SPEAKER_DATA_ON 0x00000002 3054 3055This IOCTL replaces the obsolete KVM_GET_PIT. 3056 3057 30584.73 KVM_SET_PIT2 3059----------------- 3060 3061:Capability: KVM_CAP_PIT_STATE2 3062:Architectures: x86 3063:Type: vm ioctl 3064:Parameters: struct kvm_pit_state2 (in) 3065:Returns: 0 on success, -1 on error 3066 3067Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2. 3068See KVM_GET_PIT2 for details on struct kvm_pit_state2. 3069 3070This IOCTL replaces the obsolete KVM_SET_PIT. 3071 3072 30734.74 KVM_PPC_GET_SMMU_INFO 3074-------------------------- 3075 3076:Capability: KVM_CAP_PPC_GET_SMMU_INFO 3077:Architectures: powerpc 3078:Type: vm ioctl 3079:Parameters: None 3080:Returns: 0 on success, -1 on error 3081 3082This populates and returns a structure describing the features of 3083the "Server" class MMU emulation supported by KVM. 3084This can in turn be used by userspace to generate the appropriate 3085device-tree properties for the guest operating system. 3086 3087The structure contains some global information, followed by an 3088array of supported segment page sizes:: 3089 3090 struct kvm_ppc_smmu_info { 3091 __u64 flags; 3092 __u32 slb_size; 3093 __u32 pad; 3094 struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ]; 3095 }; 3096 3097The supported flags are: 3098 3099 - KVM_PPC_PAGE_SIZES_REAL: 3100 When that flag is set, guest page sizes must "fit" the backing 3101 store page sizes. When not set, any page size in the list can 3102 be used regardless of how they are backed by userspace. 3103 3104 - KVM_PPC_1T_SEGMENTS 3105 The emulated MMU supports 1T segments in addition to the 3106 standard 256M ones. 3107 3108 - KVM_PPC_NO_HASH 3109 This flag indicates that HPT guests are not supported by KVM, 3110 thus all guests must use radix MMU mode. 3111 3112The "slb_size" field indicates how many SLB entries are supported 3113 3114The "sps" array contains 8 entries indicating the supported base 3115page sizes for a segment in increasing order. Each entry is defined 3116as follow:: 3117 3118 struct kvm_ppc_one_seg_page_size { 3119 __u32 page_shift; /* Base page shift of segment (or 0) */ 3120 __u32 slb_enc; /* SLB encoding for BookS */ 3121 struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ]; 3122 }; 3123 3124An entry with a "page_shift" of 0 is unused. Because the array is 3125organized in increasing order, a lookup can stop when encountering 3126such an entry. 3127 3128The "slb_enc" field provides the encoding to use in the SLB for the 3129page size. The bits are in positions such as the value can directly 3130be OR'ed into the "vsid" argument of the slbmte instruction. 3131 3132The "enc" array is a list which for each of those segment base page 3133size provides the list of supported actual page sizes (which can be 3134only larger or equal to the base page size), along with the 3135corresponding encoding in the hash PTE. Similarly, the array is 31368 entries sorted by increasing sizes and an entry with a "0" shift 3137is an empty entry and a terminator:: 3138 3139 struct kvm_ppc_one_page_size { 3140 __u32 page_shift; /* Page shift (or 0) */ 3141 __u32 pte_enc; /* Encoding in the HPTE (>>12) */ 3142 }; 3143 3144The "pte_enc" field provides a value that can OR'ed into the hash 3145PTE's RPN field (ie, it needs to be shifted left by 12 to OR it 3146into the hash PTE second double word). 3147 31484.75 KVM_IRQFD 3149-------------- 3150 3151:Capability: KVM_CAP_IRQFD 3152:Architectures: x86 s390 arm64 3153:Type: vm ioctl 3154:Parameters: struct kvm_irqfd (in) 3155:Returns: 0 on success, -1 on error 3156 3157Allows setting an eventfd to directly trigger a guest interrupt. 3158kvm_irqfd.fd specifies the file descriptor to use as the eventfd and 3159kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When 3160an event is triggered on the eventfd, an interrupt is injected into 3161the guest using the specified gsi pin. The irqfd is removed using 3162the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd 3163and kvm_irqfd.gsi. 3164 3165With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify 3166mechanism allowing emulation of level-triggered, irqfd-based 3167interrupts. When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an 3168additional eventfd in the kvm_irqfd.resamplefd field. When operating 3169in resample mode, posting of an interrupt through kvm_irq.fd asserts 3170the specified gsi in the irqchip. When the irqchip is resampled, such 3171as from an EOI, the gsi is de-asserted and the user is notified via 3172kvm_irqfd.resamplefd. It is the user's responsibility to re-queue 3173the interrupt if the device making use of it still requires service. 3174Note that closing the resamplefd is not sufficient to disable the 3175irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment 3176and need not be specified with KVM_IRQFD_FLAG_DEASSIGN. 3177 3178On arm64, gsi routing being supported, the following can happen: 3179 3180- in case no routing entry is associated to this gsi, injection fails 3181- in case the gsi is associated to an irqchip routing entry, 3182 irqchip.pin + 32 corresponds to the injected SPI ID. 3183- in case the gsi is associated to an MSI routing entry, the MSI 3184 message and device ID are translated into an LPI (support restricted 3185 to GICv3 ITS in-kernel emulation). 3186 31874.76 KVM_PPC_ALLOCATE_HTAB 3188-------------------------- 3189 3190:Capability: KVM_CAP_PPC_ALLOC_HTAB 3191:Architectures: powerpc 3192:Type: vm ioctl 3193:Parameters: Pointer to u32 containing hash table order (in/out) 3194:Returns: 0 on success, -1 on error 3195 3196This requests the host kernel to allocate an MMU hash table for a 3197guest using the PAPR paravirtualization interface. This only does 3198anything if the kernel is configured to use the Book 3S HV style of 3199virtualization. Otherwise the capability doesn't exist and the ioctl 3200returns an ENOTTY error. The rest of this description assumes Book 3S 3201HV. 3202 3203There must be no vcpus running when this ioctl is called; if there 3204are, it will do nothing and return an EBUSY error. 3205 3206The parameter is a pointer to a 32-bit unsigned integer variable 3207containing the order (log base 2) of the desired size of the hash 3208table, which must be between 18 and 46. On successful return from the 3209ioctl, the value will not be changed by the kernel. 3210 3211If no hash table has been allocated when any vcpu is asked to run 3212(with the KVM_RUN ioctl), the host kernel will allocate a 3213default-sized hash table (16 MB). 3214 3215If this ioctl is called when a hash table has already been allocated, 3216with a different order from the existing hash table, the existing hash 3217table will be freed and a new one allocated. If this is ioctl is 3218called when a hash table has already been allocated of the same order 3219as specified, the kernel will clear out the existing hash table (zero 3220all HPTEs). In either case, if the guest is using the virtualized 3221real-mode area (VRMA) facility, the kernel will re-create the VMRA 3222HPTEs on the next KVM_RUN of any vcpu. 3223 32244.77 KVM_S390_INTERRUPT 3225----------------------- 3226 3227:Capability: basic 3228:Architectures: s390 3229:Type: vm ioctl, vcpu ioctl 3230:Parameters: struct kvm_s390_interrupt (in) 3231:Returns: 0 on success, -1 on error 3232 3233Allows to inject an interrupt to the guest. Interrupts can be floating 3234(vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type. 3235 3236Interrupt parameters are passed via kvm_s390_interrupt:: 3237 3238 struct kvm_s390_interrupt { 3239 __u32 type; 3240 __u32 parm; 3241 __u64 parm64; 3242 }; 3243 3244type can be one of the following: 3245 3246KVM_S390_SIGP_STOP (vcpu) 3247 - sigp stop; optional flags in parm 3248KVM_S390_PROGRAM_INT (vcpu) 3249 - program check; code in parm 3250KVM_S390_SIGP_SET_PREFIX (vcpu) 3251 - sigp set prefix; prefix address in parm 3252KVM_S390_RESTART (vcpu) 3253 - restart 3254KVM_S390_INT_CLOCK_COMP (vcpu) 3255 - clock comparator interrupt 3256KVM_S390_INT_CPU_TIMER (vcpu) 3257 - CPU timer interrupt 3258KVM_S390_INT_VIRTIO (vm) 3259 - virtio external interrupt; external interrupt 3260 parameters in parm and parm64 3261KVM_S390_INT_SERVICE (vm) 3262 - sclp external interrupt; sclp parameter in parm 3263KVM_S390_INT_EMERGENCY (vcpu) 3264 - sigp emergency; source cpu in parm 3265KVM_S390_INT_EXTERNAL_CALL (vcpu) 3266 - sigp external call; source cpu in parm 3267KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm) 3268 - compound value to indicate an 3269 I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel); 3270 I/O interruption parameters in parm (subchannel) and parm64 (intparm, 3271 interruption subclass) 3272KVM_S390_MCHK (vm, vcpu) 3273 - machine check interrupt; cr 14 bits in parm, machine check interrupt 3274 code in parm64 (note that machine checks needing further payload are not 3275 supported by this ioctl) 3276 3277This is an asynchronous vcpu ioctl and can be invoked from any thread. 3278 32794.78 KVM_PPC_GET_HTAB_FD 3280------------------------ 3281 3282:Capability: KVM_CAP_PPC_HTAB_FD 3283:Architectures: powerpc 3284:Type: vm ioctl 3285:Parameters: Pointer to struct kvm_get_htab_fd (in) 3286:Returns: file descriptor number (>= 0) on success, -1 on error 3287 3288This returns a file descriptor that can be used either to read out the 3289entries in the guest's hashed page table (HPT), or to write entries to 3290initialize the HPT. The returned fd can only be written to if the 3291KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and 3292can only be read if that bit is clear. The argument struct looks like 3293this:: 3294 3295 /* For KVM_PPC_GET_HTAB_FD */ 3296 struct kvm_get_htab_fd { 3297 __u64 flags; 3298 __u64 start_index; 3299 __u64 reserved[2]; 3300 }; 3301 3302 /* Values for kvm_get_htab_fd.flags */ 3303 #define KVM_GET_HTAB_BOLTED_ONLY ((__u64)0x1) 3304 #define KVM_GET_HTAB_WRITE ((__u64)0x2) 3305 3306The 'start_index' field gives the index in the HPT of the entry at 3307which to start reading. It is ignored when writing. 3308 3309Reads on the fd will initially supply information about all 3310"interesting" HPT entries. Interesting entries are those with the 3311bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise 3312all entries. When the end of the HPT is reached, the read() will 3313return. If read() is called again on the fd, it will start again from 3314the beginning of the HPT, but will only return HPT entries that have 3315changed since they were last read. 3316 3317Data read or written is structured as a header (8 bytes) followed by a 3318series of valid HPT entries (16 bytes) each. The header indicates how 3319many valid HPT entries there are and how many invalid entries follow 3320the valid entries. The invalid entries are not represented explicitly 3321in the stream. The header format is:: 3322 3323 struct kvm_get_htab_header { 3324 __u32 index; 3325 __u16 n_valid; 3326 __u16 n_invalid; 3327 }; 3328 3329Writes to the fd create HPT entries starting at the index given in the 3330header; first 'n_valid' valid entries with contents from the data 3331written, then 'n_invalid' invalid entries, invalidating any previously 3332valid entries found. 3333 33344.79 KVM_CREATE_DEVICE 3335---------------------- 3336 3337:Capability: KVM_CAP_DEVICE_CTRL 3338:Architectures: all 3339:Type: vm ioctl 3340:Parameters: struct kvm_create_device (in/out) 3341:Returns: 0 on success, -1 on error 3342 3343Errors: 3344 3345 ====== ======================================================= 3346 ENODEV The device type is unknown or unsupported 3347 EEXIST Device already created, and this type of device may not 3348 be instantiated multiple times 3349 ====== ======================================================= 3350 3351 Other error conditions may be defined by individual device types or 3352 have their standard meanings. 3353 3354Creates an emulated device in the kernel. The file descriptor returned 3355in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR. 3356 3357If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the 3358device type is supported (not necessarily whether it can be created 3359in the current vm). 3360 3361Individual devices should not define flags. Attributes should be used 3362for specifying any behavior that is not implied by the device type 3363number. 3364 3365:: 3366 3367 struct kvm_create_device { 3368 __u32 type; /* in: KVM_DEV_TYPE_xxx */ 3369 __u32 fd; /* out: device handle */ 3370 __u32 flags; /* in: KVM_CREATE_DEVICE_xxx */ 3371 }; 3372 33734.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR 3374-------------------------------------------- 3375 3376:Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device, 3377 KVM_CAP_VCPU_ATTRIBUTES for vcpu device 3378 KVM_CAP_SYS_ATTRIBUTES for system (/dev/kvm) device (no set) 3379:Architectures: x86, arm64, s390 3380:Type: device ioctl, vm ioctl, vcpu ioctl 3381:Parameters: struct kvm_device_attr 3382:Returns: 0 on success, -1 on error 3383 3384Errors: 3385 3386 ===== ============================================================= 3387 ENXIO The group or attribute is unknown/unsupported for this device 3388 or hardware support is missing. 3389 EPERM The attribute cannot (currently) be accessed this way 3390 (e.g. read-only attribute, or attribute that only makes 3391 sense when the device is in a different state) 3392 ===== ============================================================= 3393 3394 Other error conditions may be defined by individual device types. 3395 3396Gets/sets a specified piece of device configuration and/or state. The 3397semantics are device-specific. See individual device documentation in 3398the "devices" directory. As with ONE_REG, the size of the data 3399transferred is defined by the particular attribute. 3400 3401:: 3402 3403 struct kvm_device_attr { 3404 __u32 flags; /* no flags currently defined */ 3405 __u32 group; /* device-defined */ 3406 __u64 attr; /* group-defined */ 3407 __u64 addr; /* userspace address of attr data */ 3408 }; 3409 34104.81 KVM_HAS_DEVICE_ATTR 3411------------------------ 3412 3413:Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device, 3414 KVM_CAP_VCPU_ATTRIBUTES for vcpu device 3415 KVM_CAP_SYS_ATTRIBUTES for system (/dev/kvm) device 3416:Type: device ioctl, vm ioctl, vcpu ioctl 3417:Parameters: struct kvm_device_attr 3418:Returns: 0 on success, -1 on error 3419 3420Errors: 3421 3422 ===== ============================================================= 3423 ENXIO The group or attribute is unknown/unsupported for this device 3424 or hardware support is missing. 3425 ===== ============================================================= 3426 3427Tests whether a device supports a particular attribute. A successful 3428return indicates the attribute is implemented. It does not necessarily 3429indicate that the attribute can be read or written in the device's 3430current state. "addr" is ignored. 3431 3432.. _KVM_ARM_VCPU_INIT: 3433 34344.82 KVM_ARM_VCPU_INIT 3435---------------------- 3436 3437:Capability: basic 3438:Architectures: arm64 3439:Type: vcpu ioctl 3440:Parameters: struct kvm_vcpu_init (in) 3441:Returns: 0 on success; -1 on error 3442 3443Errors: 3444 3445 ====== ================================================================= 3446 EINVAL the target is unknown, or the combination of features is invalid. 3447 ENOENT a features bit specified is unknown. 3448 ====== ================================================================= 3449 3450This tells KVM what type of CPU to present to the guest, and what 3451optional features it should have. This will cause a reset of the cpu 3452registers to their initial values. If this is not called, KVM_RUN will 3453return ENOEXEC for that vcpu. 3454 3455The initial values are defined as: 3456 - Processor state: 3457 * AArch64: EL1h, D, A, I and F bits set. All other bits 3458 are cleared. 3459 * AArch32: SVC, A, I and F bits set. All other bits are 3460 cleared. 3461 - General Purpose registers, including PC and SP: set to 0 3462 - FPSIMD/NEON registers: set to 0 3463 - SVE registers: set to 0 3464 - System registers: Reset to their architecturally defined 3465 values as for a warm reset to EL1 (resp. SVC) 3466 3467Note that because some registers reflect machine topology, all vcpus 3468should be created before this ioctl is invoked. 3469 3470Userspace can call this function multiple times for a given vcpu, including 3471after the vcpu has been run. This will reset the vcpu to its initial 3472state. All calls to this function after the initial call must use the same 3473target and same set of feature flags, otherwise EINVAL will be returned. 3474 3475Possible features: 3476 3477 - KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state. 3478 Depends on KVM_CAP_ARM_PSCI. If not set, the CPU will be powered on 3479 and execute guest code when KVM_RUN is called. 3480 - KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode. 3481 Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only). 3482 - KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 (or a future revision 3483 backward compatible with v0.2) for the CPU. 3484 Depends on KVM_CAP_ARM_PSCI_0_2. 3485 - KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU. 3486 Depends on KVM_CAP_ARM_PMU_V3. 3487 3488 - KVM_ARM_VCPU_PTRAUTH_ADDRESS: Enables Address Pointer authentication 3489 for arm64 only. 3490 Depends on KVM_CAP_ARM_PTRAUTH_ADDRESS. 3491 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are 3492 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and 3493 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be 3494 requested. 3495 3496 - KVM_ARM_VCPU_PTRAUTH_GENERIC: Enables Generic Pointer authentication 3497 for arm64 only. 3498 Depends on KVM_CAP_ARM_PTRAUTH_GENERIC. 3499 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are 3500 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and 3501 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be 3502 requested. 3503 3504 - KVM_ARM_VCPU_SVE: Enables SVE for the CPU (arm64 only). 3505 Depends on KVM_CAP_ARM_SVE. 3506 Requires KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3507 3508 * After KVM_ARM_VCPU_INIT: 3509 3510 - KVM_REG_ARM64_SVE_VLS may be read using KVM_GET_ONE_REG: the 3511 initial value of this pseudo-register indicates the best set of 3512 vector lengths possible for a vcpu on this host. 3513 3514 * Before KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3515 3516 - KVM_RUN and KVM_GET_REG_LIST are not available; 3517 3518 - KVM_GET_ONE_REG and KVM_SET_ONE_REG cannot be used to access 3519 the scalable architectural SVE registers 3520 KVM_REG_ARM64_SVE_ZREG(), KVM_REG_ARM64_SVE_PREG() or 3521 KVM_REG_ARM64_SVE_FFR; 3522 3523 - KVM_REG_ARM64_SVE_VLS may optionally be written using 3524 KVM_SET_ONE_REG, to modify the set of vector lengths available 3525 for the vcpu. 3526 3527 * After KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE): 3528 3529 - the KVM_REG_ARM64_SVE_VLS pseudo-register is immutable, and can 3530 no longer be written using KVM_SET_ONE_REG. 3531 35324.83 KVM_ARM_PREFERRED_TARGET 3533----------------------------- 3534 3535:Capability: basic 3536:Architectures: arm64 3537:Type: vm ioctl 3538:Parameters: struct kvm_vcpu_init (out) 3539:Returns: 0 on success; -1 on error 3540 3541Errors: 3542 3543 ====== ========================================== 3544 ENODEV no preferred target available for the host 3545 ====== ========================================== 3546 3547This queries KVM for preferred CPU target type which can be emulated 3548by KVM on underlying host. 3549 3550The ioctl returns struct kvm_vcpu_init instance containing information 3551about preferred CPU target type and recommended features for it. The 3552kvm_vcpu_init->features bitmap returned will have feature bits set if 3553the preferred target recommends setting these features, but this is 3554not mandatory. 3555 3556The information returned by this ioctl can be used to prepare an instance 3557of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in 3558VCPU matching underlying host. 3559 3560 35614.84 KVM_GET_REG_LIST 3562--------------------- 3563 3564:Capability: basic 3565:Architectures: arm64, mips, riscv 3566:Type: vcpu ioctl 3567:Parameters: struct kvm_reg_list (in/out) 3568:Returns: 0 on success; -1 on error 3569 3570Errors: 3571 3572 ===== ============================================================== 3573 E2BIG the reg index list is too big to fit in the array specified by 3574 the user (the number required will be written into n). 3575 ===== ============================================================== 3576 3577:: 3578 3579 struct kvm_reg_list { 3580 __u64 n; /* number of registers in reg[] */ 3581 __u64 reg[0]; 3582 }; 3583 3584This ioctl returns the guest registers that are supported for the 3585KVM_GET_ONE_REG/KVM_SET_ONE_REG calls. 3586 3587 35884.85 KVM_ARM_SET_DEVICE_ADDR (deprecated) 3589----------------------------------------- 3590 3591:Capability: KVM_CAP_ARM_SET_DEVICE_ADDR 3592:Architectures: arm64 3593:Type: vm ioctl 3594:Parameters: struct kvm_arm_device_address (in) 3595:Returns: 0 on success, -1 on error 3596 3597Errors: 3598 3599 ====== ============================================ 3600 ENODEV The device id is unknown 3601 ENXIO Device not supported on current system 3602 EEXIST Address already set 3603 E2BIG Address outside guest physical address space 3604 EBUSY Address overlaps with other device range 3605 ====== ============================================ 3606 3607:: 3608 3609 struct kvm_arm_device_addr { 3610 __u64 id; 3611 __u64 addr; 3612 }; 3613 3614Specify a device address in the guest's physical address space where guests 3615can access emulated or directly exposed devices, which the host kernel needs 3616to know about. The id field is an architecture specific identifier for a 3617specific device. 3618 3619arm64 divides the id field into two parts, a device id and an 3620address type id specific to the individual device:: 3621 3622 bits: | 63 ... 32 | 31 ... 16 | 15 ... 0 | 3623 field: | 0x00000000 | device id | addr type id | 3624 3625arm64 currently only require this when using the in-kernel GIC 3626support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2 3627as the device id. When setting the base address for the guest's 3628mapping of the VGIC virtual CPU and distributor interface, the ioctl 3629must be called after calling KVM_CREATE_IRQCHIP, but before calling 3630KVM_RUN on any of the VCPUs. Calling this ioctl twice for any of the 3631base addresses will return -EEXIST. 3632 3633Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API 3634should be used instead. 3635 3636 36374.86 KVM_PPC_RTAS_DEFINE_TOKEN 3638------------------------------ 3639 3640:Capability: KVM_CAP_PPC_RTAS 3641:Architectures: ppc 3642:Type: vm ioctl 3643:Parameters: struct kvm_rtas_token_args 3644:Returns: 0 on success, -1 on error 3645 3646Defines a token value for a RTAS (Run Time Abstraction Services) 3647service in order to allow it to be handled in the kernel. The 3648argument struct gives the name of the service, which must be the name 3649of a service that has a kernel-side implementation. If the token 3650value is non-zero, it will be associated with that service, and 3651subsequent RTAS calls by the guest specifying that token will be 3652handled by the kernel. If the token value is 0, then any token 3653associated with the service will be forgotten, and subsequent RTAS 3654calls by the guest for that service will be passed to userspace to be 3655handled. 3656 36574.87 KVM_SET_GUEST_DEBUG 3658------------------------ 3659 3660:Capability: KVM_CAP_SET_GUEST_DEBUG 3661:Architectures: x86, s390, ppc, arm64 3662:Type: vcpu ioctl 3663:Parameters: struct kvm_guest_debug (in) 3664:Returns: 0 on success; -1 on error 3665 3666:: 3667 3668 struct kvm_guest_debug { 3669 __u32 control; 3670 __u32 pad; 3671 struct kvm_guest_debug_arch arch; 3672 }; 3673 3674Set up the processor specific debug registers and configure vcpu for 3675handling guest debug events. There are two parts to the structure, the 3676first a control bitfield indicates the type of debug events to handle 3677when running. Common control bits are: 3678 3679 - KVM_GUESTDBG_ENABLE: guest debugging is enabled 3680 - KVM_GUESTDBG_SINGLESTEP: the next run should single-step 3681 3682The top 16 bits of the control field are architecture specific control 3683flags which can include the following: 3684 3685 - KVM_GUESTDBG_USE_SW_BP: using software breakpoints [x86, arm64] 3686 - KVM_GUESTDBG_USE_HW_BP: using hardware breakpoints [x86, s390] 3687 - KVM_GUESTDBG_USE_HW: using hardware debug events [arm64] 3688 - KVM_GUESTDBG_INJECT_DB: inject DB type exception [x86] 3689 - KVM_GUESTDBG_INJECT_BP: inject BP type exception [x86] 3690 - KVM_GUESTDBG_EXIT_PENDING: trigger an immediate guest exit [s390] 3691 - KVM_GUESTDBG_BLOCKIRQ: avoid injecting interrupts/NMI/SMI [x86] 3692 3693For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints 3694are enabled in memory so we need to ensure breakpoint exceptions are 3695correctly trapped and the KVM run loop exits at the breakpoint and not 3696running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP 3697we need to ensure the guest vCPUs architecture specific registers are 3698updated to the correct (supplied) values. 3699 3700The second part of the structure is architecture specific and 3701typically contains a set of debug registers. 3702 3703For arm64 the number of debug registers is implementation defined and 3704can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and 3705KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number 3706indicating the number of supported registers. 3707 3708For ppc, the KVM_CAP_PPC_GUEST_DEBUG_SSTEP capability indicates whether 3709the single-step debug event (KVM_GUESTDBG_SINGLESTEP) is supported. 3710 3711Also when supported, KVM_CAP_SET_GUEST_DEBUG2 capability indicates the 3712supported KVM_GUESTDBG_* bits in the control field. 3713 3714When debug events exit the main run loop with the reason 3715KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run 3716structure containing architecture specific debug information. 3717 37184.88 KVM_GET_EMULATED_CPUID 3719--------------------------- 3720 3721:Capability: KVM_CAP_EXT_EMUL_CPUID 3722:Architectures: x86 3723:Type: system ioctl 3724:Parameters: struct kvm_cpuid2 (in/out) 3725:Returns: 0 on success, -1 on error 3726 3727:: 3728 3729 struct kvm_cpuid2 { 3730 __u32 nent; 3731 __u32 flags; 3732 struct kvm_cpuid_entry2 entries[0]; 3733 }; 3734 3735The member 'flags' is used for passing flags from userspace. 3736 3737:: 3738 3739 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0) 3740 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */ 3741 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */ 3742 3743 struct kvm_cpuid_entry2 { 3744 __u32 function; 3745 __u32 index; 3746 __u32 flags; 3747 __u32 eax; 3748 __u32 ebx; 3749 __u32 ecx; 3750 __u32 edx; 3751 __u32 padding[3]; 3752 }; 3753 3754This ioctl returns x86 cpuid features which are emulated by 3755kvm.Userspace can use the information returned by this ioctl to query 3756which features are emulated by kvm instead of being present natively. 3757 3758Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2 3759structure with the 'nent' field indicating the number of entries in 3760the variable-size array 'entries'. If the number of entries is too low 3761to describe the cpu capabilities, an error (E2BIG) is returned. If the 3762number is too high, the 'nent' field is adjusted and an error (ENOMEM) 3763is returned. If the number is just right, the 'nent' field is adjusted 3764to the number of valid entries in the 'entries' array, which is then 3765filled. 3766 3767The entries returned are the set CPUID bits of the respective features 3768which kvm emulates, as returned by the CPUID instruction, with unknown 3769or unsupported feature bits cleared. 3770 3771Features like x2apic, for example, may not be present in the host cpu 3772but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be 3773emulated efficiently and thus not included here. 3774 3775The fields in each entry are defined as follows: 3776 3777 function: 3778 the eax value used to obtain the entry 3779 index: 3780 the ecx value used to obtain the entry (for entries that are 3781 affected by ecx) 3782 flags: 3783 an OR of zero or more of the following: 3784 3785 KVM_CPUID_FLAG_SIGNIFCANT_INDEX: 3786 if the index field is valid 3787 3788 eax, ebx, ecx, edx: 3789 3790 the values returned by the cpuid instruction for 3791 this function/index combination 3792 37934.89 KVM_S390_MEM_OP 3794-------------------- 3795 3796:Capability: KVM_CAP_S390_MEM_OP, KVM_CAP_S390_PROTECTED, KVM_CAP_S390_MEM_OP_EXTENSION 3797:Architectures: s390 3798:Type: vm ioctl, vcpu ioctl 3799:Parameters: struct kvm_s390_mem_op (in) 3800:Returns: = 0 on success, 3801 < 0 on generic error (e.g. -EFAULT or -ENOMEM), 3802 16 bit program exception code if the access causes such an exception 3803 3804Read or write data from/to the VM's memory. 