1 The Definitive KVM (Kernel-based Virtual Machine) API Documentation
2 ===================================================================
7 The kvm API is a set of ioctls that are issued to control various aspects
8 of a virtual machine. The ioctls belong to three classes:
10 - System ioctls: These query and set global attributes which affect the
11 whole kvm subsystem. In addition a system ioctl is used to create
14 - VM ioctls: These query and set attributes that affect an entire virtual
15 machine, for example memory layout. In addition a VM ioctl is used to
16 create virtual cpus (vcpus) and devices.
18 VM ioctls must be issued from the same process (address space) that was
19 used to create the VM.
21 - vcpu ioctls: These query and set attributes that control the operation
22 of a single virtual cpu.
24 vcpu ioctls should be issued from the same thread that was used to create
25 the vcpu, except for asynchronous vcpu ioctl that are marked as such in
26 the documentation. Otherwise, the first ioctl after switching threads
27 could see a performance impact.
29 - device ioctls: These query and set attributes that control the operation
32 device ioctls must be issued from the same process (address space) that
33 was used to create the VM.
38 The kvm API is centered around file descriptors. An initial
39 open("/dev/kvm") obtains a handle to the kvm subsystem; this handle
40 can be used to issue system ioctls. A KVM_CREATE_VM ioctl on this
41 handle will create a VM file descriptor which can be used to issue VM
42 ioctls. A KVM_CREATE_VCPU or KVM_CREATE_DEVICE ioctl on a VM fd will
43 create a virtual cpu or device and return a file descriptor pointing to
44 the new resource. Finally, ioctls on a vcpu or device fd can be used
45 to control the vcpu or device. For vcpus, this includes the important
46 task of actually running guest code.
48 In general file descriptors can be migrated among processes by means
49 of fork() and the SCM_RIGHTS facility of unix domain socket. These
50 kinds of tricks are explicitly not supported by kvm. While they will
51 not cause harm to the host, their actual behavior is not guaranteed by
52 the API. See "General description" for details on the ioctl usage
53 model that is supported by KVM.
55 It is important to note that althought VM ioctls may only be issued from
56 the process that created the VM, a VM's lifecycle is associated with its
57 file descriptor, not its creator (process). In other words, the VM and
58 its resources, *including the associated address space*, are not freed
59 until the last reference to the VM's file descriptor has been released.
60 For example, if fork() is issued after ioctl(KVM_CREATE_VM), the VM will
61 not be freed until both the parent (original) process and its child have
62 put their references to the VM's file descriptor.
64 Because a VM's resources are not freed until the last reference to its
65 file descriptor is released, creating additional references to a VM via
66 via fork(), dup(), etc... without careful consideration is strongly
67 discouraged and may have unwanted side effects, e.g. memory allocated
68 by and on behalf of the VM's process may not be freed/unaccounted when
75 As of Linux 2.6.22, the KVM ABI has been stabilized: no backward
76 incompatible change are allowed. However, there is an extension
77 facility that allows backward-compatible extensions to the API to be
80 The extension mechanism is not based on the Linux version number.
81 Instead, kvm defines extension identifiers and a facility to query
82 whether a particular extension identifier is available. If it is, a
83 set of ioctls is available for application use.
89 This section describes ioctls that can be used to control kvm guests.
90 For each ioctl, the following information is provided along with a
93 Capability: which KVM extension provides this ioctl. Can be 'basic',
94 which means that is will be provided by any kernel that supports
95 API version 12 (see section 4.1), a KVM_CAP_xyz constant, which
96 means availability needs to be checked with KVM_CHECK_EXTENSION
97 (see section 4.4), or 'none' which means that while not all kernels
98 support this ioctl, there's no capability bit to check its
99 availability: for kernels that don't support the ioctl,
100 the ioctl returns -ENOTTY.
102 Architectures: which instruction set architectures provide this ioctl.
103 x86 includes both i386 and x86_64.
105 Type: system, vm, or vcpu.
107 Parameters: what parameters are accepted by the ioctl.
109 Returns: the return value. General error numbers (EBADF, ENOMEM, EINVAL)
110 are not detailed, but errors with specific meanings are.
113 4.1 KVM_GET_API_VERSION
119 Returns: the constant KVM_API_VERSION (=12)
121 This identifies the API version as the stable kvm API. It is not
122 expected that this number will change. However, Linux 2.6.20 and
123 2.6.21 report earlier versions; these are not documented and not
124 supported. Applications should refuse to run if KVM_GET_API_VERSION
125 returns a value other than 12. If this check passes, all ioctls
126 described as 'basic' will be available.
134 Parameters: machine type identifier (KVM_VM_*)
135 Returns: a VM fd that can be used to control the new virtual machine.
137 The new VM has no virtual cpus and no memory.
138 You probably want to use 0 as machine type.
140 In order to create user controlled virtual machines on S390, check
141 KVM_CAP_S390_UCONTROL and use the flag KVM_VM_S390_UCONTROL as
142 privileged user (CAP_SYS_ADMIN).
144 To use hardware assisted virtualization on MIPS (VZ ASE) rather than
145 the default trap & emulate implementation (which changes the virtual
146 memory layout to fit in user mode), check KVM_CAP_MIPS_VZ and use the
150 On arm64, the physical address size for a VM (IPA Size limit) is limited
151 to 40bits by default. The limit can be configured if the host supports the
152 extension KVM_CAP_ARM_VM_IPA_SIZE. When supported, use
153 KVM_VM_TYPE_ARM_IPA_SIZE(IPA_Bits) to set the size in the machine type
154 identifier, where IPA_Bits is the maximum width of any physical
155 address used by the VM. The IPA_Bits is encoded in bits[7-0] of the
156 machine type identifier.
158 e.g, to configure a guest to use 48bit physical address size :
160 vm_fd = ioctl(dev_fd, KVM_CREATE_VM, KVM_VM_TYPE_ARM_IPA_SIZE(48));
162 The requested size (IPA_Bits) must be :
163 0 - Implies default size, 40bits (for backward compatibility)
167 N - Implies N bits, where N is a positive integer such that,
168 32 <= N <= Host_IPA_Limit
170 Host_IPA_Limit is the maximum possible value for IPA_Bits on the host and
171 is dependent on the CPU capability and the kernel configuration. The limit can
172 be retrieved using KVM_CAP_ARM_VM_IPA_SIZE of the KVM_CHECK_EXTENSION
175 Creation of the VM will fail if the requested IPA size (whether it is
176 implicit or explicit) is unsupported on the host.
178 Please note that configuring the IPA size does not affect the capability
179 exposed by the guest CPUs in ID_AA64MMFR0_EL1[PARange]. It only affects
180 size of the address translated by the stage2 level (guest physical to
181 host physical address translations).
184 4.3 KVM_GET_MSR_INDEX_LIST, KVM_GET_MSR_FEATURE_INDEX_LIST
186 Capability: basic, KVM_CAP_GET_MSR_FEATURES for KVM_GET_MSR_FEATURE_INDEX_LIST
189 Parameters: struct kvm_msr_list (in/out)
190 Returns: 0 on success; -1 on error
192 EFAULT: the msr index list cannot be read from or written to
193 E2BIG: the msr index list is to be to fit in the array specified by
196 struct kvm_msr_list {
197 __u32 nmsrs; /* number of msrs in entries */
201 The user fills in the size of the indices array in nmsrs, and in return
202 kvm adjusts nmsrs to reflect the actual number of msrs and fills in the
203 indices array with their numbers.
205 KVM_GET_MSR_INDEX_LIST returns the guest msrs that are supported. The list
206 varies by kvm version and host processor, but does not change otherwise.
208 Note: if kvm indicates supports MCE (KVM_CAP_MCE), then the MCE bank MSRs are
209 not returned in the MSR list, as different vcpus can have a different number
210 of banks, as set via the KVM_X86_SETUP_MCE ioctl.
212 KVM_GET_MSR_FEATURE_INDEX_LIST returns the list of MSRs that can be passed
213 to the KVM_GET_MSRS system ioctl. This lets userspace probe host capabilities
214 and processor features that are exposed via MSRs (e.g., VMX capabilities).
215 This list also varies by kvm version and host processor, but does not change
219 4.4 KVM_CHECK_EXTENSION
221 Capability: basic, KVM_CAP_CHECK_EXTENSION_VM for vm ioctl
223 Type: system ioctl, vm ioctl
224 Parameters: extension identifier (KVM_CAP_*)
225 Returns: 0 if unsupported; 1 (or some other positive integer) if supported
227 The API allows the application to query about extensions to the core
228 kvm API. Userspace passes an extension identifier (an integer) and
229 receives an integer that describes the extension availability.
230 Generally 0 means no and 1 means yes, but some extensions may report
231 additional information in the integer return value.
233 Based on their initialization different VMs may have different capabilities.
234 It is thus encouraged to use the vm ioctl to query for capabilities (available
235 with KVM_CAP_CHECK_EXTENSION_VM on the vm fd)
237 4.5 KVM_GET_VCPU_MMAP_SIZE
243 Returns: size of vcpu mmap area, in bytes
245 The KVM_RUN ioctl (cf.) communicates with userspace via a shared
246 memory region. This ioctl returns the size of that region. See the
247 KVM_RUN documentation for details.
250 4.6 KVM_SET_MEMORY_REGION
255 Parameters: struct kvm_memory_region (in)
256 Returns: 0 on success, -1 on error
258 This ioctl is obsolete and has been removed.
266 Parameters: vcpu id (apic id on x86)
267 Returns: vcpu fd on success, -1 on error
269 This API adds a vcpu to a virtual machine. No more than max_vcpus may be added.
270 The vcpu id is an integer in the range [0, max_vcpu_id).
272 The recommended max_vcpus value can be retrieved using the KVM_CAP_NR_VCPUS of
273 the KVM_CHECK_EXTENSION ioctl() at run-time.
274 The maximum possible value for max_vcpus can be retrieved using the
275 KVM_CAP_MAX_VCPUS of the KVM_CHECK_EXTENSION ioctl() at run-time.
277 If the KVM_CAP_NR_VCPUS does not exist, you should assume that max_vcpus is 4
279 If the KVM_CAP_MAX_VCPUS does not exist, you should assume that max_vcpus is
280 same as the value returned from KVM_CAP_NR_VCPUS.
282 The maximum possible value for max_vcpu_id can be retrieved using the
283 KVM_CAP_MAX_VCPU_ID of the KVM_CHECK_EXTENSION ioctl() at run-time.
285 If the KVM_CAP_MAX_VCPU_ID does not exist, you should assume that max_vcpu_id
286 is the same as the value returned from KVM_CAP_MAX_VCPUS.
288 On powerpc using book3s_hv mode, the vcpus are mapped onto virtual
289 threads in one or more virtual CPU cores. (This is because the
290 hardware requires all the hardware threads in a CPU core to be in the
291 same partition.) The KVM_CAP_PPC_SMT capability indicates the number
292 of vcpus per virtual core (vcore). The vcore id is obtained by
293 dividing the vcpu id by the number of vcpus per vcore. The vcpus in a
294 given vcore will always be in the same physical core as each other
295 (though that might be a different physical core from time to time).
296 Userspace can control the threading (SMT) mode of the guest by its
297 allocation of vcpu ids. For example, if userspace wants
298 single-threaded guest vcpus, it should make all vcpu ids be a multiple
299 of the number of vcpus per vcore.
301 For virtual cpus that have been created with S390 user controlled virtual
302 machines, the resulting vcpu fd can be memory mapped at page offset
303 KVM_S390_SIE_PAGE_OFFSET in order to obtain a memory map of the virtual
304 cpu's hardware control block.
307 4.8 KVM_GET_DIRTY_LOG (vm ioctl)
312 Parameters: struct kvm_dirty_log (in/out)
313 Returns: 0 on success, -1 on error
315 /* for KVM_GET_DIRTY_LOG */
316 struct kvm_dirty_log {
320 void __user *dirty_bitmap; /* one bit per page */
325 Given a memory slot, return a bitmap containing any pages dirtied
326 since the last call to this ioctl. Bit 0 is the first page in the
327 memory slot. Ensure the entire structure is cleared to avoid padding
330 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 specifies
331 the address space for which you want to return the dirty bitmap.
332 They must be less than the value that KVM_CHECK_EXTENSION returns for
333 the KVM_CAP_MULTI_ADDRESS_SPACE capability.
335 The bits in the dirty bitmap are cleared before the ioctl returns, unless
336 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is enabled. For more information,
337 see the description of the capability.
339 4.9 KVM_SET_MEMORY_ALIAS
344 Parameters: struct kvm_memory_alias (in)
345 Returns: 0 (success), -1 (error)
347 This ioctl is obsolete and has been removed.
356 Returns: 0 on success, -1 on error
358 EINTR: an unmasked signal is pending
360 This ioctl is used to run a guest virtual cpu. While there are no
361 explicit parameters, there is an implicit parameter block that can be
362 obtained by mmap()ing the vcpu fd at offset 0, with the size given by
363 KVM_GET_VCPU_MMAP_SIZE. The parameter block is formatted as a 'struct
364 kvm_run' (see below).
370 Architectures: all except ARM, arm64
372 Parameters: struct kvm_regs (out)
373 Returns: 0 on success, -1 on error
375 Reads the general purpose registers from the vcpu.
379 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
380 __u64 rax, rbx, rcx, rdx;
381 __u64 rsi, rdi, rsp, rbp;
382 __u64 r8, r9, r10, r11;
383 __u64 r12, r13, r14, r15;
389 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
400 Architectures: all except ARM, arm64
402 Parameters: struct kvm_regs (in)
403 Returns: 0 on success, -1 on error
405 Writes the general purpose registers into the vcpu.
407 See KVM_GET_REGS for the data structure.
413 Architectures: x86, ppc
415 Parameters: struct kvm_sregs (out)
416 Returns: 0 on success, -1 on error
418 Reads special registers from the vcpu.
422 struct kvm_segment cs, ds, es, fs, gs, ss;
423 struct kvm_segment tr, ldt;
424 struct kvm_dtable gdt, idt;
425 __u64 cr0, cr2, cr3, cr4, cr8;
428 __u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64];
431 /* ppc -- see arch/powerpc/include/uapi/asm/kvm.h */
433 interrupt_bitmap is a bitmap of pending external interrupts. At most
434 one bit may be set. This interrupt has been acknowledged by the APIC
435 but not yet injected into the cpu core.
441 Architectures: x86, ppc
443 Parameters: struct kvm_sregs (in)
444 Returns: 0 on success, -1 on error
446 Writes special registers into the vcpu. See KVM_GET_SREGS for the
455 Parameters: struct kvm_translation (in/out)
456 Returns: 0 on success, -1 on error
458 Translates a virtual address according to the vcpu's current address
461 struct kvm_translation {
463 __u64 linear_address;
466 __u64 physical_address;
477 Architectures: x86, ppc, mips
479 Parameters: struct kvm_interrupt (in)
480 Returns: 0 on success, negative on failure.
482 Queues a hardware interrupt vector to be injected.
484 /* for KVM_INTERRUPT */
485 struct kvm_interrupt {
492 Returns: 0 on success,
493 -EEXIST if an interrupt is already enqueued
494 -EINVAL the the irq number is invalid
495 -ENXIO if the PIC is in the kernel
496 -EFAULT if the pointer is invalid
498 Note 'irq' is an interrupt vector, not an interrupt pin or line. This
499 ioctl is useful if the in-kernel PIC is not used.
503 Queues an external interrupt to be injected. This ioctl is overleaded
504 with 3 different irq values:
508 This injects an edge type external interrupt into the guest once it's ready
509 to receive interrupts. When injected, the interrupt is done.
511 b) KVM_INTERRUPT_UNSET
513 This unsets any pending interrupt.
515 Only available with KVM_CAP_PPC_UNSET_IRQ.
517 c) KVM_INTERRUPT_SET_LEVEL
519 This injects a level type external interrupt into the guest context. The
520 interrupt stays pending until a specific ioctl with KVM_INTERRUPT_UNSET
523 Only available with KVM_CAP_PPC_IRQ_LEVEL.
525 Note that any value for 'irq' other than the ones stated above is invalid
526 and incurs unexpected behavior.
528 This is an asynchronous vcpu ioctl and can be invoked from any thread.
532 Queues an external interrupt to be injected into the virtual CPU. A negative
533 interrupt number dequeues the interrupt.
535 This is an asynchronous vcpu ioctl and can be invoked from any thread.
546 Support for this has been removed. Use KVM_SET_GUEST_DEBUG instead.
551 Capability: basic (vcpu), KVM_CAP_GET_MSR_FEATURES (system)
553 Type: system ioctl, vcpu ioctl
554 Parameters: struct kvm_msrs (in/out)
555 Returns: number of msrs successfully returned;
558 When used as a system ioctl:
559 Reads the values of MSR-based features that are available for the VM. This
560 is similar to KVM_GET_SUPPORTED_CPUID, but it returns MSR indices and values.
561 The list of msr-based features can be obtained using KVM_GET_MSR_FEATURE_INDEX_LIST
564 When used as a vcpu ioctl:
565 Reads model-specific registers from the vcpu. Supported msr indices can
566 be obtained using KVM_GET_MSR_INDEX_LIST in a system ioctl.
569 __u32 nmsrs; /* number of msrs in entries */
572 struct kvm_msr_entry entries[0];
575 struct kvm_msr_entry {
581 Application code should set the 'nmsrs' member (which indicates the
582 size of the entries array) and the 'index' member of each array entry.
583 kvm will fill in the 'data' member.
591 Parameters: struct kvm_msrs (in)
592 Returns: number of msrs successfully set (see below), -1 on error
594 Writes model-specific registers to the vcpu. See KVM_GET_MSRS for the
597 Application code should set the 'nmsrs' member (which indicates the
598 size of the entries array), and the 'index' and 'data' members of each
601 It tries to set the MSRs in array entries[] one by one. If setting an MSR
602 fails, e.g., due to setting reserved bits, the MSR isn't supported/emulated
603 by KVM, etc..., it stops processing the MSR list and returns the number of
604 MSRs that have been set successfully.
612 Parameters: struct kvm_cpuid (in)
613 Returns: 0 on success, -1 on error
615 Defines the vcpu responses to the cpuid instruction. Applications
616 should use the KVM_SET_CPUID2 ioctl if available.
619 struct kvm_cpuid_entry {
628 /* for KVM_SET_CPUID */
632 struct kvm_cpuid_entry entries[0];
636 4.21 KVM_SET_SIGNAL_MASK
641 Parameters: struct kvm_signal_mask (in)
642 Returns: 0 on success, -1 on error
644 Defines which signals are blocked during execution of KVM_RUN. This
645 signal mask temporarily overrides the threads signal mask. Any
646 unblocked signal received (except SIGKILL and SIGSTOP, which retain
647 their traditional behaviour) will cause KVM_RUN to return with -EINTR.
649 Note the signal will only be delivered if not blocked by the original
652 /* for KVM_SET_SIGNAL_MASK */
653 struct kvm_signal_mask {
664 Parameters: struct kvm_fpu (out)
665 Returns: 0 on success, -1 on error
667 Reads the floating point state from the vcpu.
669 /* for KVM_GET_FPU and KVM_SET_FPU */
674 __u8 ftwx; /* in fxsave format */
690 Parameters: struct kvm_fpu (in)
691 Returns: 0 on success, -1 on error
693 Writes the floating point state to the vcpu.
695 /* for KVM_GET_FPU and KVM_SET_FPU */
700 __u8 ftwx; /* in fxsave format */
711 4.24 KVM_CREATE_IRQCHIP
713 Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQCHIP (s390)
714 Architectures: x86, ARM, arm64, s390
717 Returns: 0 on success, -1 on error
719 Creates an interrupt controller model in the kernel.
720 On x86, creates a virtual ioapic, a virtual PIC (two PICs, nested), and sets up
721 future vcpus to have a local APIC. IRQ routing for GSIs 0-15 is set to both
722 PIC and IOAPIC; GSI 16-23 only go to the IOAPIC.
723 On ARM/arm64, a GICv2 is created. Any other GIC versions require the usage of
724 KVM_CREATE_DEVICE, which also supports creating a GICv2. Using
725 KVM_CREATE_DEVICE is preferred over KVM_CREATE_IRQCHIP for GICv2.
726 On s390, a dummy irq routing table is created.
728 Note that on s390 the KVM_CAP_S390_IRQCHIP vm capability needs to be enabled
729 before KVM_CREATE_IRQCHIP can be used.
734 Capability: KVM_CAP_IRQCHIP
735 Architectures: x86, arm, arm64
737 Parameters: struct kvm_irq_level
738 Returns: 0 on success, -1 on error
740 Sets the level of a GSI input to the interrupt controller model in the kernel.
741 On some architectures it is required that an interrupt controller model has
742 been previously created with KVM_CREATE_IRQCHIP. Note that edge-triggered
743 interrupts require the level to be set to 1 and then back to 0.
745 On real hardware, interrupt pins can be active-low or active-high. This
746 does not matter for the level field of struct kvm_irq_level: 1 always
747 means active (asserted), 0 means inactive (deasserted).
749 x86 allows the operating system to program the interrupt polarity
750 (active-low/active-high) for level-triggered interrupts, and KVM used
751 to consider the polarity. However, due to bitrot in the handling of
752 active-low interrupts, the above convention is now valid on x86 too.
753 This is signaled by KVM_CAP_X86_IOAPIC_POLARITY_IGNORED. Userspace
754 should not present interrupts to the guest as active-low unless this
755 capability is present (or unless it is not using the in-kernel irqchip,
759 ARM/arm64 can signal an interrupt either at the CPU level, or at the
760 in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to
761 use PPIs designated for specific cpus. The irq field is interpreted
764 Â bits: | 31 ... 28 | 27 ... 24 | 23 ... 16 | 15 ... 0 |
765 field: | vcpu2_index | irq_type | vcpu_index | irq_id |
767 The irq_type field has the following values:
768 - irq_type[0]: out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ
769 - irq_type[1]: in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.)
770 (the vcpu_index field is ignored)
771 - irq_type[2]: in-kernel GIC: PPI, irq_id between 16 and 31 (incl.)
773 (The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs)
775 In both cases, level is used to assert/deassert the line.
777 When KVM_CAP_ARM_IRQ_LINE_LAYOUT_2 is supported, the target vcpu is
778 identified as (256 * vcpu2_index + vcpu_index). Otherwise, vcpu2_index
781 Note that on arm/arm64, the KVM_CAP_IRQCHIP capability only conditions
782 injection of interrupts for the in-kernel irqchip. KVM_IRQ_LINE can always
783 be used for a userspace interrupt controller.
