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).
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
31 The kvm API is centered around file descriptors. An initial
32 open("/dev/kvm") obtains a handle to the kvm subsystem; this handle
33 can be used to issue system ioctls. A KVM_CREATE_VM ioctl on this
34 handle will create a VM file descriptor which can be used to issue VM
35 ioctls. A KVM_CREATE_VCPU ioctl on a VM fd will create a virtual cpu
36 and return a file descriptor pointing to it. Finally, ioctls on a vcpu
37 fd can be used to control the vcpu, including the important task of
38 actually running guest code.
40 In general file descriptors can be migrated among processes by means
41 of fork() and the SCM_RIGHTS facility of unix domain socket. These
42 kinds of tricks are explicitly not supported by kvm. While they will
43 not cause harm to the host, their actual behavior is not guaranteed by
44 the API. The only supported use is one virtual machine per process,
45 and one vcpu per thread.
51 As of Linux 2.6.22, the KVM ABI has been stabilized: no backward
52 incompatible change are allowed. However, there is an extension
53 facility that allows backward-compatible extensions to the API to be
56 The extension mechanism is not based on the Linux version number.
57 Instead, kvm defines extension identifiers and a facility to query
58 whether a particular extension identifier is available. If it is, a
59 set of ioctls is available for application use.
65 This section describes ioctls that can be used to control kvm guests.
66 For each ioctl, the following information is provided along with a
69 Capability: which KVM extension provides this ioctl. Can be 'basic',
70 which means that is will be provided by any kernel that supports
71 API version 12 (see section 4.1), a KVM_CAP_xyz constant, which
72 means availability needs to be checked with KVM_CHECK_EXTENSION
73 (see section 4.4), or 'none' which means that while not all kernels
74 support this ioctl, there's no capability bit to check its
75 availability: for kernels that don't support the ioctl,
76 the ioctl returns -ENOTTY.
78 Architectures: which instruction set architectures provide this ioctl.
79 x86 includes both i386 and x86_64.
81 Type: system, vm, or vcpu.
83 Parameters: what parameters are accepted by the ioctl.
85 Returns: the return value. General error numbers (EBADF, ENOMEM, EINVAL)
86 are not detailed, but errors with specific meanings are.
89 4.1 KVM_GET_API_VERSION
95 Returns: the constant KVM_API_VERSION (=12)
97 This identifies the API version as the stable kvm API. It is not
98 expected that this number will change. However, Linux 2.6.20 and
99 2.6.21 report earlier versions; these are not documented and not
100 supported. Applications should refuse to run if KVM_GET_API_VERSION
101 returns a value other than 12. If this check passes, all ioctls
102 described as 'basic' will be available.
110 Parameters: machine type identifier (KVM_VM_*)
111 Returns: a VM fd that can be used to control the new virtual machine.
113 The new VM has no virtual cpus and no memory.
114 You probably want to use 0 as machine type.
116 In order to create user controlled virtual machines on S390, check
117 KVM_CAP_S390_UCONTROL and use the flag KVM_VM_S390_UCONTROL as
118 privileged user (CAP_SYS_ADMIN).
120 To use hardware assisted virtualization on MIPS (VZ ASE) rather than
121 the default trap & emulate implementation (which changes the virtual
122 memory layout to fit in user mode), check KVM_CAP_MIPS_VZ and use the
126 4.3 KVM_GET_MSR_INDEX_LIST, KVM_GET_MSR_FEATURE_INDEX_LIST
128 Capability: basic, KVM_CAP_GET_MSR_FEATURES for KVM_GET_MSR_FEATURE_INDEX_LIST
131 Parameters: struct kvm_msr_list (in/out)
132 Returns: 0 on success; -1 on error
134 EFAULT: the msr index list cannot be read from or written to
135 E2BIG: the msr index list is to be to fit in the array specified by
138 struct kvm_msr_list {
139 __u32 nmsrs; /* number of msrs in entries */
143 The user fills in the size of the indices array in nmsrs, and in return
144 kvm adjusts nmsrs to reflect the actual number of msrs and fills in the
145 indices array with their numbers.
147 KVM_GET_MSR_INDEX_LIST returns the guest msrs that are supported. The list
148 varies by kvm version and host processor, but does not change otherwise.
150 Note: if kvm indicates supports MCE (KVM_CAP_MCE), then the MCE bank MSRs are
151 not returned in the MSR list, as different vcpus can have a different number
152 of banks, as set via the KVM_X86_SETUP_MCE ioctl.
154 KVM_GET_MSR_FEATURE_INDEX_LIST returns the list of MSRs that can be passed
155 to the KVM_GET_MSRS system ioctl. This lets userspace probe host capabilities
156 and processor features that are exposed via MSRs (e.g., VMX capabilities).
157 This list also varies by kvm version and host processor, but does not change
161 4.4 KVM_CHECK_EXTENSION
163 Capability: basic, KVM_CAP_CHECK_EXTENSION_VM for vm ioctl
165 Type: system ioctl, vm ioctl
166 Parameters: extension identifier (KVM_CAP_*)
167 Returns: 0 if unsupported; 1 (or some other positive integer) if supported
169 The API allows the application to query about extensions to the core
170 kvm API. Userspace passes an extension identifier (an integer) and
171 receives an integer that describes the extension availability.
172 Generally 0 means no and 1 means yes, but some extensions may report
173 additional information in the integer return value.
175 Based on their initialization different VMs may have different capabilities.
176 It is thus encouraged to use the vm ioctl to query for capabilities (available
177 with KVM_CAP_CHECK_EXTENSION_VM on the vm fd)
179 4.5 KVM_GET_VCPU_MMAP_SIZE
185 Returns: size of vcpu mmap area, in bytes
187 The KVM_RUN ioctl (cf.) communicates with userspace via a shared
188 memory region. This ioctl returns the size of that region. See the
189 KVM_RUN documentation for details.
192 4.6 KVM_SET_MEMORY_REGION
197 Parameters: struct kvm_memory_region (in)
198 Returns: 0 on success, -1 on error
200 This ioctl is obsolete and has been removed.
208 Parameters: vcpu id (apic id on x86)
209 Returns: vcpu fd on success, -1 on error
211 This API adds a vcpu to a virtual machine. No more than max_vcpus may be added.
212 The vcpu id is an integer in the range [0, max_vcpu_id).
214 The recommended max_vcpus value can be retrieved using the KVM_CAP_NR_VCPUS of
215 the KVM_CHECK_EXTENSION ioctl() at run-time.
216 The maximum possible value for max_vcpus can be retrieved using the
217 KVM_CAP_MAX_VCPUS of the KVM_CHECK_EXTENSION ioctl() at run-time.
219 If the KVM_CAP_NR_VCPUS does not exist, you should assume that max_vcpus is 4
221 If the KVM_CAP_MAX_VCPUS does not exist, you should assume that max_vcpus is
222 same as the value returned from KVM_CAP_NR_VCPUS.
224 The maximum possible value for max_vcpu_id can be retrieved using the
225 KVM_CAP_MAX_VCPU_ID of the KVM_CHECK_EXTENSION ioctl() at run-time.
227 If the KVM_CAP_MAX_VCPU_ID does not exist, you should assume that max_vcpu_id
228 is the same as the value returned from KVM_CAP_MAX_VCPUS.
230 On powerpc using book3s_hv mode, the vcpus are mapped onto virtual
231 threads in one or more virtual CPU cores. (This is because the
232 hardware requires all the hardware threads in a CPU core to be in the
233 same partition.) The KVM_CAP_PPC_SMT capability indicates the number
234 of vcpus per virtual core (vcore). The vcore id is obtained by
235 dividing the vcpu id by the number of vcpus per vcore. The vcpus in a
236 given vcore will always be in the same physical core as each other
237 (though that might be a different physical core from time to time).
238 Userspace can control the threading (SMT) mode of the guest by its
239 allocation of vcpu ids. For example, if userspace wants
240 single-threaded guest vcpus, it should make all vcpu ids be a multiple
241 of the number of vcpus per vcore.
243 For virtual cpus that have been created with S390 user controlled virtual
244 machines, the resulting vcpu fd can be memory mapped at page offset
245 KVM_S390_SIE_PAGE_OFFSET in order to obtain a memory map of the virtual
246 cpu's hardware control block.
249 4.8 KVM_GET_DIRTY_LOG (vm ioctl)
254 Parameters: struct kvm_dirty_log (in/out)
255 Returns: 0 on success, -1 on error
257 /* for KVM_GET_DIRTY_LOG */
258 struct kvm_dirty_log {
262 void __user *dirty_bitmap; /* one bit per page */
267 Given a memory slot, return a bitmap containing any pages dirtied
268 since the last call to this ioctl. Bit 0 is the first page in the
269 memory slot. Ensure the entire structure is cleared to avoid padding
272 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 specifies
273 the address space for which you want to return the dirty bitmap.
274 They must be less than the value that KVM_CHECK_EXTENSION returns for
275 the KVM_CAP_MULTI_ADDRESS_SPACE capability.
278 4.9 KVM_SET_MEMORY_ALIAS
283 Parameters: struct kvm_memory_alias (in)
284 Returns: 0 (success), -1 (error)
286 This ioctl is obsolete and has been removed.
295 Returns: 0 on success, -1 on error
297 EINTR: an unmasked signal is pending
299 This ioctl is used to run a guest virtual cpu. While there are no
300 explicit parameters, there is an implicit parameter block that can be
301 obtained by mmap()ing the vcpu fd at offset 0, with the size given by
302 KVM_GET_VCPU_MMAP_SIZE. The parameter block is formatted as a 'struct
303 kvm_run' (see below).
309 Architectures: all except ARM, arm64
311 Parameters: struct kvm_regs (out)
312 Returns: 0 on success, -1 on error
314 Reads the general purpose registers from the vcpu.
318 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
319 __u64 rax, rbx, rcx, rdx;
320 __u64 rsi, rdi, rsp, rbp;
321 __u64 r8, r9, r10, r11;
322 __u64 r12, r13, r14, r15;
328 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
339 Architectures: all except ARM, arm64
341 Parameters: struct kvm_regs (in)
342 Returns: 0 on success, -1 on error
344 Writes the general purpose registers into the vcpu.
346 See KVM_GET_REGS for the data structure.
352 Architectures: x86, ppc
354 Parameters: struct kvm_sregs (out)
355 Returns: 0 on success, -1 on error
357 Reads special registers from the vcpu.
361 struct kvm_segment cs, ds, es, fs, gs, ss;
362 struct kvm_segment tr, ldt;
363 struct kvm_dtable gdt, idt;
364 __u64 cr0, cr2, cr3, cr4, cr8;
367 __u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64];
370 /* ppc -- see arch/powerpc/include/uapi/asm/kvm.h */
372 interrupt_bitmap is a bitmap of pending external interrupts. At most
373 one bit may be set. This interrupt has been acknowledged by the APIC
374 but not yet injected into the cpu core.
380 Architectures: x86, ppc
382 Parameters: struct kvm_sregs (in)
383 Returns: 0 on success, -1 on error
385 Writes special registers into the vcpu. See KVM_GET_SREGS for the
394 Parameters: struct kvm_translation (in/out)
395 Returns: 0 on success, -1 on error
397 Translates a virtual address according to the vcpu's current address
400 struct kvm_translation {
402 __u64 linear_address;
405 __u64 physical_address;
416 Architectures: x86, ppc, mips
418 Parameters: struct kvm_interrupt (in)
419 Returns: 0 on success, negative on failure.
421 Queues a hardware interrupt vector to be injected.
423 /* for KVM_INTERRUPT */
424 struct kvm_interrupt {
431 Returns: 0 on success,
432 -EEXIST if an interrupt is already enqueued
433 -EINVAL the the irq number is invalid
434 -ENXIO if the PIC is in the kernel
435 -EFAULT if the pointer is invalid
437 Note 'irq' is an interrupt vector, not an interrupt pin or line. This
438 ioctl is useful if the in-kernel PIC is not used.
442 Queues an external interrupt to be injected. This ioctl is overleaded
443 with 3 different irq values:
447 This injects an edge type external interrupt into the guest once it's ready
448 to receive interrupts. When injected, the interrupt is done.
450 b) KVM_INTERRUPT_UNSET
452 This unsets any pending interrupt.
454 Only available with KVM_CAP_PPC_UNSET_IRQ.
456 c) KVM_INTERRUPT_SET_LEVEL
458 This injects a level type external interrupt into the guest context. The
459 interrupt stays pending until a specific ioctl with KVM_INTERRUPT_UNSET
462 Only available with KVM_CAP_PPC_IRQ_LEVEL.
464 Note that any value for 'irq' other than the ones stated above is invalid
465 and incurs unexpected behavior.
469 Queues an external interrupt to be injected into the virtual CPU. A negative
470 interrupt number dequeues the interrupt.
481 Support for this has been removed. Use KVM_SET_GUEST_DEBUG instead.
486 Capability: basic (vcpu), KVM_CAP_GET_MSR_FEATURES (system)
488 Type: system ioctl, vcpu ioctl
489 Parameters: struct kvm_msrs (in/out)
490 Returns: number of msrs successfully returned;
493 When used as a system ioctl:
494 Reads the values of MSR-based features that are available for the VM. This
495 is similar to KVM_GET_SUPPORTED_CPUID, but it returns MSR indices and values.
496 The list of msr-based features can be obtained using KVM_GET_MSR_FEATURE_INDEX_LIST
499 When used as a vcpu ioctl:
500 Reads model-specific registers from the vcpu. Supported msr indices can
501 be obtained using KVM_GET_MSR_INDEX_LIST in a system ioctl.
504 __u32 nmsrs; /* number of msrs in entries */
507 struct kvm_msr_entry entries[0];
510 struct kvm_msr_entry {
516 Application code should set the 'nmsrs' member (which indicates the
517 size of the entries array) and the 'index' member of each array entry.
518 kvm will fill in the 'data' member.
526 Parameters: struct kvm_msrs (in)
527 Returns: 0 on success, -1 on error
529 Writes model-specific registers to the vcpu. See KVM_GET_MSRS for the
532 Application code should set the 'nmsrs' member (which indicates the
533 size of the entries array), and the 'index' and 'data' members of each
542 Parameters: struct kvm_cpuid (in)
543 Returns: 0 on success, -1 on error
545 Defines the vcpu responses to the cpuid instruction. Applications
546 should use the KVM_SET_CPUID2 ioctl if available.
549 struct kvm_cpuid_entry {
558 /* for KVM_SET_CPUID */
562 struct kvm_cpuid_entry entries[0];
566 4.21 KVM_SET_SIGNAL_MASK
571 Parameters: struct kvm_signal_mask (in)
572 Returns: 0 on success, -1 on error
574 Defines which signals are blocked during execution of KVM_RUN. This
575 signal mask temporarily overrides the threads signal mask. Any
576 unblocked signal received (except SIGKILL and SIGSTOP, which retain
577 their traditional behaviour) will cause KVM_RUN to return with -EINTR.
579 Note the signal will only be delivered if not blocked by the original
582 /* for KVM_SET_SIGNAL_MASK */
583 struct kvm_signal_mask {
594 Parameters: struct kvm_fpu (out)
595 Returns: 0 on success, -1 on error
597 Reads the floating point state from the vcpu.
599 /* for KVM_GET_FPU and KVM_SET_FPU */
604 __u8 ftwx; /* in fxsave format */
620 Parameters: struct kvm_fpu (in)
621 Returns: 0 on success, -1 on error
623 Writes the floating point state to the vcpu.
625 /* for KVM_GET_FPU and KVM_SET_FPU */
630 __u8 ftwx; /* in fxsave format */
641 4.24 KVM_CREATE_IRQCHIP
643 Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQCHIP (s390)
644 Architectures: x86, ARM, arm64, s390
647 Returns: 0 on success, -1 on error
649 Creates an interrupt controller model in the kernel.
650 On x86, creates a virtual ioapic, a virtual PIC (two PICs, nested), and sets up
651 future vcpus to have a local APIC. IRQ routing for GSIs 0-15 is set to both
652 PIC and IOAPIC; GSI 16-23 only go to the IOAPIC.
653 On ARM/arm64, a GICv2 is created. Any other GIC versions require the usage of
654 KVM_CREATE_DEVICE, which also supports creating a GICv2. Using
655 KVM_CREATE_DEVICE is preferred over KVM_CREATE_IRQCHIP for GICv2.
656 On s390, a dummy irq routing table is created.
658 Note that on s390 the KVM_CAP_S390_IRQCHIP vm capability needs to be enabled
659 before KVM_CREATE_IRQCHIP can be used.
664 Capability: KVM_CAP_IRQCHIP
665 Architectures: x86, arm, arm64
667 Parameters: struct kvm_irq_level
668 Returns: 0 on success, -1 on error
670 Sets the level of a GSI input to the interrupt controller model in the kernel.
671 On some architectures it is required that an interrupt controller model has
672 been previously created with KVM_CREATE_IRQCHIP. Note that edge-triggered
673 interrupts require the level to be set to 1 and then back to 0.
675 On real hardware, interrupt pins can be active-low or active-high. This
676 does not matter for the level field of struct kvm_irq_level: 1 always
677 means active (asserted), 0 means inactive (deasserted).
679 x86 allows the operating system to program the interrupt polarity
680 (active-low/active-high) for level-triggered interrupts, and KVM used
681 to consider the polarity. However, due to bitrot in the handling of
682 active-low interrupts, the above convention is now valid on x86 too.
683 This is signaled by KVM_CAP_X86_IOAPIC_POLARITY_IGNORED. Userspace
684 should not present interrupts to the guest as active-low unless this
685 capability is present (or unless it is not using the in-kernel irqchip,
689 ARM/arm64 can signal an interrupt either at the CPU level, or at the
690 in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to
691 use PPIs designated for specific cpus. The irq field is interpreted
694 Â bits: | 31 ... 24 | 23 ... 16 | 15 ... 0 |
695 field: | irq_type | vcpu_index | irq_id |
697 The irq_type field has the following values:
698 - irq_type[0]: out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ
699 - irq_type[1]: in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.)
700 (the vcpu_index field is ignored)
701 - irq_type[2]: in-kernel GIC: PPI, irq_id between 16 and 31 (incl.)
