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 On arm64, the physical address size for a VM (IPA Size limit) is limited
127 to 40bits by default. The limit can be configured if the host supports the
128 extension KVM_CAP_ARM_VM_IPA_SIZE. When supported, use
129 KVM_VM_TYPE_ARM_IPA_SIZE(IPA_Bits) to set the size in the machine type
130 identifier, where IPA_Bits is the maximum width of any physical
131 address used by the VM. The IPA_Bits is encoded in bits[7-0] of the
132 machine type identifier.
134 e.g, to configure a guest to use 48bit physical address size :
136 vm_fd = ioctl(dev_fd, KVM_CREATE_VM, KVM_VM_TYPE_ARM_IPA_SIZE(48));
138 The requested size (IPA_Bits) must be :
139 0 - Implies default size, 40bits (for backward compatibility)
143 N - Implies N bits, where N is a positive integer such that,
144 32 <= N <= Host_IPA_Limit
146 Host_IPA_Limit is the maximum possible value for IPA_Bits on the host and
147 is dependent on the CPU capability and the kernel configuration. The limit can
148 be retrieved using KVM_CAP_ARM_VM_IPA_SIZE of the KVM_CHECK_EXTENSION
151 Please note that configuring the IPA size does not affect the capability
152 exposed by the guest CPUs in ID_AA64MMFR0_EL1[PARange]. It only affects
153 size of the address translated by the stage2 level (guest physical to
154 host physical address translations).
157 4.3 KVM_GET_MSR_INDEX_LIST, KVM_GET_MSR_FEATURE_INDEX_LIST
159 Capability: basic, KVM_CAP_GET_MSR_FEATURES for KVM_GET_MSR_FEATURE_INDEX_LIST
162 Parameters: struct kvm_msr_list (in/out)
163 Returns: 0 on success; -1 on error
165 EFAULT: the msr index list cannot be read from or written to
166 E2BIG: the msr index list is to be to fit in the array specified by
169 struct kvm_msr_list {
170 __u32 nmsrs; /* number of msrs in entries */
174 The user fills in the size of the indices array in nmsrs, and in return
175 kvm adjusts nmsrs to reflect the actual number of msrs and fills in the
176 indices array with their numbers.
178 KVM_GET_MSR_INDEX_LIST returns the guest msrs that are supported. The list
179 varies by kvm version and host processor, but does not change otherwise.
181 Note: if kvm indicates supports MCE (KVM_CAP_MCE), then the MCE bank MSRs are
182 not returned in the MSR list, as different vcpus can have a different number
183 of banks, as set via the KVM_X86_SETUP_MCE ioctl.
185 KVM_GET_MSR_FEATURE_INDEX_LIST returns the list of MSRs that can be passed
186 to the KVM_GET_MSRS system ioctl. This lets userspace probe host capabilities
187 and processor features that are exposed via MSRs (e.g., VMX capabilities).
188 This list also varies by kvm version and host processor, but does not change
192 4.4 KVM_CHECK_EXTENSION
194 Capability: basic, KVM_CAP_CHECK_EXTENSION_VM for vm ioctl
196 Type: system ioctl, vm ioctl
197 Parameters: extension identifier (KVM_CAP_*)
198 Returns: 0 if unsupported; 1 (or some other positive integer) if supported
200 The API allows the application to query about extensions to the core
201 kvm API. Userspace passes an extension identifier (an integer) and
202 receives an integer that describes the extension availability.
203 Generally 0 means no and 1 means yes, but some extensions may report
204 additional information in the integer return value.
206 Based on their initialization different VMs may have different capabilities.
207 It is thus encouraged to use the vm ioctl to query for capabilities (available
208 with KVM_CAP_CHECK_EXTENSION_VM on the vm fd)
210 4.5 KVM_GET_VCPU_MMAP_SIZE
216 Returns: size of vcpu mmap area, in bytes
218 The KVM_RUN ioctl (cf.) communicates with userspace via a shared
219 memory region. This ioctl returns the size of that region. See the
220 KVM_RUN documentation for details.
223 4.6 KVM_SET_MEMORY_REGION
228 Parameters: struct kvm_memory_region (in)
229 Returns: 0 on success, -1 on error
231 This ioctl is obsolete and has been removed.
239 Parameters: vcpu id (apic id on x86)
240 Returns: vcpu fd on success, -1 on error
242 This API adds a vcpu to a virtual machine. No more than max_vcpus may be added.
243 The vcpu id is an integer in the range [0, max_vcpu_id).
245 The recommended max_vcpus value can be retrieved using the KVM_CAP_NR_VCPUS of
246 the KVM_CHECK_EXTENSION ioctl() at run-time.
247 The maximum possible value for max_vcpus can be retrieved using the
248 KVM_CAP_MAX_VCPUS of the KVM_CHECK_EXTENSION ioctl() at run-time.
250 If the KVM_CAP_NR_VCPUS does not exist, you should assume that max_vcpus is 4
252 If the KVM_CAP_MAX_VCPUS does not exist, you should assume that max_vcpus is
253 same as the value returned from KVM_CAP_NR_VCPUS.
255 The maximum possible value for max_vcpu_id can be retrieved using the
256 KVM_CAP_MAX_VCPU_ID of the KVM_CHECK_EXTENSION ioctl() at run-time.
258 If the KVM_CAP_MAX_VCPU_ID does not exist, you should assume that max_vcpu_id
259 is the same as the value returned from KVM_CAP_MAX_VCPUS.
261 On powerpc using book3s_hv mode, the vcpus are mapped onto virtual
262 threads in one or more virtual CPU cores. (This is because the
263 hardware requires all the hardware threads in a CPU core to be in the
264 same partition.) The KVM_CAP_PPC_SMT capability indicates the number
265 of vcpus per virtual core (vcore). The vcore id is obtained by
266 dividing the vcpu id by the number of vcpus per vcore. The vcpus in a
267 given vcore will always be in the same physical core as each other
268 (though that might be a different physical core from time to time).
269 Userspace can control the threading (SMT) mode of the guest by its
270 allocation of vcpu ids. For example, if userspace wants
271 single-threaded guest vcpus, it should make all vcpu ids be a multiple
272 of the number of vcpus per vcore.
274 For virtual cpus that have been created with S390 user controlled virtual
275 machines, the resulting vcpu fd can be memory mapped at page offset
276 KVM_S390_SIE_PAGE_OFFSET in order to obtain a memory map of the virtual
277 cpu's hardware control block.
280 4.8 KVM_GET_DIRTY_LOG (vm ioctl)
285 Parameters: struct kvm_dirty_log (in/out)
286 Returns: 0 on success, -1 on error
288 /* for KVM_GET_DIRTY_LOG */
289 struct kvm_dirty_log {
293 void __user *dirty_bitmap; /* one bit per page */
298 Given a memory slot, return a bitmap containing any pages dirtied
299 since the last call to this ioctl. Bit 0 is the first page in the
300 memory slot. Ensure the entire structure is cleared to avoid padding
303 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 specifies
304 the address space for which you want to return the dirty bitmap.
305 They must be less than the value that KVM_CHECK_EXTENSION returns for
306 the KVM_CAP_MULTI_ADDRESS_SPACE capability.
309 4.9 KVM_SET_MEMORY_ALIAS
314 Parameters: struct kvm_memory_alias (in)
315 Returns: 0 (success), -1 (error)
317 This ioctl is obsolete and has been removed.
326 Returns: 0 on success, -1 on error
328 EINTR: an unmasked signal is pending
330 This ioctl is used to run a guest virtual cpu. While there are no
331 explicit parameters, there is an implicit parameter block that can be
332 obtained by mmap()ing the vcpu fd at offset 0, with the size given by
333 KVM_GET_VCPU_MMAP_SIZE. The parameter block is formatted as a 'struct
334 kvm_run' (see below).
340 Architectures: all except ARM, arm64
342 Parameters: struct kvm_regs (out)
343 Returns: 0 on success, -1 on error
345 Reads the general purpose registers from the vcpu.
349 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
350 __u64 rax, rbx, rcx, rdx;
351 __u64 rsi, rdi, rsp, rbp;
352 __u64 r8, r9, r10, r11;
353 __u64 r12, r13, r14, r15;
359 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
370 Architectures: all except ARM, arm64
372 Parameters: struct kvm_regs (in)
373 Returns: 0 on success, -1 on error
375 Writes the general purpose registers into the vcpu.
377 See KVM_GET_REGS for the data structure.
383 Architectures: x86, ppc
385 Parameters: struct kvm_sregs (out)
386 Returns: 0 on success, -1 on error
388 Reads special registers from the vcpu.
392 struct kvm_segment cs, ds, es, fs, gs, ss;
393 struct kvm_segment tr, ldt;
394 struct kvm_dtable gdt, idt;
395 __u64 cr0, cr2, cr3, cr4, cr8;
398 __u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64];
401 /* ppc -- see arch/powerpc/include/uapi/asm/kvm.h */
403 interrupt_bitmap is a bitmap of pending external interrupts. At most
404 one bit may be set. This interrupt has been acknowledged by the APIC
405 but not yet injected into the cpu core.
411 Architectures: x86, ppc
413 Parameters: struct kvm_sregs (in)
414 Returns: 0 on success, -1 on error
416 Writes special registers into the vcpu. See KVM_GET_SREGS for the
425 Parameters: struct kvm_translation (in/out)
426 Returns: 0 on success, -1 on error
428 Translates a virtual address according to the vcpu's current address
431 struct kvm_translation {
433 __u64 linear_address;
436 __u64 physical_address;
447 Architectures: x86, ppc, mips
449 Parameters: struct kvm_interrupt (in)
450 Returns: 0 on success, negative on failure.
452 Queues a hardware interrupt vector to be injected.
454 /* for KVM_INTERRUPT */
455 struct kvm_interrupt {
462 Returns: 0 on success,
463 -EEXIST if an interrupt is already enqueued
464 -EINVAL the the irq number is invalid
465 -ENXIO if the PIC is in the kernel
466 -EFAULT if the pointer is invalid
468 Note 'irq' is an interrupt vector, not an interrupt pin or line. This
469 ioctl is useful if the in-kernel PIC is not used.
473 Queues an external interrupt to be injected. This ioctl is overleaded
474 with 3 different irq values:
478 This injects an edge type external interrupt into the guest once it's ready
479 to receive interrupts. When injected, the interrupt is done.
481 b) KVM_INTERRUPT_UNSET
483 This unsets any pending interrupt.
485 Only available with KVM_CAP_PPC_UNSET_IRQ.
487 c) KVM_INTERRUPT_SET_LEVEL
489 This injects a level type external interrupt into the guest context. The
490 interrupt stays pending until a specific ioctl with KVM_INTERRUPT_UNSET
493 Only available with KVM_CAP_PPC_IRQ_LEVEL.
495 Note that any value for 'irq' other than the ones stated above is invalid
496 and incurs unexpected behavior.
500 Queues an external interrupt to be injected into the virtual CPU. A negative
501 interrupt number dequeues the interrupt.
512 Support for this has been removed. Use KVM_SET_GUEST_DEBUG instead.
517 Capability: basic (vcpu), KVM_CAP_GET_MSR_FEATURES (system)
519 Type: system ioctl, vcpu ioctl
520 Parameters: struct kvm_msrs (in/out)
521 Returns: number of msrs successfully returned;
524 When used as a system ioctl:
525 Reads the values of MSR-based features that are available for the VM. This
526 is similar to KVM_GET_SUPPORTED_CPUID, but it returns MSR indices and values.
527 The list of msr-based features can be obtained using KVM_GET_MSR_FEATURE_INDEX_LIST
530 When used as a vcpu ioctl:
531 Reads model-specific registers from the vcpu. Supported msr indices can
532 be obtained using KVM_GET_MSR_INDEX_LIST in a system ioctl.
535 __u32 nmsrs; /* number of msrs in entries */
538 struct kvm_msr_entry entries[0];
541 struct kvm_msr_entry {
547 Application code should set the 'nmsrs' member (which indicates the
548 size of the entries array) and the 'index' member of each array entry.
549 kvm will fill in the 'data' member.
557 Parameters: struct kvm_msrs (in)
558 Returns: 0 on success, -1 on error
560 Writes model-specific registers to the vcpu. See KVM_GET_MSRS for the
563 Application code should set the 'nmsrs' member (which indicates the
564 size of the entries array), and the 'index' and 'data' members of each
573 Parameters: struct kvm_cpuid (in)
574 Returns: 0 on success, -1 on error
576 Defines the vcpu responses to the cpuid instruction. Applications
577 should use the KVM_SET_CPUID2 ioctl if available.
580 struct kvm_cpuid_entry {
589 /* for KVM_SET_CPUID */
593 struct kvm_cpuid_entry entries[0];
597 4.21 KVM_SET_SIGNAL_MASK
602 Parameters: struct kvm_signal_mask (in)
603 Returns: 0 on success, -1 on error
605 Defines which signals are blocked during execution of KVM_RUN. This
606 signal mask temporarily overrides the threads signal mask. Any
607 unblocked signal received (except SIGKILL and SIGSTOP, which retain
608 their traditional behaviour) will cause KVM_RUN to return with -EINTR.
610 Note the signal will only be delivered if not blocked by the original
613 /* for KVM_SET_SIGNAL_MASK */
614 struct kvm_signal_mask {
625 Parameters: struct kvm_fpu (out)
626 Returns: 0 on success, -1 on error
628 Reads the floating point state from the vcpu.
630 /* for KVM_GET_FPU and KVM_SET_FPU */
635 __u8 ftwx; /* in fxsave format */
651 Parameters: struct kvm_fpu (in)
652 Returns: 0 on success, -1 on error
654 Writes the floating point state to the vcpu.
656 /* for KVM_GET_FPU and KVM_SET_FPU */
661 __u8 ftwx; /* in fxsave format */
672 4.24 KVM_CREATE_IRQCHIP
674 Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQCHIP (s390)
675 Architectures: x86, ARM, arm64, s390
678 Returns: 0 on success, -1 on error
680 Creates an interrupt controller model in the kernel.
681 On x86, creates a virtual ioapic, a virtual PIC (two PICs, nested), and sets up
682 future vcpus to have a local APIC. IRQ routing for GSIs 0-15 is set to both
683 PIC and IOAPIC; GSI 16-23 only go to the IOAPIC.
684 On ARM/arm64, a GICv2 is created. Any other GIC versions require the usage of
685 KVM_CREATE_DEVICE, which also supports creating a GICv2. Using
686 KVM_CREATE_DEVICE is preferred over KVM_CREATE_IRQCHIP for GICv2.
687 On s390, a dummy irq routing table is created.
689 Note that on s390 the KVM_CAP_S390_IRQCHIP vm capability needs to be enabled
690 before KVM_CREATE_IRQCHIP can be used.
