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 Only run VM ioctls from the same process (address space) that was used
21 - vcpu ioctls: These query and set attributes that control the operation
22 of a single virtual cpu.
24 Only run vcpu ioctls from the same thread that was used to create the
27 - device ioctls: These query and set attributes that control the operation
30 device ioctls must be issued from the same process (address space) that
31 was used to create the VM.
36 The kvm API is centered around file descriptors. An initial
37 open("/dev/kvm") obtains a handle to the kvm subsystem; this handle
38 can be used to issue system ioctls. A KVM_CREATE_VM ioctl on this
39 handle will create a VM file descriptor which can be used to issue VM
40 ioctls. A KVM_CREATE_VCPU or KVM_CREATE_DEVICE ioctl on a VM fd will
41 create a virtual cpu or device and return a file descriptor pointing to
42 the new resource. Finally, ioctls on a vcpu or device fd can be used
43 to control the vcpu or device. For vcpus, this includes the important
44 task of actually running guest code.
46 In general file descriptors can be migrated among processes by means
47 of fork() and the SCM_RIGHTS facility of unix domain socket. These
48 kinds of tricks are explicitly not supported by kvm. While they will
49 not cause harm to the host, their actual behavior is not guaranteed by
50 the API. The only supported use is one virtual machine per process,
51 and one vcpu per thread.
57 As of Linux 2.6.22, the KVM ABI has been stabilized: no backward
58 incompatible change are allowed. However, there is an extension
59 facility that allows backward-compatible extensions to the API to be
62 The extension mechanism is not based on the Linux version number.
63 Instead, kvm defines extension identifiers and a facility to query
64 whether a particular extension identifier is available. If it is, a
65 set of ioctls is available for application use.
71 This section describes ioctls that can be used to control kvm guests.
72 For each ioctl, the following information is provided along with a
75 Capability: which KVM extension provides this ioctl. Can be 'basic',
76 which means that is will be provided by any kernel that supports
77 API version 12 (see section 4.1), a KVM_CAP_xyz constant, which
78 means availability needs to be checked with KVM_CHECK_EXTENSION
79 (see section 4.4), or 'none' which means that while not all kernels
80 support this ioctl, there's no capability bit to check its
81 availability: for kernels that don't support the ioctl,
82 the ioctl returns -ENOTTY.
84 Architectures: which instruction set architectures provide this ioctl.
85 x86 includes both i386 and x86_64.
87 Type: system, vm, or vcpu.
89 Parameters: what parameters are accepted by the ioctl.
91 Returns: the return value. General error numbers (EBADF, ENOMEM, EINVAL)
92 are not detailed, but errors with specific meanings are.
95 4.1 KVM_GET_API_VERSION
101 Returns: the constant KVM_API_VERSION (=12)
103 This identifies the API version as the stable kvm API. It is not
104 expected that this number will change. However, Linux 2.6.20 and
105 2.6.21 report earlier versions; these are not documented and not
106 supported. Applications should refuse to run if KVM_GET_API_VERSION
107 returns a value other than 12. If this check passes, all ioctls
108 described as 'basic' will be available.
116 Parameters: machine type identifier (KVM_VM_*)
117 Returns: a VM fd that can be used to control the new virtual machine.
119 The new VM has no virtual cpus and no memory.
120 You probably want to use 0 as machine type.
122 In order to create user controlled virtual machines on S390, check
123 KVM_CAP_S390_UCONTROL and use the flag KVM_VM_S390_UCONTROL as
124 privileged user (CAP_SYS_ADMIN).
126 To use hardware assisted virtualization on MIPS (VZ ASE) rather than
127 the default trap & emulate implementation (which changes the virtual
128 memory layout to fit in user mode), check KVM_CAP_MIPS_VZ and use the
132 4.3 KVM_GET_MSR_INDEX_LIST, KVM_GET_MSR_FEATURE_INDEX_LIST
134 Capability: basic, KVM_CAP_GET_MSR_FEATURES for KVM_GET_MSR_FEATURE_INDEX_LIST
137 Parameters: struct kvm_msr_list (in/out)
138 Returns: 0 on success; -1 on error
140 EFAULT: the msr index list cannot be read from or written to
141 E2BIG: the msr index list is to be to fit in the array specified by
144 struct kvm_msr_list {
145 __u32 nmsrs; /* number of msrs in entries */
149 The user fills in the size of the indices array in nmsrs, and in return
150 kvm adjusts nmsrs to reflect the actual number of msrs and fills in the
151 indices array with their numbers.
153 KVM_GET_MSR_INDEX_LIST returns the guest msrs that are supported. The list
154 varies by kvm version and host processor, but does not change otherwise.
156 Note: if kvm indicates supports MCE (KVM_CAP_MCE), then the MCE bank MSRs are
157 not returned in the MSR list, as different vcpus can have a different number
158 of banks, as set via the KVM_X86_SETUP_MCE ioctl.
160 KVM_GET_MSR_FEATURE_INDEX_LIST returns the list of MSRs that can be passed
161 to the KVM_GET_MSRS system ioctl. This lets userspace probe host capabilities
162 and processor features that are exposed via MSRs (e.g., VMX capabilities).
163 This list also varies by kvm version and host processor, but does not change
167 4.4 KVM_CHECK_EXTENSION
169 Capability: basic, KVM_CAP_CHECK_EXTENSION_VM for vm ioctl
171 Type: system ioctl, vm ioctl
172 Parameters: extension identifier (KVM_CAP_*)
173 Returns: 0 if unsupported; 1 (or some other positive integer) if supported
175 The API allows the application to query about extensions to the core
176 kvm API. Userspace passes an extension identifier (an integer) and
177 receives an integer that describes the extension availability.
178 Generally 0 means no and 1 means yes, but some extensions may report
179 additional information in the integer return value.
181 Based on their initialization different VMs may have different capabilities.
182 It is thus encouraged to use the vm ioctl to query for capabilities (available
183 with KVM_CAP_CHECK_EXTENSION_VM on the vm fd)
185 4.5 KVM_GET_VCPU_MMAP_SIZE
191 Returns: size of vcpu mmap area, in bytes
193 The KVM_RUN ioctl (cf.) communicates with userspace via a shared
194 memory region. This ioctl returns the size of that region. See the
195 KVM_RUN documentation for details.
198 4.6 KVM_SET_MEMORY_REGION
203 Parameters: struct kvm_memory_region (in)
204 Returns: 0 on success, -1 on error
206 This ioctl is obsolete and has been removed.
214 Parameters: vcpu id (apic id on x86)
215 Returns: vcpu fd on success, -1 on error
217 This API adds a vcpu to a virtual machine. No more than max_vcpus may be added.
218 The vcpu id is an integer in the range [0, max_vcpu_id).
220 The recommended max_vcpus value can be retrieved using the KVM_CAP_NR_VCPUS of
221 the KVM_CHECK_EXTENSION ioctl() at run-time.
222 The maximum possible value for max_vcpus can be retrieved using the
223 KVM_CAP_MAX_VCPUS of the KVM_CHECK_EXTENSION ioctl() at run-time.
225 If the KVM_CAP_NR_VCPUS does not exist, you should assume that max_vcpus is 4
227 If the KVM_CAP_MAX_VCPUS does not exist, you should assume that max_vcpus is
228 same as the value returned from KVM_CAP_NR_VCPUS.
230 The maximum possible value for max_vcpu_id can be retrieved using the
231 KVM_CAP_MAX_VCPU_ID of the KVM_CHECK_EXTENSION ioctl() at run-time.
233 If the KVM_CAP_MAX_VCPU_ID does not exist, you should assume that max_vcpu_id
234 is the same as the value returned from KVM_CAP_MAX_VCPUS.
236 On powerpc using book3s_hv mode, the vcpus are mapped onto virtual
237 threads in one or more virtual CPU cores. (This is because the
238 hardware requires all the hardware threads in a CPU core to be in the
239 same partition.) The KVM_CAP_PPC_SMT capability indicates the number
240 of vcpus per virtual core (vcore). The vcore id is obtained by
241 dividing the vcpu id by the number of vcpus per vcore. The vcpus in a
242 given vcore will always be in the same physical core as each other
243 (though that might be a different physical core from time to time).
244 Userspace can control the threading (SMT) mode of the guest by its
245 allocation of vcpu ids. For example, if userspace wants
246 single-threaded guest vcpus, it should make all vcpu ids be a multiple
247 of the number of vcpus per vcore.
249 For virtual cpus that have been created with S390 user controlled virtual
250 machines, the resulting vcpu fd can be memory mapped at page offset
251 KVM_S390_SIE_PAGE_OFFSET in order to obtain a memory map of the virtual
252 cpu's hardware control block.
255 4.8 KVM_GET_DIRTY_LOG (vm ioctl)
260 Parameters: struct kvm_dirty_log (in/out)
261 Returns: 0 on success, -1 on error
263 /* for KVM_GET_DIRTY_LOG */
264 struct kvm_dirty_log {
268 void __user *dirty_bitmap; /* one bit per page */
273 Given a memory slot, return a bitmap containing any pages dirtied
274 since the last call to this ioctl. Bit 0 is the first page in the
275 memory slot. Ensure the entire structure is cleared to avoid padding
278 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 specifies
279 the address space for which you want to return the dirty bitmap.
280 They must be less than the value that KVM_CHECK_EXTENSION returns for
281 the KVM_CAP_MULTI_ADDRESS_SPACE capability.
284 4.9 KVM_SET_MEMORY_ALIAS
289 Parameters: struct kvm_memory_alias (in)
290 Returns: 0 (success), -1 (error)
292 This ioctl is obsolete and has been removed.
301 Returns: 0 on success, -1 on error
303 EINTR: an unmasked signal is pending
305 This ioctl is used to run a guest virtual cpu. While there are no
306 explicit parameters, there is an implicit parameter block that can be
307 obtained by mmap()ing the vcpu fd at offset 0, with the size given by
308 KVM_GET_VCPU_MMAP_SIZE. The parameter block is formatted as a 'struct
309 kvm_run' (see below).
315 Architectures: all except ARM, arm64
317 Parameters: struct kvm_regs (out)
318 Returns: 0 on success, -1 on error
320 Reads the general purpose registers from the vcpu.
324 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
325 __u64 rax, rbx, rcx, rdx;
326 __u64 rsi, rdi, rsp, rbp;
327 __u64 r8, r9, r10, r11;
328 __u64 r12, r13, r14, r15;
334 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
345 Architectures: all except ARM, arm64
347 Parameters: struct kvm_regs (in)
348 Returns: 0 on success, -1 on error
350 Writes the general purpose registers into the vcpu.
352 See KVM_GET_REGS for the data structure.
358 Architectures: x86, ppc
360 Parameters: struct kvm_sregs (out)
361 Returns: 0 on success, -1 on error
363 Reads special registers from the vcpu.
367 struct kvm_segment cs, ds, es, fs, gs, ss;
368 struct kvm_segment tr, ldt;
369 struct kvm_dtable gdt, idt;
370 __u64 cr0, cr2, cr3, cr4, cr8;
373 __u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64];
376 /* ppc -- see arch/powerpc/include/uapi/asm/kvm.h */
378 interrupt_bitmap is a bitmap of pending external interrupts. At most
379 one bit may be set. This interrupt has been acknowledged by the APIC
380 but not yet injected into the cpu core.
386 Architectures: x86, ppc
388 Parameters: struct kvm_sregs (in)
389 Returns: 0 on success, -1 on error
391 Writes special registers into the vcpu. See KVM_GET_SREGS for the
400 Parameters: struct kvm_translation (in/out)
401 Returns: 0 on success, -1 on error
403 Translates a virtual address according to the vcpu's current address
406 struct kvm_translation {
408 __u64 linear_address;
411 __u64 physical_address;
422 Architectures: x86, ppc, mips
424 Parameters: struct kvm_interrupt (in)
425 Returns: 0 on success, negative on failure.
427 Queues a hardware interrupt vector to be injected.
429 /* for KVM_INTERRUPT */
430 struct kvm_interrupt {
437 Returns: 0 on success,
438 -EEXIST if an interrupt is already enqueued
439 -EINVAL the the irq number is invalid
440 -ENXIO if the PIC is in the kernel
441 -EFAULT if the pointer is invalid
443 Note 'irq' is an interrupt vector, not an interrupt pin or line. This
444 ioctl is useful if the in-kernel PIC is not used.
448 Queues an external interrupt to be injected. This ioctl is overleaded
449 with 3 different irq values:
453 This injects an edge type external interrupt into the guest once it's ready
454 to receive interrupts. When injected, the interrupt is done.
456 b) KVM_INTERRUPT_UNSET
458 This unsets any pending interrupt.
460 Only available with KVM_CAP_PPC_UNSET_IRQ.
462 c) KVM_INTERRUPT_SET_LEVEL
464 This injects a level type external interrupt into the guest context. The
465 interrupt stays pending until a specific ioctl with KVM_INTERRUPT_UNSET
468 Only available with KVM_CAP_PPC_IRQ_LEVEL.
470 Note that any value for 'irq' other than the ones stated above is invalid
471 and incurs unexpected behavior.
475 Queues an external interrupt to be injected into the virtual CPU. A negative
476 interrupt number dequeues the interrupt.
487 Support for this has been removed. Use KVM_SET_GUEST_DEBUG instead.
492 Capability: basic (vcpu), KVM_CAP_GET_MSR_FEATURES (system)
494 Type: system ioctl, vcpu ioctl
495 Parameters: struct kvm_msrs (in/out)
496 Returns: number of msrs successfully returned;
499 When used as a system ioctl:
500 Reads the values of MSR-based features that are available for the VM. This
501 is similar to KVM_GET_SUPPORTED_CPUID, but it returns MSR indices and values.
502 The list of msr-based features can be obtained using KVM_GET_MSR_FEATURE_INDEX_LIST
505 When used as a vcpu ioctl:
506 Reads model-specific registers from the vcpu. Supported msr indices can
507 be obtained using KVM_GET_MSR_INDEX_LIST in a system ioctl.
510 __u32 nmsrs; /* number of msrs in entries */
513 struct kvm_msr_entry entries[0];
516 struct kvm_msr_entry {
522 Application code should set the 'nmsrs' member (which indicates the
523 size of the entries array) and the 'index' member of each array entry.
524 kvm will fill in the 'data' member.
532 Parameters: struct kvm_msrs (in)
533 Returns: 0 on success, -1 on error
535 Writes model-specific registers to the vcpu. See KVM_GET_MSRS for the
538 Application code should set the 'nmsrs' member (which indicates the
539 size of the entries array), and the 'index' and 'data' members of each
548 Parameters: struct kvm_cpuid (in)
549 Returns: 0 on success, -1 on error
551 Defines the vcpu responses to the cpuid instruction. Applications
552 should use the KVM_SET_CPUID2 ioctl if available.
555 struct kvm_cpuid_entry {
564 /* for KVM_SET_CPUID */
568 struct kvm_cpuid_entry entries[0];
572 4.21 KVM_SET_SIGNAL_MASK
577 Parameters: struct kvm_signal_mask (in)
578 Returns: 0 on success, -1 on error
580 Defines which signals are blocked during execution of KVM_RUN. This
581 signal mask temporarily overrides the threads signal mask. Any
582 unblocked signal received (except SIGKILL and SIGSTOP, which retain
583 their traditional behaviour) will cause KVM_RUN to return with -EINTR.
585 Note the signal will only be delivered if not blocked by the original
588 /* for KVM_SET_SIGNAL_MASK */
589 struct kvm_signal_mask {
600 Parameters: struct kvm_fpu (out)
601 Returns: 0 on success, -1 on error
603 Reads the floating point state from the vcpu.
605 /* for KVM_GET_FPU and KVM_SET_FPU */
610 __u8 ftwx; /* in fxsave format */
626 Parameters: struct kvm_fpu (in)
627 Returns: 0 on success, -1 on error
629 Writes the floating point state to the vcpu.
631 /* for KVM_GET_FPU and KVM_SET_FPU */
636 __u8 ftwx; /* in fxsave format */
647 4.24 KVM_CREATE_IRQCHIP
649 Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQCHIP (s390)
650 Architectures: x86, ARM, arm64, s390
653 Returns: 0 on success, -1 on error
655 Creates an interrupt controller model in the kernel.
656 On x86, creates a virtual ioapic, a virtual PIC (two PICs, nested), and sets up
657 future vcpus to have a local APIC. IRQ routing for GSIs 0-15 is set to both
658 PIC and IOAPIC; GSI 16-23 only go to the IOAPIC.
659 On ARM/arm64, a GICv2 is created. Any other GIC versions require the usage of
660 KVM_CREATE_DEVICE, which also supports creating a GICv2. Using
661 KVM_CREATE_DEVICE is preferred over KVM_CREATE_IRQCHIP for GICv2.
662 On s390, a dummy irq routing table is created.
664 Note that on s390 the KVM_CAP_S390_IRQCHIP vm capability needs to be enabled
665 before KVM_CREATE_IRQCHIP can be used.
670 Capability: KVM_CAP_IRQCHIP
671 Architectures: x86, arm, arm64
673 Parameters: struct kvm_irq_level
674 Returns: 0 on success, -1 on error
676 Sets the level of a GSI input to the interrupt controller model in the kernel.
677 On some architectures it is required that an interrupt controller model has
678 been previously created with KVM_CREATE_IRQCHIP. Note that edge-triggered
679 interrupts require the level to be set to 1 and then back to 0.
681 On real hardware, interrupt pins can be active-low or active-high. This
682 does not matter for the level field of struct kvm_irq_level: 1 always
683 means active (asserted), 0 means inactive (deasserted).
685 x86 allows the operating system to program the interrupt polarity
686 (active-low/active-high) for level-triggered interrupts, and KVM used
687 to consider the polarity. However, due to bitrot in the handling of
688 active-low interrupts, the above convention is now valid on x86 too.
689 This is signaled by KVM_CAP_X86_IOAPIC_POLARITY_IGNORED. Userspace
690 should not present interrupts to the guest as active-low unless this
691 capability is present (or unless it is not using the in-kernel irqchip,
695 ARM/arm64 can signal an interrupt either at the CPU level, or at the
696 in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to
697 use PPIs designated for specific cpus. The irq field is interpreted
700 Â bits: | 31 ... 24 | 23 ... 16 | 15 ... 0 |
701 field: | irq_type | vcpu_index | irq_id |
703 The irq_type field has the following values:
704 - irq_type[0]: out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ
705 - irq_type[1]: in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.)
706 (the vcpu_index field is ignored)
707 - irq_type[2]: in-kernel GIC: PPI, irq_id between 16 and 31 (incl.)