3805The KVM_CAP_S390_MEM_OP_EXTENSION capability specifies what functionality is 3806supported. 3807 3808Parameters are specified via the following structure:: 3809 3810 struct kvm_s390_mem_op { 3811 __u64 gaddr; /* the guest address */ 3812 __u64 flags; /* flags */ 3813 __u32 size; /* amount of bytes */ 3814 __u32 op; /* type of operation */ 3815 __u64 buf; /* buffer in userspace */ 3816 union { 3817 struct { 3818 __u8 ar; /* the access register number */ 3819 __u8 key; /* access key, ignored if flag unset */ 3820 __u8 pad1[6]; /* ignored */ 3821 __u64 old_addr; /* ignored if flag unset */ 3822 }; 3823 __u32 sida_offset; /* offset into the sida */ 3824 __u8 reserved[32]; /* ignored */ 3825 }; 3826 }; 3827 3828The start address of the memory region has to be specified in the "gaddr" 3829field, and the length of the region in the "size" field (which must not 3830be 0). The maximum value for "size" can be obtained by checking the 3831KVM_CAP_S390_MEM_OP capability. "buf" is the buffer supplied by the 3832userspace application where the read data should be written to for 3833a read access, or where the data that should be written is stored for 3834a write access. The "reserved" field is meant for future extensions. 3835Reserved and unused values are ignored. Future extension that add members must 3836introduce new flags. 3837 3838The type of operation is specified in the "op" field. Flags modifying 3839their behavior can be set in the "flags" field. Undefined flag bits must 3840be set to 0. 3841 3842Possible operations are: 3843 * ``KVM_S390_MEMOP_LOGICAL_READ`` 3844 * ``KVM_S390_MEMOP_LOGICAL_WRITE`` 3845 * ``KVM_S390_MEMOP_ABSOLUTE_READ`` 3846 * ``KVM_S390_MEMOP_ABSOLUTE_WRITE`` 3847 * ``KVM_S390_MEMOP_SIDA_READ`` 3848 * ``KVM_S390_MEMOP_SIDA_WRITE`` 3849 * ``KVM_S390_MEMOP_ABSOLUTE_CMPXCHG`` 3850 3851Logical read/write: 3852^^^^^^^^^^^^^^^^^^^ 3853 3854Access logical memory, i.e. translate the given guest address to an absolute 3855address given the state of the VCPU and use the absolute address as target of 3856the access. "ar" designates the access register number to be used; the valid 3857range is 0..15. 3858Logical accesses are permitted for the VCPU ioctl only. 3859Logical accesses are permitted for non-protected guests only. 3860 3861Supported flags: 3862 * ``KVM_S390_MEMOP_F_CHECK_ONLY`` 3863 * ``KVM_S390_MEMOP_F_INJECT_EXCEPTION`` 3864 * ``KVM_S390_MEMOP_F_SKEY_PROTECTION`` 3865 3866The KVM_S390_MEMOP_F_CHECK_ONLY flag can be set to check whether the 3867corresponding memory access would cause an access exception; however, 3868no actual access to the data in memory at the destination is performed. 3869In this case, "buf" is unused and can be NULL. 3870 3871In case an access exception occurred during the access (or would occur 3872in case of KVM_S390_MEMOP_F_CHECK_ONLY), the ioctl returns a positive 3873error number indicating the type of exception. This exception is also 3874raised directly at the corresponding VCPU if the flag 3875KVM_S390_MEMOP_F_INJECT_EXCEPTION is set. 3876On protection exceptions, unless specified otherwise, the injected 3877translation-exception identifier (TEID) indicates suppression. 3878 3879If the KVM_S390_MEMOP_F_SKEY_PROTECTION flag is set, storage key 3880protection is also in effect and may cause exceptions if accesses are 3881prohibited given the access key designated by "key"; the valid range is 0..15. 3882KVM_S390_MEMOP_F_SKEY_PROTECTION is available if KVM_CAP_S390_MEM_OP_EXTENSION 3883is > 0. 3884Since the accessed memory may span multiple pages and those pages might have 3885different storage keys, it is possible that a protection exception occurs 3886after memory has been modified. In this case, if the exception is injected, 3887the TEID does not indicate suppression. 3888 3889Absolute read/write: 3890^^^^^^^^^^^^^^^^^^^^ 3891 3892Access absolute memory. This operation is intended to be used with the 3893KVM_S390_MEMOP_F_SKEY_PROTECTION flag, to allow accessing memory and performing 3894the checks required for storage key protection as one operation (as opposed to 3895user space getting the storage keys, performing the checks, and accessing 3896memory thereafter, which could lead to a delay between check and access). 3897Absolute accesses are permitted for the VM ioctl if KVM_CAP_S390_MEM_OP_EXTENSION 3898has the KVM_S390_MEMOP_EXTENSION_CAP_BASE bit set. 3899Currently absolute accesses are not permitted for VCPU ioctls. 3900Absolute accesses are permitted for non-protected guests only. 3901 3902Supported flags: 3903 * ``KVM_S390_MEMOP_F_CHECK_ONLY`` 3904 * ``KVM_S390_MEMOP_F_SKEY_PROTECTION`` 3905 3906The semantics of the flags common with logical accesses are as for logical 3907accesses. 3908 3909Absolute cmpxchg: 3910^^^^^^^^^^^^^^^^^ 3911 3912Perform cmpxchg on absolute guest memory. Intended for use with the 3913KVM_S390_MEMOP_F_SKEY_PROTECTION flag. 3914Instead of doing an unconditional write, the access occurs only if the target 3915location contains the value pointed to by "old_addr". 3916This is performed as an atomic cmpxchg with the length specified by the "size" 3917parameter. "size" must be a power of two up to and including 16. 3918If the exchange did not take place because the target value doesn't match the 3919old value, the value "old_addr" points to is replaced by the target value. 3920User space can tell if an exchange took place by checking if this replacement 3921occurred. The cmpxchg op is permitted for the VM ioctl if 3922KVM_CAP_S390_MEM_OP_EXTENSION has flag KVM_S390_MEMOP_EXTENSION_CAP_CMPXCHG set. 3923 3924Supported flags: 3925 * ``KVM_S390_MEMOP_F_SKEY_PROTECTION`` 3926 3927SIDA read/write: 3928^^^^^^^^^^^^^^^^ 3929 3930Access the secure instruction data area which contains memory operands necessary 3931for instruction emulation for protected guests. 3932SIDA accesses are available if the KVM_CAP_S390_PROTECTED capability is available. 3933SIDA accesses are permitted for the VCPU ioctl only. 3934SIDA accesses are permitted for protected guests only. 3935 3936No flags are supported. 3937 39384.90 KVM_S390_GET_SKEYS 3939----------------------- 3940 3941:Capability: KVM_CAP_S390_SKEYS 3942:Architectures: s390 3943:Type: vm ioctl 3944:Parameters: struct kvm_s390_skeys 3945:Returns: 0 on success, KVM_S390_GET_SKEYS_NONE if guest is not using storage 3946 keys, negative value on error 3947 3948This ioctl is used to get guest storage key values on the s390 3949architecture. The ioctl takes parameters via the kvm_s390_skeys struct:: 3950 3951 struct kvm_s390_skeys { 3952 __u64 start_gfn; 3953 __u64 count; 3954 __u64 skeydata_addr; 3955 __u32 flags; 3956 __u32 reserved[9]; 3957 }; 3958 3959The start_gfn field is the number of the first guest frame whose storage keys 3960you want to get. 3961 3962The count field is the number of consecutive frames (starting from start_gfn) 3963whose storage keys to get. The count field must be at least 1 and the maximum 3964allowed value is defined as KVM_S390_SKEYS_MAX. Values outside this range 3965will cause the ioctl to return -EINVAL. 3966 3967The skeydata_addr field is the address to a buffer large enough to hold count 3968bytes. This buffer will be filled with storage key data by the ioctl. 3969 39704.91 KVM_S390_SET_SKEYS 3971----------------------- 3972 3973:Capability: KVM_CAP_S390_SKEYS 3974:Architectures: s390 3975:Type: vm ioctl 3976:Parameters: struct kvm_s390_skeys 3977:Returns: 0 on success, negative value on error 3978 3979This ioctl is used to set guest storage key values on the s390 3980architecture. The ioctl takes parameters via the kvm_s390_skeys struct. 3981See section on KVM_S390_GET_SKEYS for struct definition. 3982 3983The start_gfn field is the number of the first guest frame whose storage keys 3984you want to set. 3985 3986The count field is the number of consecutive frames (starting from start_gfn) 3987whose storage keys to get. The count field must be at least 1 and the maximum 3988allowed value is defined as KVM_S390_SKEYS_MAX. Values outside this range 3989will cause the ioctl to return -EINVAL. 3990 3991The skeydata_addr field is the address to a buffer containing count bytes of 3992storage keys. Each byte in the buffer will be set as the storage key for a 3993single frame starting at start_gfn for count frames. 3994 3995Note: If any architecturally invalid key value is found in the given data then 3996the ioctl will return -EINVAL. 3997 39984.92 KVM_S390_IRQ 3999----------------- 4000 4001:Capability: KVM_CAP_S390_INJECT_IRQ 4002:Architectures: s390 4003:Type: vcpu ioctl 4004:Parameters: struct kvm_s390_irq (in) 4005:Returns: 0 on success, -1 on error 4006 4007Errors: 4008 4009 4010 ====== ================================================================= 4011 EINVAL interrupt type is invalid 4012 type is KVM_S390_SIGP_STOP and flag parameter is invalid value, 4013 type is KVM_S390_INT_EXTERNAL_CALL and code is bigger 4014 than the maximum of VCPUs 4015 EBUSY type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped, 4016 type is KVM_S390_SIGP_STOP and a stop irq is already pending, 4017 type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt 4018 is already pending 4019 ====== ================================================================= 4020 4021Allows to inject an interrupt to the guest. 4022 4023Using struct kvm_s390_irq as a parameter allows 4024to inject additional payload which is not 4025possible via KVM_S390_INTERRUPT. 4026 4027Interrupt parameters are passed via kvm_s390_irq:: 4028 4029 struct kvm_s390_irq { 4030 __u64 type; 4031 union { 4032 struct kvm_s390_io_info io; 4033 struct kvm_s390_ext_info ext; 4034 struct kvm_s390_pgm_info pgm; 4035 struct kvm_s390_emerg_info emerg; 4036 struct kvm_s390_extcall_info extcall; 4037 struct kvm_s390_prefix_info prefix; 4038 struct kvm_s390_stop_info stop; 4039 struct kvm_s390_mchk_info mchk; 4040 char reserved[64]; 4041 } u; 4042 }; 4043 4044type can be one of the following: 4045 4046- KVM_S390_SIGP_STOP - sigp stop; parameter in .stop 4047- KVM_S390_PROGRAM_INT - program check; parameters in .pgm 4048- KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix 4049- KVM_S390_RESTART - restart; no parameters 4050- KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters 4051- KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters 4052- KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg 4053- KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall 4054- KVM_S390_MCHK - machine check interrupt; parameters in .mchk 4055 4056This is an asynchronous vcpu ioctl and can be invoked from any thread. 4057 40584.94 KVM_S390_GET_IRQ_STATE 4059--------------------------- 4060 4061:Capability: KVM_CAP_S390_IRQ_STATE 4062:Architectures: s390 4063:Type: vcpu ioctl 4064:Parameters: struct kvm_s390_irq_state (out) 4065:Returns: >= number of bytes copied into buffer, 4066 -EINVAL if buffer size is 0, 4067 -ENOBUFS if buffer size is too small to fit all pending interrupts, 4068 -EFAULT if the buffer address was invalid 4069 4070This ioctl allows userspace to retrieve the complete state of all currently 4071pending interrupts in a single buffer. Use cases include migration 4072and introspection. The parameter structure contains the address of a 4073userspace buffer and its length:: 4074 4075 struct kvm_s390_irq_state { 4076 __u64 buf; 4077 __u32 flags; /* will stay unused for compatibility reasons */ 4078 __u32 len; 4079 __u32 reserved[4]; /* will stay unused for compatibility reasons */ 4080 }; 4081 4082Userspace passes in the above struct and for each pending interrupt a 4083struct kvm_s390_irq is copied to the provided buffer. 4084 4085The structure contains a flags and a reserved field for future extensions. As 4086the kernel never checked for flags == 0 and QEMU never pre-zeroed flags and 4087reserved, these fields can not be used in the future without breaking 4088compatibility. 4089 4090If -ENOBUFS is returned the buffer provided was too small and userspace 4091may retry with a bigger buffer. 4092 40934.95 KVM_S390_SET_IRQ_STATE 4094--------------------------- 4095 4096:Capability: KVM_CAP_S390_IRQ_STATE 4097:Architectures: s390 4098:Type: vcpu ioctl 4099:Parameters: struct kvm_s390_irq_state (in) 4100:Returns: 0 on success, 4101 -EFAULT if the buffer address was invalid, 4102 -EINVAL for an invalid buffer length (see below), 4103 -EBUSY if there were already interrupts pending, 4104 errors occurring when actually injecting the 4105 interrupt. See KVM_S390_IRQ. 4106 4107This ioctl allows userspace to set the complete state of all cpu-local 4108interrupts currently pending for the vcpu. It is intended for restoring 4109interrupt state after a migration. The input parameter is a userspace buffer 4110containing a struct kvm_s390_irq_state:: 4111 4112 struct kvm_s390_irq_state { 4113 __u64 buf; 4114 __u32 flags; /* will stay unused for compatibility reasons */ 4115 __u32 len; 4116 __u32 reserved[4]; /* will stay unused for compatibility reasons */ 4117 }; 4118 4119The restrictions for flags and reserved apply as well. 4120(see KVM_S390_GET_IRQ_STATE) 4121 4122The userspace memory referenced by buf contains a struct kvm_s390_irq 4123for each interrupt to be injected into the guest. 4124If one of the interrupts could not be injected for some reason the 4125ioctl aborts. 4126 4127len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0 4128and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq), 4129which is the maximum number of possibly pending cpu-local interrupts. 4130 41314.96 KVM_SMI 4132------------ 4133 4134:Capability: KVM_CAP_X86_SMM 4135:Architectures: x86 4136:Type: vcpu ioctl 4137:Parameters: none 4138:Returns: 0 on success, -1 on error 4139 4140Queues an SMI on the thread's vcpu. 4141 41424.97 KVM_X86_SET_MSR_FILTER 4143---------------------------- 4144 4145:Capability: KVM_CAP_X86_MSR_FILTER 4146:Architectures: x86 4147:Type: vm ioctl 4148:Parameters: struct kvm_msr_filter 4149:Returns: 0 on success, < 0 on error 4150 4151:: 4152 4153 struct kvm_msr_filter_range { 4154 #define KVM_MSR_FILTER_READ (1 << 0) 4155 #define KVM_MSR_FILTER_WRITE (1 << 1) 4156 __u32 flags; 4157 __u32 nmsrs; /* number of msrs in bitmap */ 4158 __u32 base; /* MSR index the bitmap starts at */ 4159 __u8 *bitmap; /* a 1 bit allows the operations in flags, 0 denies */ 4160 }; 4161 4162 #define KVM_MSR_FILTER_MAX_RANGES 16 4163 struct kvm_msr_filter { 4164 #define KVM_MSR_FILTER_DEFAULT_ALLOW (0 << 0) 4165 #define KVM_MSR_FILTER_DEFAULT_DENY (1 << 0) 4166 __u32 flags; 4167 struct kvm_msr_filter_range ranges[KVM_MSR_FILTER_MAX_RANGES]; 4168 }; 4169 4170flags values for ``struct kvm_msr_filter_range``: 4171 4172``KVM_MSR_FILTER_READ`` 4173 4174 Filter read accesses to MSRs using the given bitmap. A 0 in the bitmap 4175 indicates that read accesses should be denied, while a 1 indicates that 4176 a read for a particular MSR should be allowed regardless of the default 4177 filter action. 4178 4179``KVM_MSR_FILTER_WRITE`` 4180 4181 Filter write accesses to MSRs using the given bitmap. A 0 in the bitmap 4182 indicates that write accesses should be denied, while a 1 indicates that 4183 a write for a particular MSR should be allowed regardless of the default 4184 filter action. 4185 4186flags values for ``struct kvm_msr_filter``: 4187 4188``KVM_MSR_FILTER_DEFAULT_ALLOW`` 4189 4190 If no filter range matches an MSR index that is getting accessed, KVM will 4191 allow accesses to all MSRs by default. 4192 4193``KVM_MSR_FILTER_DEFAULT_DENY`` 4194 4195 If no filter range matches an MSR index that is getting accessed, KVM will 4196 deny accesses to all MSRs by default. 4197 4198This ioctl allows userspace to define up to 16 bitmaps of MSR ranges to deny 4199guest MSR accesses that would normally be allowed by KVM. If an MSR is not 4200covered by a specific range, the "default" filtering behavior applies. Each 4201bitmap range covers MSRs from [base .. base+nmsrs). 4202 4203If an MSR access is denied by userspace, the resulting KVM behavior depends on 4204whether or not KVM_CAP_X86_USER_SPACE_MSR's KVM_MSR_EXIT_REASON_FILTER is 4205enabled. If KVM_MSR_EXIT_REASON_FILTER is enabled, KVM will exit to userspace 4206on denied accesses, i.e. userspace effectively intercepts the MSR access. If 4207KVM_MSR_EXIT_REASON_FILTER is not enabled, KVM will inject a #GP into the guest 4208on denied accesses. 4209 4210If an MSR access is allowed by userspace, KVM will emulate and/or virtualize 4211the access in accordance with the vCPU model. Note, KVM may still ultimately 4212inject a #GP if an access is allowed by userspace, e.g. if KVM doesn't support 4213the MSR, or to follow architectural behavior for the MSR. 4214 4215By default, KVM operates in KVM_MSR_FILTER_DEFAULT_ALLOW mode with no MSR range 4216filters. 4217 4218Calling this ioctl with an empty set of ranges (all nmsrs == 0) disables MSR 4219filtering. In that mode, ``KVM_MSR_FILTER_DEFAULT_DENY`` is invalid and causes 4220an error. 4221 4222.. warning:: 4223 MSR accesses as part of nested VM-Enter/VM-Exit are not filtered. 4224 This includes both writes to individual VMCS fields and reads/writes 4225 through the MSR lists pointed to by the VMCS. 4226 4227 x2APIC MSR accesses cannot be filtered (KVM silently ignores filters that 4228 cover any x2APIC MSRs). 4229 4230Note, invoking this ioctl while a vCPU is running is inherently racy. However, 4231KVM does guarantee that vCPUs will see either the previous filter or the new 4232filter, e.g. MSRs with identical settings in both the old and new filter will 4233have deterministic behavior. 4234 4235Similarly, if userspace wishes to intercept on denied accesses, 4236KVM_MSR_EXIT_REASON_FILTER must be enabled before activating any filters, and 4237left enabled until after all filters are deactivated. Failure to do so may 4238result in KVM injecting a #GP instead of exiting to userspace. 4239 42404.98 KVM_CREATE_SPAPR_TCE_64 4241---------------------------- 4242 4243:Capability: KVM_CAP_SPAPR_TCE_64 4244:Architectures: powerpc 4245:Type: vm ioctl 4246:Parameters: struct kvm_create_spapr_tce_64 (in) 4247:Returns: file descriptor for manipulating the created TCE table 4248 4249This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit 4250windows, described in 4.62 KVM_CREATE_SPAPR_TCE 4251 4252This capability uses extended struct in ioctl interface:: 4253 4254 /* for KVM_CAP_SPAPR_TCE_64 */ 4255 struct kvm_create_spapr_tce_64 { 4256 __u64 liobn; 4257 __u32 page_shift; 4258 __u32 flags; 4259 __u64 offset; /* in pages */ 4260 __u64 size; /* in pages */ 4261 }; 4262 4263The aim of extension is to support an additional bigger DMA window with 4264a variable page size. 4265KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and 4266a bus offset of the corresponding DMA window, @size and @offset are numbers 4267of IOMMU pages. 4268 4269@flags are not used at the moment. 4270 4271The rest of functionality is identical to KVM_CREATE_SPAPR_TCE. 4272 42734.99 KVM_REINJECT_CONTROL 4274------------------------- 4275 4276:Capability: KVM_CAP_REINJECT_CONTROL 4277:Architectures: x86 4278:Type: vm ioctl 4279:Parameters: struct kvm_reinject_control (in) 4280:Returns: 0 on success, 4281 -EFAULT if struct kvm_reinject_control cannot be read, 4282 -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier. 4283 4284i8254 (PIT) has two modes, reinject and !reinject. The default is reinject, 4285where KVM queues elapsed i8254 ticks and monitors completion of interrupt from 4286vector(s) that i8254 injects. Reinject mode dequeues a tick and injects its 4287interrupt whenever there isn't a pending interrupt from i8254. 4288!reinject mode injects an interrupt as soon as a tick arrives. 4289 4290:: 4291 4292 struct kvm_reinject_control { 4293 __u8 pit_reinject; 4294 __u8 reserved[31]; 4295 }; 4296 4297pit_reinject = 0 (!reinject mode) is recommended, unless running an old 4298operating system that uses the PIT for timing (e.g. Linux 2.4.x). 4299 43004.100 KVM_PPC_CONFIGURE_V3_MMU 4301------------------------------ 4302 4303:Capability: KVM_CAP_PPC_MMU_RADIX or KVM_CAP_PPC_MMU_HASH_V3 4304:Architectures: ppc 4305:Type: vm ioctl 4306:Parameters: struct kvm_ppc_mmuv3_cfg (in) 4307:Returns: 0 on success, 4308 -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read, 4309 -EINVAL if the configuration is invalid 4310 4311This ioctl controls whether the guest will use radix or HPT (hashed 4312page table) translation, and sets the pointer to the process table for 4313the guest. 4314 4315:: 4316 4317 struct kvm_ppc_mmuv3_cfg { 4318 __u64 flags; 4319 __u64 process_table; 4320 }; 4321 4322There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and 4323KVM_PPC_MMUV3_GTSE. KVM_PPC_MMUV3_RADIX, if set, configures the guest 4324to use radix tree translation, and if clear, to use HPT translation. 4325KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest 4326to be able to use the global TLB and SLB invalidation instructions; 4327if clear, the guest may not use these instructions. 4328 4329The process_table field specifies the address and size of the guest 4330process table, which is in the guest's space. This field is formatted 4331as the second doubleword of the partition table entry, as defined in 4332the Power ISA V3.00, Book III section 5.7.6.1. 4333 43344.101 KVM_PPC_GET_RMMU_INFO 4335--------------------------- 4336 4337:Capability: KVM_CAP_PPC_MMU_RADIX 4338:Architectures: ppc 4339:Type: vm ioctl 4340:Parameters: struct kvm_ppc_rmmu_info (out) 4341:Returns: 0 on success, 4342 -EFAULT if struct kvm_ppc_rmmu_info cannot be written, 4343 -EINVAL if no useful information can be returned 4344 4345This ioctl returns a structure containing two things: (a) a list 4346containing supported radix tree geometries, and (b) a list that maps 4347page sizes to put in the "AP" (actual page size) field for the tlbie 4348(TLB invalidate entry) instruction. 4349 4350:: 4351 4352 struct kvm_ppc_rmmu_info { 4353 struct kvm_ppc_radix_geom { 4354 __u8 page_shift; 4355 __u8 level_bits[4]; 4356 __u8 pad[3]; 4357 } geometries[8]; 4358 __u32 ap_encodings[8]; 4359 }; 4360 4361The geometries[] field gives up to 8 supported geometries for the 4362radix page table, in terms of the log base 2 of the smallest page 4363size, and the number of bits indexed at each level of the tree, from 4364the PTE level up to the PGD level in that order. Any unused entries 4365will have 0 in the page_shift field. 4366 4367The ap_encodings gives the supported page sizes and their AP field 4368encodings, encoded with the AP value in the top 3 bits and the log 4369base 2 of the page size in the bottom 6 bits. 4370 43714.102 KVM_PPC_RESIZE_HPT_PREPARE 4372-------------------------------- 4373 4374:Capability: KVM_CAP_SPAPR_RESIZE_HPT 4375:Architectures: powerpc 4376:Type: vm ioctl 4377:Parameters: struct kvm_ppc_resize_hpt (in) 4378:Returns: 0 on successful completion, 4379 >0 if a new HPT is being prepared, the value is an estimated 4380 number of milliseconds until preparation is complete, 4381 -EFAULT if struct kvm_reinject_control cannot be read, 4382 -EINVAL if the supplied shift or flags are invalid, 4383 -ENOMEM if unable to allocate the new HPT, 4384 4385Used to implement the PAPR extension for runtime resizing of a guest's 4386Hashed Page Table (HPT). Specifically this starts, stops or monitors 4387the preparation of a new potential HPT for the guest, essentially 4388implementing the H_RESIZE_HPT_PREPARE hypercall. 4389 4390:: 4391 4392 struct kvm_ppc_resize_hpt { 4393 __u64 flags; 4394 __u32 shift; 4395 __u32 pad; 4396 }; 4397 4398If called with shift > 0 when there is no pending HPT for the guest, 4399this begins preparation of a new pending HPT of size 2^(shift) bytes. 4400It then returns a positive integer with the estimated number of 4401milliseconds until preparation is complete. 4402 4403If called when there is a pending HPT whose size does not match that 4404requested in the parameters, discards the existing pending HPT and 4405creates a new one as above. 4406 4407If called when there is a pending HPT of the size requested, will: 4408 4409 * If preparation of the pending HPT is already complete, return 0 4410 * If preparation of the pending HPT has failed, return an error 4411 code, then discard the pending HPT. 4412 * If preparation of the pending HPT is still in progress, return an 4413 estimated number of milliseconds until preparation is complete. 4414 4415If called with shift == 0, discards any currently pending HPT and 4416returns 0 (i.e. cancels any in-progress preparation). 4417 4418flags is reserved for future expansion, currently setting any bits in 4419flags will result in an -EINVAL. 4420 4421Normally this will be called repeatedly with the same parameters until 4422it returns <= 0. The first call will initiate preparation, subsequent 4423ones will monitor preparation until it completes or fails. 4424 44254.103 KVM_PPC_RESIZE_HPT_COMMIT 4426------------------------------- 4427 4428:Capability: KVM_CAP_SPAPR_RESIZE_HPT 4429:Architectures: powerpc 4430:Type: vm ioctl 4431:Parameters: struct kvm_ppc_resize_hpt (in) 4432:Returns: 0 on successful completion, 4433 -EFAULT if struct kvm_reinject_control cannot be read, 4434 -EINVAL if the supplied shift or flags are invalid, 4435 -ENXIO is there is no pending HPT, or the pending HPT doesn't 4436 have the requested size, 4437 -EBUSY if the pending HPT is not fully prepared, 4438 -ENOSPC if there was a hash collision when moving existing 4439 HPT entries to the new HPT, 4440 -EIO on other error conditions 4441 4442Used to implement the PAPR extension for runtime resizing of a guest's 4443Hashed Page Table (HPT). Specifically this requests that the guest be 4444transferred to working with the new HPT, essentially implementing the 4445H_RESIZE_HPT_COMMIT hypercall. 4446 4447:: 4448 4449 struct kvm_ppc_resize_hpt { 4450 __u64 flags; 4451 __u32 shift; 4452 __u32 pad; 4453 }; 4454 4455This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has 4456returned 0 with the same parameters. In other cases 4457KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or 4458-EBUSY, though others may be possible if the preparation was started, 4459but failed). 4460 4461This will have undefined effects on the guest if it has not already 4462placed itself in a quiescent state where no vcpu will make MMU enabled 4463memory accesses. 4464 4465On successful completion, the pending HPT will become the guest's active 4466HPT and the previous HPT will be discarded. 4467 4468On failure, the guest will still be operating on its previous HPT. 4469 44704.104 KVM_X86_GET_MCE_CAP_SUPPORTED 4471----------------------------------- 4472 4473:Capability: KVM_CAP_MCE 4474:Architectures: x86 4475:Type: system ioctl 4476:Parameters: u64 mce_cap (out) 4477:Returns: 0 on success, -1 on error 4478 4479Returns supported MCE capabilities. The u64 mce_cap parameter 4480has the same format as the MSR_IA32_MCG_CAP register. Supported 4481capabilities will have the corresponding bits set. 4482 44834.105 KVM_X86_SETUP_MCE 4484----------------------- 4485 4486:Capability: KVM_CAP_MCE 4487:Architectures: x86 4488:Type: vcpu ioctl 4489:Parameters: u64 mcg_cap (in) 4490:Returns: 0 on success, 4491 -EFAULT if u64 mcg_cap cannot be read, 4492 -EINVAL if the requested number of banks is invalid, 4493 -EINVAL if requested MCE capability is not supported. 4494 4495Initializes MCE support for use. The u64 mcg_cap parameter 4496has the same format as the MSR_IA32_MCG_CAP register and 4497specifies which capabilities should be enabled. The maximum 4498supported number of error-reporting banks can be retrieved when 4499checking for KVM_CAP_MCE. The supported capabilities can be 4500retrieved with KVM_X86_GET_MCE_CAP_SUPPORTED. 4501 45024.106 KVM_X86_SET_MCE 4503--------------------- 4504 4505:Capability: KVM_CAP_MCE 4506:Architectures: x86 4507:Type: vcpu ioctl 4508:Parameters: struct kvm_x86_mce (in) 4509:Returns: 0 on success, 4510 -EFAULT if struct kvm_x86_mce cannot be read, 4511 -EINVAL if the bank number is invalid, 4512 -EINVAL if VAL bit is not set in status field. 4513 4514Inject a machine check error (MCE) into the guest. The input 4515parameter is:: 4516 4517 struct kvm_x86_mce { 4518 __u64 status; 4519 __u64 addr; 4520 __u64 misc; 4521 __u64 mcg_status; 4522 __u8 bank; 4523 __u8 pad1[7]; 4524 __u64 pad2[3]; 4525 }; 4526 4527If the MCE being reported is an uncorrected error, KVM will 4528inject it as an MCE exception into the guest. If the guest 4529MCG_STATUS register reports that an MCE is in progress, KVM 4530causes an KVM_EXIT_SHUTDOWN vmexit. 4531 4532Otherwise, if the MCE is a corrected error, KVM will just 4533store it in the corresponding bank (provided this bank is 4534not holding a previously reported uncorrected error). 