695 Capability: KVM_CAP_IRQCHIP
696 Architectures: x86, arm, arm64
698 Parameters: struct kvm_irq_level
699 Returns: 0 on success, -1 on error
701 Sets the level of a GSI input to the interrupt controller model in the kernel.
702 On some architectures it is required that an interrupt controller model has
703 been previously created with KVM_CREATE_IRQCHIP. Note that edge-triggered
704 interrupts require the level to be set to 1 and then back to 0.
706 On real hardware, interrupt pins can be active-low or active-high. This
707 does not matter for the level field of struct kvm_irq_level: 1 always
708 means active (asserted), 0 means inactive (deasserted).
710 x86 allows the operating system to program the interrupt polarity
711 (active-low/active-high) for level-triggered interrupts, and KVM used
712 to consider the polarity. However, due to bitrot in the handling of
713 active-low interrupts, the above convention is now valid on x86 too.
714 This is signaled by KVM_CAP_X86_IOAPIC_POLARITY_IGNORED. Userspace
715 should not present interrupts to the guest as active-low unless this
716 capability is present (or unless it is not using the in-kernel irqchip,
720 ARM/arm64 can signal an interrupt either at the CPU level, or at the
721 in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to
722 use PPIs designated for specific cpus. The irq field is interpreted
725 Â bits: | 31 ... 24 | 23 ... 16 | 15 ... 0 |
726 field: | irq_type | vcpu_index | irq_id |
728 The irq_type field has the following values:
729 - irq_type[0]: out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ
730 - irq_type[1]: in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.)
731 (the vcpu_index field is ignored)
732 - irq_type[2]: in-kernel GIC: PPI, irq_id between 16 and 31 (incl.)
734 (The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs)
736 In both cases, level is used to assert/deassert the line.
738 struct kvm_irq_level {
741 __s32 status; /* not used for KVM_IRQ_LEVEL */
743 __u32 level; /* 0 or 1 */
749 Capability: KVM_CAP_IRQCHIP
752 Parameters: struct kvm_irqchip (in/out)
753 Returns: 0 on success, -1 on error
755 Reads the state of a kernel interrupt controller created with
756 KVM_CREATE_IRQCHIP into a buffer provided by the caller.
759 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
762 char dummy[512]; /* reserving space */
763 struct kvm_pic_state pic;
764 struct kvm_ioapic_state ioapic;
771 Capability: KVM_CAP_IRQCHIP
774 Parameters: struct kvm_irqchip (in)
775 Returns: 0 on success, -1 on error
777 Sets the state of a kernel interrupt controller created with
778 KVM_CREATE_IRQCHIP from a buffer provided by the caller.
781 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
784 char dummy[512]; /* reserving space */
785 struct kvm_pic_state pic;
786 struct kvm_ioapic_state ioapic;
791 4.28 KVM_XEN_HVM_CONFIG
793 Capability: KVM_CAP_XEN_HVM
796 Parameters: struct kvm_xen_hvm_config (in)
797 Returns: 0 on success, -1 on error
799 Sets the MSR that the Xen HVM guest uses to initialize its hypercall
800 page, and provides the starting address and size of the hypercall
801 blobs in userspace. When the guest writes the MSR, kvm copies one
802 page of a blob (32- or 64-bit, depending on the vcpu mode) to guest
805 struct kvm_xen_hvm_config {
818 Capability: KVM_CAP_ADJUST_CLOCK
821 Parameters: struct kvm_clock_data (out)
822 Returns: 0 on success, -1 on error
824 Gets the current timestamp of kvmclock as seen by the current guest. In
825 conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios
828 When KVM_CAP_ADJUST_CLOCK is passed to KVM_CHECK_EXTENSION, it returns the
829 set of bits that KVM can return in struct kvm_clock_data's flag member.
831 The only flag defined now is KVM_CLOCK_TSC_STABLE. If set, the returned
832 value is the exact kvmclock value seen by all VCPUs at the instant
833 when KVM_GET_CLOCK was called. If clear, the returned value is simply
834 CLOCK_MONOTONIC plus a constant offset; the offset can be modified
835 with KVM_SET_CLOCK. KVM will try to make all VCPUs follow this clock,
836 but the exact value read by each VCPU could differ, because the host
839 struct kvm_clock_data {
840 __u64 clock; /* kvmclock current value */
848 Capability: KVM_CAP_ADJUST_CLOCK
851 Parameters: struct kvm_clock_data (in)
852 Returns: 0 on success, -1 on error
854 Sets the current timestamp of kvmclock to the value specified in its parameter.
855 In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios
858 struct kvm_clock_data {
859 __u64 clock; /* kvmclock current value */
865 4.31 KVM_GET_VCPU_EVENTS
867 Capability: KVM_CAP_VCPU_EVENTS
868 Extended by: KVM_CAP_INTR_SHADOW
869 Architectures: x86, arm, arm64
871 Parameters: struct kvm_vcpu_event (out)
872 Returns: 0 on success, -1 on error
876 Gets currently pending exceptions, interrupts, and NMIs as well as related
879 struct kvm_vcpu_events {
908 __u8 exception_has_payload;
909 __u64 exception_payload;
912 The following bits are defined in the flags field:
914 - KVM_VCPUEVENT_VALID_SHADOW may be set to signal that
915 interrupt.shadow contains a valid state.
917 - KVM_VCPUEVENT_VALID_SMM may be set to signal that smi contains a
920 - KVM_VCPUEVENT_VALID_PAYLOAD may be set to signal that the
921 exception_has_payload, exception_payload, and exception.pending
922 fields contain a valid state. This bit will be set whenever
923 KVM_CAP_EXCEPTION_PAYLOAD is enabled.
927 If the guest accesses a device that is being emulated by the host kernel in
928 such a way that a real device would generate a physical SError, KVM may make
929 a virtual SError pending for that VCPU. This system error interrupt remains
930 pending until the guest takes the exception by unmasking PSTATE.A.
932 Running the VCPU may cause it to take a pending SError, or make an access that
933 causes an SError to become pending. The event's description is only valid while
934 the VPCU is not running.
936 This API provides a way to read and write the pending 'event' state that is not
937 visible to the guest. To save, restore or migrate a VCPU the struct representing
938 the state can be read then written using this GET/SET API, along with the other
939 guest-visible registers. It is not possible to 'cancel' an SError that has been
942 A device being emulated in user-space may also wish to generate an SError. To do
943 this the events structure can be populated by user-space. The current state
944 should be read first, to ensure no existing SError is pending. If an existing
945 SError is pending, the architecture's 'Multiple SError interrupts' rules should
946 be followed. (2.5.3 of DDI0587.a "ARM Reliability, Availability, and
947 Serviceability (RAS) Specification").
949 SError exceptions always have an ESR value. Some CPUs have the ability to
950 specify what the virtual SError's ESR value should be. These systems will
951 advertise KVM_CAP_ARM_INJECT_SERROR_ESR. In this case exception.has_esr will
952 always have a non-zero value when read, and the agent making an SError pending
953 should specify the ISS field in the lower 24 bits of exception.serror_esr. If
954 the system supports KVM_CAP_ARM_INJECT_SERROR_ESR, but user-space sets the events
955 with exception.has_esr as zero, KVM will choose an ESR.
957 Specifying exception.has_esr on a system that does not support it will return
958 -EINVAL. Setting anything other than the lower 24bits of exception.serror_esr
961 struct kvm_vcpu_events {
965 /* Align it to 8 bytes */
972 4.32 KVM_SET_VCPU_EVENTS
974 Capability: KVM_CAP_VCPU_EVENTS
975 Extended by: KVM_CAP_INTR_SHADOW
976 Architectures: x86, arm, arm64
978 Parameters: struct kvm_vcpu_event (in)
979 Returns: 0 on success, -1 on error
983 Set pending exceptions, interrupts, and NMIs as well as related states of the
986 See KVM_GET_VCPU_EVENTS for the data structure.
988 Fields that may be modified asynchronously by running VCPUs can be excluded
989 from the update. These fields are nmi.pending, sipi_vector, smi.smm,
990 smi.pending. Keep the corresponding bits in the flags field cleared to
991 suppress overwriting the current in-kernel state. The bits are:
993 KVM_VCPUEVENT_VALID_NMI_PENDING - transfer nmi.pending to the kernel
994 KVM_VCPUEVENT_VALID_SIPI_VECTOR - transfer sipi_vector
995 KVM_VCPUEVENT_VALID_SMM - transfer the smi sub-struct.
997 If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in
998 the flags field to signal that interrupt.shadow contains a valid state and
999 shall be written into the VCPU.
1001 KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available.
1003 If KVM_CAP_EXCEPTION_PAYLOAD is enabled, KVM_VCPUEVENT_VALID_PAYLOAD
1004 can be set in the flags field to signal that the
1005 exception_has_payload, exception_payload, and exception.pending fields
1006 contain a valid state and shall be written into the VCPU.
1010 Set the pending SError exception state for this VCPU. It is not possible to
1011 'cancel' an Serror that has been made pending.
1013 See KVM_GET_VCPU_EVENTS for the data structure.
1016 4.33 KVM_GET_DEBUGREGS
1018 Capability: KVM_CAP_DEBUGREGS
1021 Parameters: struct kvm_debugregs (out)
1022 Returns: 0 on success, -1 on error
1024 Reads debug registers from the vcpu.
1026 struct kvm_debugregs {
1035 4.34 KVM_SET_DEBUGREGS
1037 Capability: KVM_CAP_DEBUGREGS
1040 Parameters: struct kvm_debugregs (in)
1041 Returns: 0 on success, -1 on error
1043 Writes debug registers into the vcpu.
1045 See KVM_GET_DEBUGREGS for the data structure. The flags field is unused
1046 yet and must be cleared on entry.
1049 4.35 KVM_SET_USER_MEMORY_REGION
1051 Capability: KVM_CAP_USER_MEM
1054 Parameters: struct kvm_userspace_memory_region (in)
1055 Returns: 0 on success, -1 on error
1057 struct kvm_userspace_memory_region {
1060 __u64 guest_phys_addr;
1061 __u64 memory_size; /* bytes */
1062 __u64 userspace_addr; /* start of the userspace allocated memory */
1065 /* for kvm_memory_region::flags */
1066 #define KVM_MEM_LOG_DIRTY_PAGES (1UL << 0)
1067 #define KVM_MEM_READONLY (1UL << 1)
1069 This ioctl allows the user to create or modify a guest physical memory
1070 slot. When changing an existing slot, it may be moved in the guest
1071 physical memory space, or its flags may be modified. It may not be
1072 resized. Slots may not overlap in guest physical address space.
1073 Bits 0-15 of "slot" specifies the slot id and this value should be
1074 less than the maximum number of user memory slots supported per VM.
1075 The maximum allowed slots can be queried using KVM_CAP_NR_MEMSLOTS,
1076 if this capability is supported by the architecture.
1078 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot"
1079 specifies the address space which is being modified. They must be
1080 less than the value that KVM_CHECK_EXTENSION returns for the
1081 KVM_CAP_MULTI_ADDRESS_SPACE capability. Slots in separate address spaces
1082 are unrelated; the restriction on overlapping slots only applies within
1085 Memory for the region is taken starting at the address denoted by the
1086 field userspace_addr, which must point at user addressable memory for
1087 the entire memory slot size. Any object may back this memory, including
1088 anonymous memory, ordinary files, and hugetlbfs.
1090 It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr
1091 be identical. This allows large pages in the guest to be backed by large
1094 The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and
1095 KVM_MEM_READONLY. The former can be set to instruct KVM to keep track of
1096 writes to memory within the slot. See KVM_GET_DIRTY_LOG ioctl to know how to
1097 use it. The latter can be set, if KVM_CAP_READONLY_MEM capability allows it,
1098 to make a new slot read-only. In this case, writes to this memory will be
1099 posted to userspace as KVM_EXIT_MMIO exits.
1101 When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of
1102 the memory region are automatically reflected into the guest. For example, an
1103 mmap() that affects the region will be made visible immediately. Another
1104 example is madvise(MADV_DROP).
1106 It is recommended to use this API instead of the KVM_SET_MEMORY_REGION ioctl.
1107 The KVM_SET_MEMORY_REGION does not allow fine grained control over memory
1108 allocation and is deprecated.
1111 4.36 KVM_SET_TSS_ADDR
1113 Capability: KVM_CAP_SET_TSS_ADDR
1116 Parameters: unsigned long tss_address (in)
1117 Returns: 0 on success, -1 on error
1119 This ioctl defines the physical address of a three-page region in the guest
1120 physical address space. The region must be within the first 4GB of the
1121 guest physical address space and must not conflict with any memory slot
1122 or any mmio address. The guest may malfunction if it accesses this memory
1125 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1126 because of a quirk in the virtualization implementation (see the internals
1127 documentation when it pops into existence).
1132 Capability: KVM_CAP_ENABLE_CAP, KVM_CAP_ENABLE_CAP_VM
1133 Architectures: x86 (only KVM_CAP_ENABLE_CAP_VM),
1134 mips (only KVM_CAP_ENABLE_CAP), ppc, s390
1135 Type: vcpu ioctl, vm ioctl (with KVM_CAP_ENABLE_CAP_VM)
1136 Parameters: struct kvm_enable_cap (in)
1137 Returns: 0 on success; -1 on error
1139 +Not all extensions are enabled by default. Using this ioctl the application
1140 can enable an extension, making it available to the guest.
1142 On systems that do not support this ioctl, it always fails. On systems that
1143 do support it, it only works for extensions that are supported for enablement.