785 struct kvm_irq_level {
788 __s32 status; /* not used for KVM_IRQ_LEVEL */
790 __u32 level; /* 0 or 1 */
796 Capability: KVM_CAP_IRQCHIP
799 Parameters: struct kvm_irqchip (in/out)
800 Returns: 0 on success, -1 on error
802 Reads the state of a kernel interrupt controller created with
803 KVM_CREATE_IRQCHIP into a buffer provided by the caller.
806 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
809 char dummy[512]; /* reserving space */
810 struct kvm_pic_state pic;
811 struct kvm_ioapic_state ioapic;
818 Capability: KVM_CAP_IRQCHIP
821 Parameters: struct kvm_irqchip (in)
822 Returns: 0 on success, -1 on error
824 Sets the state of a kernel interrupt controller created with
825 KVM_CREATE_IRQCHIP from a buffer provided by the caller.
828 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
831 char dummy[512]; /* reserving space */
832 struct kvm_pic_state pic;
833 struct kvm_ioapic_state ioapic;
838 4.28 KVM_XEN_HVM_CONFIG
840 Capability: KVM_CAP_XEN_HVM
843 Parameters: struct kvm_xen_hvm_config (in)
844 Returns: 0 on success, -1 on error
846 Sets the MSR that the Xen HVM guest uses to initialize its hypercall
847 page, and provides the starting address and size of the hypercall
848 blobs in userspace. When the guest writes the MSR, kvm copies one
849 page of a blob (32- or 64-bit, depending on the vcpu mode) to guest
852 struct kvm_xen_hvm_config {
865 Capability: KVM_CAP_ADJUST_CLOCK
868 Parameters: struct kvm_clock_data (out)
869 Returns: 0 on success, -1 on error
871 Gets the current timestamp of kvmclock as seen by the current guest. In
872 conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios
875 When KVM_CAP_ADJUST_CLOCK is passed to KVM_CHECK_EXTENSION, it returns the
876 set of bits that KVM can return in struct kvm_clock_data's flag member.
878 The only flag defined now is KVM_CLOCK_TSC_STABLE. If set, the returned
879 value is the exact kvmclock value seen by all VCPUs at the instant
880 when KVM_GET_CLOCK was called. If clear, the returned value is simply
881 CLOCK_MONOTONIC plus a constant offset; the offset can be modified
882 with KVM_SET_CLOCK. KVM will try to make all VCPUs follow this clock,
883 but the exact value read by each VCPU could differ, because the host
886 struct kvm_clock_data {
887 __u64 clock; /* kvmclock current value */
895 Capability: KVM_CAP_ADJUST_CLOCK
898 Parameters: struct kvm_clock_data (in)
899 Returns: 0 on success, -1 on error
901 Sets the current timestamp of kvmclock to the value specified in its parameter.
902 In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios
905 struct kvm_clock_data {
906 __u64 clock; /* kvmclock current value */
912 4.31 KVM_GET_VCPU_EVENTS
914 Capability: KVM_CAP_VCPU_EVENTS
915 Extended by: KVM_CAP_INTR_SHADOW
916 Architectures: x86, arm, arm64
918 Parameters: struct kvm_vcpu_event (out)
919 Returns: 0 on success, -1 on error
923 Gets currently pending exceptions, interrupts, and NMIs as well as related
926 struct kvm_vcpu_events {
955 __u8 exception_has_payload;
956 __u64 exception_payload;
959 The following bits are defined in the flags field:
961 - KVM_VCPUEVENT_VALID_SHADOW may be set to signal that
962 interrupt.shadow contains a valid state.
964 - KVM_VCPUEVENT_VALID_SMM may be set to signal that smi contains a
967 - KVM_VCPUEVENT_VALID_PAYLOAD may be set to signal that the
968 exception_has_payload, exception_payload, and exception.pending
969 fields contain a valid state. This bit will be set whenever
970 KVM_CAP_EXCEPTION_PAYLOAD is enabled.
974 If the guest accesses a device that is being emulated by the host kernel in
975 such a way that a real device would generate a physical SError, KVM may make
976 a virtual SError pending for that VCPU. This system error interrupt remains
977 pending until the guest takes the exception by unmasking PSTATE.A.
979 Running the VCPU may cause it to take a pending SError, or make an access that
980 causes an SError to become pending. The event's description is only valid while
981 the VPCU is not running.
983 This API provides a way to read and write the pending 'event' state that is not
984 visible to the guest. To save, restore or migrate a VCPU the struct representing
985 the state can be read then written using this GET/SET API, along with the other
986 guest-visible registers. It is not possible to 'cancel' an SError that has been
989 A device being emulated in user-space may also wish to generate an SError. To do
990 this the events structure can be populated by user-space. The current state
991 should be read first, to ensure no existing SError is pending. If an existing
992 SError is pending, the architecture's 'Multiple SError interrupts' rules should
993 be followed. (2.5.3 of DDI0587.a "ARM Reliability, Availability, and
994 Serviceability (RAS) Specification").
996 SError exceptions always have an ESR value. Some CPUs have the ability to
997 specify what the virtual SError's ESR value should be. These systems will
998 advertise KVM_CAP_ARM_INJECT_SERROR_ESR. In this case exception.has_esr will
999 always have a non-zero value when read, and the agent making an SError pending
1000 should specify the ISS field in the lower 24 bits of exception.serror_esr. If
1001 the system supports KVM_CAP_ARM_INJECT_SERROR_ESR, but user-space sets the events
1002 with exception.has_esr as zero, KVM will choose an ESR.
1004 Specifying exception.has_esr on a system that does not support it will return
1005 -EINVAL. Setting anything other than the lower 24bits of exception.serror_esr
1006 will return -EINVAL.
1008 struct kvm_vcpu_events {
1010 __u8 serror_pending;
1011 __u8 serror_has_esr;
1012 /* Align it to 8 bytes */
1019 4.32 KVM_SET_VCPU_EVENTS
1021 Capability: KVM_CAP_VCPU_EVENTS
1022 Extended by: KVM_CAP_INTR_SHADOW
1023 Architectures: x86, arm, arm64
1025 Parameters: struct kvm_vcpu_event (in)
1026 Returns: 0 on success, -1 on error
1030 Set pending exceptions, interrupts, and NMIs as well as related states of the
1033 See KVM_GET_VCPU_EVENTS for the data structure.
1035 Fields that may be modified asynchronously by running VCPUs can be excluded
1036 from the update. These fields are nmi.pending, sipi_vector, smi.smm,
1037 smi.pending. Keep the corresponding bits in the flags field cleared to
1038 suppress overwriting the current in-kernel state. The bits are:
1040 KVM_VCPUEVENT_VALID_NMI_PENDING - transfer nmi.pending to the kernel
1041 KVM_VCPUEVENT_VALID_SIPI_VECTOR - transfer sipi_vector
1042 KVM_VCPUEVENT_VALID_SMM - transfer the smi sub-struct.
1044 If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in
1045 the flags field to signal that interrupt.shadow contains a valid state and
1046 shall be written into the VCPU.
1048 KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available.
1050 If KVM_CAP_EXCEPTION_PAYLOAD is enabled, KVM_VCPUEVENT_VALID_PAYLOAD
1051 can be set in the flags field to signal that the
1052 exception_has_payload, exception_payload, and exception.pending fields
1053 contain a valid state and shall be written into the VCPU.
1057 Set the pending SError exception state for this VCPU. It is not possible to
1058 'cancel' an Serror that has been made pending.
1060 See KVM_GET_VCPU_EVENTS for the data structure.
1063 4.33 KVM_GET_DEBUGREGS
1065 Capability: KVM_CAP_DEBUGREGS
1068 Parameters: struct kvm_debugregs (out)
1069 Returns: 0 on success, -1 on error
1071 Reads debug registers from the vcpu.
1073 struct kvm_debugregs {
1082 4.34 KVM_SET_DEBUGREGS
1084 Capability: KVM_CAP_DEBUGREGS
1087 Parameters: struct kvm_debugregs (in)
1088 Returns: 0 on success, -1 on error
1090 Writes debug registers into the vcpu.
1092 See KVM_GET_DEBUGREGS for the data structure. The flags field is unused
1093 yet and must be cleared on entry.
1096 4.35 KVM_SET_USER_MEMORY_REGION
1098 Capability: KVM_CAP_USER_MEMORY
1101 Parameters: struct kvm_userspace_memory_region (in)
1102 Returns: 0 on success, -1 on error
1104 struct kvm_userspace_memory_region {
1107 __u64 guest_phys_addr;
1108 __u64 memory_size; /* bytes */
1109 __u64 userspace_addr; /* start of the userspace allocated memory */
1112 /* for kvm_memory_region::flags */
1113 #define KVM_MEM_LOG_DIRTY_PAGES (1UL << 0)
1114 #define KVM_MEM_READONLY (1UL << 1)
1116 This ioctl allows the user to create, modify or delete a guest physical
1117 memory slot. Bits 0-15 of "slot" specify the slot id and this value
1118 should be less than the maximum number of user memory slots supported per
1119 VM. The maximum allowed slots can be queried using KVM_CAP_NR_MEMSLOTS.
1120 Slots may not overlap in guest physical address space.
1122 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot"
1123 specifies the address space which is being modified. They must be
1124 less than the value that KVM_CHECK_EXTENSION returns for the
1125 KVM_CAP_MULTI_ADDRESS_SPACE capability. Slots in separate address spaces
1126 are unrelated; the restriction on overlapping slots only applies within
1129 Deleting a slot is done by passing zero for memory_size. When changing
1130 an existing slot, it may be moved in the guest physical memory space,
1131 or its flags may be modified, but it may not be resized.
1133 Memory for the region is taken starting at the address denoted by the
1134 field userspace_addr, which must point at user addressable memory for
1135 the entire memory slot size. Any object may back this memory, including
1136 anonymous memory, ordinary files, and hugetlbfs.
1138 On architectures that support a form of address tagging, userspace_addr must
1139 be an untagged address.
1141 It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr
1142 be identical. This allows large pages in the guest to be backed by large
1145 The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and
1146 KVM_MEM_READONLY. The former can be set to instruct KVM to keep track of
1147 writes to memory within the slot. See KVM_GET_DIRTY_LOG ioctl to know how to
1148 use it. The latter can be set, if KVM_CAP_READONLY_MEM capability allows it,
1149 to make a new slot read-only. In this case, writes to this memory will be
1150 posted to userspace as KVM_EXIT_MMIO exits.
1152 When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of
1153 the memory region are automatically reflected into the guest. For example, an
1154 mmap() that affects the region will be made visible immediately. Another
1155 example is madvise(MADV_DROP).
1157 It is recommended to use this API instead of the KVM_SET_MEMORY_REGION ioctl.
1158 The KVM_SET_MEMORY_REGION does not allow fine grained control over memory
1159 allocation and is deprecated.
1162 4.36 KVM_SET_TSS_ADDR
1164 Capability: KVM_CAP_SET_TSS_ADDR
1167 Parameters: unsigned long tss_address (in)
1168 Returns: 0 on success, -1 on error
1170 This ioctl defines the physical address of a three-page region in the guest
1171 physical address space. The region must be within the first 4GB of the
1172 guest physical address space and must not conflict with any memory slot
1173 or any mmio address. The guest may malfunction if it accesses this memory
1176 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1177 because of a quirk in the virtualization implementation (see the internals
1178 documentation when it pops into existence).
1183 Capability: KVM_CAP_ENABLE_CAP
1184 Architectures: mips, ppc, s390
1186 Parameters: struct kvm_enable_cap (in)
1187 Returns: 0 on success; -1 on error
1189 Capability: KVM_CAP_ENABLE_CAP_VM
1192 Parameters: struct kvm_enable_cap (in)
1193 Returns: 0 on success; -1 on error
1195 +Not all extensions are enabled by default. Using this ioctl the application
1196 can enable an extension, making it available to the guest.
1198 On systems that do not support this ioctl, it always fails. On systems that
1199 do support it, it only works for extensions that are supported for enablement.
1201 To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should
1204 struct kvm_enable_cap {
1208 The capability that is supposed to get enabled.
1212 A bitfield indicating future enhancements. Has to be 0 for now.
1216 Arguments for enabling a feature. If a feature needs initial values to
1217 function properly, this is the place to put them.
1222 The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl
1223 for vm-wide capabilities.
1225 4.38 KVM_GET_MP_STATE
1227 Capability: KVM_CAP_MP_STATE
1228 Architectures: x86, s390, arm, arm64
1230 Parameters: struct kvm_mp_state (out)
1231 Returns: 0 on success; -1 on error
1233 struct kvm_mp_state {
1237 Returns the vcpu's current "multiprocessing state" (though also valid on
1238 uniprocessor guests).
1240 Possible values are:
1242 - KVM_MP_STATE_RUNNABLE: the vcpu is currently running [x86,arm/arm64]
1243 - KVM_MP_STATE_UNINITIALIZED: the vcpu is an application processor (AP)
1244 which has not yet received an INIT signal [x86]
1245 - KVM_MP_STATE_INIT_RECEIVED: the vcpu has received an INIT signal, and is
1246 now ready for a SIPI [x86]
1247 - KVM_MP_STATE_HALTED: the vcpu has executed a HLT instruction and
1248 is waiting for an interrupt [x86]
1249 - KVM_MP_STATE_SIPI_RECEIVED: the vcpu has just received a SIPI (vector
1250 accessible via KVM_GET_VCPU_EVENTS) [x86]
1251 - KVM_MP_STATE_STOPPED: the vcpu is stopped [s390,arm/arm64]
1252 - KVM_MP_STATE_CHECK_STOP: the vcpu is in a special error state [s390]
1253 - KVM_MP_STATE_OPERATING: the vcpu is operating (running or halted)
1255 - KVM_MP_STATE_LOAD: the vcpu is in a special load/startup state
1258 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1259 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1260 these architectures.
1264 The only states that are valid are KVM_MP_STATE_STOPPED and
1265 KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not.
1267 4.39 KVM_SET_MP_STATE
1269 Capability: KVM_CAP_MP_STATE
1270 Architectures: x86, s390, arm, arm64
1272 Parameters: struct kvm_mp_state (in)
1273 Returns: 0 on success; -1 on error
1275 Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for
1278 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1279 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1280 these architectures.
1284 The only states that are valid are KVM_MP_STATE_STOPPED and
1285 KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not.
1287 4.40 KVM_SET_IDENTITY_MAP_ADDR
1289 Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR
1292 Parameters: unsigned long identity (in)
1293 Returns: 0 on success, -1 on error
1295 This ioctl defines the physical address of a one-page region in the guest
1296 physical address space. The region must be within the first 4GB of the
1297 guest physical address space and must not conflict with any memory slot
1298 or any mmio address. The guest may malfunction if it accesses this memory
1301 Setting the address to 0 will result in resetting the address to its default
1304 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1305 because of a quirk in the virtualization implementation (see the internals
1306 documentation when it pops into existence).
1308 Fails if any VCPU has already been created.
1310 4.41 KVM_SET_BOOT_CPU_ID
1312 Capability: KVM_CAP_SET_BOOT_CPU_ID
1315 Parameters: unsigned long vcpu_id
1316 Returns: 0 on success, -1 on error
1318 Define which vcpu is the Bootstrap Processor (BSP). Values are the same
1319 as the vcpu id in KVM_CREATE_VCPU. If this ioctl is not called, the default
1325 Capability: KVM_CAP_XSAVE
1328 Parameters: struct kvm_xsave (out)
1329 Returns: 0 on success, -1 on error
1335 This ioctl would copy current vcpu's xsave struct to the userspace.
1340 Capability: KVM_CAP_XSAVE
1343 Parameters: struct kvm_xsave (in)
1344 Returns: 0 on success, -1 on error
1350 This ioctl would copy userspace's xsave struct to the kernel.
1355 Capability: KVM_CAP_XCRS
1358 Parameters: struct kvm_xcrs (out)
1359 Returns: 0 on success, -1 on error
1370 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1374 This ioctl would copy current vcpu's xcrs to the userspace.
1379 Capability: KVM_CAP_XCRS
1382 Parameters: struct kvm_xcrs (in)
1383 Returns: 0 on success, -1 on error
1394 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1398 This ioctl would set vcpu's xcr to the value userspace specified.
1401 4.46 KVM_GET_SUPPORTED_CPUID
1403 Capability: KVM_CAP_EXT_CPUID
1406 Parameters: struct kvm_cpuid2 (in/out)
1407 Returns: 0 on success, -1 on error
1412 struct kvm_cpuid_entry2 entries[0];
1415 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
1416 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1)
1417 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2)
1419 struct kvm_cpuid_entry2 {
1430 This ioctl returns x86 cpuid features which are supported by both the
1431 hardware and kvm in its default configuration. Userspace can use the
1432 information returned by this ioctl to construct cpuid information (for
1433 KVM_SET_CPUID2) that is consistent with hardware, kernel, and
1434 userspace capabilities, and with user requirements (for example, the
1435 user may wish to constrain cpuid to emulate older hardware, or for
1436 feature consistency across a cluster).
1438 Note that certain capabilities, such as KVM_CAP_X86_DISABLE_EXITS, may
1439 expose cpuid features (e.g. MONITOR) which are not supported by kvm in
1440 its default configuration. If userspace enables such capabilities, it
1441 is responsible for modifying the results of this ioctl appropriately.
1443 Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure
1444 with the 'nent' field indicating the number of entries in the variable-size
1445 array 'entries'. If the number of entries is too low to describe the cpu
1446 capabilities, an error (E2BIG) is returned. If the number is too high,
1447 the 'nent' field is adjusted and an error (ENOMEM) is returned. If the
1448 number is just right, the 'nent' field is adjusted to the number of valid
1449 entries in the 'entries' array, which is then filled.
1451 The entries returned are the host cpuid as returned by the cpuid instruction,
1452 with unknown or unsupported features masked out. Some features (for example,
1453 x2apic), may not be present in the host cpu, but are exposed by kvm if it can
1454 emulate them efficiently. The fields in each entry are defined as follows:
1456 function: the eax value used to obtain the entry
1457 index: the ecx value used to obtain the entry (for entries that are
1459 flags: an OR of zero or more of the following:
1460 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
1461 if the index field is valid
1462 KVM_CPUID_FLAG_STATEFUL_FUNC:
1463 if cpuid for this function returns different values for successive
1464 invocations; there will be several entries with the same function,
1465 all with this flag set
1466 KVM_CPUID_FLAG_STATE_READ_NEXT:
1467 for KVM_CPUID_FLAG_STATEFUL_FUNC entries, set if this entry is
1468 the first entry to be read by a cpu
1469 eax, ebx, ecx, edx: the values returned by the cpuid instruction for
1470 this function/index combination
1472 The TSC deadline timer feature (CPUID leaf 1, ecx[24]) is always returned
1473 as false, since the feature depends on KVM_CREATE_IRQCHIP for local APIC
1474 support. Instead it is reported via
1476 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER)
1478 if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the
1479 feature in userspace, then you can enable the feature for KVM_SET_CPUID2.
1482 4.47 KVM_PPC_GET_PVINFO
1484 Capability: KVM_CAP_PPC_GET_PVINFO
1487 Parameters: struct kvm_ppc_pvinfo (out)
1488 Returns: 0 on success, !0 on error
1490 struct kvm_ppc_pvinfo {
1496 This ioctl fetches PV specific information that need to be passed to the guest
1497 using the device tree or other means from vm context.
1499 The hcall array defines 4 instructions that make up a hypercall.
1501 If any additional field gets added to this structure later on, a bit for that
1502 additional piece of information will be set in the flags bitmap.
1504 The flags bitmap is defined as:
1506 /* the host supports the ePAPR idle hcall
1507 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1<<0)
1509 4.52 KVM_SET_GSI_ROUTING
1511 Capability: KVM_CAP_IRQ_ROUTING
1512 Architectures: x86 s390 arm arm64
1514 Parameters: struct kvm_irq_routing (in)
1515 Returns: 0 on success, -1 on error
1517 Sets the GSI routing table entries, overwriting any previously set entries.
1519 On arm/arm64, GSI routing has the following limitation:
1520 - GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD.
1522 struct kvm_irq_routing {
1525 struct kvm_irq_routing_entry entries[0];
1528 No flags are specified so far, the corresponding field must be set to zero.
1530 struct kvm_irq_routing_entry {
1536 struct kvm_irq_routing_irqchip irqchip;
1537 struct kvm_irq_routing_msi msi;
1538 struct kvm_irq_routing_s390_adapter adapter;
1539 struct kvm_irq_routing_hv_sint hv_sint;
1544 /* gsi routing entry types */
1545 #define KVM_IRQ_ROUTING_IRQCHIP 1
1546 #define KVM_IRQ_ROUTING_MSI 2
1547 #define KVM_IRQ_ROUTING_S390_ADAPTER 3
1548 #define KVM_IRQ_ROUTING_HV_SINT 4
1551 - KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry
1552 type, specifies that the devid field contains a valid value. The per-VM
1553 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
1554 the device ID. If this capability is not available, userspace should
1555 never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
1558 struct kvm_irq_routing_irqchip {
1563 struct kvm_irq_routing_msi {
1573 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
1574 for the device that wrote the MSI message. For PCI, this is usually a
1575 BFD identifier in the lower 16 bits.
1577 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
1578 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
1579 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
1580 address_hi must be zero.
1582 struct kvm_irq_routing_s390_adapter {
1586 __u32 summary_offset;
1590 struct kvm_irq_routing_hv_sint {
1596 4.55 KVM_SET_TSC_KHZ
1598 Capability: KVM_CAP_TSC_CONTROL
1601 Parameters: virtual tsc_khz
1602 Returns: 0 on success, -1 on error
1604 Specifies the tsc frequency for the virtual machine. The unit of the
1608 4.56 KVM_GET_TSC_KHZ
1610 Capability: KVM_CAP_GET_TSC_KHZ
1614 Returns: virtual tsc-khz on success, negative value on error
1616 Returns the tsc frequency of the guest. The unit of the return value is
1617 KHz. If the host has unstable tsc this ioctl returns -EIO instead as an
1623 Capability: KVM_CAP_IRQCHIP
1626 Parameters: struct kvm_lapic_state (out)
1627 Returns: 0 on success, -1 on error
1629 #define KVM_APIC_REG_SIZE 0x400
1630 struct kvm_lapic_state {
1631 char regs[KVM_APIC_REG_SIZE];
1634 Reads the Local APIC registers and copies them into the input argument. The
1635 data format and layout are the same as documented in the architecture manual.