4535 45364.107 KVM_S390_GET_CMMA_BITS 4537---------------------------- 4538 4539:Capability: KVM_CAP_S390_CMMA_MIGRATION 4540:Architectures: s390 4541:Type: vm ioctl 4542:Parameters: struct kvm_s390_cmma_log (in, out) 4543:Returns: 0 on success, a negative value on error 4544 4545Errors: 4546 4547 ====== ============================================================= 4548 ENOMEM not enough memory can be allocated to complete the task 4549 ENXIO if CMMA is not enabled 4550 EINVAL if KVM_S390_CMMA_PEEK is not set but migration mode was not enabled 4551 EINVAL if KVM_S390_CMMA_PEEK is not set but dirty tracking has been 4552 disabled (and thus migration mode was automatically disabled) 4553 EFAULT if the userspace address is invalid or if no page table is 4554 present for the addresses (e.g. when using hugepages). 4555 ====== ============================================================= 4556 4557This ioctl is used to get the values of the CMMA bits on the s390 4558architecture. It is meant to be used in two scenarios: 4559 4560- During live migration to save the CMMA values. Live migration needs 4561 to be enabled via the KVM_REQ_START_MIGRATION VM property. 4562- To non-destructively peek at the CMMA values, with the flag 4563 KVM_S390_CMMA_PEEK set. 4564 4565The ioctl takes parameters via the kvm_s390_cmma_log struct. The desired 4566values are written to a buffer whose location is indicated via the "values" 4567member in the kvm_s390_cmma_log struct. The values in the input struct are 4568also updated as needed. 4569 4570Each CMMA value takes up one byte. 4571 4572:: 4573 4574 struct kvm_s390_cmma_log { 4575 __u64 start_gfn; 4576 __u32 count; 4577 __u32 flags; 4578 union { 4579 __u64 remaining; 4580 __u64 mask; 4581 }; 4582 __u64 values; 4583 }; 4584 4585start_gfn is the number of the first guest frame whose CMMA values are 4586to be retrieved, 4587 4588count is the length of the buffer in bytes, 4589 4590values points to the buffer where the result will be written to. 4591 4592If count is greater than KVM_S390_SKEYS_MAX, then it is considered to be 4593KVM_S390_SKEYS_MAX. KVM_S390_SKEYS_MAX is re-used for consistency with 4594other ioctls. 4595 4596The result is written in the buffer pointed to by the field values, and 4597the values of the input parameter are updated as follows. 4598 4599Depending on the flags, different actions are performed. The only 4600supported flag so far is KVM_S390_CMMA_PEEK. 4601 4602The default behaviour if KVM_S390_CMMA_PEEK is not set is: 4603start_gfn will indicate the first page frame whose CMMA bits were dirty. 4604It is not necessarily the same as the one passed as input, as clean pages 4605are skipped. 4606 4607count will indicate the number of bytes actually written in the buffer. 4608It can (and very often will) be smaller than the input value, since the 4609buffer is only filled until 16 bytes of clean values are found (which 4610are then not copied in the buffer). Since a CMMA migration block needs 4611the base address and the length, for a total of 16 bytes, we will send 4612back some clean data if there is some dirty data afterwards, as long as 4613the size of the clean data does not exceed the size of the header. This 4614allows to minimize the amount of data to be saved or transferred over 4615the network at the expense of more roundtrips to userspace. The next 4616invocation of the ioctl will skip over all the clean values, saving 4617potentially more than just the 16 bytes we found. 4618 4619If KVM_S390_CMMA_PEEK is set: 4620the existing storage attributes are read even when not in migration 4621mode, and no other action is performed; 4622 4623the output start_gfn will be equal to the input start_gfn, 4624 4625the output count will be equal to the input count, except if the end of 4626memory has been reached. 4627 4628In both cases: 4629the field "remaining" will indicate the total number of dirty CMMA values 4630still remaining, or 0 if KVM_S390_CMMA_PEEK is set and migration mode is 4631not enabled. 4632 4633mask is unused. 4634 4635values points to the userspace buffer where the result will be stored. 4636 46374.108 KVM_S390_SET_CMMA_BITS 4638---------------------------- 4639 4640:Capability: KVM_CAP_S390_CMMA_MIGRATION 4641:Architectures: s390 4642:Type: vm ioctl 4643:Parameters: struct kvm_s390_cmma_log (in) 4644:Returns: 0 on success, a negative value on error 4645 4646This ioctl is used to set the values of the CMMA bits on the s390 4647architecture. It is meant to be used during live migration to restore 4648the CMMA values, but there are no restrictions on its use. 4649The ioctl takes parameters via the kvm_s390_cmma_values struct. 4650Each CMMA value takes up one byte. 4651 4652:: 4653 4654 struct kvm_s390_cmma_log { 4655 __u64 start_gfn; 4656 __u32 count; 4657 __u32 flags; 4658 union { 4659 __u64 remaining; 4660 __u64 mask; 4661 }; 4662 __u64 values; 4663 }; 4664 4665start_gfn indicates the starting guest frame number, 4666 4667count indicates how many values are to be considered in the buffer, 4668 4669flags is not used and must be 0. 4670 4671mask indicates which PGSTE bits are to be considered. 4672 4673remaining is not used. 4674 4675values points to the buffer in userspace where to store the values. 4676 4677This ioctl can fail with -ENOMEM if not enough memory can be allocated to 4678complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if 4679the count field is too large (e.g. more than KVM_S390_CMMA_SIZE_MAX) or 4680if the flags field was not 0, with -EFAULT if the userspace address is 4681invalid, if invalid pages are written to (e.g. after the end of memory) 4682or if no page table is present for the addresses (e.g. when using 4683hugepages). 4684 46854.109 KVM_PPC_GET_CPU_CHAR 4686-------------------------- 4687 4688:Capability: KVM_CAP_PPC_GET_CPU_CHAR 4689:Architectures: powerpc 4690:Type: vm ioctl 4691:Parameters: struct kvm_ppc_cpu_char (out) 4692:Returns: 0 on successful completion, 4693 -EFAULT if struct kvm_ppc_cpu_char cannot be written 4694 4695This ioctl gives userspace information about certain characteristics 4696of the CPU relating to speculative execution of instructions and 4697possible information leakage resulting from speculative execution (see 4698CVE-2017-5715, CVE-2017-5753 and CVE-2017-5754). The information is 4699returned in struct kvm_ppc_cpu_char, which looks like this:: 4700 4701 struct kvm_ppc_cpu_char { 4702 __u64 character; /* characteristics of the CPU */ 4703 __u64 behaviour; /* recommended software behaviour */ 4704 __u64 character_mask; /* valid bits in character */ 4705 __u64 behaviour_mask; /* valid bits in behaviour */ 4706 }; 4707 4708For extensibility, the character_mask and behaviour_mask fields 4709indicate which bits of character and behaviour have been filled in by 4710the kernel. If the set of defined bits is extended in future then 4711userspace will be able to tell whether it is running on a kernel that 4712knows about the new bits. 4713 4714The character field describes attributes of the CPU which can help 4715with preventing inadvertent information disclosure - specifically, 4716whether there is an instruction to flash-invalidate the L1 data cache 4717(ori 30,30,0 or mtspr SPRN_TRIG2,rN), whether the L1 data cache is set 4718to a mode where entries can only be used by the thread that created 4719them, whether the bcctr[l] instruction prevents speculation, and 4720whether a speculation barrier instruction (ori 31,31,0) is provided. 4721 4722The behaviour field describes actions that software should take to 4723prevent inadvertent information disclosure, and thus describes which 4724vulnerabilities the hardware is subject to; specifically whether the 4725L1 data cache should be flushed when returning to user mode from the 4726kernel, and whether a speculation barrier should be placed between an 4727array bounds check and the array access. 4728 4729These fields use the same bit definitions as the new 4730H_GET_CPU_CHARACTERISTICS hypercall. 4731 47324.110 KVM_MEMORY_ENCRYPT_OP 4733--------------------------- 4734 4735:Capability: basic 4736:Architectures: x86 4737:Type: vm 4738:Parameters: an opaque platform specific structure (in/out) 4739:Returns: 0 on success; -1 on error 4740 4741If the platform supports creating encrypted VMs then this ioctl can be used 4742for issuing platform-specific memory encryption commands to manage those 4743encrypted VMs. 4744 4745Currently, this ioctl is used for issuing Secure Encrypted Virtualization 4746(SEV) commands on AMD Processors. The SEV commands are defined in 4747Documentation/virt/kvm/x86/amd-memory-encryption.rst. 4748 47494.111 KVM_MEMORY_ENCRYPT_REG_REGION 4750----------------------------------- 4751 4752:Capability: basic 4753:Architectures: x86 4754:Type: system 4755:Parameters: struct kvm_enc_region (in) 4756:Returns: 0 on success; -1 on error 4757 4758This ioctl can be used to register a guest memory region which may 4759contain encrypted data (e.g. guest RAM, SMRAM etc). 4760 4761It is used in the SEV-enabled guest. When encryption is enabled, a guest 4762memory region may contain encrypted data. The SEV memory encryption 4763engine uses a tweak such that two identical plaintext pages, each at 4764different locations will have differing ciphertexts. So swapping or 4765moving ciphertext of those pages will not result in plaintext being 4766swapped. So relocating (or migrating) physical backing pages for the SEV 4767guest will require some additional steps. 4768 4769Note: The current SEV key management spec does not provide commands to 4770swap or migrate (move) ciphertext pages. Hence, for now we pin the guest 4771memory region registered with the ioctl. 4772 47734.112 KVM_MEMORY_ENCRYPT_UNREG_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 unregister the guest memory region registered 4783with KVM_MEMORY_ENCRYPT_REG_REGION ioctl above. 4784 47854.113 KVM_HYPERV_EVENTFD 4786------------------------ 4787 4788:Capability: KVM_CAP_HYPERV_EVENTFD 4789:Architectures: x86 4790:Type: vm ioctl 4791:Parameters: struct kvm_hyperv_eventfd (in) 4792 4793This ioctl (un)registers an eventfd to receive notifications from the guest on 4794the specified Hyper-V connection id through the SIGNAL_EVENT hypercall, without 4795causing a user exit. SIGNAL_EVENT hypercall with non-zero event flag number 4796(bits 24-31) still triggers a KVM_EXIT_HYPERV_HCALL user exit. 4797 4798:: 4799 4800 struct kvm_hyperv_eventfd { 4801 __u32 conn_id; 4802 __s32 fd; 4803 __u32 flags; 4804 __u32 padding[3]; 4805 }; 4806 4807The conn_id field should fit within 24 bits:: 4808 4809 #define KVM_HYPERV_CONN_ID_MASK 0x00ffffff 4810 4811The acceptable values for the flags field are:: 4812 4813 #define KVM_HYPERV_EVENTFD_DEASSIGN (1 << 0) 4814 4815:Returns: 0 on success, 4816 -EINVAL if conn_id or flags is outside the allowed range, 4817 -ENOENT on deassign if the conn_id isn't registered, 4818 -EEXIST on assign if the conn_id is already registered 4819 48204.114 KVM_GET_NESTED_STATE 4821-------------------------- 4822 4823:Capability: KVM_CAP_NESTED_STATE 4824:Architectures: x86 4825:Type: vcpu ioctl 4826:Parameters: struct kvm_nested_state (in/out) 4827:Returns: 0 on success, -1 on error 4828 4829Errors: 4830 4831 ===== ============================================================= 4832 E2BIG the total state size exceeds the value of 'size' specified by 4833 the user; the size required will be written into size. 4834 ===== ============================================================= 4835 4836:: 4837 4838 struct kvm_nested_state { 4839 __u16 flags; 4840 __u16 format; 4841 __u32 size; 4842 4843 union { 4844 struct kvm_vmx_nested_state_hdr vmx; 4845 struct kvm_svm_nested_state_hdr svm; 4846 4847 /* Pad the header to 128 bytes. */ 4848 __u8 pad[120]; 4849 } hdr; 4850 4851 union { 4852 struct kvm_vmx_nested_state_data vmx[0]; 4853 struct kvm_svm_nested_state_data svm[0]; 4854 } data; 4855 }; 4856 4857 #define KVM_STATE_NESTED_GUEST_MODE 0x00000001 4858 #define KVM_STATE_NESTED_RUN_PENDING 0x00000002 4859 #define KVM_STATE_NESTED_EVMCS 0x00000004 4860 4861 #define KVM_STATE_NESTED_FORMAT_VMX 0 4862 #define KVM_STATE_NESTED_FORMAT_SVM 1 4863 4864 #define KVM_STATE_NESTED_VMX_VMCS_SIZE 0x1000 4865 4866 #define KVM_STATE_NESTED_VMX_SMM_GUEST_MODE 0x00000001 4867 #define KVM_STATE_NESTED_VMX_SMM_VMXON 0x00000002 4868 4869 #define KVM_STATE_VMX_PREEMPTION_TIMER_DEADLINE 0x00000001 4870 4871 struct kvm_vmx_nested_state_hdr { 4872 __u64 vmxon_pa; 4873 __u64 vmcs12_pa; 4874 4875 struct { 4876 __u16 flags; 4877 } smm; 4878 4879 __u32 flags; 4880 __u64 preemption_timer_deadline; 4881 }; 4882 4883 struct kvm_vmx_nested_state_data { 4884 __u8 vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE]; 4885 __u8 shadow_vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE]; 4886 }; 4887 4888This ioctl copies the vcpu's nested virtualization state from the kernel to 4889userspace. 4890 4891The maximum size of the state can be retrieved by passing KVM_CAP_NESTED_STATE 4892to the KVM_CHECK_EXTENSION ioctl(). 4893 48944.115 KVM_SET_NESTED_STATE 4895-------------------------- 4896 4897:Capability: KVM_CAP_NESTED_STATE 4898:Architectures: x86 4899:Type: vcpu ioctl 4900:Parameters: struct kvm_nested_state (in) 4901:Returns: 0 on success, -1 on error 4902 4903This copies the vcpu's kvm_nested_state struct from userspace to the kernel. 4904For the definition of struct kvm_nested_state, see KVM_GET_NESTED_STATE. 4905 49064.116 KVM_(UN)REGISTER_COALESCED_MMIO 4907------------------------------------- 4908 4909:Capability: KVM_CAP_COALESCED_MMIO (for coalesced mmio) 4910 KVM_CAP_COALESCED_PIO (for coalesced pio) 4911:Architectures: all 4912:Type: vm ioctl 4913:Parameters: struct kvm_coalesced_mmio_zone 4914:Returns: 0 on success, < 0 on error 4915 4916Coalesced I/O is a performance optimization that defers hardware 4917register write emulation so that userspace exits are avoided. It is 4918typically used to reduce the overhead of emulating frequently accessed 4919hardware registers. 4920 4921When a hardware register is configured for coalesced I/O, write accesses 4922do not exit to userspace and their value is recorded in a ring buffer 4923that is shared between kernel and userspace. 4924 4925Coalesced I/O is used if one or more write accesses to a hardware 4926register can be deferred until a read or a write to another hardware 4927register on the same device. This last access will cause a vmexit and 4928userspace will process accesses from the ring buffer before emulating 4929it. That will avoid exiting to userspace on repeated writes. 4930 4931Coalesced pio is based on coalesced mmio. There is little difference 4932between coalesced mmio and pio except that coalesced pio records accesses 4933to I/O ports. 4934 49354.117 KVM_CLEAR_DIRTY_LOG (vm ioctl) 4936------------------------------------ 4937 4938:Capability: KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 4939:Architectures: x86, arm64, mips 4940:Type: vm ioctl 4941:Parameters: struct kvm_clear_dirty_log (in) 4942:Returns: 0 on success, -1 on error 4943 4944:: 4945 4946 /* for KVM_CLEAR_DIRTY_LOG */ 4947 struct kvm_clear_dirty_log { 4948 __u32 slot; 4949 __u32 num_pages; 4950 __u64 first_page; 4951 union { 4952 void __user *dirty_bitmap; /* one bit per page */ 4953 __u64 padding; 4954 }; 4955 }; 4956 4957The ioctl clears the dirty status of pages in a memory slot, according to 4958the bitmap that is passed in struct kvm_clear_dirty_log's dirty_bitmap 4959field. Bit 0 of the bitmap corresponds to page "first_page" in the 4960memory slot, and num_pages is the size in bits of the input bitmap. 4961first_page must be a multiple of 64; num_pages must also be a multiple of 496264 unless first_page + num_pages is the size of the memory slot. For each 4963bit that is set in the input bitmap, the corresponding page is marked "clean" 4964in KVM's dirty bitmap, and dirty tracking is re-enabled for that page 4965(for example via write-protection, or by clearing the dirty bit in 4966a page table entry). 4967 4968If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies 4969the address space for which you want to clear the dirty status. See 4970KVM_SET_USER_MEMORY_REGION for details on the usage of slot field. 4971 4972This ioctl is mostly useful when KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 4973is enabled; for more information, see the description of the capability. 4974However, it can always be used as long as KVM_CHECK_EXTENSION confirms 4975that KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is present. 4976 49774.118 KVM_GET_SUPPORTED_HV_CPUID 4978-------------------------------- 4979 4980:Capability: KVM_CAP_HYPERV_CPUID (vcpu), KVM_CAP_SYS_HYPERV_CPUID (system) 4981:Architectures: x86 4982:Type: system ioctl, vcpu ioctl 4983:Parameters: struct kvm_cpuid2 (in/out) 4984:Returns: 0 on success, -1 on error 4985 4986:: 4987 4988 struct kvm_cpuid2 { 4989 __u32 nent; 4990 __u32 padding; 4991 struct kvm_cpuid_entry2 entries[0]; 4992 }; 4993 4994 struct kvm_cpuid_entry2 { 4995 __u32 function; 4996 __u32 index; 4997 __u32 flags; 4998 __u32 eax; 4999 __u32 ebx; 5000 __u32 ecx; 5001 __u32 edx; 5002 __u32 padding[3]; 5003 }; 5004 5005This ioctl returns x86 cpuid features leaves related to Hyper-V emulation in 5006KVM. Userspace can use the information returned by this ioctl to construct 5007cpuid information presented to guests consuming Hyper-V enlightenments (e.g. 5008Windows or Hyper-V guests). 5009 5010CPUID feature leaves returned by this ioctl are defined by Hyper-V Top Level 5011Functional Specification (TLFS). These leaves can't be obtained with 5012KVM_GET_SUPPORTED_CPUID ioctl because some of them intersect with KVM feature 5013leaves (0x40000000, 0x40000001). 5014 5015Currently, the following list of CPUID leaves are returned: 5016 5017 - HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS 5018 - HYPERV_CPUID_INTERFACE 5019 - HYPERV_CPUID_VERSION 5020 - HYPERV_CPUID_FEATURES 5021 - HYPERV_CPUID_ENLIGHTMENT_INFO 5022 - HYPERV_CPUID_IMPLEMENT_LIMITS 5023 - HYPERV_CPUID_NESTED_FEATURES 5024 - HYPERV_CPUID_SYNDBG_VENDOR_AND_MAX_FUNCTIONS 5025 - HYPERV_CPUID_SYNDBG_INTERFACE 5026 - HYPERV_CPUID_SYNDBG_PLATFORM_CAPABILITIES 5027 5028Userspace invokes KVM_GET_SUPPORTED_HV_CPUID by passing a kvm_cpuid2 structure 5029with the 'nent' field indicating the number of entries in the variable-size 5030array 'entries'. If the number of entries is too low to describe all Hyper-V 5031feature leaves, an error (E2BIG) is returned. If the number is more or equal 5032to the number of Hyper-V feature leaves, the 'nent' field is adjusted to the 5033number of valid entries in the 'entries' array, which is then filled. 5034 5035'index' and 'flags' fields in 'struct kvm_cpuid_entry2' are currently reserved, 5036userspace should not expect to get any particular value there. 5037 5038Note, vcpu version of KVM_GET_SUPPORTED_HV_CPUID is currently deprecated. Unlike 5039system ioctl which exposes all supported feature bits unconditionally, vcpu 5040version has the following quirks: 5041 5042- HYPERV_CPUID_NESTED_FEATURES leaf and HV_X64_ENLIGHTENED_VMCS_RECOMMENDED 5043 feature bit are only exposed when Enlightened VMCS was previously enabled 5044 on the corresponding vCPU (KVM_CAP_HYPERV_ENLIGHTENED_VMCS). 5045- HV_STIMER_DIRECT_MODE_AVAILABLE bit is only exposed with in-kernel LAPIC. 5046 (presumes KVM_CREATE_IRQCHIP has already been called). 5047 50484.119 KVM_ARM_VCPU_FINALIZE 5049--------------------------- 5050 5051:Architectures: arm64 5052:Type: vcpu ioctl 5053:Parameters: int feature (in) 5054:Returns: 0 on success, -1 on error 5055 5056Errors: 5057 5058 ====== ============================================================== 5059 EPERM feature not enabled, needs configuration, or already finalized 5060 EINVAL feature unknown or not present 5061 ====== ============================================================== 5062 5063Recognised values for feature: 5064 5065 ===== =========================================== 5066 arm64 KVM_ARM_VCPU_SVE (requires KVM_CAP_ARM_SVE) 5067 ===== =========================================== 5068 5069Finalizes the configuration of the specified vcpu feature. 5070 5071The vcpu must already have been initialised, enabling the affected feature, by 5072means of a successful KVM_ARM_VCPU_INIT call with the appropriate flag set in 5073features[]. 5074 5075For affected vcpu features, this is a mandatory step that must be performed 5076before the vcpu is fully usable. 5077 5078Between KVM_ARM_VCPU_INIT and KVM_ARM_VCPU_FINALIZE, the feature may be 5079configured by use of ioctls such as KVM_SET_ONE_REG. The exact configuration 5080that should be performed and how to do it are feature-dependent. 5081 5082Other calls that depend on a particular feature being finalized, such as 5083KVM_RUN, KVM_GET_REG_LIST, KVM_GET_ONE_REG and KVM_SET_ONE_REG, will fail with 5084-EPERM unless the feature has already been finalized by means of a 5085KVM_ARM_VCPU_FINALIZE call. 5086 5087See KVM_ARM_VCPU_INIT for details of vcpu features that require finalization 5088using this ioctl. 5089 50904.120 KVM_SET_PMU_EVENT_FILTER 5091------------------------------ 5092 5093:Capability: KVM_CAP_PMU_EVENT_FILTER 5094:Architectures: x86 5095:Type: vm ioctl 5096:Parameters: struct kvm_pmu_event_filter (in) 5097:Returns: 0 on success, -1 on error 5098 5099Errors: 5100 5101 ====== ============================================================ 5102 EFAULT args[0] cannot be accessed 5103 EINVAL args[0] contains invalid data in the filter or filter events 5104 E2BIG nevents is too large 5105 EBUSY not enough memory to allocate the filter 5106 ====== ============================================================ 5107 5108:: 5109 5110 struct kvm_pmu_event_filter { 5111 __u32 action; 5112 __u32 nevents; 5113 __u32 fixed_counter_bitmap; 5114 __u32 flags; 5115 __u32 pad[4]; 5116 __u64 events[0]; 5117 }; 5118 5119This ioctl restricts the set of PMU events the guest can program by limiting 5120which event select and unit mask combinations are permitted. 5121 5122The argument holds a list of filter events which will be allowed or denied. 5123 5124Filter events only control general purpose counters; fixed purpose counters 5125are controlled by the fixed_counter_bitmap. 5126 5127Valid values for 'flags':: 5128 5129``0`` 5130 5131To use this mode, clear the 'flags' field. 5132 5133In this mode each event will contain an event select + unit mask. 5134 5135When the guest attempts to program the PMU the guest's event select + 5136unit mask is compared against the filter events to determine whether the 5137guest should have access. 5138 5139``KVM_PMU_EVENT_FLAG_MASKED_EVENTS`` 5140:Capability: KVM_CAP_PMU_EVENT_MASKED_EVENTS 5141 5142In this mode each filter event will contain an event select, mask, match, and 5143exclude value. To encode a masked event use:: 5144 5145 KVM_PMU_ENCODE_MASKED_ENTRY() 5146 5147An encoded event will follow this layout:: 5148 5149 Bits Description 5150 ---- ----------- 5151 7:0 event select (low bits) 5152 15:8 umask match 5153 31:16 unused 5154 35:32 event select (high bits) 5155 36:54 unused 5156 55 exclude bit 5157 63:56 umask mask 5158 5159When the guest attempts to program the PMU, these steps are followed in 5160determining if the guest should have access: 5161 5162 1. Match the event select from the guest against the filter events. 5163 2. If a match is found, match the guest's unit mask to the mask and match 5164 values of the included filter events. 5165 I.e. (unit mask & mask) == match && !exclude. 5166 3. If a match is found, match the guest's unit mask to the mask and match 5167 values of the excluded filter events. 5168 I.e. (unit mask & mask) == match && exclude. 5169 4. 5170 a. If an included match is found and an excluded match is not found, filter 5171 the event. 5172 b. For everything else, do not filter the event. 5173 5. 5174 a. If the event is filtered and it's an allow list, allow the guest to 5175 program the event. 5176 b. If the event is filtered and it's a deny list, do not allow the guest to 5177 program the event. 5178 5179When setting a new pmu event filter, -EINVAL will be returned if any of the 5180unused fields are set or if any of the high bits (35:32) in the event 5181select are set when called on Intel. 5182 5183Valid values for 'action':: 5184 5185 #define KVM_PMU_EVENT_ALLOW 0 5186 #define KVM_PMU_EVENT_DENY 1 5187 5188Via this API, KVM userspace can also control the behavior of the VM's fixed 5189counters (if any) by configuring the "action" and "fixed_counter_bitmap" fields. 5190 5191Specifically, KVM follows the following pseudo-code when determining whether to 5192allow the guest FixCtr[i] to count its pre-defined fixed event:: 5193 5194 FixCtr[i]_is_allowed = (action == ALLOW) && (bitmap & BIT(i)) || 5195 (action == DENY) && !(bitmap & BIT(i)); 5196 FixCtr[i]_is_denied = !FixCtr[i]_is_allowed; 5197 5198KVM always consumes fixed_counter_bitmap, it's userspace's responsibility to 5199ensure fixed_counter_bitmap is set correctly, e.g. if userspace wants to define 5200a filter that only affects general purpose counters. 5201 5202Note, the "events" field also applies to fixed counters' hardcoded event_select 5203and unit_mask values. "fixed_counter_bitmap" has higher priority than "events" 5204if there is a contradiction between the two. 5205 52064.121 KVM_PPC_SVM_OFF 5207--------------------- 5208 5209:Capability: basic 5210:Architectures: powerpc 5211:Type: vm ioctl 5212:Parameters: none 5213:Returns: 0 on successful completion, 5214 5215Errors: 5216 5217 ====== ================================================================ 5218 EINVAL if ultravisor failed to terminate the secure guest 5219 ENOMEM if hypervisor failed to allocate new radix page tables for guest 5220 ====== ================================================================ 5221 5222This ioctl is used to turn off the secure mode of the guest or transition 5223the guest from secure mode to normal mode. This is invoked when the guest 5224is reset. This has no effect if called for a normal guest. 5225 5226This ioctl issues an ultravisor call to terminate the secure guest, 5227unpins the VPA pages and releases all the device pages that are used to 5228track the secure pages by hypervisor. 5229 52304.122 KVM_S390_NORMAL_RESET 5231--------------------------- 5232 5233:Capability: KVM_CAP_S390_VCPU_RESETS 5234:Architectures: s390 5235:Type: vcpu ioctl 5236:Parameters: none 5237:Returns: 0 5238 5239This ioctl resets VCPU registers and control structures according to 5240the cpu reset definition in the POP (Principles Of Operation). 5241 52424.123 KVM_S390_INITIAL_RESET 5243---------------------------- 5244 5245:Capability: none 5246:Architectures: s390 5247:Type: vcpu ioctl 5248:Parameters: none 5249:Returns: 0 5250 5251This ioctl resets VCPU registers and control structures according to 5252the initial cpu reset definition in the POP. However, the cpu is not 5253put into ESA mode. This reset is a superset of the normal reset. 5254 52554.124 KVM_S390_CLEAR_RESET 5256-------------------------- 5257 5258:Capability: KVM_CAP_S390_VCPU_RESETS 5259:Architectures: s390 5260:Type: vcpu ioctl 5261:Parameters: none 5262:Returns: 0 5263 5264This ioctl resets VCPU registers and control structures according to 5265the clear cpu reset definition in the POP. However, the cpu is not put 5266into ESA mode. This reset is a superset of the initial reset. 5267 5268 52694.125 KVM_S390_PV_COMMAND 5270------------------------- 5271 5272:Capability: KVM_CAP_S390_PROTECTED 5273:Architectures: s390 5274:Type: vm ioctl 5275:Parameters: struct kvm_pv_cmd 5276:Returns: 0 on success, < 0 on error 5277 5278:: 5279 5280 struct kvm_pv_cmd { 5281 __u32 cmd; /* Command to be executed */ 5282 __u16 rc; /* Ultravisor return code */ 5283 __u16 rrc; /* Ultravisor return reason code */ 5284 __u64 data; /* Data or address */ 5285 __u32 flags; /* flags for future extensions. Must be 0 for now */ 5286 __u32 reserved[3]; 5287 }; 5288 5289**Ultravisor return codes** 5290The Ultravisor return (reason) codes are provided by the kernel if a 5291Ultravisor call has been executed to achieve the results expected by 5292the command. Therefore they are independent of the IOCTL return 5293code. If KVM changes `rc`, its value will always be greater than 0 5294hence setting it to 0 before issuing a PV command is advised to be 5295able to detect a change of `rc`. 