1145 To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should
1148 struct kvm_enable_cap {
1152 The capability that is supposed to get enabled.
1156 A bitfield indicating future enhancements. Has to be 0 for now.
1160 Arguments for enabling a feature. If a feature needs initial values to
1161 function properly, this is the place to put them.
1166 The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl
1167 for vm-wide capabilities.
1169 4.38 KVM_GET_MP_STATE
1171 Capability: KVM_CAP_MP_STATE
1172 Architectures: x86, s390, arm, arm64
1174 Parameters: struct kvm_mp_state (out)
1175 Returns: 0 on success; -1 on error
1177 struct kvm_mp_state {
1181 Returns the vcpu's current "multiprocessing state" (though also valid on
1182 uniprocessor guests).
1184 Possible values are:
1186 - KVM_MP_STATE_RUNNABLE: the vcpu is currently running [x86,arm/arm64]
1187 - KVM_MP_STATE_UNINITIALIZED: the vcpu is an application processor (AP)
1188 which has not yet received an INIT signal [x86]
1189 - KVM_MP_STATE_INIT_RECEIVED: the vcpu has received an INIT signal, and is
1190 now ready for a SIPI [x86]
1191 - KVM_MP_STATE_HALTED: the vcpu has executed a HLT instruction and
1192 is waiting for an interrupt [x86]
1193 - KVM_MP_STATE_SIPI_RECEIVED: the vcpu has just received a SIPI (vector
1194 accessible via KVM_GET_VCPU_EVENTS) [x86]
1195 - KVM_MP_STATE_STOPPED: the vcpu is stopped [s390,arm/arm64]
1196 - KVM_MP_STATE_CHECK_STOP: the vcpu is in a special error state [s390]
1197 - KVM_MP_STATE_OPERATING: the vcpu is operating (running or halted)
1199 - KVM_MP_STATE_LOAD: the vcpu is in a special load/startup state
1202 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1203 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1204 these architectures.
1208 The only states that are valid are KVM_MP_STATE_STOPPED and
1209 KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not.
1211 4.39 KVM_SET_MP_STATE
1213 Capability: KVM_CAP_MP_STATE
1214 Architectures: x86, s390, arm, arm64
1216 Parameters: struct kvm_mp_state (in)
1217 Returns: 0 on success; -1 on error
1219 Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for
1222 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1223 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1224 these architectures.
1228 The only states that are valid are KVM_MP_STATE_STOPPED and
1229 KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not.
1231 4.40 KVM_SET_IDENTITY_MAP_ADDR
1233 Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR
1236 Parameters: unsigned long identity (in)
1237 Returns: 0 on success, -1 on error
1239 This ioctl defines the physical address of a one-page region in the guest
1240 physical address space. The region must be within the first 4GB of the
1241 guest physical address space and must not conflict with any memory slot
1242 or any mmio address. The guest may malfunction if it accesses this memory
1245 Setting the address to 0 will result in resetting the address to its default
1248 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1249 because of a quirk in the virtualization implementation (see the internals
1250 documentation when it pops into existence).
1252 Fails if any VCPU has already been created.
1254 4.41 KVM_SET_BOOT_CPU_ID
1256 Capability: KVM_CAP_SET_BOOT_CPU_ID
1259 Parameters: unsigned long vcpu_id
1260 Returns: 0 on success, -1 on error
1262 Define which vcpu is the Bootstrap Processor (BSP). Values are the same
1263 as the vcpu id in KVM_CREATE_VCPU. If this ioctl is not called, the default
1269 Capability: KVM_CAP_XSAVE
1272 Parameters: struct kvm_xsave (out)
1273 Returns: 0 on success, -1 on error
1279 This ioctl would copy current vcpu's xsave struct to the userspace.
1284 Capability: KVM_CAP_XSAVE
1287 Parameters: struct kvm_xsave (in)
1288 Returns: 0 on success, -1 on error
1294 This ioctl would copy userspace's xsave struct to the kernel.
1299 Capability: KVM_CAP_XCRS
1302 Parameters: struct kvm_xcrs (out)
1303 Returns: 0 on success, -1 on error
1314 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1318 This ioctl would copy current vcpu's xcrs to the userspace.
1323 Capability: KVM_CAP_XCRS
1326 Parameters: struct kvm_xcrs (in)
1327 Returns: 0 on success, -1 on error
1338 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1342 This ioctl would set vcpu's xcr to the value userspace specified.
1345 4.46 KVM_GET_SUPPORTED_CPUID
1347 Capability: KVM_CAP_EXT_CPUID
1350 Parameters: struct kvm_cpuid2 (in/out)
1351 Returns: 0 on success, -1 on error
1356 struct kvm_cpuid_entry2 entries[0];
1359 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
1360 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1)
1361 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2)
1363 struct kvm_cpuid_entry2 {
1374 This ioctl returns x86 cpuid features which are supported by both the
1375 hardware and kvm in its default configuration. Userspace can use the
1376 information returned by this ioctl to construct cpuid information (for
1377 KVM_SET_CPUID2) that is consistent with hardware, kernel, and
1378 userspace capabilities, and with user requirements (for example, the
1379 user may wish to constrain cpuid to emulate older hardware, or for
1380 feature consistency across a cluster).
1382 Note that certain capabilities, such as KVM_CAP_X86_DISABLE_EXITS, may
1383 expose cpuid features (e.g. MONITOR) which are not supported by kvm in
1384 its default configuration. If userspace enables such capabilities, it
1385 is responsible for modifying the results of this ioctl appropriately.
1387 Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure
1388 with the 'nent' field indicating the number of entries in the variable-size
1389 array 'entries'. If the number of entries is too low to describe the cpu
1390 capabilities, an error (E2BIG) is returned. If the number is too high,
1391 the 'nent' field is adjusted and an error (ENOMEM) is returned. If the
1392 number is just right, the 'nent' field is adjusted to the number of valid
1393 entries in the 'entries' array, which is then filled.
1395 The entries returned are the host cpuid as returned by the cpuid instruction,
1396 with unknown or unsupported features masked out. Some features (for example,
1397 x2apic), may not be present in the host cpu, but are exposed by kvm if it can
1398 emulate them efficiently. The fields in each entry are defined as follows:
1400 function: the eax value used to obtain the entry
1401 index: the ecx value used to obtain the entry (for entries that are
1403 flags: an OR of zero or more of the following:
1404 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
1405 if the index field is valid
1406 KVM_CPUID_FLAG_STATEFUL_FUNC:
1407 if cpuid for this function returns different values for successive
1408 invocations; there will be several entries with the same function,
1409 all with this flag set
1410 KVM_CPUID_FLAG_STATE_READ_NEXT:
1411 for KVM_CPUID_FLAG_STATEFUL_FUNC entries, set if this entry is
1412 the first entry to be read by a cpu
1413 eax, ebx, ecx, edx: the values returned by the cpuid instruction for
1414 this function/index combination
1416 The TSC deadline timer feature (CPUID leaf 1, ecx[24]) is always returned
1417 as false, since the feature depends on KVM_CREATE_IRQCHIP for local APIC
1418 support. Instead it is reported via
1420 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER)
1422 if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the
1423 feature in userspace, then you can enable the feature for KVM_SET_CPUID2.
1426 4.47 KVM_PPC_GET_PVINFO
1428 Capability: KVM_CAP_PPC_GET_PVINFO
1431 Parameters: struct kvm_ppc_pvinfo (out)
1432 Returns: 0 on success, !0 on error
1434 struct kvm_ppc_pvinfo {
1440 This ioctl fetches PV specific information that need to be passed to the guest
1441 using the device tree or other means from vm context.
1443 The hcall array defines 4 instructions that make up a hypercall.
1445 If any additional field gets added to this structure later on, a bit for that
1446 additional piece of information will be set in the flags bitmap.
1448 The flags bitmap is defined as:
1450 /* the host supports the ePAPR idle hcall
1451 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1<<0)
1453 4.52 KVM_SET_GSI_ROUTING
1455 Capability: KVM_CAP_IRQ_ROUTING
1456 Architectures: x86 s390 arm arm64
1458 Parameters: struct kvm_irq_routing (in)
1459 Returns: 0 on success, -1 on error
1461 Sets the GSI routing table entries, overwriting any previously set entries.
1463 On arm/arm64, GSI routing has the following limitation:
1464 - GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD.
1466 struct kvm_irq_routing {
1469 struct kvm_irq_routing_entry entries[0];
1472 No flags are specified so far, the corresponding field must be set to zero.
1474 struct kvm_irq_routing_entry {
1480 struct kvm_irq_routing_irqchip irqchip;
1481 struct kvm_irq_routing_msi msi;
1482 struct kvm_irq_routing_s390_adapter adapter;
1483 struct kvm_irq_routing_hv_sint hv_sint;
1488 /* gsi routing entry types */
1489 #define KVM_IRQ_ROUTING_IRQCHIP 1
1490 #define KVM_IRQ_ROUTING_MSI 2
1491 #define KVM_IRQ_ROUTING_S390_ADAPTER 3
1492 #define KVM_IRQ_ROUTING_HV_SINT 4
1495 - KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry
1496 type, specifies that the devid field contains a valid value. The per-VM
1497 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
1498 the device ID. If this capability is not available, userspace should
1499 never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
1502 struct kvm_irq_routing_irqchip {
1507 struct kvm_irq_routing_msi {
1517 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
1518 for the device that wrote the MSI message. For PCI, this is usually a
1519 BFD identifier in the lower 16 bits.
1521 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
1522 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
1523 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
1524 address_hi must be zero.
1526 struct kvm_irq_routing_s390_adapter {
1530 __u32 summary_offset;
1534 struct kvm_irq_routing_hv_sint {
1540 4.55 KVM_SET_TSC_KHZ
1542 Capability: KVM_CAP_TSC_CONTROL
1545 Parameters: virtual tsc_khz
1546 Returns: 0 on success, -1 on error
1548 Specifies the tsc frequency for the virtual machine. The unit of the
1552 4.56 KVM_GET_TSC_KHZ
1554 Capability: KVM_CAP_GET_TSC_KHZ
1558 Returns: virtual tsc-khz on success, negative value on error
1560 Returns the tsc frequency of the guest. The unit of the return value is
1561 KHz. If the host has unstable tsc this ioctl returns -EIO instead as an
1567 Capability: KVM_CAP_IRQCHIP
1570 Parameters: struct kvm_lapic_state (out)
1571 Returns: 0 on success, -1 on error
1573 #define KVM_APIC_REG_SIZE 0x400
1574 struct kvm_lapic_state {
1575 char regs[KVM_APIC_REG_SIZE];
1578 Reads the Local APIC registers and copies them into the input argument. The
1579 data format and layout are the same as documented in the architecture manual.
1581 If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is
1582 enabled, then the format of APIC_ID register depends on the APIC mode
1583 (reported by MSR_IA32_APICBASE) of its VCPU. x2APIC stores APIC ID in
1584 the APIC_ID register (bytes 32-35). xAPIC only allows an 8-bit APIC ID
1585 which is stored in bits 31-24 of the APIC register, or equivalently in
1586 byte 35 of struct kvm_lapic_state's regs field. KVM_GET_LAPIC must then
1587 be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR.
1589 If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state
1590 always uses xAPIC format.
1595 Capability: KVM_CAP_IRQCHIP
1598 Parameters: struct kvm_lapic_state (in)
1599 Returns: 0 on success, -1 on error
1601 #define KVM_APIC_REG_SIZE 0x400
1602 struct kvm_lapic_state {
1603 char regs[KVM_APIC_REG_SIZE];
1606 Copies the input argument into the Local APIC registers. The data format
1607 and layout are the same as documented in the architecture manual.
1609 The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's
1610 regs field) depends on the state of the KVM_CAP_X2APIC_API capability.
1611 See the note in KVM_GET_LAPIC.
1616 Capability: KVM_CAP_IOEVENTFD
1619 Parameters: struct kvm_ioeventfd (in)
1620 Returns: 0 on success, !0 on error
1622 This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address
1623 within the guest. A guest write in the registered address will signal the
1624 provided event instead of triggering an exit.
1626 struct kvm_ioeventfd {
1628 __u64 addr; /* legal pio/mmio address */
1629 __u32 len; /* 0, 1, 2, 4, or 8 bytes */
1635 For the special case of virtio-ccw devices on s390, the ioevent is matched
1636 to a subchannel/virtqueue tuple instead.
1638 The following flags are defined:
1640 #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch)
1641 #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio)
1642 #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign)
1643 #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \
1644 (1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify)
1646 If datamatch flag is set, the event will be signaled only if the written value
1647 to the registered address is equal to datamatch in struct kvm_ioeventfd.
1649 For virtio-ccw devices, addr contains the subchannel id and datamatch the
1652 With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and
1653 the kernel will ignore the length of guest write and may get a faster vmexit.
1654 The speedup may only apply to specific architectures, but the ioeventfd will
1659 Capability: KVM_CAP_SW_TLB
1662 Parameters: struct kvm_dirty_tlb (in)
1663 Returns: 0 on success, -1 on error
1665 struct kvm_dirty_tlb {
1670 This must be called whenever userspace has changed an entry in the shared
1671 TLB, prior to calling KVM_RUN on the associated vcpu.
1673 The "bitmap" field is the userspace address of an array. This array
1674 consists of a number of bits, equal to the total number of TLB entries as
1675 determined by the last successful call to KVM_CONFIG_TLB, rounded up to the
1676 nearest multiple of 64.
1678 Each bit corresponds to one TLB entry, ordered the same as in the shared TLB
1681 The array is little-endian: the bit 0 is the least significant bit of the
1682 first byte, bit 8 is the least significant bit of the second byte, etc.
1683 This avoids any complications with differing word sizes.
1685 The "num_dirty" field is a performance hint for KVM to determine whether it
1686 should skip processing the bitmap and just invalidate everything. It must
1687 be set to the number of set bits in the bitmap.
1690 4.62 KVM_CREATE_SPAPR_TCE
1692 Capability: KVM_CAP_SPAPR_TCE
1693 Architectures: powerpc
1695 Parameters: struct kvm_create_spapr_tce (in)
1696 Returns: file descriptor for manipulating the created TCE table
1698 This creates a virtual TCE (translation control entry) table, which
1699 is an IOMMU for PAPR-style virtual I/O. It is used to translate
1700 logical addresses used in virtual I/O into guest physical addresses,
1701 and provides a scatter/gather capability for PAPR virtual I/O.