703 (The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs)
705 In both cases, level is used to assert/deassert the line.
707 struct kvm_irq_level {
710 __s32 status; /* not used for KVM_IRQ_LEVEL */
712 __u32 level; /* 0 or 1 */
718 Capability: KVM_CAP_IRQCHIP
721 Parameters: struct kvm_irqchip (in/out)
722 Returns: 0 on success, -1 on error
724 Reads the state of a kernel interrupt controller created with
725 KVM_CREATE_IRQCHIP into a buffer provided by the caller.
728 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
731 char dummy[512]; /* reserving space */
732 struct kvm_pic_state pic;
733 struct kvm_ioapic_state ioapic;
740 Capability: KVM_CAP_IRQCHIP
743 Parameters: struct kvm_irqchip (in)
744 Returns: 0 on success, -1 on error
746 Sets the state of a kernel interrupt controller created with
747 KVM_CREATE_IRQCHIP from a buffer provided by the caller.
750 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
753 char dummy[512]; /* reserving space */
754 struct kvm_pic_state pic;
755 struct kvm_ioapic_state ioapic;
760 4.28 KVM_XEN_HVM_CONFIG
762 Capability: KVM_CAP_XEN_HVM
765 Parameters: struct kvm_xen_hvm_config (in)
766 Returns: 0 on success, -1 on error
768 Sets the MSR that the Xen HVM guest uses to initialize its hypercall
769 page, and provides the starting address and size of the hypercall
770 blobs in userspace. When the guest writes the MSR, kvm copies one
771 page of a blob (32- or 64-bit, depending on the vcpu mode) to guest
774 struct kvm_xen_hvm_config {
787 Capability: KVM_CAP_ADJUST_CLOCK
790 Parameters: struct kvm_clock_data (out)
791 Returns: 0 on success, -1 on error
793 Gets the current timestamp of kvmclock as seen by the current guest. In
794 conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios
797 When KVM_CAP_ADJUST_CLOCK is passed to KVM_CHECK_EXTENSION, it returns the
798 set of bits that KVM can return in struct kvm_clock_data's flag member.
800 The only flag defined now is KVM_CLOCK_TSC_STABLE. If set, the returned
801 value is the exact kvmclock value seen by all VCPUs at the instant
802 when KVM_GET_CLOCK was called. If clear, the returned value is simply
803 CLOCK_MONOTONIC plus a constant offset; the offset can be modified
804 with KVM_SET_CLOCK. KVM will try to make all VCPUs follow this clock,
805 but the exact value read by each VCPU could differ, because the host
808 struct kvm_clock_data {
809 __u64 clock; /* kvmclock current value */
817 Capability: KVM_CAP_ADJUST_CLOCK
820 Parameters: struct kvm_clock_data (in)
821 Returns: 0 on success, -1 on error
823 Sets the current timestamp of kvmclock to the value specified in its parameter.
824 In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios
827 struct kvm_clock_data {
828 __u64 clock; /* kvmclock current value */
834 4.31 KVM_GET_VCPU_EVENTS
836 Capability: KVM_CAP_VCPU_EVENTS
837 Extended by: KVM_CAP_INTR_SHADOW
838 Architectures: x86, arm, arm64
840 Parameters: struct kvm_vcpu_event (out)
841 Returns: 0 on success, -1 on error
845 Gets currently pending exceptions, interrupts, and NMIs as well as related
848 struct kvm_vcpu_events {
878 Only two fields are defined in the flags field:
880 - KVM_VCPUEVENT_VALID_SHADOW may be set in the flags field to signal that
881 interrupt.shadow contains a valid state.
883 - KVM_VCPUEVENT_VALID_SMM may be set in the flags field to signal that
884 smi contains a valid state.
888 If the guest accesses a device that is being emulated by the host kernel in
889 such a way that a real device would generate a physical SError, KVM may make
890 a virtual SError pending for that VCPU. This system error interrupt remains
891 pending until the guest takes the exception by unmasking PSTATE.A.
893 Running the VCPU may cause it to take a pending SError, or make an access that
894 causes an SError to become pending. The event's description is only valid while
895 the VPCU is not running.
897 This API provides a way to read and write the pending 'event' state that is not
898 visible to the guest. To save, restore or migrate a VCPU the struct representing
899 the state can be read then written using this GET/SET API, along with the other
900 guest-visible registers. It is not possible to 'cancel' an SError that has been
903 A device being emulated in user-space may also wish to generate an SError. To do
904 this the events structure can be populated by user-space. The current state
905 should be read first, to ensure no existing SError is pending. If an existing
906 SError is pending, the architecture's 'Multiple SError interrupts' rules should
907 be followed. (2.5.3 of DDI0587.a "ARM Reliability, Availability, and
908 Serviceability (RAS) Specification").
910 SError exceptions always have an ESR value. Some CPUs have the ability to
911 specify what the virtual SError's ESR value should be. These systems will
912 advertise KVM_CAP_ARM_SET_SERROR_ESR. In this case exception.has_esr will
913 always have a non-zero value when read, and the agent making an SError pending
914 should specify the ISS field in the lower 24 bits of exception.serror_esr. If
915 the system supports KVM_CAP_ARM_SET_SERROR_ESR, but user-space sets the events
916 with exception.has_esr as zero, KVM will choose an ESR.
918 Specifying exception.has_esr on a system that does not support it will return
919 -EINVAL. Setting anything other than the lower 24bits of exception.serror_esr
922 struct kvm_vcpu_events {
926 /* Align it to 8 bytes */
933 4.32 KVM_SET_VCPU_EVENTS
935 Capability: KVM_CAP_VCPU_EVENTS
936 Extended by: KVM_CAP_INTR_SHADOW
937 Architectures: x86, arm, arm64
939 Parameters: struct kvm_vcpu_event (in)
940 Returns: 0 on success, -1 on error
944 Set pending exceptions, interrupts, and NMIs as well as related states of the
947 See KVM_GET_VCPU_EVENTS for the data structure.
949 Fields that may be modified asynchronously by running VCPUs can be excluded
950 from the update. These fields are nmi.pending, sipi_vector, smi.smm,
951 smi.pending. Keep the corresponding bits in the flags field cleared to
952 suppress overwriting the current in-kernel state. The bits are:
954 KVM_VCPUEVENT_VALID_NMI_PENDING - transfer nmi.pending to the kernel
955 KVM_VCPUEVENT_VALID_SIPI_VECTOR - transfer sipi_vector
956 KVM_VCPUEVENT_VALID_SMM - transfer the smi sub-struct.
958 If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in
959 the flags field to signal that interrupt.shadow contains a valid state and
960 shall be written into the VCPU.
962 KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available.
966 Set the pending SError exception state for this VCPU. It is not possible to
967 'cancel' an Serror that has been made pending.
969 See KVM_GET_VCPU_EVENTS for the data structure.
972 4.33 KVM_GET_DEBUGREGS
974 Capability: KVM_CAP_DEBUGREGS
977 Parameters: struct kvm_debugregs (out)
978 Returns: 0 on success, -1 on error
980 Reads debug registers from the vcpu.
982 struct kvm_debugregs {
991 4.34 KVM_SET_DEBUGREGS
993 Capability: KVM_CAP_DEBUGREGS
996 Parameters: struct kvm_debugregs (in)
997 Returns: 0 on success, -1 on error
999 Writes debug registers into the vcpu.
1001 See KVM_GET_DEBUGREGS for the data structure. The flags field is unused
1002 yet and must be cleared on entry.
1005 4.35 KVM_SET_USER_MEMORY_REGION
1007 Capability: KVM_CAP_USER_MEM
1010 Parameters: struct kvm_userspace_memory_region (in)
1011 Returns: 0 on success, -1 on error
1013 struct kvm_userspace_memory_region {
1016 __u64 guest_phys_addr;
1017 __u64 memory_size; /* bytes */
1018 __u64 userspace_addr; /* start of the userspace allocated memory */
1021 /* for kvm_memory_region::flags */
1022 #define KVM_MEM_LOG_DIRTY_PAGES (1UL << 0)
1023 #define KVM_MEM_READONLY (1UL << 1)
1025 This ioctl allows the user to create or modify a guest physical memory
1026 slot. When changing an existing slot, it may be moved in the guest
1027 physical memory space, or its flags may be modified. It may not be
1028 resized. Slots may not overlap in guest physical address space.
1029 Bits 0-15 of "slot" specifies the slot id and this value should be
1030 less than the maximum number of user memory slots supported per VM.
1031 The maximum allowed slots can be queried using KVM_CAP_NR_MEMSLOTS,
1032 if this capability is supported by the architecture.
1034 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot"
1035 specifies the address space which is being modified. They must be
1036 less than the value that KVM_CHECK_EXTENSION returns for the
1037 KVM_CAP_MULTI_ADDRESS_SPACE capability. Slots in separate address spaces
1038 are unrelated; the restriction on overlapping slots only applies within
1041 Memory for the region is taken starting at the address denoted by the
1042 field userspace_addr, which must point at user addressable memory for
1043 the entire memory slot size. Any object may back this memory, including
1044 anonymous memory, ordinary files, and hugetlbfs.
1046 It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr
1047 be identical. This allows large pages in the guest to be backed by large
1050 The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and
1051 KVM_MEM_READONLY. The former can be set to instruct KVM to keep track of
1052 writes to memory within the slot. See KVM_GET_DIRTY_LOG ioctl to know how to
1053 use it. The latter can be set, if KVM_CAP_READONLY_MEM capability allows it,
1054 to make a new slot read-only. In this case, writes to this memory will be
1055 posted to userspace as KVM_EXIT_MMIO exits.
1057 When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of
1058 the memory region are automatically reflected into the guest. For example, an
1059 mmap() that affects the region will be made visible immediately. Another
1060 example is madvise(MADV_DROP).
1062 It is recommended to use this API instead of the KVM_SET_MEMORY_REGION ioctl.
1063 The KVM_SET_MEMORY_REGION does not allow fine grained control over memory
1064 allocation and is deprecated.
1067 4.36 KVM_SET_TSS_ADDR
1069 Capability: KVM_CAP_SET_TSS_ADDR
1072 Parameters: unsigned long tss_address (in)
1073 Returns: 0 on success, -1 on error
1075 This ioctl defines the physical address of a three-page region in the guest
1076 physical address space. The region must be within the first 4GB of the
1077 guest physical address space and must not conflict with any memory slot
1078 or any mmio address. The guest may malfunction if it accesses this memory
1081 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1082 because of a quirk in the virtualization implementation (see the internals
1083 documentation when it pops into existence).
1088 Capability: KVM_CAP_ENABLE_CAP, KVM_CAP_ENABLE_CAP_VM
1089 Architectures: x86 (only KVM_CAP_ENABLE_CAP_VM),
1090 mips (only KVM_CAP_ENABLE_CAP), ppc, s390
1091 Type: vcpu ioctl, vm ioctl (with KVM_CAP_ENABLE_CAP_VM)
1092 Parameters: struct kvm_enable_cap (in)
1093 Returns: 0 on success; -1 on error
1095 +Not all extensions are enabled by default. Using this ioctl the application
1096 can enable an extension, making it available to the guest.
1098 On systems that do not support this ioctl, it always fails. On systems that
1099 do support it, it only works for extensions that are supported for enablement.
1101 To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should
1104 struct kvm_enable_cap {
1108 The capability that is supposed to get enabled.
1112 A bitfield indicating future enhancements. Has to be 0 for now.
1116 Arguments for enabling a feature. If a feature needs initial values to
1117 function properly, this is the place to put them.
1122 The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl
1123 for vm-wide capabilities.
1125 4.38 KVM_GET_MP_STATE
1127 Capability: KVM_CAP_MP_STATE
1128 Architectures: x86, s390, arm, arm64
1130 Parameters: struct kvm_mp_state (out)
1131 Returns: 0 on success; -1 on error
1133 struct kvm_mp_state {
1137 Returns the vcpu's current "multiprocessing state" (though also valid on
1138 uniprocessor guests).
1140 Possible values are:
1142 - KVM_MP_STATE_RUNNABLE: the vcpu is currently running [x86,arm/arm64]
1143 - KVM_MP_STATE_UNINITIALIZED: the vcpu is an application processor (AP)
1144 which has not yet received an INIT signal [x86]
1145 - KVM_MP_STATE_INIT_RECEIVED: the vcpu has received an INIT signal, and is
1146 now ready for a SIPI [x86]
1147 - KVM_MP_STATE_HALTED: the vcpu has executed a HLT instruction and
1148 is waiting for an interrupt [x86]
1149 - KVM_MP_STATE_SIPI_RECEIVED: the vcpu has just received a SIPI (vector
1150 accessible via KVM_GET_VCPU_EVENTS) [x86]
1151 - KVM_MP_STATE_STOPPED: the vcpu is stopped [s390,arm/arm64]
1152 - KVM_MP_STATE_CHECK_STOP: the vcpu is in a special error state [s390]
1153 - KVM_MP_STATE_OPERATING: the vcpu is operating (running or halted)
1155 - KVM_MP_STATE_LOAD: the vcpu is in a special load/startup state
1158 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1159 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1160 these architectures.
1164 The only states that are valid are KVM_MP_STATE_STOPPED and
1165 KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not.
1167 4.39 KVM_SET_MP_STATE
1169 Capability: KVM_CAP_MP_STATE
1170 Architectures: x86, s390, arm, arm64
1172 Parameters: struct kvm_mp_state (in)
1173 Returns: 0 on success; -1 on error
1175 Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for
1178 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1179 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1180 these architectures.
1184 The only states that are valid are KVM_MP_STATE_STOPPED and
1185 KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not.
1187 4.40 KVM_SET_IDENTITY_MAP_ADDR
1189 Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR
1192 Parameters: unsigned long identity (in)
1193 Returns: 0 on success, -1 on error
1195 This ioctl defines the physical address of a one-page region in the guest
1196 physical address space. The region must be within the first 4GB of the
1197 guest physical address space and must not conflict with any memory slot
1198 or any mmio address. The guest may malfunction if it accesses this memory
1201 Setting the address to 0 will result in resetting the address to its default
1204 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1205 because of a quirk in the virtualization implementation (see the internals
1206 documentation when it pops into existence).
1208 Fails if any VCPU has already been created.
1210 4.41 KVM_SET_BOOT_CPU_ID
1212 Capability: KVM_CAP_SET_BOOT_CPU_ID
1215 Parameters: unsigned long vcpu_id
1216 Returns: 0 on success, -1 on error
1218 Define which vcpu is the Bootstrap Processor (BSP). Values are the same
1219 as the vcpu id in KVM_CREATE_VCPU. If this ioctl is not called, the default
1225 Capability: KVM_CAP_XSAVE
1228 Parameters: struct kvm_xsave (out)
1229 Returns: 0 on success, -1 on error
1235 This ioctl would copy current vcpu's xsave struct to the userspace.
1240 Capability: KVM_CAP_XSAVE
1243 Parameters: struct kvm_xsave (in)
1244 Returns: 0 on success, -1 on error
1250 This ioctl would copy userspace's xsave struct to the kernel.
1255 Capability: KVM_CAP_XCRS
1258 Parameters: struct kvm_xcrs (out)
1259 Returns: 0 on success, -1 on error
1270 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1274 This ioctl would copy current vcpu's xcrs to the userspace.
1279 Capability: KVM_CAP_XCRS
1282 Parameters: struct kvm_xcrs (in)
1283 Returns: 0 on success, -1 on error
1294 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1298 This ioctl would set vcpu's xcr to the value userspace specified.
1301 4.46 KVM_GET_SUPPORTED_CPUID
1303 Capability: KVM_CAP_EXT_CPUID
1306 Parameters: struct kvm_cpuid2 (in/out)
1307 Returns: 0 on success, -1 on error
1312 struct kvm_cpuid_entry2 entries[0];
1315 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
1316 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1)
1317 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2)
1319 struct kvm_cpuid_entry2 {
1330 This ioctl returns x86 cpuid features which are supported by both the
1331 hardware and kvm in its default configuration. Userspace can use the
1332 information returned by this ioctl to construct cpuid information (for
1333 KVM_SET_CPUID2) that is consistent with hardware, kernel, and
1334 userspace capabilities, and with user requirements (for example, the
1335 user may wish to constrain cpuid to emulate older hardware, or for
1336 feature consistency across a cluster).
1338 Note that certain capabilities, such as KVM_CAP_X86_DISABLE_EXITS, may
1339 expose cpuid features (e.g. MONITOR) which are not supported by kvm in
1340 its default configuration. If userspace enables such capabilities, it
1341 is responsible for modifying the results of this ioctl appropriately.
1343 Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure
1344 with the 'nent' field indicating the number of entries in the variable-size
1345 array 'entries'. If the number of entries is too low to describe the cpu
1346 capabilities, an error (E2BIG) is returned. If the number is too high,
1347 the 'nent' field is adjusted and an error (ENOMEM) is returned. If the
1348 number is just right, the 'nent' field is adjusted to the number of valid
1349 entries in the 'entries' array, which is then filled.
1351 The entries returned are the host cpuid as returned by the cpuid instruction,
1352 with unknown or unsupported features masked out. Some features (for example,
1353 x2apic), may not be present in the host cpu, but are exposed by kvm if it can
1354 emulate them efficiently. The fields in each entry are defined as follows:
1356 function: the eax value used to obtain the entry
1357 index: the ecx value used to obtain the entry (for entries that are
1359 flags: an OR of zero or more of the following:
1360 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
1361 if the index field is valid
1362 KVM_CPUID_FLAG_STATEFUL_FUNC:
1363 if cpuid for this function returns different values for successive
1364 invocations; there will be several entries with the same function,
1365 all with this flag set
1366 KVM_CPUID_FLAG_STATE_READ_NEXT:
1367 for KVM_CPUID_FLAG_STATEFUL_FUNC entries, set if this entry is
1368 the first entry to be read by a cpu
1369 eax, ebx, ecx, edx: the values returned by the cpuid instruction for
1370 this function/index combination
1372 The TSC deadline timer feature (CPUID leaf 1, ecx[24]) is always returned
1373 as false, since the feature depends on KVM_CREATE_IRQCHIP for local APIC
1374 support. Instead it is reported via
1376 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER)
1378 if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the
1379 feature in userspace, then you can enable the feature for KVM_SET_CPUID2.
1382 4.47 KVM_PPC_GET_PVINFO
1384 Capability: KVM_CAP_PPC_GET_PVINFO
1387 Parameters: struct kvm_ppc_pvinfo (out)
1388 Returns: 0 on success, !0 on error
1390 struct kvm_ppc_pvinfo {
1396 This ioctl fetches PV specific information that need to be passed to the guest
1397 using the device tree or other means from vm context.
1399 The hcall array defines 4 instructions that make up a hypercall.
1401 If any additional field gets added to this structure later on, a bit for that
1402 additional piece of information will be set in the flags bitmap.
1404 The flags bitmap is defined as:
1406 /* the host supports the ePAPR idle hcall
1407 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1<<0)
1409 4.52 KVM_SET_GSI_ROUTING
1411 Capability: KVM_CAP_IRQ_ROUTING
1412 Architectures: x86 s390 arm arm64
1414 Parameters: struct kvm_irq_routing (in)
1415 Returns: 0 on success, -1 on error
1417 Sets the GSI routing table entries, overwriting any previously set entries.
1419 On arm/arm64, GSI routing has the following limitation:
1420 - GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD.