5296 5297**cmd values:** 5298 5299KVM_PV_ENABLE 5300 Allocate memory and register the VM with the Ultravisor, thereby 5301 donating memory to the Ultravisor that will become inaccessible to 5302 KVM. All existing CPUs are converted to protected ones. After this 5303 command has succeeded, any CPU added via hotplug will become 5304 protected during its creation as well. 5305 5306 Errors: 5307 5308 ===== ============================= 5309 EINTR an unmasked signal is pending 5310 ===== ============================= 5311 5312KVM_PV_DISABLE 5313 Deregister the VM from the Ultravisor and reclaim the memory that had 5314 been donated to the Ultravisor, making it usable by the kernel again. 5315 All registered VCPUs are converted back to non-protected ones. If a 5316 previous protected VM had been prepared for asynchronous teardown with 5317 KVM_PV_ASYNC_CLEANUP_PREPARE and not subsequently torn down with 5318 KVM_PV_ASYNC_CLEANUP_PERFORM, it will be torn down in this call 5319 together with the current protected VM. 5320 5321KVM_PV_VM_SET_SEC_PARMS 5322 Pass the image header from VM memory to the Ultravisor in 5323 preparation of image unpacking and verification. 5324 5325KVM_PV_VM_UNPACK 5326 Unpack (protect and decrypt) a page of the encrypted boot image. 5327 5328KVM_PV_VM_VERIFY 5329 Verify the integrity of the unpacked image. Only if this succeeds, 5330 KVM is allowed to start protected VCPUs. 5331 5332KVM_PV_INFO 5333 :Capability: KVM_CAP_S390_PROTECTED_DUMP 5334 5335 Presents an API that provides Ultravisor related data to userspace 5336 via subcommands. len_max is the size of the user space buffer, 5337 len_written is KVM's indication of how much bytes of that buffer 5338 were actually written to. len_written can be used to determine the 5339 valid fields if more response fields are added in the future. 5340 5341 :: 5342 5343 enum pv_cmd_info_id { 5344 KVM_PV_INFO_VM, 5345 KVM_PV_INFO_DUMP, 5346 }; 5347 5348 struct kvm_s390_pv_info_header { 5349 __u32 id; 5350 __u32 len_max; 5351 __u32 len_written; 5352 __u32 reserved; 5353 }; 5354 5355 struct kvm_s390_pv_info { 5356 struct kvm_s390_pv_info_header header; 5357 struct kvm_s390_pv_info_dump dump; 5358 struct kvm_s390_pv_info_vm vm; 5359 }; 5360 5361**subcommands:** 5362 5363 KVM_PV_INFO_VM 5364 This subcommand provides basic Ultravisor information for PV 5365 hosts. These values are likely also exported as files in the sysfs 5366 firmware UV query interface but they are more easily available to 5367 programs in this API. 5368 5369 The installed calls and feature_indication members provide the 5370 installed UV calls and the UV's other feature indications. 5371 5372 The max_* members provide information about the maximum number of PV 5373 vcpus, PV guests and PV guest memory size. 5374 5375 :: 5376 5377 struct kvm_s390_pv_info_vm { 5378 __u64 inst_calls_list[4]; 5379 __u64 max_cpus; 5380 __u64 max_guests; 5381 __u64 max_guest_addr; 5382 __u64 feature_indication; 5383 }; 5384 5385 5386 KVM_PV_INFO_DUMP 5387 This subcommand provides information related to dumping PV guests. 5388 5389 :: 5390 5391 struct kvm_s390_pv_info_dump { 5392 __u64 dump_cpu_buffer_len; 5393 __u64 dump_config_mem_buffer_per_1m; 5394 __u64 dump_config_finalize_len; 5395 }; 5396 5397KVM_PV_DUMP 5398 :Capability: KVM_CAP_S390_PROTECTED_DUMP 5399 5400 Presents an API that provides calls which facilitate dumping a 5401 protected VM. 5402 5403 :: 5404 5405 struct kvm_s390_pv_dmp { 5406 __u64 subcmd; 5407 __u64 buff_addr; 5408 __u64 buff_len; 5409 __u64 gaddr; /* For dump storage state */ 5410 }; 5411 5412 **subcommands:** 5413 5414 KVM_PV_DUMP_INIT 5415 Initializes the dump process of a protected VM. If this call does 5416 not succeed all other subcommands will fail with -EINVAL. This 5417 subcommand will return -EINVAL if a dump process has not yet been 5418 completed. 5419 5420 Not all PV vms can be dumped, the owner needs to set `dump 5421 allowed` PCF bit 34 in the SE header to allow dumping. 5422 5423 KVM_PV_DUMP_CONFIG_STOR_STATE 5424 Stores `buff_len` bytes of tweak component values starting with 5425 the 1MB block specified by the absolute guest address 5426 (`gaddr`). `buff_len` needs to be `conf_dump_storage_state_len` 5427 aligned and at least >= the `conf_dump_storage_state_len` value 5428 provided by the dump uv_info data. buff_user might be written to 5429 even if an error rc is returned. For instance if we encounter a 5430 fault after writing the first page of data. 5431 5432 KVM_PV_DUMP_COMPLETE 5433 If the subcommand succeeds it completes the dump process and lets 5434 KVM_PV_DUMP_INIT be called again. 5435 5436 On success `conf_dump_finalize_len` bytes of completion data will be 5437 stored to the `buff_addr`. The completion data contains a key 5438 derivation seed, IV, tweak nonce and encryption keys as well as an 5439 authentication tag all of which are needed to decrypt the dump at a 5440 later time. 5441 5442KVM_PV_ASYNC_CLEANUP_PREPARE 5443 :Capability: KVM_CAP_S390_PROTECTED_ASYNC_DISABLE 5444 5445 Prepare the current protected VM for asynchronous teardown. Most 5446 resources used by the current protected VM will be set aside for a 5447 subsequent asynchronous teardown. The current protected VM will then 5448 resume execution immediately as non-protected. There can be at most 5449 one protected VM prepared for asynchronous teardown at any time. If 5450 a protected VM had already been prepared for teardown without 5451 subsequently calling KVM_PV_ASYNC_CLEANUP_PERFORM, this call will 5452 fail. In that case, the userspace process should issue a normal 5453 KVM_PV_DISABLE. The resources set aside with this call will need to 5454 be cleaned up with a subsequent call to KVM_PV_ASYNC_CLEANUP_PERFORM 5455 or KVM_PV_DISABLE, otherwise they will be cleaned up when KVM 5456 terminates. KVM_PV_ASYNC_CLEANUP_PREPARE can be called again as soon 5457 as cleanup starts, i.e. before KVM_PV_ASYNC_CLEANUP_PERFORM finishes. 5458 5459KVM_PV_ASYNC_CLEANUP_PERFORM 5460 :Capability: KVM_CAP_S390_PROTECTED_ASYNC_DISABLE 5461 5462 Tear down the protected VM previously prepared for teardown with 5463 KVM_PV_ASYNC_CLEANUP_PREPARE. The resources that had been set aside 5464 will be freed during the execution of this command. This PV command 5465 should ideally be issued by userspace from a separate thread. If a 5466 fatal signal is received (or the process terminates naturally), the 5467 command will terminate immediately without completing, and the normal 5468 KVM shutdown procedure will take care of cleaning up all remaining 5469 protected VMs, including the ones whose teardown was interrupted by 5470 process termination. 5471 54724.126 KVM_XEN_HVM_SET_ATTR 5473-------------------------- 5474 5475:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 5476:Architectures: x86 5477:Type: vm ioctl 5478:Parameters: struct kvm_xen_hvm_attr 5479:Returns: 0 on success, < 0 on error 5480 5481:: 5482 5483 struct kvm_xen_hvm_attr { 5484 __u16 type; 5485 __u16 pad[3]; 5486 union { 5487 __u8 long_mode; 5488 __u8 vector; 5489 __u8 runstate_update_flag; 5490 union { 5491 __u64 gfn; 5492 __u64 hva; 5493 } shared_info; 5494 struct { 5495 __u32 send_port; 5496 __u32 type; /* EVTCHNSTAT_ipi / EVTCHNSTAT_interdomain */ 5497 __u32 flags; 5498 union { 5499 struct { 5500 __u32 port; 5501 __u32 vcpu; 5502 __u32 priority; 5503 } port; 5504 struct { 5505 __u32 port; /* Zero for eventfd */ 5506 __s32 fd; 5507 } eventfd; 5508 __u32 padding[4]; 5509 } deliver; 5510 } evtchn; 5511 __u32 xen_version; 5512 __u64 pad[8]; 5513 } u; 5514 }; 5515 5516type values: 5517 5518KVM_XEN_ATTR_TYPE_LONG_MODE 5519 Sets the ABI mode of the VM to 32-bit or 64-bit (long mode). This 5520 determines the layout of the shared_info page exposed to the VM. 5521 5522KVM_XEN_ATTR_TYPE_SHARED_INFO 5523 Sets the guest physical frame number at which the Xen shared_info 5524 page resides. Note that although Xen places vcpu_info for the first 5525 32 vCPUs in the shared_info page, KVM does not automatically do so 5526 and instead requires that KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO or 5527 KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO_HVA be used explicitly even when 5528 the vcpu_info for a given vCPU resides at the "default" location 5529 in the shared_info page. This is because KVM may not be aware of 5530 the Xen CPU id which is used as the index into the vcpu_info[] 5531 array, so may know the correct default location. 5532 5533 Note that the shared_info page may be constantly written to by KVM; 5534 it contains the event channel bitmap used to deliver interrupts to 5535 a Xen guest, amongst other things. It is exempt from dirty tracking 5536 mechanisms — KVM will not explicitly mark the page as dirty each 5537 time an event channel interrupt is delivered to the guest! Thus, 5538 userspace should always assume that the designated GFN is dirty if 5539 any vCPU has been running or any event channel interrupts can be 5540 routed to the guest. 5541 5542 Setting the gfn to KVM_XEN_INVALID_GFN will disable the shared_info 5543 page. 5544 5545KVM_XEN_ATTR_TYPE_SHARED_INFO_HVA 5546 If the KVM_XEN_HVM_CONFIG_SHARED_INFO_HVA flag is also set in the 5547 Xen capabilities, then this attribute may be used to set the 5548 userspace address at which the shared_info page resides, which 5549 will always be fixed in the VMM regardless of where it is mapped 5550 in guest physical address space. This attribute should be used in 5551 preference to KVM_XEN_ATTR_TYPE_SHARED_INFO as it avoids 5552 unnecessary invalidation of an internal cache when the page is 5553 re-mapped in guest physcial address space. 5554 5555 Setting the hva to zero will disable the shared_info page. 5556 5557KVM_XEN_ATTR_TYPE_UPCALL_VECTOR 5558 Sets the exception vector used to deliver Xen event channel upcalls. 5559 This is the HVM-wide vector injected directly by the hypervisor 5560 (not through the local APIC), typically configured by a guest via 5561 HVM_PARAM_CALLBACK_IRQ. This can be disabled again (e.g. for guest 5562 SHUTDOWN_soft_reset) by setting it to zero. 5563 5564KVM_XEN_ATTR_TYPE_EVTCHN 5565 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5566 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It configures 5567 an outbound port number for interception of EVTCHNOP_send requests 5568 from the guest. A given sending port number may be directed back to 5569 a specified vCPU (by APIC ID) / port / priority on the guest, or to 5570 trigger events on an eventfd. The vCPU and priority can be changed 5571 by setting KVM_XEN_EVTCHN_UPDATE in a subsequent call, but other 5572 fields cannot change for a given sending port. A port mapping is 5573 removed by using KVM_XEN_EVTCHN_DEASSIGN in the flags field. Passing 5574 KVM_XEN_EVTCHN_RESET in the flags field removes all interception of 5575 outbound event channels. The values of the flags field are mutually 5576 exclusive and cannot be combined as a bitmask. 5577 5578KVM_XEN_ATTR_TYPE_XEN_VERSION 5579 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5580 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It configures 5581 the 32-bit version code returned to the guest when it invokes the 5582 XENVER_version call; typically (XEN_MAJOR << 16 | XEN_MINOR). PV 5583 Xen guests will often use this to as a dummy hypercall to trigger 5584 event channel delivery, so responding within the kernel without 5585 exiting to userspace is beneficial. 5586 5587KVM_XEN_ATTR_TYPE_RUNSTATE_UPDATE_FLAG 5588 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5589 support for KVM_XEN_HVM_CONFIG_RUNSTATE_UPDATE_FLAG. It enables the 5590 XEN_RUNSTATE_UPDATE flag which allows guest vCPUs to safely read 5591 other vCPUs' vcpu_runstate_info. Xen guests enable this feature via 5592 the VMASST_TYPE_runstate_update_flag of the HYPERVISOR_vm_assist 5593 hypercall. 5594 55954.127 KVM_XEN_HVM_GET_ATTR 5596-------------------------- 5597 5598:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 5599:Architectures: x86 5600:Type: vm ioctl 5601:Parameters: struct kvm_xen_hvm_attr 5602:Returns: 0 on success, < 0 on error 5603 5604Allows Xen VM attributes to be read. For the structure and types, 5605see KVM_XEN_HVM_SET_ATTR above. The KVM_XEN_ATTR_TYPE_EVTCHN 5606attribute cannot be read. 5607 56084.128 KVM_XEN_VCPU_SET_ATTR 5609--------------------------- 5610 5611:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 5612:Architectures: x86 5613:Type: vcpu ioctl 5614:Parameters: struct kvm_xen_vcpu_attr 5615:Returns: 0 on success, < 0 on error 5616 5617:: 5618 5619 struct kvm_xen_vcpu_attr { 5620 __u16 type; 5621 __u16 pad[3]; 5622 union { 5623 __u64 gpa; 5624 __u64 pad[4]; 5625 struct { 5626 __u64 state; 5627 __u64 state_entry_time; 5628 __u64 time_running; 5629 __u64 time_runnable; 5630 __u64 time_blocked; 5631 __u64 time_offline; 5632 } runstate; 5633 __u32 vcpu_id; 5634 struct { 5635 __u32 port; 5636 __u32 priority; 5637 __u64 expires_ns; 5638 } timer; 5639 __u8 vector; 5640 } u; 5641 }; 5642 5643type values: 5644 5645KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO 5646 Sets the guest physical address of the vcpu_info for a given vCPU. 5647 As with the shared_info page for the VM, the corresponding page may be 5648 dirtied at any time if event channel interrupt delivery is enabled, so 5649 userspace should always assume that the page is dirty without relying 5650 on dirty logging. Setting the gpa to KVM_XEN_INVALID_GPA will disable 5651 the vcpu_info. 5652 5653KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO_HVA 5654 If the KVM_XEN_HVM_CONFIG_SHARED_INFO_HVA flag is also set in the 5655 Xen capabilities, then this attribute may be used to set the 5656 userspace address of the vcpu_info for a given vCPU. It should 5657 only be used when the vcpu_info resides at the "default" location 5658 in the shared_info page. In this case it is safe to assume the 5659 userspace address will not change, because the shared_info page is 5660 an overlay on guest memory and remains at a fixed host address 5661 regardless of where it is mapped in guest physical address space 5662 and hence unnecessary invalidation of an internal cache may be 5663 avoided if the guest memory layout is modified. 5664 If the vcpu_info does not reside at the "default" location then 5665 it is not guaranteed to remain at the same host address and 5666 hence the aforementioned cache invalidation is required. 5667 5668KVM_XEN_VCPU_ATTR_TYPE_VCPU_TIME_INFO 5669 Sets the guest physical address of an additional pvclock structure 5670 for a given vCPU. This is typically used for guest vsyscall support. 5671 Setting the gpa to KVM_XEN_INVALID_GPA will disable the structure. 5672 5673KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR 5674 Sets the guest physical address of the vcpu_runstate_info for a given 5675 vCPU. This is how a Xen guest tracks CPU state such as steal time. 5676 Setting the gpa to KVM_XEN_INVALID_GPA will disable the runstate area. 5677 5678KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_CURRENT 5679 Sets the runstate (RUNSTATE_running/_runnable/_blocked/_offline) of 5680 the given vCPU from the .u.runstate.state member of the structure. 5681 KVM automatically accounts running and runnable time but blocked 5682 and offline states are only entered explicitly. 5683 5684KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_DATA 5685 Sets all fields of the vCPU runstate data from the .u.runstate member 5686 of the structure, including the current runstate. The state_entry_time 5687 must equal the sum of the other four times. 5688 5689KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST 5690 This *adds* the contents of the .u.runstate members of the structure 5691 to the corresponding members of the given vCPU's runstate data, thus 5692 permitting atomic adjustments to the runstate times. The adjustment 5693 to the state_entry_time must equal the sum of the adjustments to the 5694 other four times. The state field must be set to -1, or to a valid 5695 runstate value (RUNSTATE_running, RUNSTATE_runnable, RUNSTATE_blocked 5696 or RUNSTATE_offline) to set the current accounted state as of the 5697 adjusted state_entry_time. 5698 5699KVM_XEN_VCPU_ATTR_TYPE_VCPU_ID 5700 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5701 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It sets the Xen 5702 vCPU ID of the given vCPU, to allow timer-related VCPU operations to 5703 be intercepted by KVM. 5704 5705KVM_XEN_VCPU_ATTR_TYPE_TIMER 5706 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5707 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It sets the 5708 event channel port/priority for the VIRQ_TIMER of the vCPU, as well 5709 as allowing a pending timer to be saved/restored. Setting the timer 5710 port to zero disables kernel handling of the singleshot timer. 5711 5712KVM_XEN_VCPU_ATTR_TYPE_UPCALL_VECTOR 5713 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates 5714 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It sets the 5715 per-vCPU local APIC upcall vector, configured by a Xen guest with 5716 the HVMOP_set_evtchn_upcall_vector hypercall. This is typically 5717 used by Windows guests, and is distinct from the HVM-wide upcall 5718 vector configured with HVM_PARAM_CALLBACK_IRQ. It is disabled by 5719 setting the vector to zero. 5720 5721 57224.129 KVM_XEN_VCPU_GET_ATTR 5723--------------------------- 5724 5725:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO 5726:Architectures: x86 5727:Type: vcpu ioctl 5728:Parameters: struct kvm_xen_vcpu_attr 5729:Returns: 0 on success, < 0 on error 5730 5731Allows Xen vCPU attributes to be read. For the structure and types, 5732see KVM_XEN_VCPU_SET_ATTR above. 5733 5734The KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST type may not be used 5735with the KVM_XEN_VCPU_GET_ATTR ioctl. 5736 57374.130 KVM_ARM_MTE_COPY_TAGS 5738--------------------------- 5739 5740:Capability: KVM_CAP_ARM_MTE 5741:Architectures: arm64 5742:Type: vm ioctl 5743:Parameters: struct kvm_arm_copy_mte_tags 5744:Returns: number of bytes copied, < 0 on error (-EINVAL for incorrect 5745 arguments, -EFAULT if memory cannot be accessed). 5746 5747:: 5748 5749 struct kvm_arm_copy_mte_tags { 5750 __u64 guest_ipa; 5751 __u64 length; 5752 void __user *addr; 5753 __u64 flags; 5754 __u64 reserved[2]; 5755 }; 5756 5757Copies Memory Tagging Extension (MTE) tags to/from guest tag memory. The 5758``guest_ipa`` and ``length`` fields must be ``PAGE_SIZE`` aligned. 5759``length`` must not be bigger than 2^31 - PAGE_SIZE bytes. The ``addr`` 5760field must point to a buffer which the tags will be copied to or from. 5761 5762``flags`` specifies the direction of copy, either ``KVM_ARM_TAGS_TO_GUEST`` or 5763``KVM_ARM_TAGS_FROM_GUEST``. 5764 5765The size of the buffer to store the tags is ``(length / 16)`` bytes 5766(granules in MTE are 16 bytes long). Each byte contains a single tag 5767value. This matches the format of ``PTRACE_PEEKMTETAGS`` and 5768``PTRACE_POKEMTETAGS``. 5769 5770If an error occurs before any data is copied then a negative error code is 5771returned. If some tags have been copied before an error occurs then the number 5772of bytes successfully copied is returned. If the call completes successfully 5773then ``length`` is returned. 5774 57754.131 KVM_GET_SREGS2 5776-------------------- 5777 5778:Capability: KVM_CAP_SREGS2 5779:Architectures: x86 5780:Type: vcpu ioctl 5781:Parameters: struct kvm_sregs2 (out) 5782:Returns: 0 on success, -1 on error 5783 5784Reads special registers from the vcpu. 5785This ioctl (when supported) replaces the KVM_GET_SREGS. 5786 5787:: 5788 5789 struct kvm_sregs2 { 5790 /* out (KVM_GET_SREGS2) / in (KVM_SET_SREGS2) */ 5791 struct kvm_segment cs, ds, es, fs, gs, ss; 5792 struct kvm_segment tr, ldt; 5793 struct kvm_dtable gdt, idt; 5794 __u64 cr0, cr2, cr3, cr4, cr8; 5795 __u64 efer; 5796 __u64 apic_base; 5797 __u64 flags; 5798 __u64 pdptrs[4]; 5799 }; 5800 5801flags values for ``kvm_sregs2``: 5802 5803``KVM_SREGS2_FLAGS_PDPTRS_VALID`` 5804 5805 Indicates that the struct contains valid PDPTR values. 5806 5807 58084.132 KVM_SET_SREGS2 5809-------------------- 5810 5811:Capability: KVM_CAP_SREGS2 5812:Architectures: x86 5813:Type: vcpu ioctl 5814:Parameters: struct kvm_sregs2 (in) 5815:Returns: 0 on success, -1 on error 5816 5817Writes special registers into the vcpu. 5818See KVM_GET_SREGS2 for the data structures. 5819This ioctl (when supported) replaces the KVM_SET_SREGS. 5820 58214.133 KVM_GET_STATS_FD 5822---------------------- 5823 5824:Capability: KVM_CAP_STATS_BINARY_FD 5825:Architectures: all 5826:Type: vm ioctl, vcpu ioctl 5827:Parameters: none 5828:Returns: statistics file descriptor on success, < 0 on error 5829 5830Errors: 5831 5832 ====== ====================================================== 5833 ENOMEM if the fd could not be created due to lack of memory 5834 EMFILE if the number of opened files exceeds the limit 5835 ====== ====================================================== 5836 5837The returned file descriptor can be used to read VM/vCPU statistics data in 5838binary format. The data in the file descriptor consists of four blocks 5839organized as follows: 5840 5841+-------------+ 5842| Header | 5843+-------------+ 5844| id string | 5845+-------------+ 5846| Descriptors | 5847+-------------+ 5848| Stats Data | 5849+-------------+ 5850 5851Apart from the header starting at offset 0, please be aware that it is 5852not guaranteed that the four blocks are adjacent or in the above order; 5853the offsets of the id, descriptors and data blocks are found in the 5854header. However, all four blocks are aligned to 64 bit offsets in the 5855file and they do not overlap. 5856 5857All blocks except the data block are immutable. Userspace can read them 5858only one time after retrieving the file descriptor, and then use ``pread`` or 5859``lseek`` to read the statistics repeatedly. 5860 5861All data is in system endianness. 5862 5863The format of the header is as follows:: 5864 5865 struct kvm_stats_header { 5866 __u32 flags; 5867 __u32 name_size; 5868 __u32 num_desc; 5869 __u32 id_offset; 5870 __u32 desc_offset; 5871 __u32 data_offset; 5872 }; 5873 5874The ``flags`` field is not used at the moment. It is always read as 0. 5875 5876The ``name_size`` field is the size (in byte) of the statistics name string 5877(including trailing '\0') which is contained in the "id string" block and 5878appended at the end of every descriptor. 5879 5880The ``num_desc`` field is the number of descriptors that are included in the 5881descriptor block. (The actual number of values in the data block may be 5882larger, since each descriptor may comprise more than one value). 5883 5884The ``id_offset`` field is the offset of the id string from the start of the 5885file indicated by the file descriptor. It is a multiple of 8. 5886 5887The ``desc_offset`` field is the offset of the Descriptors block from the start 5888of the file indicated by the file descriptor. It is a multiple of 8. 5889 5890The ``data_offset`` field is the offset of the Stats Data block from the start 5891of the file indicated by the file descriptor. It is a multiple of 8. 5892 5893The id string block contains a string which identifies the file descriptor on 5894which KVM_GET_STATS_FD was invoked. The size of the block, including the 5895trailing ``'\0'``, is indicated by the ``name_size`` field in the header. 5896 5897The descriptors block is only needed to be read once for the lifetime of the 5898file descriptor contains a sequence of ``struct kvm_stats_desc``, each followed 5899by a string of size ``name_size``. 5900:: 5901 5902 #define KVM_STATS_TYPE_SHIFT 0 5903 #define KVM_STATS_TYPE_MASK (0xF << KVM_STATS_TYPE_SHIFT) 5904 #define KVM_STATS_TYPE_CUMULATIVE (0x0 << KVM_STATS_TYPE_SHIFT) 5905 #define KVM_STATS_TYPE_INSTANT (0x1 << KVM_STATS_TYPE_SHIFT) 5906 #define KVM_STATS_TYPE_PEAK (0x2 << KVM_STATS_TYPE_SHIFT) 5907 #define KVM_STATS_TYPE_LINEAR_HIST (0x3 << KVM_STATS_TYPE_SHIFT) 5908 #define KVM_STATS_TYPE_LOG_HIST (0x4 << KVM_STATS_TYPE_SHIFT) 5909 #define KVM_STATS_TYPE_MAX KVM_STATS_TYPE_LOG_HIST 5910 5911 #define KVM_STATS_UNIT_SHIFT 4 5912 #define KVM_STATS_UNIT_MASK (0xF << KVM_STATS_UNIT_SHIFT) 5913 #define KVM_STATS_UNIT_NONE (0x0 << KVM_STATS_UNIT_SHIFT) 5914 #define KVM_STATS_UNIT_BYTES (0x1 << KVM_STATS_UNIT_SHIFT) 5915 #define KVM_STATS_UNIT_SECONDS (0x2 << KVM_STATS_UNIT_SHIFT) 5916 #define KVM_STATS_UNIT_CYCLES (0x3 << KVM_STATS_UNIT_SHIFT) 5917 #define KVM_STATS_UNIT_BOOLEAN (0x4 << KVM_STATS_UNIT_SHIFT) 5918 #define KVM_STATS_UNIT_MAX KVM_STATS_UNIT_BOOLEAN 5919 5920 #define KVM_STATS_BASE_SHIFT 8 5921 #define KVM_STATS_BASE_MASK (0xF << KVM_STATS_BASE_SHIFT) 5922 #define KVM_STATS_BASE_POW10 (0x0 << KVM_STATS_BASE_SHIFT) 5923 #define KVM_STATS_BASE_POW2 (0x1 << KVM_STATS_BASE_SHIFT) 5924 #define KVM_STATS_BASE_MAX KVM_STATS_BASE_POW2 5925 5926 struct kvm_stats_desc { 5927 __u32 flags; 5928 __s16 exponent; 5929 __u16 size; 5930 __u32 offset; 5931 __u32 bucket_size; 5932 char name[]; 5933 }; 5934 5935The ``flags`` field contains the type and unit of the statistics data described 5936by this descriptor. Its endianness is CPU native. 5937The following flags are supported: 5938 5939Bits 0-3 of ``flags`` encode the type: 5940 5941 * ``KVM_STATS_TYPE_CUMULATIVE`` 5942 The statistics reports a cumulative count. The value of data can only be increased. 5943 Most of the counters used in KVM are of this type. 5944 The corresponding ``size`` field for this type is always 1. 5945 All cumulative statistics data are read/write. 5946 * ``KVM_STATS_TYPE_INSTANT`` 5947 The statistics reports an instantaneous value. Its value can be increased or 5948 decreased. This type is usually used as a measurement of some resources, 5949 like the number of dirty pages, the number of large pages, etc. 5950 All instant statistics are read only. 5951 The corresponding ``size`` field for this type is always 1. 5952 * ``KVM_STATS_TYPE_PEAK`` 5953 The statistics data reports a peak value, for example the maximum number 5954 of items in a hash table bucket, the longest time waited and so on. 5955 The value of data can only be increased. 5956 The corresponding ``size`` field for this type is always 1. 5957 * ``KVM_STATS_TYPE_LINEAR_HIST`` 5958 The statistic is reported as a linear histogram. The number of 5959 buckets is specified by the ``size`` field. The size of buckets is specified 5960 by the ``hist_param`` field. The range of the Nth bucket (1 <= N < ``size``) 5961 is [``hist_param``*(N-1), ``hist_param``*N), while the range of the last 5962 bucket is [``hist_param``*(``size``-1), +INF). (+INF means positive infinity 5963 value.) 5964 * ``KVM_STATS_TYPE_LOG_HIST`` 5965 The statistic is reported as a logarithmic histogram. The number of 5966 buckets is specified by the ``size`` field. The range of the first bucket is 5967 [0, 1), while the range of the last bucket is [pow(2, ``size``-2), +INF). 5968 Otherwise, The Nth bucket (1 < N < ``size``) covers 5969 [pow(2, N-2), pow(2, N-1)). 5970 5971Bits 4-7 of ``flags`` encode the unit: 5972 5973 * ``KVM_STATS_UNIT_NONE`` 5974 There is no unit for the value of statistics data. This usually means that 5975 the value is a simple counter of an event. 5976 * ``KVM_STATS_UNIT_BYTES`` 5977 It indicates that the statistics data is used to measure memory size, in the 5978 unit of Byte, KiByte, MiByte, GiByte, etc. The unit of the data is 5979 determined by the ``exponent`` field in the descriptor. 5980 * ``KVM_STATS_UNIT_SECONDS`` 5981 It indicates that the statistics data is used to measure time or latency. 5982 * ``KVM_STATS_UNIT_CYCLES`` 5983 It indicates that the statistics data is used to measure CPU clock cycles. 5984 * ``KVM_STATS_UNIT_BOOLEAN`` 5985 It indicates that the statistic will always be either 0 or 1. Boolean 5986 statistics of "peak" type will never go back from 1 to 0. Boolean 5987 statistics can be linear histograms (with two buckets) but not logarithmic 5988 histograms. 5989 5990Note that, in the case of histograms, the unit applies to the bucket 5991ranges, while the bucket value indicates how many samples fell in the 5992bucket's range. 5993 5994Bits 8-11 of ``flags``, together with ``exponent``, encode the scale of the 5995unit: 5996 5997 * ``KVM_STATS_BASE_POW10`` 5998 The scale is based on power of 10. It is used for measurement of time and 5999 CPU clock cycles. For example, an exponent of -9 can be used with 6000 ``KVM_STATS_UNIT_SECONDS`` to express that the unit is nanoseconds. 6001 * ``KVM_STATS_BASE_POW2`` 6002 The scale is based on power of 2. It is used for measurement of memory size. 6003 For example, an exponent of 20 can be used with ``KVM_STATS_UNIT_BYTES`` to 6004 express that the unit is MiB. 6005 6006The ``size`` field is the number of values of this statistics data. Its 6007value is usually 1 for most of simple statistics. 1 means it contains an 6008unsigned 64bit data. 6009 6010The ``offset`` field is the offset from the start of Data Block to the start of 6011the corresponding statistics data. 6012 6013The ``bucket_size`` field is used as a parameter for histogram statistics data. 6014It is only used by linear histogram statistics data, specifying the size of a 6015bucket in the unit expressed by bits 4-11 of ``flags`` together with ``exponent``. 