1703 /* for KVM_CAP_SPAPR_TCE */
1704 struct kvm_create_spapr_tce {
1709 The liobn field gives the logical IO bus number for which to create a
1710 TCE table. The window_size field specifies the size of the DMA window
1711 which this TCE table will translate - the table will contain one 64
1712 bit TCE entry for every 4kiB of the DMA window.
1714 When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE
1715 table has been created using this ioctl(), the kernel will handle it
1716 in real mode, updating the TCE table. H_PUT_TCE calls for other
1717 liobns will cause a vm exit and must be handled by userspace.
1719 The return value is a file descriptor which can be passed to mmap(2)
1720 to map the created TCE table into userspace. This lets userspace read
1721 the entries written by kernel-handled H_PUT_TCE calls, and also lets
1722 userspace update the TCE table directly which is useful in some
1726 4.63 KVM_ALLOCATE_RMA
1728 Capability: KVM_CAP_PPC_RMA
1729 Architectures: powerpc
1731 Parameters: struct kvm_allocate_rma (out)
1732 Returns: file descriptor for mapping the allocated RMA
1734 This allocates a Real Mode Area (RMA) from the pool allocated at boot
1735 time by the kernel. An RMA is a physically-contiguous, aligned region
1736 of memory used on older POWER processors to provide the memory which
1737 will be accessed by real-mode (MMU off) accesses in a KVM guest.
1738 POWER processors support a set of sizes for the RMA that usually
1739 includes 64MB, 128MB, 256MB and some larger powers of two.
1741 /* for KVM_ALLOCATE_RMA */
1742 struct kvm_allocate_rma {
1746 The return value is a file descriptor which can be passed to mmap(2)
1747 to map the allocated RMA into userspace. The mapped area can then be
1748 passed to the KVM_SET_USER_MEMORY_REGION ioctl to establish it as the
1749 RMA for a virtual machine. The size of the RMA in bytes (which is
1750 fixed at host kernel boot time) is returned in the rma_size field of
1751 the argument structure.
1753 The KVM_CAP_PPC_RMA capability is 1 or 2 if the KVM_ALLOCATE_RMA ioctl
1754 is supported; 2 if the processor requires all virtual machines to have
1755 an RMA, or 1 if the processor can use an RMA but doesn't require it,
1756 because it supports the Virtual RMA (VRMA) facility.
1761 Capability: KVM_CAP_USER_NMI
1765 Returns: 0 on success, -1 on error
1767 Queues an NMI on the thread's vcpu. Note this is well defined only
1768 when KVM_CREATE_IRQCHIP has not been called, since this is an interface
1769 between the virtual cpu core and virtual local APIC. After KVM_CREATE_IRQCHIP
1770 has been called, this interface is completely emulated within the kernel.
1772 To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the
1773 following algorithm:
1776 - read the local APIC's state (KVM_GET_LAPIC)
1777 - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1)
1778 - if so, issue KVM_NMI
1781 Some guests configure the LINT1 NMI input to cause a panic, aiding in
1785 4.65 KVM_S390_UCAS_MAP
1787 Capability: KVM_CAP_S390_UCONTROL
1790 Parameters: struct kvm_s390_ucas_mapping (in)
1791 Returns: 0 in case of success
1793 The parameter is defined like this:
1794 struct kvm_s390_ucas_mapping {
1800 This ioctl maps the memory at "user_addr" with the length "length" to
1801 the vcpu's address space starting at "vcpu_addr". All parameters need to
1802 be aligned by 1 megabyte.
1805 4.66 KVM_S390_UCAS_UNMAP
1807 Capability: KVM_CAP_S390_UCONTROL
1810 Parameters: struct kvm_s390_ucas_mapping (in)
1811 Returns: 0 in case of success
1813 The parameter is defined like this:
1814 struct kvm_s390_ucas_mapping {
1820 This ioctl unmaps the memory in the vcpu's address space starting at
1821 "vcpu_addr" with the length "length". The field "user_addr" is ignored.
1822 All parameters need to be aligned by 1 megabyte.
1825 4.67 KVM_S390_VCPU_FAULT
1827 Capability: KVM_CAP_S390_UCONTROL
1830 Parameters: vcpu absolute address (in)
1831 Returns: 0 in case of success
1833 This call creates a page table entry on the virtual cpu's address space
1834 (for user controlled virtual machines) or the virtual machine's address
1835 space (for regular virtual machines). This only works for minor faults,
1836 thus it's recommended to access subject memory page via the user page
1837 table upfront. This is useful to handle validity intercepts for user
1838 controlled virtual machines to fault in the virtual cpu's lowcore pages
1839 prior to calling the KVM_RUN ioctl.
1842 4.68 KVM_SET_ONE_REG
1844 Capability: KVM_CAP_ONE_REG
1847 Parameters: struct kvm_one_reg (in)
1848 Returns: 0 on success, negative value on failure
1850 struct kvm_one_reg {
1855 Using this ioctl, a single vcpu register can be set to a specific value
1856 defined by user space with the passed in struct kvm_one_reg, where id
1857 refers to the register identifier as described below and addr is a pointer
1858 to a variable with the respective size. There can be architecture agnostic
1859 and architecture specific registers. Each have their own range of operation
1860 and their own constants and width. To keep track of the implemented
1861 registers, find a list below:
1863 Arch | Register | Width (bits)
1865 PPC | KVM_REG_PPC_HIOR | 64
1866 PPC | KVM_REG_PPC_IAC1 | 64
1867 PPC | KVM_REG_PPC_IAC2 | 64
1868 PPC | KVM_REG_PPC_IAC3 | 64
1869 PPC | KVM_REG_PPC_IAC4 | 64
1870 PPC | KVM_REG_PPC_DAC1 | 64
1871 PPC | KVM_REG_PPC_DAC2 | 64
1872 PPC | KVM_REG_PPC_DABR | 64
1873 PPC | KVM_REG_PPC_DSCR | 64
1874 PPC | KVM_REG_PPC_PURR | 64
1875 PPC | KVM_REG_PPC_SPURR | 64
1876 PPC | KVM_REG_PPC_DAR | 64
1877 PPC | KVM_REG_PPC_DSISR | 32
1878 PPC | KVM_REG_PPC_AMR | 64
1879 PPC | KVM_REG_PPC_UAMOR | 64
1880 PPC | KVM_REG_PPC_MMCR0 | 64
1881 PPC | KVM_REG_PPC_MMCR1 | 64
1882 PPC | KVM_REG_PPC_MMCRA | 64
1883 PPC | KVM_REG_PPC_MMCR2 | 64
1884 PPC | KVM_REG_PPC_MMCRS | 64
1885 PPC | KVM_REG_PPC_SIAR | 64
1886 PPC | KVM_REG_PPC_SDAR | 64
1887 PPC | KVM_REG_PPC_SIER | 64
1888 PPC | KVM_REG_PPC_PMC1 | 32
1889 PPC | KVM_REG_PPC_PMC2 | 32
1890 PPC | KVM_REG_PPC_PMC3 | 32
1891 PPC | KVM_REG_PPC_PMC4 | 32
1892 PPC | KVM_REG_PPC_PMC5 | 32
1893 PPC | KVM_REG_PPC_PMC6 | 32
1894 PPC | KVM_REG_PPC_PMC7 | 32
1895 PPC | KVM_REG_PPC_PMC8 | 32
1896 PPC | KVM_REG_PPC_FPR0 | 64
1898 PPC | KVM_REG_PPC_FPR31 | 64
1899 PPC | KVM_REG_PPC_VR0 | 128
1901 PPC | KVM_REG_PPC_VR31 | 128
1902 PPC | KVM_REG_PPC_VSR0 | 128
1904 PPC | KVM_REG_PPC_VSR31 | 128
1905 PPC | KVM_REG_PPC_FPSCR | 64
1906 PPC | KVM_REG_PPC_VSCR | 32
1907 PPC | KVM_REG_PPC_VPA_ADDR | 64
1908 PPC | KVM_REG_PPC_VPA_SLB | 128
1909 PPC | KVM_REG_PPC_VPA_DTL | 128
1910 PPC | KVM_REG_PPC_EPCR | 32
1911 PPC | KVM_REG_PPC_EPR | 32
1912 PPC | KVM_REG_PPC_TCR | 32
1913 PPC | KVM_REG_PPC_TSR | 32
1914 PPC | KVM_REG_PPC_OR_TSR | 32
1915 PPC | KVM_REG_PPC_CLEAR_TSR | 32
1916 PPC | KVM_REG_PPC_MAS0 | 32
1917 PPC | KVM_REG_PPC_MAS1 | 32
1918 PPC | KVM_REG_PPC_MAS2 | 64
1919 PPC | KVM_REG_PPC_MAS7_3 | 64
1920 PPC | KVM_REG_PPC_MAS4 | 32
1921 PPC | KVM_REG_PPC_MAS6 | 32
1922 PPC | KVM_REG_PPC_MMUCFG | 32
1923 PPC | KVM_REG_PPC_TLB0CFG | 32
1924 PPC | KVM_REG_PPC_TLB1CFG | 32
1925 PPC | KVM_REG_PPC_TLB2CFG | 32
1926 PPC | KVM_REG_PPC_TLB3CFG | 32
1927 PPC | KVM_REG_PPC_TLB0PS | 32
1928 PPC | KVM_REG_PPC_TLB1PS | 32
1929 PPC | KVM_REG_PPC_TLB2PS | 32
1930 PPC | KVM_REG_PPC_TLB3PS | 32
1931 PPC | KVM_REG_PPC_EPTCFG | 32
1932 PPC | KVM_REG_PPC_ICP_STATE | 64
1933 PPC | KVM_REG_PPC_TB_OFFSET | 64
1934 PPC | KVM_REG_PPC_SPMC1 | 32
1935 PPC | KVM_REG_PPC_SPMC2 | 32
1936 PPC | KVM_REG_PPC_IAMR | 64
1937 PPC | KVM_REG_PPC_TFHAR | 64
1938 PPC | KVM_REG_PPC_TFIAR | 64
1939 PPC | KVM_REG_PPC_TEXASR | 64
1940 PPC | KVM_REG_PPC_FSCR | 64
1941 PPC | KVM_REG_PPC_PSPB | 32
1942 PPC | KVM_REG_PPC_EBBHR | 64
1943 PPC | KVM_REG_PPC_EBBRR | 64
1944 PPC | KVM_REG_PPC_BESCR | 64
1945 PPC | KVM_REG_PPC_TAR | 64
1946 PPC | KVM_REG_PPC_DPDES | 64
1947 PPC | KVM_REG_PPC_DAWR | 64
1948 PPC | KVM_REG_PPC_DAWRX | 64
1949 PPC | KVM_REG_PPC_CIABR | 64
1950 PPC | KVM_REG_PPC_IC | 64
1951 PPC | KVM_REG_PPC_VTB | 64
1952 PPC | KVM_REG_PPC_CSIGR | 64
1953 PPC | KVM_REG_PPC_TACR | 64
1954 PPC | KVM_REG_PPC_TCSCR | 64
1955 PPC | KVM_REG_PPC_PID | 64
1956 PPC | KVM_REG_PPC_ACOP | 64
1957 PPC | KVM_REG_PPC_VRSAVE | 32
1958 PPC | KVM_REG_PPC_LPCR | 32
1959 PPC | KVM_REG_PPC_LPCR_64 | 64
1960 PPC | KVM_REG_PPC_PPR | 64
1961 PPC | KVM_REG_PPC_ARCH_COMPAT | 32
1962 PPC | KVM_REG_PPC_DABRX | 32
1963 PPC | KVM_REG_PPC_WORT | 64
1964 PPC | KVM_REG_PPC_SPRG9 | 64
1965 PPC | KVM_REG_PPC_DBSR | 32
1966 PPC | KVM_REG_PPC_TIDR | 64
1967 PPC | KVM_REG_PPC_PSSCR | 64
1968 PPC | KVM_REG_PPC_DEC_EXPIRY | 64
1969 PPC | KVM_REG_PPC_PTCR | 64
1970 PPC | KVM_REG_PPC_TM_GPR0 | 64
1972 PPC | KVM_REG_PPC_TM_GPR31 | 64
1973 PPC | KVM_REG_PPC_TM_VSR0 | 128
1975 PPC | KVM_REG_PPC_TM_VSR63 | 128
1976 PPC | KVM_REG_PPC_TM_CR | 64
1977 PPC | KVM_REG_PPC_TM_LR | 64
1978 PPC | KVM_REG_PPC_TM_CTR | 64
1979 PPC | KVM_REG_PPC_TM_FPSCR | 64
1980 PPC | KVM_REG_PPC_TM_AMR | 64
1981 PPC | KVM_REG_PPC_TM_PPR | 64
1982 PPC | KVM_REG_PPC_TM_VRSAVE | 64
1983 PPC | KVM_REG_PPC_TM_VSCR | 32
1984 PPC | KVM_REG_PPC_TM_DSCR | 64
1985 PPC | KVM_REG_PPC_TM_TAR | 64
1986 PPC | KVM_REG_PPC_TM_XER | 64
1988 MIPS | KVM_REG_MIPS_R0 | 64
1990 MIPS | KVM_REG_MIPS_R31 | 64
1991 MIPS | KVM_REG_MIPS_HI | 64
1992 MIPS | KVM_REG_MIPS_LO | 64
1993 MIPS | KVM_REG_MIPS_PC | 64
1994 MIPS | KVM_REG_MIPS_CP0_INDEX | 32
1995 MIPS | KVM_REG_MIPS_CP0_ENTRYLO0 | 64
1996 MIPS | KVM_REG_MIPS_CP0_ENTRYLO1 | 64
1997 MIPS | KVM_REG_MIPS_CP0_CONTEXT | 64
1998 MIPS | KVM_REG_MIPS_CP0_CONTEXTCONFIG| 32
1999 MIPS | KVM_REG_MIPS_CP0_USERLOCAL | 64
2000 MIPS | KVM_REG_MIPS_CP0_XCONTEXTCONFIG| 64
2001 MIPS | KVM_REG_MIPS_CP0_PAGEMASK | 32
2002 MIPS | KVM_REG_MIPS_CP0_PAGEGRAIN | 32
2003 MIPS | KVM_REG_MIPS_CP0_SEGCTL0 | 64
2004 MIPS | KVM_REG_MIPS_CP0_SEGCTL1 | 64
2005 MIPS | KVM_REG_MIPS_CP0_SEGCTL2 | 64
2006 MIPS | KVM_REG_MIPS_CP0_PWBASE | 64
2007 MIPS | KVM_REG_MIPS_CP0_PWFIELD | 64
2008 MIPS | KVM_REG_MIPS_CP0_PWSIZE | 64
2009 MIPS | KVM_REG_MIPS_CP0_WIRED | 32
2010 MIPS | KVM_REG_MIPS_CP0_PWCTL | 32
2011 MIPS | KVM_REG_MIPS_CP0_HWRENA | 32
2012 MIPS | KVM_REG_MIPS_CP0_BADVADDR | 64
2013 MIPS | KVM_REG_MIPS_CP0_BADINSTR | 32
2014 MIPS | KVM_REG_MIPS_CP0_BADINSTRP | 32
2015 MIPS | KVM_REG_MIPS_CP0_COUNT | 32
2016 MIPS | KVM_REG_MIPS_CP0_ENTRYHI | 64
2017 MIPS | KVM_REG_MIPS_CP0_COMPARE | 32
2018 MIPS | KVM_REG_MIPS_CP0_STATUS | 32
2019 MIPS | KVM_REG_MIPS_CP0_INTCTL | 32
2020 MIPS | KVM_REG_MIPS_CP0_CAUSE | 32
2021 MIPS | KVM_REG_MIPS_CP0_EPC | 64
2022 MIPS | KVM_REG_MIPS_CP0_PRID | 32
2023 MIPS | KVM_REG_MIPS_CP0_EBASE | 64
2024 MIPS | KVM_REG_MIPS_CP0_CONFIG | 32
2025 MIPS | KVM_REG_MIPS_CP0_CONFIG1 | 32
2026 MIPS | KVM_REG_MIPS_CP0_CONFIG2 | 32
2027 MIPS | KVM_REG_MIPS_CP0_CONFIG3 | 32
2028 MIPS | KVM_REG_MIPS_CP0_CONFIG4 | 32
2029 MIPS | KVM_REG_MIPS_CP0_CONFIG5 | 32
2030 MIPS | KVM_REG_MIPS_CP0_CONFIG7 | 32
2031 MIPS | KVM_REG_MIPS_CP0_XCONTEXT | 64
2032 MIPS | KVM_REG_MIPS_CP0_ERROREPC | 64
2033 MIPS | KVM_REG_MIPS_CP0_KSCRATCH1 | 64
2034 MIPS | KVM_REG_MIPS_CP0_KSCRATCH2 | 64
2035 MIPS | KVM_REG_MIPS_CP0_KSCRATCH3 | 64
2036 MIPS | KVM_REG_MIPS_CP0_KSCRATCH4 | 64
2037 MIPS | KVM_REG_MIPS_CP0_KSCRATCH5 | 64
2038 MIPS | KVM_REG_MIPS_CP0_KSCRATCH6 | 64
2039 MIPS | KVM_REG_MIPS_CP0_MAAR(0..63) | 64
2040 MIPS | KVM_REG_MIPS_COUNT_CTL | 64
2041 MIPS | KVM_REG_MIPS_COUNT_RESUME | 64
2042 MIPS | KVM_REG_MIPS_COUNT_HZ | 64
2043 MIPS | KVM_REG_MIPS_FPR_32(0..31) | 32
2044 MIPS | KVM_REG_MIPS_FPR_64(0..31) | 64
2045 MIPS | KVM_REG_MIPS_VEC_128(0..31) | 128
2046 MIPS | KVM_REG_MIPS_FCR_IR | 32
2047 MIPS | KVM_REG_MIPS_FCR_CSR | 32