709 (The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs)
711 In both cases, level is used to assert/deassert the line.
713 struct kvm_irq_level {
716 __s32 status; /* not used for KVM_IRQ_LEVEL */
718 __u32 level; /* 0 or 1 */
724 Capability: KVM_CAP_IRQCHIP
727 Parameters: struct kvm_irqchip (in/out)
728 Returns: 0 on success, -1 on error
730 Reads the state of a kernel interrupt controller created with
731 KVM_CREATE_IRQCHIP into a buffer provided by the caller.
734 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
737 char dummy[512]; /* reserving space */
738 struct kvm_pic_state pic;
739 struct kvm_ioapic_state ioapic;
746 Capability: KVM_CAP_IRQCHIP
749 Parameters: struct kvm_irqchip (in)
750 Returns: 0 on success, -1 on error
752 Sets the state of a kernel interrupt controller created with
753 KVM_CREATE_IRQCHIP from a buffer provided by the caller.
756 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
759 char dummy[512]; /* reserving space */
760 struct kvm_pic_state pic;
761 struct kvm_ioapic_state ioapic;
766 4.28 KVM_XEN_HVM_CONFIG
768 Capability: KVM_CAP_XEN_HVM
771 Parameters: struct kvm_xen_hvm_config (in)
772 Returns: 0 on success, -1 on error
774 Sets the MSR that the Xen HVM guest uses to initialize its hypercall
775 page, and provides the starting address and size of the hypercall
776 blobs in userspace. When the guest writes the MSR, kvm copies one
777 page of a blob (32- or 64-bit, depending on the vcpu mode) to guest
780 struct kvm_xen_hvm_config {
793 Capability: KVM_CAP_ADJUST_CLOCK
796 Parameters: struct kvm_clock_data (out)
797 Returns: 0 on success, -1 on error
799 Gets the current timestamp of kvmclock as seen by the current guest. In
800 conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios
803 When KVM_CAP_ADJUST_CLOCK is passed to KVM_CHECK_EXTENSION, it returns the
804 set of bits that KVM can return in struct kvm_clock_data's flag member.
806 The only flag defined now is KVM_CLOCK_TSC_STABLE. If set, the returned
807 value is the exact kvmclock value seen by all VCPUs at the instant
808 when KVM_GET_CLOCK was called. If clear, the returned value is simply
809 CLOCK_MONOTONIC plus a constant offset; the offset can be modified
810 with KVM_SET_CLOCK. KVM will try to make all VCPUs follow this clock,
811 but the exact value read by each VCPU could differ, because the host
814 struct kvm_clock_data {
815 __u64 clock; /* kvmclock current value */
823 Capability: KVM_CAP_ADJUST_CLOCK
826 Parameters: struct kvm_clock_data (in)
827 Returns: 0 on success, -1 on error
829 Sets the current timestamp of kvmclock to the value specified in its parameter.
830 In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios
833 struct kvm_clock_data {
834 __u64 clock; /* kvmclock current value */
840 4.31 KVM_GET_VCPU_EVENTS
842 Capability: KVM_CAP_VCPU_EVENTS
843 Extended by: KVM_CAP_INTR_SHADOW
844 Architectures: x86, arm, arm64
846 Parameters: struct kvm_vcpu_event (out)
847 Returns: 0 on success, -1 on error
851 Gets currently pending exceptions, interrupts, and NMIs as well as related
854 struct kvm_vcpu_events {
884 Only two fields are defined in the flags field:
886 - KVM_VCPUEVENT_VALID_SHADOW may be set in the flags field to signal that
887 interrupt.shadow contains a valid state.
889 - KVM_VCPUEVENT_VALID_SMM may be set in the flags field to signal that
890 smi contains a valid state.
894 If the guest accesses a device that is being emulated by the host kernel in
895 such a way that a real device would generate a physical SError, KVM may make
896 a virtual SError pending for that VCPU. This system error interrupt remains
897 pending until the guest takes the exception by unmasking PSTATE.A.
899 Running the VCPU may cause it to take a pending SError, or make an access that
900 causes an SError to become pending. The event's description is only valid while
901 the VPCU is not running.
903 This API provides a way to read and write the pending 'event' state that is not
904 visible to the guest. To save, restore or migrate a VCPU the struct representing
905 the state can be read then written using this GET/SET API, along with the other
906 guest-visible registers. It is not possible to 'cancel' an SError that has been
909 A device being emulated in user-space may also wish to generate an SError. To do
910 this the events structure can be populated by user-space. The current state
911 should be read first, to ensure no existing SError is pending. If an existing
912 SError is pending, the architecture's 'Multiple SError interrupts' rules should
913 be followed. (2.5.3 of DDI0587.a "ARM Reliability, Availability, and
914 Serviceability (RAS) Specification").
916 SError exceptions always have an ESR value. Some CPUs have the ability to
917 specify what the virtual SError's ESR value should be. These systems will
918 advertise KVM_CAP_ARM_INJECT_SERROR_ESR. In this case exception.has_esr will
919 always have a non-zero value when read, and the agent making an SError pending
920 should specify the ISS field in the lower 24 bits of exception.serror_esr. If
921 the system supports KVM_CAP_ARM_INJECT_SERROR_ESR, but user-space sets the events
922 with exception.has_esr as zero, KVM will choose an ESR.
924 Specifying exception.has_esr on a system that does not support it will return
925 -EINVAL. Setting anything other than the lower 24bits of exception.serror_esr
928 struct kvm_vcpu_events {
932 /* Align it to 8 bytes */
939 4.32 KVM_SET_VCPU_EVENTS
941 Capability: KVM_CAP_VCPU_EVENTS
942 Extended by: KVM_CAP_INTR_SHADOW
943 Architectures: x86, arm, arm64
945 Parameters: struct kvm_vcpu_event (in)
946 Returns: 0 on success, -1 on error
950 Set pending exceptions, interrupts, and NMIs as well as related states of the
953 See KVM_GET_VCPU_EVENTS for the data structure.
955 Fields that may be modified asynchronously by running VCPUs can be excluded
956 from the update. These fields are nmi.pending, sipi_vector, smi.smm,
957 smi.pending. Keep the corresponding bits in the flags field cleared to
958 suppress overwriting the current in-kernel state. The bits are:
960 KVM_VCPUEVENT_VALID_NMI_PENDING - transfer nmi.pending to the kernel
961 KVM_VCPUEVENT_VALID_SIPI_VECTOR - transfer sipi_vector
962 KVM_VCPUEVENT_VALID_SMM - transfer the smi sub-struct.
964 If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in
965 the flags field to signal that interrupt.shadow contains a valid state and
966 shall be written into the VCPU.
968 KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available.
972 Set the pending SError exception state for this VCPU. It is not possible to
973 'cancel' an Serror that has been made pending.
975 See KVM_GET_VCPU_EVENTS for the data structure.
978 4.33 KVM_GET_DEBUGREGS
980 Capability: KVM_CAP_DEBUGREGS
983 Parameters: struct kvm_debugregs (out)
984 Returns: 0 on success, -1 on error
986 Reads debug registers from the vcpu.
988 struct kvm_debugregs {
997 4.34 KVM_SET_DEBUGREGS
999 Capability: KVM_CAP_DEBUGREGS
1002 Parameters: struct kvm_debugregs (in)
1003 Returns: 0 on success, -1 on error
1005 Writes debug registers into the vcpu.
1007 See KVM_GET_DEBUGREGS for the data structure. The flags field is unused
1008 yet and must be cleared on entry.
1011 4.35 KVM_SET_USER_MEMORY_REGION
1013 Capability: KVM_CAP_USER_MEM
1016 Parameters: struct kvm_userspace_memory_region (in)
1017 Returns: 0 on success, -1 on error
1019 struct kvm_userspace_memory_region {
1022 __u64 guest_phys_addr;
1023 __u64 memory_size; /* bytes */
1024 __u64 userspace_addr; /* start of the userspace allocated memory */
1027 /* for kvm_memory_region::flags */
1028 #define KVM_MEM_LOG_DIRTY_PAGES (1UL << 0)
1029 #define KVM_MEM_READONLY (1UL << 1)
1031 This ioctl allows the user to create or modify a guest physical memory
1032 slot. When changing an existing slot, it may be moved in the guest
1033 physical memory space, or its flags may be modified. It may not be
1034 resized. Slots may not overlap in guest physical address space.
1035 Bits 0-15 of "slot" specifies the slot id and this value should be
1036 less than the maximum number of user memory slots supported per VM.
1037 The maximum allowed slots can be queried using KVM_CAP_NR_MEMSLOTS,
1038 if this capability is supported by the architecture.
1040 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot"
1041 specifies the address space which is being modified. They must be
1042 less than the value that KVM_CHECK_EXTENSION returns for the
1043 KVM_CAP_MULTI_ADDRESS_SPACE capability. Slots in separate address spaces
1044 are unrelated; the restriction on overlapping slots only applies within
1047 Memory for the region is taken starting at the address denoted by the
1048 field userspace_addr, which must point at user addressable memory for
1049 the entire memory slot size. Any object may back this memory, including
1050 anonymous memory, ordinary files, and hugetlbfs.
1052 It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr
1053 be identical. This allows large pages in the guest to be backed by large
1056 The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and
1057 KVM_MEM_READONLY. The former can be set to instruct KVM to keep track of
1058 writes to memory within the slot. See KVM_GET_DIRTY_LOG ioctl to know how to
1059 use it. The latter can be set, if KVM_CAP_READONLY_MEM capability allows it,
1060 to make a new slot read-only. In this case, writes to this memory will be
1061 posted to userspace as KVM_EXIT_MMIO exits.
1063 When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of
1064 the memory region are automatically reflected into the guest. For example, an
1065 mmap() that affects the region will be made visible immediately. Another
1066 example is madvise(MADV_DROP).
1068 It is recommended to use this API instead of the KVM_SET_MEMORY_REGION ioctl.
1069 The KVM_SET_MEMORY_REGION does not allow fine grained control over memory
1070 allocation and is deprecated.
1073 4.36 KVM_SET_TSS_ADDR
1075 Capability: KVM_CAP_SET_TSS_ADDR
1078 Parameters: unsigned long tss_address (in)
1079 Returns: 0 on success, -1 on error
1081 This ioctl defines the physical address of a three-page region in the guest
1082 physical address space. The region must be within the first 4GB of the
1083 guest physical address space and must not conflict with any memory slot
1084 or any mmio address. The guest may malfunction if it accesses this memory
1087 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1088 because of a quirk in the virtualization implementation (see the internals
1089 documentation when it pops into existence).
1094 Capability: KVM_CAP_ENABLE_CAP, KVM_CAP_ENABLE_CAP_VM
1095 Architectures: x86 (only KVM_CAP_ENABLE_CAP_VM),
1096 mips (only KVM_CAP_ENABLE_CAP), ppc, s390
1097 Type: vcpu ioctl, vm ioctl (with KVM_CAP_ENABLE_CAP_VM)
1098 Parameters: struct kvm_enable_cap (in)
1099 Returns: 0 on success; -1 on error
1101 +Not all extensions are enabled by default. Using this ioctl the application
1102 can enable an extension, making it available to the guest.
1104 On systems that do not support this ioctl, it always fails. On systems that
1105 do support it, it only works for extensions that are supported for enablement.
1107 To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should
1110 struct kvm_enable_cap {
1114 The capability that is supposed to get enabled.
1118 A bitfield indicating future enhancements. Has to be 0 for now.
1122 Arguments for enabling a feature. If a feature needs initial values to
1123 function properly, this is the place to put them.
1128 The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl
1129 for vm-wide capabilities.
1131 4.38 KVM_GET_MP_STATE
1133 Capability: KVM_CAP_MP_STATE
1134 Architectures: x86, s390, arm, arm64
1136 Parameters: struct kvm_mp_state (out)
1137 Returns: 0 on success; -1 on error
1139 struct kvm_mp_state {
1143 Returns the vcpu's current "multiprocessing state" (though also valid on
1144 uniprocessor guests).
1146 Possible values are:
1148 - KVM_MP_STATE_RUNNABLE: the vcpu is currently running [x86,arm/arm64]
1149 - KVM_MP_STATE_UNINITIALIZED: the vcpu is an application processor (AP)
1150 which has not yet received an INIT signal [x86]
1151 - KVM_MP_STATE_INIT_RECEIVED: the vcpu has received an INIT signal, and is
1152 now ready for a SIPI [x86]
1153 - KVM_MP_STATE_HALTED: the vcpu has executed a HLT instruction and
1154 is waiting for an interrupt [x86]
1155 - KVM_MP_STATE_SIPI_RECEIVED: the vcpu has just received a SIPI (vector
1156 accessible via KVM_GET_VCPU_EVENTS) [x86]
1157 - KVM_MP_STATE_STOPPED: the vcpu is stopped [s390,arm/arm64]
1158 - KVM_MP_STATE_CHECK_STOP: the vcpu is in a special error state [s390]
1159 - KVM_MP_STATE_OPERATING: the vcpu is operating (running or halted)
1161 - KVM_MP_STATE_LOAD: the vcpu is in a special load/startup state
1164 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1165 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1166 these architectures.
1170 The only states that are valid are KVM_MP_STATE_STOPPED and
1171 KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not.
1173 4.39 KVM_SET_MP_STATE
1175 Capability: KVM_CAP_MP_STATE
1176 Architectures: x86, s390, arm, arm64
1178 Parameters: struct kvm_mp_state (in)
1179 Returns: 0 on success; -1 on error
1181 Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for
1184 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1185 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1186 these architectures.
1190 The only states that are valid are KVM_MP_STATE_STOPPED and
1191 KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not.
1193 4.40 KVM_SET_IDENTITY_MAP_ADDR
1195 Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR
1198 Parameters: unsigned long identity (in)
1199 Returns: 0 on success, -1 on error
1201 This ioctl defines the physical address of a one-page region in the guest
1202 physical address space. The region must be within the first 4GB of the
1203 guest physical address space and must not conflict with any memory slot
1204 or any mmio address. The guest may malfunction if it accesses this memory
1207 Setting the address to 0 will result in resetting the address to its default
1210 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1211 because of a quirk in the virtualization implementation (see the internals
1212 documentation when it pops into existence).
1214 Fails if any VCPU has already been created.
1216 4.41 KVM_SET_BOOT_CPU_ID
1218 Capability: KVM_CAP_SET_BOOT_CPU_ID
1221 Parameters: unsigned long vcpu_id
1222 Returns: 0 on success, -1 on error
1224 Define which vcpu is the Bootstrap Processor (BSP). Values are the same
1225 as the vcpu id in KVM_CREATE_VCPU. If this ioctl is not called, the default
1231 Capability: KVM_CAP_XSAVE
1234 Parameters: struct kvm_xsave (out)
1235 Returns: 0 on success, -1 on error
1241 This ioctl would copy current vcpu's xsave struct to the userspace.
1246 Capability: KVM_CAP_XSAVE
1249 Parameters: struct kvm_xsave (in)
1250 Returns: 0 on success, -1 on error
1256 This ioctl would copy userspace's xsave struct to the kernel.
1261 Capability: KVM_CAP_XCRS
1264 Parameters: struct kvm_xcrs (out)
1265 Returns: 0 on success, -1 on error
1276 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1280 This ioctl would copy current vcpu's xcrs to the userspace.
1285 Capability: KVM_CAP_XCRS
1288 Parameters: struct kvm_xcrs (in)
1289 Returns: 0 on success, -1 on error
1300 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1304 This ioctl would set vcpu's xcr to the value userspace specified.
1307 4.46 KVM_GET_SUPPORTED_CPUID
1309 Capability: KVM_CAP_EXT_CPUID
1312 Parameters: struct kvm_cpuid2 (in/out)
1313 Returns: 0 on success, -1 on error
1318 struct kvm_cpuid_entry2 entries[0];
1321 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
1322 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1)
1323 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2)
1325 struct kvm_cpuid_entry2 {
1336 This ioctl returns x86 cpuid features which are supported by both the
1337 hardware and kvm in its default configuration. Userspace can use the
1338 information returned by this ioctl to construct cpuid information (for
1339 KVM_SET_CPUID2) that is consistent with hardware, kernel, and
1340 userspace capabilities, and with user requirements (for example, the
1341 user may wish to constrain cpuid to emulate older hardware, or for
1342 feature consistency across a cluster).
1344 Note that certain capabilities, such as KVM_CAP_X86_DISABLE_EXITS, may
1345 expose cpuid features (e.g. MONITOR) which are not supported by kvm in
1346 its default configuration. If userspace enables such capabilities, it
1347 is responsible for modifying the results of this ioctl appropriately.
1349 Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure
1350 with the 'nent' field indicating the number of entries in the variable-size
1351 array 'entries'. If the number of entries is too low to describe the cpu
1352 capabilities, an error (E2BIG) is returned. If the number is too high,
1353 the 'nent' field is adjusted and an error (ENOMEM) is returned. If the
1354 number is just right, the 'nent' field is adjusted to the number of valid
1355 entries in the 'entries' array, which is then filled.
1357 The entries returned are the host cpuid as returned by the cpuid instruction,
1358 with unknown or unsupported features masked out. Some features (for example,
1359 x2apic), may not be present in the host cpu, but are exposed by kvm if it can
1360 emulate them efficiently. The fields in each entry are defined as follows:
1362 function: the eax value used to obtain the entry
1363 index: the ecx value used to obtain the entry (for entries that are
1365 flags: an OR of zero or more of the following:
1366 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
1367 if the index field is valid
1368 KVM_CPUID_FLAG_STATEFUL_FUNC:
1369 if cpuid for this function returns different values for successive
1370 invocations; there will be several entries with the same function,
1371 all with this flag set
1372 KVM_CPUID_FLAG_STATE_READ_NEXT:
1373 for KVM_CPUID_FLAG_STATEFUL_FUNC entries, set if this entry is
1374 the first entry to be read by a cpu
1375 eax, ebx, ecx, edx: the values returned by the cpuid instruction for
1376 this function/index combination
1378 The TSC deadline timer feature (CPUID leaf 1, ecx[24]) is always returned
1379 as false, since the feature depends on KVM_CREATE_IRQCHIP for local APIC
1380 support. Instead it is reported via
1382 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER)
1384 if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the
1385 feature in userspace, then you can enable the feature for KVM_SET_CPUID2.
1388 4.47 KVM_PPC_GET_PVINFO
1390 Capability: KVM_CAP_PPC_GET_PVINFO
1393 Parameters: struct kvm_ppc_pvinfo (out)
1394 Returns: 0 on success, !0 on error
1396 struct kvm_ppc_pvinfo {
1402 This ioctl fetches PV specific information that need to be passed to the guest
1403 using the device tree or other means from vm context.
1405 The hcall array defines 4 instructions that make up a hypercall.
1407 If any additional field gets added to this structure later on, a bit for that
1408 additional piece of information will be set in the flags bitmap.
1410 The flags bitmap is defined as:
1412 /* the host supports the ePAPR idle hcall
1413 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1<<0)
1415 4.52 KVM_SET_GSI_ROUTING
1417 Capability: KVM_CAP_IRQ_ROUTING
1418 Architectures: x86 s390 arm arm64
1420 Parameters: struct kvm_irq_routing (in)
1421 Returns: 0 on success, -1 on error
1423 Sets the GSI routing table entries, overwriting any previously set entries.
1425 On arm/arm64, GSI routing has the following limitation:
1426 - GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD.
1428 struct kvm_irq_routing {
1431 struct kvm_irq_routing_entry entries[0];
1434 No flags are specified so far, the corresponding field must be set to zero.
1436 struct kvm_irq_routing_entry {
1442 struct kvm_irq_routing_irqchip irqchip;
1443 struct kvm_irq_routing_msi msi;
1444 struct kvm_irq_routing_s390_adapter adapter;
1445 struct kvm_irq_routing_hv_sint hv_sint;
1450 /* gsi routing entry types */
1451 #define KVM_IRQ_ROUTING_IRQCHIP 1
1452 #define KVM_IRQ_ROUTING_MSI 2
1453 #define KVM_IRQ_ROUTING_S390_ADAPTER 3
1454 #define KVM_IRQ_ROUTING_HV_SINT 4
1457 - KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry
1458 type, specifies that the devid field contains a valid value. The per-VM
1459 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
1460 the device ID. If this capability is not available, userspace should
1461 never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
1464 struct kvm_irq_routing_irqchip {
1469 struct kvm_irq_routing_msi {
1479 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
1480 for the device that wrote the MSI message. For PCI, this is usually a
1481 BFD identifier in the lower 16 bits.
1483 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
1484 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
1485 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
1486 address_hi must be zero.