1422 struct kvm_irq_routing {
1425 struct kvm_irq_routing_entry entries[0];
1428 No flags are specified so far, the corresponding field must be set to zero.
1430 struct kvm_irq_routing_entry {
1436 struct kvm_irq_routing_irqchip irqchip;
1437 struct kvm_irq_routing_msi msi;
1438 struct kvm_irq_routing_s390_adapter adapter;
1439 struct kvm_irq_routing_hv_sint hv_sint;
1444 /* gsi routing entry types */
1445 #define KVM_IRQ_ROUTING_IRQCHIP 1
1446 #define KVM_IRQ_ROUTING_MSI 2
1447 #define KVM_IRQ_ROUTING_S390_ADAPTER 3
1448 #define KVM_IRQ_ROUTING_HV_SINT 4
1451 - KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry
1452 type, specifies that the devid field contains a valid value. The per-VM
1453 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
1454 the device ID. If this capability is not available, userspace should
1455 never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
1458 struct kvm_irq_routing_irqchip {
1463 struct kvm_irq_routing_msi {
1473 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
1474 for the device that wrote the MSI message. For PCI, this is usually a
1475 BFD identifier in the lower 16 bits.
1477 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
1478 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
1479 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
1480 address_hi must be zero.
1482 struct kvm_irq_routing_s390_adapter {
1486 __u32 summary_offset;
1490 struct kvm_irq_routing_hv_sint {
1496 4.55 KVM_SET_TSC_KHZ
1498 Capability: KVM_CAP_TSC_CONTROL
1501 Parameters: virtual tsc_khz
1502 Returns: 0 on success, -1 on error
1504 Specifies the tsc frequency for the virtual machine. The unit of the
1508 4.56 KVM_GET_TSC_KHZ
1510 Capability: KVM_CAP_GET_TSC_KHZ
1514 Returns: virtual tsc-khz on success, negative value on error
1516 Returns the tsc frequency of the guest. The unit of the return value is
1517 KHz. If the host has unstable tsc this ioctl returns -EIO instead as an
1523 Capability: KVM_CAP_IRQCHIP
1526 Parameters: struct kvm_lapic_state (out)
1527 Returns: 0 on success, -1 on error
1529 #define KVM_APIC_REG_SIZE 0x400
1530 struct kvm_lapic_state {
1531 char regs[KVM_APIC_REG_SIZE];
1534 Reads the Local APIC registers and copies them into the input argument. The
1535 data format and layout are the same as documented in the architecture manual.
1537 If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is
1538 enabled, then the format of APIC_ID register depends on the APIC mode
1539 (reported by MSR_IA32_APICBASE) of its VCPU. x2APIC stores APIC ID in
1540 the APIC_ID register (bytes 32-35). xAPIC only allows an 8-bit APIC ID
1541 which is stored in bits 31-24 of the APIC register, or equivalently in
1542 byte 35 of struct kvm_lapic_state's regs field. KVM_GET_LAPIC must then
1543 be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR.
1545 If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state
1546 always uses xAPIC format.
1551 Capability: KVM_CAP_IRQCHIP
1554 Parameters: struct kvm_lapic_state (in)
1555 Returns: 0 on success, -1 on error
1557 #define KVM_APIC_REG_SIZE 0x400
1558 struct kvm_lapic_state {
1559 char regs[KVM_APIC_REG_SIZE];
1562 Copies the input argument into the Local APIC registers. The data format
1563 and layout are the same as documented in the architecture manual.
1565 The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's
1566 regs field) depends on the state of the KVM_CAP_X2APIC_API capability.
1567 See the note in KVM_GET_LAPIC.
1572 Capability: KVM_CAP_IOEVENTFD
1575 Parameters: struct kvm_ioeventfd (in)
1576 Returns: 0 on success, !0 on error
1578 This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address
1579 within the guest. A guest write in the registered address will signal the
1580 provided event instead of triggering an exit.
1582 struct kvm_ioeventfd {
1584 __u64 addr; /* legal pio/mmio address */
1585 __u32 len; /* 0, 1, 2, 4, or 8 bytes */
1591 For the special case of virtio-ccw devices on s390, the ioevent is matched
1592 to a subchannel/virtqueue tuple instead.
1594 The following flags are defined:
1596 #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch)
1597 #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio)
1598 #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign)
1599 #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \
1600 (1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify)
1602 If datamatch flag is set, the event will be signaled only if the written value
1603 to the registered address is equal to datamatch in struct kvm_ioeventfd.
1605 For virtio-ccw devices, addr contains the subchannel id and datamatch the
1608 With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and
1609 the kernel will ignore the length of guest write and may get a faster vmexit.
1610 The speedup may only apply to specific architectures, but the ioeventfd will
1615 Capability: KVM_CAP_SW_TLB
1618 Parameters: struct kvm_dirty_tlb (in)
1619 Returns: 0 on success, -1 on error
1621 struct kvm_dirty_tlb {
1626 This must be called whenever userspace has changed an entry in the shared
1627 TLB, prior to calling KVM_RUN on the associated vcpu.
1629 The "bitmap" field is the userspace address of an array. This array
1630 consists of a number of bits, equal to the total number of TLB entries as
1631 determined by the last successful call to KVM_CONFIG_TLB, rounded up to the
1632 nearest multiple of 64.
1634 Each bit corresponds to one TLB entry, ordered the same as in the shared TLB
1637 The array is little-endian: the bit 0 is the least significant bit of the
1638 first byte, bit 8 is the least significant bit of the second byte, etc.
1639 This avoids any complications with differing word sizes.
1641 The "num_dirty" field is a performance hint for KVM to determine whether it
1642 should skip processing the bitmap and just invalidate everything. It must
1643 be set to the number of set bits in the bitmap.
1646 4.62 KVM_CREATE_SPAPR_TCE
1648 Capability: KVM_CAP_SPAPR_TCE
1649 Architectures: powerpc
1651 Parameters: struct kvm_create_spapr_tce (in)
1652 Returns: file descriptor for manipulating the created TCE table
1654 This creates a virtual TCE (translation control entry) table, which
1655 is an IOMMU for PAPR-style virtual I/O. It is used to translate
1656 logical addresses used in virtual I/O into guest physical addresses,
1657 and provides a scatter/gather capability for PAPR virtual I/O.
1659 /* for KVM_CAP_SPAPR_TCE */
1660 struct kvm_create_spapr_tce {
1665 The liobn field gives the logical IO bus number for which to create a
1666 TCE table. The window_size field specifies the size of the DMA window
1667 which this TCE table will translate - the table will contain one 64
1668 bit TCE entry for every 4kiB of the DMA window.
1670 When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE
1671 table has been created using this ioctl(), the kernel will handle it
1672 in real mode, updating the TCE table. H_PUT_TCE calls for other
1673 liobns will cause a vm exit and must be handled by userspace.
1675 The return value is a file descriptor which can be passed to mmap(2)
1676 to map the created TCE table into userspace. This lets userspace read
1677 the entries written by kernel-handled H_PUT_TCE calls, and also lets
1678 userspace update the TCE table directly which is useful in some
1682 4.63 KVM_ALLOCATE_RMA
1684 Capability: KVM_CAP_PPC_RMA
1685 Architectures: powerpc
1687 Parameters: struct kvm_allocate_rma (out)
1688 Returns: file descriptor for mapping the allocated RMA
1690 This allocates a Real Mode Area (RMA) from the pool allocated at boot
1691 time by the kernel. An RMA is a physically-contiguous, aligned region
1692 of memory used on older POWER processors to provide the memory which
1693 will be accessed by real-mode (MMU off) accesses in a KVM guest.
1694 POWER processors support a set of sizes for the RMA that usually
1695 includes 64MB, 128MB, 256MB and some larger powers of two.
1697 /* for KVM_ALLOCATE_RMA */
1698 struct kvm_allocate_rma {
1702 The return value is a file descriptor which can be passed to mmap(2)
1703 to map the allocated RMA into userspace. The mapped area can then be
1704 passed to the KVM_SET_USER_MEMORY_REGION ioctl to establish it as the
1705 RMA for a virtual machine. The size of the RMA in bytes (which is
1706 fixed at host kernel boot time) is returned in the rma_size field of
1707 the argument structure.
1709 The KVM_CAP_PPC_RMA capability is 1 or 2 if the KVM_ALLOCATE_RMA ioctl
1710 is supported; 2 if the processor requires all virtual machines to have
1711 an RMA, or 1 if the processor can use an RMA but doesn't require it,
1712 because it supports the Virtual RMA (VRMA) facility.
1717 Capability: KVM_CAP_USER_NMI
1721 Returns: 0 on success, -1 on error
1723 Queues an NMI on the thread's vcpu. Note this is well defined only
1724 when KVM_CREATE_IRQCHIP has not been called, since this is an interface
1725 between the virtual cpu core and virtual local APIC. After KVM_CREATE_IRQCHIP
1726 has been called, this interface is completely emulated within the kernel.
1728 To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the
1729 following algorithm:
1732 - read the local APIC's state (KVM_GET_LAPIC)
1733 - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1)
1734 - if so, issue KVM_NMI
1737 Some guests configure the LINT1 NMI input to cause a panic, aiding in
1741 4.65 KVM_S390_UCAS_MAP
1743 Capability: KVM_CAP_S390_UCONTROL
1746 Parameters: struct kvm_s390_ucas_mapping (in)
1747 Returns: 0 in case of success
1749 The parameter is defined like this:
1750 struct kvm_s390_ucas_mapping {
1756 This ioctl maps the memory at "user_addr" with the length "length" to
1757 the vcpu's address space starting at "vcpu_addr". All parameters need to
1758 be aligned by 1 megabyte.
1761 4.66 KVM_S390_UCAS_UNMAP
1763 Capability: KVM_CAP_S390_UCONTROL
1766 Parameters: struct kvm_s390_ucas_mapping (in)
1767 Returns: 0 in case of success
1769 The parameter is defined like this:
1770 struct kvm_s390_ucas_mapping {
1776 This ioctl unmaps the memory in the vcpu's address space starting at
1777 "vcpu_addr" with the length "length". The field "user_addr" is ignored.
1778 All parameters need to be aligned by 1 megabyte.
1781 4.67 KVM_S390_VCPU_FAULT
1783 Capability: KVM_CAP_S390_UCONTROL
1786 Parameters: vcpu absolute address (in)
1787 Returns: 0 in case of success
1789 This call creates a page table entry on the virtual cpu's address space
1790 (for user controlled virtual machines) or the virtual machine's address
1791 space (for regular virtual machines). This only works for minor faults,
1792 thus it's recommended to access subject memory page via the user page
1793 table upfront. This is useful to handle validity intercepts for user
1794 controlled virtual machines to fault in the virtual cpu's lowcore pages
1795 prior to calling the KVM_RUN ioctl.
1798 4.68 KVM_SET_ONE_REG
1800 Capability: KVM_CAP_ONE_REG
1803 Parameters: struct kvm_one_reg (in)
1804 Returns: 0 on success, negative value on failure
1806 struct kvm_one_reg {
1811 Using this ioctl, a single vcpu register can be set to a specific value
1812 defined by user space with the passed in struct kvm_one_reg, where id
1813 refers to the register identifier as described below and addr is a pointer
1814 to a variable with the respective size. There can be architecture agnostic
1815 and architecture specific registers. Each have their own range of operation
1816 and their own constants and width. To keep track of the implemented
1817 registers, find a list below:
1819 Arch | Register | Width (bits)
1821 PPC | KVM_REG_PPC_HIOR | 64
1822 PPC | KVM_REG_PPC_IAC1 | 64
1823 PPC | KVM_REG_PPC_IAC2 | 64
1824 PPC | KVM_REG_PPC_IAC3 | 64
1825 PPC | KVM_REG_PPC_IAC4 | 64
1826 PPC | KVM_REG_PPC_DAC1 | 64
1827 PPC | KVM_REG_PPC_DAC2 | 64
1828 PPC | KVM_REG_PPC_DABR | 64
1829 PPC | KVM_REG_PPC_DSCR | 64
1830 PPC | KVM_REG_PPC_PURR | 64
1831 PPC | KVM_REG_PPC_SPURR | 64
1832 PPC | KVM_REG_PPC_DAR | 64
1833 PPC | KVM_REG_PPC_DSISR | 32
1834 PPC | KVM_REG_PPC_AMR | 64
1835 PPC | KVM_REG_PPC_UAMOR | 64
1836 PPC | KVM_REG_PPC_MMCR0 | 64
1837 PPC | KVM_REG_PPC_MMCR1 | 64
1838 PPC | KVM_REG_PPC_MMCRA | 64
1839 PPC | KVM_REG_PPC_MMCR2 | 64
1840 PPC | KVM_REG_PPC_MMCRS | 64
1841 PPC | KVM_REG_PPC_SIAR | 64
1842 PPC | KVM_REG_PPC_SDAR | 64
1843 PPC | KVM_REG_PPC_SIER | 64
1844 PPC | KVM_REG_PPC_PMC1 | 32
1845 PPC | KVM_REG_PPC_PMC2 | 32
1846 PPC | KVM_REG_PPC_PMC3 | 32
1847 PPC | KVM_REG_PPC_PMC4 | 32
1848 PPC | KVM_REG_PPC_PMC5 | 32
1849 PPC | KVM_REG_PPC_PMC6 | 32
1850 PPC | KVM_REG_PPC_PMC7 | 32
1851 PPC | KVM_REG_PPC_PMC8 | 32
1852 PPC | KVM_REG_PPC_FPR0 | 64
1854 PPC | KVM_REG_PPC_FPR31 | 64
1855 PPC | KVM_REG_PPC_VR0 | 128
1857 PPC | KVM_REG_PPC_VR31 | 128
1858 PPC | KVM_REG_PPC_VSR0 | 128
1860 PPC | KVM_REG_PPC_VSR31 | 128
1861 PPC | KVM_REG_PPC_FPSCR | 64
1862 PPC | KVM_REG_PPC_VSCR | 32
1863 PPC | KVM_REG_PPC_VPA_ADDR | 64
1864 PPC | KVM_REG_PPC_VPA_SLB | 128
1865 PPC | KVM_REG_PPC_VPA_DTL | 128
1866 PPC | KVM_REG_PPC_EPCR | 32
1867 PPC | KVM_REG_PPC_EPR | 32
1868 PPC | KVM_REG_PPC_TCR | 32
1869 PPC | KVM_REG_PPC_TSR | 32
1870 PPC | KVM_REG_PPC_OR_TSR | 32
1871 PPC | KVM_REG_PPC_CLEAR_TSR | 32
1872 PPC | KVM_REG_PPC_MAS0 | 32
1873 PPC | KVM_REG_PPC_MAS1 | 32
1874 PPC | KVM_REG_PPC_MAS2 | 64
1875 PPC | KVM_REG_PPC_MAS7_3 | 64
1876 PPC | KVM_REG_PPC_MAS4 | 32
1877 PPC | KVM_REG_PPC_MAS6 | 32
1878 PPC | KVM_REG_PPC_MMUCFG | 32
1879 PPC | KVM_REG_PPC_TLB0CFG | 32
1880 PPC | KVM_REG_PPC_TLB1CFG | 32
1881 PPC | KVM_REG_PPC_TLB2CFG | 32
1882 PPC | KVM_REG_PPC_TLB3CFG | 32
1883 PPC | KVM_REG_PPC_TLB0PS | 32
1884 PPC | KVM_REG_PPC_TLB1PS | 32
1885 PPC | KVM_REG_PPC_TLB2PS | 32
1886 PPC | KVM_REG_PPC_TLB3PS | 32
1887 PPC | KVM_REG_PPC_EPTCFG | 32
1888 PPC | KVM_REG_PPC_ICP_STATE | 64
1889 PPC | KVM_REG_PPC_TB_OFFSET | 64
1890 PPC | KVM_REG_PPC_SPMC1 | 32
1891 PPC | KVM_REG_PPC_SPMC2 | 32
1892 PPC | KVM_REG_PPC_IAMR | 64
1893 PPC | KVM_REG_PPC_TFHAR | 64
1894 PPC | KVM_REG_PPC_TFIAR | 64
1895 PPC | KVM_REG_PPC_TEXASR | 64
1896 PPC | KVM_REG_PPC_FSCR | 64
1897 PPC | KVM_REG_PPC_PSPB | 32
1898 PPC | KVM_REG_PPC_EBBHR | 64
1899 PPC | KVM_REG_PPC_EBBRR | 64
1900 PPC | KVM_REG_PPC_BESCR | 64
1901 PPC | KVM_REG_PPC_TAR | 64
1902 PPC | KVM_REG_PPC_DPDES | 64
1903 PPC | KVM_REG_PPC_DAWR | 64
1904 PPC | KVM_REG_PPC_DAWRX | 64
1905 PPC | KVM_REG_PPC_CIABR | 64
1906 PPC | KVM_REG_PPC_IC | 64
1907 PPC | KVM_REG_PPC_VTB | 64
1908 PPC | KVM_REG_PPC_CSIGR | 64
1909 PPC | KVM_REG_PPC_TACR | 64
1910 PPC | KVM_REG_PPC_TCSCR | 64
1911 PPC | KVM_REG_PPC_PID | 64
1912 PPC | KVM_REG_PPC_ACOP | 64
1913 PPC | KVM_REG_PPC_VRSAVE | 32
1914 PPC | KVM_REG_PPC_LPCR | 32
1915 PPC | KVM_REG_PPC_LPCR_64 | 64
1916 PPC | KVM_REG_PPC_PPR | 64
1917 PPC | KVM_REG_PPC_ARCH_COMPAT | 32
1918 PPC | KVM_REG_PPC_DABRX | 32
1919 PPC | KVM_REG_PPC_WORT | 64
1920 PPC | KVM_REG_PPC_SPRG9 | 64
1921 PPC | KVM_REG_PPC_DBSR | 32
1922 PPC | KVM_REG_PPC_TIDR | 64
1923 PPC | KVM_REG_PPC_PSSCR | 64
1924 PPC | KVM_REG_PPC_DEC_EXPIRY | 64
1925 PPC | KVM_REG_PPC_TM_GPR0 | 64
1927 PPC | KVM_REG_PPC_TM_GPR31 | 64
1928 PPC | KVM_REG_PPC_TM_VSR0 | 128
1930 PPC | KVM_REG_PPC_TM_VSR63 | 128
1931 PPC | KVM_REG_PPC_TM_CR | 64
1932 PPC | KVM_REG_PPC_TM_LR | 64
1933 PPC | KVM_REG_PPC_TM_CTR | 64
1934 PPC | KVM_REG_PPC_TM_FPSCR | 64
1935 PPC | KVM_REG_PPC_TM_AMR | 64
1936 PPC | KVM_REG_PPC_TM_PPR | 64
1937 PPC | KVM_REG_PPC_TM_VRSAVE | 64
1938 PPC | KVM_REG_PPC_TM_VSCR | 32
1939 PPC | KVM_REG_PPC_TM_DSCR | 64
1940 PPC | KVM_REG_PPC_TM_TAR | 64
1941 PPC | KVM_REG_PPC_TM_XER | 64
1943 MIPS | KVM_REG_MIPS_R0 | 64
1945 MIPS | KVM_REG_MIPS_R31 | 64
1946 MIPS | KVM_REG_MIPS_HI | 64
1947 MIPS | KVM_REG_MIPS_LO | 64
1948 MIPS | KVM_REG_MIPS_PC | 64
1949 MIPS | KVM_REG_MIPS_CP0_INDEX | 32
1950 MIPS | KVM_REG_MIPS_CP0_ENTRYLO0 | 64
1951 MIPS | KVM_REG_MIPS_CP0_ENTRYLO1 | 64
1952 MIPS | KVM_REG_MIPS_CP0_CONTEXT | 64
1953 MIPS | KVM_REG_MIPS_CP0_CONTEXTCONFIG| 32
1954 MIPS | KVM_REG_MIPS_CP0_USERLOCAL | 64
1955 MIPS | KVM_REG_MIPS_CP0_XCONTEXTCONFIG| 64
1956 MIPS | KVM_REG_MIPS_CP0_PAGEMASK | 32
1957 MIPS | KVM_REG_MIPS_CP0_PAGEGRAIN | 32
1958 MIPS | KVM_REG_MIPS_CP0_SEGCTL0 | 64
1959 MIPS | KVM_REG_MIPS_CP0_SEGCTL1 | 64
1960 MIPS | KVM_REG_MIPS_CP0_SEGCTL2 | 64
1961 MIPS | KVM_REG_MIPS_CP0_PWBASE | 64
1962 MIPS | KVM_REG_MIPS_CP0_PWFIELD | 64
1963 MIPS | KVM_REG_MIPS_CP0_PWSIZE | 64
1964 MIPS | KVM_REG_MIPS_CP0_WIRED | 32
1965 MIPS | KVM_REG_MIPS_CP0_PWCTL | 32
1966 MIPS | KVM_REG_MIPS_CP0_HWRENA | 32
1967 MIPS | KVM_REG_MIPS_CP0_BADVADDR | 64
1968 MIPS | KVM_REG_MIPS_CP0_BADINSTR | 32
1969 MIPS | KVM_REG_MIPS_CP0_BADINSTRP | 32
1970 MIPS | KVM_REG_MIPS_CP0_COUNT | 32
1971 MIPS | KVM_REG_MIPS_CP0_ENTRYHI | 64
1972 MIPS | KVM_REG_MIPS_CP0_COMPARE | 32
1973 MIPS | KVM_REG_MIPS_CP0_STATUS | 32
1974 MIPS | KVM_REG_MIPS_CP0_INTCTL | 32
1975 MIPS | KVM_REG_MIPS_CP0_CAUSE | 32
1976 MIPS | KVM_REG_MIPS_CP0_EPC | 64
1977 MIPS | KVM_REG_MIPS_CP0_PRID | 32
1978 MIPS | KVM_REG_MIPS_CP0_EBASE | 64
1979 MIPS | KVM_REG_MIPS_CP0_CONFIG | 32
1980 MIPS | KVM_REG_MIPS_CP0_CONFIG1 | 32
1981 MIPS | KVM_REG_MIPS_CP0_CONFIG2 | 32
1982 MIPS | KVM_REG_MIPS_CP0_CONFIG3 | 32
1983 MIPS | KVM_REG_MIPS_CP0_CONFIG4 | 32