2048 MIPS | KVM_REG_MIPS_MSA_IR | 32
2049 MIPS | KVM_REG_MIPS_MSA_CSR | 32
2051 ARM registers are mapped using the lower 32 bits. The upper 16 of that
2052 is the register group type, or coprocessor number:
2054 ARM core registers have the following id bit patterns:
2055 0x4020 0000 0010 <index into the kvm_regs struct:16>
2057 ARM 32-bit CP15 registers have the following id bit patterns:
2058 0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3>
2060 ARM 64-bit CP15 registers have the following id bit patterns:
2061 0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3>
2063 ARM CCSIDR registers are demultiplexed by CSSELR value:
2064 0x4020 0000 0011 00 <csselr:8>
2066 ARM 32-bit VFP control registers have the following id bit patterns:
2067 0x4020 0000 0012 1 <regno:12>
2069 ARM 64-bit FP registers have the following id bit patterns:
2070 0x4030 0000 0012 0 <regno:12>
2072 ARM firmware pseudo-registers have the following bit pattern:
2073 0x4030 0000 0014 <regno:16>
2076 arm64 registers are mapped using the lower 32 bits. The upper 16 of
2077 that is the register group type, or coprocessor number:
2079 arm64 core/FP-SIMD registers have the following id bit patterns. Note
2080 that the size of the access is variable, as the kvm_regs structure
2081 contains elements ranging from 32 to 128 bits. The index is a 32bit
2082 value in the kvm_regs structure seen as a 32bit array.
2083 0x60x0 0000 0010 <index into the kvm_regs struct:16>
2085 arm64 CCSIDR registers are demultiplexed by CSSELR value:
2086 0x6020 0000 0011 00 <csselr:8>
2088 arm64 system registers have the following id bit patterns:
2089 0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3>
2091 arm64 firmware pseudo-registers have the following bit pattern:
2092 0x6030 0000 0014 <regno:16>
2095 MIPS registers are mapped using the lower 32 bits. The upper 16 of that is
2096 the register group type:
2098 MIPS core registers (see above) have the following id bit patterns:
2099 0x7030 0000 0000 <reg:16>
2101 MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit
2102 patterns depending on whether they're 32-bit or 64-bit registers:
2103 0x7020 0000 0001 00 <reg:5> <sel:3> (32-bit)
2104 0x7030 0000 0001 00 <reg:5> <sel:3> (64-bit)
2106 Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_MIPS_CP0_ENTRYLO1 are the MIPS64
2107 versions of the EntryLo registers regardless of the word size of the host
2108 hardware, host kernel, guest, and whether XPA is present in the guest, i.e.
2109 with the RI and XI bits (if they exist) in bits 63 and 62 respectively, and
2110 the PFNX field starting at bit 30.
2112 MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) above) have the following id bit
2114 0x7030 0000 0001 01 <reg:8>
2116 MIPS KVM control registers (see above) have the following id bit patterns:
2117 0x7030 0000 0002 <reg:16>
2119 MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following
2120 id bit patterns depending on the size of the register being accessed. They are
2121 always accessed according to the current guest FPU mode (Status.FR and
2122 Config5.FRE), i.e. as the guest would see them, and they become unpredictable
2123 if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector
2124 registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they
2125 overlap the FPU registers:
2126 0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers)
2127 0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers)
2128 0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers)
2130 MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the
2131 following id bit patterns:
2132 0x7020 0000 0003 01 <0:3> <reg:5>
2134 MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the
2135 following id bit patterns:
2136 0x7020 0000 0003 02 <0:3> <reg:5>
2139 4.69 KVM_GET_ONE_REG
2141 Capability: KVM_CAP_ONE_REG
2144 Parameters: struct kvm_one_reg (in and out)
2145 Returns: 0 on success, negative value on failure
2147 This ioctl allows to receive the value of a single register implemented
2148 in a vcpu. The register to read is indicated by the "id" field of the
2149 kvm_one_reg struct passed in. On success, the register value can be found
2150 at the memory location pointed to by "addr".
2152 The list of registers accessible using this interface is identical to the
2156 4.70 KVM_KVMCLOCK_CTRL
2158 Capability: KVM_CAP_KVMCLOCK_CTRL
2159 Architectures: Any that implement pvclocks (currently x86 only)
2162 Returns: 0 on success, -1 on error
2164 This signals to the host kernel that the specified guest is being paused by
2165 userspace. The host will set a flag in the pvclock structure that is checked
2166 from the soft lockup watchdog. The flag is part of the pvclock structure that
2167 is shared between guest and host, specifically the second bit of the flags
2168 field of the pvclock_vcpu_time_info structure. It will be set exclusively by
2169 the host and read/cleared exclusively by the guest. The guest operation of
2170 checking and clearing the flag must an atomic operation so
2171 load-link/store-conditional, or equivalent must be used. There are two cases
2172 where the guest will clear the flag: when the soft lockup watchdog timer resets
2173 itself or when a soft lockup is detected. This ioctl can be called any time
2174 after pausing the vcpu, but before it is resumed.
2179 Capability: KVM_CAP_SIGNAL_MSI
2180 Architectures: x86 arm arm64
2182 Parameters: struct kvm_msi (in)
2183 Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error
2185 Directly inject a MSI message. Only valid with in-kernel irqchip that handles
2197 flags: KVM_MSI_VALID_DEVID: devid contains a valid value. The per-VM
2198 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
2199 the device ID. If this capability is not available, userspace
2200 should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
2202 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
2203 for the device that wrote the MSI message. For PCI, this is usually a
2204 BFD identifier in the lower 16 bits.
2206 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
2207 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
2208 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
2209 address_hi must be zero.
2212 4.71 KVM_CREATE_PIT2
2214 Capability: KVM_CAP_PIT2
2217 Parameters: struct kvm_pit_config (in)
2218 Returns: 0 on success, -1 on error
2220 Creates an in-kernel device model for the i8254 PIT. This call is only valid
2221 after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following
2222 parameters have to be passed:
2224 struct kvm_pit_config {
2231 #define KVM_PIT_SPEAKER_DUMMY 1 /* emulate speaker port stub */
2233 PIT timer interrupts may use a per-VM kernel thread for injection. If it
2234 exists, this thread will have a name of the following pattern:
2236 kvm-pit/<owner-process-pid>
2238 When running a guest with elevated priorities, the scheduling parameters of
2239 this thread may have to be adjusted accordingly.
2241 This IOCTL replaces the obsolete KVM_CREATE_PIT.
2246 Capability: KVM_CAP_PIT_STATE2
2249 Parameters: struct kvm_pit_state2 (out)
2250 Returns: 0 on success, -1 on error
2252 Retrieves the state of the in-kernel PIT model. Only valid after
2253 KVM_CREATE_PIT2. The state is returned in the following structure:
2255 struct kvm_pit_state2 {
2256 struct kvm_pit_channel_state channels[3];
2263 /* disable PIT in HPET legacy mode */
2264 #define KVM_PIT_FLAGS_HPET_LEGACY 0x00000001
2266 This IOCTL replaces the obsolete KVM_GET_PIT.
2271 Capability: KVM_CAP_PIT_STATE2
2274 Parameters: struct kvm_pit_state2 (in)
2275 Returns: 0 on success, -1 on error
2277 Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2.
2278 See KVM_GET_PIT2 for details on struct kvm_pit_state2.
2280 This IOCTL replaces the obsolete KVM_SET_PIT.
2283 4.74 KVM_PPC_GET_SMMU_INFO
2285 Capability: KVM_CAP_PPC_GET_SMMU_INFO
2286 Architectures: powerpc
2289 Returns: 0 on success, -1 on error
2291 This populates and returns a structure describing the features of
2292 the "Server" class MMU emulation supported by KVM.
2293 This can in turn be used by userspace to generate the appropriate
2294 device-tree properties for the guest operating system.
2296 The structure contains some global information, followed by an
2297 array of supported segment page sizes:
2299 struct kvm_ppc_smmu_info {
2303 struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ];
2306 The supported flags are:
2308 - KVM_PPC_PAGE_SIZES_REAL:
2309 When that flag is set, guest page sizes must "fit" the backing
2310 store page sizes. When not set, any page size in the list can
2311 be used regardless of how they are backed by userspace.
2313 - KVM_PPC_1T_SEGMENTS
2314 The emulated MMU supports 1T segments in addition to the
2318 This flag indicates that HPT guests are not supported by KVM,
2319 thus all guests must use radix MMU mode.
2321 The "slb_size" field indicates how many SLB entries are supported
2323 The "sps" array contains 8 entries indicating the supported base
2324 page sizes for a segment in increasing order. Each entry is defined
2327 struct kvm_ppc_one_seg_page_size {
2328 __u32 page_shift; /* Base page shift of segment (or 0) */
2329 __u32 slb_enc; /* SLB encoding for BookS */
2330 struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ];
2333 An entry with a "page_shift" of 0 is unused. Because the array is
2334 organized in increasing order, a lookup can stop when encoutering
2337 The "slb_enc" field provides the encoding to use in the SLB for the
2338 page size. The bits are in positions such as the value can directly
2339 be OR'ed into the "vsid" argument of the slbmte instruction.
2341 The "enc" array is a list which for each of those segment base page
2342 size provides the list of supported actual page sizes (which can be
2343 only larger or equal to the base page size), along with the
2344 corresponding encoding in the hash PTE. Similarly, the array is
2345 8 entries sorted by increasing sizes and an entry with a "0" shift
2346 is an empty entry and a terminator:
2348 struct kvm_ppc_one_page_size {
2349 __u32 page_shift; /* Page shift (or 0) */
2350 __u32 pte_enc; /* Encoding in the HPTE (>>12) */
2353 The "pte_enc" field provides a value that can OR'ed into the hash
2354 PTE's RPN field (ie, it needs to be shifted left by 12 to OR it
2355 into the hash PTE second double word).
2359 Capability: KVM_CAP_IRQFD
2360 Architectures: x86 s390 arm arm64
2362 Parameters: struct kvm_irqfd (in)
2363 Returns: 0 on success, -1 on error
2365 Allows setting an eventfd to directly trigger a guest interrupt.
2366 kvm_irqfd.fd specifies the file descriptor to use as the eventfd and
2367 kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When
2368 an event is triggered on the eventfd, an interrupt is injected into
2369 the guest using the specified gsi pin. The irqfd is removed using
2370 the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd
2373 With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify
2374 mechanism allowing emulation of level-triggered, irqfd-based
2375 interrupts. When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an
2376 additional eventfd in the kvm_irqfd.resamplefd field. When operating
2377 in resample mode, posting of an interrupt through kvm_irq.fd asserts
2378 the specified gsi in the irqchip. When the irqchip is resampled, such
2379 as from an EOI, the gsi is de-asserted and the user is notified via
2380 kvm_irqfd.resamplefd. It is the user's responsibility to re-queue
2381 the interrupt if the device making use of it still requires service.