1637 If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is
1638 enabled, then the format of APIC_ID register depends on the APIC mode
1639 (reported by MSR_IA32_APICBASE) of its VCPU. x2APIC stores APIC ID in
1640 the APIC_ID register (bytes 32-35). xAPIC only allows an 8-bit APIC ID
1641 which is stored in bits 31-24 of the APIC register, or equivalently in
1642 byte 35 of struct kvm_lapic_state's regs field. KVM_GET_LAPIC must then
1643 be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR.
1645 If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state
1646 always uses xAPIC format.
1651 Capability: KVM_CAP_IRQCHIP
1654 Parameters: struct kvm_lapic_state (in)
1655 Returns: 0 on success, -1 on error
1657 #define KVM_APIC_REG_SIZE 0x400
1658 struct kvm_lapic_state {
1659 char regs[KVM_APIC_REG_SIZE];
1662 Copies the input argument into the Local APIC registers. The data format
1663 and layout are the same as documented in the architecture manual.
1665 The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's
1666 regs field) depends on the state of the KVM_CAP_X2APIC_API capability.
1667 See the note in KVM_GET_LAPIC.
1672 Capability: KVM_CAP_IOEVENTFD
1675 Parameters: struct kvm_ioeventfd (in)
1676 Returns: 0 on success, !0 on error
1678 This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address
1679 within the guest. A guest write in the registered address will signal the
1680 provided event instead of triggering an exit.
1682 struct kvm_ioeventfd {
1684 __u64 addr; /* legal pio/mmio address */
1685 __u32 len; /* 0, 1, 2, 4, or 8 bytes */
1691 For the special case of virtio-ccw devices on s390, the ioevent is matched
1692 to a subchannel/virtqueue tuple instead.
1694 The following flags are defined:
1696 #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch)
1697 #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio)
1698 #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign)
1699 #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \
1700 (1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify)
1702 If datamatch flag is set, the event will be signaled only if the written value
1703 to the registered address is equal to datamatch in struct kvm_ioeventfd.
1705 For virtio-ccw devices, addr contains the subchannel id and datamatch the
1708 With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and
1709 the kernel will ignore the length of guest write and may get a faster vmexit.
1710 The speedup may only apply to specific architectures, but the ioeventfd will
1715 Capability: KVM_CAP_SW_TLB
1718 Parameters: struct kvm_dirty_tlb (in)
1719 Returns: 0 on success, -1 on error
1721 struct kvm_dirty_tlb {
1726 This must be called whenever userspace has changed an entry in the shared
1727 TLB, prior to calling KVM_RUN on the associated vcpu.
1729 The "bitmap" field is the userspace address of an array. This array
1730 consists of a number of bits, equal to the total number of TLB entries as
1731 determined by the last successful call to KVM_CONFIG_TLB, rounded up to the
1732 nearest multiple of 64.
1734 Each bit corresponds to one TLB entry, ordered the same as in the shared TLB
1737 The array is little-endian: the bit 0 is the least significant bit of the
1738 first byte, bit 8 is the least significant bit of the second byte, etc.
1739 This avoids any complications with differing word sizes.
1741 The "num_dirty" field is a performance hint for KVM to determine whether it
1742 should skip processing the bitmap and just invalidate everything. It must
1743 be set to the number of set bits in the bitmap.
1746 4.62 KVM_CREATE_SPAPR_TCE
1748 Capability: KVM_CAP_SPAPR_TCE
1749 Architectures: powerpc
1751 Parameters: struct kvm_create_spapr_tce (in)
1752 Returns: file descriptor for manipulating the created TCE table
1754 This creates a virtual TCE (translation control entry) table, which
1755 is an IOMMU for PAPR-style virtual I/O. It is used to translate
1756 logical addresses used in virtual I/O into guest physical addresses,
1757 and provides a scatter/gather capability for PAPR virtual I/O.
1759 /* for KVM_CAP_SPAPR_TCE */
1760 struct kvm_create_spapr_tce {
1765 The liobn field gives the logical IO bus number for which to create a
1766 TCE table. The window_size field specifies the size of the DMA window
1767 which this TCE table will translate - the table will contain one 64
1768 bit TCE entry for every 4kiB of the DMA window.
1770 When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE
1771 table has been created using this ioctl(), the kernel will handle it
1772 in real mode, updating the TCE table. H_PUT_TCE calls for other
1773 liobns will cause a vm exit and must be handled by userspace.
1775 The return value is a file descriptor which can be passed to mmap(2)
1776 to map the created TCE table into userspace. This lets userspace read
1777 the entries written by kernel-handled H_PUT_TCE calls, and also lets
1778 userspace update the TCE table directly which is useful in some
1782 4.63 KVM_ALLOCATE_RMA
1784 Capability: KVM_CAP_PPC_RMA
1785 Architectures: powerpc
1787 Parameters: struct kvm_allocate_rma (out)
1788 Returns: file descriptor for mapping the allocated RMA
1790 This allocates a Real Mode Area (RMA) from the pool allocated at boot
1791 time by the kernel. An RMA is a physically-contiguous, aligned region
1792 of memory used on older POWER processors to provide the memory which
1793 will be accessed by real-mode (MMU off) accesses in a KVM guest.
1794 POWER processors support a set of sizes for the RMA that usually
1795 includes 64MB, 128MB, 256MB and some larger powers of two.
1797 /* for KVM_ALLOCATE_RMA */
1798 struct kvm_allocate_rma {
1802 The return value is a file descriptor which can be passed to mmap(2)
1803 to map the allocated RMA into userspace. The mapped area can then be
1804 passed to the KVM_SET_USER_MEMORY_REGION ioctl to establish it as the
1805 RMA for a virtual machine. The size of the RMA in bytes (which is
1806 fixed at host kernel boot time) is returned in the rma_size field of
1807 the argument structure.
1809 The KVM_CAP_PPC_RMA capability is 1 or 2 if the KVM_ALLOCATE_RMA ioctl
1810 is supported; 2 if the processor requires all virtual machines to have
1811 an RMA, or 1 if the processor can use an RMA but doesn't require it,
1812 because it supports the Virtual RMA (VRMA) facility.
1817 Capability: KVM_CAP_USER_NMI
1821 Returns: 0 on success, -1 on error
1823 Queues an NMI on the thread's vcpu. Note this is well defined only
1824 when KVM_CREATE_IRQCHIP has not been called, since this is an interface
1825 between the virtual cpu core and virtual local APIC. After KVM_CREATE_IRQCHIP
1826 has been called, this interface is completely emulated within the kernel.
1828 To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the
1829 following algorithm:
1832 - read the local APIC's state (KVM_GET_LAPIC)
1833 - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1)
1834 - if so, issue KVM_NMI
1837 Some guests configure the LINT1 NMI input to cause a panic, aiding in
1841 4.65 KVM_S390_UCAS_MAP
1843 Capability: KVM_CAP_S390_UCONTROL
1846 Parameters: struct kvm_s390_ucas_mapping (in)
1847 Returns: 0 in case of success
1849 The parameter is defined like this:
1850 struct kvm_s390_ucas_mapping {
1856 This ioctl maps the memory at "user_addr" with the length "length" to
1857 the vcpu's address space starting at "vcpu_addr". All parameters need to
1858 be aligned by 1 megabyte.
1861 4.66 KVM_S390_UCAS_UNMAP
1863 Capability: KVM_CAP_S390_UCONTROL
1866 Parameters: struct kvm_s390_ucas_mapping (in)
1867 Returns: 0 in case of success
1869 The parameter is defined like this:
1870 struct kvm_s390_ucas_mapping {
1876 This ioctl unmaps the memory in the vcpu's address space starting at
1877 "vcpu_addr" with the length "length". The field "user_addr" is ignored.
1878 All parameters need to be aligned by 1 megabyte.
1881 4.67 KVM_S390_VCPU_FAULT
1883 Capability: KVM_CAP_S390_UCONTROL
1886 Parameters: vcpu absolute address (in)
1887 Returns: 0 in case of success
1889 This call creates a page table entry on the virtual cpu's address space
1890 (for user controlled virtual machines) or the virtual machine's address
1891 space (for regular virtual machines). This only works for minor faults,
1892 thus it's recommended to access subject memory page via the user page
1893 table upfront. This is useful to handle validity intercepts for user
1894 controlled virtual machines to fault in the virtual cpu's lowcore pages
1895 prior to calling the KVM_RUN ioctl.
1898 4.68 KVM_SET_ONE_REG
1900 Capability: KVM_CAP_ONE_REG
1903 Parameters: struct kvm_one_reg (in)
1904 Returns: 0 on success, negative value on failure
1906 Â ENOENT: Â Â no such register
1907 Â EINVAL: Â Â invalid register ID, or no such register
1908 Â EPERM: Â Â Â (arm64) register access not allowed before vcpu finalization
1909 (These error codes are indicative only: do not rely on a specific error
1910 code being returned in a specific situation.)
1912 struct kvm_one_reg {
1917 Using this ioctl, a single vcpu register can be set to a specific value
1918 defined by user space with the passed in struct kvm_one_reg, where id
1919 refers to the register identifier as described below and addr is a pointer
1920 to a variable with the respective size. There can be architecture agnostic
1921 and architecture specific registers. Each have their own range of operation
1922 and their own constants and width. To keep track of the implemented
1923 registers, find a list below:
1925 Arch | Register | Width (bits)
1927 PPC | KVM_REG_PPC_HIOR | 64
1928 PPC | KVM_REG_PPC_IAC1 | 64
1929 PPC | KVM_REG_PPC_IAC2 | 64
1930 PPC | KVM_REG_PPC_IAC3 | 64
1931 PPC | KVM_REG_PPC_IAC4 | 64
1932 PPC | KVM_REG_PPC_DAC1 | 64
1933 PPC | KVM_REG_PPC_DAC2 | 64
1934 PPC | KVM_REG_PPC_DABR | 64
1935 PPC | KVM_REG_PPC_DSCR | 64
1936 PPC | KVM_REG_PPC_PURR | 64
1937 PPC | KVM_REG_PPC_SPURR | 64
1938 PPC | KVM_REG_PPC_DAR | 64
1939 PPC | KVM_REG_PPC_DSISR | 32
1940 PPC | KVM_REG_PPC_AMR | 64
1941 PPC | KVM_REG_PPC_UAMOR | 64
1942 PPC | KVM_REG_PPC_MMCR0 | 64
1943 PPC | KVM_REG_PPC_MMCR1 | 64
1944 PPC | KVM_REG_PPC_MMCRA | 64
1945 PPC | KVM_REG_PPC_MMCR2 | 64
1946 PPC | KVM_REG_PPC_MMCRS | 64
1947 PPC | KVM_REG_PPC_SIAR | 64
1948 PPC | KVM_REG_PPC_SDAR | 64
1949 PPC | KVM_REG_PPC_SIER | 64
1950 PPC | KVM_REG_PPC_PMC1 | 32
1951 PPC | KVM_REG_PPC_PMC2 | 32
1952 PPC | KVM_REG_PPC_PMC3 | 32
1953 PPC | KVM_REG_PPC_PMC4 | 32
1954 PPC | KVM_REG_PPC_PMC5 | 32
1955 PPC | KVM_REG_PPC_PMC6 | 32
1956 PPC | KVM_REG_PPC_PMC7 | 32
1957 PPC | KVM_REG_PPC_PMC8 | 32
1958 PPC | KVM_REG_PPC_FPR0 | 64
1960 PPC | KVM_REG_PPC_FPR31 | 64
1961 PPC | KVM_REG_PPC_VR0 | 128
1963 PPC | KVM_REG_PPC_VR31 | 128
1964 PPC | KVM_REG_PPC_VSR0 | 128
1966 PPC | KVM_REG_PPC_VSR31 | 128
1967 PPC | KVM_REG_PPC_FPSCR | 64
1968 PPC | KVM_REG_PPC_VSCR | 32
1969 PPC | KVM_REG_PPC_VPA_ADDR | 64
1970 PPC | KVM_REG_PPC_VPA_SLB | 128
1971 PPC | KVM_REG_PPC_VPA_DTL | 128
1972 PPC | KVM_REG_PPC_EPCR | 32
1973 PPC | KVM_REG_PPC_EPR | 32
1974 PPC | KVM_REG_PPC_TCR | 32
1975 PPC | KVM_REG_PPC_TSR | 32
1976 PPC | KVM_REG_PPC_OR_TSR | 32
1977 PPC | KVM_REG_PPC_CLEAR_TSR | 32
1978 PPC | KVM_REG_PPC_MAS0 | 32
1979 PPC | KVM_REG_PPC_MAS1 | 32
1980 PPC | KVM_REG_PPC_MAS2 | 64
1981 PPC | KVM_REG_PPC_MAS7_3 | 64
1982 PPC | KVM_REG_PPC_MAS4 | 32
1983 PPC | KVM_REG_PPC_MAS6 | 32
1984 PPC | KVM_REG_PPC_MMUCFG | 32
1985 PPC | KVM_REG_PPC_TLB0CFG | 32
1986 PPC | KVM_REG_PPC_TLB1CFG | 32
1987 PPC | KVM_REG_PPC_TLB2CFG | 32
1988 PPC | KVM_REG_PPC_TLB3CFG | 32
1989 PPC | KVM_REG_PPC_TLB0PS | 32
1990 PPC | KVM_REG_PPC_TLB1PS | 32
1991 PPC | KVM_REG_PPC_TLB2PS | 32
1992 PPC | KVM_REG_PPC_TLB3PS | 32
1993 PPC | KVM_REG_PPC_EPTCFG | 32
1994 PPC | KVM_REG_PPC_ICP_STATE | 64
1995 PPC | KVM_REG_PPC_VP_STATE | 128
1996 PPC | KVM_REG_PPC_TB_OFFSET | 64
1997 PPC | KVM_REG_PPC_SPMC1 | 32
1998 PPC | KVM_REG_PPC_SPMC2 | 32
1999 PPC | KVM_REG_PPC_IAMR | 64
2000 PPC | KVM_REG_PPC_TFHAR | 64
2001 PPC | KVM_REG_PPC_TFIAR | 64
2002 PPC | KVM_REG_PPC_TEXASR | 64
2003 PPC | KVM_REG_PPC_FSCR | 64
2004 PPC | KVM_REG_PPC_PSPB | 32
2005 PPC | KVM_REG_PPC_EBBHR | 64
2006 PPC | KVM_REG_PPC_EBBRR | 64
2007 PPC | KVM_REG_PPC_BESCR | 64
2008 PPC | KVM_REG_PPC_TAR | 64
2009 PPC | KVM_REG_PPC_DPDES | 64
2010 PPC | KVM_REG_PPC_DAWR | 64
2011 PPC | KVM_REG_PPC_DAWRX | 64
2012 PPC | KVM_REG_PPC_CIABR | 64
2013 PPC | KVM_REG_PPC_IC | 64
2014 PPC | KVM_REG_PPC_VTB | 64
2015 PPC | KVM_REG_PPC_CSIGR | 64
2016 PPC | KVM_REG_PPC_TACR | 64
2017 PPC | KVM_REG_PPC_TCSCR | 64
2018 PPC | KVM_REG_PPC_PID | 64
2019 PPC | KVM_REG_PPC_ACOP | 64
2020 PPC | KVM_REG_PPC_VRSAVE | 32
2021 PPC | KVM_REG_PPC_LPCR | 32
2022 PPC | KVM_REG_PPC_LPCR_64 | 64
2023 PPC | KVM_REG_PPC_PPR | 64
2024 PPC | KVM_REG_PPC_ARCH_COMPAT | 32
2025 PPC | KVM_REG_PPC_DABRX | 32
2026 PPC | KVM_REG_PPC_WORT | 64
2027 PPC | KVM_REG_PPC_SPRG9 | 64
2028 PPC | KVM_REG_PPC_DBSR | 32
2029 PPC | KVM_REG_PPC_TIDR | 64
2030 PPC | KVM_REG_PPC_PSSCR | 64
2031 PPC | KVM_REG_PPC_DEC_EXPIRY | 64
2032 PPC | KVM_REG_PPC_PTCR | 64
2033 PPC | KVM_REG_PPC_TM_GPR0 | 64
2035 PPC | KVM_REG_PPC_TM_GPR31 | 64
2036 PPC | KVM_REG_PPC_TM_VSR0 | 128
2038 PPC | KVM_REG_PPC_TM_VSR63 | 128
2039 PPC | KVM_REG_PPC_TM_CR | 64
2040 PPC | KVM_REG_PPC_TM_LR | 64
2041 PPC | KVM_REG_PPC_TM_CTR | 64
2042 PPC | KVM_REG_PPC_TM_FPSCR | 64
2043 PPC | KVM_REG_PPC_TM_AMR | 64
2044 PPC | KVM_REG_PPC_TM_PPR | 64
2045 PPC | KVM_REG_PPC_TM_VRSAVE | 64
2046 PPC | KVM_REG_PPC_TM_VSCR | 32
2047 PPC | KVM_REG_PPC_TM_DSCR | 64
2048 PPC | KVM_REG_PPC_TM_TAR | 64
2049 PPC | KVM_REG_PPC_TM_XER | 64
2051 MIPS | KVM_REG_MIPS_R0 | 64
2053 MIPS | KVM_REG_MIPS_R31 | 64
2054 MIPS | KVM_REG_MIPS_HI | 64
2055 MIPS | KVM_REG_MIPS_LO | 64
2056 MIPS | KVM_REG_MIPS_PC | 64
2057 MIPS | KVM_REG_MIPS_CP0_INDEX | 32
2058 MIPS | KVM_REG_MIPS_CP0_ENTRYLO0 | 64
2059 MIPS | KVM_REG_MIPS_CP0_ENTRYLO1 | 64
2060 MIPS | KVM_REG_MIPS_CP0_CONTEXT | 64
2061 MIPS | KVM_REG_MIPS_CP0_CONTEXTCONFIG| 32
2062 MIPS | KVM_REG_MIPS_CP0_USERLOCAL | 64
2063 MIPS | KVM_REG_MIPS_CP0_XCONTEXTCONFIG| 64
2064 MIPS | KVM_REG_MIPS_CP0_PAGEMASK | 32
2065 MIPS | KVM_REG_MIPS_CP0_PAGEGRAIN | 32
2066 MIPS | KVM_REG_MIPS_CP0_SEGCTL0 | 64
2067 MIPS | KVM_REG_MIPS_CP0_SEGCTL1 | 64
2068 MIPS | KVM_REG_MIPS_CP0_SEGCTL2 | 64
2069 MIPS | KVM_REG_MIPS_CP0_PWBASE | 64
2070 MIPS | KVM_REG_MIPS_CP0_PWFIELD | 64
2071 MIPS | KVM_REG_MIPS_CP0_PWSIZE | 64
2072 MIPS | KVM_REG_MIPS_CP0_WIRED | 32
2073 MIPS | KVM_REG_MIPS_CP0_PWCTL | 32
2074 MIPS | KVM_REG_MIPS_CP0_HWRENA | 32
2075 MIPS | KVM_REG_MIPS_CP0_BADVADDR | 64
2076 MIPS | KVM_REG_MIPS_CP0_BADINSTR | 32
2077 MIPS | KVM_REG_MIPS_CP0_BADINSTRP | 32
2078 MIPS | KVM_REG_MIPS_CP0_COUNT | 32
2079 MIPS | KVM_REG_MIPS_CP0_ENTRYHI | 64
2080 MIPS | KVM_REG_MIPS_CP0_COMPARE | 32
2081 MIPS | KVM_REG_MIPS_CP0_STATUS | 32
2082 MIPS | KVM_REG_MIPS_CP0_INTCTL | 32
2083 MIPS | KVM_REG_MIPS_CP0_CAUSE | 32
2084 MIPS | KVM_REG_MIPS_CP0_EPC | 64
2085 MIPS | KVM_REG_MIPS_CP0_PRID | 32
2086 MIPS | KVM_REG_MIPS_CP0_EBASE | 64
2087 MIPS | KVM_REG_MIPS_CP0_CONFIG | 32
2088 MIPS | KVM_REG_MIPS_CP0_CONFIG1 | 32
2089 MIPS | KVM_REG_MIPS_CP0_CONFIG2 | 32
2090 MIPS | KVM_REG_MIPS_CP0_CONFIG3 | 32
2091 MIPS | KVM_REG_MIPS_CP0_CONFIG4 | 32
2092 MIPS | KVM_REG_MIPS_CP0_CONFIG5 | 32
2093 MIPS | KVM_REG_MIPS_CP0_CONFIG7 | 32
2094 MIPS | KVM_REG_MIPS_CP0_XCONTEXT | 64
2095 MIPS | KVM_REG_MIPS_CP0_ERROREPC | 64
2096 MIPS | KVM_REG_MIPS_CP0_KSCRATCH1 | 64
2097 MIPS | KVM_REG_MIPS_CP0_KSCRATCH2 | 64
2098 MIPS | KVM_REG_MIPS_CP0_KSCRATCH3 | 64
2099 MIPS | KVM_REG_MIPS_CP0_KSCRATCH4 | 64
2100 MIPS | KVM_REG_MIPS_CP0_KSCRATCH5 | 64
2101 MIPS | KVM_REG_MIPS_CP0_KSCRATCH6 | 64
2102 MIPS | KVM_REG_MIPS_CP0_MAAR(0..63) | 64
2103 MIPS | KVM_REG_MIPS_COUNT_CTL | 64
2104 MIPS | KVM_REG_MIPS_COUNT_RESUME | 64
2105 MIPS | KVM_REG_MIPS_COUNT_HZ | 64
2106 MIPS | KVM_REG_MIPS_FPR_32(0..31) | 32
2107 MIPS | KVM_REG_MIPS_FPR_64(0..31) | 64
2108 MIPS | KVM_REG_MIPS_VEC_128(0..31) | 128
2109 MIPS | KVM_REG_MIPS_FCR_IR | 32
2110 MIPS | KVM_REG_MIPS_FCR_CSR | 32
2111 MIPS | KVM_REG_MIPS_MSA_IR | 32
2112 MIPS | KVM_REG_MIPS_MSA_CSR | 32
2114 ARM registers are mapped using the lower 32 bits. The upper 16 of that
2115 is the register group type, or coprocessor number:
2117 ARM core registers have the following id bit patterns:
2118 0x4020 0000 0010 <index into the kvm_regs struct:16>
2120 ARM 32-bit CP15 registers have the following id bit patterns:
2121 0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3>
2123 ARM 64-bit CP15 registers have the following id bit patterns:
2124 0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3>
2126 ARM CCSIDR registers are demultiplexed by CSSELR value:
2127 0x4020 0000 0011 00 <csselr:8>
2129 ARM 32-bit VFP control registers have the following id bit patterns:
2130 0x4020 0000 0012 1 <regno:12>
2132 ARM 64-bit FP registers have the following id bit patterns:
2133 0x4030 0000 0012 0 <regno:12>
2135 ARM firmware pseudo-registers have the following bit pattern:
2136 0x4030 0000 0014 <regno:16>
2139 arm64 registers are mapped using the lower 32 bits. The upper 16 of
2140 that is the register group type, or coprocessor number:
2142 arm64 core/FP-SIMD registers have the following id bit patterns. Note
2143 that the size of the access is variable, as the kvm_regs structure
2144 contains elements ranging from 32 to 128 bits. The index is a 32bit
2145 value in the kvm_regs structure seen as a 32bit array.