1984 MIPS | KVM_REG_MIPS_CP0_CONFIG5 | 32
1985 MIPS | KVM_REG_MIPS_CP0_CONFIG7 | 32
1986 MIPS | KVM_REG_MIPS_CP0_XCONTEXT | 64
1987 MIPS | KVM_REG_MIPS_CP0_ERROREPC | 64
1988 MIPS | KVM_REG_MIPS_CP0_KSCRATCH1 | 64
1989 MIPS | KVM_REG_MIPS_CP0_KSCRATCH2 | 64
1990 MIPS | KVM_REG_MIPS_CP0_KSCRATCH3 | 64
1991 MIPS | KVM_REG_MIPS_CP0_KSCRATCH4 | 64
1992 MIPS | KVM_REG_MIPS_CP0_KSCRATCH5 | 64
1993 MIPS | KVM_REG_MIPS_CP0_KSCRATCH6 | 64
1994 MIPS | KVM_REG_MIPS_CP0_MAAR(0..63) | 64
1995 MIPS | KVM_REG_MIPS_COUNT_CTL | 64
1996 MIPS | KVM_REG_MIPS_COUNT_RESUME | 64
1997 MIPS | KVM_REG_MIPS_COUNT_HZ | 64
1998 MIPS | KVM_REG_MIPS_FPR_32(0..31) | 32
1999 MIPS | KVM_REG_MIPS_FPR_64(0..31) | 64
2000 MIPS | KVM_REG_MIPS_VEC_128(0..31) | 128
2001 MIPS | KVM_REG_MIPS_FCR_IR | 32
2002 MIPS | KVM_REG_MIPS_FCR_CSR | 32
2003 MIPS | KVM_REG_MIPS_MSA_IR | 32
2004 MIPS | KVM_REG_MIPS_MSA_CSR | 32
2006 ARM registers are mapped using the lower 32 bits. The upper 16 of that
2007 is the register group type, or coprocessor number:
2009 ARM core registers have the following id bit patterns:
2010 0x4020 0000 0010 <index into the kvm_regs struct:16>
2012 ARM 32-bit CP15 registers have the following id bit patterns:
2013 0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3>
2015 ARM 64-bit CP15 registers have the following id bit patterns:
2016 0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3>
2018 ARM CCSIDR registers are demultiplexed by CSSELR value:
2019 0x4020 0000 0011 00 <csselr:8>
2021 ARM 32-bit VFP control registers have the following id bit patterns:
2022 0x4020 0000 0012 1 <regno:12>
2024 ARM 64-bit FP registers have the following id bit patterns:
2025 0x4030 0000 0012 0 <regno:12>
2027 ARM firmware pseudo-registers have the following bit pattern:
2028 0x4030 0000 0014 <regno:16>
2031 arm64 registers are mapped using the lower 32 bits. The upper 16 of
2032 that is the register group type, or coprocessor number:
2034 arm64 core/FP-SIMD registers have the following id bit patterns. Note
2035 that the size of the access is variable, as the kvm_regs structure
2036 contains elements ranging from 32 to 128 bits. The index is a 32bit
2037 value in the kvm_regs structure seen as a 32bit array.
2038 0x60x0 0000 0010 <index into the kvm_regs struct:16>
2040 arm64 CCSIDR registers are demultiplexed by CSSELR value:
2041 0x6020 0000 0011 00 <csselr:8>
2043 arm64 system registers have the following id bit patterns:
2044 0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3>
2046 arm64 firmware pseudo-registers have the following bit pattern:
2047 0x6030 0000 0014 <regno:16>
2050 MIPS registers are mapped using the lower 32 bits. The upper 16 of that is
2051 the register group type:
2053 MIPS core registers (see above) have the following id bit patterns:
2054 0x7030 0000 0000 <reg:16>
2056 MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit
2057 patterns depending on whether they're 32-bit or 64-bit registers:
2058 0x7020 0000 0001 00 <reg:5> <sel:3> (32-bit)
2059 0x7030 0000 0001 00 <reg:5> <sel:3> (64-bit)
2061 Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_MIPS_CP0_ENTRYLO1 are the MIPS64
2062 versions of the EntryLo registers regardless of the word size of the host
2063 hardware, host kernel, guest, and whether XPA is present in the guest, i.e.
2064 with the RI and XI bits (if they exist) in bits 63 and 62 respectively, and
2065 the PFNX field starting at bit 30.
2067 MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) above) have the following id bit
2069 0x7030 0000 0001 01 <reg:8>
2071 MIPS KVM control registers (see above) have the following id bit patterns:
2072 0x7030 0000 0002 <reg:16>
2074 MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following
2075 id bit patterns depending on the size of the register being accessed. They are
2076 always accessed according to the current guest FPU mode (Status.FR and
2077 Config5.FRE), i.e. as the guest would see them, and they become unpredictable
2078 if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector
2079 registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they
2080 overlap the FPU registers:
2081 0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers)
2082 0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers)
2083 0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers)
2085 MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the
2086 following id bit patterns:
2087 0x7020 0000 0003 01 <0:3> <reg:5>
2089 MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the
2090 following id bit patterns:
2091 0x7020 0000 0003 02 <0:3> <reg:5>
2094 4.69 KVM_GET_ONE_REG
2096 Capability: KVM_CAP_ONE_REG
2099 Parameters: struct kvm_one_reg (in and out)
2100 Returns: 0 on success, negative value on failure
2102 This ioctl allows to receive the value of a single register implemented
2103 in a vcpu. The register to read is indicated by the "id" field of the
2104 kvm_one_reg struct passed in. On success, the register value can be found
2105 at the memory location pointed to by "addr".
2107 The list of registers accessible using this interface is identical to the
2111 4.70 KVM_KVMCLOCK_CTRL
2113 Capability: KVM_CAP_KVMCLOCK_CTRL
2114 Architectures: Any that implement pvclocks (currently x86 only)
2117 Returns: 0 on success, -1 on error
2119 This signals to the host kernel that the specified guest is being paused by
2120 userspace. The host will set a flag in the pvclock structure that is checked
2121 from the soft lockup watchdog. The flag is part of the pvclock structure that
2122 is shared between guest and host, specifically the second bit of the flags
2123 field of the pvclock_vcpu_time_info structure. It will be set exclusively by
2124 the host and read/cleared exclusively by the guest. The guest operation of
2125 checking and clearing the flag must an atomic operation so
2126 load-link/store-conditional, or equivalent must be used. There are two cases
2127 where the guest will clear the flag: when the soft lockup watchdog timer resets
2128 itself or when a soft lockup is detected. This ioctl can be called any time
2129 after pausing the vcpu, but before it is resumed.
2134 Capability: KVM_CAP_SIGNAL_MSI
2135 Architectures: x86 arm arm64
2137 Parameters: struct kvm_msi (in)
2138 Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error
2140 Directly inject a MSI message. Only valid with in-kernel irqchip that handles
2152 flags: KVM_MSI_VALID_DEVID: devid contains a valid value. The per-VM
2153 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
2154 the device ID. If this capability is not available, userspace
2155 should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
2157 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
2158 for the device that wrote the MSI message. For PCI, this is usually a
2159 BFD identifier in the lower 16 bits.
2161 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
2162 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
2163 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
2164 address_hi must be zero.
2167 4.71 KVM_CREATE_PIT2
2169 Capability: KVM_CAP_PIT2
2172 Parameters: struct kvm_pit_config (in)
2173 Returns: 0 on success, -1 on error
2175 Creates an in-kernel device model for the i8254 PIT. This call is only valid
2176 after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following
2177 parameters have to be passed:
2179 struct kvm_pit_config {
2186 #define KVM_PIT_SPEAKER_DUMMY 1 /* emulate speaker port stub */
2188 PIT timer interrupts may use a per-VM kernel thread for injection. If it
2189 exists, this thread will have a name of the following pattern:
2191 kvm-pit/<owner-process-pid>
2193 When running a guest with elevated priorities, the scheduling parameters of
2194 this thread may have to be adjusted accordingly.
2196 This IOCTL replaces the obsolete KVM_CREATE_PIT.
2201 Capability: KVM_CAP_PIT_STATE2
2204 Parameters: struct kvm_pit_state2 (out)
2205 Returns: 0 on success, -1 on error
2207 Retrieves the state of the in-kernel PIT model. Only valid after
2208 KVM_CREATE_PIT2. The state is returned in the following structure:
2210 struct kvm_pit_state2 {
2211 struct kvm_pit_channel_state channels[3];
2218 /* disable PIT in HPET legacy mode */
2219 #define KVM_PIT_FLAGS_HPET_LEGACY 0x00000001
2221 This IOCTL replaces the obsolete KVM_GET_PIT.
2226 Capability: KVM_CAP_PIT_STATE2
2229 Parameters: struct kvm_pit_state2 (in)
2230 Returns: 0 on success, -1 on error
2232 Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2.
2233 See KVM_GET_PIT2 for details on struct kvm_pit_state2.
2235 This IOCTL replaces the obsolete KVM_SET_PIT.
2238 4.74 KVM_PPC_GET_SMMU_INFO
2240 Capability: KVM_CAP_PPC_GET_SMMU_INFO
2241 Architectures: powerpc
2244 Returns: 0 on success, -1 on error
2246 This populates and returns a structure describing the features of
2247 the "Server" class MMU emulation supported by KVM.
2248 This can in turn be used by userspace to generate the appropriate
2249 device-tree properties for the guest operating system.
2251 The structure contains some global information, followed by an
2252 array of supported segment page sizes:
2254 struct kvm_ppc_smmu_info {
2258 struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ];
2261 The supported flags are:
2263 - KVM_PPC_PAGE_SIZES_REAL:
2264 When that flag is set, guest page sizes must "fit" the backing
2265 store page sizes. When not set, any page size in the list can
2266 be used regardless of how they are backed by userspace.
2268 - KVM_PPC_1T_SEGMENTS
2269 The emulated MMU supports 1T segments in addition to the
2272 The "slb_size" field indicates how many SLB entries are supported
2274 The "sps" array contains 8 entries indicating the supported base
2275 page sizes for a segment in increasing order. Each entry is defined
2278 struct kvm_ppc_one_seg_page_size {
2279 __u32 page_shift; /* Base page shift of segment (or 0) */
2280 __u32 slb_enc; /* SLB encoding for BookS */
2281 struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ];
2284 An entry with a "page_shift" of 0 is unused. Because the array is
2285 organized in increasing order, a lookup can stop when encoutering
2288 The "slb_enc" field provides the encoding to use in the SLB for the
2289 page size. The bits are in positions such as the value can directly
2290 be OR'ed into the "vsid" argument of the slbmte instruction.
2292 The "enc" array is a list which for each of those segment base page
2293 size provides the list of supported actual page sizes (which can be
2294 only larger or equal to the base page size), along with the
2295 corresponding encoding in the hash PTE. Similarly, the array is
2296 8 entries sorted by increasing sizes and an entry with a "0" shift
2297 is an empty entry and a terminator:
2299 struct kvm_ppc_one_page_size {
2300 __u32 page_shift; /* Page shift (or 0) */
2301 __u32 pte_enc; /* Encoding in the HPTE (>>12) */
2304 The "pte_enc" field provides a value that can OR'ed into the hash
2305 PTE's RPN field (ie, it needs to be shifted left by 12 to OR it
2306 into the hash PTE second double word).
2310 Capability: KVM_CAP_IRQFD
2311 Architectures: x86 s390 arm arm64
2313 Parameters: struct kvm_irqfd (in)
2314 Returns: 0 on success, -1 on error
2316 Allows setting an eventfd to directly trigger a guest interrupt.
2317 kvm_irqfd.fd specifies the file descriptor to use as the eventfd and
2318 kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When
2319 an event is triggered on the eventfd, an interrupt is injected into
2320 the guest using the specified gsi pin. The irqfd is removed using
2321 the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd
2324 With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify
2325 mechanism allowing emulation of level-triggered, irqfd-based
2326 interrupts. When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an
2327 additional eventfd in the kvm_irqfd.resamplefd field. When operating
2328 in resample mode, posting of an interrupt through kvm_irq.fd asserts
2329 the specified gsi in the irqchip. When the irqchip is resampled, such
2330 as from an EOI, the gsi is de-asserted and the user is notified via
2331 kvm_irqfd.resamplefd. It is the user's responsibility to re-queue
2332 the interrupt if the device making use of it still requires service.
2333 Note that closing the resamplefd is not sufficient to disable the
2334 irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment
2335 and need not be specified with KVM_IRQFD_FLAG_DEASSIGN.
2337 On arm/arm64, gsi routing being supported, the following can happen:
2338 - in case no routing entry is associated to this gsi, injection fails
2339 - in case the gsi is associated to an irqchip routing entry,
2340 irqchip.pin + 32 corresponds to the injected SPI ID.
2341 - in case the gsi is associated to an MSI routing entry, the MSI
2342 message and device ID are translated into an LPI (support restricted
2343 to GICv3 ITS in-kernel emulation).
2345 4.76 KVM_PPC_ALLOCATE_HTAB
2347 Capability: KVM_CAP_PPC_ALLOC_HTAB
2348 Architectures: powerpc
2350 Parameters: Pointer to u32 containing hash table order (in/out)
2351 Returns: 0 on success, -1 on error
2353 This requests the host kernel to allocate an MMU hash table for a
2354 guest using the PAPR paravirtualization interface. This only does
2355 anything if the kernel is configured to use the Book 3S HV style of
2356 virtualization. Otherwise the capability doesn't exist and the ioctl
2357 returns an ENOTTY error. The rest of this description assumes Book 3S
2360 There must be no vcpus running when this ioctl is called; if there
2361 are, it will do nothing and return an EBUSY error.
2363 The parameter is a pointer to a 32-bit unsigned integer variable
2364 containing the order (log base 2) of the desired size of the hash
2365 table, which must be between 18 and 46. On successful return from the
2366 ioctl, the value will not be changed by the kernel.
2368 If no hash table has been allocated when any vcpu is asked to run
2369 (with the KVM_RUN ioctl), the host kernel will allocate a
2370 default-sized hash table (16 MB).
2372 If this ioctl is called when a hash table has already been allocated,
2373 with a different order from the existing hash table, the existing hash
2374 table will be freed and a new one allocated. If this is ioctl is
2375 called when a hash table has already been allocated of the same order
2376 as specified, the kernel will clear out the existing hash table (zero
2377 all HPTEs). In either case, if the guest is using the virtualized
2378 real-mode area (VRMA) facility, the kernel will re-create the VMRA
2379 HPTEs on the next KVM_RUN of any vcpu.
2381 4.77 KVM_S390_INTERRUPT
2385 Type: vm ioctl, vcpu ioctl
2386 Parameters: struct kvm_s390_interrupt (in)
2387 Returns: 0 on success, -1 on error
2389 Allows to inject an interrupt to the guest. Interrupts can be floating
2390 (vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type.
2392 Interrupt parameters are passed via kvm_s390_interrupt:
2394 struct kvm_s390_interrupt {
2400 type can be one of the following:
2402 KVM_S390_SIGP_STOP (vcpu) - sigp stop; optional flags in parm
2403 KVM_S390_PROGRAM_INT (vcpu) - program check; code in parm
2404 KVM_S390_SIGP_SET_PREFIX (vcpu) - sigp set prefix; prefix address in parm
2405 KVM_S390_RESTART (vcpu) - restart
2406 KVM_S390_INT_CLOCK_COMP (vcpu) - clock comparator interrupt
2407 KVM_S390_INT_CPU_TIMER (vcpu) - CPU timer interrupt
2408 KVM_S390_INT_VIRTIO (vm) - virtio external interrupt; external interrupt
2409 parameters in parm and parm64
2410 KVM_S390_INT_SERVICE (vm) - sclp external interrupt; sclp parameter in parm
2411 KVM_S390_INT_EMERGENCY (vcpu) - sigp emergency; source cpu in parm
2412 KVM_S390_INT_EXTERNAL_CALL (vcpu) - sigp external call; source cpu in parm
2413 KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm) - compound value to indicate an
2414 I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel);
2415 I/O interruption parameters in parm (subchannel) and parm64 (intparm,
2416 interruption subclass)
2417 KVM_S390_MCHK (vm, vcpu) - machine check interrupt; cr 14 bits in parm,
2418 machine check interrupt code in parm64 (note that
2419 machine checks needing further payload are not
2420 supported by this ioctl)
2422 Note that the vcpu ioctl is asynchronous to vcpu execution.
2424 4.78 KVM_PPC_GET_HTAB_FD
2426 Capability: KVM_CAP_PPC_HTAB_FD
2427 Architectures: powerpc
2429 Parameters: Pointer to struct kvm_get_htab_fd (in)
2430 Returns: file descriptor number (>= 0) on success, -1 on error
2432 This returns a file descriptor that can be used either to read out the
2433 entries in the guest's hashed page table (HPT), or to write entries to
2434 initialize the HPT. The returned fd can only be written to if the
2435 KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and
2436 can only be read if that bit is clear. The argument struct looks like
2439 /* For KVM_PPC_GET_HTAB_FD */
2440 struct kvm_get_htab_fd {
2446 /* Values for kvm_get_htab_fd.flags */
2447 #define KVM_GET_HTAB_BOLTED_ONLY ((__u64)0x1)
2448 #define KVM_GET_HTAB_WRITE ((__u64)0x2)
2450 The `start_index' field gives the index in the HPT of the entry at
2451 which to start reading. It is ignored when writing.