2382 Note that closing the resamplefd is not sufficient to disable the
2383 irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment
2384 and need not be specified with KVM_IRQFD_FLAG_DEASSIGN.
2386 On arm/arm64, gsi routing being supported, the following can happen:
2387 - in case no routing entry is associated to this gsi, injection fails
2388 - in case the gsi is associated to an irqchip routing entry,
2389 irqchip.pin + 32 corresponds to the injected SPI ID.
2390 - in case the gsi is associated to an MSI routing entry, the MSI
2391 message and device ID are translated into an LPI (support restricted
2392 to GICv3 ITS in-kernel emulation).
2394 4.76 KVM_PPC_ALLOCATE_HTAB
2396 Capability: KVM_CAP_PPC_ALLOC_HTAB
2397 Architectures: powerpc
2399 Parameters: Pointer to u32 containing hash table order (in/out)
2400 Returns: 0 on success, -1 on error
2402 This requests the host kernel to allocate an MMU hash table for a
2403 guest using the PAPR paravirtualization interface. This only does
2404 anything if the kernel is configured to use the Book 3S HV style of
2405 virtualization. Otherwise the capability doesn't exist and the ioctl
2406 returns an ENOTTY error. The rest of this description assumes Book 3S
2409 There must be no vcpus running when this ioctl is called; if there
2410 are, it will do nothing and return an EBUSY error.
2412 The parameter is a pointer to a 32-bit unsigned integer variable
2413 containing the order (log base 2) of the desired size of the hash
2414 table, which must be between 18 and 46. On successful return from the
2415 ioctl, the value will not be changed by the kernel.
2417 If no hash table has been allocated when any vcpu is asked to run
2418 (with the KVM_RUN ioctl), the host kernel will allocate a
2419 default-sized hash table (16 MB).
2421 If this ioctl is called when a hash table has already been allocated,
2422 with a different order from the existing hash table, the existing hash
2423 table will be freed and a new one allocated. If this is ioctl is
2424 called when a hash table has already been allocated of the same order
2425 as specified, the kernel will clear out the existing hash table (zero
2426 all HPTEs). In either case, if the guest is using the virtualized
2427 real-mode area (VRMA) facility, the kernel will re-create the VMRA
2428 HPTEs on the next KVM_RUN of any vcpu.
2430 4.77 KVM_S390_INTERRUPT
2434 Type: vm ioctl, vcpu ioctl
2435 Parameters: struct kvm_s390_interrupt (in)
2436 Returns: 0 on success, -1 on error
2438 Allows to inject an interrupt to the guest. Interrupts can be floating
2439 (vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type.
2441 Interrupt parameters are passed via kvm_s390_interrupt:
2443 struct kvm_s390_interrupt {
2449 type can be one of the following:
2451 KVM_S390_SIGP_STOP (vcpu) - sigp stop; optional flags in parm
2452 KVM_S390_PROGRAM_INT (vcpu) - program check; code in parm
2453 KVM_S390_SIGP_SET_PREFIX (vcpu) - sigp set prefix; prefix address in parm
2454 KVM_S390_RESTART (vcpu) - restart
2455 KVM_S390_INT_CLOCK_COMP (vcpu) - clock comparator interrupt
2456 KVM_S390_INT_CPU_TIMER (vcpu) - CPU timer interrupt
2457 KVM_S390_INT_VIRTIO (vm) - virtio external interrupt; external interrupt
2458 parameters in parm and parm64
2459 KVM_S390_INT_SERVICE (vm) - sclp external interrupt; sclp parameter in parm
2460 KVM_S390_INT_EMERGENCY (vcpu) - sigp emergency; source cpu in parm
2461 KVM_S390_INT_EXTERNAL_CALL (vcpu) - sigp external call; source cpu in parm
2462 KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm) - compound value to indicate an
2463 I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel);
2464 I/O interruption parameters in parm (subchannel) and parm64 (intparm,
2465 interruption subclass)
2466 KVM_S390_MCHK (vm, vcpu) - machine check interrupt; cr 14 bits in parm,
2467 machine check interrupt code in parm64 (note that
2468 machine checks needing further payload are not
2469 supported by this ioctl)
2471 Note that the vcpu ioctl is asynchronous to vcpu execution.
2473 4.78 KVM_PPC_GET_HTAB_FD
2475 Capability: KVM_CAP_PPC_HTAB_FD
2476 Architectures: powerpc
2478 Parameters: Pointer to struct kvm_get_htab_fd (in)
2479 Returns: file descriptor number (>= 0) on success, -1 on error
2481 This returns a file descriptor that can be used either to read out the
2482 entries in the guest's hashed page table (HPT), or to write entries to
2483 initialize the HPT. The returned fd can only be written to if the
2484 KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and
2485 can only be read if that bit is clear. The argument struct looks like
2488 /* For KVM_PPC_GET_HTAB_FD */
2489 struct kvm_get_htab_fd {
2495 /* Values for kvm_get_htab_fd.flags */
2496 #define KVM_GET_HTAB_BOLTED_ONLY ((__u64)0x1)
2497 #define KVM_GET_HTAB_WRITE ((__u64)0x2)
2499 The `start_index' field gives the index in the HPT of the entry at
2500 which to start reading. It is ignored when writing.
2502 Reads on the fd will initially supply information about all
2503 "interesting" HPT entries. Interesting entries are those with the
2504 bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise
2505 all entries. When the end of the HPT is reached, the read() will
2506 return. If read() is called again on the fd, it will start again from
2507 the beginning of the HPT, but will only return HPT entries that have
2508 changed since they were last read.
2510 Data read or written is structured as a header (8 bytes) followed by a
2511 series of valid HPT entries (16 bytes) each. The header indicates how
2512 many valid HPT entries there are and how many invalid entries follow
2513 the valid entries. The invalid entries are not represented explicitly
2514 in the stream. The header format is:
2516 struct kvm_get_htab_header {
2522 Writes to the fd create HPT entries starting at the index given in the
2523 header; first `n_valid' valid entries with contents from the data
2524 written, then `n_invalid' invalid entries, invalidating any previously
2525 valid entries found.
2527 4.79 KVM_CREATE_DEVICE
2529 Capability: KVM_CAP_DEVICE_CTRL
2531 Parameters: struct kvm_create_device (in/out)
2532 Returns: 0 on success, -1 on error
2534 ENODEV: The device type is unknown or unsupported
2535 EEXIST: Device already created, and this type of device may not
2536 be instantiated multiple times
2538 Other error conditions may be defined by individual device types or
2539 have their standard meanings.
2541 Creates an emulated device in the kernel. The file descriptor returned
2542 in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR.
2544 If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the
2545 device type is supported (not necessarily whether it can be created
2548 Individual devices should not define flags. Attributes should be used
2549 for specifying any behavior that is not implied by the device type
2552 struct kvm_create_device {
2553 __u32 type; /* in: KVM_DEV_TYPE_xxx */
2554 __u32 fd; /* out: device handle */
2555 __u32 flags; /* in: KVM_CREATE_DEVICE_xxx */
2558 4.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR
2560 Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
2561 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
2562 Type: device ioctl, vm ioctl, vcpu ioctl
2563 Parameters: struct kvm_device_attr
2564 Returns: 0 on success, -1 on error
2566 ENXIO: The group or attribute is unknown/unsupported for this device
2567 or hardware support is missing.
2568 EPERM: The attribute cannot (currently) be accessed this way
2569 (e.g. read-only attribute, or attribute that only makes
2570 sense when the device is in a different state)
2572 Other error conditions may be defined by individual device types.
2574 Gets/sets a specified piece of device configuration and/or state. The
2575 semantics are device-specific. See individual device documentation in
2576 the "devices" directory. As with ONE_REG, the size of the data
2577 transferred is defined by the particular attribute.
2579 struct kvm_device_attr {
2580 __u32 flags; /* no flags currently defined */
2581 __u32 group; /* device-defined */
2582 __u64 attr; /* group-defined */
2583 __u64 addr; /* userspace address of attr data */
2586 4.81 KVM_HAS_DEVICE_ATTR
2588 Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
2589 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
2590 Type: device ioctl, vm ioctl, vcpu ioctl
2591 Parameters: struct kvm_device_attr
2592 Returns: 0 on success, -1 on error
2594 ENXIO: The group or attribute is unknown/unsupported for this device
2595 or hardware support is missing.
2597 Tests whether a device supports a particular attribute. A successful
2598 return indicates the attribute is implemented. It does not necessarily
2599 indicate that the attribute can be read or written in the device's
2600 current state. "addr" is ignored.
2602 4.82 KVM_ARM_VCPU_INIT
2605 Architectures: arm, arm64
2607 Parameters: struct kvm_vcpu_init (in)
2608 Returns: 0 on success; -1 on error
2610 Â EINVAL: Â Â Â the target is unknown, or the combination of features is invalid.
2611 Â ENOENT: Â Â Â a features bit specified is unknown.
2613 This tells KVM what type of CPU to present to the guest, and what
2614 optional features it should have. Â This will cause a reset of the cpu
2615 registers to their initial values. Â If this is not called, KVM_RUN will
2616 return ENOEXEC for that vcpu.
2618 Note that because some registers reflect machine topology, all vcpus
2619 should be created before this ioctl is invoked.
2621 Userspace can call this function multiple times for a given vcpu, including
2622 after the vcpu has been run. This will reset the vcpu to its initial
2623 state. All calls to this function after the initial call must use the same
2624 target and same set of feature flags, otherwise EINVAL will be returned.
2627 - KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state.
2628 Depends on KVM_CAP_ARM_PSCI. If not set, the CPU will be powered on
2629 and execute guest code when KVM_RUN is called.
2630 - KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode.
2631 Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only).
2632 - KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 (or a future revision
2633 backward compatible with v0.2) for the CPU.
2634 Depends on KVM_CAP_ARM_PSCI_0_2.
2635 - KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU.
2636 Depends on KVM_CAP_ARM_PMU_V3.
2639 4.83 KVM_ARM_PREFERRED_TARGET
2642 Architectures: arm, arm64
2644 Parameters: struct struct kvm_vcpu_init (out)
2645 Returns: 0 on success; -1 on error
2647 ENODEV: no preferred target available for the host
2649 This queries KVM for preferred CPU target type which can be emulated
2650 by KVM on underlying host.
2652 The ioctl returns struct kvm_vcpu_init instance containing information
2653 about preferred CPU target type and recommended features for it. The
2654 kvm_vcpu_init->features bitmap returned will have feature bits set if
2655 the preferred target recommends setting these features, but this is
2658 The information returned by this ioctl can be used to prepare an instance
2659 of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in
2660 in VCPU matching underlying host.
2663 4.84 KVM_GET_REG_LIST
2666 Architectures: arm, arm64, mips
2668 Parameters: struct kvm_reg_list (in/out)
2669 Returns: 0 on success; -1 on error
2671 Â E2BIG: Â Â Â Â the reg index list is too big to fit in the array specified by
2672 Â Â Â Â Â Â Â Â Â Â Â Â the user (the number required will be written into n).
2674 struct kvm_reg_list {
2675 __u64 n; /* number of registers in reg[] */
2679 This ioctl returns the guest registers that are supported for the
2680 KVM_GET_ONE_REG/KVM_SET_ONE_REG calls.
2683 4.85 KVM_ARM_SET_DEVICE_ADDR (deprecated)
2685 Capability: KVM_CAP_ARM_SET_DEVICE_ADDR
2686 Architectures: arm, arm64
2688 Parameters: struct kvm_arm_device_address (in)
2689 Returns: 0 on success, -1 on error
2691 ENODEV: The device id is unknown
2692 ENXIO: Device not supported on current system
2693 EEXIST: Address already set
2694 E2BIG: Address outside guest physical address space
2695 EBUSY: Address overlaps with other device range
2697 struct kvm_arm_device_addr {
2702 Specify a device address in the guest's physical address space where guests
2703 can access emulated or directly exposed devices, which the host kernel needs
2704 to know about. The id field is an architecture specific identifier for a
2707 ARM/arm64 divides the id field into two parts, a device id and an
2708 address type id specific to the individual device.
2710 Â bits: | 63 ... 32 | 31 ... 16 | 15 ... 0 |
2711 field: | 0x00000000 | device id | addr type id |
2713 ARM/arm64 currently only require this when using the in-kernel GIC
2714 support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2
2715 as the device id. When setting the base address for the guest's
2716 mapping of the VGIC virtual CPU and distributor interface, the ioctl
2717 must be called after calling KVM_CREATE_IRQCHIP, but before calling
2718 KVM_RUN on any of the VCPUs. Calling this ioctl twice for any of the
2719 base addresses will return -EEXIST.
2721 Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API
2722 should be used instead.