2146 0x60x0 0000 0010 <index into the kvm_regs struct:16>
2149 Encoding Register Bits kvm_regs member
2150 ----------------------------------------------------------------
2151 0x6030 0000 0010 0000 X0 64 regs.regs[0]
2152 0x6030 0000 0010 0002 X1 64 regs.regs[1]
2154 0x6030 0000 0010 003c X30 64 regs.regs[30]
2155 0x6030 0000 0010 003e SP 64 regs.sp
2156 0x6030 0000 0010 0040 PC 64 regs.pc
2157 0x6030 0000 0010 0042 PSTATE 64 regs.pstate
2158 0x6030 0000 0010 0044 SP_EL1 64 sp_el1
2159 0x6030 0000 0010 0046 ELR_EL1 64 elr_el1
2160 0x6030 0000 0010 0048 SPSR_EL1 64 spsr[KVM_SPSR_EL1] (alias SPSR_SVC)
2161 0x6030 0000 0010 004a SPSR_ABT 64 spsr[KVM_SPSR_ABT]
2162 0x6030 0000 0010 004c SPSR_UND 64 spsr[KVM_SPSR_UND]
2163 0x6030 0000 0010 004e SPSR_IRQ 64 spsr[KVM_SPSR_IRQ]
2164 0x6060 0000 0010 0050 SPSR_FIQ 64 spsr[KVM_SPSR_FIQ]
2165 0x6040 0000 0010 0054 V0 128 fp_regs.vregs[0] (*)
2166 0x6040 0000 0010 0058 V1 128 fp_regs.vregs[1] (*)
2168 0x6040 0000 0010 00d0 V31 128 fp_regs.vregs[31] (*)
2169 0x6020 0000 0010 00d4 FPSR 32 fp_regs.fpsr
2170 0x6020 0000 0010 00d5 FPCR 32 fp_regs.fpcr
2172 (*) These encodings are not accepted for SVE-enabled vcpus. See
2175 The equivalent register content can be accessed via bits [127:0] of
2176 the corresponding SVE Zn registers instead for vcpus that have SVE
2177 enabled (see below).
2179 arm64 CCSIDR registers are demultiplexed by CSSELR value:
2180 0x6020 0000 0011 00 <csselr:8>
2182 arm64 system registers have the following id bit patterns:
2183 0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3>
2185 arm64 firmware pseudo-registers have the following bit pattern:
2186 0x6030 0000 0014 <regno:16>
2188 arm64 SVE registers have the following bit patterns:
2189 0x6080 0000 0015 00 <n:5> <slice:5> Zn bits[2048*slice + 2047 : 2048*slice]
2190 0x6050 0000 0015 04 <n:4> <slice:5> Pn bits[256*slice + 255 : 256*slice]
2191 0x6050 0000 0015 060 <slice:5> FFR bits[256*slice + 255 : 256*slice]
2192 0x6060 0000 0015 ffff KVM_REG_ARM64_SVE_VLS pseudo-register
2194 Access to register IDs where 2048 * slice >= 128 * max_vq will fail with
2195 ENOENT. max_vq is the vcpu's maximum supported vector length in 128-bit
2196 quadwords: see (**) below.
2198 These registers are only accessible on vcpus for which SVE is enabled.
2199 See KVM_ARM_VCPU_INIT for details.
2201 In addition, except for KVM_REG_ARM64_SVE_VLS, these registers are not
2202 accessible until the vcpu's SVE configuration has been finalized
2203 using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). See KVM_ARM_VCPU_INIT
2204 and KVM_ARM_VCPU_FINALIZE for more information about this procedure.
2206 KVM_REG_ARM64_SVE_VLS is a pseudo-register that allows the set of vector
2207 lengths supported by the vcpu to be discovered and configured by
2208 userspace. When transferred to or from user memory via KVM_GET_ONE_REG
2209 or KVM_SET_ONE_REG, the value of this register is of type
2210 __u64[KVM_ARM64_SVE_VLS_WORDS], and encodes the set of vector lengths as
2213 __u64 vector_lengths[KVM_ARM64_SVE_VLS_WORDS];
2215 if (vq >= SVE_VQ_MIN && vq <= SVE_VQ_MAX &&
2216 ((vector_lengths[(vq - KVM_ARM64_SVE_VQ_MIN) / 64] >>
2217 ((vq - KVM_ARM64_SVE_VQ_MIN) % 64)) & 1))
2218 /* Vector length vq * 16 bytes supported */
2220 /* Vector length vq * 16 bytes not supported */
2222 (**) The maximum value vq for which the above condition is true is
2223 max_vq. This is the maximum vector length available to the guest on
2224 this vcpu, and determines which register slices are visible through
2225 this ioctl interface.
2227 (See Documentation/arm64/sve.rst for an explanation of the "vq"
2230 KVM_REG_ARM64_SVE_VLS is only accessible after KVM_ARM_VCPU_INIT.
2231 KVM_ARM_VCPU_INIT initialises it to the best set of vector lengths that
2234 Userspace may subsequently modify it if desired until the vcpu's SVE
2235 configuration is finalized using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE).
2237 Apart from simply removing all vector lengths from the host set that
2238 exceed some value, support for arbitrarily chosen sets of vector lengths
2239 is hardware-dependent and may not be available. Attempting to configure
2240 an invalid set of vector lengths via KVM_SET_ONE_REG will fail with
2243 After the vcpu's SVE configuration is finalized, further attempts to
2244 write this register will fail with EPERM.
2247 MIPS registers are mapped using the lower 32 bits. The upper 16 of that is
2248 the register group type:
2250 MIPS core registers (see above) have the following id bit patterns:
2251 0x7030 0000 0000 <reg:16>
2253 MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit
2254 patterns depending on whether they're 32-bit or 64-bit registers:
2255 0x7020 0000 0001 00 <reg:5> <sel:3> (32-bit)
2256 0x7030 0000 0001 00 <reg:5> <sel:3> (64-bit)
2258 Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_MIPS_CP0_ENTRYLO1 are the MIPS64
2259 versions of the EntryLo registers regardless of the word size of the host
2260 hardware, host kernel, guest, and whether XPA is present in the guest, i.e.
2261 with the RI and XI bits (if they exist) in bits 63 and 62 respectively, and
2262 the PFNX field starting at bit 30.
2264 MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) above) have the following id bit
2266 0x7030 0000 0001 01 <reg:8>
2268 MIPS KVM control registers (see above) have the following id bit patterns:
2269 0x7030 0000 0002 <reg:16>
2271 MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following
2272 id bit patterns depending on the size of the register being accessed. They are
2273 always accessed according to the current guest FPU mode (Status.FR and
2274 Config5.FRE), i.e. as the guest would see them, and they become unpredictable
2275 if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector
2276 registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they
2277 overlap the FPU registers:
2278 0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers)
2279 0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers)
2280 0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers)
2282 MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the
2283 following id bit patterns:
2284 0x7020 0000 0003 01 <0:3> <reg:5>
2286 MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the
2287 following id bit patterns:
2288 0x7020 0000 0003 02 <0:3> <reg:5>
2291 4.69 KVM_GET_ONE_REG
2293 Capability: KVM_CAP_ONE_REG
2296 Parameters: struct kvm_one_reg (in and out)
2297 Returns: 0 on success, negative value on failure
2299 Â ENOENT: Â Â no such register
2300 Â EINVAL: Â Â invalid register ID, or no such register
2301 Â EPERM: Â Â Â (arm64) register access not allowed before vcpu finalization
2302 (These error codes are indicative only: do not rely on a specific error
2303 code being returned in a specific situation.)
2305 This ioctl allows to receive the value of a single register implemented
2306 in a vcpu. The register to read is indicated by the "id" field of the
2307 kvm_one_reg struct passed in. On success, the register value can be found
2308 at the memory location pointed to by "addr".
2310 The list of registers accessible using this interface is identical to the
2314 4.70 KVM_KVMCLOCK_CTRL
2316 Capability: KVM_CAP_KVMCLOCK_CTRL
2317 Architectures: Any that implement pvclocks (currently x86 only)
2320 Returns: 0 on success, -1 on error
2322 This signals to the host kernel that the specified guest is being paused by
2323 userspace. The host will set a flag in the pvclock structure that is checked
2324 from the soft lockup watchdog. The flag is part of the pvclock structure that
2325 is shared between guest and host, specifically the second bit of the flags
2326 field of the pvclock_vcpu_time_info structure. It will be set exclusively by
2327 the host and read/cleared exclusively by the guest. The guest operation of
2328 checking and clearing the flag must an atomic operation so
2329 load-link/store-conditional, or equivalent must be used. There are two cases
2330 where the guest will clear the flag: when the soft lockup watchdog timer resets
2331 itself or when a soft lockup is detected. This ioctl can be called any time
2332 after pausing the vcpu, but before it is resumed.
2337 Capability: KVM_CAP_SIGNAL_MSI
2338 Architectures: x86 arm arm64
2340 Parameters: struct kvm_msi (in)
2341 Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error
2343 Directly inject a MSI message. Only valid with in-kernel irqchip that handles
2355 flags: KVM_MSI_VALID_DEVID: devid contains a valid value. The per-VM
2356 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
2357 the device ID. If this capability is not available, userspace
2358 should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
2360 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
2361 for the device that wrote the MSI message. For PCI, this is usually a
2362 BFD identifier in the lower 16 bits.
2364 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
2365 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
2366 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
2367 address_hi must be zero.
2370 4.71 KVM_CREATE_PIT2
2372 Capability: KVM_CAP_PIT2
2375 Parameters: struct kvm_pit_config (in)
2376 Returns: 0 on success, -1 on error
2378 Creates an in-kernel device model for the i8254 PIT. This call is only valid
2379 after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following
2380 parameters have to be passed:
2382 struct kvm_pit_config {
2389 #define KVM_PIT_SPEAKER_DUMMY 1 /* emulate speaker port stub */
2391 PIT timer interrupts may use a per-VM kernel thread for injection. If it
2392 exists, this thread will have a name of the following pattern:
2394 kvm-pit/<owner-process-pid>
2396 When running a guest with elevated priorities, the scheduling parameters of
2397 this thread may have to be adjusted accordingly.
2399 This IOCTL replaces the obsolete KVM_CREATE_PIT.
2404 Capability: KVM_CAP_PIT_STATE2
2407 Parameters: struct kvm_pit_state2 (out)
2408 Returns: 0 on success, -1 on error
2410 Retrieves the state of the in-kernel PIT model. Only valid after
2411 KVM_CREATE_PIT2. The state is returned in the following structure:
2413 struct kvm_pit_state2 {
2414 struct kvm_pit_channel_state channels[3];
2421 /* disable PIT in HPET legacy mode */
2422 #define KVM_PIT_FLAGS_HPET_LEGACY 0x00000001
2424 This IOCTL replaces the obsolete KVM_GET_PIT.
2429 Capability: KVM_CAP_PIT_STATE2
2432 Parameters: struct kvm_pit_state2 (in)
2433 Returns: 0 on success, -1 on error
2435 Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2.
2436 See KVM_GET_PIT2 for details on struct kvm_pit_state2.
2438 This IOCTL replaces the obsolete KVM_SET_PIT.
2441 4.74 KVM_PPC_GET_SMMU_INFO
2443 Capability: KVM_CAP_PPC_GET_SMMU_INFO
2444 Architectures: powerpc
2447 Returns: 0 on success, -1 on error
2449 This populates and returns a structure describing the features of
2450 the "Server" class MMU emulation supported by KVM.
2451 This can in turn be used by userspace to generate the appropriate
2452 device-tree properties for the guest operating system.
2454 The structure contains some global information, followed by an
2455 array of supported segment page sizes:
2457 struct kvm_ppc_smmu_info {
2461 struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ];
2464 The supported flags are:
2466 - KVM_PPC_PAGE_SIZES_REAL:
2467 When that flag is set, guest page sizes must "fit" the backing
2468 store page sizes. When not set, any page size in the list can
2469 be used regardless of how they are backed by userspace.
2471 - KVM_PPC_1T_SEGMENTS
2472 The emulated MMU supports 1T segments in addition to the
2476 This flag indicates that HPT guests are not supported by KVM,
2477 thus all guests must use radix MMU mode.
2479 The "slb_size" field indicates how many SLB entries are supported
2481 The "sps" array contains 8 entries indicating the supported base
2482 page sizes for a segment in increasing order. Each entry is defined
2485 struct kvm_ppc_one_seg_page_size {
2486 __u32 page_shift; /* Base page shift of segment (or 0) */
2487 __u32 slb_enc; /* SLB encoding for BookS */
2488 struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ];
2491 An entry with a "page_shift" of 0 is unused. Because the array is
2492 organized in increasing order, a lookup can stop when encoutering
2495 The "slb_enc" field provides the encoding to use in the SLB for the
2496 page size. The bits are in positions such as the value can directly
2497 be OR'ed into the "vsid" argument of the slbmte instruction.
2499 The "enc" array is a list which for each of those segment base page
2500 size provides the list of supported actual page sizes (which can be
2501 only larger or equal to the base page size), along with the
2502 corresponding encoding in the hash PTE. Similarly, the array is
2503 8 entries sorted by increasing sizes and an entry with a "0" shift
2504 is an empty entry and a terminator:
2506 struct kvm_ppc_one_page_size {
2507 __u32 page_shift; /* Page shift (or 0) */
2508 __u32 pte_enc; /* Encoding in the HPTE (>>12) */
2511 The "pte_enc" field provides a value that can OR'ed into the hash
2512 PTE's RPN field (ie, it needs to be shifted left by 12 to OR it
2513 into the hash PTE second double word).
2517 Capability: KVM_CAP_IRQFD
2518 Architectures: x86 s390 arm arm64
2520 Parameters: struct kvm_irqfd (in)
2521 Returns: 0 on success, -1 on error
2523 Allows setting an eventfd to directly trigger a guest interrupt.
2524 kvm_irqfd.fd specifies the file descriptor to use as the eventfd and
2525 kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When
2526 an event is triggered on the eventfd, an interrupt is injected into
2527 the guest using the specified gsi pin. The irqfd is removed using
2528 the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd
2531 With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify
2532 mechanism allowing emulation of level-triggered, irqfd-based
2533 interrupts. When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an
2534 additional eventfd in the kvm_irqfd.resamplefd field. When operating
2535 in resample mode, posting of an interrupt through kvm_irq.fd asserts
2536 the specified gsi in the irqchip. When the irqchip is resampled, such
2537 as from an EOI, the gsi is de-asserted and the user is notified via
2538 kvm_irqfd.resamplefd. It is the user's responsibility to re-queue
2539 the interrupt if the device making use of it still requires service.
2540 Note that closing the resamplefd is not sufficient to disable the
2541 irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment
2542 and need not be specified with KVM_IRQFD_FLAG_DEASSIGN.
2544 On arm/arm64, gsi routing being supported, the following can happen:
2545 - in case no routing entry is associated to this gsi, injection fails
2546 - in case the gsi is associated to an irqchip routing entry,
2547 irqchip.pin + 32 corresponds to the injected SPI ID.
2548 - in case the gsi is associated to an MSI routing entry, the MSI
2549 message and device ID are translated into an LPI (support restricted
2550 to GICv3 ITS in-kernel emulation).
2552 4.76 KVM_PPC_ALLOCATE_HTAB
2554 Capability: KVM_CAP_PPC_ALLOC_HTAB
2555 Architectures: powerpc
2557 Parameters: Pointer to u32 containing hash table order (in/out)
2558 Returns: 0 on success, -1 on error
2560 This requests the host kernel to allocate an MMU hash table for a
2561 guest using the PAPR paravirtualization interface. This only does
2562 anything if the kernel is configured to use the Book 3S HV style of
2563 virtualization. Otherwise the capability doesn't exist and the ioctl
2564 returns an ENOTTY error. The rest of this description assumes Book 3S
2567 There must be no vcpus running when this ioctl is called; if there
2568 are, it will do nothing and return an EBUSY error.
2570 The parameter is a pointer to a 32-bit unsigned integer variable
2571 containing the order (log base 2) of the desired size of the hash
2572 table, which must be between 18 and 46. On successful return from the
2573 ioctl, the value will not be changed by the kernel.
2575 If no hash table has been allocated when any vcpu is asked to run
2576 (with the KVM_RUN ioctl), the host kernel will allocate a
2577 default-sized hash table (16 MB).
2579 If this ioctl is called when a hash table has already been allocated,
2580 with a different order from the existing hash table, the existing hash
2581 table will be freed and a new one allocated. If this is ioctl is
2582 called when a hash table has already been allocated of the same order
2583 as specified, the kernel will clear out the existing hash table (zero
2584 all HPTEs). In either case, if the guest is using the virtualized
2585 real-mode area (VRMA) facility, the kernel will re-create the VMRA
2586 HPTEs on the next KVM_RUN of any vcpu.
2588 4.77 KVM_S390_INTERRUPT
2592 Type: vm ioctl, vcpu ioctl
2593 Parameters: struct kvm_s390_interrupt (in)
2594 Returns: 0 on success, -1 on error
2596 Allows to inject an interrupt to the guest. Interrupts can be floating
2597 (vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type.
2599 Interrupt parameters are passed via kvm_s390_interrupt:
2601 struct kvm_s390_interrupt {
2607 type can be one of the following:
2609 KVM_S390_SIGP_STOP (vcpu) - sigp stop; optional flags in parm
2610 KVM_S390_PROGRAM_INT (vcpu) - program check; code in parm
2611 KVM_S390_SIGP_SET_PREFIX (vcpu) - sigp set prefix; prefix address in parm
2612 KVM_S390_RESTART (vcpu) - restart
2613 KVM_S390_INT_CLOCK_COMP (vcpu) - clock comparator interrupt
2614 KVM_S390_INT_CPU_TIMER (vcpu) - CPU timer interrupt
2615 KVM_S390_INT_VIRTIO (vm) - virtio external interrupt; external interrupt
2616 parameters in parm and parm64
2617 KVM_S390_INT_SERVICE (vm) - sclp external interrupt; sclp parameter in parm
2618 KVM_S390_INT_EMERGENCY (vcpu) - sigp emergency; source cpu in parm
2619 KVM_S390_INT_EXTERNAL_CALL (vcpu) - sigp external call; source cpu in parm
2620 KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm) - compound value to indicate an
2621 I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel);
2622 I/O interruption parameters in parm (subchannel) and parm64 (intparm,
2623 interruption subclass)
2624 KVM_S390_MCHK (vm, vcpu) - machine check interrupt; cr 14 bits in parm,
2625 machine check interrupt code in parm64 (note that
2626 machine checks needing further payload are not
2627 supported by this ioctl)
2629 This is an asynchronous vcpu ioctl and can be invoked from any thread.
2631 4.78 KVM_PPC_GET_HTAB_FD
2633 Capability: KVM_CAP_PPC_HTAB_FD
2634 Architectures: powerpc
2636 Parameters: Pointer to struct kvm_get_htab_fd (in)
2637 Returns: file descriptor number (>= 0) on success, -1 on error
2639 This returns a file descriptor that can be used either to read out the
2640 entries in the guest's hashed page table (HPT), or to write entries to
2641 initialize the HPT. The returned fd can only be written to if the
2642 KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and
2643 can only be read if that bit is clear. The argument struct looks like
2646 /* For KVM_PPC_GET_HTAB_FD */
2647 struct kvm_get_htab_fd {
2653 /* Values for kvm_get_htab_fd.flags */
2654 #define KVM_GET_HTAB_BOLTED_ONLY ((__u64)0x1)
2655 #define KVM_GET_HTAB_WRITE ((__u64)0x2)
2657 The `start_index' field gives the index in the HPT of the entry at
2658 which to start reading. It is ignored when writing.
2660 Reads on the fd will initially supply information about all
2661 "interesting" HPT entries. Interesting entries are those with the
2662 bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise
2663 all entries. When the end of the HPT is reached, the read() will
2664 return. If read() is called again on the fd, it will start again from
2665 the beginning of the HPT, but will only return HPT entries that have
2666 changed since they were last read.
2668 Data read or written is structured as a header (8 bytes) followed by a
2669 series of valid HPT entries (16 bytes) each. The header indicates how
2670 many valid HPT entries there are and how many invalid entries follow
2671 the valid entries. The invalid entries are not represented explicitly
2672 in the stream. The header format is:
2674 struct kvm_get_htab_header {
2680 Writes to the fd create HPT entries starting at the index given in the
2681 header; first `n_valid' valid entries with contents from the data
2682 written, then `n_invalid' invalid entries, invalidating any previously
2683 valid entries found.
2685 4.79 KVM_CREATE_DEVICE
2687 Capability: KVM_CAP_DEVICE_CTRL
2689 Parameters: struct kvm_create_device (in/out)
2690 Returns: 0 on success, -1 on error
2692 ENODEV: The device type is unknown or unsupported
2693 EEXIST: Device already created, and this type of device may not
2694 be instantiated multiple times
2696 Other error conditions may be defined by individual device types or
2697 have their standard meanings.
2699 Creates an emulated device in the kernel. The file descriptor returned
2700 in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR.