1488 struct kvm_irq_routing_s390_adapter {
1492 __u32 summary_offset;
1496 struct kvm_irq_routing_hv_sint {
1502 4.55 KVM_SET_TSC_KHZ
1504 Capability: KVM_CAP_TSC_CONTROL
1507 Parameters: virtual tsc_khz
1508 Returns: 0 on success, -1 on error
1510 Specifies the tsc frequency for the virtual machine. The unit of the
1514 4.56 KVM_GET_TSC_KHZ
1516 Capability: KVM_CAP_GET_TSC_KHZ
1520 Returns: virtual tsc-khz on success, negative value on error
1522 Returns the tsc frequency of the guest. The unit of the return value is
1523 KHz. If the host has unstable tsc this ioctl returns -EIO instead as an
1529 Capability: KVM_CAP_IRQCHIP
1532 Parameters: struct kvm_lapic_state (out)
1533 Returns: 0 on success, -1 on error
1535 #define KVM_APIC_REG_SIZE 0x400
1536 struct kvm_lapic_state {
1537 char regs[KVM_APIC_REG_SIZE];
1540 Reads the Local APIC registers and copies them into the input argument. The
1541 data format and layout are the same as documented in the architecture manual.
1543 If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is
1544 enabled, then the format of APIC_ID register depends on the APIC mode
1545 (reported by MSR_IA32_APICBASE) of its VCPU. x2APIC stores APIC ID in
1546 the APIC_ID register (bytes 32-35). xAPIC only allows an 8-bit APIC ID
1547 which is stored in bits 31-24 of the APIC register, or equivalently in
1548 byte 35 of struct kvm_lapic_state's regs field. KVM_GET_LAPIC must then
1549 be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR.
1551 If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state
1552 always uses xAPIC format.
1557 Capability: KVM_CAP_IRQCHIP
1560 Parameters: struct kvm_lapic_state (in)
1561 Returns: 0 on success, -1 on error
1563 #define KVM_APIC_REG_SIZE 0x400
1564 struct kvm_lapic_state {
1565 char regs[KVM_APIC_REG_SIZE];
1568 Copies the input argument into the Local APIC registers. The data format
1569 and layout are the same as documented in the architecture manual.
1571 The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's
1572 regs field) depends on the state of the KVM_CAP_X2APIC_API capability.
1573 See the note in KVM_GET_LAPIC.
1578 Capability: KVM_CAP_IOEVENTFD
1581 Parameters: struct kvm_ioeventfd (in)
1582 Returns: 0 on success, !0 on error
1584 This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address
1585 within the guest. A guest write in the registered address will signal the
1586 provided event instead of triggering an exit.
1588 struct kvm_ioeventfd {
1590 __u64 addr; /* legal pio/mmio address */
1591 __u32 len; /* 0, 1, 2, 4, or 8 bytes */
1597 For the special case of virtio-ccw devices on s390, the ioevent is matched
1598 to a subchannel/virtqueue tuple instead.
1600 The following flags are defined:
1602 #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch)
1603 #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio)
1604 #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign)
1605 #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \
1606 (1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify)
1608 If datamatch flag is set, the event will be signaled only if the written value
1609 to the registered address is equal to datamatch in struct kvm_ioeventfd.
1611 For virtio-ccw devices, addr contains the subchannel id and datamatch the
1614 With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and
1615 the kernel will ignore the length of guest write and may get a faster vmexit.
1616 The speedup may only apply to specific architectures, but the ioeventfd will
1621 Capability: KVM_CAP_SW_TLB
1624 Parameters: struct kvm_dirty_tlb (in)
1625 Returns: 0 on success, -1 on error
1627 struct kvm_dirty_tlb {
1632 This must be called whenever userspace has changed an entry in the shared
1633 TLB, prior to calling KVM_RUN on the associated vcpu.
1635 The "bitmap" field is the userspace address of an array. This array
1636 consists of a number of bits, equal to the total number of TLB entries as
1637 determined by the last successful call to KVM_CONFIG_TLB, rounded up to the
1638 nearest multiple of 64.
1640 Each bit corresponds to one TLB entry, ordered the same as in the shared TLB
1643 The array is little-endian: the bit 0 is the least significant bit of the
1644 first byte, bit 8 is the least significant bit of the second byte, etc.
1645 This avoids any complications with differing word sizes.
1647 The "num_dirty" field is a performance hint for KVM to determine whether it
1648 should skip processing the bitmap and just invalidate everything. It must
1649 be set to the number of set bits in the bitmap.
1652 4.62 KVM_CREATE_SPAPR_TCE
1654 Capability: KVM_CAP_SPAPR_TCE
1655 Architectures: powerpc
1657 Parameters: struct kvm_create_spapr_tce (in)
1658 Returns: file descriptor for manipulating the created TCE table
1660 This creates a virtual TCE (translation control entry) table, which
1661 is an IOMMU for PAPR-style virtual I/O. It is used to translate
1662 logical addresses used in virtual I/O into guest physical addresses,
1663 and provides a scatter/gather capability for PAPR virtual I/O.
1665 /* for KVM_CAP_SPAPR_TCE */
1666 struct kvm_create_spapr_tce {
1671 The liobn field gives the logical IO bus number for which to create a
1672 TCE table. The window_size field specifies the size of the DMA window
1673 which this TCE table will translate - the table will contain one 64
1674 bit TCE entry for every 4kiB of the DMA window.
1676 When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE
1677 table has been created using this ioctl(), the kernel will handle it
1678 in real mode, updating the TCE table. H_PUT_TCE calls for other
1679 liobns will cause a vm exit and must be handled by userspace.
1681 The return value is a file descriptor which can be passed to mmap(2)
1682 to map the created TCE table into userspace. This lets userspace read
1683 the entries written by kernel-handled H_PUT_TCE calls, and also lets
1684 userspace update the TCE table directly which is useful in some
1688 4.63 KVM_ALLOCATE_RMA
1690 Capability: KVM_CAP_PPC_RMA
1691 Architectures: powerpc
1693 Parameters: struct kvm_allocate_rma (out)
1694 Returns: file descriptor for mapping the allocated RMA
1696 This allocates a Real Mode Area (RMA) from the pool allocated at boot
1697 time by the kernel. An RMA is a physically-contiguous, aligned region
1698 of memory used on older POWER processors to provide the memory which
1699 will be accessed by real-mode (MMU off) accesses in a KVM guest.
1700 POWER processors support a set of sizes for the RMA that usually
1701 includes 64MB, 128MB, 256MB and some larger powers of two.
1703 /* for KVM_ALLOCATE_RMA */
1704 struct kvm_allocate_rma {
1708 The return value is a file descriptor which can be passed to mmap(2)
1709 to map the allocated RMA into userspace. The mapped area can then be
1710 passed to the KVM_SET_USER_MEMORY_REGION ioctl to establish it as the
1711 RMA for a virtual machine. The size of the RMA in bytes (which is
1712 fixed at host kernel boot time) is returned in the rma_size field of
1713 the argument structure.
1715 The KVM_CAP_PPC_RMA capability is 1 or 2 if the KVM_ALLOCATE_RMA ioctl
1716 is supported; 2 if the processor requires all virtual machines to have
1717 an RMA, or 1 if the processor can use an RMA but doesn't require it,
1718 because it supports the Virtual RMA (VRMA) facility.
1723 Capability: KVM_CAP_USER_NMI
1727 Returns: 0 on success, -1 on error
1729 Queues an NMI on the thread's vcpu. Note this is well defined only
1730 when KVM_CREATE_IRQCHIP has not been called, since this is an interface
1731 between the virtual cpu core and virtual local APIC. After KVM_CREATE_IRQCHIP
1732 has been called, this interface is completely emulated within the kernel.
1734 To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the
1735 following algorithm:
1738 - read the local APIC's state (KVM_GET_LAPIC)
1739 - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1)
1740 - if so, issue KVM_NMI
1743 Some guests configure the LINT1 NMI input to cause a panic, aiding in
1747 4.65 KVM_S390_UCAS_MAP
1749 Capability: KVM_CAP_S390_UCONTROL
1752 Parameters: struct kvm_s390_ucas_mapping (in)
1753 Returns: 0 in case of success
1755 The parameter is defined like this:
1756 struct kvm_s390_ucas_mapping {
1762 This ioctl maps the memory at "user_addr" with the length "length" to
1763 the vcpu's address space starting at "vcpu_addr". All parameters need to
1764 be aligned by 1 megabyte.
1767 4.66 KVM_S390_UCAS_UNMAP
1769 Capability: KVM_CAP_S390_UCONTROL
1772 Parameters: struct kvm_s390_ucas_mapping (in)
1773 Returns: 0 in case of success
1775 The parameter is defined like this:
1776 struct kvm_s390_ucas_mapping {
1782 This ioctl unmaps the memory in the vcpu's address space starting at
1783 "vcpu_addr" with the length "length". The field "user_addr" is ignored.
1784 All parameters need to be aligned by 1 megabyte.
1787 4.67 KVM_S390_VCPU_FAULT
1789 Capability: KVM_CAP_S390_UCONTROL
1792 Parameters: vcpu absolute address (in)
1793 Returns: 0 in case of success
1795 This call creates a page table entry on the virtual cpu's address space
1796 (for user controlled virtual machines) or the virtual machine's address
1797 space (for regular virtual machines). This only works for minor faults,
1798 thus it's recommended to access subject memory page via the user page
1799 table upfront. This is useful to handle validity intercepts for user
1800 controlled virtual machines to fault in the virtual cpu's lowcore pages
1801 prior to calling the KVM_RUN ioctl.
1804 4.68 KVM_SET_ONE_REG
1806 Capability: KVM_CAP_ONE_REG
1809 Parameters: struct kvm_one_reg (in)
1810 Returns: 0 on success, negative value on failure
1812 struct kvm_one_reg {
1817 Using this ioctl, a single vcpu register can be set to a specific value
1818 defined by user space with the passed in struct kvm_one_reg, where id
1819 refers to the register identifier as described below and addr is a pointer
1820 to a variable with the respective size. There can be architecture agnostic
1821 and architecture specific registers. Each have their own range of operation
1822 and their own constants and width. To keep track of the implemented
1823 registers, find a list below:
1825 Arch | Register | Width (bits)
1827 PPC | KVM_REG_PPC_HIOR | 64
1828 PPC | KVM_REG_PPC_IAC1 | 64
1829 PPC | KVM_REG_PPC_IAC2 | 64
1830 PPC | KVM_REG_PPC_IAC3 | 64
1831 PPC | KVM_REG_PPC_IAC4 | 64
1832 PPC | KVM_REG_PPC_DAC1 | 64
1833 PPC | KVM_REG_PPC_DAC2 | 64
1834 PPC | KVM_REG_PPC_DABR | 64
1835 PPC | KVM_REG_PPC_DSCR | 64
1836 PPC | KVM_REG_PPC_PURR | 64
1837 PPC | KVM_REG_PPC_SPURR | 64
1838 PPC | KVM_REG_PPC_DAR | 64
1839 PPC | KVM_REG_PPC_DSISR | 32
1840 PPC | KVM_REG_PPC_AMR | 64
1841 PPC | KVM_REG_PPC_UAMOR | 64
1842 PPC | KVM_REG_PPC_MMCR0 | 64
1843 PPC | KVM_REG_PPC_MMCR1 | 64
1844 PPC | KVM_REG_PPC_MMCRA | 64
1845 PPC | KVM_REG_PPC_MMCR2 | 64
1846 PPC | KVM_REG_PPC_MMCRS | 64
1847 PPC | KVM_REG_PPC_SIAR | 64
1848 PPC | KVM_REG_PPC_SDAR | 64
1849 PPC | KVM_REG_PPC_SIER | 64
1850 PPC | KVM_REG_PPC_PMC1 | 32
1851 PPC | KVM_REG_PPC_PMC2 | 32
1852 PPC | KVM_REG_PPC_PMC3 | 32
1853 PPC | KVM_REG_PPC_PMC4 | 32
1854 PPC | KVM_REG_PPC_PMC5 | 32
1855 PPC | KVM_REG_PPC_PMC6 | 32
1856 PPC | KVM_REG_PPC_PMC7 | 32
1857 PPC | KVM_REG_PPC_PMC8 | 32
1858 PPC | KVM_REG_PPC_FPR0 | 64
1860 PPC | KVM_REG_PPC_FPR31 | 64
1861 PPC | KVM_REG_PPC_VR0 | 128
1863 PPC | KVM_REG_PPC_VR31 | 128
1864 PPC | KVM_REG_PPC_VSR0 | 128
1866 PPC | KVM_REG_PPC_VSR31 | 128
1867 PPC | KVM_REG_PPC_FPSCR | 64
1868 PPC | KVM_REG_PPC_VSCR | 32
1869 PPC | KVM_REG_PPC_VPA_ADDR | 64
1870 PPC | KVM_REG_PPC_VPA_SLB | 128
1871 PPC | KVM_REG_PPC_VPA_DTL | 128
1872 PPC | KVM_REG_PPC_EPCR | 32
1873 PPC | KVM_REG_PPC_EPR | 32
1874 PPC | KVM_REG_PPC_TCR | 32
1875 PPC | KVM_REG_PPC_TSR | 32
1876 PPC | KVM_REG_PPC_OR_TSR | 32
1877 PPC | KVM_REG_PPC_CLEAR_TSR | 32
1878 PPC | KVM_REG_PPC_MAS0 | 32
1879 PPC | KVM_REG_PPC_MAS1 | 32
1880 PPC | KVM_REG_PPC_MAS2 | 64
1881 PPC | KVM_REG_PPC_MAS7_3 | 64
1882 PPC | KVM_REG_PPC_MAS4 | 32
1883 PPC | KVM_REG_PPC_MAS6 | 32
1884 PPC | KVM_REG_PPC_MMUCFG | 32
1885 PPC | KVM_REG_PPC_TLB0CFG | 32
1886 PPC | KVM_REG_PPC_TLB1CFG | 32
1887 PPC | KVM_REG_PPC_TLB2CFG | 32
1888 PPC | KVM_REG_PPC_TLB3CFG | 32
1889 PPC | KVM_REG_PPC_TLB0PS | 32
1890 PPC | KVM_REG_PPC_TLB1PS | 32
1891 PPC | KVM_REG_PPC_TLB2PS | 32
1892 PPC | KVM_REG_PPC_TLB3PS | 32
1893 PPC | KVM_REG_PPC_EPTCFG | 32
1894 PPC | KVM_REG_PPC_ICP_STATE | 64
1895 PPC | KVM_REG_PPC_TB_OFFSET | 64
1896 PPC | KVM_REG_PPC_SPMC1 | 32
1897 PPC | KVM_REG_PPC_SPMC2 | 32
1898 PPC | KVM_REG_PPC_IAMR | 64
1899 PPC | KVM_REG_PPC_TFHAR | 64
1900 PPC | KVM_REG_PPC_TFIAR | 64
1901 PPC | KVM_REG_PPC_TEXASR | 64
1902 PPC | KVM_REG_PPC_FSCR | 64
1903 PPC | KVM_REG_PPC_PSPB | 32
1904 PPC | KVM_REG_PPC_EBBHR | 64
1905 PPC | KVM_REG_PPC_EBBRR | 64
1906 PPC | KVM_REG_PPC_BESCR | 64
1907 PPC | KVM_REG_PPC_TAR | 64
1908 PPC | KVM_REG_PPC_DPDES | 64
1909 PPC | KVM_REG_PPC_DAWR | 64
1910 PPC | KVM_REG_PPC_DAWRX | 64
1911 PPC | KVM_REG_PPC_CIABR | 64
1912 PPC | KVM_REG_PPC_IC | 64
1913 PPC | KVM_REG_PPC_VTB | 64
1914 PPC | KVM_REG_PPC_CSIGR | 64
1915 PPC | KVM_REG_PPC_TACR | 64
1916 PPC | KVM_REG_PPC_TCSCR | 64
1917 PPC | KVM_REG_PPC_PID | 64
1918 PPC | KVM_REG_PPC_ACOP | 64
1919 PPC | KVM_REG_PPC_VRSAVE | 32
1920 PPC | KVM_REG_PPC_LPCR | 32
1921 PPC | KVM_REG_PPC_LPCR_64 | 64
1922 PPC | KVM_REG_PPC_PPR | 64
1923 PPC | KVM_REG_PPC_ARCH_COMPAT | 32
1924 PPC | KVM_REG_PPC_DABRX | 32
1925 PPC | KVM_REG_PPC_WORT | 64
1926 PPC | KVM_REG_PPC_SPRG9 | 64
1927 PPC | KVM_REG_PPC_DBSR | 32
1928 PPC | KVM_REG_PPC_TIDR | 64
1929 PPC | KVM_REG_PPC_PSSCR | 64
1930 PPC | KVM_REG_PPC_DEC_EXPIRY | 64
1931 PPC | KVM_REG_PPC_TM_GPR0 | 64
1933 PPC | KVM_REG_PPC_TM_GPR31 | 64
1934 PPC | KVM_REG_PPC_TM_VSR0 | 128
1936 PPC | KVM_REG_PPC_TM_VSR63 | 128
1937 PPC | KVM_REG_PPC_TM_CR | 64
1938 PPC | KVM_REG_PPC_TM_LR | 64
1939 PPC | KVM_REG_PPC_TM_CTR | 64
1940 PPC | KVM_REG_PPC_TM_FPSCR | 64
1941 PPC | KVM_REG_PPC_TM_AMR | 64
1942 PPC | KVM_REG_PPC_TM_PPR | 64
1943 PPC | KVM_REG_PPC_TM_VRSAVE | 64
1944 PPC | KVM_REG_PPC_TM_VSCR | 32
1945 PPC | KVM_REG_PPC_TM_DSCR | 64
1946 PPC | KVM_REG_PPC_TM_TAR | 64
1947 PPC | KVM_REG_PPC_TM_XER | 64
1949 MIPS | KVM_REG_MIPS_R0 | 64
1951 MIPS | KVM_REG_MIPS_R31 | 64
1952 MIPS | KVM_REG_MIPS_HI | 64
1953 MIPS | KVM_REG_MIPS_LO | 64
1954 MIPS | KVM_REG_MIPS_PC | 64
1955 MIPS | KVM_REG_MIPS_CP0_INDEX | 32
1956 MIPS | KVM_REG_MIPS_CP0_ENTRYLO0 | 64
1957 MIPS | KVM_REG_MIPS_CP0_ENTRYLO1 | 64
1958 MIPS | KVM_REG_MIPS_CP0_CONTEXT | 64
1959 MIPS | KVM_REG_MIPS_CP0_CONTEXTCONFIG| 32
1960 MIPS | KVM_REG_MIPS_CP0_USERLOCAL | 64
1961 MIPS | KVM_REG_MIPS_CP0_XCONTEXTCONFIG| 64
1962 MIPS | KVM_REG_MIPS_CP0_PAGEMASK | 32
1963 MIPS | KVM_REG_MIPS_CP0_PAGEGRAIN | 32
1964 MIPS | KVM_REG_MIPS_CP0_SEGCTL0 | 64
1965 MIPS | KVM_REG_MIPS_CP0_SEGCTL1 | 64
1966 MIPS | KVM_REG_MIPS_CP0_SEGCTL2 | 64
1967 MIPS | KVM_REG_MIPS_CP0_PWBASE | 64
1968 MIPS | KVM_REG_MIPS_CP0_PWFIELD | 64
1969 MIPS | KVM_REG_MIPS_CP0_PWSIZE | 64
1970 MIPS | KVM_REG_MIPS_CP0_WIRED | 32
1971 MIPS | KVM_REG_MIPS_CP0_PWCTL | 32
1972 MIPS | KVM_REG_MIPS_CP0_HWRENA | 32
1973 MIPS | KVM_REG_MIPS_CP0_BADVADDR | 64
1974 MIPS | KVM_REG_MIPS_CP0_BADINSTR | 32
1975 MIPS | KVM_REG_MIPS_CP0_BADINSTRP | 32
1976 MIPS | KVM_REG_MIPS_CP0_COUNT | 32
1977 MIPS | KVM_REG_MIPS_CP0_ENTRYHI | 64
1978 MIPS | KVM_REG_MIPS_CP0_COMPARE | 32
1979 MIPS | KVM_REG_MIPS_CP0_STATUS | 32
1980 MIPS | KVM_REG_MIPS_CP0_INTCTL | 32
1981 MIPS | KVM_REG_MIPS_CP0_CAUSE | 32
1982 MIPS | KVM_REG_MIPS_CP0_EPC | 64
1983 MIPS | KVM_REG_MIPS_CP0_PRID | 32
1984 MIPS | KVM_REG_MIPS_CP0_EBASE | 64
1985 MIPS | KVM_REG_MIPS_CP0_CONFIG | 32
1986 MIPS | KVM_REG_MIPS_CP0_CONFIG1 | 32
1987 MIPS | KVM_REG_MIPS_CP0_CONFIG2 | 32
1988 MIPS | KVM_REG_MIPS_CP0_CONFIG3 | 32
1989 MIPS | KVM_REG_MIPS_CP0_CONFIG4 | 32
1990 MIPS | KVM_REG_MIPS_CP0_CONFIG5 | 32
1991 MIPS | KVM_REG_MIPS_CP0_CONFIG7 | 32
1992 MIPS | KVM_REG_MIPS_CP0_XCONTEXT | 64
1993 MIPS | KVM_REG_MIPS_CP0_ERROREPC | 64
1994 MIPS | KVM_REG_MIPS_CP0_KSCRATCH1 | 64
1995 MIPS | KVM_REG_MIPS_CP0_KSCRATCH2 | 64
1996 MIPS | KVM_REG_MIPS_CP0_KSCRATCH3 | 64
1997 MIPS | KVM_REG_MIPS_CP0_KSCRATCH4 | 64
1998 MIPS | KVM_REG_MIPS_CP0_KSCRATCH5 | 64
1999 MIPS | KVM_REG_MIPS_CP0_KSCRATCH6 | 64
2000 MIPS | KVM_REG_MIPS_CP0_MAAR(0..63) | 64
2001 MIPS | KVM_REG_MIPS_COUNT_CTL | 64
2002 MIPS | KVM_REG_MIPS_COUNT_RESUME | 64
2003 MIPS | KVM_REG_MIPS_COUNT_HZ | 64
2004 MIPS | KVM_REG_MIPS_FPR_32(0..31) | 32
2005 MIPS | KVM_REG_MIPS_FPR_64(0..31) | 64
2006 MIPS | KVM_REG_MIPS_VEC_128(0..31) | 128
2007 MIPS | KVM_REG_MIPS_FCR_IR | 32
2008 MIPS | KVM_REG_MIPS_FCR_CSR | 32
2009 MIPS | KVM_REG_MIPS_MSA_IR | 32
2010 MIPS | KVM_REG_MIPS_MSA_CSR | 32
2012 ARM registers are mapped using the lower 32 bits. The upper 16 of that
2013 is the register group type, or coprocessor number:
2015 ARM core registers have the following id bit patterns:
2016 0x4020 0000 0010 <index into the kvm_regs struct:16>
2018 ARM 32-bit CP15 registers have the following id bit patterns:
2019 0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3>
2021 ARM 64-bit CP15 registers have the following id bit patterns:
2022 0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3>
2024 ARM CCSIDR registers are demultiplexed by CSSELR value:
2025 0x4020 0000 0011 00 <csselr:8>
2027 ARM 32-bit VFP control registers have the following id bit patterns:
2028 0x4020 0000 0012 1 <regno:12>
2030 ARM 64-bit FP registers have the following id bit patterns:
2031 0x4030 0000 0012 0 <regno:12>
2033 ARM firmware pseudo-registers have the following bit pattern:
2034 0x4030 0000 0014 <regno:16>
2037 arm64 registers are mapped using the lower 32 bits. The upper 16 of
2038 that is the register group type, or coprocessor number:
2040 arm64 core/FP-SIMD registers have the following id bit patterns. Note
2041 that the size of the access is variable, as the kvm_regs structure
2042 contains elements ranging from 32 to 128 bits. The index is a 32bit
2043 value in the kvm_regs structure seen as a 32bit array.