2725 4.86 KVM_PPC_RTAS_DEFINE_TOKEN
2727 Capability: KVM_CAP_PPC_RTAS
2730 Parameters: struct kvm_rtas_token_args
2731 Returns: 0 on success, -1 on error
2733 Defines a token value for a RTAS (Run Time Abstraction Services)
2734 service in order to allow it to be handled in the kernel. The
2735 argument struct gives the name of the service, which must be the name
2736 of a service that has a kernel-side implementation. If the token
2737 value is non-zero, it will be associated with that service, and
2738 subsequent RTAS calls by the guest specifying that token will be
2739 handled by the kernel. If the token value is 0, then any token
2740 associated with the service will be forgotten, and subsequent RTAS
2741 calls by the guest for that service will be passed to userspace to be
2744 4.87 KVM_SET_GUEST_DEBUG
2746 Capability: KVM_CAP_SET_GUEST_DEBUG
2747 Architectures: x86, s390, ppc, arm64
2749 Parameters: struct kvm_guest_debug (in)
2750 Returns: 0 on success; -1 on error
2752 struct kvm_guest_debug {
2755 struct kvm_guest_debug_arch arch;
2758 Set up the processor specific debug registers and configure vcpu for
2759 handling guest debug events. There are two parts to the structure, the
2760 first a control bitfield indicates the type of debug events to handle
2761 when running. Common control bits are:
2763 - KVM_GUESTDBG_ENABLE: guest debugging is enabled
2764 - KVM_GUESTDBG_SINGLESTEP: the next run should single-step
2766 The top 16 bits of the control field are architecture specific control
2767 flags which can include the following:
2769 - KVM_GUESTDBG_USE_SW_BP: using software breakpoints [x86, arm64]
2770 - KVM_GUESTDBG_USE_HW_BP: using hardware breakpoints [x86, s390, arm64]
2771 - KVM_GUESTDBG_INJECT_DB: inject DB type exception [x86]
2772 - KVM_GUESTDBG_INJECT_BP: inject BP type exception [x86]
2773 - KVM_GUESTDBG_EXIT_PENDING: trigger an immediate guest exit [s390]
2775 For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints
2776 are enabled in memory so we need to ensure breakpoint exceptions are
2777 correctly trapped and the KVM run loop exits at the breakpoint and not
2778 running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP
2779 we need to ensure the guest vCPUs architecture specific registers are
2780 updated to the correct (supplied) values.
2782 The second part of the structure is architecture specific and
2783 typically contains a set of debug registers.
2785 For arm64 the number of debug registers is implementation defined and
2786 can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and
2787 KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number
2788 indicating the number of supported registers.
2790 When debug events exit the main run loop with the reason
2791 KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run
2792 structure containing architecture specific debug information.
2794 4.88 KVM_GET_EMULATED_CPUID
2796 Capability: KVM_CAP_EXT_EMUL_CPUID
2799 Parameters: struct kvm_cpuid2 (in/out)
2800 Returns: 0 on success, -1 on error
2805 struct kvm_cpuid_entry2 entries[0];
2808 The member 'flags' is used for passing flags from userspace.
2810 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
2811 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1)
2812 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2)
2814 struct kvm_cpuid_entry2 {
2825 This ioctl returns x86 cpuid features which are emulated by
2826 kvm.Userspace can use the information returned by this ioctl to query
2827 which features are emulated by kvm instead of being present natively.
2829 Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2
2830 structure with the 'nent' field indicating the number of entries in
2831 the variable-size array 'entries'. If the number of entries is too low
2832 to describe the cpu capabilities, an error (E2BIG) is returned. If the
2833 number is too high, the 'nent' field is adjusted and an error (ENOMEM)
2834 is returned. If the number is just right, the 'nent' field is adjusted
2835 to the number of valid entries in the 'entries' array, which is then
2838 The entries returned are the set CPUID bits of the respective features
2839 which kvm emulates, as returned by the CPUID instruction, with unknown
2840 or unsupported feature bits cleared.
2842 Features like x2apic, for example, may not be present in the host cpu
2843 but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be
2844 emulated efficiently and thus not included here.
2846 The fields in each entry are defined as follows:
2848 function: the eax value used to obtain the entry
2849 index: the ecx value used to obtain the entry (for entries that are
2851 flags: an OR of zero or more of the following:
2852 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
2853 if the index field is valid
2854 KVM_CPUID_FLAG_STATEFUL_FUNC:
2855 if cpuid for this function returns different values for successive
2856 invocations; there will be several entries with the same function,
2857 all with this flag set
2858 KVM_CPUID_FLAG_STATE_READ_NEXT:
2859 for KVM_CPUID_FLAG_STATEFUL_FUNC entries, set if this entry is
2860 the first entry to be read by a cpu
2861 eax, ebx, ecx, edx: the values returned by the cpuid instruction for
2862 this function/index combination
2864 4.89 KVM_S390_MEM_OP
2866 Capability: KVM_CAP_S390_MEM_OP
2869 Parameters: struct kvm_s390_mem_op (in)
2870 Returns: = 0 on success,
2871 < 0 on generic error (e.g. -EFAULT or -ENOMEM),
2872 > 0 if an exception occurred while walking the page tables
2874 Read or write data from/to the logical (virtual) memory of a VCPU.
2876 Parameters are specified via the following structure:
2878 struct kvm_s390_mem_op {
2879 __u64 gaddr; /* the guest address */
2880 __u64 flags; /* flags */
2881 __u32 size; /* amount of bytes */
2882 __u32 op; /* type of operation */
2883 __u64 buf; /* buffer in userspace */
2884 __u8 ar; /* the access register number */
2885 __u8 reserved[31]; /* should be set to 0 */
2888 The type of operation is specified in the "op" field. It is either
2889 KVM_S390_MEMOP_LOGICAL_READ for reading from logical memory space or
2890 KVM_S390_MEMOP_LOGICAL_WRITE for writing to logical memory space. The
2891 KVM_S390_MEMOP_F_CHECK_ONLY flag can be set in the "flags" field to check
2892 whether the corresponding memory access would create an access exception
2893 (without touching the data in the memory at the destination). In case an
2894 access exception occurred while walking the MMU tables of the guest, the
2895 ioctl returns a positive error number to indicate the type of exception.
2896 This exception is also raised directly at the corresponding VCPU if the
2897 flag KVM_S390_MEMOP_F_INJECT_EXCEPTION is set in the "flags" field.
2899 The start address of the memory region has to be specified in the "gaddr"
2900 field, and the length of the region in the "size" field. "buf" is the buffer
2901 supplied by the userspace application where the read data should be written
2902 to for KVM_S390_MEMOP_LOGICAL_READ, or where the data that should be written
2903 is stored for a KVM_S390_MEMOP_LOGICAL_WRITE. "buf" is unused and can be NULL
2904 when KVM_S390_MEMOP_F_CHECK_ONLY is specified. "ar" designates the access
2905 register number to be used.
2907 The "reserved" field is meant for future extensions. It is not used by
2908 KVM with the currently defined set of flags.
2910 4.90 KVM_S390_GET_SKEYS
2912 Capability: KVM_CAP_S390_SKEYS
2915 Parameters: struct kvm_s390_skeys
2916 Returns: 0 on success, KVM_S390_GET_KEYS_NONE if guest is not using storage
2917 keys, negative value on error
2919 This ioctl is used to get guest storage key values on the s390
2920 architecture. The ioctl takes parameters via the kvm_s390_skeys struct.
2922 struct kvm_s390_skeys {
2925 __u64 skeydata_addr;
2930 The start_gfn field is the number of the first guest frame whose storage keys
2933 The count field is the number of consecutive frames (starting from start_gfn)
2934 whose storage keys to get. The count field must be at least 1 and the maximum
2935 allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range
2936 will cause the ioctl to return -EINVAL.
2938 The skeydata_addr field is the address to a buffer large enough to hold count
2939 bytes. This buffer will be filled with storage key data by the ioctl.
2941 4.91 KVM_S390_SET_SKEYS
2943 Capability: KVM_CAP_S390_SKEYS
2946 Parameters: struct kvm_s390_skeys
2947 Returns: 0 on success, negative value on error
2949 This ioctl is used to set guest storage key values on the s390
2950 architecture. The ioctl takes parameters via the kvm_s390_skeys struct.
2951 See section on KVM_S390_GET_SKEYS for struct definition.
2953 The start_gfn field is the number of the first guest frame whose storage keys
2956 The count field is the number of consecutive frames (starting from start_gfn)
2957 whose storage keys to get. The count field must be at least 1 and the maximum
2958 allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range
2959 will cause the ioctl to return -EINVAL.
2961 The skeydata_addr field is the address to a buffer containing count bytes of
2962 storage keys. Each byte in the buffer will be set as the storage key for a
2963 single frame starting at start_gfn for count frames.
2965 Note: If any architecturally invalid key value is found in the given data then
2966 the ioctl will return -EINVAL.
2970 Capability: KVM_CAP_S390_INJECT_IRQ
2973 Parameters: struct kvm_s390_irq (in)
2974 Returns: 0 on success, -1 on error
2976 EINVAL: interrupt type is invalid
2977 type is KVM_S390_SIGP_STOP and flag parameter is invalid value
2978 type is KVM_S390_INT_EXTERNAL_CALL and code is bigger
2979 than the maximum of VCPUs
2980 EBUSY: type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped
2981 type is KVM_S390_SIGP_STOP and a stop irq is already pending
2982 type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt
2985 Allows to inject an interrupt to the guest.
2987 Using struct kvm_s390_irq as a parameter allows
2988 to inject additional payload which is not
2989 possible via KVM_S390_INTERRUPT.
2991 Interrupt parameters are passed via kvm_s390_irq:
2993 struct kvm_s390_irq {
2996 struct kvm_s390_io_info io;
2997 struct kvm_s390_ext_info ext;
2998 struct kvm_s390_pgm_info pgm;
2999 struct kvm_s390_emerg_info emerg;
3000 struct kvm_s390_extcall_info extcall;
3001 struct kvm_s390_prefix_info prefix;
3002 struct kvm_s390_stop_info stop;
3003 struct kvm_s390_mchk_info mchk;
3008 type can be one of the following:
3010 KVM_S390_SIGP_STOP - sigp stop; parameter in .stop
3011 KVM_S390_PROGRAM_INT - program check; parameters in .pgm
3012 KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix
3013 KVM_S390_RESTART - restart; no parameters
3014 KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters
3015 KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters
3016 KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg
3017 KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall
3018 KVM_S390_MCHK - machine check interrupt; parameters in .mchk
3021 Note that the vcpu ioctl is asynchronous to vcpu execution.
3023 4.94 KVM_S390_GET_IRQ_STATE
3025 Capability: KVM_CAP_S390_IRQ_STATE
3028 Parameters: struct kvm_s390_irq_state (out)
3029 Returns: >= number of bytes copied into buffer,
3030 -EINVAL if buffer size is 0,
3031 -ENOBUFS if buffer size is too small to fit all pending interrupts,
3032 -EFAULT if the buffer address was invalid
3034 This ioctl allows userspace to retrieve the complete state of all currently
3035 pending interrupts in a single buffer. Use cases include migration
3036 and introspection. The parameter structure contains the address of a
3037 userspace buffer and its length:
3039 struct kvm_s390_irq_state {
3041 __u32 flags; /* will stay unused for compatibility reasons */
3043 __u32 reserved[4]; /* will stay unused for compatibility reasons */
3046 Userspace passes in the above struct and for each pending interrupt a
3047 struct kvm_s390_irq is copied to the provided buffer.
3049 The structure contains a flags and a reserved field for future extensions. As
3050 the kernel never checked for flags == 0 and QEMU never pre-zeroed flags and
3051 reserved, these fields can not be used in the future without breaking
3054 If -ENOBUFS is returned the buffer provided was too small and userspace
3055 may retry with a bigger buffer.
3057 4.95 KVM_S390_SET_IRQ_STATE
3059 Capability: KVM_CAP_S390_IRQ_STATE
3062 Parameters: struct kvm_s390_irq_state (in)
3063 Returns: 0 on success,
3064 -EFAULT if the buffer address was invalid,
3065 -EINVAL for an invalid buffer length (see below),
3066 -EBUSY if there were already interrupts pending,
3067 errors occurring when actually injecting the
3068 interrupt. See KVM_S390_IRQ.
3070 This ioctl allows userspace to set the complete state of all cpu-local
3071 interrupts currently pending for the vcpu. It is intended for restoring
3072 interrupt state after a migration. The input parameter is a userspace buffer
3073 containing a struct kvm_s390_irq_state:
3075 struct kvm_s390_irq_state {
3077 __u32 flags; /* will stay unused for compatibility reasons */
3079 __u32 reserved[4]; /* will stay unused for compatibility reasons */
3082 The restrictions for flags and reserved apply as well.
3083 (see KVM_S390_GET_IRQ_STATE)
3085 The userspace memory referenced by buf contains a struct kvm_s390_irq
3086 for each interrupt to be injected into the guest.
3087 If one of the interrupts could not be injected for some reason the
3090 len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0
3091 and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq),
3092 which is the maximum number of possibly pending cpu-local interrupts.
3096 Capability: KVM_CAP_X86_SMM
3100 Returns: 0 on success, -1 on error
3102 Queues an SMI on the thread's vcpu.
3104 4.97 KVM_CAP_PPC_MULTITCE
3106 Capability: KVM_CAP_PPC_MULTITCE
3110 This capability means the kernel is capable of handling hypercalls
3111 H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user
3112 space. This significantly accelerates DMA operations for PPC KVM guests.
3113 User space should expect that its handlers for these hypercalls
3114 are not going to be called if user space previously registered LIOBN
3115 in KVM (via KVM_CREATE_SPAPR_TCE or similar calls).
3117 In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest,
3118 user space might have to advertise it for the guest. For example,
3119 IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is
3120 present in the "ibm,hypertas-functions" device-tree property.
3122 The hypercalls mentioned above may or may not be processed successfully
3123 in the kernel based fast path. If they can not be handled by the kernel,
3124 they will get passed on to user space. So user space still has to have
3125 an implementation for these despite the in kernel acceleration.
3127 This capability is always enabled.
3129 4.98 KVM_CREATE_SPAPR_TCE_64
3131 Capability: KVM_CAP_SPAPR_TCE_64
3132 Architectures: powerpc
3134 Parameters: struct kvm_create_spapr_tce_64 (in)
3135 Returns: file descriptor for manipulating the created TCE table
3137 This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit
3138 windows, described in 4.62 KVM_CREATE_SPAPR_TCE
3140 This capability uses extended struct in ioctl interface:
3142 /* for KVM_CAP_SPAPR_TCE_64 */
3143 struct kvm_create_spapr_tce_64 {
3147 __u64 offset; /* in pages */
3148 __u64 size; /* in pages */
3151 The aim of extension is to support an additional bigger DMA window with
3152 a variable page size.
3153 KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and
3154 a bus offset of the corresponding DMA window, @size and @offset are numbers
3157 @flags are not used at the moment.
3159 The rest of functionality is identical to KVM_CREATE_SPAPR_TCE.
3161 4.99 KVM_REINJECT_CONTROL
3163 Capability: KVM_CAP_REINJECT_CONTROL
3166 Parameters: struct kvm_reinject_control (in)
3167 Returns: 0 on success,
3168 -EFAULT if struct kvm_reinject_control cannot be read,
3169 -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier.