2702 If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the
2703 device type is supported (not necessarily whether it can be created
2706 Individual devices should not define flags. Attributes should be used
2707 for specifying any behavior that is not implied by the device type
2710 struct kvm_create_device {
2711 __u32 type; /* in: KVM_DEV_TYPE_xxx */
2712 __u32 fd; /* out: device handle */
2713 __u32 flags; /* in: KVM_CREATE_DEVICE_xxx */
2716 4.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR
2718 Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
2719 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
2720 Type: device ioctl, vm ioctl, vcpu ioctl
2721 Parameters: struct kvm_device_attr
2722 Returns: 0 on success, -1 on error
2724 ENXIO: The group or attribute is unknown/unsupported for this device
2725 or hardware support is missing.
2726 EPERM: The attribute cannot (currently) be accessed this way
2727 (e.g. read-only attribute, or attribute that only makes
2728 sense when the device is in a different state)
2730 Other error conditions may be defined by individual device types.
2732 Gets/sets a specified piece of device configuration and/or state. The
2733 semantics are device-specific. See individual device documentation in
2734 the "devices" directory. As with ONE_REG, the size of the data
2735 transferred is defined by the particular attribute.
2737 struct kvm_device_attr {
2738 __u32 flags; /* no flags currently defined */
2739 __u32 group; /* device-defined */
2740 __u64 attr; /* group-defined */
2741 __u64 addr; /* userspace address of attr data */
2744 4.81 KVM_HAS_DEVICE_ATTR
2746 Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
2747 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
2748 Type: device ioctl, vm ioctl, vcpu ioctl
2749 Parameters: struct kvm_device_attr
2750 Returns: 0 on success, -1 on error
2752 ENXIO: The group or attribute is unknown/unsupported for this device
2753 or hardware support is missing.
2755 Tests whether a device supports a particular attribute. A successful
2756 return indicates the attribute is implemented. It does not necessarily
2757 indicate that the attribute can be read or written in the device's
2758 current state. "addr" is ignored.
2760 4.82 KVM_ARM_VCPU_INIT
2763 Architectures: arm, arm64
2765 Parameters: struct kvm_vcpu_init (in)
2766 Returns: 0 on success; -1 on error
2768 Â EINVAL: Â Â Â the target is unknown, or the combination of features is invalid.
2769 Â ENOENT: Â Â Â a features bit specified is unknown.
2771 This tells KVM what type of CPU to present to the guest, and what
2772 optional features it should have. Â This will cause a reset of the cpu
2773 registers to their initial values. Â If this is not called, KVM_RUN will
2774 return ENOEXEC for that vcpu.
2776 Note that because some registers reflect machine topology, all vcpus
2777 should be created before this ioctl is invoked.
2779 Userspace can call this function multiple times for a given vcpu, including
2780 after the vcpu has been run. This will reset the vcpu to its initial
2781 state. All calls to this function after the initial call must use the same
2782 target and same set of feature flags, otherwise EINVAL will be returned.
2785 - KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state.
2786 Depends on KVM_CAP_ARM_PSCI. If not set, the CPU will be powered on
2787 and execute guest code when KVM_RUN is called.
2788 - KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode.
2789 Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only).
2790 - KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 (or a future revision
2791 backward compatible with v0.2) for the CPU.
2792 Depends on KVM_CAP_ARM_PSCI_0_2.
2793 - KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU.
2794 Depends on KVM_CAP_ARM_PMU_V3.
2796 - KVM_ARM_VCPU_PTRAUTH_ADDRESS: Enables Address Pointer authentication
2798 Depends on KVM_CAP_ARM_PTRAUTH_ADDRESS.
2799 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are
2800 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and
2801 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be
2804 - KVM_ARM_VCPU_PTRAUTH_GENERIC: Enables Generic Pointer authentication
2806 Depends on KVM_CAP_ARM_PTRAUTH_GENERIC.
2807 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are
2808 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and
2809 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be
2812 - KVM_ARM_VCPU_SVE: Enables SVE for the CPU (arm64 only).
2813 Depends on KVM_CAP_ARM_SVE.
2814 Requires KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
2816 * After KVM_ARM_VCPU_INIT:
2818 - KVM_REG_ARM64_SVE_VLS may be read using KVM_GET_ONE_REG: the
2819 initial value of this pseudo-register indicates the best set of
2820 vector lengths possible for a vcpu on this host.
2822 * Before KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
2824 - KVM_RUN and KVM_GET_REG_LIST are not available;
2826 - KVM_GET_ONE_REG and KVM_SET_ONE_REG cannot be used to access
2827 the scalable archietctural SVE registers
2828 KVM_REG_ARM64_SVE_ZREG(), KVM_REG_ARM64_SVE_PREG() or
2829 KVM_REG_ARM64_SVE_FFR;
2831 - KVM_REG_ARM64_SVE_VLS may optionally be written using
2832 KVM_SET_ONE_REG, to modify the set of vector lengths available
2835 * After KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
2837 - the KVM_REG_ARM64_SVE_VLS pseudo-register is immutable, and can
2838 no longer be written using KVM_SET_ONE_REG.
2840 4.83 KVM_ARM_PREFERRED_TARGET
2843 Architectures: arm, arm64
2845 Parameters: struct struct kvm_vcpu_init (out)
2846 Returns: 0 on success; -1 on error
2848 ENODEV: no preferred target available for the host
2850 This queries KVM for preferred CPU target type which can be emulated
2851 by KVM on underlying host.
2853 The ioctl returns struct kvm_vcpu_init instance containing information
2854 about preferred CPU target type and recommended features for it. The
2855 kvm_vcpu_init->features bitmap returned will have feature bits set if
2856 the preferred target recommends setting these features, but this is
2859 The information returned by this ioctl can be used to prepare an instance
2860 of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in
2861 in VCPU matching underlying host.
2864 4.84 KVM_GET_REG_LIST
2867 Architectures: arm, arm64, mips
2869 Parameters: struct kvm_reg_list (in/out)
2870 Returns: 0 on success; -1 on error
2872 Â E2BIG: Â Â Â Â the reg index list is too big to fit in the array specified by
2873 Â Â Â Â Â Â Â Â Â Â Â Â the user (the number required will be written into n).
2875 struct kvm_reg_list {
2876 __u64 n; /* number of registers in reg[] */
2880 This ioctl returns the guest registers that are supported for the
2881 KVM_GET_ONE_REG/KVM_SET_ONE_REG calls.
2884 4.85 KVM_ARM_SET_DEVICE_ADDR (deprecated)
2886 Capability: KVM_CAP_ARM_SET_DEVICE_ADDR
2887 Architectures: arm, arm64
2889 Parameters: struct kvm_arm_device_address (in)
2890 Returns: 0 on success, -1 on error
2892 ENODEV: The device id is unknown
2893 ENXIO: Device not supported on current system
2894 EEXIST: Address already set
2895 E2BIG: Address outside guest physical address space
2896 EBUSY: Address overlaps with other device range
2898 struct kvm_arm_device_addr {
2903 Specify a device address in the guest's physical address space where guests
2904 can access emulated or directly exposed devices, which the host kernel needs
2905 to know about. The id field is an architecture specific identifier for a
2908 ARM/arm64 divides the id field into two parts, a device id and an
2909 address type id specific to the individual device.
2911 Â bits: | 63 ... 32 | 31 ... 16 | 15 ... 0 |
2912 field: | 0x00000000 | device id | addr type id |
2914 ARM/arm64 currently only require this when using the in-kernel GIC
2915 support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2
2916 as the device id. When setting the base address for the guest's
2917 mapping of the VGIC virtual CPU and distributor interface, the ioctl
2918 must be called after calling KVM_CREATE_IRQCHIP, but before calling
2919 KVM_RUN on any of the VCPUs. Calling this ioctl twice for any of the
2920 base addresses will return -EEXIST.
2922 Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API
2923 should be used instead.
2926 4.86 KVM_PPC_RTAS_DEFINE_TOKEN
2928 Capability: KVM_CAP_PPC_RTAS
2931 Parameters: struct kvm_rtas_token_args
2932 Returns: 0 on success, -1 on error
2934 Defines a token value for a RTAS (Run Time Abstraction Services)
2935 service in order to allow it to be handled in the kernel. The
2936 argument struct gives the name of the service, which must be the name
2937 of a service that has a kernel-side implementation. If the token
2938 value is non-zero, it will be associated with that service, and
2939 subsequent RTAS calls by the guest specifying that token will be
2940 handled by the kernel. If the token value is 0, then any token
2941 associated with the service will be forgotten, and subsequent RTAS
2942 calls by the guest for that service will be passed to userspace to be
2945 4.87 KVM_SET_GUEST_DEBUG
2947 Capability: KVM_CAP_SET_GUEST_DEBUG
2948 Architectures: x86, s390, ppc, arm64
2950 Parameters: struct kvm_guest_debug (in)
2951 Returns: 0 on success; -1 on error
2953 struct kvm_guest_debug {
2956 struct kvm_guest_debug_arch arch;
2959 Set up the processor specific debug registers and configure vcpu for
2960 handling guest debug events. There are two parts to the structure, the
2961 first a control bitfield indicates the type of debug events to handle
2962 when running. Common control bits are:
2964 - KVM_GUESTDBG_ENABLE: guest debugging is enabled
2965 - KVM_GUESTDBG_SINGLESTEP: the next run should single-step
2967 The top 16 bits of the control field are architecture specific control
2968 flags which can include the following:
2970 - KVM_GUESTDBG_USE_SW_BP: using software breakpoints [x86, arm64]
2971 - KVM_GUESTDBG_USE_HW_BP: using hardware breakpoints [x86, s390, arm64]
2972 - KVM_GUESTDBG_INJECT_DB: inject DB type exception [x86]
2973 - KVM_GUESTDBG_INJECT_BP: inject BP type exception [x86]
2974 - KVM_GUESTDBG_EXIT_PENDING: trigger an immediate guest exit [s390]
2976 For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints
2977 are enabled in memory so we need to ensure breakpoint exceptions are
2978 correctly trapped and the KVM run loop exits at the breakpoint and not
2979 running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP
2980 we need to ensure the guest vCPUs architecture specific registers are
2981 updated to the correct (supplied) values.
2983 The second part of the structure is architecture specific and
2984 typically contains a set of debug registers.
2986 For arm64 the number of debug registers is implementation defined and
2987 can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and
2988 KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number
2989 indicating the number of supported registers.
2991 When debug events exit the main run loop with the reason
2992 KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run
2993 structure containing architecture specific debug information.
2995 4.88 KVM_GET_EMULATED_CPUID
2997 Capability: KVM_CAP_EXT_EMUL_CPUID
3000 Parameters: struct kvm_cpuid2 (in/out)
3001 Returns: 0 on success, -1 on error
3006 struct kvm_cpuid_entry2 entries[0];
3009 The member 'flags' is used for passing flags from userspace.
3011 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
3012 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1)
3013 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2)
3015 struct kvm_cpuid_entry2 {
3026 This ioctl returns x86 cpuid features which are emulated by
3027 kvm.Userspace can use the information returned by this ioctl to query
3028 which features are emulated by kvm instead of being present natively.
3030 Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2
3031 structure with the 'nent' field indicating the number of entries in
3032 the variable-size array 'entries'. If the number of entries is too low
3033 to describe the cpu capabilities, an error (E2BIG) is returned. If the
3034 number is too high, the 'nent' field is adjusted and an error (ENOMEM)
3035 is returned. If the number is just right, the 'nent' field is adjusted
3036 to the number of valid entries in the 'entries' array, which is then
3039 The entries returned are the set CPUID bits of the respective features
3040 which kvm emulates, as returned by the CPUID instruction, with unknown
3041 or unsupported feature bits cleared.
3043 Features like x2apic, for example, may not be present in the host cpu
3044 but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be
3045 emulated efficiently and thus not included here.
3047 The fields in each entry are defined as follows:
3049 function: the eax value used to obtain the entry
3050 index: the ecx value used to obtain the entry (for entries that are
3052 flags: an OR of zero or more of the following:
3053 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
3054 if the index field is valid
3055 KVM_CPUID_FLAG_STATEFUL_FUNC:
3056 if cpuid for this function returns different values for successive
3057 invocations; there will be several entries with the same function,
3058 all with this flag set
3059 KVM_CPUID_FLAG_STATE_READ_NEXT:
3060 for KVM_CPUID_FLAG_STATEFUL_FUNC entries, set if this entry is
3061 the first entry to be read by a cpu
3062 eax, ebx, ecx, edx: the values returned by the cpuid instruction for
3063 this function/index combination
3065 4.89 KVM_S390_MEM_OP
3067 Capability: KVM_CAP_S390_MEM_OP
3070 Parameters: struct kvm_s390_mem_op (in)
3071 Returns: = 0 on success,
3072 < 0 on generic error (e.g. -EFAULT or -ENOMEM),
3073 > 0 if an exception occurred while walking the page tables
3075 Read or write data from/to the logical (virtual) memory of a VCPU.
3077 Parameters are specified via the following structure:
3079 struct kvm_s390_mem_op {
3080 __u64 gaddr; /* the guest address */
3081 __u64 flags; /* flags */
3082 __u32 size; /* amount of bytes */
3083 __u32 op; /* type of operation */
3084 __u64 buf; /* buffer in userspace */
3085 __u8 ar; /* the access register number */
3086 __u8 reserved[31]; /* should be set to 0 */
3089 The type of operation is specified in the "op" field. It is either
3090 KVM_S390_MEMOP_LOGICAL_READ for reading from logical memory space or
3091 KVM_S390_MEMOP_LOGICAL_WRITE for writing to logical memory space. The
3092 KVM_S390_MEMOP_F_CHECK_ONLY flag can be set in the "flags" field to check
3093 whether the corresponding memory access would create an access exception
3094 (without touching the data in the memory at the destination). In case an
3095 access exception occurred while walking the MMU tables of the guest, the
3096 ioctl returns a positive error number to indicate the type of exception.
3097 This exception is also raised directly at the corresponding VCPU if the
3098 flag KVM_S390_MEMOP_F_INJECT_EXCEPTION is set in the "flags" field.
3100 The start address of the memory region has to be specified in the "gaddr"
3101 field, and the length of the region in the "size" field (which must not
3102 be 0). The maximum value for "size" can be obtained by checking the
3103 KVM_CAP_S390_MEM_OP capability. "buf" is the buffer supplied by the
3104 userspace application where the read data should be written to for
3105 KVM_S390_MEMOP_LOGICAL_READ, or where the data that should be written is
3106 stored for a KVM_S390_MEMOP_LOGICAL_WRITE. When KVM_S390_MEMOP_F_CHECK_ONLY
3107 is specified, "buf" is unused and can be NULL. "ar" designates the access
3108 register number to be used; the valid range is 0..15.
3110 The "reserved" field is meant for future extensions. It is not used by
3111 KVM with the currently defined set of flags.
3113 4.90 KVM_S390_GET_SKEYS
3115 Capability: KVM_CAP_S390_SKEYS
3118 Parameters: struct kvm_s390_skeys
3119 Returns: 0 on success, KVM_S390_GET_KEYS_NONE if guest is not using storage
3120 keys, negative value on error
3122 This ioctl is used to get guest storage key values on the s390
3123 architecture. The ioctl takes parameters via the kvm_s390_skeys struct.
3125 struct kvm_s390_skeys {
3128 __u64 skeydata_addr;
3133 The start_gfn field is the number of the first guest frame whose storage keys
3136 The count field is the number of consecutive frames (starting from start_gfn)
3137 whose storage keys to get. The count field must be at least 1 and the maximum
3138 allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range
3139 will cause the ioctl to return -EINVAL.
3141 The skeydata_addr field is the address to a buffer large enough to hold count
3142 bytes. This buffer will be filled with storage key data by the ioctl.
3144 4.91 KVM_S390_SET_SKEYS
3146 Capability: KVM_CAP_S390_SKEYS
3149 Parameters: struct kvm_s390_skeys
3150 Returns: 0 on success, negative value on error
3152 This ioctl is used to set guest storage key values on the s390
3153 architecture. The ioctl takes parameters via the kvm_s390_skeys struct.
3154 See section on KVM_S390_GET_SKEYS for struct definition.
3156 The start_gfn field is the number of the first guest frame whose storage keys
3159 The count field is the number of consecutive frames (starting from start_gfn)
3160 whose storage keys to get. The count field must be at least 1 and the maximum
3161 allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range
3162 will cause the ioctl to return -EINVAL.
3164 The skeydata_addr field is the address to a buffer containing count bytes of
3165 storage keys. Each byte in the buffer will be set as the storage key for a
3166 single frame starting at start_gfn for count frames.
3168 Note: If any architecturally invalid key value is found in the given data then
3169 the ioctl will return -EINVAL.
3173 Capability: KVM_CAP_S390_INJECT_IRQ
3176 Parameters: struct kvm_s390_irq (in)
3177 Returns: 0 on success, -1 on error
3179 EINVAL: interrupt type is invalid
3180 type is KVM_S390_SIGP_STOP and flag parameter is invalid value
3181 type is KVM_S390_INT_EXTERNAL_CALL and code is bigger
3182 than the maximum of VCPUs
3183 EBUSY: type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped
3184 type is KVM_S390_SIGP_STOP and a stop irq is already pending
3185 type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt
3188 Allows to inject an interrupt to the guest.
3190 Using struct kvm_s390_irq as a parameter allows
3191 to inject additional payload which is not
3192 possible via KVM_S390_INTERRUPT.
3194 Interrupt parameters are passed via kvm_s390_irq:
3196 struct kvm_s390_irq {
3199 struct kvm_s390_io_info io;
3200 struct kvm_s390_ext_info ext;
3201 struct kvm_s390_pgm_info pgm;
3202 struct kvm_s390_emerg_info emerg;
3203 struct kvm_s390_extcall_info extcall;
3204 struct kvm_s390_prefix_info prefix;
3205 struct kvm_s390_stop_info stop;
3206 struct kvm_s390_mchk_info mchk;
3211 type can be one of the following:
3213 KVM_S390_SIGP_STOP - sigp stop; parameter in .stop
3214 KVM_S390_PROGRAM_INT - program check; parameters in .pgm
3215 KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix
3216 KVM_S390_RESTART - restart; no parameters
3217 KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters
3218 KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters
3219 KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg
3220 KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall
3221 KVM_S390_MCHK - machine check interrupt; parameters in .mchk
3223 This is an asynchronous vcpu ioctl and can be invoked from any thread.
3225 4.94 KVM_S390_GET_IRQ_STATE
3227 Capability: KVM_CAP_S390_IRQ_STATE
3230 Parameters: struct kvm_s390_irq_state (out)
3231 Returns: >= number of bytes copied into buffer,
3232 -EINVAL if buffer size is 0,
3233 -ENOBUFS if buffer size is too small to fit all pending interrupts,
3234 -EFAULT if the buffer address was invalid
3236 This ioctl allows userspace to retrieve the complete state of all currently
3237 pending interrupts in a single buffer. Use cases include migration
3238 and introspection. The parameter structure contains the address of a
3239 userspace buffer and its length:
3241 struct kvm_s390_irq_state {
3243 __u32 flags; /* will stay unused for compatibility reasons */
3245 __u32 reserved[4]; /* will stay unused for compatibility reasons */
3248 Userspace passes in the above struct and for each pending interrupt a
3249 struct kvm_s390_irq is copied to the provided buffer.
3251 The structure contains a flags and a reserved field for future extensions. As
3252 the kernel never checked for flags == 0 and QEMU never pre-zeroed flags and
3253 reserved, these fields can not be used in the future without breaking
3256 If -ENOBUFS is returned the buffer provided was too small and userspace
3257 may retry with a bigger buffer.
3259 4.95 KVM_S390_SET_IRQ_STATE
3261 Capability: KVM_CAP_S390_IRQ_STATE
3264 Parameters: struct kvm_s390_irq_state (in)
3265 Returns: 0 on success,
3266 -EFAULT if the buffer address was invalid,
3267 -EINVAL for an invalid buffer length (see below),
3268 -EBUSY if there were already interrupts pending,
3269 errors occurring when actually injecting the
3270 interrupt. See KVM_S390_IRQ.
3272 This ioctl allows userspace to set the complete state of all cpu-local
3273 interrupts currently pending for the vcpu. It is intended for restoring
3274 interrupt state after a migration. The input parameter is a userspace buffer
3275 containing a struct kvm_s390_irq_state:
3277 struct kvm_s390_irq_state {
3279 __u32 flags; /* will stay unused for compatibility reasons */
3281 __u32 reserved[4]; /* will stay unused for compatibility reasons */
3284 The restrictions for flags and reserved apply as well.
3285 (see KVM_S390_GET_IRQ_STATE)
3287 The userspace memory referenced by buf contains a struct kvm_s390_irq
3288 for each interrupt to be injected into the guest.
3289 If one of the interrupts could not be injected for some reason the
3292 len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0
3293 and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq),
3294 which is the maximum number of possibly pending cpu-local interrupts.
3298 Capability: KVM_CAP_X86_SMM
3302 Returns: 0 on success, -1 on error
3304 Queues an SMI on the thread's vcpu.
3306 4.97 KVM_CAP_PPC_MULTITCE
3308 Capability: KVM_CAP_PPC_MULTITCE
3312 This capability means the kernel is capable of handling hypercalls
3313 H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user
3314 space. This significantly accelerates DMA operations for PPC KVM guests.
3315 User space should expect that its handlers for these hypercalls
3316 are not going to be called if user space previously registered LIOBN
3317 in KVM (via KVM_CREATE_SPAPR_TCE or similar calls).
3319 In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest,
3320 user space might have to advertise it for the guest. For example,
3321 IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is
3322 present in the "ibm,hypertas-functions" device-tree property.
3324 The hypercalls mentioned above may or may not be processed successfully
3325 in the kernel based fast path. If they can not be handled by the kernel,
3326 they will get passed on to user space. So user space still has to have
3327 an implementation for these despite the in kernel acceleration.
3329 This capability is always enabled.
3331 4.98 KVM_CREATE_SPAPR_TCE_64
3333 Capability: KVM_CAP_SPAPR_TCE_64
3334 Architectures: powerpc
3336 Parameters: struct kvm_create_spapr_tce_64 (in)
3337 Returns: file descriptor for manipulating the created TCE table
3339 This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit
3340 windows, described in 4.62 KVM_CREATE_SPAPR_TCE
3342 This capability uses extended struct in ioctl interface:
3344 /* for KVM_CAP_SPAPR_TCE_64 */
3345 struct kvm_create_spapr_tce_64 {
3349 __u64 offset; /* in pages */
3350 __u64 size; /* in pages */
3353 The aim of extension is to support an additional bigger DMA window with
3354 a variable page size.