6016 6017The ``name`` field is the name string of the statistics data. The name string 6018starts at the end of ``struct kvm_stats_desc``. The maximum length including 6019the trailing ``'\0'``, is indicated by ``name_size`` in the header. 6020 6021The Stats Data block contains an array of 64-bit values in the same order 6022as the descriptors in Descriptors block. 6023 60244.134 KVM_GET_XSAVE2 6025-------------------- 6026 6027:Capability: KVM_CAP_XSAVE2 6028:Architectures: x86 6029:Type: vcpu ioctl 6030:Parameters: struct kvm_xsave (out) 6031:Returns: 0 on success, -1 on error 6032 6033 6034:: 6035 6036 struct kvm_xsave { 6037 __u32 region[1024]; 6038 __u32 extra[0]; 6039 }; 6040 6041This ioctl would copy current vcpu's xsave struct to the userspace. It 6042copies as many bytes as are returned by KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) 6043when invoked on the vm file descriptor. The size value returned by 6044KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) will always be at least 4096. 6045Currently, it is only greater than 4096 if a dynamic feature has been 6046enabled with ``arch_prctl()``, but this may change in the future. 6047 6048The offsets of the state save areas in struct kvm_xsave follow the contents 6049of CPUID leaf 0xD on the host. 6050 60514.135 KVM_XEN_HVM_EVTCHN_SEND 6052----------------------------- 6053 6054:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_EVTCHN_SEND 6055:Architectures: x86 6056:Type: vm ioctl 6057:Parameters: struct kvm_irq_routing_xen_evtchn 6058:Returns: 0 on success, < 0 on error 6059 6060 6061:: 6062 6063 struct kvm_irq_routing_xen_evtchn { 6064 __u32 port; 6065 __u32 vcpu; 6066 __u32 priority; 6067 }; 6068 6069This ioctl injects an event channel interrupt directly to the guest vCPU. 6070 60714.136 KVM_S390_PV_CPU_COMMAND 6072----------------------------- 6073 6074:Capability: KVM_CAP_S390_PROTECTED_DUMP 6075:Architectures: s390 6076:Type: vcpu ioctl 6077:Parameters: none 6078:Returns: 0 on success, < 0 on error 6079 6080This ioctl closely mirrors `KVM_S390_PV_COMMAND` but handles requests 6081for vcpus. It re-uses the kvm_s390_pv_dmp struct and hence also shares 6082the command ids. 6083 6084**command:** 6085 6086KVM_PV_DUMP 6087 Presents an API that provides calls which facilitate dumping a vcpu 6088 of a protected VM. 6089 6090**subcommand:** 6091 6092KVM_PV_DUMP_CPU 6093 Provides encrypted dump data like register values. 6094 The length of the returned data is provided by uv_info.guest_cpu_stor_len. 6095 60964.137 KVM_S390_ZPCI_OP 6097---------------------- 6098 6099:Capability: KVM_CAP_S390_ZPCI_OP 6100:Architectures: s390 6101:Type: vm ioctl 6102:Parameters: struct kvm_s390_zpci_op (in) 6103:Returns: 0 on success, <0 on error 6104 6105Used to manage hardware-assisted virtualization features for zPCI devices. 6106 6107Parameters are specified via the following structure:: 6108 6109 struct kvm_s390_zpci_op { 6110 /* in */ 6111 __u32 fh; /* target device */ 6112 __u8 op; /* operation to perform */ 6113 __u8 pad[3]; 6114 union { 6115 /* for KVM_S390_ZPCIOP_REG_AEN */ 6116 struct { 6117 __u64 ibv; /* Guest addr of interrupt bit vector */ 6118 __u64 sb; /* Guest addr of summary bit */ 6119 __u32 flags; 6120 __u32 noi; /* Number of interrupts */ 6121 __u8 isc; /* Guest interrupt subclass */ 6122 __u8 sbo; /* Offset of guest summary bit vector */ 6123 __u16 pad; 6124 } reg_aen; 6125 __u64 reserved[8]; 6126 } u; 6127 }; 6128 6129The type of operation is specified in the "op" field. 6130KVM_S390_ZPCIOP_REG_AEN is used to register the VM for adapter event 6131notification interpretation, which will allow firmware delivery of adapter 6132events directly to the vm, with KVM providing a backup delivery mechanism; 6133KVM_S390_ZPCIOP_DEREG_AEN is used to subsequently disable interpretation of 6134adapter event notifications. 6135 6136The target zPCI function must also be specified via the "fh" field. For the 6137KVM_S390_ZPCIOP_REG_AEN operation, additional information to establish firmware 6138delivery must be provided via the "reg_aen" struct. 6139 6140The "pad" and "reserved" fields may be used for future extensions and should be 6141set to 0s by userspace. 6142 61434.138 KVM_ARM_SET_COUNTER_OFFSET 6144-------------------------------- 6145 6146:Capability: KVM_CAP_COUNTER_OFFSET 6147:Architectures: arm64 6148:Type: vm ioctl 6149:Parameters: struct kvm_arm_counter_offset (in) 6150:Returns: 0 on success, < 0 on error 6151 6152This capability indicates that userspace is able to apply a single VM-wide 6153offset to both the virtual and physical counters as viewed by the guest 6154using the KVM_ARM_SET_CNT_OFFSET ioctl and the following data structure: 6155 6156:: 6157 6158 struct kvm_arm_counter_offset { 6159 __u64 counter_offset; 6160 __u64 reserved; 6161 }; 6162 6163The offset describes a number of counter cycles that are subtracted from 6164both virtual and physical counter views (similar to the effects of the 6165CNTVOFF_EL2 and CNTPOFF_EL2 system registers, but only global). The offset 6166always applies to all vcpus (already created or created after this ioctl) 6167for this VM. 6168 6169It is userspace's responsibility to compute the offset based, for example, 6170on previous values of the guest counters. 6171 6172Any value other than 0 for the "reserved" field may result in an error 6173(-EINVAL) being returned. This ioctl can also return -EBUSY if any vcpu 6174ioctl is issued concurrently. 6175 6176Note that using this ioctl results in KVM ignoring subsequent userspace 6177writes to the CNTVCT_EL0 and CNTPCT_EL0 registers using the SET_ONE_REG 6178interface. No error will be returned, but the resulting offset will not be 6179applied. 6180 6181.. _KVM_ARM_GET_REG_WRITABLE_MASKS: 6182 61834.139 KVM_ARM_GET_REG_WRITABLE_MASKS 6184------------------------------------------- 6185 6186:Capability: KVM_CAP_ARM_SUPPORTED_REG_MASK_RANGES 6187:Architectures: arm64 6188:Type: vm ioctl 6189:Parameters: struct reg_mask_range (in/out) 6190:Returns: 0 on success, < 0 on error 6191 6192 6193:: 6194 6195 #define KVM_ARM_FEATURE_ID_RANGE 0 6196 #define KVM_ARM_FEATURE_ID_RANGE_SIZE (3 * 8 * 8) 6197 6198 struct reg_mask_range { 6199 __u64 addr; /* Pointer to mask array */ 6200 __u32 range; /* Requested range */ 6201 __u32 reserved[13]; 6202 }; 6203 6204This ioctl copies the writable masks for a selected range of registers to 6205userspace. 6206 6207The ``addr`` field is a pointer to the destination array where KVM copies 6208the writable masks. 6209 6210The ``range`` field indicates the requested range of registers. 6211``KVM_CHECK_EXTENSION`` for the ``KVM_CAP_ARM_SUPPORTED_REG_MASK_RANGES`` 6212capability returns the supported ranges, expressed as a set of flags. Each 6213flag's bit index represents a possible value for the ``range`` field. 6214All other values are reserved for future use and KVM may return an error. 6215 6216The ``reserved[13]`` array is reserved for future use and should be 0, or 6217KVM may return an error. 6218 6219KVM_ARM_FEATURE_ID_RANGE (0) 6220^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6221 6222The Feature ID range is defined as the AArch64 System register space with 6223op0==3, op1=={0, 1, 3}, CRn==0, CRm=={0-7}, op2=={0-7}. 6224 6225The mask returned array pointed to by ``addr`` is indexed by the macro 6226``ARM64_FEATURE_ID_RANGE_IDX(op0, op1, crn, crm, op2)``, allowing userspace 6227to know what fields can be changed for the system register described by 6228``op0, op1, crn, crm, op2``. KVM rejects ID register values that describe a 6229superset of the features supported by the system. 6230 62314.140 KVM_SET_USER_MEMORY_REGION2 6232--------------------------------- 6233 6234:Capability: KVM_CAP_USER_MEMORY2 6235:Architectures: all 6236:Type: vm ioctl 6237:Parameters: struct kvm_userspace_memory_region2 (in) 6238:Returns: 0 on success, -1 on error 6239 6240KVM_SET_USER_MEMORY_REGION2 is an extension to KVM_SET_USER_MEMORY_REGION that 6241allows mapping guest_memfd memory into a guest. All fields shared with 6242KVM_SET_USER_MEMORY_REGION identically. Userspace can set KVM_MEM_GUEST_MEMFD 6243in flags to have KVM bind the memory region to a given guest_memfd range of 6244[guest_memfd_offset, guest_memfd_offset + memory_size]. The target guest_memfd 6245must point at a file created via KVM_CREATE_GUEST_MEMFD on the current VM, and 6246the target range must not be bound to any other memory region. All standard 6247bounds checks apply (use common sense). 6248 6249:: 6250 6251 struct kvm_userspace_memory_region2 { 6252 __u32 slot; 6253 __u32 flags; 6254 __u64 guest_phys_addr; 6255 __u64 memory_size; /* bytes */ 6256 __u64 userspace_addr; /* start of the userspace allocated memory */ 6257 __u64 guest_memfd_offset; 6258 __u32 guest_memfd; 6259 __u32 pad1; 6260 __u64 pad2[14]; 6261 }; 6262 6263A KVM_MEM_GUEST_MEMFD region _must_ have a valid guest_memfd (private memory) and 6264userspace_addr (shared memory). However, "valid" for userspace_addr simply 6265means that the address itself must be a legal userspace address. The backing 6266mapping for userspace_addr is not required to be valid/populated at the time of 6267KVM_SET_USER_MEMORY_REGION2, e.g. shared memory can be lazily mapped/allocated 6268on-demand. 6269 6270When mapping a gfn into the guest, KVM selects shared vs. private, i.e consumes 6271userspace_addr vs. guest_memfd, based on the gfn's KVM_MEMORY_ATTRIBUTE_PRIVATE 6272state. At VM creation time, all memory is shared, i.e. the PRIVATE attribute 6273is '0' for all gfns. Userspace can control whether memory is shared/private by 6274toggling KVM_MEMORY_ATTRIBUTE_PRIVATE via KVM_SET_MEMORY_ATTRIBUTES as needed. 6275 62764.141 KVM_SET_MEMORY_ATTRIBUTES 6277------------------------------- 6278 6279:Capability: KVM_CAP_MEMORY_ATTRIBUTES 6280:Architectures: x86 6281:Type: vm ioctl 6282:Parameters: struct kvm_memory_attributes (in) 6283:Returns: 0 on success, <0 on error 6284 6285KVM_SET_MEMORY_ATTRIBUTES allows userspace to set memory attributes for a range 6286of guest physical memory. 6287 6288:: 6289 6290 struct kvm_memory_attributes { 6291 __u64 address; 6292 __u64 size; 6293 __u64 attributes; 6294 __u64 flags; 6295 }; 6296 6297 #define KVM_MEMORY_ATTRIBUTE_PRIVATE (1ULL << 3) 6298 6299The address and size must be page aligned. The supported attributes can be 6300retrieved via ioctl(KVM_CHECK_EXTENSION) on KVM_CAP_MEMORY_ATTRIBUTES. If 6301executed on a VM, KVM_CAP_MEMORY_ATTRIBUTES precisely returns the attributes 6302supported by that VM. If executed at system scope, KVM_CAP_MEMORY_ATTRIBUTES 6303returns all attributes supported by KVM. The only attribute defined at this 6304time is KVM_MEMORY_ATTRIBUTE_PRIVATE, which marks the associated gfn as being 6305guest private memory. 6306 6307Note, there is no "get" API. Userspace is responsible for explicitly tracking 6308the state of a gfn/page as needed. 6309 6310The "flags" field is reserved for future extensions and must be '0'. 6311 63124.142 KVM_CREATE_GUEST_MEMFD 6313---------------------------- 6314 6315:Capability: KVM_CAP_GUEST_MEMFD 6316:Architectures: none 6317:Type: vm ioctl 6318:Parameters: struct kvm_create_guest_memfd(in) 6319:Returns: A file descriptor on success, <0 on error 6320 6321KVM_CREATE_GUEST_MEMFD creates an anonymous file and returns a file descriptor 6322that refers to it. guest_memfd files are roughly analogous to files created 6323via memfd_create(), e.g. guest_memfd files live in RAM, have volatile storage, 6324and are automatically released when the last reference is dropped. Unlike 6325"regular" memfd_create() files, guest_memfd files are bound to their owning 6326virtual machine (see below), cannot be mapped, read, or written by userspace, 6327and cannot be resized (guest_memfd files do however support PUNCH_HOLE). 6328 6329:: 6330 6331 struct kvm_create_guest_memfd { 6332 __u64 size; 6333 __u64 flags; 6334 __u64 reserved[6]; 6335 }; 6336 6337Conceptually, the inode backing a guest_memfd file represents physical memory, 6338i.e. is coupled to the virtual machine as a thing, not to a "struct kvm". The 6339file itself, which is bound to a "struct kvm", is that instance's view of the 6340underlying memory, e.g. effectively provides the translation of guest addresses 6341to host memory. This allows for use cases where multiple KVM structures are 6342used to manage a single virtual machine, e.g. when performing intrahost 6343migration of a virtual machine. 6344 6345KVM currently only supports mapping guest_memfd via KVM_SET_USER_MEMORY_REGION2, 6346and more specifically via the guest_memfd and guest_memfd_offset fields in 6347"struct kvm_userspace_memory_region2", where guest_memfd_offset is the offset 6348into the guest_memfd instance. For a given guest_memfd file, there can be at 6349most one mapping per page, i.e. binding multiple memory regions to a single 6350guest_memfd range is not allowed (any number of memory regions can be bound to 6351a single guest_memfd file, but the bound ranges must not overlap). 6352 6353See KVM_SET_USER_MEMORY_REGION2 for additional details. 6354 63555. The kvm_run structure 6356======================== 6357 6358Application code obtains a pointer to the kvm_run structure by 6359mmap()ing a vcpu fd. From that point, application code can control 6360execution by changing fields in kvm_run prior to calling the KVM_RUN 6361ioctl, and obtain information about the reason KVM_RUN returned by 6362looking up structure members. 6363 6364:: 6365 6366 struct kvm_run { 6367 /* in */ 6368 __u8 request_interrupt_window; 6369 6370Request that KVM_RUN return when it becomes possible to inject external 6371interrupts into the guest. Useful in conjunction with KVM_INTERRUPT. 6372 6373:: 6374 6375 __u8 immediate_exit; 6376 6377This field is polled once when KVM_RUN starts; if non-zero, KVM_RUN 6378exits immediately, returning -EINTR. In the common scenario where a 6379signal is used to "kick" a VCPU out of KVM_RUN, this field can be used 6380to avoid usage of KVM_SET_SIGNAL_MASK, which has worse scalability. 6381Rather than blocking the signal outside KVM_RUN, userspace can set up 6382a signal handler that sets run->immediate_exit to a non-zero value. 6383 6384This field is ignored if KVM_CAP_IMMEDIATE_EXIT is not available. 6385 6386:: 6387 6388 __u8 padding1[6]; 6389 6390 /* out */ 6391 __u32 exit_reason; 6392 6393When KVM_RUN has returned successfully (return value 0), this informs 6394application code why KVM_RUN has returned. Allowable values for this 6395field are detailed below. 6396 6397:: 6398 6399 __u8 ready_for_interrupt_injection; 6400 6401If request_interrupt_window has been specified, this field indicates 6402an interrupt can be injected now with KVM_INTERRUPT. 6403 6404:: 6405 6406 __u8 if_flag; 6407 6408The value of the current interrupt flag. Only valid if in-kernel 6409local APIC is not used. 6410 6411:: 6412 6413 __u16 flags; 6414 6415More architecture-specific flags detailing state of the VCPU that may 6416affect the device's behavior. Current defined flags:: 6417 6418 /* x86, set if the VCPU is in system management mode */ 6419 #define KVM_RUN_X86_SMM (1 << 0) 6420 /* x86, set if bus lock detected in VM */ 6421 #define KVM_RUN_BUS_LOCK (1 << 1) 6422 /* arm64, set for KVM_EXIT_DEBUG */ 6423 #define KVM_DEBUG_ARCH_HSR_HIGH_VALID (1 << 0) 6424 6425:: 6426 6427 /* in (pre_kvm_run), out (post_kvm_run) */ 6428 __u64 cr8; 6429 6430The value of the cr8 register. Only valid if in-kernel local APIC is 6431not used. Both input and output. 6432 6433:: 6434 6435 __u64 apic_base; 6436 6437The value of the APIC BASE msr. Only valid if in-kernel local 6438APIC is not used. Both input and output. 6439 6440:: 6441 6442 union { 6443 /* KVM_EXIT_UNKNOWN */ 6444 struct { 6445 __u64 hardware_exit_reason; 6446 } hw; 6447 6448If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown 6449reasons. Further architecture-specific information is available in 6450hardware_exit_reason. 6451 6452:: 6453 6454 /* KVM_EXIT_FAIL_ENTRY */ 6455 struct { 6456 __u64 hardware_entry_failure_reason; 6457 __u32 cpu; /* if KVM_LAST_CPU */ 6458 } fail_entry; 6459 6460If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due 6461to unknown reasons. Further architecture-specific information is 6462available in hardware_entry_failure_reason. 6463 6464:: 6465 6466 /* KVM_EXIT_EXCEPTION */ 6467 struct { 6468 __u32 exception; 6469 __u32 error_code; 6470 } ex; 6471 6472Unused. 6473 6474:: 6475 6476 /* KVM_EXIT_IO */ 6477 struct { 6478 #define KVM_EXIT_IO_IN 0 6479 #define KVM_EXIT_IO_OUT 1 6480 __u8 direction; 6481 __u8 size; /* bytes */ 6482 __u16 port; 6483 __u32 count; 6484 __u64 data_offset; /* relative to kvm_run start */ 6485 } io; 6486 6487If exit_reason is KVM_EXIT_IO, then the vcpu has 6488executed a port I/O instruction which could not be satisfied by kvm. 6489data_offset describes where the data is located (KVM_EXIT_IO_OUT) or 6490where kvm expects application code to place the data for the next 6491KVM_RUN invocation (KVM_EXIT_IO_IN). Data format is a packed array. 6492 6493:: 6494 6495 /* KVM_EXIT_DEBUG */ 6496 struct { 6497 struct kvm_debug_exit_arch arch; 6498 } debug; 6499 6500If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event 6501for which architecture specific information is returned. 6502 6503:: 6504 6505 /* KVM_EXIT_MMIO */ 6506 struct { 6507 __u64 phys_addr; 6508 __u8 data[8]; 6509 __u32 len; 6510 __u8 is_write; 6511 } mmio; 6512 6513If exit_reason is KVM_EXIT_MMIO, then the vcpu has 6514executed a memory-mapped I/O instruction which could not be satisfied 6515by kvm. The 'data' member contains the written data if 'is_write' is 6516true, and should be filled by application code otherwise. 6517 6518The 'data' member contains, in its first 'len' bytes, the value as it would 6519appear if the VCPU performed a load or store of the appropriate width directly 6520to the byte array. 6521 6522.. note:: 6523 6524 For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR, KVM_EXIT_XEN, 6525 KVM_EXIT_EPR, KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR the corresponding 6526 operations are complete (and guest state is consistent) only after userspace 6527 has re-entered the kernel with KVM_RUN. The kernel side will first finish 6528 incomplete operations and then check for pending signals. 6529 6530 The pending state of the operation is not preserved in state which is 6531 visible to userspace, thus userspace should ensure that the operation is 6532 completed before performing a live migration. Userspace can re-enter the 6533 guest with an unmasked signal pending or with the immediate_exit field set 6534 to complete pending operations without allowing any further instructions 6535 to be executed. 6536 6537:: 6538 6539 /* KVM_EXIT_HYPERCALL */ 6540 struct { 6541 __u64 nr; 6542 __u64 args[6]; 6543 __u64 ret; 6544 __u64 flags; 6545 } hypercall; 6546 6547 6548It is strongly recommended that userspace use ``KVM_EXIT_IO`` (x86) or 6549``KVM_EXIT_MMIO`` (all except s390) to implement functionality that 6550requires a guest to interact with host userspace. 6551 6552.. note:: KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO. 6553 6554For arm64: 6555---------- 6556 6557SMCCC exits can be enabled depending on the configuration of the SMCCC 6558filter. See the Documentation/virt/kvm/devices/vm.rst 6559``KVM_ARM_SMCCC_FILTER`` for more details. 6560 6561``nr`` contains the function ID of the guest's SMCCC call. Userspace is 6562expected to use the ``KVM_GET_ONE_REG`` ioctl to retrieve the call 6563parameters from the vCPU's GPRs. 6564 6565Definition of ``flags``: 6566 - ``KVM_HYPERCALL_EXIT_SMC``: Indicates that the guest used the SMC 6567 conduit to initiate the SMCCC call. If this bit is 0 then the guest 6568 used the HVC conduit for the SMCCC call. 6569 6570 - ``KVM_HYPERCALL_EXIT_16BIT``: Indicates that the guest used a 16bit 6571 instruction to initiate the SMCCC call. If this bit is 0 then the 6572 guest used a 32bit instruction. An AArch64 guest always has this 6573 bit set to 0. 6574 6575At the point of exit, PC points to the instruction immediately following 6576the trapping instruction. 6577 6578:: 6579 6580 /* KVM_EXIT_TPR_ACCESS */ 6581 struct { 6582 __u64 rip; 6583 __u32 is_write; 6584 __u32 pad; 6585 } tpr_access; 6586 6587To be documented (KVM_TPR_ACCESS_REPORTING). 6588 6589:: 6590 6591 /* KVM_EXIT_S390_SIEIC */ 6592 struct { 6593 __u8 icptcode; 6594 __u64 mask; /* psw upper half */ 6595 __u64 addr; /* psw lower half */ 6596 __u16 ipa; 6597 __u32 ipb; 6598 } s390_sieic; 6599 6600s390 specific. 6601 6602:: 6603 6604 /* KVM_EXIT_S390_RESET */ 6605 #define KVM_S390_RESET_POR 1 6606 #define KVM_S390_RESET_CLEAR 2 6607 #define KVM_S390_RESET_SUBSYSTEM 4 6608 #define KVM_S390_RESET_CPU_INIT 8 6609 #define KVM_S390_RESET_IPL 16 6610 __u64 s390_reset_flags; 6611 6612s390 specific. 6613 6614:: 6615 6616 /* KVM_EXIT_S390_UCONTROL */ 6617 struct { 6618 __u64 trans_exc_code; 6619 __u32 pgm_code; 6620 } s390_ucontrol; 6621 6622s390 specific. A page fault has occurred for a user controlled virtual 6623machine (KVM_VM_S390_UNCONTROL) on its host page table that cannot be 6624resolved by the kernel. 6625The program code and the translation exception code that were placed 6626in the cpu's lowcore are presented here as defined by the z Architecture 6627Principles of Operation Book in the Chapter for Dynamic Address Translation 6628(DAT) 6629 6630:: 6631 6632 /* KVM_EXIT_DCR */ 6633 struct { 6634 __u32 dcrn; 6635 __u32 data; 6636 __u8 is_write; 6637 } dcr; 6638 6639Deprecated - was used for 440 KVM. 6640 6641:: 6642 6643 /* KVM_EXIT_OSI */ 6644 struct { 6645 __u64 gprs[32]; 6646 } osi; 6647 6648MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch 6649hypercalls and exit with this exit struct that contains all the guest gprs. 6650 6651If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall. 6652Userspace can now handle the hypercall and when it's done modify the gprs as 6653necessary. Upon guest entry all guest GPRs will then be replaced by the values 6654in this struct. 6655 6656:: 6657 6658 /* KVM_EXIT_PAPR_HCALL */ 6659 struct { 6660 __u64 nr; 6661 __u64 ret; 6662 __u64 args[9]; 6663 } papr_hcall; 6664 6665This is used on 64-bit PowerPC when emulating a pSeries partition, 6666e.g. with the 'pseries' machine type in qemu. It occurs when the 6667guest does a hypercall using the 'sc 1' instruction. The 'nr' field 6668contains the hypercall number (from the guest R3), and 'args' contains 6669the arguments (from the guest R4 - R12). Userspace should put the 6670return code in 'ret' and any extra returned values in args[]. 6671The possible hypercalls are defined in the Power Architecture Platform 6672Requirements (PAPR) document available from www.power.org (free 6673developer registration required to access it). 6674 6675:: 6676 6677 /* KVM_EXIT_S390_TSCH */ 6678 struct { 6679 __u16 subchannel_id; 6680 __u16 subchannel_nr; 6681 __u32 io_int_parm; 6682 __u32 io_int_word; 6683 __u32 ipb; 6684 __u8 dequeued; 6685 } s390_tsch; 6686 6687s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled 6688and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O 6689interrupt for the target subchannel has been dequeued and subchannel_id, 6690subchannel_nr, io_int_parm and io_int_word contain the parameters for that 6691interrupt. ipb is needed for instruction parameter decoding. 6692 6693:: 6694 6695 /* KVM_EXIT_EPR */ 6696 struct { 6697 __u32 epr; 6698 } epr; 6699 6700On FSL BookE PowerPC chips, the interrupt controller has a fast patch 6701interrupt acknowledge path to the core. When the core successfully 6702delivers an interrupt, it automatically populates the EPR register with 6703the interrupt vector number and acknowledges the interrupt inside 6704the interrupt controller. 6705 6706In case the interrupt controller lives in user space, we need to do 6707the interrupt acknowledge cycle through it to fetch the next to be 6708delivered interrupt vector using this exit. 6709 6710It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an 6711external interrupt has just been delivered into the guest. User space 6712should put the acknowledged interrupt vector into the 'epr' field. 6713 6714:: 6715 6716 /* KVM_EXIT_SYSTEM_EVENT */ 6717 struct { 6718 #define KVM_SYSTEM_EVENT_SHUTDOWN 1 6719 #define KVM_SYSTEM_EVENT_RESET 2 6720 #define KVM_SYSTEM_EVENT_CRASH 3 6721 #define KVM_SYSTEM_EVENT_WAKEUP 4 6722 #define KVM_SYSTEM_EVENT_SUSPEND 5 6723 #define KVM_SYSTEM_EVENT_SEV_TERM 6 6724 __u32 type; 6725 __u32 ndata; 6726 __u64 data[16]; 6727 } system_event; 6728 6729If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered 6730a system-level event using some architecture specific mechanism (hypercall 6731or some special instruction). In case of ARM64, this is triggered using 6732HVC instruction based PSCI call from the vcpu. 6733 6734The 'type' field describes the system-level event type. 6735Valid values for 'type' are: 6736 6737 - KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the 6738 VM. Userspace is not obliged to honour this, and if it does honour 6739 this does not need to destroy the VM synchronously (ie it may call 6740 KVM_RUN again before shutdown finally occurs). 6741 - KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM. 6742 As with SHUTDOWN, userspace can choose to ignore the request, or 6743 to schedule the reset to occur in the future and may call KVM_RUN again. 6744 - KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest 6745 has requested a crash condition maintenance. Userspace can choose 6746 to ignore the request, or to gather VM memory core dump and/or 6747 reset/shutdown of the VM. 6748 - KVM_SYSTEM_EVENT_SEV_TERM -- an AMD SEV guest requested termination. 6749 The guest physical address of the guest's GHCB is stored in `data[0]`. 6750 - KVM_SYSTEM_EVENT_WAKEUP -- the exiting vCPU is in a suspended state and 6751 KVM has recognized a wakeup event. Userspace may honor this event by 6752 marking the exiting vCPU as runnable, or deny it and call KVM_RUN again. 6753 - KVM_SYSTEM_EVENT_SUSPEND -- the guest has requested a suspension of 6754 the VM. 6755 6756If KVM_CAP_SYSTEM_EVENT_DATA is present, the 'data' field can contain 6757architecture specific information for the system-level event. Only 6758the first `ndata` items (possibly zero) of the data array are valid. 6759 6760 - for arm64, data[0] is set to KVM_SYSTEM_EVENT_RESET_FLAG_PSCI_RESET2 if 6761 the guest issued a SYSTEM_RESET2 call according to v1.1 of the PSCI 6762 specification. 6763 6764 - for RISC-V, data[0] is set to the value of the second argument of the 6765 ``sbi_system_reset`` call. 6766 6767Previous versions of Linux defined a `flags` member in this struct. The 6768field is now aliased to `data[0]`. Userspace can assume that it is only 6769written if ndata is greater than 0. 6770 6771For arm/arm64: 6772-------------- 6773 6774KVM_SYSTEM_EVENT_SUSPEND exits are enabled with the 6775KVM_CAP_ARM_SYSTEM_SUSPEND VM capability. If a guest invokes the PSCI 6776SYSTEM_SUSPEND function, KVM will exit to userspace with this event 6777type. 6778 6779It is the sole responsibility of userspace to implement the PSCI 6780SYSTEM_SUSPEND call according to ARM DEN0022D.b 5.19 "SYSTEM_SUSPEND". 6781KVM does not change the vCPU's state before exiting to userspace, so 6782the call parameters are left in-place in the vCPU registers. 6783 6784Userspace is _required_ to take action for such an exit. It must 6785either: 6786 6787 - Honor the guest request to suspend the VM. Userspace can request 6788 in-kernel emulation of suspension by setting the calling vCPU's 6789 state to KVM_MP_STATE_SUSPENDED. Userspace must configure the vCPU's 6790 state according to the parameters passed to the PSCI function when 6791 the calling vCPU is resumed. See ARM DEN0022D.b 5.19.1 "Intended use" 6792 for details on the function parameters. 6793 6794 - Deny the guest request to suspend the VM. See ARM DEN0022D.b 5.19.2 6795 "Caller responsibilities" for possible return values. 6796 6797:: 6798 6799 /* KVM_EXIT_IOAPIC_EOI */ 6800 struct { 6801 __u8 vector; 6802 } eoi; 6803 6804Indicates that the VCPU's in-kernel local APIC received an EOI for a 6805level-triggered IOAPIC interrupt. This exit only triggers when the 6806IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled); 6807the userspace IOAPIC should process the EOI and retrigger the interrupt if 6808it is still asserted. Vector is the LAPIC interrupt vector for which the 6809EOI was received. 