2453 Reads on the fd will initially supply information about all
2454 "interesting" HPT entries. Interesting entries are those with the
2455 bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise
2456 all entries. When the end of the HPT is reached, the read() will
2457 return. If read() is called again on the fd, it will start again from
2458 the beginning of the HPT, but will only return HPT entries that have
2459 changed since they were last read.
2461 Data read or written is structured as a header (8 bytes) followed by a
2462 series of valid HPT entries (16 bytes) each. The header indicates how
2463 many valid HPT entries there are and how many invalid entries follow
2464 the valid entries. The invalid entries are not represented explicitly
2465 in the stream. The header format is:
2467 struct kvm_get_htab_header {
2473 Writes to the fd create HPT entries starting at the index given in the
2474 header; first `n_valid' valid entries with contents from the data
2475 written, then `n_invalid' invalid entries, invalidating any previously
2476 valid entries found.
2478 4.79 KVM_CREATE_DEVICE
2480 Capability: KVM_CAP_DEVICE_CTRL
2482 Parameters: struct kvm_create_device (in/out)
2483 Returns: 0 on success, -1 on error
2485 ENODEV: The device type is unknown or unsupported
2486 EEXIST: Device already created, and this type of device may not
2487 be instantiated multiple times
2489 Other error conditions may be defined by individual device types or
2490 have their standard meanings.
2492 Creates an emulated device in the kernel. The file descriptor returned
2493 in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR.
2495 If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the
2496 device type is supported (not necessarily whether it can be created
2499 Individual devices should not define flags. Attributes should be used
2500 for specifying any behavior that is not implied by the device type
2503 struct kvm_create_device {
2504 __u32 type; /* in: KVM_DEV_TYPE_xxx */
2505 __u32 fd; /* out: device handle */
2506 __u32 flags; /* in: KVM_CREATE_DEVICE_xxx */
2509 4.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR
2511 Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
2512 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
2513 Type: device ioctl, vm ioctl, vcpu ioctl
2514 Parameters: struct kvm_device_attr
2515 Returns: 0 on success, -1 on error
2517 ENXIO: The group or attribute is unknown/unsupported for this device
2518 or hardware support is missing.
2519 EPERM: The attribute cannot (currently) be accessed this way
2520 (e.g. read-only attribute, or attribute that only makes
2521 sense when the device is in a different state)
2523 Other error conditions may be defined by individual device types.
2525 Gets/sets a specified piece of device configuration and/or state. The
2526 semantics are device-specific. See individual device documentation in
2527 the "devices" directory. As with ONE_REG, the size of the data
2528 transferred is defined by the particular attribute.
2530 struct kvm_device_attr {
2531 __u32 flags; /* no flags currently defined */
2532 __u32 group; /* device-defined */
2533 __u64 attr; /* group-defined */
2534 __u64 addr; /* userspace address of attr data */
2537 4.81 KVM_HAS_DEVICE_ATTR
2539 Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
2540 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
2541 Type: device ioctl, vm ioctl, vcpu ioctl
2542 Parameters: struct kvm_device_attr
2543 Returns: 0 on success, -1 on error
2545 ENXIO: The group or attribute is unknown/unsupported for this device
2546 or hardware support is missing.
2548 Tests whether a device supports a particular attribute. A successful
2549 return indicates the attribute is implemented. It does not necessarily
2550 indicate that the attribute can be read or written in the device's
2551 current state. "addr" is ignored.
2553 4.82 KVM_ARM_VCPU_INIT
2556 Architectures: arm, arm64
2558 Parameters: struct kvm_vcpu_init (in)
2559 Returns: 0 on success; -1 on error
2561 Â EINVAL: Â Â Â the target is unknown, or the combination of features is invalid.
2562 Â ENOENT: Â Â Â a features bit specified is unknown.
2564 This tells KVM what type of CPU to present to the guest, and what
2565 optional features it should have. Â This will cause a reset of the cpu
2566 registers to their initial values. Â If this is not called, KVM_RUN will
2567 return ENOEXEC for that vcpu.
2569 Note that because some registers reflect machine topology, all vcpus
2570 should be created before this ioctl is invoked.
2572 Userspace can call this function multiple times for a given vcpu, including
2573 after the vcpu has been run. This will reset the vcpu to its initial
2574 state. All calls to this function after the initial call must use the same
2575 target and same set of feature flags, otherwise EINVAL will be returned.
2578 - KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state.
2579 Depends on KVM_CAP_ARM_PSCI. If not set, the CPU will be powered on
2580 and execute guest code when KVM_RUN is called.
2581 - KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode.
2582 Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only).
2583 - KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 (or a future revision
2584 backward compatible with v0.2) for the CPU.
2585 Depends on KVM_CAP_ARM_PSCI_0_2.
2586 - KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU.
2587 Depends on KVM_CAP_ARM_PMU_V3.
2590 4.83 KVM_ARM_PREFERRED_TARGET
2593 Architectures: arm, arm64
2595 Parameters: struct struct kvm_vcpu_init (out)
2596 Returns: 0 on success; -1 on error
2598 ENODEV: no preferred target available for the host
2600 This queries KVM for preferred CPU target type which can be emulated
2601 by KVM on underlying host.
2603 The ioctl returns struct kvm_vcpu_init instance containing information
2604 about preferred CPU target type and recommended features for it. The
2605 kvm_vcpu_init->features bitmap returned will have feature bits set if
2606 the preferred target recommends setting these features, but this is
2609 The information returned by this ioctl can be used to prepare an instance
2610 of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in
2611 in VCPU matching underlying host.
2614 4.84 KVM_GET_REG_LIST
2617 Architectures: arm, arm64, mips
2619 Parameters: struct kvm_reg_list (in/out)
2620 Returns: 0 on success; -1 on error
2622 Â E2BIG: Â Â Â Â the reg index list is too big to fit in the array specified by
2623 Â Â Â Â Â Â Â Â Â Â Â Â the user (the number required will be written into n).
2625 struct kvm_reg_list {
2626 __u64 n; /* number of registers in reg[] */
2630 This ioctl returns the guest registers that are supported for the
2631 KVM_GET_ONE_REG/KVM_SET_ONE_REG calls.
2634 4.85 KVM_ARM_SET_DEVICE_ADDR (deprecated)
2636 Capability: KVM_CAP_ARM_SET_DEVICE_ADDR
2637 Architectures: arm, arm64
2639 Parameters: struct kvm_arm_device_address (in)
2640 Returns: 0 on success, -1 on error
2642 ENODEV: The device id is unknown
2643 ENXIO: Device not supported on current system
2644 EEXIST: Address already set
2645 E2BIG: Address outside guest physical address space
2646 EBUSY: Address overlaps with other device range
2648 struct kvm_arm_device_addr {
2653 Specify a device address in the guest's physical address space where guests
2654 can access emulated or directly exposed devices, which the host kernel needs
2655 to know about. The id field is an architecture specific identifier for a
2658 ARM/arm64 divides the id field into two parts, a device id and an
2659 address type id specific to the individual device.
2661 Â bits: | 63 ... 32 | 31 ... 16 | 15 ... 0 |
2662 field: | 0x00000000 | device id | addr type id |
2664 ARM/arm64 currently only require this when using the in-kernel GIC
2665 support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2
2666 as the device id. When setting the base address for the guest's
2667 mapping of the VGIC virtual CPU and distributor interface, the ioctl
2668 must be called after calling KVM_CREATE_IRQCHIP, but before calling
2669 KVM_RUN on any of the VCPUs. Calling this ioctl twice for any of the
2670 base addresses will return -EEXIST.
2672 Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API
2673 should be used instead.
2676 4.86 KVM_PPC_RTAS_DEFINE_TOKEN
2678 Capability: KVM_CAP_PPC_RTAS
2681 Parameters: struct kvm_rtas_token_args
2682 Returns: 0 on success, -1 on error
2684 Defines a token value for a RTAS (Run Time Abstraction Services)
2685 service in order to allow it to be handled in the kernel. The
2686 argument struct gives the name of the service, which must be the name
2687 of a service that has a kernel-side implementation. If the token
2688 value is non-zero, it will be associated with that service, and
2689 subsequent RTAS calls by the guest specifying that token will be
2690 handled by the kernel. If the token value is 0, then any token
2691 associated with the service will be forgotten, and subsequent RTAS
2692 calls by the guest for that service will be passed to userspace to be
2695 4.87 KVM_SET_GUEST_DEBUG
2697 Capability: KVM_CAP_SET_GUEST_DEBUG
2698 Architectures: x86, s390, ppc, arm64
2700 Parameters: struct kvm_guest_debug (in)
2701 Returns: 0 on success; -1 on error
2703 struct kvm_guest_debug {
2706 struct kvm_guest_debug_arch arch;
2709 Set up the processor specific debug registers and configure vcpu for
2710 handling guest debug events. There are two parts to the structure, the
2711 first a control bitfield indicates the type of debug events to handle
2712 when running. Common control bits are:
2714 - KVM_GUESTDBG_ENABLE: guest debugging is enabled
2715 - KVM_GUESTDBG_SINGLESTEP: the next run should single-step
2717 The top 16 bits of the control field are architecture specific control
2718 flags which can include the following:
2720 - KVM_GUESTDBG_USE_SW_BP: using software breakpoints [x86, arm64]
2721 - KVM_GUESTDBG_USE_HW_BP: using hardware breakpoints [x86, s390, arm64]
2722 - KVM_GUESTDBG_INJECT_DB: inject DB type exception [x86]
2723 - KVM_GUESTDBG_INJECT_BP: inject BP type exception [x86]
2724 - KVM_GUESTDBG_EXIT_PENDING: trigger an immediate guest exit [s390]
2726 For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints
2727 are enabled in memory so we need to ensure breakpoint exceptions are
2728 correctly trapped and the KVM run loop exits at the breakpoint and not
2729 running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP
2730 we need to ensure the guest vCPUs architecture specific registers are
2731 updated to the correct (supplied) values.
2733 The second part of the structure is architecture specific and
2734 typically contains a set of debug registers.
2736 For arm64 the number of debug registers is implementation defined and
2737 can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and
2738 KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number
2739 indicating the number of supported registers.
2741 When debug events exit the main run loop with the reason
2742 KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run
2743 structure containing architecture specific debug information.
2745 4.88 KVM_GET_EMULATED_CPUID
2747 Capability: KVM_CAP_EXT_EMUL_CPUID
2750 Parameters: struct kvm_cpuid2 (in/out)
2751 Returns: 0 on success, -1 on error
2756 struct kvm_cpuid_entry2 entries[0];
2759 The member 'flags' is used for passing flags from userspace.
2761 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
2762 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1)
2763 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2)
2765 struct kvm_cpuid_entry2 {
2776 This ioctl returns x86 cpuid features which are emulated by
2777 kvm.Userspace can use the information returned by this ioctl to query
2778 which features are emulated by kvm instead of being present natively.
2780 Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2
2781 structure with the 'nent' field indicating the number of entries in
2782 the variable-size array 'entries'. If the number of entries is too low
2783 to describe the cpu capabilities, an error (E2BIG) is returned. If the
2784 number is too high, the 'nent' field is adjusted and an error (ENOMEM)
2785 is returned. If the number is just right, the 'nent' field is adjusted
2786 to the number of valid entries in the 'entries' array, which is then
2789 The entries returned are the set CPUID bits of the respective features
2790 which kvm emulates, as returned by the CPUID instruction, with unknown
2791 or unsupported feature bits cleared.
2793 Features like x2apic, for example, may not be present in the host cpu
2794 but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be
2795 emulated efficiently and thus not included here.
2797 The fields in each entry are defined as follows:
2799 function: the eax value used to obtain the entry
2800 index: the ecx value used to obtain the entry (for entries that are
2802 flags: an OR of zero or more of the following:
2803 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
2804 if the index field is valid
2805 KVM_CPUID_FLAG_STATEFUL_FUNC:
2806 if cpuid for this function returns different values for successive
2807 invocations; there will be several entries with the same function,
2808 all with this flag set
2809 KVM_CPUID_FLAG_STATE_READ_NEXT:
2810 for KVM_CPUID_FLAG_STATEFUL_FUNC entries, set if this entry is
2811 the first entry to be read by a cpu
2812 eax, ebx, ecx, edx: the values returned by the cpuid instruction for
2813 this function/index combination
2815 4.89 KVM_S390_MEM_OP
2817 Capability: KVM_CAP_S390_MEM_OP
2820 Parameters: struct kvm_s390_mem_op (in)
2821 Returns: = 0 on success,
2822 < 0 on generic error (e.g. -EFAULT or -ENOMEM),
2823 > 0 if an exception occurred while walking the page tables
2825 Read or write data from/to the logical (virtual) memory of a VCPU.
2827 Parameters are specified via the following structure:
2829 struct kvm_s390_mem_op {
2830 __u64 gaddr; /* the guest address */
2831 __u64 flags; /* flags */
2832 __u32 size; /* amount of bytes */
2833 __u32 op; /* type of operation */
2834 __u64 buf; /* buffer in userspace */
2835 __u8 ar; /* the access register number */
2836 __u8 reserved[31]; /* should be set to 0 */
2839 The type of operation is specified in the "op" field. It is either
2840 KVM_S390_MEMOP_LOGICAL_READ for reading from logical memory space or
2841 KVM_S390_MEMOP_LOGICAL_WRITE for writing to logical memory space. The
2842 KVM_S390_MEMOP_F_CHECK_ONLY flag can be set in the "flags" field to check
2843 whether the corresponding memory access would create an access exception
2844 (without touching the data in the memory at the destination). In case an
2845 access exception occurred while walking the MMU tables of the guest, the
2846 ioctl returns a positive error number to indicate the type of exception.
2847 This exception is also raised directly at the corresponding VCPU if the
2848 flag KVM_S390_MEMOP_F_INJECT_EXCEPTION is set in the "flags" field.
2850 The start address of the memory region has to be specified in the "gaddr"
2851 field, and the length of the region in the "size" field. "buf" is the buffer
2852 supplied by the userspace application where the read data should be written
2853 to for KVM_S390_MEMOP_LOGICAL_READ, or where the data that should be written
2854 is stored for a KVM_S390_MEMOP_LOGICAL_WRITE. "buf" is unused and can be NULL
2855 when KVM_S390_MEMOP_F_CHECK_ONLY is specified. "ar" designates the access
2856 register number to be used.
2858 The "reserved" field is meant for future extensions. It is not used by
2859 KVM with the currently defined set of flags.
2861 4.90 KVM_S390_GET_SKEYS
2863 Capability: KVM_CAP_S390_SKEYS
2866 Parameters: struct kvm_s390_skeys
2867 Returns: 0 on success, KVM_S390_GET_KEYS_NONE if guest is not using storage
2868 keys, negative value on error
2870 This ioctl is used to get guest storage key values on the s390
2871 architecture. The ioctl takes parameters via the kvm_s390_skeys struct.
2873 struct kvm_s390_skeys {
2876 __u64 skeydata_addr;
2881 The start_gfn field is the number of the first guest frame whose storage keys
2884 The count field is the number of consecutive frames (starting from start_gfn)
2885 whose storage keys to get. The count field must be at least 1 and the maximum
2886 allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range
2887 will cause the ioctl to return -EINVAL.
2889 The skeydata_addr field is the address to a buffer large enough to hold count
2890 bytes. This buffer will be filled with storage key data by the ioctl.
2892 4.91 KVM_S390_SET_SKEYS
2894 Capability: KVM_CAP_S390_SKEYS
2897 Parameters: struct kvm_s390_skeys
2898 Returns: 0 on success, negative value on error
2900 This ioctl is used to set guest storage key values on the s390
2901 architecture. The ioctl takes parameters via the kvm_s390_skeys struct.
2902 See section on KVM_S390_GET_SKEYS for struct definition.
2904 The start_gfn field is the number of the first guest frame whose storage keys
2907 The count field is the number of consecutive frames (starting from start_gfn)
2908 whose storage keys to get. The count field must be at least 1 and the maximum
2909 allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range
2910 will cause the ioctl to return -EINVAL.
2912 The skeydata_addr field is the address to a buffer containing count bytes of
2913 storage keys. Each byte in the buffer will be set as the storage key for a
2914 single frame starting at start_gfn for count frames.
2916 Note: If any architecturally invalid key value is found in the given data then
2917 the ioctl will return -EINVAL.
2921 Capability: KVM_CAP_S390_INJECT_IRQ
2924 Parameters: struct kvm_s390_irq (in)
2925 Returns: 0 on success, -1 on error
2927 EINVAL: interrupt type is invalid
2928 type is KVM_S390_SIGP_STOP and flag parameter is invalid value
2929 type is KVM_S390_INT_EXTERNAL_CALL and code is bigger
2930 than the maximum of VCPUs
2931 EBUSY: type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped
2932 type is KVM_S390_SIGP_STOP and a stop irq is already pending
2933 type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt
2936 Allows to inject an interrupt to the guest.
2938 Using struct kvm_s390_irq as a parameter allows
2939 to inject additional payload which is not
2940 possible via KVM_S390_INTERRUPT.