2044 0x60x0 0000 0010 <index into the kvm_regs struct:16>
2046 arm64 CCSIDR registers are demultiplexed by CSSELR value:
2047 0x6020 0000 0011 00 <csselr:8>
2049 arm64 system registers have the following id bit patterns:
2050 0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3>
2052 arm64 firmware pseudo-registers have the following bit pattern:
2053 0x6030 0000 0014 <regno:16>
2056 MIPS registers are mapped using the lower 32 bits. The upper 16 of that is
2057 the register group type:
2059 MIPS core registers (see above) have the following id bit patterns:
2060 0x7030 0000 0000 <reg:16>
2062 MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit
2063 patterns depending on whether they're 32-bit or 64-bit registers:
2064 0x7020 0000 0001 00 <reg:5> <sel:3> (32-bit)
2065 0x7030 0000 0001 00 <reg:5> <sel:3> (64-bit)
2067 Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_MIPS_CP0_ENTRYLO1 are the MIPS64
2068 versions of the EntryLo registers regardless of the word size of the host
2069 hardware, host kernel, guest, and whether XPA is present in the guest, i.e.
2070 with the RI and XI bits (if they exist) in bits 63 and 62 respectively, and
2071 the PFNX field starting at bit 30.
2073 MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) above) have the following id bit
2075 0x7030 0000 0001 01 <reg:8>
2077 MIPS KVM control registers (see above) have the following id bit patterns:
2078 0x7030 0000 0002 <reg:16>
2080 MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following
2081 id bit patterns depending on the size of the register being accessed. They are
2082 always accessed according to the current guest FPU mode (Status.FR and
2083 Config5.FRE), i.e. as the guest would see them, and they become unpredictable
2084 if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector
2085 registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they
2086 overlap the FPU registers:
2087 0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers)
2088 0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers)
2089 0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers)
2091 MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the
2092 following id bit patterns:
2093 0x7020 0000 0003 01 <0:3> <reg:5>
2095 MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the
2096 following id bit patterns:
2097 0x7020 0000 0003 02 <0:3> <reg:5>
2100 4.69 KVM_GET_ONE_REG
2102 Capability: KVM_CAP_ONE_REG
2105 Parameters: struct kvm_one_reg (in and out)
2106 Returns: 0 on success, negative value on failure
2108 This ioctl allows to receive the value of a single register implemented
2109 in a vcpu. The register to read is indicated by the "id" field of the
2110 kvm_one_reg struct passed in. On success, the register value can be found
2111 at the memory location pointed to by "addr".
2113 The list of registers accessible using this interface is identical to the
2117 4.70 KVM_KVMCLOCK_CTRL
2119 Capability: KVM_CAP_KVMCLOCK_CTRL
2120 Architectures: Any that implement pvclocks (currently x86 only)
2123 Returns: 0 on success, -1 on error
2125 This signals to the host kernel that the specified guest is being paused by
2126 userspace. The host will set a flag in the pvclock structure that is checked
2127 from the soft lockup watchdog. The flag is part of the pvclock structure that
2128 is shared between guest and host, specifically the second bit of the flags
2129 field of the pvclock_vcpu_time_info structure. It will be set exclusively by
2130 the host and read/cleared exclusively by the guest. The guest operation of
2131 checking and clearing the flag must an atomic operation so
2132 load-link/store-conditional, or equivalent must be used. There are two cases
2133 where the guest will clear the flag: when the soft lockup watchdog timer resets
2134 itself or when a soft lockup is detected. This ioctl can be called any time
2135 after pausing the vcpu, but before it is resumed.
2140 Capability: KVM_CAP_SIGNAL_MSI
2141 Architectures: x86 arm arm64
2143 Parameters: struct kvm_msi (in)
2144 Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error
2146 Directly inject a MSI message. Only valid with in-kernel irqchip that handles
2158 flags: KVM_MSI_VALID_DEVID: devid contains a valid value. The per-VM
2159 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
2160 the device ID. If this capability is not available, userspace
2161 should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
2163 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
2164 for the device that wrote the MSI message. For PCI, this is usually a
2165 BFD identifier in the lower 16 bits.
2167 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
2168 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
2169 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
2170 address_hi must be zero.
2173 4.71 KVM_CREATE_PIT2
2175 Capability: KVM_CAP_PIT2
2178 Parameters: struct kvm_pit_config (in)
2179 Returns: 0 on success, -1 on error
2181 Creates an in-kernel device model for the i8254 PIT. This call is only valid
2182 after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following
2183 parameters have to be passed:
2185 struct kvm_pit_config {
2192 #define KVM_PIT_SPEAKER_DUMMY 1 /* emulate speaker port stub */
2194 PIT timer interrupts may use a per-VM kernel thread for injection. If it
2195 exists, this thread will have a name of the following pattern:
2197 kvm-pit/<owner-process-pid>
2199 When running a guest with elevated priorities, the scheduling parameters of
2200 this thread may have to be adjusted accordingly.
2202 This IOCTL replaces the obsolete KVM_CREATE_PIT.
2207 Capability: KVM_CAP_PIT_STATE2
2210 Parameters: struct kvm_pit_state2 (out)
2211 Returns: 0 on success, -1 on error
2213 Retrieves the state of the in-kernel PIT model. Only valid after
2214 KVM_CREATE_PIT2. The state is returned in the following structure:
2216 struct kvm_pit_state2 {
2217 struct kvm_pit_channel_state channels[3];
2224 /* disable PIT in HPET legacy mode */
2225 #define KVM_PIT_FLAGS_HPET_LEGACY 0x00000001
2227 This IOCTL replaces the obsolete KVM_GET_PIT.
2232 Capability: KVM_CAP_PIT_STATE2
2235 Parameters: struct kvm_pit_state2 (in)
2236 Returns: 0 on success, -1 on error
2238 Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2.
2239 See KVM_GET_PIT2 for details on struct kvm_pit_state2.
2241 This IOCTL replaces the obsolete KVM_SET_PIT.
2244 4.74 KVM_PPC_GET_SMMU_INFO
2246 Capability: KVM_CAP_PPC_GET_SMMU_INFO
2247 Architectures: powerpc
2250 Returns: 0 on success, -1 on error
2252 This populates and returns a structure describing the features of
2253 the "Server" class MMU emulation supported by KVM.
2254 This can in turn be used by userspace to generate the appropriate
2255 device-tree properties for the guest operating system.
2257 The structure contains some global information, followed by an
2258 array of supported segment page sizes:
2260 struct kvm_ppc_smmu_info {
2264 struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ];
2267 The supported flags are:
2269 - KVM_PPC_PAGE_SIZES_REAL:
2270 When that flag is set, guest page sizes must "fit" the backing
2271 store page sizes. When not set, any page size in the list can
2272 be used regardless of how they are backed by userspace.
2274 - KVM_PPC_1T_SEGMENTS
2275 The emulated MMU supports 1T segments in addition to the
2278 The "slb_size" field indicates how many SLB entries are supported
2280 The "sps" array contains 8 entries indicating the supported base
2281 page sizes for a segment in increasing order. Each entry is defined
2284 struct kvm_ppc_one_seg_page_size {
2285 __u32 page_shift; /* Base page shift of segment (or 0) */
2286 __u32 slb_enc; /* SLB encoding for BookS */
2287 struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ];
2290 An entry with a "page_shift" of 0 is unused. Because the array is
2291 organized in increasing order, a lookup can stop when encoutering
2294 The "slb_enc" field provides the encoding to use in the SLB for the
2295 page size. The bits are in positions such as the value can directly
2296 be OR'ed into the "vsid" argument of the slbmte instruction.
2298 The "enc" array is a list which for each of those segment base page
2299 size provides the list of supported actual page sizes (which can be
2300 only larger or equal to the base page size), along with the
2301 corresponding encoding in the hash PTE. Similarly, the array is
2302 8 entries sorted by increasing sizes and an entry with a "0" shift
2303 is an empty entry and a terminator:
2305 struct kvm_ppc_one_page_size {
2306 __u32 page_shift; /* Page shift (or 0) */
2307 __u32 pte_enc; /* Encoding in the HPTE (>>12) */
2310 The "pte_enc" field provides a value that can OR'ed into the hash
2311 PTE's RPN field (ie, it needs to be shifted left by 12 to OR it
2312 into the hash PTE second double word).
2316 Capability: KVM_CAP_IRQFD
2317 Architectures: x86 s390 arm arm64
2319 Parameters: struct kvm_irqfd (in)
2320 Returns: 0 on success, -1 on error
2322 Allows setting an eventfd to directly trigger a guest interrupt.
2323 kvm_irqfd.fd specifies the file descriptor to use as the eventfd and
2324 kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When
2325 an event is triggered on the eventfd, an interrupt is injected into
2326 the guest using the specified gsi pin. The irqfd is removed using
2327 the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd
2330 With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify
2331 mechanism allowing emulation of level-triggered, irqfd-based
2332 interrupts. When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an
2333 additional eventfd in the kvm_irqfd.resamplefd field. When operating
2334 in resample mode, posting of an interrupt through kvm_irq.fd asserts
2335 the specified gsi in the irqchip. When the irqchip is resampled, such
2336 as from an EOI, the gsi is de-asserted and the user is notified via
2337 kvm_irqfd.resamplefd. It is the user's responsibility to re-queue
2338 the interrupt if the device making use of it still requires service.
2339 Note that closing the resamplefd is not sufficient to disable the
2340 irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment
2341 and need not be specified with KVM_IRQFD_FLAG_DEASSIGN.
2343 On arm/arm64, gsi routing being supported, the following can happen:
2344 - in case no routing entry is associated to this gsi, injection fails
2345 - in case the gsi is associated to an irqchip routing entry,
2346 irqchip.pin + 32 corresponds to the injected SPI ID.
2347 - in case the gsi is associated to an MSI routing entry, the MSI
2348 message and device ID are translated into an LPI (support restricted
2349 to GICv3 ITS in-kernel emulation).
2351 4.76 KVM_PPC_ALLOCATE_HTAB
2353 Capability: KVM_CAP_PPC_ALLOC_HTAB
2354 Architectures: powerpc
2356 Parameters: Pointer to u32 containing hash table order (in/out)
2357 Returns: 0 on success, -1 on error
2359 This requests the host kernel to allocate an MMU hash table for a
2360 guest using the PAPR paravirtualization interface. This only does
2361 anything if the kernel is configured to use the Book 3S HV style of
2362 virtualization. Otherwise the capability doesn't exist and the ioctl
2363 returns an ENOTTY error. The rest of this description assumes Book 3S
2366 There must be no vcpus running when this ioctl is called; if there
2367 are, it will do nothing and return an EBUSY error.
2369 The parameter is a pointer to a 32-bit unsigned integer variable
2370 containing the order (log base 2) of the desired size of the hash
2371 table, which must be between 18 and 46. On successful return from the
2372 ioctl, the value will not be changed by the kernel.
2374 If no hash table has been allocated when any vcpu is asked to run
2375 (with the KVM_RUN ioctl), the host kernel will allocate a
2376 default-sized hash table (16 MB).
2378 If this ioctl is called when a hash table has already been allocated,
2379 with a different order from the existing hash table, the existing hash
2380 table will be freed and a new one allocated. If this is ioctl is
2381 called when a hash table has already been allocated of the same order
2382 as specified, the kernel will clear out the existing hash table (zero
2383 all HPTEs). In either case, if the guest is using the virtualized
2384 real-mode area (VRMA) facility, the kernel will re-create the VMRA
2385 HPTEs on the next KVM_RUN of any vcpu.
2387 4.77 KVM_S390_INTERRUPT
2391 Type: vm ioctl, vcpu ioctl
2392 Parameters: struct kvm_s390_interrupt (in)
2393 Returns: 0 on success, -1 on error
2395 Allows to inject an interrupt to the guest. Interrupts can be floating
2396 (vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type.
2398 Interrupt parameters are passed via kvm_s390_interrupt:
2400 struct kvm_s390_interrupt {
2406 type can be one of the following:
2408 KVM_S390_SIGP_STOP (vcpu) - sigp stop; optional flags in parm
2409 KVM_S390_PROGRAM_INT (vcpu) - program check; code in parm
2410 KVM_S390_SIGP_SET_PREFIX (vcpu) - sigp set prefix; prefix address in parm
2411 KVM_S390_RESTART (vcpu) - restart
2412 KVM_S390_INT_CLOCK_COMP (vcpu) - clock comparator interrupt
2413 KVM_S390_INT_CPU_TIMER (vcpu) - CPU timer interrupt
2414 KVM_S390_INT_VIRTIO (vm) - virtio external interrupt; external interrupt
2415 parameters in parm and parm64
2416 KVM_S390_INT_SERVICE (vm) - sclp external interrupt; sclp parameter in parm
2417 KVM_S390_INT_EMERGENCY (vcpu) - sigp emergency; source cpu in parm
2418 KVM_S390_INT_EXTERNAL_CALL (vcpu) - sigp external call; source cpu in parm
2419 KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm) - compound value to indicate an
2420 I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel);
2421 I/O interruption parameters in parm (subchannel) and parm64 (intparm,
2422 interruption subclass)
2423 KVM_S390_MCHK (vm, vcpu) - machine check interrupt; cr 14 bits in parm,
2424 machine check interrupt code in parm64 (note that
2425 machine checks needing further payload are not
2426 supported by this ioctl)
2428 Note that the vcpu ioctl is asynchronous to vcpu execution.
2430 4.78 KVM_PPC_GET_HTAB_FD
2432 Capability: KVM_CAP_PPC_HTAB_FD
2433 Architectures: powerpc
2435 Parameters: Pointer to struct kvm_get_htab_fd (in)
2436 Returns: file descriptor number (>= 0) on success, -1 on error
2438 This returns a file descriptor that can be used either to read out the
2439 entries in the guest's hashed page table (HPT), or to write entries to
2440 initialize the HPT. The returned fd can only be written to if the
2441 KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and
2442 can only be read if that bit is clear. The argument struct looks like
2445 /* For KVM_PPC_GET_HTAB_FD */
2446 struct kvm_get_htab_fd {
2452 /* Values for kvm_get_htab_fd.flags */
2453 #define KVM_GET_HTAB_BOLTED_ONLY ((__u64)0x1)
2454 #define KVM_GET_HTAB_WRITE ((__u64)0x2)
2456 The `start_index' field gives the index in the HPT of the entry at
2457 which to start reading. It is ignored when writing.