6810 6811:: 6812 6813 struct kvm_hyperv_exit { 6814 #define KVM_EXIT_HYPERV_SYNIC 1 6815 #define KVM_EXIT_HYPERV_HCALL 2 6816 #define KVM_EXIT_HYPERV_SYNDBG 3 6817 __u32 type; 6818 __u32 pad1; 6819 union { 6820 struct { 6821 __u32 msr; 6822 __u32 pad2; 6823 __u64 control; 6824 __u64 evt_page; 6825 __u64 msg_page; 6826 } synic; 6827 struct { 6828 __u64 input; 6829 __u64 result; 6830 __u64 params[2]; 6831 } hcall; 6832 struct { 6833 __u32 msr; 6834 __u32 pad2; 6835 __u64 control; 6836 __u64 status; 6837 __u64 send_page; 6838 __u64 recv_page; 6839 __u64 pending_page; 6840 } syndbg; 6841 } u; 6842 }; 6843 /* KVM_EXIT_HYPERV */ 6844 struct kvm_hyperv_exit hyperv; 6845 6846Indicates that the VCPU exits into userspace to process some tasks 6847related to Hyper-V emulation. 6848 6849Valid values for 'type' are: 6850 6851 - KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about 6852 6853Hyper-V SynIC state change. Notification is used to remap SynIC 6854event/message pages and to enable/disable SynIC messages/events processing 6855in userspace. 6856 6857 - KVM_EXIT_HYPERV_SYNDBG -- synchronously notify user-space about 6858 6859Hyper-V Synthetic debugger state change. Notification is used to either update 6860the pending_page location or to send a control command (send the buffer located 6861in send_page or recv a buffer to recv_page). 6862 6863:: 6864 6865 /* KVM_EXIT_ARM_NISV */ 6866 struct { 6867 __u64 esr_iss; 6868 __u64 fault_ipa; 6869 } arm_nisv; 6870 6871Used on arm64 systems. If a guest accesses memory not in a memslot, 6872KVM will typically return to userspace and ask it to do MMIO emulation on its 6873behalf. However, for certain classes of instructions, no instruction decode 6874(direction, length of memory access) is provided, and fetching and decoding 6875the instruction from the VM is overly complicated to live in the kernel. 6876 6877Historically, when this situation occurred, KVM would print a warning and kill 6878the VM. KVM assumed that if the guest accessed non-memslot memory, it was 6879trying to do I/O, which just couldn't be emulated, and the warning message was 6880phrased accordingly. However, what happened more often was that a guest bug 6881caused access outside the guest memory areas which should lead to a more 6882meaningful warning message and an external abort in the guest, if the access 6883did not fall within an I/O window. 6884 6885Userspace implementations can query for KVM_CAP_ARM_NISV_TO_USER, and enable 6886this capability at VM creation. Once this is done, these types of errors will 6887instead return to userspace with KVM_EXIT_ARM_NISV, with the valid bits from 6888the ESR_EL2 in the esr_iss field, and the faulting IPA in the fault_ipa field. 6889Userspace can either fix up the access if it's actually an I/O access by 6890decoding the instruction from guest memory (if it's very brave) and continue 6891executing the guest, or it can decide to suspend, dump, or restart the guest. 6892 6893Note that KVM does not skip the faulting instruction as it does for 6894KVM_EXIT_MMIO, but userspace has to emulate any change to the processing state 6895if it decides to decode and emulate the instruction. 6896 6897This feature isn't available to protected VMs, as userspace does not 6898have access to the state that is required to perform the emulation. 6899Instead, a data abort exception is directly injected in the guest. 6900Note that although KVM_CAP_ARM_NISV_TO_USER will be reported if 6901queried outside of a protected VM context, the feature will not be 6902exposed if queried on a protected VM file descriptor. 6903 6904:: 6905 6906 /* KVM_EXIT_X86_RDMSR / KVM_EXIT_X86_WRMSR */ 6907 struct { 6908 __u8 error; /* user -> kernel */ 6909 __u8 pad[7]; 6910 __u32 reason; /* kernel -> user */ 6911 __u32 index; /* kernel -> user */ 6912 __u64 data; /* kernel <-> user */ 6913 } msr; 6914 6915Used on x86 systems. When the VM capability KVM_CAP_X86_USER_SPACE_MSR is 6916enabled, MSR accesses to registers that would invoke a #GP by KVM kernel code 6917may instead trigger a KVM_EXIT_X86_RDMSR exit for reads and KVM_EXIT_X86_WRMSR 6918exit for writes. 6919 6920The "reason" field specifies why the MSR interception occurred. Userspace will 6921only receive MSR exits when a particular reason was requested during through 6922ENABLE_CAP. Currently valid exit reasons are: 6923 6924============================ ======================================== 6925 KVM_MSR_EXIT_REASON_UNKNOWN access to MSR that is unknown to KVM 6926 KVM_MSR_EXIT_REASON_INVAL access to invalid MSRs or reserved bits 6927 KVM_MSR_EXIT_REASON_FILTER access blocked by KVM_X86_SET_MSR_FILTER 6928============================ ======================================== 6929 6930For KVM_EXIT_X86_RDMSR, the "index" field tells userspace which MSR the guest 6931wants to read. To respond to this request with a successful read, userspace 6932writes the respective data into the "data" field and must continue guest 6933execution to ensure the read data is transferred into guest register state. 6934 6935If the RDMSR request was unsuccessful, userspace indicates that with a "1" in 6936the "error" field. This will inject a #GP into the guest when the VCPU is 6937executed again. 6938 6939For KVM_EXIT_X86_WRMSR, the "index" field tells userspace which MSR the guest 6940wants to write. Once finished processing the event, userspace must continue 6941vCPU execution. If the MSR write was unsuccessful, userspace also sets the 6942"error" field to "1". 6943 6944See KVM_X86_SET_MSR_FILTER for details on the interaction with MSR filtering. 6945 6946:: 6947 6948 6949 struct kvm_xen_exit { 6950 #define KVM_EXIT_XEN_HCALL 1 6951 __u32 type; 6952 union { 6953 struct { 6954 __u32 longmode; 6955 __u32 cpl; 6956 __u64 input; 6957 __u64 result; 6958 __u64 params[6]; 6959 } hcall; 6960 } u; 6961 }; 6962 /* KVM_EXIT_XEN */ 6963 struct kvm_hyperv_exit xen; 6964 6965Indicates that the VCPU exits into userspace to process some tasks 6966related to Xen emulation. 6967 6968Valid values for 'type' are: 6969 6970 - KVM_EXIT_XEN_HCALL -- synchronously notify user-space about Xen hypercall. 6971 Userspace is expected to place the hypercall result into the appropriate 6972 field before invoking KVM_RUN again. 6973 6974:: 6975 6976 /* KVM_EXIT_RISCV_SBI */ 6977 struct { 6978 unsigned long extension_id; 6979 unsigned long function_id; 6980 unsigned long args[6]; 6981 unsigned long ret[2]; 6982 } riscv_sbi; 6983 6984If exit reason is KVM_EXIT_RISCV_SBI then it indicates that the VCPU has 6985done a SBI call which is not handled by KVM RISC-V kernel module. The details 6986of the SBI call are available in 'riscv_sbi' member of kvm_run structure. The 6987'extension_id' field of 'riscv_sbi' represents SBI extension ID whereas the 6988'function_id' field represents function ID of given SBI extension. The 'args' 6989array field of 'riscv_sbi' represents parameters for the SBI call and 'ret' 6990array field represents return values. The userspace should update the return 6991values of SBI call before resuming the VCPU. For more details on RISC-V SBI 6992spec refer, https://github.com/riscv/riscv-sbi-doc. 6993 6994:: 6995 6996 /* KVM_EXIT_MEMORY_FAULT */ 6997 struct { 6998 #define KVM_MEMORY_EXIT_FLAG_PRIVATE (1ULL << 3) 6999 __u64 flags; 7000 __u64 gpa; 7001 __u64 size; 7002 } memory_fault; 7003 7004KVM_EXIT_MEMORY_FAULT indicates the vCPU has encountered a memory fault that 7005could not be resolved by KVM. The 'gpa' and 'size' (in bytes) describe the 7006guest physical address range [gpa, gpa + size) of the fault. The 'flags' field 7007describes properties of the faulting access that are likely pertinent: 7008 7009 - KVM_MEMORY_EXIT_FLAG_PRIVATE - When set, indicates the memory fault occurred 7010 on a private memory access. When clear, indicates the fault occurred on a 7011 shared access. 7012 7013Note! KVM_EXIT_MEMORY_FAULT is unique among all KVM exit reasons in that it 7014accompanies a return code of '-1', not '0'! errno will always be set to EFAULT 7015or EHWPOISON when KVM exits with KVM_EXIT_MEMORY_FAULT, userspace should assume 7016kvm_run.exit_reason is stale/undefined for all other error numbers. 7017 7018:: 7019 7020 /* KVM_EXIT_NOTIFY */ 7021 struct { 7022 #define KVM_NOTIFY_CONTEXT_INVALID (1 << 0) 7023 __u32 flags; 7024 } notify; 7025 7026Used on x86 systems. When the VM capability KVM_CAP_X86_NOTIFY_VMEXIT is 7027enabled, a VM exit generated if no event window occurs in VM non-root mode 7028for a specified amount of time. Once KVM_X86_NOTIFY_VMEXIT_USER is set when 7029enabling the cap, it would exit to userspace with the exit reason 7030KVM_EXIT_NOTIFY for further handling. The "flags" field contains more 7031detailed info. 7032 7033The valid value for 'flags' is: 7034 7035 - KVM_NOTIFY_CONTEXT_INVALID -- the VM context is corrupted and not valid 7036 in VMCS. It would run into unknown result if resume the target VM. 7037 7038:: 7039 7040 /* Fix the size of the union. */ 7041 char padding[256]; 7042 }; 7043 7044 /* 7045 * shared registers between kvm and userspace. 7046 * kvm_valid_regs specifies the register classes set by the host 7047 * kvm_dirty_regs specified the register classes dirtied by userspace 7048 * struct kvm_sync_regs is architecture specific, as well as the 7049 * bits for kvm_valid_regs and kvm_dirty_regs 7050 */ 7051 __u64 kvm_valid_regs; 7052 __u64 kvm_dirty_regs; 7053 union { 7054 struct kvm_sync_regs regs; 7055 char padding[SYNC_REGS_SIZE_BYTES]; 7056 } s; 7057 7058If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access 7059certain guest registers without having to call SET/GET_*REGS. Thus we can 7060avoid some system call overhead if userspace has to handle the exit. 7061Userspace can query the validity of the structure by checking 7062kvm_valid_regs for specific bits. These bits are architecture specific 7063and usually define the validity of a groups of registers. (e.g. one bit 7064for general purpose registers) 7065 7066Please note that the kernel is allowed to use the kvm_run structure as the 7067primary storage for certain register types. Therefore, the kernel may use the 7068values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set. 7069 7070 70716. Capabilities that can be enabled on vCPUs 7072============================================ 7073 7074There are certain capabilities that change the behavior of the virtual CPU or 7075the virtual machine when enabled. To enable them, please see section 4.37. 7076Below you can find a list of capabilities and what their effect on the vCPU or 7077the virtual machine is when enabling them. 7078 7079The following information is provided along with the description: 7080 7081 Architectures: 7082 which instruction set architectures provide this ioctl. 7083 x86 includes both i386 and x86_64. 7084 7085 Target: 7086 whether this is a per-vcpu or per-vm capability. 7087 7088 Parameters: 7089 what parameters are accepted by the capability. 7090 7091 Returns: 7092 the return value. General error numbers (EBADF, ENOMEM, EINVAL) 7093 are not detailed, but errors with specific meanings are. 7094 7095 70966.1 KVM_CAP_PPC_OSI 7097------------------- 7098 7099:Architectures: ppc 7100:Target: vcpu 7101:Parameters: none 7102:Returns: 0 on success; -1 on error 7103 7104This capability enables interception of OSI hypercalls that otherwise would 7105be treated as normal system calls to be injected into the guest. OSI hypercalls 7106were invented by Mac-on-Linux to have a standardized communication mechanism 7107between the guest and the host. 7108 7109When this capability is enabled, KVM_EXIT_OSI can occur. 7110 7111 71126.2 KVM_CAP_PPC_PAPR 7113-------------------- 7114 7115:Architectures: ppc 7116:Target: vcpu 7117:Parameters: none 7118:Returns: 0 on success; -1 on error 7119 7120This capability enables interception of PAPR hypercalls. PAPR hypercalls are 7121done using the hypercall instruction "sc 1". 7122 7123It also sets the guest privilege level to "supervisor" mode. Usually the guest 7124runs in "hypervisor" privilege mode with a few missing features. 7125 7126In addition to the above, it changes the semantics of SDR1. In this mode, the 7127HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the 7128HTAB invisible to the guest. 7129 7130When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur. 7131 7132 71336.3 KVM_CAP_SW_TLB 7134------------------ 7135 7136:Architectures: ppc 7137:Target: vcpu 7138:Parameters: args[0] is the address of a struct kvm_config_tlb 7139:Returns: 0 on success; -1 on error 7140 7141:: 7142 7143 struct kvm_config_tlb { 7144 __u64 params; 7145 __u64 array; 7146 __u32 mmu_type; 7147 __u32 array_len; 7148 }; 7149 7150Configures the virtual CPU's TLB array, establishing a shared memory area 7151between userspace and KVM. The "params" and "array" fields are userspace 7152addresses of mmu-type-specific data structures. The "array_len" field is an 7153safety mechanism, and should be set to the size in bytes of the memory that 7154userspace has reserved for the array. It must be at least the size dictated 7155by "mmu_type" and "params". 7156 7157While KVM_RUN is active, the shared region is under control of KVM. Its 7158contents are undefined, and any modification by userspace results in 7159boundedly undefined behavior. 7160 7161On return from KVM_RUN, the shared region will reflect the current state of 7162the guest's TLB. If userspace makes any changes, it must call KVM_DIRTY_TLB 7163to tell KVM which entries have been changed, prior to calling KVM_RUN again 7164on this vcpu. 7165 7166For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV: 7167 7168 - The "params" field is of type "struct kvm_book3e_206_tlb_params". 7169 - The "array" field points to an array of type "struct 7170 kvm_book3e_206_tlb_entry". 7171 - The array consists of all entries in the first TLB, followed by all 7172 entries in the second TLB. 7173 - Within a TLB, entries are ordered first by increasing set number. Within a 7174 set, entries are ordered by way (increasing ESEL). 7175 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1) 7176 where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value. 7177 - The tsize field of mas1 shall be set to 4K on TLB0, even though the 7178 hardware ignores this value for TLB0. 7179 71806.4 KVM_CAP_S390_CSS_SUPPORT 7181---------------------------- 7182 7183:Architectures: s390 7184:Target: vcpu 7185:Parameters: none 7186:Returns: 0 on success; -1 on error 7187 7188This capability enables support for handling of channel I/O instructions. 7189 7190TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are 7191handled in-kernel, while the other I/O instructions are passed to userspace. 7192 7193When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST 7194SUBCHANNEL intercepts. 7195 7196Note that even though this capability is enabled per-vcpu, the complete 7197virtual machine is affected. 7198 71996.5 KVM_CAP_PPC_EPR 7200------------------- 7201 7202:Architectures: ppc 7203:Target: vcpu 7204:Parameters: args[0] defines whether the proxy facility is active 7205:Returns: 0 on success; -1 on error 7206 7207This capability enables or disables the delivery of interrupts through the 7208external proxy facility. 7209 7210When enabled (args[0] != 0), every time the guest gets an external interrupt 7211delivered, it automatically exits into user space with a KVM_EXIT_EPR exit 7212to receive the topmost interrupt vector. 7213 7214When disabled (args[0] == 0), behavior is as if this facility is unsupported. 7215 7216When this capability is enabled, KVM_EXIT_EPR can occur. 7217 72186.6 KVM_CAP_IRQ_MPIC 7219-------------------- 7220 7221:Architectures: ppc 7222:Parameters: args[0] is the MPIC device fd; 7223 args[1] is the MPIC CPU number for this vcpu 7224 7225This capability connects the vcpu to an in-kernel MPIC device. 7226 72276.7 KVM_CAP_IRQ_XICS 7228-------------------- 7229 7230:Architectures: ppc 7231:Target: vcpu 7232:Parameters: args[0] is the XICS device fd; 7233 args[1] is the XICS CPU number (server ID) for this vcpu 7234 7235This capability connects the vcpu to an in-kernel XICS device. 7236 72376.8 KVM_CAP_S390_IRQCHIP 7238------------------------ 7239 7240:Architectures: s390 7241:Target: vm 7242:Parameters: none 7243 7244This capability enables the in-kernel irqchip for s390. Please refer to 7245"4.24 KVM_CREATE_IRQCHIP" for details. 7246 72476.9 KVM_CAP_MIPS_FPU 7248-------------------- 7249 7250:Architectures: mips 7251:Target: vcpu 7252:Parameters: args[0] is reserved for future use (should be 0). 7253 7254This capability allows the use of the host Floating Point Unit by the guest. It 7255allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is 7256done the ``KVM_REG_MIPS_FPR_*`` and ``KVM_REG_MIPS_FCR_*`` registers can be 7257accessed (depending on the current guest FPU register mode), and the Status.FR, 7258Config5.FRE bits are accessible via the KVM API and also from the guest, 7259depending on them being supported by the FPU. 7260 72616.10 KVM_CAP_MIPS_MSA 7262--------------------- 7263 7264:Architectures: mips 7265:Target: vcpu 7266:Parameters: args[0] is reserved for future use (should be 0). 7267 7268This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest. 7269It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest. 7270Once this is done the ``KVM_REG_MIPS_VEC_*`` and ``KVM_REG_MIPS_MSA_*`` 7271registers can be accessed, and the Config5.MSAEn bit is accessible via the 7272KVM API and also from the guest. 7273 72746.74 KVM_CAP_SYNC_REGS 7275---------------------- 7276 7277:Architectures: s390, x86 7278:Target: s390: always enabled, x86: vcpu 7279:Parameters: none 7280:Returns: x86: KVM_CHECK_EXTENSION returns a bit-array indicating which register 7281 sets are supported 7282 (bitfields defined in arch/x86/include/uapi/asm/kvm.h). 7283 7284As described above in the kvm_sync_regs struct info in section 5 (kvm_run): 7285KVM_CAP_SYNC_REGS "allow[s] userspace to access certain guest registers 7286without having to call SET/GET_*REGS". This reduces overhead by eliminating 7287repeated ioctl calls for setting and/or getting register values. This is 7288particularly important when userspace is making synchronous guest state 7289modifications, e.g. when emulating and/or intercepting instructions in 7290userspace. 7291 7292For s390 specifics, please refer to the source code. 7293 7294For x86: 7295 7296- the register sets to be copied out to kvm_run are selectable 7297 by userspace (rather that all sets being copied out for every exit). 7298- vcpu_events are available in addition to regs and sregs. 7299 7300For x86, the 'kvm_valid_regs' field of struct kvm_run is overloaded to 7301function as an input bit-array field set by userspace to indicate the 7302specific register sets to be copied out on the next exit. 7303 7304To indicate when userspace has modified values that should be copied into 7305the vCPU, the all architecture bitarray field, 'kvm_dirty_regs' must be set. 7306This is done using the same bitflags as for the 'kvm_valid_regs' field. 7307If the dirty bit is not set, then the register set values will not be copied 7308into the vCPU even if they've been modified. 7309 7310Unused bitfields in the bitarrays must be set to zero. 7311 7312:: 7313 7314 struct kvm_sync_regs { 7315 struct kvm_regs regs; 7316 struct kvm_sregs sregs; 7317 struct kvm_vcpu_events events; 7318 }; 7319 73206.75 KVM_CAP_PPC_IRQ_XIVE 7321------------------------- 7322 7323:Architectures: ppc 7324:Target: vcpu 7325:Parameters: args[0] is the XIVE device fd; 7326 args[1] is the XIVE CPU number (server ID) for this vcpu 7327 7328This capability connects the vcpu to an in-kernel XIVE device. 7329 73307. Capabilities that can be enabled on VMs 7331========================================== 7332 7333There are certain capabilities that change the behavior of the virtual 7334machine when enabled. To enable them, please see section 4.37. Below 7335you can find a list of capabilities and what their effect on the VM 7336is when enabling them. 7337 7338The following information is provided along with the description: 7339 7340 Architectures: 7341 which instruction set architectures provide this ioctl. 7342 x86 includes both i386 and x86_64. 7343 7344 Parameters: 7345 what parameters are accepted by the capability. 7346 7347 Returns: 7348 the return value. General error numbers (EBADF, ENOMEM, EINVAL) 7349 are not detailed, but errors with specific meanings are. 7350 7351 73527.1 KVM_CAP_PPC_ENABLE_HCALL 7353---------------------------- 7354 7355:Architectures: ppc 7356:Parameters: args[0] is the sPAPR hcall number; 7357 args[1] is 0 to disable, 1 to enable in-kernel handling 7358 7359This capability controls whether individual sPAPR hypercalls (hcalls) 7360get handled by the kernel or not. Enabling or disabling in-kernel 7361handling of an hcall is effective across the VM. On creation, an 7362initial set of hcalls are enabled for in-kernel handling, which 7363consists of those hcalls for which in-kernel handlers were implemented 7364before this capability was implemented. If disabled, the kernel will 7365not to attempt to handle the hcall, but will always exit to userspace 7366to handle it. Note that it may not make sense to enable some and 7367disable others of a group of related hcalls, but KVM does not prevent 7368userspace from doing that. 7369 7370If the hcall number specified is not one that has an in-kernel 7371implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL 7372error. 7373 73747.2 KVM_CAP_S390_USER_SIGP 7375-------------------------- 7376 7377:Architectures: s390 7378:Parameters: none 7379 7380This capability controls which SIGP orders will be handled completely in user 7381space. With this capability enabled, all fast orders will be handled completely 7382in the kernel: 7383 7384- SENSE 7385- SENSE RUNNING 7386- EXTERNAL CALL 7387- EMERGENCY SIGNAL 7388- CONDITIONAL EMERGENCY SIGNAL 7389 7390All other orders will be handled completely in user space. 7391 7392Only privileged operation exceptions will be checked for in the kernel (or even 7393in the hardware prior to interception). If this capability is not enabled, the 7394old way of handling SIGP orders is used (partially in kernel and user space). 7395 73967.3 KVM_CAP_S390_VECTOR_REGISTERS 7397--------------------------------- 7398 7399:Architectures: s390 7400:Parameters: none 7401:Returns: 0 on success, negative value on error 7402 7403Allows use of the vector registers introduced with z13 processor, and 7404provides for the synchronization between host and user space. Will 7405return -EINVAL if the machine does not support vectors. 7406 74077.4 KVM_CAP_S390_USER_STSI 7408-------------------------- 7409 7410:Architectures: s390 7411:Parameters: none 7412 7413This capability allows post-handlers for the STSI instruction. After 7414initial handling in the kernel, KVM exits to user space with 7415KVM_EXIT_S390_STSI to allow user space to insert further data. 7416 7417Before exiting to userspace, kvm handlers should fill in s390_stsi field of 7418vcpu->run:: 7419 7420 struct { 7421 __u64 addr; 7422 __u8 ar; 7423 __u8 reserved; 7424 __u8 fc; 7425 __u8 sel1; 7426 __u16 sel2; 7427 } s390_stsi; 7428 7429 @addr - guest address of STSI SYSIB 7430 @fc - function code 7431 @sel1 - selector 1 7432 @sel2 - selector 2 7433 @ar - access register number 7434 7435KVM handlers should exit to userspace with rc = -EREMOTE. 7436 74377.5 KVM_CAP_SPLIT_IRQCHIP 7438------------------------- 7439 7440:Architectures: x86 7441:Parameters: args[0] - number of routes reserved for userspace IOAPICs 7442:Returns: 0 on success, -1 on error 7443 7444Create a local apic for each processor in the kernel. This can be used 7445instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the 7446IOAPIC and PIC (and also the PIT, even though this has to be enabled 7447separately). 7448 7449This capability also enables in kernel routing of interrupt requests; 7450when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are 7451used in the IRQ routing table. The first args[0] MSI routes are reserved 7452for the IOAPIC pins. Whenever the LAPIC receives an EOI for these routes, 7453a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace. 7454 7455Fails if VCPU has already been created, or if the irqchip is already in the 7456kernel (i.e. KVM_CREATE_IRQCHIP has already been called). 7457 74587.6 KVM_CAP_S390_RI 7459------------------- 7460 7461:Architectures: s390 7462:Parameters: none 7463 7464Allows use of runtime-instrumentation introduced with zEC12 processor. 7465Will return -EINVAL if the machine does not support runtime-instrumentation. 7466Will return -EBUSY if a VCPU has already been created. 7467 74687.7 KVM_CAP_X2APIC_API 7469---------------------- 7470 7471:Architectures: x86 7472:Parameters: args[0] - features that should be enabled 7473:Returns: 0 on success, -EINVAL when args[0] contains invalid features 7474 7475Valid feature flags in args[0] are:: 7476 7477 #define KVM_X2APIC_API_USE_32BIT_IDS (1ULL << 0) 7478 #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK (1ULL << 1) 7479 7480Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of 7481KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC, 7482allowing the use of 32-bit APIC IDs. See KVM_CAP_X2APIC_API in their 7483respective sections. 7484 7485KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work 7486in logical mode or with more than 255 VCPUs. Otherwise, KVM treats 0xff 7487as a broadcast even in x2APIC mode in order to support physical x2APIC 7488without interrupt remapping. This is undesirable in logical mode, 7489where 0xff represents CPUs 0-7 in cluster 0. 7490 74917.8 KVM_CAP_S390_USER_INSTR0 7492---------------------------- 7493 7494:Architectures: s390 7495:Parameters: none 7496 7497With this capability enabled, all illegal instructions 0x0000 (2 bytes) will 7498be intercepted and forwarded to user space. User space can use this 7499mechanism e.g. to realize 2-byte software breakpoints. The kernel will 7500not inject an operating exception for these instructions, user space has 7501to take care of that. 7502 7503This capability can be enabled dynamically even if VCPUs were already 7504created and are running. 7505 75067.9 KVM_CAP_S390_GS 7507------------------- 7508 7509:Architectures: s390 7510:Parameters: none 7511:Returns: 0 on success; -EINVAL if the machine does not support 7512 guarded storage; -EBUSY if a VCPU has already been created. 7513 7514Allows use of guarded storage for the KVM guest. 7515 75167.10 KVM_CAP_S390_AIS 7517--------------------- 7518 7519:Architectures: s390 7520:Parameters: none 7521 7522Allow use of adapter-interruption suppression. 7523:Returns: 0 on success; -EBUSY if a VCPU has already been created. 7524 75257.11 KVM_CAP_PPC_SMT 7526-------------------- 7527 7528:Architectures: ppc 7529:Parameters: vsmt_mode, flags 7530 7531Enabling this capability on a VM provides userspace with a way to set 7532the desired virtual SMT mode (i.e. the number of virtual CPUs per 7533virtual core). The virtual SMT mode, vsmt_mode, must be a power of 2 7534between 1 and 8. On POWER8, vsmt_mode must also be no greater than 7535the number of threads per subcore for the host. Currently flags must 7536be 0. A successful call to enable this capability will result in 7537vsmt_mode being returned when the KVM_CAP_PPC_SMT capability is 7538subsequently queried for the VM. This capability is only supported by 7539HV KVM, and can only be set before any VCPUs have been created. 7540The KVM_CAP_PPC_SMT_POSSIBLE capability indicates which virtual SMT 7541modes are available. 7542 75437.12 KVM_CAP_PPC_FWNMI 7544---------------------- 7545 7546:Architectures: ppc 7547:Parameters: none 7548 7549With this capability a machine check exception in the guest address 7550space will cause KVM to exit the guest with NMI exit reason. This 7551enables QEMU to build error log and branch to guest kernel registered 7552machine check handling routine. Without this capability KVM will 7553branch to guests' 0x200 interrupt vector. 7554 75557.13 KVM_CAP_X86_DISABLE_EXITS 7556------------------------------ 7557 7558:Architectures: x86 7559:Parameters: args[0] defines which exits are disabled 7560:Returns: 0 on success, -EINVAL when args[0] contains invalid exits 7561 7562Valid bits in args[0] are:: 7563 7564 #define KVM_X86_DISABLE_EXITS_MWAIT (1 << 0) 7565 #define KVM_X86_DISABLE_EXITS_HLT (1 << 1) 7566 #define KVM_X86_DISABLE_EXITS_PAUSE (1 << 2) 7567 #define KVM_X86_DISABLE_EXITS_CSTATE (1 << 3) 7568 7569Enabling this capability on a VM provides userspace with a way to no 7570longer intercept some instructions for improved latency in some 7571workloads, and is suggested when vCPUs are associated to dedicated 7572physical CPUs. More bits can be added in the future; userspace can 7573just pass the KVM_CHECK_EXTENSION result to KVM_ENABLE_CAP to disable 7574all such vmexits. 7575 7576Do not enable KVM_FEATURE_PV_UNHALT if you disable HLT exits. 