3355 KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and
3356 a bus offset of the corresponding DMA window, @size and @offset are numbers
3359 @flags are not used at the moment.
3361 The rest of functionality is identical to KVM_CREATE_SPAPR_TCE.
3363 4.99 KVM_REINJECT_CONTROL
3365 Capability: KVM_CAP_REINJECT_CONTROL
3368 Parameters: struct kvm_reinject_control (in)
3369 Returns: 0 on success,
3370 -EFAULT if struct kvm_reinject_control cannot be read,
3371 -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier.
3373 i8254 (PIT) has two modes, reinject and !reinject. The default is reinject,
3374 where KVM queues elapsed i8254 ticks and monitors completion of interrupt from
3375 vector(s) that i8254 injects. Reinject mode dequeues a tick and injects its
3376 interrupt whenever there isn't a pending interrupt from i8254.
3377 !reinject mode injects an interrupt as soon as a tick arrives.
3379 struct kvm_reinject_control {
3384 pit_reinject = 0 (!reinject mode) is recommended, unless running an old
3385 operating system that uses the PIT for timing (e.g. Linux 2.4.x).
3387 4.100 KVM_PPC_CONFIGURE_V3_MMU
3389 Capability: KVM_CAP_PPC_RADIX_MMU or KVM_CAP_PPC_HASH_MMU_V3
3392 Parameters: struct kvm_ppc_mmuv3_cfg (in)
3393 Returns: 0 on success,
3394 -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read,
3395 -EINVAL if the configuration is invalid
3397 This ioctl controls whether the guest will use radix or HPT (hashed
3398 page table) translation, and sets the pointer to the process table for
3401 struct kvm_ppc_mmuv3_cfg {
3403 __u64 process_table;
3406 There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and
3407 KVM_PPC_MMUV3_GTSE. KVM_PPC_MMUV3_RADIX, if set, configures the guest
3408 to use radix tree translation, and if clear, to use HPT translation.
3409 KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest
3410 to be able to use the global TLB and SLB invalidation instructions;
3411 if clear, the guest may not use these instructions.
3413 The process_table field specifies the address and size of the guest
3414 process table, which is in the guest's space. This field is formatted
3415 as the second doubleword of the partition table entry, as defined in
3416 the Power ISA V3.00, Book III section 5.7.6.1.
3418 4.101 KVM_PPC_GET_RMMU_INFO
3420 Capability: KVM_CAP_PPC_RADIX_MMU
3423 Parameters: struct kvm_ppc_rmmu_info (out)
3424 Returns: 0 on success,
3425 -EFAULT if struct kvm_ppc_rmmu_info cannot be written,
3426 -EINVAL if no useful information can be returned
3428 This ioctl returns a structure containing two things: (a) a list
3429 containing supported radix tree geometries, and (b) a list that maps
3430 page sizes to put in the "AP" (actual page size) field for the tlbie
3431 (TLB invalidate entry) instruction.
3433 struct kvm_ppc_rmmu_info {
3434 struct kvm_ppc_radix_geom {
3439 __u32 ap_encodings[8];
3442 The geometries[] field gives up to 8 supported geometries for the
3443 radix page table, in terms of the log base 2 of the smallest page
3444 size, and the number of bits indexed at each level of the tree, from
3445 the PTE level up to the PGD level in that order. Any unused entries
3446 will have 0 in the page_shift field.
3448 The ap_encodings gives the supported page sizes and their AP field
3449 encodings, encoded with the AP value in the top 3 bits and the log
3450 base 2 of the page size in the bottom 6 bits.
3452 4.102 KVM_PPC_RESIZE_HPT_PREPARE
3454 Capability: KVM_CAP_SPAPR_RESIZE_HPT
3455 Architectures: powerpc
3457 Parameters: struct kvm_ppc_resize_hpt (in)
3458 Returns: 0 on successful completion,
3459 >0 if a new HPT is being prepared, the value is an estimated
3460 number of milliseconds until preparation is complete
3461 -EFAULT if struct kvm_reinject_control cannot be read,
3462 -EINVAL if the supplied shift or flags are invalid
3463 -ENOMEM if unable to allocate the new HPT
3464 -ENOSPC if there was a hash collision when moving existing
3465 HPT entries to the new HPT
3466 -EIO on other error conditions
3468 Used to implement the PAPR extension for runtime resizing of a guest's
3469 Hashed Page Table (HPT). Specifically this starts, stops or monitors
3470 the preparation of a new potential HPT for the guest, essentially
3471 implementing the H_RESIZE_HPT_PREPARE hypercall.
3473 If called with shift > 0 when there is no pending HPT for the guest,
3474 this begins preparation of a new pending HPT of size 2^(shift) bytes.
3475 It then returns a positive integer with the estimated number of
3476 milliseconds until preparation is complete.
3478 If called when there is a pending HPT whose size does not match that
3479 requested in the parameters, discards the existing pending HPT and
3480 creates a new one as above.
3482 If called when there is a pending HPT of the size requested, will:
3483 * If preparation of the pending HPT is already complete, return 0
3484 * If preparation of the pending HPT has failed, return an error
3485 code, then discard the pending HPT.
3486 * If preparation of the pending HPT is still in progress, return an
3487 estimated number of milliseconds until preparation is complete.
3489 If called with shift == 0, discards any currently pending HPT and
3490 returns 0 (i.e. cancels any in-progress preparation).
3492 flags is reserved for future expansion, currently setting any bits in
3493 flags will result in an -EINVAL.
3495 Normally this will be called repeatedly with the same parameters until
3496 it returns <= 0. The first call will initiate preparation, subsequent
3497 ones will monitor preparation until it completes or fails.
3499 struct kvm_ppc_resize_hpt {
3505 4.103 KVM_PPC_RESIZE_HPT_COMMIT
3507 Capability: KVM_CAP_SPAPR_RESIZE_HPT
3508 Architectures: powerpc
3510 Parameters: struct kvm_ppc_resize_hpt (in)
3511 Returns: 0 on successful completion,
3512 -EFAULT if struct kvm_reinject_control cannot be read,
3513 -EINVAL if the supplied shift or flags are invalid
3514 -ENXIO is there is no pending HPT, or the pending HPT doesn't
3515 have the requested size
3516 -EBUSY if the pending HPT is not fully prepared
3517 -ENOSPC if there was a hash collision when moving existing
3518 HPT entries to the new HPT
3519 -EIO on other error conditions
3521 Used to implement the PAPR extension for runtime resizing of a guest's
3522 Hashed Page Table (HPT). Specifically this requests that the guest be
3523 transferred to working with the new HPT, essentially implementing the
3524 H_RESIZE_HPT_COMMIT hypercall.
3526 This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has
3527 returned 0 with the same parameters. In other cases
3528 KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or
3529 -EBUSY, though others may be possible if the preparation was started,
3532 This will have undefined effects on the guest if it has not already
3533 placed itself in a quiescent state where no vcpu will make MMU enabled
3536 On succsful completion, the pending HPT will become the guest's active
3537 HPT and the previous HPT will be discarded.
3539 On failure, the guest will still be operating on its previous HPT.
3541 struct kvm_ppc_resize_hpt {
3547 4.104 KVM_X86_GET_MCE_CAP_SUPPORTED
3549 Capability: KVM_CAP_MCE
3552 Parameters: u64 mce_cap (out)
3553 Returns: 0 on success, -1 on error
3555 Returns supported MCE capabilities. The u64 mce_cap parameter
3556 has the same format as the MSR_IA32_MCG_CAP register. Supported
3557 capabilities will have the corresponding bits set.
3559 4.105 KVM_X86_SETUP_MCE
3561 Capability: KVM_CAP_MCE
3564 Parameters: u64 mcg_cap (in)
3565 Returns: 0 on success,
3566 -EFAULT if u64 mcg_cap cannot be read,
3567 -EINVAL if the requested number of banks is invalid,
3568 -EINVAL if requested MCE capability is not supported.
3570 Initializes MCE support for use. The u64 mcg_cap parameter
3571 has the same format as the MSR_IA32_MCG_CAP register and
3572 specifies which capabilities should be enabled. The maximum
3573 supported number of error-reporting banks can be retrieved when
3574 checking for KVM_CAP_MCE. The supported capabilities can be
3575 retrieved with KVM_X86_GET_MCE_CAP_SUPPORTED.
3577 4.106 KVM_X86_SET_MCE
3579 Capability: KVM_CAP_MCE
3582 Parameters: struct kvm_x86_mce (in)
3583 Returns: 0 on success,
3584 -EFAULT if struct kvm_x86_mce cannot be read,
3585 -EINVAL if the bank number is invalid,
3586 -EINVAL if VAL bit is not set in status field.
3588 Inject a machine check error (MCE) into the guest. The input
3591 struct kvm_x86_mce {
3601 If the MCE being reported is an uncorrected error, KVM will
3602 inject it as an MCE exception into the guest. If the guest
3603 MCG_STATUS register reports that an MCE is in progress, KVM
3604 causes an KVM_EXIT_SHUTDOWN vmexit.
3606 Otherwise, if the MCE is a corrected error, KVM will just
3607 store it in the corresponding bank (provided this bank is
3608 not holding a previously reported uncorrected error).
3610 4.107 KVM_S390_GET_CMMA_BITS
3612 Capability: KVM_CAP_S390_CMMA_MIGRATION
3615 Parameters: struct kvm_s390_cmma_log (in, out)
3616 Returns: 0 on success, a negative value on error
3618 This ioctl is used to get the values of the CMMA bits on the s390
3619 architecture. It is meant to be used in two scenarios:
3620 - During live migration to save the CMMA values. Live migration needs
3621 to be enabled via the KVM_REQ_START_MIGRATION VM property.
3622 - To non-destructively peek at the CMMA values, with the flag
3623 KVM_S390_CMMA_PEEK set.
3625 The ioctl takes parameters via the kvm_s390_cmma_log struct. The desired
3626 values are written to a buffer whose location is indicated via the "values"
3627 member in the kvm_s390_cmma_log struct. The values in the input struct are
3628 also updated as needed.
3629 Each CMMA value takes up one byte.
3631 struct kvm_s390_cmma_log {
3642 start_gfn is the number of the first guest frame whose CMMA values are
3645 count is the length of the buffer in bytes,
3647 values points to the buffer where the result will be written to.
3649 If count is greater than KVM_S390_SKEYS_MAX, then it is considered to be
3650 KVM_S390_SKEYS_MAX. KVM_S390_SKEYS_MAX is re-used for consistency with
3653 The result is written in the buffer pointed to by the field values, and
3654 the values of the input parameter are updated as follows.
3656 Depending on the flags, different actions are performed. The only
3657 supported flag so far is KVM_S390_CMMA_PEEK.
3659 The default behaviour if KVM_S390_CMMA_PEEK is not set is:
3660 start_gfn will indicate the first page frame whose CMMA bits were dirty.
3661 It is not necessarily the same as the one passed as input, as clean pages
3664 count will indicate the number of bytes actually written in the buffer.
3665 It can (and very often will) be smaller than the input value, since the
3666 buffer is only filled until 16 bytes of clean values are found (which
3667 are then not copied in the buffer). Since a CMMA migration block needs
3668 the base address and the length, for a total of 16 bytes, we will send
3669 back some clean data if there is some dirty data afterwards, as long as
3670 the size of the clean data does not exceed the size of the header. This
3671 allows to minimize the amount of data to be saved or transferred over
3672 the network at the expense of more roundtrips to userspace. The next
3673 invocation of the ioctl will skip over all the clean values, saving
3674 potentially more than just the 16 bytes we found.
3676 If KVM_S390_CMMA_PEEK is set:
3677 the existing storage attributes are read even when not in migration
3678 mode, and no other action is performed;
3680 the output start_gfn will be equal to the input start_gfn,
3682 the output count will be equal to the input count, except if the end of
3683 memory has been reached.
3686 the field "remaining" will indicate the total number of dirty CMMA values
3687 still remaining, or 0 if KVM_S390_CMMA_PEEK is set and migration mode is
3692 values points to the userspace buffer where the result will be stored.
3694 This ioctl can fail with -ENOMEM if not enough memory can be allocated to
3695 complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
3696 KVM_S390_CMMA_PEEK is not set but migration mode was not enabled, with
3697 -EFAULT if the userspace address is invalid or if no page table is
3698 present for the addresses (e.g. when using hugepages).
3700 4.108 KVM_S390_SET_CMMA_BITS
3702 Capability: KVM_CAP_S390_CMMA_MIGRATION
3705 Parameters: struct kvm_s390_cmma_log (in)
3706 Returns: 0 on success, a negative value on error
3708 This ioctl is used to set the values of the CMMA bits on the s390
3709 architecture. It is meant to be used during live migration to restore
3710 the CMMA values, but there are no restrictions on its use.
3711 The ioctl takes parameters via the kvm_s390_cmma_values struct.
3712 Each CMMA value takes up one byte.
3714 struct kvm_s390_cmma_log {
3725 start_gfn indicates the starting guest frame number,
3727 count indicates how many values are to be considered in the buffer,
3729 flags is not used and must be 0.
3731 mask indicates which PGSTE bits are to be considered.
3733 remaining is not used.
3735 values points to the buffer in userspace where to store the values.
3737 This ioctl can fail with -ENOMEM if not enough memory can be allocated to
3738 complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
3739 the count field is too large (e.g. more than KVM_S390_CMMA_SIZE_MAX) or
3740 if the flags field was not 0, with -EFAULT if the userspace address is
3741 invalid, if invalid pages are written to (e.g. after the end of memory)
3742 or if no page table is present for the addresses (e.g. when using
3745 4.109 KVM_PPC_GET_CPU_CHAR
3747 Capability: KVM_CAP_PPC_GET_CPU_CHAR
3748 Architectures: powerpc
3750 Parameters: struct kvm_ppc_cpu_char (out)
3751 Returns: 0 on successful completion
3752 -EFAULT if struct kvm_ppc_cpu_char cannot be written
3754 This ioctl gives userspace information about certain characteristics
3755 of the CPU relating to speculative execution of instructions and
3756 possible information leakage resulting from speculative execution (see
3757 CVE-2017-5715, CVE-2017-5753 and CVE-2017-5754). The information is
3758 returned in struct kvm_ppc_cpu_char, which looks like this:
3760 struct kvm_ppc_cpu_char {
3761 __u64 character; /* characteristics of the CPU */
3762 __u64 behaviour; /* recommended software behaviour */
3763 __u64 character_mask; /* valid bits in character */
3764 __u64 behaviour_mask; /* valid bits in behaviour */
3767 For extensibility, the character_mask and behaviour_mask fields
3768 indicate which bits of character and behaviour have been filled in by
3769 the kernel. If the set of defined bits is extended in future then
3770 userspace will be able to tell whether it is running on a kernel that
3771 knows about the new bits.
3773 The character field describes attributes of the CPU which can help
3774 with preventing inadvertent information disclosure - specifically,
3775 whether there is an instruction to flash-invalidate the L1 data cache
3776 (ori 30,30,0 or mtspr SPRN_TRIG2,rN), whether the L1 data cache is set
3777 to a mode where entries can only be used by the thread that created
3778 them, whether the bcctr[l] instruction prevents speculation, and
3779 whether a speculation barrier instruction (ori 31,31,0) is provided.
3781 The behaviour field describes actions that software should take to
3782 prevent inadvertent information disclosure, and thus describes which
3783 vulnerabilities the hardware is subject to; specifically whether the
3784 L1 data cache should be flushed when returning to user mode from the
3785 kernel, and whether a speculation barrier should be placed between an
3786 array bounds check and the array access.
3788 These fields use the same bit definitions as the new
3789 H_GET_CPU_CHARACTERISTICS hypercall.
3791 4.110 KVM_MEMORY_ENCRYPT_OP
3796 Parameters: an opaque platform specific structure (in/out)
3797 Returns: 0 on success; -1 on error
3799 If the platform supports creating encrypted VMs then this ioctl can be used
3800 for issuing platform-specific memory encryption commands to manage those
3803 Currently, this ioctl is used for issuing Secure Encrypted Virtualization
3804 (SEV) commands on AMD Processors. The SEV commands are defined in
3805 Documentation/virt/kvm/amd-memory-encryption.rst.
3807 4.111 KVM_MEMORY_ENCRYPT_REG_REGION
3812 Parameters: struct kvm_enc_region (in)
3813 Returns: 0 on success; -1 on error
3815 This ioctl can be used to register a guest memory region which may
3816 contain encrypted data (e.g. guest RAM, SMRAM etc).
3818 It is used in the SEV-enabled guest. When encryption is enabled, a guest
3819 memory region may contain encrypted data. The SEV memory encryption
3820 engine uses a tweak such that two identical plaintext pages, each at
3821 different locations will have differing ciphertexts. So swapping or
3822 moving ciphertext of those pages will not result in plaintext being
3823 swapped. So relocating (or migrating) physical backing pages for the SEV
3824 guest will require some additional steps.
3826 Note: The current SEV key management spec does not provide commands to
3827 swap or migrate (move) ciphertext pages. Hence, for now we pin the guest
3828 memory region registered with the ioctl.
3830 4.112 KVM_MEMORY_ENCRYPT_UNREG_REGION
3835 Parameters: struct kvm_enc_region (in)
3836 Returns: 0 on success; -1 on error
3838 This ioctl can be used to unregister the guest memory region registered
3839 with KVM_MEMORY_ENCRYPT_REG_REGION ioctl above.
3841 4.113 KVM_HYPERV_EVENTFD
3843 Capability: KVM_CAP_HYPERV_EVENTFD
3846 Parameters: struct kvm_hyperv_eventfd (in)
3848 This ioctl (un)registers an eventfd to receive notifications from the guest on
3849 the specified Hyper-V connection id through the SIGNAL_EVENT hypercall, without
3850 causing a user exit. SIGNAL_EVENT hypercall with non-zero event flag number
3851 (bits 24-31) still triggers a KVM_EXIT_HYPERV_HCALL user exit.
3853 struct kvm_hyperv_eventfd {
3860 The conn_id field should fit within 24 bits:
3862 #define KVM_HYPERV_CONN_ID_MASK 0x00ffffff
3864 The acceptable values for the flags field are:
3866 #define KVM_HYPERV_EVENTFD_DEASSIGN (1 << 0)
3868 Returns: 0 on success,
3869 -EINVAL if conn_id or flags is outside the allowed range
3870 -ENOENT on deassign if the conn_id isn't registered
3871 -EEXIST on assign if the conn_id is already registered
3873 4.114 KVM_GET_NESTED_STATE
3875 Capability: KVM_CAP_NESTED_STATE
3878 Parameters: struct kvm_nested_state (in/out)
3879 Returns: 0 on success, -1 on error
3881 E2BIG: the total state size exceeds the value of 'size' specified by
3882 the user; the size required will be written into size.
3884 struct kvm_nested_state {
3890 struct kvm_vmx_nested_state_hdr vmx;
3891 struct kvm_svm_nested_state_hdr svm;
3893 /* Pad the header to 128 bytes. */
3898 struct kvm_vmx_nested_state_data vmx[0];
3899 struct kvm_svm_nested_state_data svm[0];
3903 #define KVM_STATE_NESTED_GUEST_MODE 0x00000001
3904 #define KVM_STATE_NESTED_RUN_PENDING 0x00000002
3905 #define KVM_STATE_NESTED_EVMCS 0x00000004
3907 #define KVM_STATE_NESTED_FORMAT_VMX 0
3908 #define KVM_STATE_NESTED_FORMAT_SVM 1
3910 #define KVM_STATE_NESTED_VMX_VMCS_SIZE 0x1000
3912 #define KVM_STATE_NESTED_VMX_SMM_GUEST_MODE 0x00000001
3913 #define KVM_STATE_NESTED_VMX_SMM_VMXON 0x00000002
3915 struct kvm_vmx_nested_state_hdr {
3924 struct kvm_vmx_nested_state_data {
3925 __u8 vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE];
3926 __u8 shadow_vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE];
3929 This ioctl copies the vcpu's nested virtualization state from the kernel to
3932 The maximum size of the state can be retrieved by passing KVM_CAP_NESTED_STATE
3933 to the KVM_CHECK_EXTENSION ioctl().
3935 4.115 KVM_SET_NESTED_STATE
3937 Capability: KVM_CAP_NESTED_STATE
3940 Parameters: struct kvm_nested_state (in)
3941 Returns: 0 on success, -1 on error
3943 This copies the vcpu's kvm_nested_state struct from userspace to the kernel.
3944 For the definition of struct kvm_nested_state, see KVM_GET_NESTED_STATE.
3946 4.116 KVM_(UN)REGISTER_COALESCED_MMIO
3948 Capability: KVM_CAP_COALESCED_MMIO (for coalesced mmio)
3949 KVM_CAP_COALESCED_PIO (for coalesced pio)
3952 Parameters: struct kvm_coalesced_mmio_zone
3953 Returns: 0 on success, < 0 on error
3955 Coalesced I/O is a performance optimization that defers hardware
3956 register write emulation so that userspace exits are avoided. It is
3957 typically used to reduce the overhead of emulating frequently accessed
3960 When a hardware register is configured for coalesced I/O, write accesses
3961 do not exit to userspace and their value is recorded in a ring buffer
3962 that is shared between kernel and userspace.
3964 Coalesced I/O is used if one or more write accesses to a hardware
3965 register can be deferred until a read or a write to another hardware
3966 register on the same device. This last access will cause a vmexit and
3967 userspace will process accesses from the ring buffer before emulating
3968 it. That will avoid exiting to userspace on repeated writes.
3970 Coalesced pio is based on coalesced mmio. There is little difference
3971 between coalesced mmio and pio except that coalesced pio records accesses
3974 4.117 KVM_CLEAR_DIRTY_LOG (vm ioctl)
3976 Capability: KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
3977 Architectures: x86, arm, arm64, mips
3979 Parameters: struct kvm_dirty_log (in)
3980 Returns: 0 on success, -1 on error
3982 /* for KVM_CLEAR_DIRTY_LOG */
3983 struct kvm_clear_dirty_log {
3988 void __user *dirty_bitmap; /* one bit per page */
3993 The ioctl clears the dirty status of pages in a memory slot, according to
3994 the bitmap that is passed in struct kvm_clear_dirty_log's dirty_bitmap
3995 field. Bit 0 of the bitmap corresponds to page "first_page" in the
3996 memory slot, and num_pages is the size in bits of the input bitmap.