2459 Reads on the fd will initially supply information about all
2460 "interesting" HPT entries. Interesting entries are those with the
2461 bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise
2462 all entries. When the end of the HPT is reached, the read() will
2463 return. If read() is called again on the fd, it will start again from
2464 the beginning of the HPT, but will only return HPT entries that have
2465 changed since they were last read.
2467 Data read or written is structured as a header (8 bytes) followed by a
2468 series of valid HPT entries (16 bytes) each. The header indicates how
2469 many valid HPT entries there are and how many invalid entries follow
2470 the valid entries. The invalid entries are not represented explicitly
2471 in the stream. The header format is:
2473 struct kvm_get_htab_header {
2479 Writes to the fd create HPT entries starting at the index given in the
2480 header; first `n_valid' valid entries with contents from the data
2481 written, then `n_invalid' invalid entries, invalidating any previously
2482 valid entries found.
2484 4.79 KVM_CREATE_DEVICE
2486 Capability: KVM_CAP_DEVICE_CTRL
2488 Parameters: struct kvm_create_device (in/out)
2489 Returns: 0 on success, -1 on error
2491 ENODEV: The device type is unknown or unsupported
2492 EEXIST: Device already created, and this type of device may not
2493 be instantiated multiple times
2495 Other error conditions may be defined by individual device types or
2496 have their standard meanings.
2498 Creates an emulated device in the kernel. The file descriptor returned
2499 in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR.
2501 If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the
2502 device type is supported (not necessarily whether it can be created
2505 Individual devices should not define flags. Attributes should be used
2506 for specifying any behavior that is not implied by the device type
2509 struct kvm_create_device {
2510 __u32 type; /* in: KVM_DEV_TYPE_xxx */
2511 __u32 fd; /* out: device handle */
2512 __u32 flags; /* in: KVM_CREATE_DEVICE_xxx */
2515 4.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR
2517 Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
2518 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
2519 Type: device ioctl, vm ioctl, vcpu ioctl
2520 Parameters: struct kvm_device_attr
2521 Returns: 0 on success, -1 on error
2523 ENXIO: The group or attribute is unknown/unsupported for this device
2524 or hardware support is missing.
2525 EPERM: The attribute cannot (currently) be accessed this way
2526 (e.g. read-only attribute, or attribute that only makes
2527 sense when the device is in a different state)
2529 Other error conditions may be defined by individual device types.
2531 Gets/sets a specified piece of device configuration and/or state. The
2532 semantics are device-specific. See individual device documentation in
2533 the "devices" directory. As with ONE_REG, the size of the data
2534 transferred is defined by the particular attribute.
2536 struct kvm_device_attr {
2537 __u32 flags; /* no flags currently defined */
2538 __u32 group; /* device-defined */
2539 __u64 attr; /* group-defined */
2540 __u64 addr; /* userspace address of attr data */
2543 4.81 KVM_HAS_DEVICE_ATTR
2545 Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
2546 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
2547 Type: device ioctl, vm ioctl, vcpu ioctl
2548 Parameters: struct kvm_device_attr
2549 Returns: 0 on success, -1 on error
2551 ENXIO: The group or attribute is unknown/unsupported for this device
2552 or hardware support is missing.
2554 Tests whether a device supports a particular attribute. A successful
2555 return indicates the attribute is implemented. It does not necessarily
2556 indicate that the attribute can be read or written in the device's
2557 current state. "addr" is ignored.
2559 4.82 KVM_ARM_VCPU_INIT
2562 Architectures: arm, arm64
2564 Parameters: struct kvm_vcpu_init (in)
2565 Returns: 0 on success; -1 on error
2567 Â EINVAL: Â Â Â the target is unknown, or the combination of features is invalid.
2568 Â ENOENT: Â Â Â a features bit specified is unknown.
2570 This tells KVM what type of CPU to present to the guest, and what
2571 optional features it should have. Â This will cause a reset of the cpu
2572 registers to their initial values. Â If this is not called, KVM_RUN will
2573 return ENOEXEC for that vcpu.
2575 Note that because some registers reflect machine topology, all vcpus
2576 should be created before this ioctl is invoked.
2578 Userspace can call this function multiple times for a given vcpu, including
2579 after the vcpu has been run. This will reset the vcpu to its initial
2580 state. All calls to this function after the initial call must use the same
2581 target and same set of feature flags, otherwise EINVAL will be returned.
2584 - KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state.
2585 Depends on KVM_CAP_ARM_PSCI. If not set, the CPU will be powered on
2586 and execute guest code when KVM_RUN is called.
2587 - KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode.
2588 Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only).
2589 - KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 (or a future revision
2590 backward compatible with v0.2) for the CPU.
2591 Depends on KVM_CAP_ARM_PSCI_0_2.
2592 - KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU.
2593 Depends on KVM_CAP_ARM_PMU_V3.
2596 4.83 KVM_ARM_PREFERRED_TARGET
2599 Architectures: arm, arm64
2601 Parameters: struct struct kvm_vcpu_init (out)
2602 Returns: 0 on success; -1 on error
2604 ENODEV: no preferred target available for the host
2606 This queries KVM for preferred CPU target type which can be emulated
2607 by KVM on underlying host.
2609 The ioctl returns struct kvm_vcpu_init instance containing information
2610 about preferred CPU target type and recommended features for it. The
2611 kvm_vcpu_init->features bitmap returned will have feature bits set if
2612 the preferred target recommends setting these features, but this is
2615 The information returned by this ioctl can be used to prepare an instance
2616 of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in
2617 in VCPU matching underlying host.
2620 4.84 KVM_GET_REG_LIST
2623 Architectures: arm, arm64, mips
2625 Parameters: struct kvm_reg_list (in/out)
2626 Returns: 0 on success; -1 on error
2628 Â E2BIG: Â Â Â Â the reg index list is too big to fit in the array specified by
2629 Â Â Â Â Â Â Â Â Â Â Â Â the user (the number required will be written into n).
2631 struct kvm_reg_list {
2632 __u64 n; /* number of registers in reg[] */
2636 This ioctl returns the guest registers that are supported for the
2637 KVM_GET_ONE_REG/KVM_SET_ONE_REG calls.
2640 4.85 KVM_ARM_SET_DEVICE_ADDR (deprecated)
2642 Capability: KVM_CAP_ARM_SET_DEVICE_ADDR
2643 Architectures: arm, arm64
2645 Parameters: struct kvm_arm_device_address (in)
2646 Returns: 0 on success, -1 on error
2648 ENODEV: The device id is unknown
2649 ENXIO: Device not supported on current system
2650 EEXIST: Address already set
2651 E2BIG: Address outside guest physical address space
2652 EBUSY: Address overlaps with other device range
2654 struct kvm_arm_device_addr {
2659 Specify a device address in the guest's physical address space where guests
2660 can access emulated or directly exposed devices, which the host kernel needs
2661 to know about. The id field is an architecture specific identifier for a
2664 ARM/arm64 divides the id field into two parts, a device id and an
2665 address type id specific to the individual device.
2667 Â bits: | 63 ... 32 | 31 ... 16 | 15 ... 0 |
2668 field: | 0x00000000 | device id | addr type id |
2670 ARM/arm64 currently only require this when using the in-kernel GIC
2671 support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2
2672 as the device id. When setting the base address for the guest's
2673 mapping of the VGIC virtual CPU and distributor interface, the ioctl
2674 must be called after calling KVM_CREATE_IRQCHIP, but before calling
2675 KVM_RUN on any of the VCPUs. Calling this ioctl twice for any of the
2676 base addresses will return -EEXIST.
2678 Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API
2679 should be used instead.
2682 4.86 KVM_PPC_RTAS_DEFINE_TOKEN
2684 Capability: KVM_CAP_PPC_RTAS
2687 Parameters: struct kvm_rtas_token_args
2688 Returns: 0 on success, -1 on error
2690 Defines a token value for a RTAS (Run Time Abstraction Services)
2691 service in order to allow it to be handled in the kernel. The
2692 argument struct gives the name of the service, which must be the name
2693 of a service that has a kernel-side implementation. If the token
2694 value is non-zero, it will be associated with that service, and
2695 subsequent RTAS calls by the guest specifying that token will be
2696 handled by the kernel. If the token value is 0, then any token
2697 associated with the service will be forgotten, and subsequent RTAS
2698 calls by the guest for that service will be passed to userspace to be
2701 4.87 KVM_SET_GUEST_DEBUG
2703 Capability: KVM_CAP_SET_GUEST_DEBUG
2704 Architectures: x86, s390, ppc, arm64
2706 Parameters: struct kvm_guest_debug (in)
2707 Returns: 0 on success; -1 on error
2709 struct kvm_guest_debug {
2712 struct kvm_guest_debug_arch arch;
2715 Set up the processor specific debug registers and configure vcpu for
2716 handling guest debug events. There are two parts to the structure, the
2717 first a control bitfield indicates the type of debug events to handle
2718 when running. Common control bits are:
2720 - KVM_GUESTDBG_ENABLE: guest debugging is enabled
2721 - KVM_GUESTDBG_SINGLESTEP: the next run should single-step
2723 The top 16 bits of the control field are architecture specific control
2724 flags which can include the following:
2726 - KVM_GUESTDBG_USE_SW_BP: using software breakpoints [x86, arm64]
2727 - KVM_GUESTDBG_USE_HW_BP: using hardware breakpoints [x86, s390, arm64]
2728 - KVM_GUESTDBG_INJECT_DB: inject DB type exception [x86]
2729 - KVM_GUESTDBG_INJECT_BP: inject BP type exception [x86]
2730 - KVM_GUESTDBG_EXIT_PENDING: trigger an immediate guest exit [s390]
2732 For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints
2733 are enabled in memory so we need to ensure breakpoint exceptions are
2734 correctly trapped and the KVM run loop exits at the breakpoint and not
2735 running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP
2736 we need to ensure the guest vCPUs architecture specific registers are
2737 updated to the correct (supplied) values.
2739 The second part of the structure is architecture specific and
2740 typically contains a set of debug registers.
2742 For arm64 the number of debug registers is implementation defined and
2743 can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and
2744 KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number
2745 indicating the number of supported registers.
2747 When debug events exit the main run loop with the reason
2748 KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run
2749 structure containing architecture specific debug information.
2751 4.88 KVM_GET_EMULATED_CPUID
2753 Capability: KVM_CAP_EXT_EMUL_CPUID
2756 Parameters: struct kvm_cpuid2 (in/out)
2757 Returns: 0 on success, -1 on error
2762 struct kvm_cpuid_entry2 entries[0];
2765 The member 'flags' is used for passing flags from userspace.
2767 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
2768 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1)
2769 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2)
2771 struct kvm_cpuid_entry2 {
2782 This ioctl returns x86 cpuid features which are emulated by
2783 kvm.Userspace can use the information returned by this ioctl to query
2784 which features are emulated by kvm instead of being present natively.
2786 Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2
2787 structure with the 'nent' field indicating the number of entries in
2788 the variable-size array 'entries'. If the number of entries is too low
2789 to describe the cpu capabilities, an error (E2BIG) is returned. If the
2790 number is too high, the 'nent' field is adjusted and an error (ENOMEM)
2791 is returned. If the number is just right, the 'nent' field is adjusted
2792 to the number of valid entries in the 'entries' array, which is then
2795 The entries returned are the set CPUID bits of the respective features
2796 which kvm emulates, as returned by the CPUID instruction, with unknown
2797 or unsupported feature bits cleared.
2799 Features like x2apic, for example, may not be present in the host cpu
2800 but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be
2801 emulated efficiently and thus not included here.
2803 The fields in each entry are defined as follows:
2805 function: the eax value used to obtain the entry
2806 index: the ecx value used to obtain the entry (for entries that are
2808 flags: an OR of zero or more of the following:
2809 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
2810 if the index field is valid
2811 KVM_CPUID_FLAG_STATEFUL_FUNC:
2812 if cpuid for this function returns different values for successive
2813 invocations; there will be several entries with the same function,
2814 all with this flag set
2815 KVM_CPUID_FLAG_STATE_READ_NEXT:
2816 for KVM_CPUID_FLAG_STATEFUL_FUNC entries, set if this entry is
2817 the first entry to be read by a cpu
2818 eax, ebx, ecx, edx: the values returned by the cpuid instruction for
2819 this function/index combination
2821 4.89 KVM_S390_MEM_OP
2823 Capability: KVM_CAP_S390_MEM_OP
2826 Parameters: struct kvm_s390_mem_op (in)
2827 Returns: = 0 on success,
2828 < 0 on generic error (e.g. -EFAULT or -ENOMEM),
2829 > 0 if an exception occurred while walking the page tables
2831 Read or write data from/to the logical (virtual) memory of a VCPU.
2833 Parameters are specified via the following structure:
2835 struct kvm_s390_mem_op {
2836 __u64 gaddr; /* the guest address */
2837 __u64 flags; /* flags */
2838 __u32 size; /* amount of bytes */
2839 __u32 op; /* type of operation */
2840 __u64 buf; /* buffer in userspace */
2841 __u8 ar; /* the access register number */
2842 __u8 reserved[31]; /* should be set to 0 */
2845 The type of operation is specified in the "op" field. It is either
2846 KVM_S390_MEMOP_LOGICAL_READ for reading from logical memory space or
2847 KVM_S390_MEMOP_LOGICAL_WRITE for writing to logical memory space. The
2848 KVM_S390_MEMOP_F_CHECK_ONLY flag can be set in the "flags" field to check
2849 whether the corresponding memory access would create an access exception
2850 (without touching the data in the memory at the destination). In case an
2851 access exception occurred while walking the MMU tables of the guest, the
2852 ioctl returns a positive error number to indicate the type of exception.
2853 This exception is also raised directly at the corresponding VCPU if the
2854 flag KVM_S390_MEMOP_F_INJECT_EXCEPTION is set in the "flags" field.
2856 The start address of the memory region has to be specified in the "gaddr"
2857 field, and the length of the region in the "size" field. "buf" is the buffer
2858 supplied by the userspace application where the read data should be written
2859 to for KVM_S390_MEMOP_LOGICAL_READ, or where the data that should be written
2860 is stored for a KVM_S390_MEMOP_LOGICAL_WRITE. "buf" is unused and can be NULL
2861 when KVM_S390_MEMOP_F_CHECK_ONLY is specified. "ar" designates the access
2862 register number to be used.
2864 The "reserved" field is meant for future extensions. It is not used by
2865 KVM with the currently defined set of flags.
2867 4.90 KVM_S390_GET_SKEYS
2869 Capability: KVM_CAP_S390_SKEYS
2872 Parameters: struct kvm_s390_skeys
2873 Returns: 0 on success, KVM_S390_GET_KEYS_NONE if guest is not using storage
2874 keys, negative value on error
2876 This ioctl is used to get guest storage key values on the s390
2877 architecture. The ioctl takes parameters via the kvm_s390_skeys struct.
2879 struct kvm_s390_skeys {
2882 __u64 skeydata_addr;
2887 The start_gfn field is the number of the first guest frame whose storage keys
2890 The count field is the number of consecutive frames (starting from start_gfn)
2891 whose storage keys to get. The count field must be at least 1 and the maximum
2892 allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range
2893 will cause the ioctl to return -EINVAL.
2895 The skeydata_addr field is the address to a buffer large enough to hold count
2896 bytes. This buffer will be filled with storage key data by the ioctl.
2898 4.91 KVM_S390_SET_SKEYS
2900 Capability: KVM_CAP_S390_SKEYS
2903 Parameters: struct kvm_s390_skeys
2904 Returns: 0 on success, negative value on error
2906 This ioctl is used to set guest storage key values on the s390
2907 architecture. The ioctl takes parameters via the kvm_s390_skeys struct.
2908 See section on KVM_S390_GET_SKEYS for struct definition.
2910 The start_gfn field is the number of the first guest frame whose storage keys
2913 The count field is the number of consecutive frames (starting from start_gfn)
2914 whose storage keys to get. The count field must be at least 1 and the maximum
2915 allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range
2916 will cause the ioctl to return -EINVAL.
2918 The skeydata_addr field is the address to a buffer containing count bytes of
2919 storage keys. Each byte in the buffer will be set as the storage key for a
2920 single frame starting at start_gfn for count frames.
2922 Note: If any architecturally invalid key value is found in the given data then
2923 the ioctl will return -EINVAL.
2927 Capability: KVM_CAP_S390_INJECT_IRQ
2930 Parameters: struct kvm_s390_irq (in)
2931 Returns: 0 on success, -1 on error
2933 EINVAL: interrupt type is invalid
2934 type is KVM_S390_SIGP_STOP and flag parameter is invalid value
2935 type is KVM_S390_INT_EXTERNAL_CALL and code is bigger
2936 than the maximum of VCPUs
2937 EBUSY: type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped
2938 type is KVM_S390_SIGP_STOP and a stop irq is already pending
2939 type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt
2942 Allows to inject an interrupt to the guest.
2944 Using struct kvm_s390_irq as a parameter allows
2945 to inject additional payload which is not
2946 possible via KVM_S390_INTERRUPT.
2948 Interrupt parameters are passed via kvm_s390_irq:
2950 struct kvm_s390_irq {
2953 struct kvm_s390_io_info io;
2954 struct kvm_s390_ext_info ext;
2955 struct kvm_s390_pgm_info pgm;
2956 struct kvm_s390_emerg_info emerg;
2957 struct kvm_s390_extcall_info extcall;
2958 struct kvm_s390_prefix_info prefix;
2959 struct kvm_s390_stop_info stop;
2960 struct kvm_s390_mchk_info mchk;
2965 type can be one of the following:
2967 KVM_S390_SIGP_STOP - sigp stop; parameter in .stop
2968 KVM_S390_PROGRAM_INT - program check; parameters in .pgm
2969 KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix
2970 KVM_S390_RESTART - restart; no parameters
2971 KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters
2972 KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters
2973 KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg
2974 KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall
2975 KVM_S390_MCHK - machine check interrupt; parameters in .mchk
2978 Note that the vcpu ioctl is asynchronous to vcpu execution.
2980 4.94 KVM_S390_GET_IRQ_STATE
2982 Capability: KVM_CAP_S390_IRQ_STATE
2985 Parameters: struct kvm_s390_irq_state (out)
2986 Returns: >= number of bytes copied into buffer,
2987 -EINVAL if buffer size is 0,
2988 -ENOBUFS if buffer size is too small to fit all pending interrupts,
2989 -EFAULT if the buffer address was invalid
2991 This ioctl allows userspace to retrieve the complete state of all currently
2992 pending interrupts in a single buffer. Use cases include migration
2993 and introspection. The parameter structure contains the address of a
2994 userspace buffer and its length:
2996 struct kvm_s390_irq_state {
2998 __u32 flags; /* will stay unused for compatibility reasons */
3000 __u32 reserved[4]; /* will stay unused for compatibility reasons */
3003 Userspace passes in the above struct and for each pending interrupt a
3004 struct kvm_s390_irq is copied to the provided buffer.
3006 The structure contains a flags and a reserved field for future extensions. As
3007 the kernel never checked for flags == 0 and QEMU never pre-zeroed flags and
3008 reserved, these fields can not be used in the future without breaking
3011 If -ENOBUFS is returned the buffer provided was too small and userspace
3012 may retry with a bigger buffer.