7577 75787.14 KVM_CAP_S390_HPAGE_1M 7579-------------------------- 7580 7581:Architectures: s390 7582:Parameters: none 7583:Returns: 0 on success, -EINVAL if hpage module parameter was not set 7584 or cmma is enabled, or the VM has the KVM_VM_S390_UCONTROL 7585 flag set 7586 7587With this capability the KVM support for memory backing with 1m pages 7588through hugetlbfs can be enabled for a VM. After the capability is 7589enabled, cmma can't be enabled anymore and pfmfi and the storage key 7590interpretation are disabled. If cmma has already been enabled or the 7591hpage module parameter is not set to 1, -EINVAL is returned. 7592 7593While it is generally possible to create a huge page backed VM without 7594this capability, the VM will not be able to run. 7595 75967.15 KVM_CAP_MSR_PLATFORM_INFO 7597------------------------------ 7598 7599:Architectures: x86 7600:Parameters: args[0] whether feature should be enabled or not 7601 7602With this capability, a guest may read the MSR_PLATFORM_INFO MSR. Otherwise, 7603a #GP would be raised when the guest tries to access. Currently, this 7604capability does not enable write permissions of this MSR for the guest. 7605 76067.16 KVM_CAP_PPC_NESTED_HV 7607-------------------------- 7608 7609:Architectures: ppc 7610:Parameters: none 7611:Returns: 0 on success, -EINVAL when the implementation doesn't support 7612 nested-HV virtualization. 7613 7614HV-KVM on POWER9 and later systems allows for "nested-HV" 7615virtualization, which provides a way for a guest VM to run guests that 7616can run using the CPU's supervisor mode (privileged non-hypervisor 7617state). Enabling this capability on a VM depends on the CPU having 7618the necessary functionality and on the facility being enabled with a 7619kvm-hv module parameter. 7620 76217.17 KVM_CAP_EXCEPTION_PAYLOAD 7622------------------------------ 7623 7624:Architectures: x86 7625:Parameters: args[0] whether feature should be enabled or not 7626 7627With this capability enabled, CR2 will not be modified prior to the 7628emulated VM-exit when L1 intercepts a #PF exception that occurs in 7629L2. Similarly, for kvm-intel only, DR6 will not be modified prior to 7630the emulated VM-exit when L1 intercepts a #DB exception that occurs in 7631L2. As a result, when KVM_GET_VCPU_EVENTS reports a pending #PF (or 7632#DB) exception for L2, exception.has_payload will be set and the 7633faulting address (or the new DR6 bits*) will be reported in the 7634exception_payload field. Similarly, when userspace injects a #PF (or 7635#DB) into L2 using KVM_SET_VCPU_EVENTS, it is expected to set 7636exception.has_payload and to put the faulting address - or the new DR6 7637bits\ [#]_ - in the exception_payload field. 7638 7639This capability also enables exception.pending in struct 7640kvm_vcpu_events, which allows userspace to distinguish between pending 7641and injected exceptions. 7642 7643 7644.. [#] For the new DR6 bits, note that bit 16 is set iff the #DB exception 7645 will clear DR6.RTM. 7646 76477.18 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 7648-------------------------------------- 7649 7650:Architectures: x86, arm64, mips 7651:Parameters: args[0] whether feature should be enabled or not 7652 7653Valid flags are:: 7654 7655 #define KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (1 << 0) 7656 #define KVM_DIRTY_LOG_INITIALLY_SET (1 << 1) 7657 7658With KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE is set, KVM_GET_DIRTY_LOG will not 7659automatically clear and write-protect all pages that are returned as dirty. 7660Rather, userspace will have to do this operation separately using 7661KVM_CLEAR_DIRTY_LOG. 7662 7663At the cost of a slightly more complicated operation, this provides better 7664scalability and responsiveness for two reasons. First, 7665KVM_CLEAR_DIRTY_LOG ioctl can operate on a 64-page granularity rather 7666than requiring to sync a full memslot; this ensures that KVM does not 7667take spinlocks for an extended period of time. Second, in some cases a 7668large amount of time can pass between a call to KVM_GET_DIRTY_LOG and 7669userspace actually using the data in the page. Pages can be modified 7670during this time, which is inefficient for both the guest and userspace: 7671the guest will incur a higher penalty due to write protection faults, 7672while userspace can see false reports of dirty pages. Manual reprotection 7673helps reducing this time, improving guest performance and reducing the 7674number of dirty log false positives. 7675 7676With KVM_DIRTY_LOG_INITIALLY_SET set, all the bits of the dirty bitmap 7677will be initialized to 1 when created. This also improves performance because 7678dirty logging can be enabled gradually in small chunks on the first call 7679to KVM_CLEAR_DIRTY_LOG. KVM_DIRTY_LOG_INITIALLY_SET depends on 7680KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (it is also only available on 7681x86 and arm64 for now). 7682 7683KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 was previously available under the name 7684KVM_CAP_MANUAL_DIRTY_LOG_PROTECT, but the implementation had bugs that make 7685it hard or impossible to use it correctly. The availability of 7686KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 signals that those bugs are fixed. 7687Userspace should not try to use KVM_CAP_MANUAL_DIRTY_LOG_PROTECT. 7688 76897.19 KVM_CAP_PPC_SECURE_GUEST 7690------------------------------ 7691 7692:Architectures: ppc 7693 7694This capability indicates that KVM is running on a host that has 7695ultravisor firmware and thus can support a secure guest. On such a 7696system, a guest can ask the ultravisor to make it a secure guest, 7697one whose memory is inaccessible to the host except for pages which 7698are explicitly requested to be shared with the host. The ultravisor 7699notifies KVM when a guest requests to become a secure guest, and KVM 7700has the opportunity to veto the transition. 7701 7702If present, this capability can be enabled for a VM, meaning that KVM 7703will allow the transition to secure guest mode. Otherwise KVM will 7704veto the transition. 7705 77067.20 KVM_CAP_HALT_POLL 7707---------------------- 7708 7709:Architectures: all 7710:Target: VM 7711:Parameters: args[0] is the maximum poll time in nanoseconds 7712:Returns: 0 on success; -1 on error 7713 7714KVM_CAP_HALT_POLL overrides the kvm.halt_poll_ns module parameter to set the 7715maximum halt-polling time for all vCPUs in the target VM. This capability can 7716be invoked at any time and any number of times to dynamically change the 7717maximum halt-polling time. 7718 7719See Documentation/virt/kvm/halt-polling.rst for more information on halt 7720polling. 7721 77227.21 KVM_CAP_X86_USER_SPACE_MSR 7723------------------------------- 7724 7725:Architectures: x86 7726:Target: VM 7727:Parameters: args[0] contains the mask of KVM_MSR_EXIT_REASON_* events to report 7728:Returns: 0 on success; -1 on error 7729 7730This capability allows userspace to intercept RDMSR and WRMSR instructions if 7731access to an MSR is denied. By default, KVM injects #GP on denied accesses. 7732 7733When a guest requests to read or write an MSR, KVM may not implement all MSRs 7734that are relevant to a respective system. It also does not differentiate by 7735CPU type. 7736 7737To allow more fine grained control over MSR handling, userspace may enable 7738this capability. With it enabled, MSR accesses that match the mask specified in 7739args[0] and would trigger a #GP inside the guest will instead trigger 7740KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR exit notifications. Userspace 7741can then implement model specific MSR handling and/or user notifications 7742to inform a user that an MSR was not emulated/virtualized by KVM. 7743 7744The valid mask flags are: 7745 7746============================ =============================================== 7747 KVM_MSR_EXIT_REASON_UNKNOWN intercept accesses to unknown (to KVM) MSRs 7748 KVM_MSR_EXIT_REASON_INVAL intercept accesses that are architecturally 7749 invalid according to the vCPU model and/or mode 7750 KVM_MSR_EXIT_REASON_FILTER intercept accesses that are denied by userspace 7751 via KVM_X86_SET_MSR_FILTER 7752============================ =============================================== 7753 77547.22 KVM_CAP_X86_BUS_LOCK_EXIT 7755------------------------------- 7756 7757:Architectures: x86 7758:Target: VM 7759:Parameters: args[0] defines the policy used when bus locks detected in guest 7760:Returns: 0 on success, -EINVAL when args[0] contains invalid bits 7761 7762Valid bits in args[0] are:: 7763 7764 #define KVM_BUS_LOCK_DETECTION_OFF (1 << 0) 7765 #define KVM_BUS_LOCK_DETECTION_EXIT (1 << 1) 7766 7767Enabling this capability on a VM provides userspace with a way to select 7768a policy to handle the bus locks detected in guest. Userspace can obtain 7769the supported modes from the result of KVM_CHECK_EXTENSION and define it 7770through the KVM_ENABLE_CAP. 7771 7772KVM_BUS_LOCK_DETECTION_OFF and KVM_BUS_LOCK_DETECTION_EXIT are supported 7773currently and mutually exclusive with each other. More bits can be added in 7774the future. 7775 7776With KVM_BUS_LOCK_DETECTION_OFF set, bus locks in guest will not cause vm exits 7777so that no additional actions are needed. This is the default mode. 7778 7779With KVM_BUS_LOCK_DETECTION_EXIT set, vm exits happen when bus lock detected 7780in VM. KVM just exits to userspace when handling them. Userspace can enforce 7781its own throttling or other policy based mitigations. 7782 7783This capability is aimed to address the thread that VM can exploit bus locks to 7784degree the performance of the whole system. Once the userspace enable this 7785capability and select the KVM_BUS_LOCK_DETECTION_EXIT mode, KVM will set the 7786KVM_RUN_BUS_LOCK flag in vcpu-run->flags field and exit to userspace. Concerning 7787the bus lock vm exit can be preempted by a higher priority VM exit, the exit 7788notifications to userspace can be KVM_EXIT_BUS_LOCK or other reasons. 7789KVM_RUN_BUS_LOCK flag is used to distinguish between them. 7790 77917.23 KVM_CAP_PPC_DAWR1 7792---------------------- 7793 7794:Architectures: ppc 7795:Parameters: none 7796:Returns: 0 on success, -EINVAL when CPU doesn't support 2nd DAWR 7797 7798This capability can be used to check / enable 2nd DAWR feature provided 7799by POWER10 processor. 7800 7801 78027.24 KVM_CAP_VM_COPY_ENC_CONTEXT_FROM 7803------------------------------------- 7804 7805Architectures: x86 SEV enabled 7806Type: vm 7807Parameters: args[0] is the fd of the source vm 7808Returns: 0 on success; ENOTTY on error 7809 7810This capability enables userspace to copy encryption context from the vm 7811indicated by the fd to the vm this is called on. 7812 7813This is intended to support in-guest workloads scheduled by the host. This 7814allows the in-guest workload to maintain its own NPTs and keeps the two vms 7815from accidentally clobbering each other with interrupts and the like (separate 7816APIC/MSRs/etc). 7817 78187.25 KVM_CAP_SGX_ATTRIBUTE 7819-------------------------- 7820 7821:Architectures: x86 7822:Target: VM 7823:Parameters: args[0] is a file handle of a SGX attribute file in securityfs 7824:Returns: 0 on success, -EINVAL if the file handle is invalid or if a requested 7825 attribute is not supported by KVM. 7826 7827KVM_CAP_SGX_ATTRIBUTE enables a userspace VMM to grant a VM access to one or 7828more privileged enclave attributes. args[0] must hold a file handle to a valid 7829SGX attribute file corresponding to an attribute that is supported/restricted 7830by KVM (currently only PROVISIONKEY). 7831 7832The SGX subsystem restricts access to a subset of enclave attributes to provide 7833additional security for an uncompromised kernel, e.g. use of the PROVISIONKEY 7834is restricted to deter malware from using the PROVISIONKEY to obtain a stable 7835system fingerprint. To prevent userspace from circumventing such restrictions 7836by running an enclave in a VM, KVM prevents access to privileged attributes by 7837default. 7838 7839See Documentation/arch/x86/sgx.rst for more details. 7840 78417.26 KVM_CAP_PPC_RPT_INVALIDATE 7842------------------------------- 7843 7844:Capability: KVM_CAP_PPC_RPT_INVALIDATE 7845:Architectures: ppc 7846:Type: vm 7847 7848This capability indicates that the kernel is capable of handling 7849H_RPT_INVALIDATE hcall. 7850 7851In order to enable the use of H_RPT_INVALIDATE in the guest, 7852user space might have to advertise it for the guest. For example, 7853IBM pSeries (sPAPR) guest starts using it if "hcall-rpt-invalidate" is 7854present in the "ibm,hypertas-functions" device-tree property. 7855 7856This capability is enabled for hypervisors on platforms like POWER9 7857that support radix MMU. 7858 78597.27 KVM_CAP_EXIT_ON_EMULATION_FAILURE 7860-------------------------------------- 7861 7862:Architectures: x86 7863:Parameters: args[0] whether the feature should be enabled or not 7864 7865When this capability is enabled, an emulation failure will result in an exit 7866to userspace with KVM_INTERNAL_ERROR (except when the emulator was invoked 7867to handle a VMware backdoor instruction). Furthermore, KVM will now provide up 7868to 15 instruction bytes for any exit to userspace resulting from an emulation 7869failure. When these exits to userspace occur use the emulation_failure struct 7870instead of the internal struct. They both have the same layout, but the 7871emulation_failure struct matches the content better. It also explicitly 7872defines the 'flags' field which is used to describe the fields in the struct 7873that are valid (ie: if KVM_INTERNAL_ERROR_EMULATION_FLAG_INSTRUCTION_BYTES is 7874set in the 'flags' field then both 'insn_size' and 'insn_bytes' have valid data 7875in them.) 7876 78777.28 KVM_CAP_ARM_MTE 7878-------------------- 7879 7880:Architectures: arm64 7881:Parameters: none 7882 7883This capability indicates that KVM (and the hardware) supports exposing the 7884Memory Tagging Extensions (MTE) to the guest. It must also be enabled by the 7885VMM before creating any VCPUs to allow the guest access. Note that MTE is only 7886available to a guest running in AArch64 mode and enabling this capability will 7887cause attempts to create AArch32 VCPUs to fail. 7888 7889When enabled the guest is able to access tags associated with any memory given 7890to the guest. KVM will ensure that the tags are maintained during swap or 7891hibernation of the host; however the VMM needs to manually save/restore the 7892tags as appropriate if the VM is migrated. 7893 7894When this capability is enabled all memory in memslots must be mapped as 7895``MAP_ANONYMOUS`` or with a RAM-based file mapping (``tmpfs``, ``memfd``), 7896attempts to create a memslot with an invalid mmap will result in an 7897-EINVAL return. 7898 7899When enabled the VMM may make use of the ``KVM_ARM_MTE_COPY_TAGS`` ioctl to 7900perform a bulk copy of tags to/from the guest. 7901 79027.29 KVM_CAP_VM_MOVE_ENC_CONTEXT_FROM 7903------------------------------------- 7904 7905Architectures: x86 SEV enabled 7906Type: vm 7907Parameters: args[0] is the fd of the source vm 7908Returns: 0 on success 7909 7910This capability enables userspace to migrate the encryption context from the VM 7911indicated by the fd to the VM this is called on. 7912 7913This is intended to support intra-host migration of VMs between userspace VMMs, 7914upgrading the VMM process without interrupting the guest. 7915 79167.30 KVM_CAP_PPC_AIL_MODE_3 7917------------------------------- 7918 7919:Capability: KVM_CAP_PPC_AIL_MODE_3 7920:Architectures: ppc 7921:Type: vm 7922 7923This capability indicates that the kernel supports the mode 3 setting for the 7924"Address Translation Mode on Interrupt" aka "Alternate Interrupt Location" 7925resource that is controlled with the H_SET_MODE hypercall. 7926 7927This capability allows a guest kernel to use a better-performance mode for 7928handling interrupts and system calls. 7929 79307.31 KVM_CAP_DISABLE_QUIRKS2 7931---------------------------- 7932 7933:Capability: KVM_CAP_DISABLE_QUIRKS2 7934:Parameters: args[0] - set of KVM quirks to disable 7935:Architectures: x86 7936:Type: vm 7937 7938This capability, if enabled, will cause KVM to disable some behavior 7939quirks. 7940 7941Calling KVM_CHECK_EXTENSION for this capability returns a bitmask of 7942quirks that can be disabled in KVM. 7943 7944The argument to KVM_ENABLE_CAP for this capability is a bitmask of 7945quirks to disable, and must be a subset of the bitmask returned by 7946KVM_CHECK_EXTENSION. 7947 7948The valid bits in cap.args[0] are: 7949 7950=================================== ============================================ 7951 KVM_X86_QUIRK_LINT0_REENABLED By default, the reset value for the LVT 7952 LINT0 register is 0x700 (APIC_MODE_EXTINT). 7953 When this quirk is disabled, the reset value 7954 is 0x10000 (APIC_LVT_MASKED). 7955 7956 KVM_X86_QUIRK_CD_NW_CLEARED By default, KVM clears CR0.CD and CR0.NW. 7957 When this quirk is disabled, KVM does not 7958 change the value of CR0.CD and CR0.NW. 7959 7960 KVM_X86_QUIRK_LAPIC_MMIO_HOLE By default, the MMIO LAPIC interface is 7961 available even when configured for x2APIC 7962 mode. When this quirk is disabled, KVM 7963 disables the MMIO LAPIC interface if the 7964 LAPIC is in x2APIC mode. 7965 7966 KVM_X86_QUIRK_OUT_7E_INC_RIP By default, KVM pre-increments %rip before 7967 exiting to userspace for an OUT instruction 7968 to port 0x7e. When this quirk is disabled, 7969 KVM does not pre-increment %rip before 7970 exiting to userspace. 7971 7972 KVM_X86_QUIRK_MISC_ENABLE_NO_MWAIT When this quirk is disabled, KVM sets 7973 CPUID.01H:ECX[bit 3] (MONITOR/MWAIT) if 7974 IA32_MISC_ENABLE[bit 18] (MWAIT) is set. 7975 Additionally, when this quirk is disabled, 7976 KVM clears CPUID.01H:ECX[bit 3] if 7977 IA32_MISC_ENABLE[bit 18] is cleared. 7978 7979 KVM_X86_QUIRK_FIX_HYPERCALL_INSN By default, KVM rewrites guest 7980 VMMCALL/VMCALL instructions to match the 7981 vendor's hypercall instruction for the 7982 system. When this quirk is disabled, KVM 7983 will no longer rewrite invalid guest 7984 hypercall instructions. Executing the 7985 incorrect hypercall instruction will 7986 generate a #UD within the guest. 7987 7988KVM_X86_QUIRK_MWAIT_NEVER_UD_FAULTS By default, KVM emulates MONITOR/MWAIT (if 7989 they are intercepted) as NOPs regardless of 7990 whether or not MONITOR/MWAIT are supported 7991 according to guest CPUID. When this quirk 7992 is disabled and KVM_X86_DISABLE_EXITS_MWAIT 7993 is not set (MONITOR/MWAIT are intercepted), 7994 KVM will inject a #UD on MONITOR/MWAIT if 7995 they're unsupported per guest CPUID. Note, 7996 KVM will modify MONITOR/MWAIT support in 7997 guest CPUID on writes to MISC_ENABLE if 7998 KVM_X86_QUIRK_MISC_ENABLE_NO_MWAIT is 7999 disabled. 8000=================================== ============================================ 8001 80027.32 KVM_CAP_MAX_VCPU_ID 8003------------------------ 8004 8005:Architectures: x86 8006:Target: VM 8007:Parameters: args[0] - maximum APIC ID value set for current VM 8008:Returns: 0 on success, -EINVAL if args[0] is beyond KVM_MAX_VCPU_IDS 8009 supported in KVM or if it has been set. 8010 8011This capability allows userspace to specify maximum possible APIC ID 8012assigned for current VM session prior to the creation of vCPUs, saving 8013memory for data structures indexed by the APIC ID. Userspace is able 8014to calculate the limit to APIC ID values from designated 8015CPU topology. 8016 8017The value can be changed only until KVM_ENABLE_CAP is set to a nonzero 8018value or until a vCPU is created. Upon creation of the first vCPU, 8019if the value was set to zero or KVM_ENABLE_CAP was not invoked, KVM 8020uses the return value of KVM_CHECK_EXTENSION(KVM_CAP_MAX_VCPU_ID) as 8021the maximum APIC ID. 8022 80237.33 KVM_CAP_X86_NOTIFY_VMEXIT 8024------------------------------ 8025 8026:Architectures: x86 8027:Target: VM 8028:Parameters: args[0] is the value of notify window as well as some flags 8029:Returns: 0 on success, -EINVAL if args[0] contains invalid flags or notify 8030 VM exit is unsupported. 8031 8032Bits 63:32 of args[0] are used for notify window. 8033Bits 31:0 of args[0] are for some flags. Valid bits are:: 8034 8035 #define KVM_X86_NOTIFY_VMEXIT_ENABLED (1 << 0) 8036 #define KVM_X86_NOTIFY_VMEXIT_USER (1 << 1) 8037 8038This capability allows userspace to configure the notify VM exit on/off 8039in per-VM scope during VM creation. Notify VM exit is disabled by default. 8040When userspace sets KVM_X86_NOTIFY_VMEXIT_ENABLED bit in args[0], VMM will 8041enable this feature with the notify window provided, which will generate 8042a VM exit if no event window occurs in VM non-root mode for a specified of 8043time (notify window). 8044 8045If KVM_X86_NOTIFY_VMEXIT_USER is set in args[0], upon notify VM exits happen, 8046KVM would exit to userspace for handling. 8047 8048This capability is aimed to mitigate the threat that malicious VMs can 8049cause CPU stuck (due to event windows don't open up) and make the CPU 8050unavailable to host or other VMs. 8051 80527.34 KVM_CAP_MEMORY_FAULT_INFO 8053------------------------------ 8054 8055:Architectures: x86 8056:Returns: Informational only, -EINVAL on direct KVM_ENABLE_CAP. 8057 8058The presence of this capability indicates that KVM_RUN will fill 8059kvm_run.memory_fault if KVM cannot resolve a guest page fault VM-Exit, e.g. if 8060there is a valid memslot but no backing VMA for the corresponding host virtual 8061address. 8062 8063The information in kvm_run.memory_fault is valid if and only if KVM_RUN returns 8064an error with errno=EFAULT or errno=EHWPOISON *and* kvm_run.exit_reason is set 8065to KVM_EXIT_MEMORY_FAULT. 8066 8067Note: Userspaces which attempt to resolve memory faults so that they can retry 8068KVM_RUN are encouraged to guard against repeatedly receiving the same 8069error/annotated fault. 8070 8071See KVM_EXIT_MEMORY_FAULT for more information. 8072 80738. Other capabilities. 8074====================== 8075 8076This section lists capabilities that give information about other 8077features of the KVM implementation. 8078 80798.1 KVM_CAP_PPC_HWRNG 8080--------------------- 8081 8082:Architectures: ppc 8083 8084This capability, if KVM_CHECK_EXTENSION indicates that it is 8085available, means that the kernel has an implementation of the 8086H_RANDOM hypercall backed by a hardware random-number generator. 8087If present, the kernel H_RANDOM handler can be enabled for guest use 8088with the KVM_CAP_PPC_ENABLE_HCALL capability. 8089 80908.2 KVM_CAP_HYPERV_SYNIC 8091------------------------ 8092 8093:Architectures: x86 8094 8095This capability, if KVM_CHECK_EXTENSION indicates that it is 8096available, means that the kernel has an implementation of the 8097Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is 8098used to support Windows Hyper-V based guest paravirt drivers(VMBus). 8099 8100In order to use SynIC, it has to be activated by setting this 8101capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this 8102will disable the use of APIC hardware virtualization even if supported 8103by the CPU, as it's incompatible with SynIC auto-EOI behavior. 8104 81058.3 KVM_CAP_PPC_MMU_RADIX 8106------------------------- 8107 8108:Architectures: ppc 8109 8110This capability, if KVM_CHECK_EXTENSION indicates that it is 8111available, means that the kernel can support guests using the 8112radix MMU defined in Power ISA V3.00 (as implemented in the POWER9 8113processor). 8114 81158.4 KVM_CAP_PPC_MMU_HASH_V3 8116--------------------------- 8117 8118:Architectures: ppc 8119 8120This capability, if KVM_CHECK_EXTENSION indicates that it is 8121available, means that the kernel can support guests using the 8122hashed page table MMU defined in Power ISA V3.00 (as implemented in 8123the POWER9 processor), including in-memory segment tables. 8124 81258.5 KVM_CAP_MIPS_VZ 8126------------------- 8127 8128:Architectures: mips 8129 8130This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that 8131it is available, means that full hardware assisted virtualization capabilities 8132of the hardware are available for use through KVM. An appropriate 8133KVM_VM_MIPS_* type must be passed to KVM_CREATE_VM to create a VM which 8134utilises it. 8135 8136If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is 8137available, it means that the VM is using full hardware assisted virtualization 8138capabilities of the hardware. This is useful to check after creating a VM with 8139KVM_VM_MIPS_DEFAULT. 8140 8141The value returned by KVM_CHECK_EXTENSION should be compared against known 8142values (see below). All other values are reserved. This is to allow for the 8143possibility of other hardware assisted virtualization implementations which 8144may be incompatible with the MIPS VZ ASE. 8145 8146== ========================================================================== 8147 0 The trap & emulate implementation is in use to run guest code in user 8148 mode. Guest virtual memory segments are rearranged to fit the guest in the 8149 user mode address space. 8150 8151 1 The MIPS VZ ASE is in use, providing full hardware assisted 8152 virtualization, including standard guest virtual memory segments. 8153== ========================================================================== 8154 81558.6 KVM_CAP_MIPS_TE 8156------------------- 8157 8158:Architectures: mips 8159 8160This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that 8161it is available, means that the trap & emulate implementation is available to 8162run guest code in user mode, even if KVM_CAP_MIPS_VZ indicates that hardware 8163assisted virtualisation is also available. KVM_VM_MIPS_TE (0) must be passed 8164to KVM_CREATE_VM to create a VM which utilises it. 8165 8166If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is 8167available, it means that the VM is using trap & emulate. 8168 81698.7 KVM_CAP_MIPS_64BIT 8170---------------------- 8171 8172:Architectures: mips 8173 8174This capability indicates the supported architecture type of the guest, i.e. the 8175supported register and address width. 8176 8177The values returned when this capability is checked by KVM_CHECK_EXTENSION on a 8178kvm VM handle correspond roughly to the CP0_Config.AT register field, and should 8179be checked specifically against known values (see below). All other values are 8180reserved. 8181 8182== ======================================================================== 8183 0 MIPS32 or microMIPS32. 8184 Both registers and addresses are 32-bits wide. 8185 It will only be possible to run 32-bit guest code. 8186 8187 1 MIPS64 or microMIPS64 with access only to 32-bit compatibility segments. 8188 Registers are 64-bits wide, but addresses are 32-bits wide. 8189 64-bit guest code may run but cannot access MIPS64 memory segments. 8190 It will also be possible to run 32-bit guest code. 8191 8192 2 MIPS64 or microMIPS64 with access to all address segments. 8193 Both registers and addresses are 64-bits wide. 8194 It will be possible to run 64-bit or 32-bit guest code. 8195== ======================================================================== 8196 81978.9 KVM_CAP_ARM_USER_IRQ 8198------------------------ 8199 8200:Architectures: arm64 8201 8202This capability, if KVM_CHECK_EXTENSION indicates that it is available, means 8203that if userspace creates a VM without an in-kernel interrupt controller, it 8204will be notified of changes to the output level of in-kernel emulated devices, 8205which can generate virtual interrupts, presented to the VM. 8206For such VMs, on every return to userspace, the kernel 8207updates the vcpu's run->s.regs.device_irq_level field to represent the actual 8208output level of the device. 8209 8210Whenever kvm detects a change in the device output level, kvm guarantees at 8211least one return to userspace before running the VM. This exit could either 8212be a KVM_EXIT_INTR or any other exit event, like KVM_EXIT_MMIO. This way, 8213userspace can always sample the device output level and re-compute the state of 8214the userspace interrupt controller. Userspace should always check the state 8215of run->s.regs.device_irq_level on every kvm exit. 8216The value in run->s.regs.device_irq_level can represent both level and edge 8217triggered interrupt signals, depending on the device. Edge triggered interrupt 8218signals will exit to userspace with the bit in run->s.regs.device_irq_level 8219set exactly once per edge signal. 8220 8221The field run->s.regs.device_irq_level is available independent of 8222run->kvm_valid_regs or run->kvm_dirty_regs bits. 8223 8224If KVM_CAP_ARM_USER_IRQ is supported, the KVM_CHECK_EXTENSION ioctl returns a 8225number larger than 0 indicating the version of this capability is implemented 8226and thereby which bits in run->s.regs.device_irq_level can signal values. 