2942 Interrupt parameters are passed via kvm_s390_irq:
2944 struct kvm_s390_irq {
2947 struct kvm_s390_io_info io;
2948 struct kvm_s390_ext_info ext;
2949 struct kvm_s390_pgm_info pgm;
2950 struct kvm_s390_emerg_info emerg;
2951 struct kvm_s390_extcall_info extcall;
2952 struct kvm_s390_prefix_info prefix;
2953 struct kvm_s390_stop_info stop;
2954 struct kvm_s390_mchk_info mchk;
2959 type can be one of the following:
2961 KVM_S390_SIGP_STOP - sigp stop; parameter in .stop
2962 KVM_S390_PROGRAM_INT - program check; parameters in .pgm
2963 KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix
2964 KVM_S390_RESTART - restart; no parameters
2965 KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters
2966 KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters
2967 KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg
2968 KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall
2969 KVM_S390_MCHK - machine check interrupt; parameters in .mchk
2972 Note that the vcpu ioctl is asynchronous to vcpu execution.
2974 4.94 KVM_S390_GET_IRQ_STATE
2976 Capability: KVM_CAP_S390_IRQ_STATE
2979 Parameters: struct kvm_s390_irq_state (out)
2980 Returns: >= number of bytes copied into buffer,
2981 -EINVAL if buffer size is 0,
2982 -ENOBUFS if buffer size is too small to fit all pending interrupts,
2983 -EFAULT if the buffer address was invalid
2985 This ioctl allows userspace to retrieve the complete state of all currently
2986 pending interrupts in a single buffer. Use cases include migration
2987 and introspection. The parameter structure contains the address of a
2988 userspace buffer and its length:
2990 struct kvm_s390_irq_state {
2992 __u32 flags; /* will stay unused for compatibility reasons */
2994 __u32 reserved[4]; /* will stay unused for compatibility reasons */
2997 Userspace passes in the above struct and for each pending interrupt a
2998 struct kvm_s390_irq is copied to the provided buffer.
3000 The structure contains a flags and a reserved field for future extensions. As
3001 the kernel never checked for flags == 0 and QEMU never pre-zeroed flags and
3002 reserved, these fields can not be used in the future without breaking
3005 If -ENOBUFS is returned the buffer provided was too small and userspace
3006 may retry with a bigger buffer.
3008 4.95 KVM_S390_SET_IRQ_STATE
3010 Capability: KVM_CAP_S390_IRQ_STATE
3013 Parameters: struct kvm_s390_irq_state (in)
3014 Returns: 0 on success,
3015 -EFAULT if the buffer address was invalid,
3016 -EINVAL for an invalid buffer length (see below),
3017 -EBUSY if there were already interrupts pending,
3018 errors occurring when actually injecting the
3019 interrupt. See KVM_S390_IRQ.
3021 This ioctl allows userspace to set the complete state of all cpu-local
3022 interrupts currently pending for the vcpu. It is intended for restoring
3023 interrupt state after a migration. The input parameter is a userspace buffer
3024 containing a struct kvm_s390_irq_state:
3026 struct kvm_s390_irq_state {
3028 __u32 flags; /* will stay unused for compatibility reasons */
3030 __u32 reserved[4]; /* will stay unused for compatibility reasons */
3033 The restrictions for flags and reserved apply as well.
3034 (see KVM_S390_GET_IRQ_STATE)
3036 The userspace memory referenced by buf contains a struct kvm_s390_irq
3037 for each interrupt to be injected into the guest.
3038 If one of the interrupts could not be injected for some reason the
3041 len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0
3042 and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq),
3043 which is the maximum number of possibly pending cpu-local interrupts.
3047 Capability: KVM_CAP_X86_SMM
3051 Returns: 0 on success, -1 on error
3053 Queues an SMI on the thread's vcpu.
3055 4.97 KVM_CAP_PPC_MULTITCE
3057 Capability: KVM_CAP_PPC_MULTITCE
3061 This capability means the kernel is capable of handling hypercalls
3062 H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user
3063 space. This significantly accelerates DMA operations for PPC KVM guests.
3064 User space should expect that its handlers for these hypercalls
3065 are not going to be called if user space previously registered LIOBN
3066 in KVM (via KVM_CREATE_SPAPR_TCE or similar calls).
3068 In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest,
3069 user space might have to advertise it for the guest. For example,
3070 IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is
3071 present in the "ibm,hypertas-functions" device-tree property.
3073 The hypercalls mentioned above may or may not be processed successfully
3074 in the kernel based fast path. If they can not be handled by the kernel,
3075 they will get passed on to user space. So user space still has to have
3076 an implementation for these despite the in kernel acceleration.
3078 This capability is always enabled.
3080 4.98 KVM_CREATE_SPAPR_TCE_64
3082 Capability: KVM_CAP_SPAPR_TCE_64
3083 Architectures: powerpc
3085 Parameters: struct kvm_create_spapr_tce_64 (in)
3086 Returns: file descriptor for manipulating the created TCE table
3088 This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit
3089 windows, described in 4.62 KVM_CREATE_SPAPR_TCE
3091 This capability uses extended struct in ioctl interface:
3093 /* for KVM_CAP_SPAPR_TCE_64 */
3094 struct kvm_create_spapr_tce_64 {
3098 __u64 offset; /* in pages */
3099 __u64 size; /* in pages */
3102 The aim of extension is to support an additional bigger DMA window with
3103 a variable page size.
3104 KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and
3105 a bus offset of the corresponding DMA window, @size and @offset are numbers
3108 @flags are not used at the moment.
3110 The rest of functionality is identical to KVM_CREATE_SPAPR_TCE.
3112 4.99 KVM_REINJECT_CONTROL
3114 Capability: KVM_CAP_REINJECT_CONTROL
3117 Parameters: struct kvm_reinject_control (in)
3118 Returns: 0 on success,
3119 -EFAULT if struct kvm_reinject_control cannot be read,
3120 -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier.
3122 i8254 (PIT) has two modes, reinject and !reinject. The default is reinject,
3123 where KVM queues elapsed i8254 ticks and monitors completion of interrupt from
3124 vector(s) that i8254 injects. Reinject mode dequeues a tick and injects its
3125 interrupt whenever there isn't a pending interrupt from i8254.
3126 !reinject mode injects an interrupt as soon as a tick arrives.
3128 struct kvm_reinject_control {
3133 pit_reinject = 0 (!reinject mode) is recommended, unless running an old
3134 operating system that uses the PIT for timing (e.g. Linux 2.4.x).
3136 4.100 KVM_PPC_CONFIGURE_V3_MMU
3138 Capability: KVM_CAP_PPC_RADIX_MMU or KVM_CAP_PPC_HASH_MMU_V3
3141 Parameters: struct kvm_ppc_mmuv3_cfg (in)
3142 Returns: 0 on success,
3143 -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read,
3144 -EINVAL if the configuration is invalid
3146 This ioctl controls whether the guest will use radix or HPT (hashed
3147 page table) translation, and sets the pointer to the process table for
3150 struct kvm_ppc_mmuv3_cfg {
3152 __u64 process_table;
3155 There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and
3156 KVM_PPC_MMUV3_GTSE. KVM_PPC_MMUV3_RADIX, if set, configures the guest
3157 to use radix tree translation, and if clear, to use HPT translation.
3158 KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest
3159 to be able to use the global TLB and SLB invalidation instructions;
3160 if clear, the guest may not use these instructions.
3162 The process_table field specifies the address and size of the guest
3163 process table, which is in the guest's space. This field is formatted
3164 as the second doubleword of the partition table entry, as defined in
3165 the Power ISA V3.00, Book III section 5.7.6.1.
3167 4.101 KVM_PPC_GET_RMMU_INFO
3169 Capability: KVM_CAP_PPC_RADIX_MMU
3172 Parameters: struct kvm_ppc_rmmu_info (out)
3173 Returns: 0 on success,
3174 -EFAULT if struct kvm_ppc_rmmu_info cannot be written,
3175 -EINVAL if no useful information can be returned
3177 This ioctl returns a structure containing two things: (a) a list
3178 containing supported radix tree geometries, and (b) a list that maps
3179 page sizes to put in the "AP" (actual page size) field for the tlbie
3180 (TLB invalidate entry) instruction.
3182 struct kvm_ppc_rmmu_info {
3183 struct kvm_ppc_radix_geom {
3188 __u32 ap_encodings[8];
3191 The geometries[] field gives up to 8 supported geometries for the
3192 radix page table, in terms of the log base 2 of the smallest page
3193 size, and the number of bits indexed at each level of the tree, from
3194 the PTE level up to the PGD level in that order. Any unused entries
3195 will have 0 in the page_shift field.
3197 The ap_encodings gives the supported page sizes and their AP field
3198 encodings, encoded with the AP value in the top 3 bits and the log
3199 base 2 of the page size in the bottom 6 bits.
3201 4.102 KVM_PPC_RESIZE_HPT_PREPARE
3203 Capability: KVM_CAP_SPAPR_RESIZE_HPT
3204 Architectures: powerpc
3206 Parameters: struct kvm_ppc_resize_hpt (in)
3207 Returns: 0 on successful completion,
3208 >0 if a new HPT is being prepared, the value is an estimated
3209 number of milliseconds until preparation is complete
3210 -EFAULT if struct kvm_reinject_control cannot be read,
3211 -EINVAL if the supplied shift or flags are invalid
3212 -ENOMEM if unable to allocate the new HPT
3213 -ENOSPC if there was a hash collision when moving existing
3214 HPT entries to the new HPT
3215 -EIO on other error conditions
3217 Used to implement the PAPR extension for runtime resizing of a guest's
3218 Hashed Page Table (HPT). Specifically this starts, stops or monitors
3219 the preparation of a new potential HPT for the guest, essentially
3220 implementing the H_RESIZE_HPT_PREPARE hypercall.
3222 If called with shift > 0 when there is no pending HPT for the guest,
3223 this begins preparation of a new pending HPT of size 2^(shift) bytes.
3224 It then returns a positive integer with the estimated number of
3225 milliseconds until preparation is complete.
3227 If called when there is a pending HPT whose size does not match that
3228 requested in the parameters, discards the existing pending HPT and
3229 creates a new one as above.
3231 If called when there is a pending HPT of the size requested, will:
3232 * If preparation of the pending HPT is already complete, return 0
3233 * If preparation of the pending HPT has failed, return an error
3234 code, then discard the pending HPT.
3235 * If preparation of the pending HPT is still in progress, return an
3236 estimated number of milliseconds until preparation is complete.
3238 If called with shift == 0, discards any currently pending HPT and
3239 returns 0 (i.e. cancels any in-progress preparation).
3241 flags is reserved for future expansion, currently setting any bits in
3242 flags will result in an -EINVAL.
3244 Normally this will be called repeatedly with the same parameters until
3245 it returns <= 0. The first call will initiate preparation, subsequent
3246 ones will monitor preparation until it completes or fails.
3248 struct kvm_ppc_resize_hpt {
3254 4.103 KVM_PPC_RESIZE_HPT_COMMIT
3256 Capability: KVM_CAP_SPAPR_RESIZE_HPT
3257 Architectures: powerpc
3259 Parameters: struct kvm_ppc_resize_hpt (in)
3260 Returns: 0 on successful completion,
3261 -EFAULT if struct kvm_reinject_control cannot be read,
3262 -EINVAL if the supplied shift or flags are invalid
3263 -ENXIO is there is no pending HPT, or the pending HPT doesn't
3264 have the requested size
3265 -EBUSY if the pending HPT is not fully prepared
3266 -ENOSPC if there was a hash collision when moving existing
3267 HPT entries to the new HPT
3268 -EIO on other error conditions
3270 Used to implement the PAPR extension for runtime resizing of a guest's
3271 Hashed Page Table (HPT). Specifically this requests that the guest be
3272 transferred to working with the new HPT, essentially implementing the
3273 H_RESIZE_HPT_COMMIT hypercall.
3275 This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has
3276 returned 0 with the same parameters. In other cases
3277 KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or
3278 -EBUSY, though others may be possible if the preparation was started,
3281 This will have undefined effects on the guest if it has not already
3282 placed itself in a quiescent state where no vcpu will make MMU enabled
3285 On succsful completion, the pending HPT will become the guest's active
3286 HPT and the previous HPT will be discarded.
3288 On failure, the guest will still be operating on its previous HPT.
3290 struct kvm_ppc_resize_hpt {
3296 4.104 KVM_X86_GET_MCE_CAP_SUPPORTED
3298 Capability: KVM_CAP_MCE
3301 Parameters: u64 mce_cap (out)
3302 Returns: 0 on success, -1 on error
3304 Returns supported MCE capabilities. The u64 mce_cap parameter
3305 has the same format as the MSR_IA32_MCG_CAP register. Supported
3306 capabilities will have the corresponding bits set.
3308 4.105 KVM_X86_SETUP_MCE
3310 Capability: KVM_CAP_MCE
3313 Parameters: u64 mcg_cap (in)
3314 Returns: 0 on success,
3315 -EFAULT if u64 mcg_cap cannot be read,
3316 -EINVAL if the requested number of banks is invalid,
3317 -EINVAL if requested MCE capability is not supported.
3319 Initializes MCE support for use. The u64 mcg_cap parameter
3320 has the same format as the MSR_IA32_MCG_CAP register and
3321 specifies which capabilities should be enabled. The maximum
3322 supported number of error-reporting banks can be retrieved when
3323 checking for KVM_CAP_MCE. The supported capabilities can be
3324 retrieved with KVM_X86_GET_MCE_CAP_SUPPORTED.
3326 4.106 KVM_X86_SET_MCE
3328 Capability: KVM_CAP_MCE
3331 Parameters: struct kvm_x86_mce (in)
3332 Returns: 0 on success,
3333 -EFAULT if struct kvm_x86_mce cannot be read,
3334 -EINVAL if the bank number is invalid,
3335 -EINVAL if VAL bit is not set in status field.
3337 Inject a machine check error (MCE) into the guest. The input
3340 struct kvm_x86_mce {
3350 If the MCE being reported is an uncorrected error, KVM will
3351 inject it as an MCE exception into the guest. If the guest
3352 MCG_STATUS register reports that an MCE is in progress, KVM
3353 causes an KVM_EXIT_SHUTDOWN vmexit.
3355 Otherwise, if the MCE is a corrected error, KVM will just
3356 store it in the corresponding bank (provided this bank is
3357 not holding a previously reported uncorrected error).
3359 4.107 KVM_S390_GET_CMMA_BITS
3361 Capability: KVM_CAP_S390_CMMA_MIGRATION
3364 Parameters: struct kvm_s390_cmma_log (in, out)
3365 Returns: 0 on success, a negative value on error
3367 This ioctl is used to get the values of the CMMA bits on the s390
3368 architecture. It is meant to be used in two scenarios:
3369 - During live migration to save the CMMA values. Live migration needs
3370 to be enabled via the KVM_REQ_START_MIGRATION VM property.
3371 - To non-destructively peek at the CMMA values, with the flag
3372 KVM_S390_CMMA_PEEK set.
3374 The ioctl takes parameters via the kvm_s390_cmma_log struct. The desired
3375 values are written to a buffer whose location is indicated via the "values"
3376 member in the kvm_s390_cmma_log struct. The values in the input struct are
3377 also updated as needed.
3378 Each CMMA value takes up one byte.
3380 struct kvm_s390_cmma_log {
3391 start_gfn is the number of the first guest frame whose CMMA values are
3394 count is the length of the buffer in bytes,
3396 values points to the buffer where the result will be written to.
3398 If count is greater than KVM_S390_SKEYS_MAX, then it is considered to be
3399 KVM_S390_SKEYS_MAX. KVM_S390_SKEYS_MAX is re-used for consistency with
3402 The result is written in the buffer pointed to by the field values, and
3403 the values of the input parameter are updated as follows.
3405 Depending on the flags, different actions are performed. The only
3406 supported flag so far is KVM_S390_CMMA_PEEK.
3408 The default behaviour if KVM_S390_CMMA_PEEK is not set is:
3409 start_gfn will indicate the first page frame whose CMMA bits were dirty.
3410 It is not necessarily the same as the one passed as input, as clean pages
3413 count will indicate the number of bytes actually written in the buffer.
3414 It can (and very often will) be smaller than the input value, since the
3415 buffer is only filled until 16 bytes of clean values are found (which
3416 are then not copied in the buffer). Since a CMMA migration block needs
3417 the base address and the length, for a total of 16 bytes, we will send
3418 back some clean data if there is some dirty data afterwards, as long as
3419 the size of the clean data does not exceed the size of the header. This
3420 allows to minimize the amount of data to be saved or transferred over
3421 the network at the expense of more roundtrips to userspace. The next
3422 invocation of the ioctl will skip over all the clean values, saving
3423 potentially more than just the 16 bytes we found.
3425 If KVM_S390_CMMA_PEEK is set:
3426 the existing storage attributes are read even when not in migration
3427 mode, and no other action is performed;
3429 the output start_gfn will be equal to the input start_gfn,
3431 the output count will be equal to the input count, except if the end of
3432 memory has been reached.
3435 the field "remaining" will indicate the total number of dirty CMMA values
3436 still remaining, or 0 if KVM_S390_CMMA_PEEK is set and migration mode is
3441 values points to the userspace buffer where the result will be stored.
3443 This ioctl can fail with -ENOMEM if not enough memory can be allocated to
3444 complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
3445 KVM_S390_CMMA_PEEK is not set but migration mode was not enabled, with
3446 -EFAULT if the userspace address is invalid or if no page table is
3447 present for the addresses (e.g. when using hugepages).
3449 4.108 KVM_S390_SET_CMMA_BITS
3451 Capability: KVM_CAP_S390_CMMA_MIGRATION
3454 Parameters: struct kvm_s390_cmma_log (in)
3455 Returns: 0 on success, a negative value on error
3457 This ioctl is used to set the values of the CMMA bits on the s390
3458 architecture. It is meant to be used during live migration to restore
3459 the CMMA values, but there are no restrictions on its use.
3460 The ioctl takes parameters via the kvm_s390_cmma_values struct.
3461 Each CMMA value takes up one byte.
3463 struct kvm_s390_cmma_log {
3474 start_gfn indicates the starting guest frame number,
3476 count indicates how many values are to be considered in the buffer,
3478 flags is not used and must be 0.
3480 mask indicates which PGSTE bits are to be considered.
3482 remaining is not used.
3484 values points to the buffer in userspace where to store the values.
3486 This ioctl can fail with -ENOMEM if not enough memory can be allocated to
3487 complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
3488 the count field is too large (e.g. more than KVM_S390_CMMA_SIZE_MAX) or
3489 if the flags field was not 0, with -EFAULT if the userspace address is
3490 invalid, if invalid pages are written to (e.g. after the end of memory)
3491 or if no page table is present for the addresses (e.g. when using
3494 4.109 KVM_PPC_GET_CPU_CHAR
3496 Capability: KVM_CAP_PPC_GET_CPU_CHAR
3497 Architectures: powerpc
3499 Parameters: struct kvm_ppc_cpu_char (out)
3500 Returns: 0 on successful completion
3501 -EFAULT if struct kvm_ppc_cpu_char cannot be written
3503 This ioctl gives userspace information about certain characteristics
3504 of the CPU relating to speculative execution of instructions and
3505 possible information leakage resulting from speculative execution (see
3506 CVE-2017-5715, CVE-2017-5753 and CVE-2017-5754). The information is
3507 returned in struct kvm_ppc_cpu_char, which looks like this:
3509 struct kvm_ppc_cpu_char {
3510 __u64 character; /* characteristics of the CPU */
3511 __u64 behaviour; /* recommended software behaviour */
3512 __u64 character_mask; /* valid bits in character */
3513 __u64 behaviour_mask; /* valid bits in behaviour */
3516 For extensibility, the character_mask and behaviour_mask fields
3517 indicate which bits of character and behaviour have been filled in by
3518 the kernel. If the set of defined bits is extended in future then
3519 userspace will be able to tell whether it is running on a kernel that
3520 knows about the new bits.