3014 4.95 KVM_S390_SET_IRQ_STATE
3016 Capability: KVM_CAP_S390_IRQ_STATE
3019 Parameters: struct kvm_s390_irq_state (in)
3020 Returns: 0 on success,
3021 -EFAULT if the buffer address was invalid,
3022 -EINVAL for an invalid buffer length (see below),
3023 -EBUSY if there were already interrupts pending,
3024 errors occurring when actually injecting the
3025 interrupt. See KVM_S390_IRQ.
3027 This ioctl allows userspace to set the complete state of all cpu-local
3028 interrupts currently pending for the vcpu. It is intended for restoring
3029 interrupt state after a migration. The input parameter is a userspace buffer
3030 containing a struct kvm_s390_irq_state:
3032 struct kvm_s390_irq_state {
3034 __u32 flags; /* will stay unused for compatibility reasons */
3036 __u32 reserved[4]; /* will stay unused for compatibility reasons */
3039 The restrictions for flags and reserved apply as well.
3040 (see KVM_S390_GET_IRQ_STATE)
3042 The userspace memory referenced by buf contains a struct kvm_s390_irq
3043 for each interrupt to be injected into the guest.
3044 If one of the interrupts could not be injected for some reason the
3047 len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0
3048 and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq),
3049 which is the maximum number of possibly pending cpu-local interrupts.
3053 Capability: KVM_CAP_X86_SMM
3057 Returns: 0 on success, -1 on error
3059 Queues an SMI on the thread's vcpu.
3061 4.97 KVM_CAP_PPC_MULTITCE
3063 Capability: KVM_CAP_PPC_MULTITCE
3067 This capability means the kernel is capable of handling hypercalls
3068 H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user
3069 space. This significantly accelerates DMA operations for PPC KVM guests.
3070 User space should expect that its handlers for these hypercalls
3071 are not going to be called if user space previously registered LIOBN
3072 in KVM (via KVM_CREATE_SPAPR_TCE or similar calls).
3074 In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest,
3075 user space might have to advertise it for the guest. For example,
3076 IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is
3077 present in the "ibm,hypertas-functions" device-tree property.
3079 The hypercalls mentioned above may or may not be processed successfully
3080 in the kernel based fast path. If they can not be handled by the kernel,
3081 they will get passed on to user space. So user space still has to have
3082 an implementation for these despite the in kernel acceleration.
3084 This capability is always enabled.
3086 4.98 KVM_CREATE_SPAPR_TCE_64
3088 Capability: KVM_CAP_SPAPR_TCE_64
3089 Architectures: powerpc
3091 Parameters: struct kvm_create_spapr_tce_64 (in)
3092 Returns: file descriptor for manipulating the created TCE table
3094 This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit
3095 windows, described in 4.62 KVM_CREATE_SPAPR_TCE
3097 This capability uses extended struct in ioctl interface:
3099 /* for KVM_CAP_SPAPR_TCE_64 */
3100 struct kvm_create_spapr_tce_64 {
3104 __u64 offset; /* in pages */
3105 __u64 size; /* in pages */
3108 The aim of extension is to support an additional bigger DMA window with
3109 a variable page size.
3110 KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and
3111 a bus offset of the corresponding DMA window, @size and @offset are numbers
3114 @flags are not used at the moment.
3116 The rest of functionality is identical to KVM_CREATE_SPAPR_TCE.
3118 4.99 KVM_REINJECT_CONTROL
3120 Capability: KVM_CAP_REINJECT_CONTROL
3123 Parameters: struct kvm_reinject_control (in)
3124 Returns: 0 on success,
3125 -EFAULT if struct kvm_reinject_control cannot be read,
3126 -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier.
3128 i8254 (PIT) has two modes, reinject and !reinject. The default is reinject,
3129 where KVM queues elapsed i8254 ticks and monitors completion of interrupt from
3130 vector(s) that i8254 injects. Reinject mode dequeues a tick and injects its
3131 interrupt whenever there isn't a pending interrupt from i8254.
3132 !reinject mode injects an interrupt as soon as a tick arrives.
3134 struct kvm_reinject_control {
3139 pit_reinject = 0 (!reinject mode) is recommended, unless running an old
3140 operating system that uses the PIT for timing (e.g. Linux 2.4.x).
3142 4.100 KVM_PPC_CONFIGURE_V3_MMU
3144 Capability: KVM_CAP_PPC_RADIX_MMU or KVM_CAP_PPC_HASH_MMU_V3
3147 Parameters: struct kvm_ppc_mmuv3_cfg (in)
3148 Returns: 0 on success,
3149 -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read,
3150 -EINVAL if the configuration is invalid
3152 This ioctl controls whether the guest will use radix or HPT (hashed
3153 page table) translation, and sets the pointer to the process table for
3156 struct kvm_ppc_mmuv3_cfg {
3158 __u64 process_table;
3161 There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and
3162 KVM_PPC_MMUV3_GTSE. KVM_PPC_MMUV3_RADIX, if set, configures the guest
3163 to use radix tree translation, and if clear, to use HPT translation.
3164 KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest
3165 to be able to use the global TLB and SLB invalidation instructions;
3166 if clear, the guest may not use these instructions.
3168 The process_table field specifies the address and size of the guest
3169 process table, which is in the guest's space. This field is formatted
3170 as the second doubleword of the partition table entry, as defined in
3171 the Power ISA V3.00, Book III section 5.7.6.1.
3173 4.101 KVM_PPC_GET_RMMU_INFO
3175 Capability: KVM_CAP_PPC_RADIX_MMU
3178 Parameters: struct kvm_ppc_rmmu_info (out)
3179 Returns: 0 on success,
3180 -EFAULT if struct kvm_ppc_rmmu_info cannot be written,
3181 -EINVAL if no useful information can be returned
3183 This ioctl returns a structure containing two things: (a) a list
3184 containing supported radix tree geometries, and (b) a list that maps
3185 page sizes to put in the "AP" (actual page size) field for the tlbie
3186 (TLB invalidate entry) instruction.
3188 struct kvm_ppc_rmmu_info {
3189 struct kvm_ppc_radix_geom {
3194 __u32 ap_encodings[8];
3197 The geometries[] field gives up to 8 supported geometries for the
3198 radix page table, in terms of the log base 2 of the smallest page
3199 size, and the number of bits indexed at each level of the tree, from
3200 the PTE level up to the PGD level in that order. Any unused entries
3201 will have 0 in the page_shift field.
3203 The ap_encodings gives the supported page sizes and their AP field
3204 encodings, encoded with the AP value in the top 3 bits and the log
3205 base 2 of the page size in the bottom 6 bits.
3207 4.102 KVM_PPC_RESIZE_HPT_PREPARE
3209 Capability: KVM_CAP_SPAPR_RESIZE_HPT
3210 Architectures: powerpc
3212 Parameters: struct kvm_ppc_resize_hpt (in)
3213 Returns: 0 on successful completion,
3214 >0 if a new HPT is being prepared, the value is an estimated
3215 number of milliseconds until preparation is complete
3216 -EFAULT if struct kvm_reinject_control cannot be read,
3217 -EINVAL if the supplied shift or flags are invalid
3218 -ENOMEM if unable to allocate the new HPT
3219 -ENOSPC if there was a hash collision when moving existing
3220 HPT entries to the new HPT
3221 -EIO on other error conditions
3223 Used to implement the PAPR extension for runtime resizing of a guest's
3224 Hashed Page Table (HPT). Specifically this starts, stops or monitors
3225 the preparation of a new potential HPT for the guest, essentially
3226 implementing the H_RESIZE_HPT_PREPARE hypercall.
3228 If called with shift > 0 when there is no pending HPT for the guest,
3229 this begins preparation of a new pending HPT of size 2^(shift) bytes.
3230 It then returns a positive integer with the estimated number of
3231 milliseconds until preparation is complete.
3233 If called when there is a pending HPT whose size does not match that
3234 requested in the parameters, discards the existing pending HPT and
3235 creates a new one as above.
3237 If called when there is a pending HPT of the size requested, will:
3238 * If preparation of the pending HPT is already complete, return 0
3239 * If preparation of the pending HPT has failed, return an error
3240 code, then discard the pending HPT.
3241 * If preparation of the pending HPT is still in progress, return an
3242 estimated number of milliseconds until preparation is complete.
3244 If called with shift == 0, discards any currently pending HPT and
3245 returns 0 (i.e. cancels any in-progress preparation).
3247 flags is reserved for future expansion, currently setting any bits in
3248 flags will result in an -EINVAL.
3250 Normally this will be called repeatedly with the same parameters until
3251 it returns <= 0. The first call will initiate preparation, subsequent
3252 ones will monitor preparation until it completes or fails.
3254 struct kvm_ppc_resize_hpt {
3260 4.103 KVM_PPC_RESIZE_HPT_COMMIT
3262 Capability: KVM_CAP_SPAPR_RESIZE_HPT
3263 Architectures: powerpc
3265 Parameters: struct kvm_ppc_resize_hpt (in)
3266 Returns: 0 on successful completion,
3267 -EFAULT if struct kvm_reinject_control cannot be read,
3268 -EINVAL if the supplied shift or flags are invalid
3269 -ENXIO is there is no pending HPT, or the pending HPT doesn't
3270 have the requested size
3271 -EBUSY if the pending HPT is not fully prepared
3272 -ENOSPC if there was a hash collision when moving existing
3273 HPT entries to the new HPT
3274 -EIO on other error conditions
3276 Used to implement the PAPR extension for runtime resizing of a guest's
3277 Hashed Page Table (HPT). Specifically this requests that the guest be
3278 transferred to working with the new HPT, essentially implementing the
3279 H_RESIZE_HPT_COMMIT hypercall.
3281 This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has
3282 returned 0 with the same parameters. In other cases
3283 KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or
3284 -EBUSY, though others may be possible if the preparation was started,
3287 This will have undefined effects on the guest if it has not already
3288 placed itself in a quiescent state where no vcpu will make MMU enabled
3291 On succsful completion, the pending HPT will become the guest's active
3292 HPT and the previous HPT will be discarded.
3294 On failure, the guest will still be operating on its previous HPT.
3296 struct kvm_ppc_resize_hpt {
3302 4.104 KVM_X86_GET_MCE_CAP_SUPPORTED
3304 Capability: KVM_CAP_MCE
3307 Parameters: u64 mce_cap (out)
3308 Returns: 0 on success, -1 on error
3310 Returns supported MCE capabilities. The u64 mce_cap parameter
3311 has the same format as the MSR_IA32_MCG_CAP register. Supported
3312 capabilities will have the corresponding bits set.
3314 4.105 KVM_X86_SETUP_MCE
3316 Capability: KVM_CAP_MCE
3319 Parameters: u64 mcg_cap (in)
3320 Returns: 0 on success,
3321 -EFAULT if u64 mcg_cap cannot be read,
3322 -EINVAL if the requested number of banks is invalid,
3323 -EINVAL if requested MCE capability is not supported.
3325 Initializes MCE support for use. The u64 mcg_cap parameter
3326 has the same format as the MSR_IA32_MCG_CAP register and
3327 specifies which capabilities should be enabled. The maximum
3328 supported number of error-reporting banks can be retrieved when
3329 checking for KVM_CAP_MCE. The supported capabilities can be
3330 retrieved with KVM_X86_GET_MCE_CAP_SUPPORTED.
3332 4.106 KVM_X86_SET_MCE
3334 Capability: KVM_CAP_MCE
3337 Parameters: struct kvm_x86_mce (in)
3338 Returns: 0 on success,
3339 -EFAULT if struct kvm_x86_mce cannot be read,
3340 -EINVAL if the bank number is invalid,
3341 -EINVAL if VAL bit is not set in status field.
3343 Inject a machine check error (MCE) into the guest. The input
3346 struct kvm_x86_mce {
3356 If the MCE being reported is an uncorrected error, KVM will
3357 inject it as an MCE exception into the guest. If the guest
3358 MCG_STATUS register reports that an MCE is in progress, KVM
3359 causes an KVM_EXIT_SHUTDOWN vmexit.
3361 Otherwise, if the MCE is a corrected error, KVM will just
3362 store it in the corresponding bank (provided this bank is
3363 not holding a previously reported uncorrected error).
3365 4.107 KVM_S390_GET_CMMA_BITS
3367 Capability: KVM_CAP_S390_CMMA_MIGRATION
3370 Parameters: struct kvm_s390_cmma_log (in, out)
3371 Returns: 0 on success, a negative value on error
3373 This ioctl is used to get the values of the CMMA bits on the s390
3374 architecture. It is meant to be used in two scenarios:
3375 - During live migration to save the CMMA values. Live migration needs
3376 to be enabled via the KVM_REQ_START_MIGRATION VM property.
3377 - To non-destructively peek at the CMMA values, with the flag
3378 KVM_S390_CMMA_PEEK set.
3380 The ioctl takes parameters via the kvm_s390_cmma_log struct. The desired
3381 values are written to a buffer whose location is indicated via the "values"
3382 member in the kvm_s390_cmma_log struct. The values in the input struct are
3383 also updated as needed.
3384 Each CMMA value takes up one byte.
3386 struct kvm_s390_cmma_log {
3397 start_gfn is the number of the first guest frame whose CMMA values are
3400 count is the length of the buffer in bytes,
3402 values points to the buffer where the result will be written to.
3404 If count is greater than KVM_S390_SKEYS_MAX, then it is considered to be
3405 KVM_S390_SKEYS_MAX. KVM_S390_SKEYS_MAX is re-used for consistency with
3408 The result is written in the buffer pointed to by the field values, and
3409 the values of the input parameter are updated as follows.
3411 Depending on the flags, different actions are performed. The only
3412 supported flag so far is KVM_S390_CMMA_PEEK.
3414 The default behaviour if KVM_S390_CMMA_PEEK is not set is:
3415 start_gfn will indicate the first page frame whose CMMA bits were dirty.
3416 It is not necessarily the same as the one passed as input, as clean pages
3419 count will indicate the number of bytes actually written in the buffer.
3420 It can (and very often will) be smaller than the input value, since the
3421 buffer is only filled until 16 bytes of clean values are found (which
3422 are then not copied in the buffer). Since a CMMA migration block needs
3423 the base address and the length, for a total of 16 bytes, we will send
3424 back some clean data if there is some dirty data afterwards, as long as
3425 the size of the clean data does not exceed the size of the header. This
3426 allows to minimize the amount of data to be saved or transferred over
3427 the network at the expense of more roundtrips to userspace. The next
3428 invocation of the ioctl will skip over all the clean values, saving
3429 potentially more than just the 16 bytes we found.
3431 If KVM_S390_CMMA_PEEK is set:
3432 the existing storage attributes are read even when not in migration
3433 mode, and no other action is performed;
3435 the output start_gfn will be equal to the input start_gfn,
3437 the output count will be equal to the input count, except if the end of
3438 memory has been reached.
3441 the field "remaining" will indicate the total number of dirty CMMA values
3442 still remaining, or 0 if KVM_S390_CMMA_PEEK is set and migration mode is
3447 values points to the userspace buffer where the result will be stored.
3449 This ioctl can fail with -ENOMEM if not enough memory can be allocated to
3450 complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
3451 KVM_S390_CMMA_PEEK is not set but migration mode was not enabled, with
3452 -EFAULT if the userspace address is invalid or if no page table is
3453 present for the addresses (e.g. when using hugepages).
3455 4.108 KVM_S390_SET_CMMA_BITS
3457 Capability: KVM_CAP_S390_CMMA_MIGRATION
3460 Parameters: struct kvm_s390_cmma_log (in)
3461 Returns: 0 on success, a negative value on error
3463 This ioctl is used to set the values of the CMMA bits on the s390
3464 architecture. It is meant to be used during live migration to restore
3465 the CMMA values, but there are no restrictions on its use.
3466 The ioctl takes parameters via the kvm_s390_cmma_values struct.
3467 Each CMMA value takes up one byte.
3469 struct kvm_s390_cmma_log {
3480 start_gfn indicates the starting guest frame number,
3482 count indicates how many values are to be considered in the buffer,
3484 flags is not used and must be 0.
3486 mask indicates which PGSTE bits are to be considered.
3488 remaining is not used.
3490 values points to the buffer in userspace where to store the values.
3492 This ioctl can fail with -ENOMEM if not enough memory can be allocated to
3493 complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
3494 the count field is too large (e.g. more than KVM_S390_CMMA_SIZE_MAX) or
3495 if the flags field was not 0, with -EFAULT if the userspace address is
3496 invalid, if invalid pages are written to (e.g. after the end of memory)
3497 or if no page table is present for the addresses (e.g. when using
3500 4.109 KVM_PPC_GET_CPU_CHAR
3502 Capability: KVM_CAP_PPC_GET_CPU_CHAR
3503 Architectures: powerpc
3505 Parameters: struct kvm_ppc_cpu_char (out)
3506 Returns: 0 on successful completion
3507 -EFAULT if struct kvm_ppc_cpu_char cannot be written
3509 This ioctl gives userspace information about certain characteristics
3510 of the CPU relating to speculative execution of instructions and
3511 possible information leakage resulting from speculative execution (see
3512 CVE-2017-5715, CVE-2017-5753 and CVE-2017-5754). The information is
3513 returned in struct kvm_ppc_cpu_char, which looks like this:
3515 struct kvm_ppc_cpu_char {
3516 __u64 character; /* characteristics of the CPU */
3517 __u64 behaviour; /* recommended software behaviour */
3518 __u64 character_mask; /* valid bits in character */
3519 __u64 behaviour_mask; /* valid bits in behaviour */
3522 For extensibility, the character_mask and behaviour_mask fields
3523 indicate which bits of character and behaviour have been filled in by
3524 the kernel. If the set of defined bits is extended in future then
3525 userspace will be able to tell whether it is running on a kernel that
3526 knows about the new bits.
3528 The character field describes attributes of the CPU which can help
3529 with preventing inadvertent information disclosure - specifically,
3530 whether there is an instruction to flash-invalidate the L1 data cache
3531 (ori 30,30,0 or mtspr SPRN_TRIG2,rN), whether the L1 data cache is set
3532 to a mode where entries can only be used by the thread that created
3533 them, whether the bcctr[l] instruction prevents speculation, and
3534 whether a speculation barrier instruction (ori 31,31,0) is provided.
3536 The behaviour field describes actions that software should take to
3537 prevent inadvertent information disclosure, and thus describes which
3538 vulnerabilities the hardware is subject to; specifically whether the
3539 L1 data cache should be flushed when returning to user mode from the
3540 kernel, and whether a speculation barrier should be placed between an
3541 array bounds check and the array access.
3543 These fields use the same bit definitions as the new
3544 H_GET_CPU_CHARACTERISTICS hypercall.
3546 4.110 KVM_MEMORY_ENCRYPT_OP
3551 Parameters: an opaque platform specific structure (in/out)
3552 Returns: 0 on success; -1 on error
3554 If the platform supports creating encrypted VMs then this ioctl can be used
3555 for issuing platform-specific memory encryption commands to manage those
3558 Currently, this ioctl is used for issuing Secure Encrypted Virtualization
3559 (SEV) commands on AMD Processors. The SEV commands are defined in
3560 Documentation/virtual/kvm/amd-memory-encryption.rst.
3562 4.111 KVM_MEMORY_ENCRYPT_REG_REGION
3567 Parameters: struct kvm_enc_region (in)
3568 Returns: 0 on success; -1 on error
3570 This ioctl can be used to register a guest memory region which may
3571 contain encrypted data (e.g. guest RAM, SMRAM etc).