8227 8228Currently the following bits are defined for the device_irq_level bitmap:: 8229 8230 KVM_CAP_ARM_USER_IRQ >= 1: 8231 8232 KVM_ARM_DEV_EL1_VTIMER - EL1 virtual timer 8233 KVM_ARM_DEV_EL1_PTIMER - EL1 physical timer 8234 KVM_ARM_DEV_PMU - ARM PMU overflow interrupt signal 8235 8236Future versions of kvm may implement additional events. These will get 8237indicated by returning a higher number from KVM_CHECK_EXTENSION and will be 8238listed above. 8239 82408.10 KVM_CAP_PPC_SMT_POSSIBLE 8241----------------------------- 8242 8243:Architectures: ppc 8244 8245Querying this capability returns a bitmap indicating the possible 8246virtual SMT modes that can be set using KVM_CAP_PPC_SMT. If bit N 8247(counting from the right) is set, then a virtual SMT mode of 2^N is 8248available. 8249 82508.11 KVM_CAP_HYPERV_SYNIC2 8251-------------------------- 8252 8253:Architectures: x86 8254 8255This capability enables a newer version of Hyper-V Synthetic interrupt 8256controller (SynIC). The only difference with KVM_CAP_HYPERV_SYNIC is that KVM 8257doesn't clear SynIC message and event flags pages when they are enabled by 8258writing to the respective MSRs. 8259 82608.12 KVM_CAP_HYPERV_VP_INDEX 8261---------------------------- 8262 8263:Architectures: x86 8264 8265This capability indicates that userspace can load HV_X64_MSR_VP_INDEX msr. Its 8266value is used to denote the target vcpu for a SynIC interrupt. For 8267compatibility, KVM initializes this msr to KVM's internal vcpu index. When this 8268capability is absent, userspace can still query this msr's value. 8269 82708.13 KVM_CAP_S390_AIS_MIGRATION 8271------------------------------- 8272 8273:Architectures: s390 8274:Parameters: none 8275 8276This capability indicates if the flic device will be able to get/set the 8277AIS states for migration via the KVM_DEV_FLIC_AISM_ALL attribute and allows 8278to discover this without having to create a flic device. 8279 82808.14 KVM_CAP_S390_PSW 8281--------------------- 8282 8283:Architectures: s390 8284 8285This capability indicates that the PSW is exposed via the kvm_run structure. 8286 82878.15 KVM_CAP_S390_GMAP 8288---------------------- 8289 8290:Architectures: s390 8291 8292This capability indicates that the user space memory used as guest mapping can 8293be anywhere in the user memory address space, as long as the memory slots are 8294aligned and sized to a segment (1MB) boundary. 8295 82968.16 KVM_CAP_S390_COW 8297--------------------- 8298 8299:Architectures: s390 8300 8301This capability indicates that the user space memory used as guest mapping can 8302use copy-on-write semantics as well as dirty pages tracking via read-only page 8303tables. 8304 83058.17 KVM_CAP_S390_BPB 8306--------------------- 8307 8308:Architectures: s390 8309 8310This capability indicates that kvm will implement the interfaces to handle 8311reset, migration and nested KVM for branch prediction blocking. The stfle 8312facility 82 should not be provided to the guest without this capability. 8313 83148.18 KVM_CAP_HYPERV_TLBFLUSH 8315---------------------------- 8316 8317:Architectures: x86 8318 8319This capability indicates that KVM supports paravirtualized Hyper-V TLB Flush 8320hypercalls: 8321HvFlushVirtualAddressSpace, HvFlushVirtualAddressSpaceEx, 8322HvFlushVirtualAddressList, HvFlushVirtualAddressListEx. 8323 83248.19 KVM_CAP_ARM_INJECT_SERROR_ESR 8325---------------------------------- 8326 8327:Architectures: arm64 8328 8329This capability indicates that userspace can specify (via the 8330KVM_SET_VCPU_EVENTS ioctl) the syndrome value reported to the guest when it 8331takes a virtual SError interrupt exception. 8332If KVM advertises this capability, userspace can only specify the ISS field for 8333the ESR syndrome. Other parts of the ESR, such as the EC are generated by the 8334CPU when the exception is taken. If this virtual SError is taken to EL1 using 8335AArch64, this value will be reported in the ISS field of ESR_ELx. 8336 8337See KVM_CAP_VCPU_EVENTS for more details. 8338 83398.20 KVM_CAP_HYPERV_SEND_IPI 8340---------------------------- 8341 8342:Architectures: x86 8343 8344This capability indicates that KVM supports paravirtualized Hyper-V IPI send 8345hypercalls: 8346HvCallSendSyntheticClusterIpi, HvCallSendSyntheticClusterIpiEx. 8347 83488.21 KVM_CAP_HYPERV_DIRECT_TLBFLUSH 8349----------------------------------- 8350 8351:Architectures: x86 8352 8353This capability indicates that KVM running on top of Hyper-V hypervisor 8354enables Direct TLB flush for its guests meaning that TLB flush 8355hypercalls are handled by Level 0 hypervisor (Hyper-V) bypassing KVM. 8356Due to the different ABI for hypercall parameters between Hyper-V and 8357KVM, enabling this capability effectively disables all hypercall 8358handling by KVM (as some KVM hypercall may be mistakenly treated as TLB 8359flush hypercalls by Hyper-V) so userspace should disable KVM identification 8360in CPUID and only exposes Hyper-V identification. In this case, guest 8361thinks it's running on Hyper-V and only use Hyper-V hypercalls. 8362 83638.22 KVM_CAP_S390_VCPU_RESETS 8364----------------------------- 8365 8366:Architectures: s390 8367 8368This capability indicates that the KVM_S390_NORMAL_RESET and 8369KVM_S390_CLEAR_RESET ioctls are available. 8370 83718.23 KVM_CAP_S390_PROTECTED 8372--------------------------- 8373 8374:Architectures: s390 8375 8376This capability indicates that the Ultravisor has been initialized and 8377KVM can therefore start protected VMs. 8378This capability governs the KVM_S390_PV_COMMAND ioctl and the 8379KVM_MP_STATE_LOAD MP_STATE. KVM_SET_MP_STATE can fail for protected 8380guests when the state change is invalid. 8381 83828.24 KVM_CAP_STEAL_TIME 8383----------------------- 8384 8385:Architectures: arm64, x86 8386 8387This capability indicates that KVM supports steal time accounting. 8388When steal time accounting is supported it may be enabled with 8389architecture-specific interfaces. This capability and the architecture- 8390specific interfaces must be consistent, i.e. if one says the feature 8391is supported, than the other should as well and vice versa. For arm64 8392see Documentation/virt/kvm/devices/vcpu.rst "KVM_ARM_VCPU_PVTIME_CTRL". 8393For x86 see Documentation/virt/kvm/x86/msr.rst "MSR_KVM_STEAL_TIME". 8394 83958.25 KVM_CAP_S390_DIAG318 8396------------------------- 8397 8398:Architectures: s390 8399 8400This capability enables a guest to set information about its control program 8401(i.e. guest kernel type and version). The information is helpful during 8402system/firmware service events, providing additional data about the guest 8403environments running on the machine. 8404 8405The information is associated with the DIAGNOSE 0x318 instruction, which sets 8406an 8-byte value consisting of a one-byte Control Program Name Code (CPNC) and 8407a 7-byte Control Program Version Code (CPVC). The CPNC determines what 8408environment the control program is running in (e.g. Linux, z/VM...), and the 8409CPVC is used for information specific to OS (e.g. Linux version, Linux 8410distribution...) 8411 8412If this capability is available, then the CPNC and CPVC can be synchronized 8413between KVM and userspace via the sync regs mechanism (KVM_SYNC_DIAG318). 8414 84158.26 KVM_CAP_X86_USER_SPACE_MSR 8416------------------------------- 8417 8418:Architectures: x86 8419 8420This capability indicates that KVM supports deflection of MSR reads and 8421writes to user space. It can be enabled on a VM level. If enabled, MSR 8422accesses that would usually trigger a #GP by KVM into the guest will 8423instead get bounced to user space through the KVM_EXIT_X86_RDMSR and 8424KVM_EXIT_X86_WRMSR exit notifications. 8425 84268.27 KVM_CAP_X86_MSR_FILTER 8427--------------------------- 8428 8429:Architectures: x86 8430 8431This capability indicates that KVM supports that accesses to user defined MSRs 8432may be rejected. With this capability exposed, KVM exports new VM ioctl 8433KVM_X86_SET_MSR_FILTER which user space can call to specify bitmaps of MSR 8434ranges that KVM should deny access to. 8435 8436In combination with KVM_CAP_X86_USER_SPACE_MSR, this allows user space to 8437trap and emulate MSRs that are outside of the scope of KVM as well as 8438limit the attack surface on KVM's MSR emulation code. 8439 84408.28 KVM_CAP_ENFORCE_PV_FEATURE_CPUID 8441------------------------------------- 8442 8443Architectures: x86 8444 8445When enabled, KVM will disable paravirtual features provided to the 8446guest according to the bits in the KVM_CPUID_FEATURES CPUID leaf 8447(0x40000001). Otherwise, a guest may use the paravirtual features 8448regardless of what has actually been exposed through the CPUID leaf. 8449 84508.29 KVM_CAP_DIRTY_LOG_RING/KVM_CAP_DIRTY_LOG_RING_ACQ_REL 8451---------------------------------------------------------- 8452 8453:Architectures: x86, arm64 8454:Parameters: args[0] - size of the dirty log ring 8455 8456KVM is capable of tracking dirty memory using ring buffers that are 8457mmapped into userspace; there is one dirty ring per vcpu. 8458 8459The dirty ring is available to userspace as an array of 8460``struct kvm_dirty_gfn``. Each dirty entry is defined as:: 8461 8462 struct kvm_dirty_gfn { 8463 __u32 flags; 8464 __u32 slot; /* as_id | slot_id */ 8465 __u64 offset; 8466 }; 8467 8468The following values are defined for the flags field to define the 8469current state of the entry:: 8470 8471 #define KVM_DIRTY_GFN_F_DIRTY BIT(0) 8472 #define KVM_DIRTY_GFN_F_RESET BIT(1) 8473 #define KVM_DIRTY_GFN_F_MASK 0x3 8474 8475Userspace should call KVM_ENABLE_CAP ioctl right after KVM_CREATE_VM 8476ioctl to enable this capability for the new guest and set the size of 8477the rings. Enabling the capability is only allowed before creating any 8478vCPU, and the size of the ring must be a power of two. The larger the 8479ring buffer, the less likely the ring is full and the VM is forced to 8480exit to userspace. The optimal size depends on the workload, but it is 8481recommended that it be at least 64 KiB (4096 entries). 8482 8483Just like for dirty page bitmaps, the buffer tracks writes to 8484all user memory regions for which the KVM_MEM_LOG_DIRTY_PAGES flag was 8485set in KVM_SET_USER_MEMORY_REGION. Once a memory region is registered 8486with the flag set, userspace can start harvesting dirty pages from the 8487ring buffer. 8488 8489An entry in the ring buffer can be unused (flag bits ``00``), 8490dirty (flag bits ``01``) or harvested (flag bits ``1X``). The 8491state machine for the entry is as follows:: 8492 8493 dirtied harvested reset 8494 00 -----------> 01 -------------> 1X -------+ 8495 ^ | 8496 | | 8497 +------------------------------------------+ 8498 8499To harvest the dirty pages, userspace accesses the mmapped ring buffer 8500to read the dirty GFNs. If the flags has the DIRTY bit set (at this stage 8501the RESET bit must be cleared), then it means this GFN is a dirty GFN. 8502The userspace should harvest this GFN and mark the flags from state 8503``01b`` to ``1Xb`` (bit 0 will be ignored by KVM, but bit 1 must be set 8504to show that this GFN is harvested and waiting for a reset), and move 8505on to the next GFN. The userspace should continue to do this until the 8506flags of a GFN have the DIRTY bit cleared, meaning that it has harvested 8507all the dirty GFNs that were available. 8508 8509Note that on weakly ordered architectures, userspace accesses to the 8510ring buffer (and more specifically the 'flags' field) must be ordered, 8511using load-acquire/store-release accessors when available, or any 8512other memory barrier that will ensure this ordering. 8513 8514It's not necessary for userspace to harvest the all dirty GFNs at once. 8515However it must collect the dirty GFNs in sequence, i.e., the userspace 8516program cannot skip one dirty GFN to collect the one next to it. 8517 8518After processing one or more entries in the ring buffer, userspace 8519calls the VM ioctl KVM_RESET_DIRTY_RINGS to notify the kernel about 8520it, so that the kernel will reprotect those collected GFNs. 8521Therefore, the ioctl must be called *before* reading the content of 8522the dirty pages. 8523 8524The dirty ring can get full. When it happens, the KVM_RUN of the 8525vcpu will return with exit reason KVM_EXIT_DIRTY_LOG_FULL. 8526 8527The dirty ring interface has a major difference comparing to the 8528KVM_GET_DIRTY_LOG interface in that, when reading the dirty ring from 8529userspace, it's still possible that the kernel has not yet flushed the 8530processor's dirty page buffers into the kernel buffer (with dirty bitmaps, the 8531flushing is done by the KVM_GET_DIRTY_LOG ioctl). To achieve that, one 8532needs to kick the vcpu out of KVM_RUN using a signal. The resulting 8533vmexit ensures that all dirty GFNs are flushed to the dirty rings. 8534 8535NOTE: KVM_CAP_DIRTY_LOG_RING_ACQ_REL is the only capability that 8536should be exposed by weakly ordered architecture, in order to indicate 8537the additional memory ordering requirements imposed on userspace when 8538reading the state of an entry and mutating it from DIRTY to HARVESTED. 8539Architecture with TSO-like ordering (such as x86) are allowed to 8540expose both KVM_CAP_DIRTY_LOG_RING and KVM_CAP_DIRTY_LOG_RING_ACQ_REL 8541to userspace. 8542 8543After enabling the dirty rings, the userspace needs to detect the 8544capability of KVM_CAP_DIRTY_LOG_RING_WITH_BITMAP to see whether the 8545ring structures can be backed by per-slot bitmaps. With this capability 8546advertised, it means the architecture can dirty guest pages without 8547vcpu/ring context, so that some of the dirty information will still be 8548maintained in the bitmap structure. KVM_CAP_DIRTY_LOG_RING_WITH_BITMAP 8549can't be enabled if the capability of KVM_CAP_DIRTY_LOG_RING_ACQ_REL 8550hasn't been enabled, or any memslot has been existing. 8551 8552Note that the bitmap here is only a backup of the ring structure. The 8553use of the ring and bitmap combination is only beneficial if there is 8554only a very small amount of memory that is dirtied out of vcpu/ring 8555context. Otherwise, the stand-alone per-slot bitmap mechanism needs to 8556be considered. 8557 8558To collect dirty bits in the backup bitmap, userspace can use the same 8559KVM_GET_DIRTY_LOG ioctl. KVM_CLEAR_DIRTY_LOG isn't needed as long as all 8560the generation of the dirty bits is done in a single pass. Collecting 8561the dirty bitmap should be the very last thing that the VMM does before 8562considering the state as complete. VMM needs to ensure that the dirty 8563state is final and avoid missing dirty pages from another ioctl ordered 8564after the bitmap collection. 8565 8566NOTE: Multiple examples of using the backup bitmap: (1) save vgic/its 8567tables through command KVM_DEV_ARM_{VGIC_GRP_CTRL, ITS_SAVE_TABLES} on 8568KVM device "kvm-arm-vgic-its". (2) restore vgic/its tables through 8569command KVM_DEV_ARM_{VGIC_GRP_CTRL, ITS_RESTORE_TABLES} on KVM device 8570"kvm-arm-vgic-its". VGICv3 LPI pending status is restored. (3) save 8571vgic3 pending table through KVM_DEV_ARM_VGIC_{GRP_CTRL, SAVE_PENDING_TABLES} 8572command on KVM device "kvm-arm-vgic-v3". 8573 85748.30 KVM_CAP_XEN_HVM 8575-------------------- 8576 8577:Architectures: x86 8578 8579This capability indicates the features that Xen supports for hosting Xen 8580PVHVM guests. Valid flags are:: 8581 8582 #define KVM_XEN_HVM_CONFIG_HYPERCALL_MSR (1 << 0) 8583 #define KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL (1 << 1) 8584 #define KVM_XEN_HVM_CONFIG_SHARED_INFO (1 << 2) 8585 #define KVM_XEN_HVM_CONFIG_RUNSTATE (1 << 3) 8586 #define KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL (1 << 4) 8587 #define KVM_XEN_HVM_CONFIG_EVTCHN_SEND (1 << 5) 8588 #define KVM_XEN_HVM_CONFIG_RUNSTATE_UPDATE_FLAG (1 << 6) 8589 #define KVM_XEN_HVM_CONFIG_PVCLOCK_TSC_UNSTABLE (1 << 7) 8590 8591The KVM_XEN_HVM_CONFIG_HYPERCALL_MSR flag indicates that the KVM_XEN_HVM_CONFIG 8592ioctl is available, for the guest to set its hypercall page. 8593 8594If KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL is also set, the same flag may also be 8595provided in the flags to KVM_XEN_HVM_CONFIG, without providing hypercall page 8596contents, to request that KVM generate hypercall page content automatically 8597and also enable interception of guest hypercalls with KVM_EXIT_XEN. 8598 8599The KVM_XEN_HVM_CONFIG_SHARED_INFO flag indicates the availability of the 8600KVM_XEN_HVM_SET_ATTR, KVM_XEN_HVM_GET_ATTR, KVM_XEN_VCPU_SET_ATTR and 8601KVM_XEN_VCPU_GET_ATTR ioctls, as well as the delivery of exception vectors 8602for event channel upcalls when the evtchn_upcall_pending field of a vcpu's 8603vcpu_info is set. 8604 8605The KVM_XEN_HVM_CONFIG_RUNSTATE flag indicates that the runstate-related 8606features KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR/_CURRENT/_DATA/_ADJUST are 8607supported by the KVM_XEN_VCPU_SET_ATTR/KVM_XEN_VCPU_GET_ATTR ioctls. 8608 8609The KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL flag indicates that IRQ routing entries 8610of the type KVM_IRQ_ROUTING_XEN_EVTCHN are supported, with the priority 8611field set to indicate 2 level event channel delivery. 8612 8613The KVM_XEN_HVM_CONFIG_EVTCHN_SEND flag indicates that KVM supports 8614injecting event channel events directly into the guest with the 8615KVM_XEN_HVM_EVTCHN_SEND ioctl. It also indicates support for the 8616KVM_XEN_ATTR_TYPE_EVTCHN/XEN_VERSION HVM attributes and the 8617KVM_XEN_VCPU_ATTR_TYPE_VCPU_ID/TIMER/UPCALL_VECTOR vCPU attributes. 8618related to event channel delivery, timers, and the XENVER_version 8619interception. 8620 8621The KVM_XEN_HVM_CONFIG_RUNSTATE_UPDATE_FLAG flag indicates that KVM supports 8622the KVM_XEN_ATTR_TYPE_RUNSTATE_UPDATE_FLAG attribute in the KVM_XEN_SET_ATTR 8623and KVM_XEN_GET_ATTR ioctls. This controls whether KVM will set the 8624XEN_RUNSTATE_UPDATE flag in guest memory mapped vcpu_runstate_info during 8625updates of the runstate information. Note that versions of KVM which support 8626the RUNSTATE feature above, but not the RUNSTATE_UPDATE_FLAG feature, will 8627always set the XEN_RUNSTATE_UPDATE flag when updating the guest structure, 8628which is perhaps counterintuitive. When this flag is advertised, KVM will 8629behave more correctly, not using the XEN_RUNSTATE_UPDATE flag until/unless 8630specifically enabled (by the guest making the hypercall, causing the VMM 8631to enable the KVM_XEN_ATTR_TYPE_RUNSTATE_UPDATE_FLAG attribute). 8632 8633The KVM_XEN_HVM_CONFIG_PVCLOCK_TSC_UNSTABLE flag indicates that KVM supports 8634clearing the PVCLOCK_TSC_STABLE_BIT flag in Xen pvclock sources. This will be 8635done when the KVM_CAP_XEN_HVM ioctl sets the 8636KVM_XEN_HVM_CONFIG_PVCLOCK_TSC_UNSTABLE flag. 8637 86388.31 KVM_CAP_PPC_MULTITCE 8639------------------------- 8640 8641:Capability: KVM_CAP_PPC_MULTITCE 8642:Architectures: ppc 8643:Type: vm 8644 8645This capability means the kernel is capable of handling hypercalls 8646H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user 8647space. This significantly accelerates DMA operations for PPC KVM guests. 8648User space should expect that its handlers for these hypercalls 8649are not going to be called if user space previously registered LIOBN 8650in KVM (via KVM_CREATE_SPAPR_TCE or similar calls). 8651 8652In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest, 8653user space might have to advertise it for the guest. For example, 8654IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is 8655present in the "ibm,hypertas-functions" device-tree property. 8656 8657The hypercalls mentioned above may or may not be processed successfully 8658in the kernel based fast path. If they can not be handled by the kernel, 8659they will get passed on to user space. So user space still has to have 8660an implementation for these despite the in kernel acceleration. 8661 8662This capability is always enabled. 8663 86648.32 KVM_CAP_PTP_KVM 8665-------------------- 8666 8667:Architectures: arm64 8668 8669This capability indicates that the KVM virtual PTP service is 8670supported in the host. A VMM can check whether the service is 8671available to the guest on migration. 8672 86738.33 KVM_CAP_HYPERV_ENFORCE_CPUID 8674--------------------------------- 8675 8676Architectures: x86 8677 8678When enabled, KVM will disable emulated Hyper-V features provided to the 8679guest according to the bits Hyper-V CPUID feature leaves. Otherwise, all 8680currently implemented Hyper-V features are provided unconditionally when 8681Hyper-V identification is set in the HYPERV_CPUID_INTERFACE (0x40000001) 8682leaf. 8683 86848.34 KVM_CAP_EXIT_HYPERCALL 8685--------------------------- 8686 8687:Capability: KVM_CAP_EXIT_HYPERCALL 8688:Architectures: x86 8689:Type: vm 8690 8691This capability, if enabled, will cause KVM to exit to userspace 8692with KVM_EXIT_HYPERCALL exit reason to process some hypercalls. 8693 8694Calling KVM_CHECK_EXTENSION for this capability will return a bitmask 8695of hypercalls that can be configured to exit to userspace. 8696Right now, the only such hypercall is KVM_HC_MAP_GPA_RANGE. 8697 8698The argument to KVM_ENABLE_CAP is also a bitmask, and must be a subset 8699of the result of KVM_CHECK_EXTENSION. KVM will forward to userspace 8700the hypercalls whose corresponding bit is in the argument, and return 8701ENOSYS for the others. 8702 87038.35 KVM_CAP_PMU_CAPABILITY 8704--------------------------- 8705 8706:Capability: KVM_CAP_PMU_CAPABILITY 8707:Architectures: x86 8708:Type: vm 8709:Parameters: arg[0] is bitmask of PMU virtualization capabilities. 8710:Returns: 0 on success, -EINVAL when arg[0] contains invalid bits 8711 8712This capability alters PMU virtualization in KVM. 8713 8714Calling KVM_CHECK_EXTENSION for this capability returns a bitmask of 8715PMU virtualization capabilities that can be adjusted on a VM. 8716 8717The argument to KVM_ENABLE_CAP is also a bitmask and selects specific 8718PMU virtualization capabilities to be applied to the VM. This can 8719only be invoked on a VM prior to the creation of VCPUs. 8720 8721At this time, KVM_PMU_CAP_DISABLE is the only capability. Setting 8722this capability will disable PMU virtualization for that VM. Usermode 8723should adjust CPUID leaf 0xA to reflect that the PMU is disabled. 8724 87258.36 KVM_CAP_ARM_SYSTEM_SUSPEND 8726------------------------------- 8727 8728:Capability: KVM_CAP_ARM_SYSTEM_SUSPEND 8729:Architectures: arm64 8730:Type: vm 8731 8732When enabled, KVM will exit to userspace with KVM_EXIT_SYSTEM_EVENT of 8733type KVM_SYSTEM_EVENT_SUSPEND to process the guest suspend request. 8734 87358.37 KVM_CAP_S390_PROTECTED_DUMP 8736-------------------------------- 8737 8738:Capability: KVM_CAP_S390_PROTECTED_DUMP 8739:Architectures: s390 8740:Type: vm 8741 8742This capability indicates that KVM and the Ultravisor support dumping 8743PV guests. The `KVM_PV_DUMP` command is available for the 8744`KVM_S390_PV_COMMAND` ioctl and the `KVM_PV_INFO` command provides 8745dump related UV data. Also the vcpu ioctl `KVM_S390_PV_CPU_COMMAND` is 8746available and supports the `KVM_PV_DUMP_CPU` subcommand. 8747 87488.38 KVM_CAP_VM_DISABLE_NX_HUGE_PAGES 8749------------------------------------- 8750 8751:Capability: KVM_CAP_VM_DISABLE_NX_HUGE_PAGES 8752:Architectures: x86 8753:Type: vm 8754:Parameters: arg[0] must be 0. 8755:Returns: 0 on success, -EPERM if the userspace process does not 8756 have CAP_SYS_BOOT, -EINVAL if args[0] is not 0 or any vCPUs have been 8757 created. 8758 8759This capability disables the NX huge pages mitigation for iTLB MULTIHIT. 8760 8761The capability has no effect if the nx_huge_pages module parameter is not set. 8762 8763This capability may only be set before any vCPUs are created. 8764 87658.39 KVM_CAP_S390_CPU_TOPOLOGY 8766------------------------------ 8767 8768:Capability: KVM_CAP_S390_CPU_TOPOLOGY 8769:Architectures: s390 8770:Type: vm 8771 8772This capability indicates that KVM will provide the S390 CPU Topology 8773facility which consist of the interpretation of the PTF instruction for 8774the function code 2 along with interception and forwarding of both the 8775PTF instruction with function codes 0 or 1 and the STSI(15,1,x) 8776instruction to the userland hypervisor. 8777 8778The stfle facility 11, CPU Topology facility, should not be indicated 8779to the guest without this capability. 8780 8781When this capability is present, KVM provides a new attribute group 8782on vm fd, KVM_S390_VM_CPU_TOPOLOGY. 8783This new attribute allows to get, set or clear the Modified Change 8784Topology Report (MTCR) bit of the SCA through the kvm_device_attr 8785structure. 8786 8787When getting the Modified Change Topology Report value, the attr->addr 8788must point to a byte where the value will be stored or retrieved from. 8789 87908.40 KVM_CAP_ARM_EAGER_SPLIT_CHUNK_SIZE 8791--------------------------------------- 8792 8793:Capability: KVM_CAP_ARM_EAGER_SPLIT_CHUNK_SIZE 8794:Architectures: arm64 8795:Type: vm 8796:Parameters: arg[0] is the new split chunk size. 8797:Returns: 0 on success, -EINVAL if any memslot was already created. 8798 8799This capability sets the chunk size used in Eager Page Splitting. 8800 8801Eager Page Splitting improves the performance of dirty-logging (used 8802in live migrations) when guest memory is backed by huge-pages. It 8803avoids splitting huge-pages (into PAGE_SIZE pages) on fault, by doing 8804it eagerly when enabling dirty logging (with the 8805KVM_MEM_LOG_DIRTY_PAGES flag for a memory region), or when using 8806KVM_CLEAR_DIRTY_LOG. 8807 8808The chunk size specifies how many pages to break at a time, using a 8809single allocation for each chunk. Bigger the chunk size, more pages 8810need to be allocated ahead of time. 8811 8812The chunk size needs to be a valid block size. The list of acceptable 8813block sizes is exposed in KVM_CAP_ARM_SUPPORTED_BLOCK_SIZES as a 881464-bit bitmap (each bit describing a block size). The default value is 88150, to disable the eager page splitting. 8816 88178.41 KVM_CAP_VM_TYPES 8818--------------------- 8819 8820:Capability: KVM_CAP_MEMORY_ATTRIBUTES 8821:Architectures: x86 8822:Type: system ioctl 8823 8824This capability returns a bitmap of support VM types. The 1-setting of bit @n 8825means the VM type with value @n is supported. Possible values of @n are:: 8826 8827 #define KVM_X86_DEFAULT_VM 0 8828 #define KVM_X86_SW_PROTECTED_VM 1 8829 #define KVM_X86_SEV_VM 2 8830 #define KVM_X86_SEV_ES_VM 3 8831 8832Note, KVM_X86_SW_PROTECTED_VM is currently only for development and testing. 8833Do not use KVM_X86_SW_PROTECTED_VM for "real" VMs, and especially not in 8834production. The behavior and effective ABI for software-protected VMs is 8835unstable. 8836 88379. Known KVM API problems 8838========================= 8839 8840In some cases, KVM's API has some inconsistencies or common pitfalls 8841that userspace need to be aware of. This section details some of 8842these issues. 8843 8844Most of them are architecture specific, so the section is split by 8845architecture. 8846 88479.1. x86 8848-------- 8849 8850``KVM_GET_SUPPORTED_CPUID`` issues 8851^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 8852 8853In general, ``KVM_GET_SUPPORTED_CPUID`` is designed so that it is possible 8854to take its result and pass it directly to ``KVM_SET_CPUID2``. This section 8855documents some cases in which that requires some care. 8856 8857Local APIC features 8858~~~~~~~~~~~~~~~~~~~ 8859 8860CPU[EAX=1]:ECX[21] (X2APIC) is reported by ``KVM_GET_SUPPORTED_CPUID``, 8861but it can only be enabled if ``KVM_CREATE_IRQCHIP`` or 8862``KVM_ENABLE_CAP(KVM_CAP_IRQCHIP_SPLIT)`` are used to enable in-kernel emulation of 8863the local APIC. 8864 8865The same is true for the ``KVM_FEATURE_PV_UNHALT`` paravirtualized feature. 8866 8867CPU[EAX=1]:ECX[24] (TSC_DEADLINE) is not reported by ``KVM_GET_SUPPORTED_CPUID``. 8868It can be enabled if ``KVM_CAP_TSC_DEADLINE_TIMER`` is present and the kernel 8869has enabled in-kernel emulation of the local APIC. 8870 8871CPU topology 8872~~~~~~~~~~~~ 8873 8874Several CPUID values include topology information for the host CPU: 88750x0b and 0x1f for Intel systems, 0x8000001e for AMD systems. Different 8876versions of KVM return different values for this information and userspace 8877should not rely on it. Currently they return all zeroes. 8878 8879If userspace wishes to set up a guest topology, it should be careful that 8880the values of these three leaves differ for each CPU. In particular, 8881the APIC ID is found in EDX for all subleaves of 0x0b and 0x1f, and in EAX 8882for 0x8000001e; the latter also encodes the core id and node id in bits 88837:0 of EBX and ECX respectively. 8884 8885Obsolete ioctls and capabilities 8886^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 8887 8888KVM_CAP_DISABLE_QUIRKS does not let userspace know which quirks are actually 8889available. Use ``KVM_CHECK_EXTENSION(KVM_CAP_DISABLE_QUIRKS2)`` instead if 8890available. 8891 8892Ordering of KVM_GET_*/KVM_SET_* ioctls 8893^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 8894 8895TBD 8896