3997 first_page must be a multiple of 64; num_pages must also be a multiple of
3998 64 unless first_page + num_pages is the size of the memory slot. For each
3999 bit that is set in the input bitmap, the corresponding page is marked "clean"
4000 in KVM's dirty bitmap, and dirty tracking is re-enabled for that page
4001 (for example via write-protection, or by clearing the dirty bit in
4002 a page table entry).
4004 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 specifies
4005 the address space for which you want to return the dirty bitmap.
4006 They must be less than the value that KVM_CHECK_EXTENSION returns for
4007 the KVM_CAP_MULTI_ADDRESS_SPACE capability.
4009 This ioctl is mostly useful when KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
4010 is enabled; for more information, see the description of the capability.
4011 However, it can always be used as long as KVM_CHECK_EXTENSION confirms
4012 that KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is present.
4014 4.118 KVM_GET_SUPPORTED_HV_CPUID
4016 Capability: KVM_CAP_HYPERV_CPUID
4019 Parameters: struct kvm_cpuid2 (in/out)
4020 Returns: 0 on success, -1 on error
4025 struct kvm_cpuid_entry2 entries[0];
4028 struct kvm_cpuid_entry2 {
4039 This ioctl returns x86 cpuid features leaves related to Hyper-V emulation in
4040 KVM. Userspace can use the information returned by this ioctl to construct
4041 cpuid information presented to guests consuming Hyper-V enlightenments (e.g.
4042 Windows or Hyper-V guests).
4044 CPUID feature leaves returned by this ioctl are defined by Hyper-V Top Level
4045 Functional Specification (TLFS). These leaves can't be obtained with
4046 KVM_GET_SUPPORTED_CPUID ioctl because some of them intersect with KVM feature
4047 leaves (0x40000000, 0x40000001).
4049 Currently, the following list of CPUID leaves are returned:
4050 HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS
4051 HYPERV_CPUID_INTERFACE
4052 HYPERV_CPUID_VERSION
4053 HYPERV_CPUID_FEATURES
4054 HYPERV_CPUID_ENLIGHTMENT_INFO
4055 HYPERV_CPUID_IMPLEMENT_LIMITS
4056 HYPERV_CPUID_NESTED_FEATURES
4058 HYPERV_CPUID_NESTED_FEATURES leaf is only exposed when Enlightened VMCS was
4059 enabled on the corresponding vCPU (KVM_CAP_HYPERV_ENLIGHTENED_VMCS).
4061 Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure
4062 with the 'nent' field indicating the number of entries in the variable-size
4063 array 'entries'. If the number of entries is too low to describe all Hyper-V
4064 feature leaves, an error (E2BIG) is returned. If the number is more or equal
4065 to the number of Hyper-V feature leaves, the 'nent' field is adjusted to the
4066 number of valid entries in the 'entries' array, which is then filled.
4068 'index' and 'flags' fields in 'struct kvm_cpuid_entry2' are currently reserved,
4069 userspace should not expect to get any particular value there.
4071 4.119 KVM_ARM_VCPU_FINALIZE
4073 Architectures: arm, arm64
4075 Parameters: int feature (in)
4076 Returns: 0 on success, -1 on error
4078 EPERM: feature not enabled, needs configuration, or already finalized
4079 EINVAL: feature unknown or not present
4081 Recognised values for feature:
4082 arm64 KVM_ARM_VCPU_SVE (requires KVM_CAP_ARM_SVE)
4084 Finalizes the configuration of the specified vcpu feature.
4086 The vcpu must already have been initialised, enabling the affected feature, by
4087 means of a successful KVM_ARM_VCPU_INIT call with the appropriate flag set in
4090 For affected vcpu features, this is a mandatory step that must be performed
4091 before the vcpu is fully usable.
4093 Between KVM_ARM_VCPU_INIT and KVM_ARM_VCPU_FINALIZE, the feature may be
4094 configured by use of ioctls such as KVM_SET_ONE_REG. The exact configuration
4095 that should be performaned and how to do it are feature-dependent.
4097 Other calls that depend on a particular feature being finalized, such as
4098 KVM_RUN, KVM_GET_REG_LIST, KVM_GET_ONE_REG and KVM_SET_ONE_REG, will fail with
4099 -EPERM unless the feature has already been finalized by means of a
4100 KVM_ARM_VCPU_FINALIZE call.
4102 See KVM_ARM_VCPU_INIT for details of vcpu features that require finalization
4105 4.120 KVM_SET_PMU_EVENT_FILTER
4107 Capability: KVM_CAP_PMU_EVENT_FILTER
4110 Parameters: struct kvm_pmu_event_filter (in)
4111 Returns: 0 on success, -1 on error
4113 struct kvm_pmu_event_filter {
4116 __u32 fixed_counter_bitmap;
4122 This ioctl restricts the set of PMU events that the guest can program.
4123 The argument holds a list of events which will be allowed or denied.
4124 The eventsel+umask of each event the guest attempts to program is compared
4125 against the events field to determine whether the guest should have access.
4126 The events field only controls general purpose counters; fixed purpose
4127 counters are controlled by the fixed_counter_bitmap.
4129 No flags are defined yet, the field must be zero.
4131 Valid values for 'action':
4132 #define KVM_PMU_EVENT_ALLOW 0
4133 #define KVM_PMU_EVENT_DENY 1
4136 5. The kvm_run structure
4137 ------------------------
4139 Application code obtains a pointer to the kvm_run structure by
4140 mmap()ing a vcpu fd. From that point, application code can control
4141 execution by changing fields in kvm_run prior to calling the KVM_RUN
4142 ioctl, and obtain information about the reason KVM_RUN returned by
4143 looking up structure members.
4147 __u8 request_interrupt_window;
4149 Request that KVM_RUN return when it becomes possible to inject external
4150 interrupts into the guest. Useful in conjunction with KVM_INTERRUPT.
4152 __u8 immediate_exit;
4154 This field is polled once when KVM_RUN starts; if non-zero, KVM_RUN
4155 exits immediately, returning -EINTR. In the common scenario where a
4156 signal is used to "kick" a VCPU out of KVM_RUN, this field can be used
4157 to avoid usage of KVM_SET_SIGNAL_MASK, which has worse scalability.
4158 Rather than blocking the signal outside KVM_RUN, userspace can set up
4159 a signal handler that sets run->immediate_exit to a non-zero value.
4161 This field is ignored if KVM_CAP_IMMEDIATE_EXIT is not available.
4168 When KVM_RUN has returned successfully (return value 0), this informs
4169 application code why KVM_RUN has returned. Allowable values for this
4170 field are detailed below.
4172 __u8 ready_for_interrupt_injection;
4174 If request_interrupt_window has been specified, this field indicates
4175 an interrupt can be injected now with KVM_INTERRUPT.
4179 The value of the current interrupt flag. Only valid if in-kernel
4180 local APIC is not used.
4184 More architecture-specific flags detailing state of the VCPU that may
4185 affect the device's behavior. The only currently defined flag is
4186 KVM_RUN_X86_SMM, which is valid on x86 machines and is set if the
4187 VCPU is in system management mode.
4189 /* in (pre_kvm_run), out (post_kvm_run) */
4192 The value of the cr8 register. Only valid if in-kernel local APIC is
4193 not used. Both input and output.
4197 The value of the APIC BASE msr. Only valid if in-kernel local
4198 APIC is not used. Both input and output.
4201 /* KVM_EXIT_UNKNOWN */
4203 __u64 hardware_exit_reason;
4206 If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown
4207 reasons. Further architecture-specific information is available in
4208 hardware_exit_reason.
4210 /* KVM_EXIT_FAIL_ENTRY */
4212 __u64 hardware_entry_failure_reason;
4215 If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due
4216 to unknown reasons. Further architecture-specific information is
4217 available in hardware_entry_failure_reason.
4219 /* KVM_EXIT_EXCEPTION */
4229 #define KVM_EXIT_IO_IN 0
4230 #define KVM_EXIT_IO_OUT 1
4232 __u8 size; /* bytes */
4235 __u64 data_offset; /* relative to kvm_run start */
4238 If exit_reason is KVM_EXIT_IO, then the vcpu has
4239 executed a port I/O instruction which could not be satisfied by kvm.
4240 data_offset describes where the data is located (KVM_EXIT_IO_OUT) or
4241 where kvm expects application code to place the data for the next
4242 KVM_RUN invocation (KVM_EXIT_IO_IN). Data format is a packed array.
4244 /* KVM_EXIT_DEBUG */
4246 struct kvm_debug_exit_arch arch;
4249 If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event
4250 for which architecture specific information is returned.
4260 If exit_reason is KVM_EXIT_MMIO, then the vcpu has
4261 executed a memory-mapped I/O instruction which could not be satisfied
4262 by kvm. The 'data' member contains the written data if 'is_write' is
4263 true, and should be filled by application code otherwise.
4265 The 'data' member contains, in its first 'len' bytes, the value as it would
4266 appear if the VCPU performed a load or store of the appropriate width directly
4269 NOTE: For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR and
4270 KVM_EXIT_EPR the corresponding
4271 operations are complete (and guest state is consistent) only after userspace
4272 has re-entered the kernel with KVM_RUN. The kernel side will first finish
4273 incomplete operations and then check for pending signals. Userspace
4274 can re-enter the guest with an unmasked signal pending to complete
4277 /* KVM_EXIT_HYPERCALL */
4286 Unused. This was once used for 'hypercall to userspace'. To implement
4287 such functionality, use KVM_EXIT_IO (x86) or KVM_EXIT_MMIO (all except s390).
4288 Note KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO.
4290 /* KVM_EXIT_TPR_ACCESS */
4297 To be documented (KVM_TPR_ACCESS_REPORTING).
4299 /* KVM_EXIT_S390_SIEIC */
4302 __u64 mask; /* psw upper half */
4303 __u64 addr; /* psw lower half */
4310 /* KVM_EXIT_S390_RESET */
4311 #define KVM_S390_RESET_POR 1
4312 #define KVM_S390_RESET_CLEAR 2
4313 #define KVM_S390_RESET_SUBSYSTEM 4
4314 #define KVM_S390_RESET_CPU_INIT 8
4315 #define KVM_S390_RESET_IPL 16
4316 __u64 s390_reset_flags;
4320 /* KVM_EXIT_S390_UCONTROL */
4322 __u64 trans_exc_code;
4326 s390 specific. A page fault has occurred for a user controlled virtual
4327 machine (KVM_VM_S390_UNCONTROL) on it's host page table that cannot be
4328 resolved by the kernel.
4329 The program code and the translation exception code that were placed
4330 in the cpu's lowcore are presented here as defined by the z Architecture
4331 Principles of Operation Book in the Chapter for Dynamic Address Translation
4341 Deprecated - was used for 440 KVM.
4348 MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch
4349 hypercalls and exit with this exit struct that contains all the guest gprs.
4351 If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall.
4352 Userspace can now handle the hypercall and when it's done modify the gprs as
4353 necessary. Upon guest entry all guest GPRs will then be replaced by the values
4356 /* KVM_EXIT_PAPR_HCALL */
4363 This is used on 64-bit PowerPC when emulating a pSeries partition,
4364 e.g. with the 'pseries' machine type in qemu. It occurs when the
4365 guest does a hypercall using the 'sc 1' instruction. The 'nr' field
4366 contains the hypercall number (from the guest R3), and 'args' contains
4367 the arguments (from the guest R4 - R12). Userspace should put the
4368 return code in 'ret' and any extra returned values in args[].
4369 The possible hypercalls are defined in the Power Architecture Platform
4370 Requirements (PAPR) document available from www.power.org (free
4371 developer registration required to access it).
4373 /* KVM_EXIT_S390_TSCH */
4375 __u16 subchannel_id;
4376 __u16 subchannel_nr;
4383 s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled
4384 and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O
4385 interrupt for the target subchannel has been dequeued and subchannel_id,
4386 subchannel_nr, io_int_parm and io_int_word contain the parameters for that
4387 interrupt. ipb is needed for instruction parameter decoding.
4394 On FSL BookE PowerPC chips, the interrupt controller has a fast patch
4395 interrupt acknowledge path to the core. When the core successfully
4396 delivers an interrupt, it automatically populates the EPR register with
4397 the interrupt vector number and acknowledges the interrupt inside
4398 the interrupt controller.
4400 In case the interrupt controller lives in user space, we need to do
4401 the interrupt acknowledge cycle through it to fetch the next to be
4402 delivered interrupt vector using this exit.
4404 It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an
4405 external interrupt has just been delivered into the guest. User space
4406 should put the acknowledged interrupt vector into the 'epr' field.
4408 /* KVM_EXIT_SYSTEM_EVENT */
4410 #define KVM_SYSTEM_EVENT_SHUTDOWN 1
4411 #define KVM_SYSTEM_EVENT_RESET 2
4412 #define KVM_SYSTEM_EVENT_CRASH 3
4417 If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered
4418 a system-level event using some architecture specific mechanism (hypercall
4419 or some special instruction). In case of ARM/ARM64, this is triggered using
4420 HVC instruction based PSCI call from the vcpu. The 'type' field describes
4421 the system-level event type. The 'flags' field describes architecture
4422 specific flags for the system-level event.
4424 Valid values for 'type' are:
4425 KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the
4426 VM. Userspace is not obliged to honour this, and if it does honour
4427 this does not need to destroy the VM synchronously (ie it may call
4428 KVM_RUN again before shutdown finally occurs).
4429 KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM.
4430 As with SHUTDOWN, userspace can choose to ignore the request, or
4431 to schedule the reset to occur in the future and may call KVM_RUN again.
4432 KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest
4433 has requested a crash condition maintenance. Userspace can choose
4434 to ignore the request, or to gather VM memory core dump and/or
4435 reset/shutdown of the VM.
4437 /* KVM_EXIT_IOAPIC_EOI */
4442 Indicates that the VCPU's in-kernel local APIC received an EOI for a
4443 level-triggered IOAPIC interrupt. This exit only triggers when the
4444 IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled);
4445 the userspace IOAPIC should process the EOI and retrigger the interrupt if
4446 it is still asserted. Vector is the LAPIC interrupt vector for which the
4449 struct kvm_hyperv_exit {
4450 #define KVM_EXIT_HYPERV_SYNIC 1
4451 #define KVM_EXIT_HYPERV_HCALL 2
4469 /* KVM_EXIT_HYPERV */
4470 struct kvm_hyperv_exit hyperv;
4471 Indicates that the VCPU exits into userspace to process some tasks
4472 related to Hyper-V emulation.
4473 Valid values for 'type' are:
4474 KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about
4475 Hyper-V SynIC state change. Notification is used to remap SynIC
4476 event/message pages and to enable/disable SynIC messages/events processing
4479 /* Fix the size of the union. */
4484 * shared registers between kvm and userspace.
4485 * kvm_valid_regs specifies the register classes set by the host
4486 * kvm_dirty_regs specified the register classes dirtied by userspace
4487 * struct kvm_sync_regs is architecture specific, as well as the
4488 * bits for kvm_valid_regs and kvm_dirty_regs
4490 __u64 kvm_valid_regs;
4491 __u64 kvm_dirty_regs;
4493 struct kvm_sync_regs regs;
4494 char padding[SYNC_REGS_SIZE_BYTES];
4497 If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access
4498 certain guest registers without having to call SET/GET_*REGS. Thus we can
4499 avoid some system call overhead if userspace has to handle the exit.
4500 Userspace can query the validity of the structure by checking
4501 kvm_valid_regs for specific bits. These bits are architecture specific
4502 and usually define the validity of a groups of registers. (e.g. one bit
4503 for general purpose registers)
4505 Please note that the kernel is allowed to use the kvm_run structure as the
4506 primary storage for certain register types. Therefore, the kernel may use the
4507 values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set.
4513 6. Capabilities that can be enabled on vCPUs
4514 --------------------------------------------
4516 There are certain capabilities that change the behavior of the virtual CPU or
4517 the virtual machine when enabled. To enable them, please see section 4.37.
4518 Below you can find a list of capabilities and what their effect on the vCPU or
4519 the virtual machine is when enabling them.
4521 The following information is provided along with the description:
4523 Architectures: which instruction set architectures provide this ioctl.
4524 x86 includes both i386 and x86_64.
4526 Target: whether this is a per-vcpu or per-vm capability.
4528 Parameters: what parameters are accepted by the capability.
4530 Returns: the return value. General error numbers (EBADF, ENOMEM, EINVAL)
4531 are not detailed, but errors with specific meanings are.
4539 Returns: 0 on success; -1 on error
4541 This capability enables interception of OSI hypercalls that otherwise would
4542 be treated as normal system calls to be injected into the guest. OSI hypercalls
4543 were invented by Mac-on-Linux to have a standardized communication mechanism
4544 between the guest and the host.
4546 When this capability is enabled, KVM_EXIT_OSI can occur.
4549 6.2 KVM_CAP_PPC_PAPR
4554 Returns: 0 on success; -1 on error
4556 This capability enables interception of PAPR hypercalls. PAPR hypercalls are
4557 done using the hypercall instruction "sc 1".
4559 It also sets the guest privilege level to "supervisor" mode. Usually the guest
4560 runs in "hypervisor" privilege mode with a few missing features.
4562 In addition to the above, it changes the semantics of SDR1. In this mode, the
4563 HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the
4564 HTAB invisible to the guest.
4566 When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur.
4573 Parameters: args[0] is the address of a struct kvm_config_tlb
4574 Returns: 0 on success; -1 on error
4576 struct kvm_config_tlb {
4583 Configures the virtual CPU's TLB array, establishing a shared memory area
4584 between userspace and KVM. The "params" and "array" fields are userspace
4585 addresses of mmu-type-specific data structures. The "array_len" field is an
4586 safety mechanism, and should be set to the size in bytes of the memory that
4587 userspace has reserved for the array. It must be at least the size dictated
4588 by "mmu_type" and "params".
4590 While KVM_RUN is active, the shared region is under control of KVM. Its
4591 contents are undefined, and any modification by userspace results in
4592 boundedly undefined behavior.
4594 On return from KVM_RUN, the shared region will reflect the current state of
4595 the guest's TLB. If userspace makes any changes, it must call KVM_DIRTY_TLB
4596 to tell KVM which entries have been changed, prior to calling KVM_RUN again
4599 For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV:
4600 - The "params" field is of type "struct kvm_book3e_206_tlb_params".
4601 - The "array" field points to an array of type "struct
4602 kvm_book3e_206_tlb_entry".
4603 - The array consists of all entries in the first TLB, followed by all
4604 entries in the second TLB.
4605 - Within a TLB, entries are ordered first by increasing set number. Within a
4606 set, entries are ordered by way (increasing ESEL).
4607 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1)
4608 where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value.
4609 - The tsize field of mas1 shall be set to 4K on TLB0, even though the
4610 hardware ignores this value for TLB0.
4612 6.4 KVM_CAP_S390_CSS_SUPPORT
4617 Returns: 0 on success; -1 on error
4619 This capability enables support for handling of channel I/O instructions.
4621 TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are
4622 handled in-kernel, while the other I/O instructions are passed to userspace.
4624 When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST
4625 SUBCHANNEL intercepts.
4627 Note that even though this capability is enabled per-vcpu, the complete
4628 virtual machine is affected.
4634 Parameters: args[0] defines whether the proxy facility is active
4635 Returns: 0 on success; -1 on error
4637 This capability enables or disables the delivery of interrupts through the
4638 external proxy facility.
4640 When enabled (args[0] != 0), every time the guest gets an external interrupt
4641 delivered, it automatically exits into user space with a KVM_EXIT_EPR exit
4642 to receive the topmost interrupt vector.
4644 When disabled (args[0] == 0), behavior is as if this facility is unsupported.
4646 When this capability is enabled, KVM_EXIT_EPR can occur.
4648 6.6 KVM_CAP_IRQ_MPIC
4651 Parameters: args[0] is the MPIC device fd
4652 args[1] is the MPIC CPU number for this vcpu
4654 This capability connects the vcpu to an in-kernel MPIC device.
4656 6.7 KVM_CAP_IRQ_XICS
4660 Parameters: args[0] is the XICS device fd
4661 args[1] is the XICS CPU number (server ID) for this vcpu
4663 This capability connects the vcpu to an in-kernel XICS device.
4665 6.8 KVM_CAP_S390_IRQCHIP
4671 This capability enables the in-kernel irqchip for s390. Please refer to
4672 "4.24 KVM_CREATE_IRQCHIP" for details.
4674 6.9 KVM_CAP_MIPS_FPU
4678 Parameters: args[0] is reserved for future use (should be 0).
4680 This capability allows the use of the host Floating Point Unit by the guest. It
4681 allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is
4682 done the KVM_REG_MIPS_FPR_* and KVM_REG_MIPS_FCR_* registers can be accessed
4683 (depending on the current guest FPU register mode), and the Status.FR,
4684 Config5.FRE bits are accessible via the KVM API and also from the guest,
4685 depending on them being supported by the FPU.
4687 6.10 KVM_CAP_MIPS_MSA
4691 Parameters: args[0] is reserved for future use (should be 0).
4693 This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest.
4694 It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest.
4695 Once this is done the KVM_REG_MIPS_VEC_* and KVM_REG_MIPS_MSA_* registers can be
4696 accessed, and the Config5.MSAEn bit is accessible via the KVM API and also from
4699 6.74 KVM_CAP_SYNC_REGS
4700 Architectures: s390, x86
4701 Target: s390: always enabled, x86: vcpu
4703 Returns: x86: KVM_CHECK_EXTENSION returns a bit-array indicating which register
4704 sets are supported (bitfields defined in arch/x86/include/uapi/asm/kvm.h).
4706 As described above in the kvm_sync_regs struct info in section 5 (kvm_run):
4707 KVM_CAP_SYNC_REGS "allow[s] userspace to access certain guest registers
4708 without having to call SET/GET_*REGS". This reduces overhead by eliminating
4709 repeated ioctl calls for setting and/or getting register values. This is
4710 particularly important when userspace is making synchronous guest state
4711 modifications, e.g. when emulating and/or intercepting instructions in
4714 For s390 specifics, please refer to the source code.
4717 - the register sets to be copied out to kvm_run are selectable
4718 by userspace (rather that all sets being copied out for every exit).
4719 - vcpu_events are available in addition to regs and sregs.
4721 For x86, the 'kvm_valid_regs' field of struct kvm_run is overloaded to
4722 function as an input bit-array field set by userspace to indicate the
4723 specific register sets to be copied out on the next exit.