3573 It is used in the SEV-enabled guest. When encryption is enabled, a guest
3574 memory region may contain encrypted data. The SEV memory encryption
3575 engine uses a tweak such that two identical plaintext pages, each at
3576 different locations will have differing ciphertexts. So swapping or
3577 moving ciphertext of those pages will not result in plaintext being
3578 swapped. So relocating (or migrating) physical backing pages for the SEV
3579 guest will require some additional steps.
3581 Note: The current SEV key management spec does not provide commands to
3582 swap or migrate (move) ciphertext pages. Hence, for now we pin the guest
3583 memory region registered with the ioctl.
3585 4.112 KVM_MEMORY_ENCRYPT_UNREG_REGION
3590 Parameters: struct kvm_enc_region (in)
3591 Returns: 0 on success; -1 on error
3593 This ioctl can be used to unregister the guest memory region registered
3594 with KVM_MEMORY_ENCRYPT_REG_REGION ioctl above.
3596 4.113 KVM_HYPERV_EVENTFD
3598 Capability: KVM_CAP_HYPERV_EVENTFD
3601 Parameters: struct kvm_hyperv_eventfd (in)
3603 This ioctl (un)registers an eventfd to receive notifications from the guest on
3604 the specified Hyper-V connection id through the SIGNAL_EVENT hypercall, without
3605 causing a user exit. SIGNAL_EVENT hypercall with non-zero event flag number
3606 (bits 24-31) still triggers a KVM_EXIT_HYPERV_HCALL user exit.
3608 struct kvm_hyperv_eventfd {
3615 The conn_id field should fit within 24 bits:
3617 #define KVM_HYPERV_CONN_ID_MASK 0x00ffffff
3619 The acceptable values for the flags field are:
3621 #define KVM_HYPERV_EVENTFD_DEASSIGN (1 << 0)
3623 Returns: 0 on success,
3624 -EINVAL if conn_id or flags is outside the allowed range
3625 -ENOENT on deassign if the conn_id isn't registered
3626 -EEXIST on assign if the conn_id is already registered
3628 4.114 KVM_GET_NESTED_STATE
3630 Capability: KVM_CAP_NESTED_STATE
3633 Parameters: struct kvm_nested_state (in/out)
3634 Returns: 0 on success, -1 on error
3636 E2BIG: the total state size (including the fixed-size part of struct
3637 kvm_nested_state) exceeds the value of 'size' specified by
3638 the user; the size required will be written into size.
3640 struct kvm_nested_state {
3645 struct kvm_vmx_nested_state vmx;
3646 struct kvm_svm_nested_state svm;
3652 #define KVM_STATE_NESTED_GUEST_MODE 0x00000001
3653 #define KVM_STATE_NESTED_RUN_PENDING 0x00000002
3655 #define KVM_STATE_NESTED_SMM_GUEST_MODE 0x00000001
3656 #define KVM_STATE_NESTED_SMM_VMXON 0x00000002
3658 struct kvm_vmx_nested_state {
3667 This ioctl copies the vcpu's nested virtualization state from the kernel to
3670 The maximum size of the state, including the fixed-size part of struct
3671 kvm_nested_state, can be retrieved by passing KVM_CAP_NESTED_STATE to
3672 the KVM_CHECK_EXTENSION ioctl().
3674 4.115 KVM_SET_NESTED_STATE
3676 Capability: KVM_CAP_NESTED_STATE
3679 Parameters: struct kvm_nested_state (in)
3680 Returns: 0 on success, -1 on error
3682 This copies the vcpu's kvm_nested_state struct from userspace to the kernel. For
3683 the definition of struct kvm_nested_state, see KVM_GET_NESTED_STATE.
3685 5. The kvm_run structure
3686 ------------------------
3688 Application code obtains a pointer to the kvm_run structure by
3689 mmap()ing a vcpu fd. From that point, application code can control
3690 execution by changing fields in kvm_run prior to calling the KVM_RUN
3691 ioctl, and obtain information about the reason KVM_RUN returned by
3692 looking up structure members.
3696 __u8 request_interrupt_window;
3698 Request that KVM_RUN return when it becomes possible to inject external
3699 interrupts into the guest. Useful in conjunction with KVM_INTERRUPT.
3701 __u8 immediate_exit;
3703 This field is polled once when KVM_RUN starts; if non-zero, KVM_RUN
3704 exits immediately, returning -EINTR. In the common scenario where a
3705 signal is used to "kick" a VCPU out of KVM_RUN, this field can be used
3706 to avoid usage of KVM_SET_SIGNAL_MASK, which has worse scalability.
3707 Rather than blocking the signal outside KVM_RUN, userspace can set up
3708 a signal handler that sets run->immediate_exit to a non-zero value.
3710 This field is ignored if KVM_CAP_IMMEDIATE_EXIT is not available.
3717 When KVM_RUN has returned successfully (return value 0), this informs
3718 application code why KVM_RUN has returned. Allowable values for this
3719 field are detailed below.
3721 __u8 ready_for_interrupt_injection;
3723 If request_interrupt_window has been specified, this field indicates
3724 an interrupt can be injected now with KVM_INTERRUPT.
3728 The value of the current interrupt flag. Only valid if in-kernel
3729 local APIC is not used.
3733 More architecture-specific flags detailing state of the VCPU that may
3734 affect the device's behavior. The only currently defined flag is
3735 KVM_RUN_X86_SMM, which is valid on x86 machines and is set if the
3736 VCPU is in system management mode.
3738 /* in (pre_kvm_run), out (post_kvm_run) */
3741 The value of the cr8 register. Only valid if in-kernel local APIC is
3742 not used. Both input and output.
3746 The value of the APIC BASE msr. Only valid if in-kernel local
3747 APIC is not used. Both input and output.
3750 /* KVM_EXIT_UNKNOWN */
3752 __u64 hardware_exit_reason;
3755 If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown
3756 reasons. Further architecture-specific information is available in
3757 hardware_exit_reason.
3759 /* KVM_EXIT_FAIL_ENTRY */
3761 __u64 hardware_entry_failure_reason;
3764 If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due
3765 to unknown reasons. Further architecture-specific information is
3766 available in hardware_entry_failure_reason.
3768 /* KVM_EXIT_EXCEPTION */
3778 #define KVM_EXIT_IO_IN 0
3779 #define KVM_EXIT_IO_OUT 1
3781 __u8 size; /* bytes */
3784 __u64 data_offset; /* relative to kvm_run start */
3787 If exit_reason is KVM_EXIT_IO, then the vcpu has
3788 executed a port I/O instruction which could not be satisfied by kvm.
3789 data_offset describes where the data is located (KVM_EXIT_IO_OUT) or
3790 where kvm expects application code to place the data for the next
3791 KVM_RUN invocation (KVM_EXIT_IO_IN). Data format is a packed array.
3793 /* KVM_EXIT_DEBUG */
3795 struct kvm_debug_exit_arch arch;
3798 If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event
3799 for which architecture specific information is returned.
3809 If exit_reason is KVM_EXIT_MMIO, then the vcpu has
3810 executed a memory-mapped I/O instruction which could not be satisfied
3811 by kvm. The 'data' member contains the written data if 'is_write' is
3812 true, and should be filled by application code otherwise.
3814 The 'data' member contains, in its first 'len' bytes, the value as it would
3815 appear if the VCPU performed a load or store of the appropriate width directly
3818 NOTE: For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR and
3819 KVM_EXIT_EPR the corresponding
3820 operations are complete (and guest state is consistent) only after userspace
3821 has re-entered the kernel with KVM_RUN. The kernel side will first finish
3822 incomplete operations and then check for pending signals. Userspace
3823 can re-enter the guest with an unmasked signal pending to complete
3826 /* KVM_EXIT_HYPERCALL */
3835 Unused. This was once used for 'hypercall to userspace'. To implement
3836 such functionality, use KVM_EXIT_IO (x86) or KVM_EXIT_MMIO (all except s390).
3837 Note KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO.
3839 /* KVM_EXIT_TPR_ACCESS */
3846 To be documented (KVM_TPR_ACCESS_REPORTING).
3848 /* KVM_EXIT_S390_SIEIC */
3851 __u64 mask; /* psw upper half */
3852 __u64 addr; /* psw lower half */
3859 /* KVM_EXIT_S390_RESET */
3860 #define KVM_S390_RESET_POR 1
3861 #define KVM_S390_RESET_CLEAR 2
3862 #define KVM_S390_RESET_SUBSYSTEM 4
3863 #define KVM_S390_RESET_CPU_INIT 8
3864 #define KVM_S390_RESET_IPL 16
3865 __u64 s390_reset_flags;
3869 /* KVM_EXIT_S390_UCONTROL */
3871 __u64 trans_exc_code;
3875 s390 specific. A page fault has occurred for a user controlled virtual
3876 machine (KVM_VM_S390_UNCONTROL) on it's host page table that cannot be
3877 resolved by the kernel.
3878 The program code and the translation exception code that were placed
3879 in the cpu's lowcore are presented here as defined by the z Architecture
3880 Principles of Operation Book in the Chapter for Dynamic Address Translation
3890 Deprecated - was used for 440 KVM.
3897 MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch
3898 hypercalls and exit with this exit struct that contains all the guest gprs.
3900 If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall.
3901 Userspace can now handle the hypercall and when it's done modify the gprs as
3902 necessary. Upon guest entry all guest GPRs will then be replaced by the values
3905 /* KVM_EXIT_PAPR_HCALL */
3912 This is used on 64-bit PowerPC when emulating a pSeries partition,
3913 e.g. with the 'pseries' machine type in qemu. It occurs when the
3914 guest does a hypercall using the 'sc 1' instruction. The 'nr' field
3915 contains the hypercall number (from the guest R3), and 'args' contains
3916 the arguments (from the guest R4 - R12). Userspace should put the
3917 return code in 'ret' and any extra returned values in args[].
3918 The possible hypercalls are defined in the Power Architecture Platform
3919 Requirements (PAPR) document available from www.power.org (free
3920 developer registration required to access it).
3922 /* KVM_EXIT_S390_TSCH */
3924 __u16 subchannel_id;
3925 __u16 subchannel_nr;
3932 s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled
3933 and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O
3934 interrupt for the target subchannel has been dequeued and subchannel_id,
3935 subchannel_nr, io_int_parm and io_int_word contain the parameters for that
3936 interrupt. ipb is needed for instruction parameter decoding.
3943 On FSL BookE PowerPC chips, the interrupt controller has a fast patch
3944 interrupt acknowledge path to the core. When the core successfully
3945 delivers an interrupt, it automatically populates the EPR register with
3946 the interrupt vector number and acknowledges the interrupt inside
3947 the interrupt controller.
3949 In case the interrupt controller lives in user space, we need to do
3950 the interrupt acknowledge cycle through it to fetch the next to be
3951 delivered interrupt vector using this exit.
3953 It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an
3954 external interrupt has just been delivered into the guest. User space
3955 should put the acknowledged interrupt vector into the 'epr' field.
3957 /* KVM_EXIT_SYSTEM_EVENT */
3959 #define KVM_SYSTEM_EVENT_SHUTDOWN 1
3960 #define KVM_SYSTEM_EVENT_RESET 2
3961 #define KVM_SYSTEM_EVENT_CRASH 3
3966 If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered
3967 a system-level event using some architecture specific mechanism (hypercall
3968 or some special instruction). In case of ARM/ARM64, this is triggered using
3969 HVC instruction based PSCI call from the vcpu. The 'type' field describes
3970 the system-level event type. The 'flags' field describes architecture
3971 specific flags for the system-level event.
3973 Valid values for 'type' are:
3974 KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the
3975 VM. Userspace is not obliged to honour this, and if it does honour
3976 this does not need to destroy the VM synchronously (ie it may call
3977 KVM_RUN again before shutdown finally occurs).
3978 KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM.
3979 As with SHUTDOWN, userspace can choose to ignore the request, or
3980 to schedule the reset to occur in the future and may call KVM_RUN again.
3981 KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest
3982 has requested a crash condition maintenance. Userspace can choose
3983 to ignore the request, or to gather VM memory core dump and/or
3984 reset/shutdown of the VM.
3986 /* KVM_EXIT_IOAPIC_EOI */
3991 Indicates that the VCPU's in-kernel local APIC received an EOI for a
3992 level-triggered IOAPIC interrupt. This exit only triggers when the
3993 IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled);
3994 the userspace IOAPIC should process the EOI and retrigger the interrupt if
3995 it is still asserted. Vector is the LAPIC interrupt vector for which the
3998 struct kvm_hyperv_exit {
3999 #define KVM_EXIT_HYPERV_SYNIC 1
4000 #define KVM_EXIT_HYPERV_HCALL 2
4018 /* KVM_EXIT_HYPERV */
4019 struct kvm_hyperv_exit hyperv;
4020 Indicates that the VCPU exits into userspace to process some tasks
4021 related to Hyper-V emulation.
4022 Valid values for 'type' are:
4023 KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about
4024 Hyper-V SynIC state change. Notification is used to remap SynIC
4025 event/message pages and to enable/disable SynIC messages/events processing
4028 /* Fix the size of the union. */
4033 * shared registers between kvm and userspace.
4034 * kvm_valid_regs specifies the register classes set by the host
4035 * kvm_dirty_regs specified the register classes dirtied by userspace
4036 * struct kvm_sync_regs is architecture specific, as well as the
4037 * bits for kvm_valid_regs and kvm_dirty_regs
4039 __u64 kvm_valid_regs;
4040 __u64 kvm_dirty_regs;
4042 struct kvm_sync_regs regs;
4043 char padding[SYNC_REGS_SIZE_BYTES];
4046 If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access
4047 certain guest registers without having to call SET/GET_*REGS. Thus we can
4048 avoid some system call overhead if userspace has to handle the exit.
4049 Userspace can query the validity of the structure by checking
4050 kvm_valid_regs for specific bits. These bits are architecture specific
4051 and usually define the validity of a groups of registers. (e.g. one bit
4052 for general purpose registers)
4054 Please note that the kernel is allowed to use the kvm_run structure as the
4055 primary storage for certain register types. Therefore, the kernel may use the
4056 values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set.
4062 6. Capabilities that can be enabled on vCPUs
4063 --------------------------------------------
4065 There are certain capabilities that change the behavior of the virtual CPU or
4066 the virtual machine when enabled. To enable them, please see section 4.37.
4067 Below you can find a list of capabilities and what their effect on the vCPU or
4068 the virtual machine is when enabling them.
4070 The following information is provided along with the description:
4072 Architectures: which instruction set architectures provide this ioctl.
4073 x86 includes both i386 and x86_64.
4075 Target: whether this is a per-vcpu or per-vm capability.
4077 Parameters: what parameters are accepted by the capability.
4079 Returns: the return value. General error numbers (EBADF, ENOMEM, EINVAL)
4080 are not detailed, but errors with specific meanings are.
4088 Returns: 0 on success; -1 on error
4090 This capability enables interception of OSI hypercalls that otherwise would
4091 be treated as normal system calls to be injected into the guest. OSI hypercalls
4092 were invented by Mac-on-Linux to have a standardized communication mechanism
4093 between the guest and the host.
4095 When this capability is enabled, KVM_EXIT_OSI can occur.
4098 6.2 KVM_CAP_PPC_PAPR
4103 Returns: 0 on success; -1 on error
4105 This capability enables interception of PAPR hypercalls. PAPR hypercalls are
4106 done using the hypercall instruction "sc 1".
4108 It also sets the guest privilege level to "supervisor" mode. Usually the guest
4109 runs in "hypervisor" privilege mode with a few missing features.
4111 In addition to the above, it changes the semantics of SDR1. In this mode, the
4112 HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the
4113 HTAB invisible to the guest.
4115 When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur.
4122 Parameters: args[0] is the address of a struct kvm_config_tlb
4123 Returns: 0 on success; -1 on error
4125 struct kvm_config_tlb {
4132 Configures the virtual CPU's TLB array, establishing a shared memory area
4133 between userspace and KVM. The "params" and "array" fields are userspace
4134 addresses of mmu-type-specific data structures. The "array_len" field is an
4135 safety mechanism, and should be set to the size in bytes of the memory that
4136 userspace has reserved for the array. It must be at least the size dictated
4137 by "mmu_type" and "params".
4139 While KVM_RUN is active, the shared region is under control of KVM. Its
4140 contents are undefined, and any modification by userspace results in
4141 boundedly undefined behavior.
4143 On return from KVM_RUN, the shared region will reflect the current state of
4144 the guest's TLB. If userspace makes any changes, it must call KVM_DIRTY_TLB
4145 to tell KVM which entries have been changed, prior to calling KVM_RUN again
4148 For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV:
4149 - The "params" field is of type "struct kvm_book3e_206_tlb_params".
4150 - The "array" field points to an array of type "struct
4151 kvm_book3e_206_tlb_entry".
4152 - The array consists of all entries in the first TLB, followed by all
4153 entries in the second TLB.
4154 - Within a TLB, entries are ordered first by increasing set number. Within a
4155 set, entries are ordered by way (increasing ESEL).
4156 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1)
4157 where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value.
4158 - The tsize field of mas1 shall be set to 4K on TLB0, even though the
4159 hardware ignores this value for TLB0.
4161 6.4 KVM_CAP_S390_CSS_SUPPORT
4166 Returns: 0 on success; -1 on error
4168 This capability enables support for handling of channel I/O instructions.
4170 TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are
4171 handled in-kernel, while the other I/O instructions are passed to userspace.
4173 When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST
4174 SUBCHANNEL intercepts.
4176 Note that even though this capability is enabled per-vcpu, the complete
4177 virtual machine is affected.
4183 Parameters: args[0] defines whether the proxy facility is active
4184 Returns: 0 on success; -1 on error
4186 This capability enables or disables the delivery of interrupts through the
4187 external proxy facility.
4189 When enabled (args[0] != 0), every time the guest gets an external interrupt
4190 delivered, it automatically exits into user space with a KVM_EXIT_EPR exit
4191 to receive the topmost interrupt vector.
4193 When disabled (args[0] == 0), behavior is as if this facility is unsupported.
4195 When this capability is enabled, KVM_EXIT_EPR can occur.
4197 6.6 KVM_CAP_IRQ_MPIC
4200 Parameters: args[0] is the MPIC device fd
4201 args[1] is the MPIC CPU number for this vcpu
4203 This capability connects the vcpu to an in-kernel MPIC device.
4205 6.7 KVM_CAP_IRQ_XICS
4209 Parameters: args[0] is the XICS device fd
4210 args[1] is the XICS CPU number (server ID) for this vcpu
4212 This capability connects the vcpu to an in-kernel XICS device.
4214 6.8 KVM_CAP_S390_IRQCHIP
4220 This capability enables the in-kernel irqchip for s390. Please refer to
4221 "4.24 KVM_CREATE_IRQCHIP" for details.
4223 6.9 KVM_CAP_MIPS_FPU
4227 Parameters: args[0] is reserved for future use (should be 0).
4229 This capability allows the use of the host Floating Point Unit by the guest. It
4230 allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is
4231 done the KVM_REG_MIPS_FPR_* and KVM_REG_MIPS_FCR_* registers can be accessed
4232 (depending on the current guest FPU register mode), and the Status.FR,
4233 Config5.FRE bits are accessible via the KVM API and also from the guest,
4234 depending on them being supported by the FPU.