3522 The character field describes attributes of the CPU which can help
3523 with preventing inadvertent information disclosure - specifically,
3524 whether there is an instruction to flash-invalidate the L1 data cache
3525 (ori 30,30,0 or mtspr SPRN_TRIG2,rN), whether the L1 data cache is set
3526 to a mode where entries can only be used by the thread that created
3527 them, whether the bcctr[l] instruction prevents speculation, and
3528 whether a speculation barrier instruction (ori 31,31,0) is provided.
3530 The behaviour field describes actions that software should take to
3531 prevent inadvertent information disclosure, and thus describes which
3532 vulnerabilities the hardware is subject to; specifically whether the
3533 L1 data cache should be flushed when returning to user mode from the
3534 kernel, and whether a speculation barrier should be placed between an
3535 array bounds check and the array access.
3537 These fields use the same bit definitions as the new
3538 H_GET_CPU_CHARACTERISTICS hypercall.
3540 4.110 KVM_MEMORY_ENCRYPT_OP
3545 Parameters: an opaque platform specific structure (in/out)
3546 Returns: 0 on success; -1 on error
3548 If the platform supports creating encrypted VMs then this ioctl can be used
3549 for issuing platform-specific memory encryption commands to manage those
3552 Currently, this ioctl is used for issuing Secure Encrypted Virtualization
3553 (SEV) commands on AMD Processors. The SEV commands are defined in
3554 Documentation/virtual/kvm/amd-memory-encryption.rst.
3556 4.111 KVM_MEMORY_ENCRYPT_REG_REGION
3561 Parameters: struct kvm_enc_region (in)
3562 Returns: 0 on success; -1 on error
3564 This ioctl can be used to register a guest memory region which may
3565 contain encrypted data (e.g. guest RAM, SMRAM etc).
3567 It is used in the SEV-enabled guest. When encryption is enabled, a guest
3568 memory region may contain encrypted data. The SEV memory encryption
3569 engine uses a tweak such that two identical plaintext pages, each at
3570 different locations will have differing ciphertexts. So swapping or
3571 moving ciphertext of those pages will not result in plaintext being
3572 swapped. So relocating (or migrating) physical backing pages for the SEV
3573 guest will require some additional steps.
3575 Note: The current SEV key management spec does not provide commands to
3576 swap or migrate (move) ciphertext pages. Hence, for now we pin the guest
3577 memory region registered with the ioctl.
3579 4.112 KVM_MEMORY_ENCRYPT_UNREG_REGION
3584 Parameters: struct kvm_enc_region (in)
3585 Returns: 0 on success; -1 on error
3587 This ioctl can be used to unregister the guest memory region registered
3588 with KVM_MEMORY_ENCRYPT_REG_REGION ioctl above.
3590 4.113 KVM_HYPERV_EVENTFD
3592 Capability: KVM_CAP_HYPERV_EVENTFD
3595 Parameters: struct kvm_hyperv_eventfd (in)
3597 This ioctl (un)registers an eventfd to receive notifications from the guest on
3598 the specified Hyper-V connection id through the SIGNAL_EVENT hypercall, without
3599 causing a user exit. SIGNAL_EVENT hypercall with non-zero event flag number
3600 (bits 24-31) still triggers a KVM_EXIT_HYPERV_HCALL user exit.
3602 struct kvm_hyperv_eventfd {
3609 The conn_id field should fit within 24 bits:
3611 #define KVM_HYPERV_CONN_ID_MASK 0x00ffffff
3613 The acceptable values for the flags field are:
3615 #define KVM_HYPERV_EVENTFD_DEASSIGN (1 << 0)
3617 Returns: 0 on success,
3618 -EINVAL if conn_id or flags is outside the allowed range
3619 -ENOENT on deassign if the conn_id isn't registered
3620 -EEXIST on assign if the conn_id is already registered
3623 5. The kvm_run structure
3624 ------------------------
3626 Application code obtains a pointer to the kvm_run structure by
3627 mmap()ing a vcpu fd. From that point, application code can control
3628 execution by changing fields in kvm_run prior to calling the KVM_RUN
3629 ioctl, and obtain information about the reason KVM_RUN returned by
3630 looking up structure members.
3634 __u8 request_interrupt_window;
3636 Request that KVM_RUN return when it becomes possible to inject external
3637 interrupts into the guest. Useful in conjunction with KVM_INTERRUPT.
3639 __u8 immediate_exit;
3641 This field is polled once when KVM_RUN starts; if non-zero, KVM_RUN
3642 exits immediately, returning -EINTR. In the common scenario where a
3643 signal is used to "kick" a VCPU out of KVM_RUN, this field can be used
3644 to avoid usage of KVM_SET_SIGNAL_MASK, which has worse scalability.
3645 Rather than blocking the signal outside KVM_RUN, userspace can set up
3646 a signal handler that sets run->immediate_exit to a non-zero value.
3648 This field is ignored if KVM_CAP_IMMEDIATE_EXIT is not available.
3655 When KVM_RUN has returned successfully (return value 0), this informs
3656 application code why KVM_RUN has returned. Allowable values for this
3657 field are detailed below.
3659 __u8 ready_for_interrupt_injection;
3661 If request_interrupt_window has been specified, this field indicates
3662 an interrupt can be injected now with KVM_INTERRUPT.
3666 The value of the current interrupt flag. Only valid if in-kernel
3667 local APIC is not used.
3671 More architecture-specific flags detailing state of the VCPU that may
3672 affect the device's behavior. The only currently defined flag is
3673 KVM_RUN_X86_SMM, which is valid on x86 machines and is set if the
3674 VCPU is in system management mode.
3676 /* in (pre_kvm_run), out (post_kvm_run) */
3679 The value of the cr8 register. Only valid if in-kernel local APIC is
3680 not used. Both input and output.
3684 The value of the APIC BASE msr. Only valid if in-kernel local
3685 APIC is not used. Both input and output.
3688 /* KVM_EXIT_UNKNOWN */
3690 __u64 hardware_exit_reason;
3693 If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown
3694 reasons. Further architecture-specific information is available in
3695 hardware_exit_reason.
3697 /* KVM_EXIT_FAIL_ENTRY */
3699 __u64 hardware_entry_failure_reason;
3702 If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due
3703 to unknown reasons. Further architecture-specific information is
3704 available in hardware_entry_failure_reason.
3706 /* KVM_EXIT_EXCEPTION */
3716 #define KVM_EXIT_IO_IN 0
3717 #define KVM_EXIT_IO_OUT 1
3719 __u8 size; /* bytes */
3722 __u64 data_offset; /* relative to kvm_run start */
3725 If exit_reason is KVM_EXIT_IO, then the vcpu has
3726 executed a port I/O instruction which could not be satisfied by kvm.
3727 data_offset describes where the data is located (KVM_EXIT_IO_OUT) or
3728 where kvm expects application code to place the data for the next
3729 KVM_RUN invocation (KVM_EXIT_IO_IN). Data format is a packed array.
3731 /* KVM_EXIT_DEBUG */
3733 struct kvm_debug_exit_arch arch;
3736 If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event
3737 for which architecture specific information is returned.
3747 If exit_reason is KVM_EXIT_MMIO, then the vcpu has
3748 executed a memory-mapped I/O instruction which could not be satisfied
3749 by kvm. The 'data' member contains the written data if 'is_write' is
3750 true, and should be filled by application code otherwise.
3752 The 'data' member contains, in its first 'len' bytes, the value as it would
3753 appear if the VCPU performed a load or store of the appropriate width directly
3756 NOTE: For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR and
3757 KVM_EXIT_EPR the corresponding
3758 operations are complete (and guest state is consistent) only after userspace
3759 has re-entered the kernel with KVM_RUN. The kernel side will first finish
3760 incomplete operations and then check for pending signals. Userspace
3761 can re-enter the guest with an unmasked signal pending to complete
3764 /* KVM_EXIT_HYPERCALL */
3773 Unused. This was once used for 'hypercall to userspace'. To implement
3774 such functionality, use KVM_EXIT_IO (x86) or KVM_EXIT_MMIO (all except s390).
3775 Note KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO.
3777 /* KVM_EXIT_TPR_ACCESS */
3784 To be documented (KVM_TPR_ACCESS_REPORTING).
3786 /* KVM_EXIT_S390_SIEIC */
3789 __u64 mask; /* psw upper half */
3790 __u64 addr; /* psw lower half */
3797 /* KVM_EXIT_S390_RESET */
3798 #define KVM_S390_RESET_POR 1
3799 #define KVM_S390_RESET_CLEAR 2
3800 #define KVM_S390_RESET_SUBSYSTEM 4
3801 #define KVM_S390_RESET_CPU_INIT 8
3802 #define KVM_S390_RESET_IPL 16
3803 __u64 s390_reset_flags;
3807 /* KVM_EXIT_S390_UCONTROL */
3809 __u64 trans_exc_code;
3813 s390 specific. A page fault has occurred for a user controlled virtual
3814 machine (KVM_VM_S390_UNCONTROL) on it's host page table that cannot be
3815 resolved by the kernel.
3816 The program code and the translation exception code that were placed
3817 in the cpu's lowcore are presented here as defined by the z Architecture
3818 Principles of Operation Book in the Chapter for Dynamic Address Translation
3828 Deprecated - was used for 440 KVM.
3835 MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch
3836 hypercalls and exit with this exit struct that contains all the guest gprs.
3838 If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall.
3839 Userspace can now handle the hypercall and when it's done modify the gprs as
3840 necessary. Upon guest entry all guest GPRs will then be replaced by the values
3843 /* KVM_EXIT_PAPR_HCALL */
3850 This is used on 64-bit PowerPC when emulating a pSeries partition,
3851 e.g. with the 'pseries' machine type in qemu. It occurs when the
3852 guest does a hypercall using the 'sc 1' instruction. The 'nr' field
3853 contains the hypercall number (from the guest R3), and 'args' contains
3854 the arguments (from the guest R4 - R12). Userspace should put the
3855 return code in 'ret' and any extra returned values in args[].
3856 The possible hypercalls are defined in the Power Architecture Platform
3857 Requirements (PAPR) document available from www.power.org (free
3858 developer registration required to access it).
3860 /* KVM_EXIT_S390_TSCH */
3862 __u16 subchannel_id;
3863 __u16 subchannel_nr;
3870 s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled
3871 and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O
3872 interrupt for the target subchannel has been dequeued and subchannel_id,
3873 subchannel_nr, io_int_parm and io_int_word contain the parameters for that
3874 interrupt. ipb is needed for instruction parameter decoding.
3881 On FSL BookE PowerPC chips, the interrupt controller has a fast patch
3882 interrupt acknowledge path to the core. When the core successfully
3883 delivers an interrupt, it automatically populates the EPR register with
3884 the interrupt vector number and acknowledges the interrupt inside
3885 the interrupt controller.
3887 In case the interrupt controller lives in user space, we need to do
3888 the interrupt acknowledge cycle through it to fetch the next to be
3889 delivered interrupt vector using this exit.
3891 It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an
3892 external interrupt has just been delivered into the guest. User space
3893 should put the acknowledged interrupt vector into the 'epr' field.
3895 /* KVM_EXIT_SYSTEM_EVENT */
3897 #define KVM_SYSTEM_EVENT_SHUTDOWN 1
3898 #define KVM_SYSTEM_EVENT_RESET 2
3899 #define KVM_SYSTEM_EVENT_CRASH 3
3904 If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered
3905 a system-level event using some architecture specific mechanism (hypercall
3906 or some special instruction). In case of ARM/ARM64, this is triggered using
3907 HVC instruction based PSCI call from the vcpu. The 'type' field describes
3908 the system-level event type. The 'flags' field describes architecture
3909 specific flags for the system-level event.
3911 Valid values for 'type' are:
3912 KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the
3913 VM. Userspace is not obliged to honour this, and if it does honour
3914 this does not need to destroy the VM synchronously (ie it may call
3915 KVM_RUN again before shutdown finally occurs).
3916 KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM.
3917 As with SHUTDOWN, userspace can choose to ignore the request, or
3918 to schedule the reset to occur in the future and may call KVM_RUN again.
3919 KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest
3920 has requested a crash condition maintenance. Userspace can choose
3921 to ignore the request, or to gather VM memory core dump and/or
3922 reset/shutdown of the VM.
3924 /* KVM_EXIT_IOAPIC_EOI */
3929 Indicates that the VCPU's in-kernel local APIC received an EOI for a
3930 level-triggered IOAPIC interrupt. This exit only triggers when the
3931 IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled);
3932 the userspace IOAPIC should process the EOI and retrigger the interrupt if
3933 it is still asserted. Vector is the LAPIC interrupt vector for which the
3936 struct kvm_hyperv_exit {
3937 #define KVM_EXIT_HYPERV_SYNIC 1
3938 #define KVM_EXIT_HYPERV_HCALL 2
3954 /* KVM_EXIT_HYPERV */
3955 struct kvm_hyperv_exit hyperv;
3956 Indicates that the VCPU exits into userspace to process some tasks
3957 related to Hyper-V emulation.
3958 Valid values for 'type' are:
3959 KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about
3960 Hyper-V SynIC state change. Notification is used to remap SynIC
3961 event/message pages and to enable/disable SynIC messages/events processing
3964 /* Fix the size of the union. */
3969 * shared registers between kvm and userspace.
3970 * kvm_valid_regs specifies the register classes set by the host
3971 * kvm_dirty_regs specified the register classes dirtied by userspace
3972 * struct kvm_sync_regs is architecture specific, as well as the
3973 * bits for kvm_valid_regs and kvm_dirty_regs
3975 __u64 kvm_valid_regs;
3976 __u64 kvm_dirty_regs;
3978 struct kvm_sync_regs regs;
3979 char padding[SYNC_REGS_SIZE_BYTES];
3982 If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access
3983 certain guest registers without having to call SET/GET_*REGS. Thus we can
3984 avoid some system call overhead if userspace has to handle the exit.
3985 Userspace can query the validity of the structure by checking
3986 kvm_valid_regs for specific bits. These bits are architecture specific
3987 and usually define the validity of a groups of registers. (e.g. one bit
3988 for general purpose registers)
3990 Please note that the kernel is allowed to use the kvm_run structure as the
3991 primary storage for certain register types. Therefore, the kernel may use the
3992 values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set.
3998 6. Capabilities that can be enabled on vCPUs
3999 --------------------------------------------
4001 There are certain capabilities that change the behavior of the virtual CPU or
4002 the virtual machine when enabled. To enable them, please see section 4.37.
4003 Below you can find a list of capabilities and what their effect on the vCPU or
4004 the virtual machine is when enabling them.
4006 The following information is provided along with the description:
4008 Architectures: which instruction set architectures provide this ioctl.
4009 x86 includes both i386 and x86_64.
4011 Target: whether this is a per-vcpu or per-vm capability.
4013 Parameters: what parameters are accepted by the capability.
4015 Returns: the return value. General error numbers (EBADF, ENOMEM, EINVAL)
4016 are not detailed, but errors with specific meanings are.
4024 Returns: 0 on success; -1 on error
4026 This capability enables interception of OSI hypercalls that otherwise would
4027 be treated as normal system calls to be injected into the guest. OSI hypercalls
4028 were invented by Mac-on-Linux to have a standardized communication mechanism
4029 between the guest and the host.
4031 When this capability is enabled, KVM_EXIT_OSI can occur.
4034 6.2 KVM_CAP_PPC_PAPR
4039 Returns: 0 on success; -1 on error
4041 This capability enables interception of PAPR hypercalls. PAPR hypercalls are
4042 done using the hypercall instruction "sc 1".
4044 It also sets the guest privilege level to "supervisor" mode. Usually the guest
4045 runs in "hypervisor" privilege mode with a few missing features.
4047 In addition to the above, it changes the semantics of SDR1. In this mode, the
4048 HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the
4049 HTAB invisible to the guest.
4051 When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur.
4058 Parameters: args[0] is the address of a struct kvm_config_tlb
4059 Returns: 0 on success; -1 on error
4061 struct kvm_config_tlb {
4068 Configures the virtual CPU's TLB array, establishing a shared memory area
4069 between userspace and KVM. The "params" and "array" fields are userspace
4070 addresses of mmu-type-specific data structures. The "array_len" field is an
4071 safety mechanism, and should be set to the size in bytes of the memory that
4072 userspace has reserved for the array. It must be at least the size dictated
4073 by "mmu_type" and "params".
4075 While KVM_RUN is active, the shared region is under control of KVM. Its
4076 contents are undefined, and any modification by userspace results in
4077 boundedly undefined behavior.
4079 On return from KVM_RUN, the shared region will reflect the current state of
4080 the guest's TLB. If userspace makes any changes, it must call KVM_DIRTY_TLB
4081 to tell KVM which entries have been changed, prior to calling KVM_RUN again
4084 For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV:
4085 - The "params" field is of type "struct kvm_book3e_206_tlb_params".
4086 - The "array" field points to an array of type "struct
4087 kvm_book3e_206_tlb_entry".
4088 - The array consists of all entries in the first TLB, followed by all
4089 entries in the second TLB.
4090 - Within a TLB, entries are ordered first by increasing set number. Within a
4091 set, entries are ordered by way (increasing ESEL).
4092 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1)
4093 where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value.
4094 - The tsize field of mas1 shall be set to 4K on TLB0, even though the
4095 hardware ignores this value for TLB0.
4097 6.4 KVM_CAP_S390_CSS_SUPPORT
4102 Returns: 0 on success; -1 on error
4104 This capability enables support for handling of channel I/O instructions.
4106 TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are
4107 handled in-kernel, while the other I/O instructions are passed to userspace.
4109 When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST
4110 SUBCHANNEL intercepts.
4112 Note that even though this capability is enabled per-vcpu, the complete
4113 virtual machine is affected.
4119 Parameters: args[0] defines whether the proxy facility is active
4120 Returns: 0 on success; -1 on error
4122 This capability enables or disables the delivery of interrupts through the
4123 external proxy facility.
4125 When enabled (args[0] != 0), every time the guest gets an external interrupt
4126 delivered, it automatically exits into user space with a KVM_EXIT_EPR exit
4127 to receive the topmost interrupt vector.
4129 When disabled (args[0] == 0), behavior is as if this facility is unsupported.
4131 When this capability is enabled, KVM_EXIT_EPR can occur.
4133 6.6 KVM_CAP_IRQ_MPIC
4136 Parameters: args[0] is the MPIC device fd
4137 args[1] is the MPIC CPU number for this vcpu
4139 This capability connects the vcpu to an in-kernel MPIC device.
4141 6.7 KVM_CAP_IRQ_XICS
4145 Parameters: args[0] is the XICS device fd
4146 args[1] is the XICS CPU number (server ID) for this vcpu
4148 This capability connects the vcpu to an in-kernel XICS device.
4150 6.8 KVM_CAP_S390_IRQCHIP
4156 This capability enables the in-kernel irqchip for s390. Please refer to
4157 "4.24 KVM_CREATE_IRQCHIP" for details.