4725 To indicate when userspace has modified values that should be copied into
4726 the vCPU, the all architecture bitarray field, 'kvm_dirty_regs' must be set.
4727 This is done using the same bitflags as for the 'kvm_valid_regs' field.
4728 If the dirty bit is not set, then the register set values will not be copied
4729 into the vCPU even if they've been modified.
4731 Unused bitfields in the bitarrays must be set to zero.
4733 struct kvm_sync_regs {
4734 struct kvm_regs regs;
4735 struct kvm_sregs sregs;
4736 struct kvm_vcpu_events events;
4739 6.75 KVM_CAP_PPC_IRQ_XIVE
4743 Parameters: args[0] is the XIVE device fd
4744 args[1] is the XIVE CPU number (server ID) for this vcpu
4746 This capability connects the vcpu to an in-kernel XIVE device.
4748 7. Capabilities that can be enabled on VMs
4749 ------------------------------------------
4751 There are certain capabilities that change the behavior of the virtual
4752 machine when enabled. To enable them, please see section 4.37. Below
4753 you can find a list of capabilities and what their effect on the VM
4754 is when enabling them.
4756 The following information is provided along with the description:
4758 Architectures: which instruction set architectures provide this ioctl.
4759 x86 includes both i386 and x86_64.
4761 Parameters: what parameters are accepted by the capability.
4763 Returns: the return value. General error numbers (EBADF, ENOMEM, EINVAL)
4764 are not detailed, but errors with specific meanings are.
4767 7.1 KVM_CAP_PPC_ENABLE_HCALL
4770 Parameters: args[0] is the sPAPR hcall number
4771 args[1] is 0 to disable, 1 to enable in-kernel handling
4773 This capability controls whether individual sPAPR hypercalls (hcalls)
4774 get handled by the kernel or not. Enabling or disabling in-kernel
4775 handling of an hcall is effective across the VM. On creation, an
4776 initial set of hcalls are enabled for in-kernel handling, which
4777 consists of those hcalls for which in-kernel handlers were implemented
4778 before this capability was implemented. If disabled, the kernel will
4779 not to attempt to handle the hcall, but will always exit to userspace
4780 to handle it. Note that it may not make sense to enable some and
4781 disable others of a group of related hcalls, but KVM does not prevent
4782 userspace from doing that.
4784 If the hcall number specified is not one that has an in-kernel
4785 implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL
4788 7.2 KVM_CAP_S390_USER_SIGP
4793 This capability controls which SIGP orders will be handled completely in user
4794 space. With this capability enabled, all fast orders will be handled completely
4800 - CONDITIONAL EMERGENCY SIGNAL
4802 All other orders will be handled completely in user space.
4804 Only privileged operation exceptions will be checked for in the kernel (or even
4805 in the hardware prior to interception). If this capability is not enabled, the
4806 old way of handling SIGP orders is used (partially in kernel and user space).
4808 7.3 KVM_CAP_S390_VECTOR_REGISTERS
4812 Returns: 0 on success, negative value on error
4814 Allows use of the vector registers introduced with z13 processor, and
4815 provides for the synchronization between host and user space. Will
4816 return -EINVAL if the machine does not support vectors.
4818 7.4 KVM_CAP_S390_USER_STSI
4823 This capability allows post-handlers for the STSI instruction. After
4824 initial handling in the kernel, KVM exits to user space with
4825 KVM_EXIT_S390_STSI to allow user space to insert further data.
4827 Before exiting to userspace, kvm handlers should fill in s390_stsi field of
4838 @addr - guest address of STSI SYSIB
4842 @ar - access register number
4844 KVM handlers should exit to userspace with rc = -EREMOTE.
4846 7.5 KVM_CAP_SPLIT_IRQCHIP
4849 Parameters: args[0] - number of routes reserved for userspace IOAPICs
4850 Returns: 0 on success, -1 on error
4852 Create a local apic for each processor in the kernel. This can be used
4853 instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the
4854 IOAPIC and PIC (and also the PIT, even though this has to be enabled
4857 This capability also enables in kernel routing of interrupt requests;
4858 when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are
4859 used in the IRQ routing table. The first args[0] MSI routes are reserved
4860 for the IOAPIC pins. Whenever the LAPIC receives an EOI for these routes,
4861 a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace.
4863 Fails if VCPU has already been created, or if the irqchip is already in the
4864 kernel (i.e. KVM_CREATE_IRQCHIP has already been called).
4871 Allows use of runtime-instrumentation introduced with zEC12 processor.
4872 Will return -EINVAL if the machine does not support runtime-instrumentation.
4873 Will return -EBUSY if a VCPU has already been created.
4875 7.7 KVM_CAP_X2APIC_API
4878 Parameters: args[0] - features that should be enabled
4879 Returns: 0 on success, -EINVAL when args[0] contains invalid features
4881 Valid feature flags in args[0] are
4883 #define KVM_X2APIC_API_USE_32BIT_IDS (1ULL << 0)
4884 #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK (1ULL << 1)
4886 Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of
4887 KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC,
4888 allowing the use of 32-bit APIC IDs. See KVM_CAP_X2APIC_API in their
4889 respective sections.
4891 KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work
4892 in logical mode or with more than 255 VCPUs. Otherwise, KVM treats 0xff
4893 as a broadcast even in x2APIC mode in order to support physical x2APIC
4894 without interrupt remapping. This is undesirable in logical mode,
4895 where 0xff represents CPUs 0-7 in cluster 0.
4897 7.8 KVM_CAP_S390_USER_INSTR0
4902 With this capability enabled, all illegal instructions 0x0000 (2 bytes) will
4903 be intercepted and forwarded to user space. User space can use this
4904 mechanism e.g. to realize 2-byte software breakpoints. The kernel will
4905 not inject an operating exception for these instructions, user space has
4906 to take care of that.
4908 This capability can be enabled dynamically even if VCPUs were already
4909 created and are running.
4915 Returns: 0 on success; -EINVAL if the machine does not support
4916 guarded storage; -EBUSY if a VCPU has already been created.
4918 Allows use of guarded storage for the KVM guest.
4920 7.10 KVM_CAP_S390_AIS
4925 Allow use of adapter-interruption suppression.
4926 Returns: 0 on success; -EBUSY if a VCPU has already been created.
4928 7.11 KVM_CAP_PPC_SMT
4931 Parameters: vsmt_mode, flags
4933 Enabling this capability on a VM provides userspace with a way to set
4934 the desired virtual SMT mode (i.e. the number of virtual CPUs per
4935 virtual core). The virtual SMT mode, vsmt_mode, must be a power of 2
4936 between 1 and 8. On POWER8, vsmt_mode must also be no greater than
4937 the number of threads per subcore for the host. Currently flags must
4938 be 0. A successful call to enable this capability will result in
4939 vsmt_mode being returned when the KVM_CAP_PPC_SMT capability is
4940 subsequently queried for the VM. This capability is only supported by
4941 HV KVM, and can only be set before any VCPUs have been created.
4942 The KVM_CAP_PPC_SMT_POSSIBLE capability indicates which virtual SMT
4943 modes are available.
4945 7.12 KVM_CAP_PPC_FWNMI
4950 With this capability a machine check exception in the guest address
4951 space will cause KVM to exit the guest with NMI exit reason. This
4952 enables QEMU to build error log and branch to guest kernel registered
4953 machine check handling routine. Without this capability KVM will
4954 branch to guests' 0x200 interrupt vector.
4956 7.13 KVM_CAP_X86_DISABLE_EXITS
4959 Parameters: args[0] defines which exits are disabled
4960 Returns: 0 on success, -EINVAL when args[0] contains invalid exits
4962 Valid bits in args[0] are
4964 #define KVM_X86_DISABLE_EXITS_MWAIT (1 << 0)
4965 #define KVM_X86_DISABLE_EXITS_HLT (1 << 1)
4966 #define KVM_X86_DISABLE_EXITS_PAUSE (1 << 2)
4967 #define KVM_X86_DISABLE_EXITS_CSTATE (1 << 3)
4969 Enabling this capability on a VM provides userspace with a way to no
4970 longer intercept some instructions for improved latency in some
4971 workloads, and is suggested when vCPUs are associated to dedicated
4972 physical CPUs. More bits can be added in the future; userspace can
4973 just pass the KVM_CHECK_EXTENSION result to KVM_ENABLE_CAP to disable
4976 Do not enable KVM_FEATURE_PV_UNHALT if you disable HLT exits.
4978 7.14 KVM_CAP_S390_HPAGE_1M
4982 Returns: 0 on success, -EINVAL if hpage module parameter was not set
4983 or cmma is enabled, or the VM has the KVM_VM_S390_UCONTROL
4986 With this capability the KVM support for memory backing with 1m pages
4987 through hugetlbfs can be enabled for a VM. After the capability is
4988 enabled, cmma can't be enabled anymore and pfmfi and the storage key
4989 interpretation are disabled. If cmma has already been enabled or the
4990 hpage module parameter is not set to 1, -EINVAL is returned.
4992 While it is generally possible to create a huge page backed VM without
4993 this capability, the VM will not be able to run.
4995 7.15 KVM_CAP_MSR_PLATFORM_INFO
4998 Parameters: args[0] whether feature should be enabled or not
5000 With this capability, a guest may read the MSR_PLATFORM_INFO MSR. Otherwise,
5001 a #GP would be raised when the guest tries to access. Currently, this
5002 capability does not enable write permissions of this MSR for the guest.
5004 7.16 KVM_CAP_PPC_NESTED_HV
5008 Returns: 0 on success, -EINVAL when the implementation doesn't support
5009 nested-HV virtualization.
5011 HV-KVM on POWER9 and later systems allows for "nested-HV"
5012 virtualization, which provides a way for a guest VM to run guests that
5013 can run using the CPU's supervisor mode (privileged non-hypervisor
5014 state). Enabling this capability on a VM depends on the CPU having
5015 the necessary functionality and on the facility being enabled with a
5016 kvm-hv module parameter.
5018 7.17 KVM_CAP_EXCEPTION_PAYLOAD
5021 Parameters: args[0] whether feature should be enabled or not
5023 With this capability enabled, CR2 will not be modified prior to the
5024 emulated VM-exit when L1 intercepts a #PF exception that occurs in
5025 L2. Similarly, for kvm-intel only, DR6 will not be modified prior to
5026 the emulated VM-exit when L1 intercepts a #DB exception that occurs in
5027 L2. As a result, when KVM_GET_VCPU_EVENTS reports a pending #PF (or
5028 #DB) exception for L2, exception.has_payload will be set and the
5029 faulting address (or the new DR6 bits*) will be reported in the
5030 exception_payload field. Similarly, when userspace injects a #PF (or
5031 #DB) into L2 using KVM_SET_VCPU_EVENTS, it is expected to set
5032 exception.has_payload and to put the faulting address (or the new DR6
5033 bits*) in the exception_payload field.
5035 This capability also enables exception.pending in struct
5036 kvm_vcpu_events, which allows userspace to distinguish between pending
5037 and injected exceptions.
5040 * For the new DR6 bits, note that bit 16 is set iff the #DB exception
5043 7.18 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
5045 Architectures: x86, arm, arm64, mips
5046 Parameters: args[0] whether feature should be enabled or not
5048 With this capability enabled, KVM_GET_DIRTY_LOG will not automatically
5049 clear and write-protect all pages that are returned as dirty.
5050 Rather, userspace will have to do this operation separately using
5051 KVM_CLEAR_DIRTY_LOG.
5053 At the cost of a slightly more complicated operation, this provides better
5054 scalability and responsiveness for two reasons. First,
5055 KVM_CLEAR_DIRTY_LOG ioctl can operate on a 64-page granularity rather
5056 than requiring to sync a full memslot; this ensures that KVM does not
5057 take spinlocks for an extended period of time. Second, in some cases a
5058 large amount of time can pass between a call to KVM_GET_DIRTY_LOG and
5059 userspace actually using the data in the page. Pages can be modified
5060 during this time, which is inefficint for both the guest and userspace:
5061 the guest will incur a higher penalty due to write protection faults,
5062 while userspace can see false reports of dirty pages. Manual reprotection
5063 helps reducing this time, improving guest performance and reducing the
5064 number of dirty log false positives.
5066 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 was previously available under the name
5067 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT, but the implementation had bugs that make
5068 it hard or impossible to use it correctly. The availability of
5069 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 signals that those bugs are fixed.
5070 Userspace should not try to use KVM_CAP_MANUAL_DIRTY_LOG_PROTECT.
5072 8. Other capabilities.
5073 ----------------------
5075 This section lists capabilities that give information about other
5076 features of the KVM implementation.
5078 8.1 KVM_CAP_PPC_HWRNG
5082 This capability, if KVM_CHECK_EXTENSION indicates that it is
5083 available, means that that the kernel has an implementation of the
5084 H_RANDOM hypercall backed by a hardware random-number generator.
5085 If present, the kernel H_RANDOM handler can be enabled for guest use
5086 with the KVM_CAP_PPC_ENABLE_HCALL capability.
5088 8.2 KVM_CAP_HYPERV_SYNIC
5091 This capability, if KVM_CHECK_EXTENSION indicates that it is
5092 available, means that that the kernel has an implementation of the
5093 Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is
5094 used to support Windows Hyper-V based guest paravirt drivers(VMBus).
5096 In order to use SynIC, it has to be activated by setting this
5097 capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this
5098 will disable the use of APIC hardware virtualization even if supported
5099 by the CPU, as it's incompatible with SynIC auto-EOI behavior.
5101 8.3 KVM_CAP_PPC_RADIX_MMU
5105 This capability, if KVM_CHECK_EXTENSION indicates that it is
5106 available, means that that the kernel can support guests using the
5107 radix MMU defined in Power ISA V3.00 (as implemented in the POWER9
5110 8.4 KVM_CAP_PPC_HASH_MMU_V3
5114 This capability, if KVM_CHECK_EXTENSION indicates that it is
5115 available, means that that the kernel can support guests using the
5116 hashed page table MMU defined in Power ISA V3.00 (as implemented in
5117 the POWER9 processor), including in-memory segment tables.
5123 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
5124 it is available, means that full hardware assisted virtualization capabilities
5125 of the hardware are available for use through KVM. An appropriate
5126 KVM_VM_MIPS_* type must be passed to KVM_CREATE_VM to create a VM which
5129 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
5130 available, it means that the VM is using full hardware assisted virtualization
5131 capabilities of the hardware. This is useful to check after creating a VM with
5132 KVM_VM_MIPS_DEFAULT.
5134 The value returned by KVM_CHECK_EXTENSION should be compared against known
5135 values (see below). All other values are reserved. This is to allow for the
5136 possibility of other hardware assisted virtualization implementations which
5137 may be incompatible with the MIPS VZ ASE.
5139 0: The trap & emulate implementation is in use to run guest code in user
5140 mode. Guest virtual memory segments are rearranged to fit the guest in the
5141 user mode address space.
5143 1: The MIPS VZ ASE is in use, providing full hardware assisted
5144 virtualization, including standard guest virtual memory segments.
5150 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
5151 it is available, means that the trap & emulate implementation is available to
5152 run guest code in user mode, even if KVM_CAP_MIPS_VZ indicates that hardware
5153 assisted virtualisation is also available. KVM_VM_MIPS_TE (0) must be passed
5154 to KVM_CREATE_VM to create a VM which utilises it.
5156 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
5157 available, it means that the VM is using trap & emulate.
5159 8.7 KVM_CAP_MIPS_64BIT
5163 This capability indicates the supported architecture type of the guest, i.e. the
5164 supported register and address width.
5166 The values returned when this capability is checked by KVM_CHECK_EXTENSION on a
5167 kvm VM handle correspond roughly to the CP0_Config.AT register field, and should
5168 be checked specifically against known values (see below). All other values are
5171 0: MIPS32 or microMIPS32.
5172 Both registers and addresses are 32-bits wide.
5173 It will only be possible to run 32-bit guest code.
5175 1: MIPS64 or microMIPS64 with access only to 32-bit compatibility segments.
5176 Registers are 64-bits wide, but addresses are 32-bits wide.
5177 64-bit guest code may run but cannot access MIPS64 memory segments.
5178 It will also be possible to run 32-bit guest code.
5180 2: MIPS64 or microMIPS64 with access to all address segments.
5181 Both registers and addresses are 64-bits wide.
5182 It will be possible to run 64-bit or 32-bit guest code.
5184 8.9 KVM_CAP_ARM_USER_IRQ
5186 Architectures: arm, arm64
5187 This capability, if KVM_CHECK_EXTENSION indicates that it is available, means
5188 that if userspace creates a VM without an in-kernel interrupt controller, it
5189 will be notified of changes to the output level of in-kernel emulated devices,
5190 which can generate virtual interrupts, presented to the VM.
5191 For such VMs, on every return to userspace, the kernel
5192 updates the vcpu's run->s.regs.device_irq_level field to represent the actual
5193 output level of the device.
5195 Whenever kvm detects a change in the device output level, kvm guarantees at
5196 least one return to userspace before running the VM. This exit could either
5197 be a KVM_EXIT_INTR or any other exit event, like KVM_EXIT_MMIO. This way,
5198 userspace can always sample the device output level and re-compute the state of
5199 the userspace interrupt controller. Userspace should always check the state
5200 of run->s.regs.device_irq_level on every kvm exit.
5201 The value in run->s.regs.device_irq_level can represent both level and edge
5202 triggered interrupt signals, depending on the device. Edge triggered interrupt
5203 signals will exit to userspace with the bit in run->s.regs.device_irq_level
5204 set exactly once per edge signal.
5206 The field run->s.regs.device_irq_level is available independent of
5207 run->kvm_valid_regs or run->kvm_dirty_regs bits.
5209 If KVM_CAP_ARM_USER_IRQ is supported, the KVM_CHECK_EXTENSION ioctl returns a
5210 number larger than 0 indicating the version of this capability is implemented
5211 and thereby which bits in in run->s.regs.device_irq_level can signal values.
5213 Currently the following bits are defined for the device_irq_level bitmap:
5215 KVM_CAP_ARM_USER_IRQ >= 1:
5217 KVM_ARM_DEV_EL1_VTIMER - EL1 virtual timer
5218 KVM_ARM_DEV_EL1_PTIMER - EL1 physical timer
5219 KVM_ARM_DEV_PMU - ARM PMU overflow interrupt signal
5221 Future versions of kvm may implement additional events. These will get
5222 indicated by returning a higher number from KVM_CHECK_EXTENSION and will be
5225 8.10 KVM_CAP_PPC_SMT_POSSIBLE
5229 Querying this capability returns a bitmap indicating the possible
5230 virtual SMT modes that can be set using KVM_CAP_PPC_SMT. If bit N
5231 (counting from the right) is set, then a virtual SMT mode of 2^N is
5234 8.11 KVM_CAP_HYPERV_SYNIC2
5238 This capability enables a newer version of Hyper-V Synthetic interrupt
5239 controller (SynIC). The only difference with KVM_CAP_HYPERV_SYNIC is that KVM
5240 doesn't clear SynIC message and event flags pages when they are enabled by
5241 writing to the respective MSRs.
5243 8.12 KVM_CAP_HYPERV_VP_INDEX
5247 This capability indicates that userspace can load HV_X64_MSR_VP_INDEX msr. Its
5248 value is used to denote the target vcpu for a SynIC interrupt. For
5249 compatibilty, KVM initializes this msr to KVM's internal vcpu index. When this
5250 capability is absent, userspace can still query this msr's value.
5252 8.13 KVM_CAP_S390_AIS_MIGRATION
5257 This capability indicates if the flic device will be able to get/set the
5258 AIS states for migration via the KVM_DEV_FLIC_AISM_ALL attribute and allows
5259 to discover this without having to create a flic device.
5261 8.14 KVM_CAP_S390_PSW
5265 This capability indicates that the PSW is exposed via the kvm_run structure.
5267 8.15 KVM_CAP_S390_GMAP
5271 This capability indicates that the user space memory used as guest mapping can
5272 be anywhere in the user memory address space, as long as the memory slots are
5273 aligned and sized to a segment (1MB) boundary.
5275 8.16 KVM_CAP_S390_COW
5279 This capability indicates that the user space memory used as guest mapping can
5280 use copy-on-write semantics as well as dirty pages tracking via read-only page
5283 8.17 KVM_CAP_S390_BPB
5287 This capability indicates that kvm will implement the interfaces to handle
5288 reset, migration and nested KVM for branch prediction blocking. The stfle
5289 facility 82 should not be provided to the guest without this capability.
5291 8.18 KVM_CAP_HYPERV_TLBFLUSH
5295 This capability indicates that KVM supports paravirtualized Hyper-V TLB Flush
5297 HvFlushVirtualAddressSpace, HvFlushVirtualAddressSpaceEx,
5298 HvFlushVirtualAddressList, HvFlushVirtualAddressListEx.
5300 8.19 KVM_CAP_ARM_INJECT_SERROR_ESR
5302 Architectures: arm, arm64
5304 This capability indicates that userspace can specify (via the
5305 KVM_SET_VCPU_EVENTS ioctl) the syndrome value reported to the guest when it
5306 takes a virtual SError interrupt exception.
5307 If KVM advertises this capability, userspace can only specify the ISS field for
5308 the ESR syndrome. Other parts of the ESR, such as the EC are generated by the
5309 CPU when the exception is taken. If this virtual SError is taken to EL1 using
5310 AArch64, this value will be reported in the ISS field of ESR_ELx.
5312 See KVM_CAP_VCPU_EVENTS for more details.
5313 8.20 KVM_CAP_HYPERV_SEND_IPI
5317 This capability indicates that KVM supports paravirtualized Hyper-V IPI send
5319 HvCallSendSyntheticClusterIpi, HvCallSendSyntheticClusterIpiEx.
5320 8.21 KVM_CAP_HYPERV_DIRECT_TLBFLUSH
5324 This capability indicates that KVM running on top of Hyper-V hypervisor
5325 enables Direct TLB flush for its guests meaning that TLB flush
5326 hypercalls are handled by Level 0 hypervisor (Hyper-V) bypassing KVM.
5327 Due to the different ABI for hypercall parameters between Hyper-V and
5328 KVM, enabling this capability effectively disables all hypercall
5329 handling by KVM (as some KVM hypercall may be mistakenly treated as TLB
5330 flush hypercalls by Hyper-V) so userspace should disable KVM identification
5331 in CPUID and only exposes Hyper-V identification. In this case, guest
5332 thinks it's running on Hyper-V and only use Hyper-V hypercalls.