4236 6.10 KVM_CAP_MIPS_MSA
4240 Parameters: args[0] is reserved for future use (should be 0).
4242 This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest.
4243 It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest.
4244 Once this is done the KVM_REG_MIPS_VEC_* and KVM_REG_MIPS_MSA_* registers can be
4245 accessed, and the Config5.MSAEn bit is accessible via the KVM API and also from
4248 6.74 KVM_CAP_SYNC_REGS
4249 Architectures: s390, x86
4250 Target: s390: always enabled, x86: vcpu
4252 Returns: x86: KVM_CHECK_EXTENSION returns a bit-array indicating which register
4253 sets are supported (bitfields defined in arch/x86/include/uapi/asm/kvm.h).
4255 As described above in the kvm_sync_regs struct info in section 5 (kvm_run):
4256 KVM_CAP_SYNC_REGS "allow[s] userspace to access certain guest registers
4257 without having to call SET/GET_*REGS". This reduces overhead by eliminating
4258 repeated ioctl calls for setting and/or getting register values. This is
4259 particularly important when userspace is making synchronous guest state
4260 modifications, e.g. when emulating and/or intercepting instructions in
4263 For s390 specifics, please refer to the source code.
4266 - the register sets to be copied out to kvm_run are selectable
4267 by userspace (rather that all sets being copied out for every exit).
4268 - vcpu_events are available in addition to regs and sregs.
4270 For x86, the 'kvm_valid_regs' field of struct kvm_run is overloaded to
4271 function as an input bit-array field set by userspace to indicate the
4272 specific register sets to be copied out on the next exit.
4274 To indicate when userspace has modified values that should be copied into
4275 the vCPU, the all architecture bitarray field, 'kvm_dirty_regs' must be set.
4276 This is done using the same bitflags as for the 'kvm_valid_regs' field.
4277 If the dirty bit is not set, then the register set values will not be copied
4278 into the vCPU even if they've been modified.
4280 Unused bitfields in the bitarrays must be set to zero.
4282 struct kvm_sync_regs {
4283 struct kvm_regs regs;
4284 struct kvm_sregs sregs;
4285 struct kvm_vcpu_events events;
4288 7. Capabilities that can be enabled on VMs
4289 ------------------------------------------
4291 There are certain capabilities that change the behavior of the virtual
4292 machine when enabled. To enable them, please see section 4.37. Below
4293 you can find a list of capabilities and what their effect on the VM
4294 is when enabling them.
4296 The following information is provided along with the description:
4298 Architectures: which instruction set architectures provide this ioctl.
4299 x86 includes both i386 and x86_64.
4301 Parameters: what parameters are accepted by the capability.
4303 Returns: the return value. General error numbers (EBADF, ENOMEM, EINVAL)
4304 are not detailed, but errors with specific meanings are.
4307 7.1 KVM_CAP_PPC_ENABLE_HCALL
4310 Parameters: args[0] is the sPAPR hcall number
4311 args[1] is 0 to disable, 1 to enable in-kernel handling
4313 This capability controls whether individual sPAPR hypercalls (hcalls)
4314 get handled by the kernel or not. Enabling or disabling in-kernel
4315 handling of an hcall is effective across the VM. On creation, an
4316 initial set of hcalls are enabled for in-kernel handling, which
4317 consists of those hcalls for which in-kernel handlers were implemented
4318 before this capability was implemented. If disabled, the kernel will
4319 not to attempt to handle the hcall, but will always exit to userspace
4320 to handle it. Note that it may not make sense to enable some and
4321 disable others of a group of related hcalls, but KVM does not prevent
4322 userspace from doing that.
4324 If the hcall number specified is not one that has an in-kernel
4325 implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL
4328 7.2 KVM_CAP_S390_USER_SIGP
4333 This capability controls which SIGP orders will be handled completely in user
4334 space. With this capability enabled, all fast orders will be handled completely
4340 - CONDITIONAL EMERGENCY SIGNAL
4342 All other orders will be handled completely in user space.
4344 Only privileged operation exceptions will be checked for in the kernel (or even
4345 in the hardware prior to interception). If this capability is not enabled, the
4346 old way of handling SIGP orders is used (partially in kernel and user space).
4348 7.3 KVM_CAP_S390_VECTOR_REGISTERS
4352 Returns: 0 on success, negative value on error
4354 Allows use of the vector registers introduced with z13 processor, and
4355 provides for the synchronization between host and user space. Will
4356 return -EINVAL if the machine does not support vectors.
4358 7.4 KVM_CAP_S390_USER_STSI
4363 This capability allows post-handlers for the STSI instruction. After
4364 initial handling in the kernel, KVM exits to user space with
4365 KVM_EXIT_S390_STSI to allow user space to insert further data.
4367 Before exiting to userspace, kvm handlers should fill in s390_stsi field of
4378 @addr - guest address of STSI SYSIB
4382 @ar - access register number
4384 KVM handlers should exit to userspace with rc = -EREMOTE.
4386 7.5 KVM_CAP_SPLIT_IRQCHIP
4389 Parameters: args[0] - number of routes reserved for userspace IOAPICs
4390 Returns: 0 on success, -1 on error
4392 Create a local apic for each processor in the kernel. This can be used
4393 instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the
4394 IOAPIC and PIC (and also the PIT, even though this has to be enabled
4397 This capability also enables in kernel routing of interrupt requests;
4398 when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are
4399 used in the IRQ routing table. The first args[0] MSI routes are reserved
4400 for the IOAPIC pins. Whenever the LAPIC receives an EOI for these routes,
4401 a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace.
4403 Fails if VCPU has already been created, or if the irqchip is already in the
4404 kernel (i.e. KVM_CREATE_IRQCHIP has already been called).
4411 Allows use of runtime-instrumentation introduced with zEC12 processor.
4412 Will return -EINVAL if the machine does not support runtime-instrumentation.
4413 Will return -EBUSY if a VCPU has already been created.
4415 7.7 KVM_CAP_X2APIC_API
4418 Parameters: args[0] - features that should be enabled
4419 Returns: 0 on success, -EINVAL when args[0] contains invalid features
4421 Valid feature flags in args[0] are
4423 #define KVM_X2APIC_API_USE_32BIT_IDS (1ULL << 0)
4424 #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK (1ULL << 1)
4426 Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of
4427 KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC,
4428 allowing the use of 32-bit APIC IDs. See KVM_CAP_X2APIC_API in their
4429 respective sections.
4431 KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work
4432 in logical mode or with more than 255 VCPUs. Otherwise, KVM treats 0xff
4433 as a broadcast even in x2APIC mode in order to support physical x2APIC
4434 without interrupt remapping. This is undesirable in logical mode,
4435 where 0xff represents CPUs 0-7 in cluster 0.
4437 7.8 KVM_CAP_S390_USER_INSTR0
4442 With this capability enabled, all illegal instructions 0x0000 (2 bytes) will
4443 be intercepted and forwarded to user space. User space can use this
4444 mechanism e.g. to realize 2-byte software breakpoints. The kernel will
4445 not inject an operating exception for these instructions, user space has
4446 to take care of that.
4448 This capability can be enabled dynamically even if VCPUs were already
4449 created and are running.
4455 Returns: 0 on success; -EINVAL if the machine does not support
4456 guarded storage; -EBUSY if a VCPU has already been created.
4458 Allows use of guarded storage for the KVM guest.
4460 7.10 KVM_CAP_S390_AIS
4465 Allow use of adapter-interruption suppression.
4466 Returns: 0 on success; -EBUSY if a VCPU has already been created.
4468 7.11 KVM_CAP_PPC_SMT
4471 Parameters: vsmt_mode, flags
4473 Enabling this capability on a VM provides userspace with a way to set
4474 the desired virtual SMT mode (i.e. the number of virtual CPUs per
4475 virtual core). The virtual SMT mode, vsmt_mode, must be a power of 2
4476 between 1 and 8. On POWER8, vsmt_mode must also be no greater than
4477 the number of threads per subcore for the host. Currently flags must
4478 be 0. A successful call to enable this capability will result in
4479 vsmt_mode being returned when the KVM_CAP_PPC_SMT capability is
4480 subsequently queried for the VM. This capability is only supported by
4481 HV KVM, and can only be set before any VCPUs have been created.
4482 The KVM_CAP_PPC_SMT_POSSIBLE capability indicates which virtual SMT
4483 modes are available.
4485 7.12 KVM_CAP_PPC_FWNMI
4490 With this capability a machine check exception in the guest address
4491 space will cause KVM to exit the guest with NMI exit reason. This
4492 enables QEMU to build error log and branch to guest kernel registered
4493 machine check handling routine. Without this capability KVM will
4494 branch to guests' 0x200 interrupt vector.
4496 7.13 KVM_CAP_X86_DISABLE_EXITS
4499 Parameters: args[0] defines which exits are disabled
4500 Returns: 0 on success, -EINVAL when args[0] contains invalid exits
4502 Valid bits in args[0] are
4504 #define KVM_X86_DISABLE_EXITS_MWAIT (1 << 0)
4505 #define KVM_X86_DISABLE_EXITS_HLT (1 << 1)
4507 Enabling this capability on a VM provides userspace with a way to no
4508 longer intercept some instructions for improved latency in some
4509 workloads, and is suggested when vCPUs are associated to dedicated
4510 physical CPUs. More bits can be added in the future; userspace can
4511 just pass the KVM_CHECK_EXTENSION result to KVM_ENABLE_CAP to disable
4514 Do not enable KVM_FEATURE_PV_UNHALT if you disable HLT exits.
4516 7.14 KVM_CAP_S390_HPAGE_1M
4520 Returns: 0 on success, -EINVAL if hpage module parameter was not set
4521 or cmma is enabled, or the VM has the KVM_VM_S390_UCONTROL
4524 With this capability the KVM support for memory backing with 1m pages
4525 through hugetlbfs can be enabled for a VM. After the capability is
4526 enabled, cmma can't be enabled anymore and pfmfi and the storage key
4527 interpretation are disabled. If cmma has already been enabled or the
4528 hpage module parameter is not set to 1, -EINVAL is returned.
4530 While it is generally possible to create a huge page backed VM without
4531 this capability, the VM will not be able to run.
4533 7.14 KVM_CAP_MSR_PLATFORM_INFO
4536 Parameters: args[0] whether feature should be enabled or not
4538 With this capability, a guest may read the MSR_PLATFORM_INFO MSR. Otherwise,
4539 a #GP would be raised when the guest tries to access. Currently, this
4540 capability does not enable write permissions of this MSR for the guest.
4542 8. Other capabilities.
4543 ----------------------
4545 This section lists capabilities that give information about other
4546 features of the KVM implementation.
4548 8.1 KVM_CAP_PPC_HWRNG
4552 This capability, if KVM_CHECK_EXTENSION indicates that it is
4553 available, means that that the kernel has an implementation of the
4554 H_RANDOM hypercall backed by a hardware random-number generator.
4555 If present, the kernel H_RANDOM handler can be enabled for guest use
4556 with the KVM_CAP_PPC_ENABLE_HCALL capability.
4558 8.2 KVM_CAP_HYPERV_SYNIC
4561 This capability, if KVM_CHECK_EXTENSION indicates that it is
4562 available, means that that the kernel has an implementation of the
4563 Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is
4564 used to support Windows Hyper-V based guest paravirt drivers(VMBus).
4566 In order to use SynIC, it has to be activated by setting this
4567 capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this
4568 will disable the use of APIC hardware virtualization even if supported
4569 by the CPU, as it's incompatible with SynIC auto-EOI behavior.
4571 8.3 KVM_CAP_PPC_RADIX_MMU
4575 This capability, if KVM_CHECK_EXTENSION indicates that it is
4576 available, means that that the kernel can support guests using the
4577 radix MMU defined in Power ISA V3.00 (as implemented in the POWER9
4580 8.4 KVM_CAP_PPC_HASH_MMU_V3
4584 This capability, if KVM_CHECK_EXTENSION indicates that it is
4585 available, means that that the kernel can support guests using the
4586 hashed page table MMU defined in Power ISA V3.00 (as implemented in
4587 the POWER9 processor), including in-memory segment tables.
4593 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
4594 it is available, means that full hardware assisted virtualization capabilities
4595 of the hardware are available for use through KVM. An appropriate
4596 KVM_VM_MIPS_* type must be passed to KVM_CREATE_VM to create a VM which
4599 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
4600 available, it means that the VM is using full hardware assisted virtualization
4601 capabilities of the hardware. This is useful to check after creating a VM with
4602 KVM_VM_MIPS_DEFAULT.
4604 The value returned by KVM_CHECK_EXTENSION should be compared against known
4605 values (see below). All other values are reserved. This is to allow for the
4606 possibility of other hardware assisted virtualization implementations which
4607 may be incompatible with the MIPS VZ ASE.
4609 0: The trap & emulate implementation is in use to run guest code in user
4610 mode. Guest virtual memory segments are rearranged to fit the guest in the
4611 user mode address space.
4613 1: The MIPS VZ ASE is in use, providing full hardware assisted
4614 virtualization, including standard guest virtual memory segments.
4620 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
4621 it is available, means that the trap & emulate implementation is available to
4622 run guest code in user mode, even if KVM_CAP_MIPS_VZ indicates that hardware
4623 assisted virtualisation is also available. KVM_VM_MIPS_TE (0) must be passed
4624 to KVM_CREATE_VM to create a VM which utilises it.
4626 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
4627 available, it means that the VM is using trap & emulate.
4629 8.7 KVM_CAP_MIPS_64BIT
4633 This capability indicates the supported architecture type of the guest, i.e. the
4634 supported register and address width.
4636 The values returned when this capability is checked by KVM_CHECK_EXTENSION on a
4637 kvm VM handle correspond roughly to the CP0_Config.AT register field, and should
4638 be checked specifically against known values (see below). All other values are
4641 0: MIPS32 or microMIPS32.
4642 Both registers and addresses are 32-bits wide.
4643 It will only be possible to run 32-bit guest code.
4645 1: MIPS64 or microMIPS64 with access only to 32-bit compatibility segments.
4646 Registers are 64-bits wide, but addresses are 32-bits wide.
4647 64-bit guest code may run but cannot access MIPS64 memory segments.
4648 It will also be possible to run 32-bit guest code.
4650 2: MIPS64 or microMIPS64 with access to all address segments.
4651 Both registers and addresses are 64-bits wide.
4652 It will be possible to run 64-bit or 32-bit guest code.
4654 8.9 KVM_CAP_ARM_USER_IRQ
4656 Architectures: arm, arm64
4657 This capability, if KVM_CHECK_EXTENSION indicates that it is available, means
4658 that if userspace creates a VM without an in-kernel interrupt controller, it
4659 will be notified of changes to the output level of in-kernel emulated devices,
4660 which can generate virtual interrupts, presented to the VM.
4661 For such VMs, on every return to userspace, the kernel
4662 updates the vcpu's run->s.regs.device_irq_level field to represent the actual
4663 output level of the device.
4665 Whenever kvm detects a change in the device output level, kvm guarantees at
4666 least one return to userspace before running the VM. This exit could either
4667 be a KVM_EXIT_INTR or any other exit event, like KVM_EXIT_MMIO. This way,
4668 userspace can always sample the device output level and re-compute the state of
4669 the userspace interrupt controller. Userspace should always check the state
4670 of run->s.regs.device_irq_level on every kvm exit.
4671 The value in run->s.regs.device_irq_level can represent both level and edge
4672 triggered interrupt signals, depending on the device. Edge triggered interrupt
4673 signals will exit to userspace with the bit in run->s.regs.device_irq_level
4674 set exactly once per edge signal.
4676 The field run->s.regs.device_irq_level is available independent of
4677 run->kvm_valid_regs or run->kvm_dirty_regs bits.
4679 If KVM_CAP_ARM_USER_IRQ is supported, the KVM_CHECK_EXTENSION ioctl returns a
4680 number larger than 0 indicating the version of this capability is implemented
4681 and thereby which bits in in run->s.regs.device_irq_level can signal values.
4683 Currently the following bits are defined for the device_irq_level bitmap:
4685 KVM_CAP_ARM_USER_IRQ >= 1:
4687 KVM_ARM_DEV_EL1_VTIMER - EL1 virtual timer
4688 KVM_ARM_DEV_EL1_PTIMER - EL1 physical timer
4689 KVM_ARM_DEV_PMU - ARM PMU overflow interrupt signal
4691 Future versions of kvm may implement additional events. These will get
4692 indicated by returning a higher number from KVM_CHECK_EXTENSION and will be
4695 8.10 KVM_CAP_PPC_SMT_POSSIBLE
4699 Querying this capability returns a bitmap indicating the possible
4700 virtual SMT modes that can be set using KVM_CAP_PPC_SMT. If bit N
4701 (counting from the right) is set, then a virtual SMT mode of 2^N is
4704 8.11 KVM_CAP_HYPERV_SYNIC2
4708 This capability enables a newer version of Hyper-V Synthetic interrupt
4709 controller (SynIC). The only difference with KVM_CAP_HYPERV_SYNIC is that KVM
4710 doesn't clear SynIC message and event flags pages when they are enabled by
4711 writing to the respective MSRs.
4713 8.12 KVM_CAP_HYPERV_VP_INDEX
4717 This capability indicates that userspace can load HV_X64_MSR_VP_INDEX msr. Its
4718 value is used to denote the target vcpu for a SynIC interrupt. For
4719 compatibilty, KVM initializes this msr to KVM's internal vcpu index. When this
4720 capability is absent, userspace can still query this msr's value.
4722 8.13 KVM_CAP_S390_AIS_MIGRATION
4727 This capability indicates if the flic device will be able to get/set the
4728 AIS states for migration via the KVM_DEV_FLIC_AISM_ALL attribute and allows
4729 to discover this without having to create a flic device.
4731 8.14 KVM_CAP_S390_PSW
4735 This capability indicates that the PSW is exposed via the kvm_run structure.
4737 8.15 KVM_CAP_S390_GMAP
4741 This capability indicates that the user space memory used as guest mapping can
4742 be anywhere in the user memory address space, as long as the memory slots are
4743 aligned and sized to a segment (1MB) boundary.
4745 8.16 KVM_CAP_S390_COW
4749 This capability indicates that the user space memory used as guest mapping can
4750 use copy-on-write semantics as well as dirty pages tracking via read-only page
4753 8.17 KVM_CAP_S390_BPB
4757 This capability indicates that kvm will implement the interfaces to handle
4758 reset, migration and nested KVM for branch prediction blocking. The stfle
4759 facility 82 should not be provided to the guest without this capability.
4761 8.18 KVM_CAP_HYPERV_TLBFLUSH
4765 This capability indicates that KVM supports paravirtualized Hyper-V TLB Flush
4767 HvFlushVirtualAddressSpace, HvFlushVirtualAddressSpaceEx,
4768 HvFlushVirtualAddressList, HvFlushVirtualAddressListEx.
4770 8.19 KVM_CAP_ARM_INJECT_SERROR_ESR
4772 Architectures: arm, arm64
4774 This capability indicates that userspace can specify (via the
4775 KVM_SET_VCPU_EVENTS ioctl) the syndrome value reported to the guest when it
4776 takes a virtual SError interrupt exception.
4777 If KVM advertises this capability, userspace can only specify the ISS field for
4778 the ESR syndrome. Other parts of the ESR, such as the EC are generated by the
4779 CPU when the exception is taken. If this virtual SError is taken to EL1 using
4780 AArch64, this value will be reported in the ISS field of ESR_ELx.
4782 See KVM_CAP_VCPU_EVENTS for more details.