4159 6.9 KVM_CAP_MIPS_FPU
4163 Parameters: args[0] is reserved for future use (should be 0).
4165 This capability allows the use of the host Floating Point Unit by the guest. It
4166 allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is
4167 done the KVM_REG_MIPS_FPR_* and KVM_REG_MIPS_FCR_* registers can be accessed
4168 (depending on the current guest FPU register mode), and the Status.FR,
4169 Config5.FRE bits are accessible via the KVM API and also from the guest,
4170 depending on them being supported by the FPU.
4172 6.10 KVM_CAP_MIPS_MSA
4176 Parameters: args[0] is reserved for future use (should be 0).
4178 This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest.
4179 It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest.
4180 Once this is done the KVM_REG_MIPS_VEC_* and KVM_REG_MIPS_MSA_* registers can be
4181 accessed, and the Config5.MSAEn bit is accessible via the KVM API and also from
4184 6.74 KVM_CAP_SYNC_REGS
4185 Architectures: s390, x86
4186 Target: s390: always enabled, x86: vcpu
4188 Returns: x86: KVM_CHECK_EXTENSION returns a bit-array indicating which register
4189 sets are supported (bitfields defined in arch/x86/include/uapi/asm/kvm.h).
4191 As described above in the kvm_sync_regs struct info in section 5 (kvm_run):
4192 KVM_CAP_SYNC_REGS "allow[s] userspace to access certain guest registers
4193 without having to call SET/GET_*REGS". This reduces overhead by eliminating
4194 repeated ioctl calls for setting and/or getting register values. This is
4195 particularly important when userspace is making synchronous guest state
4196 modifications, e.g. when emulating and/or intercepting instructions in
4199 For s390 specifics, please refer to the source code.
4202 - the register sets to be copied out to kvm_run are selectable
4203 by userspace (rather that all sets being copied out for every exit).
4204 - vcpu_events are available in addition to regs and sregs.
4206 For x86, the 'kvm_valid_regs' field of struct kvm_run is overloaded to
4207 function as an input bit-array field set by userspace to indicate the
4208 specific register sets to be copied out on the next exit.
4210 To indicate when userspace has modified values that should be copied into
4211 the vCPU, the all architecture bitarray field, 'kvm_dirty_regs' must be set.
4212 This is done using the same bitflags as for the 'kvm_valid_regs' field.
4213 If the dirty bit is not set, then the register set values will not be copied
4214 into the vCPU even if they've been modified.
4216 Unused bitfields in the bitarrays must be set to zero.
4218 struct kvm_sync_regs {
4219 struct kvm_regs regs;
4220 struct kvm_sregs sregs;
4221 struct kvm_vcpu_events events;
4224 7. Capabilities that can be enabled on VMs
4225 ------------------------------------------
4227 There are certain capabilities that change the behavior of the virtual
4228 machine when enabled. To enable them, please see section 4.37. Below
4229 you can find a list of capabilities and what their effect on the VM
4230 is when enabling them.
4232 The following information is provided along with the description:
4234 Architectures: which instruction set architectures provide this ioctl.
4235 x86 includes both i386 and x86_64.
4237 Parameters: what parameters are accepted by the capability.
4239 Returns: the return value. General error numbers (EBADF, ENOMEM, EINVAL)
4240 are not detailed, but errors with specific meanings are.
4243 7.1 KVM_CAP_PPC_ENABLE_HCALL
4246 Parameters: args[0] is the sPAPR hcall number
4247 args[1] is 0 to disable, 1 to enable in-kernel handling
4249 This capability controls whether individual sPAPR hypercalls (hcalls)
4250 get handled by the kernel or not. Enabling or disabling in-kernel
4251 handling of an hcall is effective across the VM. On creation, an
4252 initial set of hcalls are enabled for in-kernel handling, which
4253 consists of those hcalls for which in-kernel handlers were implemented
4254 before this capability was implemented. If disabled, the kernel will
4255 not to attempt to handle the hcall, but will always exit to userspace
4256 to handle it. Note that it may not make sense to enable some and
4257 disable others of a group of related hcalls, but KVM does not prevent
4258 userspace from doing that.
4260 If the hcall number specified is not one that has an in-kernel
4261 implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL
4264 7.2 KVM_CAP_S390_USER_SIGP
4269 This capability controls which SIGP orders will be handled completely in user
4270 space. With this capability enabled, all fast orders will be handled completely
4276 - CONDITIONAL EMERGENCY SIGNAL
4278 All other orders will be handled completely in user space.
4280 Only privileged operation exceptions will be checked for in the kernel (or even
4281 in the hardware prior to interception). If this capability is not enabled, the
4282 old way of handling SIGP orders is used (partially in kernel and user space).
4284 7.3 KVM_CAP_S390_VECTOR_REGISTERS
4288 Returns: 0 on success, negative value on error
4290 Allows use of the vector registers introduced with z13 processor, and
4291 provides for the synchronization between host and user space. Will
4292 return -EINVAL if the machine does not support vectors.
4294 7.4 KVM_CAP_S390_USER_STSI
4299 This capability allows post-handlers for the STSI instruction. After
4300 initial handling in the kernel, KVM exits to user space with
4301 KVM_EXIT_S390_STSI to allow user space to insert further data.
4303 Before exiting to userspace, kvm handlers should fill in s390_stsi field of
4314 @addr - guest address of STSI SYSIB
4318 @ar - access register number
4320 KVM handlers should exit to userspace with rc = -EREMOTE.
4322 7.5 KVM_CAP_SPLIT_IRQCHIP
4325 Parameters: args[0] - number of routes reserved for userspace IOAPICs
4326 Returns: 0 on success, -1 on error
4328 Create a local apic for each processor in the kernel. This can be used
4329 instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the
4330 IOAPIC and PIC (and also the PIT, even though this has to be enabled
4333 This capability also enables in kernel routing of interrupt requests;
4334 when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are
4335 used in the IRQ routing table. The first args[0] MSI routes are reserved
4336 for the IOAPIC pins. Whenever the LAPIC receives an EOI for these routes,
4337 a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace.
4339 Fails if VCPU has already been created, or if the irqchip is already in the
4340 kernel (i.e. KVM_CREATE_IRQCHIP has already been called).
4347 Allows use of runtime-instrumentation introduced with zEC12 processor.
4348 Will return -EINVAL if the machine does not support runtime-instrumentation.
4349 Will return -EBUSY if a VCPU has already been created.
4351 7.7 KVM_CAP_X2APIC_API
4354 Parameters: args[0] - features that should be enabled
4355 Returns: 0 on success, -EINVAL when args[0] contains invalid features
4357 Valid feature flags in args[0] are
4359 #define KVM_X2APIC_API_USE_32BIT_IDS (1ULL << 0)
4360 #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK (1ULL << 1)
4362 Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of
4363 KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC,
4364 allowing the use of 32-bit APIC IDs. See KVM_CAP_X2APIC_API in their
4365 respective sections.
4367 KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work
4368 in logical mode or with more than 255 VCPUs. Otherwise, KVM treats 0xff
4369 as a broadcast even in x2APIC mode in order to support physical x2APIC
4370 without interrupt remapping. This is undesirable in logical mode,
4371 where 0xff represents CPUs 0-7 in cluster 0.
4373 7.8 KVM_CAP_S390_USER_INSTR0
4378 With this capability enabled, all illegal instructions 0x0000 (2 bytes) will
4379 be intercepted and forwarded to user space. User space can use this
4380 mechanism e.g. to realize 2-byte software breakpoints. The kernel will
4381 not inject an operating exception for these instructions, user space has
4382 to take care of that.
4384 This capability can be enabled dynamically even if VCPUs were already
4385 created and are running.
4391 Returns: 0 on success; -EINVAL if the machine does not support
4392 guarded storage; -EBUSY if a VCPU has already been created.
4394 Allows use of guarded storage for the KVM guest.
4396 7.10 KVM_CAP_S390_AIS
4401 Allow use of adapter-interruption suppression.
4402 Returns: 0 on success; -EBUSY if a VCPU has already been created.
4404 7.11 KVM_CAP_PPC_SMT
4407 Parameters: vsmt_mode, flags
4409 Enabling this capability on a VM provides userspace with a way to set
4410 the desired virtual SMT mode (i.e. the number of virtual CPUs per
4411 virtual core). The virtual SMT mode, vsmt_mode, must be a power of 2
4412 between 1 and 8. On POWER8, vsmt_mode must also be no greater than
4413 the number of threads per subcore for the host. Currently flags must
4414 be 0. A successful call to enable this capability will result in
4415 vsmt_mode being returned when the KVM_CAP_PPC_SMT capability is
4416 subsequently queried for the VM. This capability is only supported by
4417 HV KVM, and can only be set before any VCPUs have been created.
4418 The KVM_CAP_PPC_SMT_POSSIBLE capability indicates which virtual SMT
4419 modes are available.
4421 7.12 KVM_CAP_PPC_FWNMI
4426 With this capability a machine check exception in the guest address
4427 space will cause KVM to exit the guest with NMI exit reason. This
4428 enables QEMU to build error log and branch to guest kernel registered
4429 machine check handling routine. Without this capability KVM will
4430 branch to guests' 0x200 interrupt vector.
4432 7.13 KVM_CAP_X86_DISABLE_EXITS
4435 Parameters: args[0] defines which exits are disabled
4436 Returns: 0 on success, -EINVAL when args[0] contains invalid exits
4438 Valid bits in args[0] are
4440 #define KVM_X86_DISABLE_EXITS_MWAIT (1 << 0)
4441 #define KVM_X86_DISABLE_EXITS_HLT (1 << 1)
4443 Enabling this capability on a VM provides userspace with a way to no
4444 longer intercept some instructions for improved latency in some
4445 workloads, and is suggested when vCPUs are associated to dedicated
4446 physical CPUs. More bits can be added in the future; userspace can
4447 just pass the KVM_CHECK_EXTENSION result to KVM_ENABLE_CAP to disable
4450 Do not enable KVM_FEATURE_PV_UNHALT if you disable HLT exits.
4452 8. Other capabilities.
4453 ----------------------
4455 This section lists capabilities that give information about other
4456 features of the KVM implementation.
4458 8.1 KVM_CAP_PPC_HWRNG
4462 This capability, if KVM_CHECK_EXTENSION indicates that it is
4463 available, means that that the kernel has an implementation of the
4464 H_RANDOM hypercall backed by a hardware random-number generator.
4465 If present, the kernel H_RANDOM handler can be enabled for guest use
4466 with the KVM_CAP_PPC_ENABLE_HCALL capability.
4468 8.2 KVM_CAP_HYPERV_SYNIC
4471 This capability, if KVM_CHECK_EXTENSION indicates that it is
4472 available, means that that the kernel has an implementation of the
4473 Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is
4474 used to support Windows Hyper-V based guest paravirt drivers(VMBus).
4476 In order to use SynIC, it has to be activated by setting this
4477 capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this
4478 will disable the use of APIC hardware virtualization even if supported
4479 by the CPU, as it's incompatible with SynIC auto-EOI behavior.
4481 8.3 KVM_CAP_PPC_RADIX_MMU
4485 This capability, if KVM_CHECK_EXTENSION indicates that it is
4486 available, means that that the kernel can support guests using the
4487 radix MMU defined in Power ISA V3.00 (as implemented in the POWER9
4490 8.4 KVM_CAP_PPC_HASH_MMU_V3
4494 This capability, if KVM_CHECK_EXTENSION indicates that it is
4495 available, means that that the kernel can support guests using the
4496 hashed page table MMU defined in Power ISA V3.00 (as implemented in
4497 the POWER9 processor), including in-memory segment tables.
4503 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
4504 it is available, means that full hardware assisted virtualization capabilities
4505 of the hardware are available for use through KVM. An appropriate
4506 KVM_VM_MIPS_* type must be passed to KVM_CREATE_VM to create a VM which
4509 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
4510 available, it means that the VM is using full hardware assisted virtualization
4511 capabilities of the hardware. This is useful to check after creating a VM with
4512 KVM_VM_MIPS_DEFAULT.
4514 The value returned by KVM_CHECK_EXTENSION should be compared against known
4515 values (see below). All other values are reserved. This is to allow for the
4516 possibility of other hardware assisted virtualization implementations which
4517 may be incompatible with the MIPS VZ ASE.
4519 0: The trap & emulate implementation is in use to run guest code in user
4520 mode. Guest virtual memory segments are rearranged to fit the guest in the
4521 user mode address space.
4523 1: The MIPS VZ ASE is in use, providing full hardware assisted
4524 virtualization, including standard guest virtual memory segments.
4530 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
4531 it is available, means that the trap & emulate implementation is available to
4532 run guest code in user mode, even if KVM_CAP_MIPS_VZ indicates that hardware
4533 assisted virtualisation is also available. KVM_VM_MIPS_TE (0) must be passed
4534 to KVM_CREATE_VM to create a VM which utilises it.
4536 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
4537 available, it means that the VM is using trap & emulate.
4539 8.7 KVM_CAP_MIPS_64BIT
4543 This capability indicates the supported architecture type of the guest, i.e. the
4544 supported register and address width.
4546 The values returned when this capability is checked by KVM_CHECK_EXTENSION on a
4547 kvm VM handle correspond roughly to the CP0_Config.AT register field, and should
4548 be checked specifically against known values (see below). All other values are
4551 0: MIPS32 or microMIPS32.
4552 Both registers and addresses are 32-bits wide.
4553 It will only be possible to run 32-bit guest code.
4555 1: MIPS64 or microMIPS64 with access only to 32-bit compatibility segments.
4556 Registers are 64-bits wide, but addresses are 32-bits wide.
4557 64-bit guest code may run but cannot access MIPS64 memory segments.
4558 It will also be possible to run 32-bit guest code.
4560 2: MIPS64 or microMIPS64 with access to all address segments.
4561 Both registers and addresses are 64-bits wide.
4562 It will be possible to run 64-bit or 32-bit guest code.
4564 8.9 KVM_CAP_ARM_USER_IRQ
4566 Architectures: arm, arm64
4567 This capability, if KVM_CHECK_EXTENSION indicates that it is available, means
4568 that if userspace creates a VM without an in-kernel interrupt controller, it
4569 will be notified of changes to the output level of in-kernel emulated devices,
4570 which can generate virtual interrupts, presented to the VM.
4571 For such VMs, on every return to userspace, the kernel
4572 updates the vcpu's run->s.regs.device_irq_level field to represent the actual
4573 output level of the device.
4575 Whenever kvm detects a change in the device output level, kvm guarantees at
4576 least one return to userspace before running the VM. This exit could either
4577 be a KVM_EXIT_INTR or any other exit event, like KVM_EXIT_MMIO. This way,
4578 userspace can always sample the device output level and re-compute the state of
4579 the userspace interrupt controller. Userspace should always check the state
4580 of run->s.regs.device_irq_level on every kvm exit.
4581 The value in run->s.regs.device_irq_level can represent both level and edge
4582 triggered interrupt signals, depending on the device. Edge triggered interrupt
4583 signals will exit to userspace with the bit in run->s.regs.device_irq_level
4584 set exactly once per edge signal.
4586 The field run->s.regs.device_irq_level is available independent of
4587 run->kvm_valid_regs or run->kvm_dirty_regs bits.
4589 If KVM_CAP_ARM_USER_IRQ is supported, the KVM_CHECK_EXTENSION ioctl returns a
4590 number larger than 0 indicating the version of this capability is implemented
4591 and thereby which bits in in run->s.regs.device_irq_level can signal values.
4593 Currently the following bits are defined for the device_irq_level bitmap:
4595 KVM_CAP_ARM_USER_IRQ >= 1:
4597 KVM_ARM_DEV_EL1_VTIMER - EL1 virtual timer
4598 KVM_ARM_DEV_EL1_PTIMER - EL1 physical timer
4599 KVM_ARM_DEV_PMU - ARM PMU overflow interrupt signal
4601 Future versions of kvm may implement additional events. These will get
4602 indicated by returning a higher number from KVM_CHECK_EXTENSION and will be
4605 8.10 KVM_CAP_PPC_SMT_POSSIBLE
4609 Querying this capability returns a bitmap indicating the possible
4610 virtual SMT modes that can be set using KVM_CAP_PPC_SMT. If bit N
4611 (counting from the right) is set, then a virtual SMT mode of 2^N is
4614 8.11 KVM_CAP_HYPERV_SYNIC2
4618 This capability enables a newer version of Hyper-V Synthetic interrupt
4619 controller (SynIC). The only difference with KVM_CAP_HYPERV_SYNIC is that KVM
4620 doesn't clear SynIC message and event flags pages when they are enabled by
4621 writing to the respective MSRs.
4623 8.12 KVM_CAP_HYPERV_VP_INDEX
4627 This capability indicates that userspace can load HV_X64_MSR_VP_INDEX msr. Its
4628 value is used to denote the target vcpu for a SynIC interrupt. For
4629 compatibilty, KVM initializes this msr to KVM's internal vcpu index. When this
4630 capability is absent, userspace can still query this msr's value.
4632 8.13 KVM_CAP_S390_AIS_MIGRATION
4637 This capability indicates if the flic device will be able to get/set the
4638 AIS states for migration via the KVM_DEV_FLIC_AISM_ALL attribute and allows
4639 to discover this without having to create a flic device.
4641 8.14 KVM_CAP_S390_PSW
4645 This capability indicates that the PSW is exposed via the kvm_run structure.
4647 8.15 KVM_CAP_S390_GMAP
4651 This capability indicates that the user space memory used as guest mapping can
4652 be anywhere in the user memory address space, as long as the memory slots are
4653 aligned and sized to a segment (1MB) boundary.
4655 8.16 KVM_CAP_S390_COW
4659 This capability indicates that the user space memory used as guest mapping can
4660 use copy-on-write semantics as well as dirty pages tracking via read-only page
4663 8.17 KVM_CAP_S390_BPB
4667 This capability indicates that kvm will implement the interfaces to handle
4668 reset, migration and nested KVM for branch prediction blocking. The stfle
4669 facility 82 should not be provided to the guest without this capability.
4671 8.18 KVM_CAP_HYPERV_TLBFLUSH
4675 This capability indicates that KVM supports paravirtualized Hyper-V TLB Flush
4677 HvFlushVirtualAddressSpace, HvFlushVirtualAddressSpaceEx,
4678 HvFlushVirtualAddressList, HvFlushVirtualAddressListEx.
4680 8.19 KVM_CAP_ARM_SET_SERROR_ESR
4682 Architectures: arm, arm64
4684 This capability indicates that userspace can specify (via the
4685 KVM_SET_VCPU_EVENTS ioctl) the syndrome value reported to the guest when it
4686 takes a virtual SError interrupt exception.
4687 If KVM advertises this capability, userspace can only specify the ISS field for
4688 the ESR syndrome. Other parts of the ESR, such as the EC are generated by the
4689 CPU when the exception is taken. If this virtual SError is taken to EL1 using
4690 AArch64, this value will be reported in the ISS field of ESR_ELx.
4692 See KVM_CAP_VCPU_EVENTS for more details.