3171 i8254 (PIT) has two modes, reinject and !reinject. The default is reinject,
3172 where KVM queues elapsed i8254 ticks and monitors completion of interrupt from
3173 vector(s) that i8254 injects. Reinject mode dequeues a tick and injects its
3174 interrupt whenever there isn't a pending interrupt from i8254.
3175 !reinject mode injects an interrupt as soon as a tick arrives.
3177 struct kvm_reinject_control {
3182 pit_reinject = 0 (!reinject mode) is recommended, unless running an old
3183 operating system that uses the PIT for timing (e.g. Linux 2.4.x).
3185 4.100 KVM_PPC_CONFIGURE_V3_MMU
3187 Capability: KVM_CAP_PPC_RADIX_MMU or KVM_CAP_PPC_HASH_MMU_V3
3190 Parameters: struct kvm_ppc_mmuv3_cfg (in)
3191 Returns: 0 on success,
3192 -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read,
3193 -EINVAL if the configuration is invalid
3195 This ioctl controls whether the guest will use radix or HPT (hashed
3196 page table) translation, and sets the pointer to the process table for
3199 struct kvm_ppc_mmuv3_cfg {
3201 __u64 process_table;
3204 There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and
3205 KVM_PPC_MMUV3_GTSE. KVM_PPC_MMUV3_RADIX, if set, configures the guest
3206 to use radix tree translation, and if clear, to use HPT translation.
3207 KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest
3208 to be able to use the global TLB and SLB invalidation instructions;
3209 if clear, the guest may not use these instructions.
3211 The process_table field specifies the address and size of the guest
3212 process table, which is in the guest's space. This field is formatted
3213 as the second doubleword of the partition table entry, as defined in
3214 the Power ISA V3.00, Book III section 5.7.6.1.
3216 4.101 KVM_PPC_GET_RMMU_INFO
3218 Capability: KVM_CAP_PPC_RADIX_MMU
3221 Parameters: struct kvm_ppc_rmmu_info (out)
3222 Returns: 0 on success,
3223 -EFAULT if struct kvm_ppc_rmmu_info cannot be written,
3224 -EINVAL if no useful information can be returned
3226 This ioctl returns a structure containing two things: (a) a list
3227 containing supported radix tree geometries, and (b) a list that maps
3228 page sizes to put in the "AP" (actual page size) field for the tlbie
3229 (TLB invalidate entry) instruction.
3231 struct kvm_ppc_rmmu_info {
3232 struct kvm_ppc_radix_geom {
3237 __u32 ap_encodings[8];
3240 The geometries[] field gives up to 8 supported geometries for the
3241 radix page table, in terms of the log base 2 of the smallest page
3242 size, and the number of bits indexed at each level of the tree, from
3243 the PTE level up to the PGD level in that order. Any unused entries
3244 will have 0 in the page_shift field.
3246 The ap_encodings gives the supported page sizes and their AP field
3247 encodings, encoded with the AP value in the top 3 bits and the log
3248 base 2 of the page size in the bottom 6 bits.
3250 4.102 KVM_PPC_RESIZE_HPT_PREPARE
3252 Capability: KVM_CAP_SPAPR_RESIZE_HPT
3253 Architectures: powerpc
3255 Parameters: struct kvm_ppc_resize_hpt (in)
3256 Returns: 0 on successful completion,
3257 >0 if a new HPT is being prepared, the value is an estimated
3258 number of milliseconds until preparation is complete
3259 -EFAULT if struct kvm_reinject_control cannot be read,
3260 -EINVAL if the supplied shift or flags are invalid
3261 -ENOMEM if unable to allocate the new HPT
3262 -ENOSPC if there was a hash collision when moving existing
3263 HPT entries to the new HPT
3264 -EIO on other error conditions
3266 Used to implement the PAPR extension for runtime resizing of a guest's
3267 Hashed Page Table (HPT). Specifically this starts, stops or monitors
3268 the preparation of a new potential HPT for the guest, essentially
3269 implementing the H_RESIZE_HPT_PREPARE hypercall.
3271 If called with shift > 0 when there is no pending HPT for the guest,
3272 this begins preparation of a new pending HPT of size 2^(shift) bytes.
3273 It then returns a positive integer with the estimated number of
3274 milliseconds until preparation is complete.
3276 If called when there is a pending HPT whose size does not match that
3277 requested in the parameters, discards the existing pending HPT and
3278 creates a new one as above.
3280 If called when there is a pending HPT of the size requested, will:
3281 * If preparation of the pending HPT is already complete, return 0
3282 * If preparation of the pending HPT has failed, return an error
3283 code, then discard the pending HPT.
3284 * If preparation of the pending HPT is still in progress, return an
3285 estimated number of milliseconds until preparation is complete.
3287 If called with shift == 0, discards any currently pending HPT and
3288 returns 0 (i.e. cancels any in-progress preparation).
3290 flags is reserved for future expansion, currently setting any bits in
3291 flags will result in an -EINVAL.
3293 Normally this will be called repeatedly with the same parameters until
3294 it returns <= 0. The first call will initiate preparation, subsequent
3295 ones will monitor preparation until it completes or fails.
3297 struct kvm_ppc_resize_hpt {
3303 4.103 KVM_PPC_RESIZE_HPT_COMMIT
3305 Capability: KVM_CAP_SPAPR_RESIZE_HPT
3306 Architectures: powerpc
3308 Parameters: struct kvm_ppc_resize_hpt (in)
3309 Returns: 0 on successful completion,
3310 -EFAULT if struct kvm_reinject_control cannot be read,
3311 -EINVAL if the supplied shift or flags are invalid
3312 -ENXIO is there is no pending HPT, or the pending HPT doesn't
3313 have the requested size
3314 -EBUSY if the pending HPT is not fully prepared
3315 -ENOSPC if there was a hash collision when moving existing
3316 HPT entries to the new HPT
3317 -EIO on other error conditions
3319 Used to implement the PAPR extension for runtime resizing of a guest's
3320 Hashed Page Table (HPT). Specifically this requests that the guest be
3321 transferred to working with the new HPT, essentially implementing the
3322 H_RESIZE_HPT_COMMIT hypercall.
3324 This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has
3325 returned 0 with the same parameters. In other cases
3326 KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or
3327 -EBUSY, though others may be possible if the preparation was started,
3330 This will have undefined effects on the guest if it has not already
3331 placed itself in a quiescent state where no vcpu will make MMU enabled
3334 On succsful completion, the pending HPT will become the guest's active
3335 HPT and the previous HPT will be discarded.
3337 On failure, the guest will still be operating on its previous HPT.
3339 struct kvm_ppc_resize_hpt {
3345 4.104 KVM_X86_GET_MCE_CAP_SUPPORTED
3347 Capability: KVM_CAP_MCE
3350 Parameters: u64 mce_cap (out)
3351 Returns: 0 on success, -1 on error
3353 Returns supported MCE capabilities. The u64 mce_cap parameter
3354 has the same format as the MSR_IA32_MCG_CAP register. Supported
3355 capabilities will have the corresponding bits set.
3357 4.105 KVM_X86_SETUP_MCE
3359 Capability: KVM_CAP_MCE
3362 Parameters: u64 mcg_cap (in)
3363 Returns: 0 on success,
3364 -EFAULT if u64 mcg_cap cannot be read,
3365 -EINVAL if the requested number of banks is invalid,
3366 -EINVAL if requested MCE capability is not supported.
3368 Initializes MCE support for use. The u64 mcg_cap parameter
3369 has the same format as the MSR_IA32_MCG_CAP register and
3370 specifies which capabilities should be enabled. The maximum
3371 supported number of error-reporting banks can be retrieved when
3372 checking for KVM_CAP_MCE. The supported capabilities can be
3373 retrieved with KVM_X86_GET_MCE_CAP_SUPPORTED.
3375 4.106 KVM_X86_SET_MCE
3377 Capability: KVM_CAP_MCE
3380 Parameters: struct kvm_x86_mce (in)
3381 Returns: 0 on success,
3382 -EFAULT if struct kvm_x86_mce cannot be read,
3383 -EINVAL if the bank number is invalid,
3384 -EINVAL if VAL bit is not set in status field.
3386 Inject a machine check error (MCE) into the guest. The input
3389 struct kvm_x86_mce {
3399 If the MCE being reported is an uncorrected error, KVM will
3400 inject it as an MCE exception into the guest. If the guest
3401 MCG_STATUS register reports that an MCE is in progress, KVM
3402 causes an KVM_EXIT_SHUTDOWN vmexit.
3404 Otherwise, if the MCE is a corrected error, KVM will just
3405 store it in the corresponding bank (provided this bank is
3406 not holding a previously reported uncorrected error).
3408 4.107 KVM_S390_GET_CMMA_BITS
3410 Capability: KVM_CAP_S390_CMMA_MIGRATION
3413 Parameters: struct kvm_s390_cmma_log (in, out)
3414 Returns: 0 on success, a negative value on error
3416 This ioctl is used to get the values of the CMMA bits on the s390
3417 architecture. It is meant to be used in two scenarios:
3418 - During live migration to save the CMMA values. Live migration needs
3419 to be enabled via the KVM_REQ_START_MIGRATION VM property.
3420 - To non-destructively peek at the CMMA values, with the flag
3421 KVM_S390_CMMA_PEEK set.
3423 The ioctl takes parameters via the kvm_s390_cmma_log struct. The desired
3424 values are written to a buffer whose location is indicated via the "values"
3425 member in the kvm_s390_cmma_log struct. The values in the input struct are
3426 also updated as needed.
3427 Each CMMA value takes up one byte.
3429 struct kvm_s390_cmma_log {
3440 start_gfn is the number of the first guest frame whose CMMA values are
3443 count is the length of the buffer in bytes,
3445 values points to the buffer where the result will be written to.
3447 If count is greater than KVM_S390_SKEYS_MAX, then it is considered to be
3448 KVM_S390_SKEYS_MAX. KVM_S390_SKEYS_MAX is re-used for consistency with
3451 The result is written in the buffer pointed to by the field values, and
3452 the values of the input parameter are updated as follows.
3454 Depending on the flags, different actions are performed. The only
3455 supported flag so far is KVM_S390_CMMA_PEEK.
3457 The default behaviour if KVM_S390_CMMA_PEEK is not set is:
3458 start_gfn will indicate the first page frame whose CMMA bits were dirty.
3459 It is not necessarily the same as the one passed as input, as clean pages
3462 count will indicate the number of bytes actually written in the buffer.
3463 It can (and very often will) be smaller than the input value, since the
3464 buffer is only filled until 16 bytes of clean values are found (which
3465 are then not copied in the buffer). Since a CMMA migration block needs
3466 the base address and the length, for a total of 16 bytes, we will send
3467 back some clean data if there is some dirty data afterwards, as long as
3468 the size of the clean data does not exceed the size of the header. This
3469 allows to minimize the amount of data to be saved or transferred over
3470 the network at the expense of more roundtrips to userspace. The next
3471 invocation of the ioctl will skip over all the clean values, saving
3472 potentially more than just the 16 bytes we found.
3474 If KVM_S390_CMMA_PEEK is set:
3475 the existing storage attributes are read even when not in migration
3476 mode, and no other action is performed;
3478 the output start_gfn will be equal to the input start_gfn,
3480 the output count will be equal to the input count, except if the end of
3481 memory has been reached.
3484 the field "remaining" will indicate the total number of dirty CMMA values
3485 still remaining, or 0 if KVM_S390_CMMA_PEEK is set and migration mode is
3490 values points to the userspace buffer where the result will be stored.
3492 This ioctl can fail with -ENOMEM if not enough memory can be allocated to
3493 complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
3494 KVM_S390_CMMA_PEEK is not set but migration mode was not enabled, with
3495 -EFAULT if the userspace address is invalid or if no page table is
3496 present for the addresses (e.g. when using hugepages).
3498 4.108 KVM_S390_SET_CMMA_BITS
3500 Capability: KVM_CAP_S390_CMMA_MIGRATION
3503 Parameters: struct kvm_s390_cmma_log (in)
3504 Returns: 0 on success, a negative value on error
3506 This ioctl is used to set the values of the CMMA bits on the s390
3507 architecture. It is meant to be used during live migration to restore
3508 the CMMA values, but there are no restrictions on its use.
3509 The ioctl takes parameters via the kvm_s390_cmma_values struct.
3510 Each CMMA value takes up one byte.
3512 struct kvm_s390_cmma_log {
3523 start_gfn indicates the starting guest frame number,
3525 count indicates how many values are to be considered in the buffer,
3527 flags is not used and must be 0.
3529 mask indicates which PGSTE bits are to be considered.
3531 remaining is not used.
3533 values points to the buffer in userspace where to store the values.
3535 This ioctl can fail with -ENOMEM if not enough memory can be allocated to
3536 complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
3537 the count field is too large (e.g. more than KVM_S390_CMMA_SIZE_MAX) or
3538 if the flags field was not 0, with -EFAULT if the userspace address is
3539 invalid, if invalid pages are written to (e.g. after the end of memory)
3540 or if no page table is present for the addresses (e.g. when using
3543 4.109 KVM_PPC_GET_CPU_CHAR
3545 Capability: KVM_CAP_PPC_GET_CPU_CHAR
3546 Architectures: powerpc
3548 Parameters: struct kvm_ppc_cpu_char (out)
3549 Returns: 0 on successful completion
3550 -EFAULT if struct kvm_ppc_cpu_char cannot be written
3552 This ioctl gives userspace information about certain characteristics
3553 of the CPU relating to speculative execution of instructions and
3554 possible information leakage resulting from speculative execution (see
3555 CVE-2017-5715, CVE-2017-5753 and CVE-2017-5754). The information is
3556 returned in struct kvm_ppc_cpu_char, which looks like this:
3558 struct kvm_ppc_cpu_char {
3559 __u64 character; /* characteristics of the CPU */
3560 __u64 behaviour; /* recommended software behaviour */
3561 __u64 character_mask; /* valid bits in character */
3562 __u64 behaviour_mask; /* valid bits in behaviour */
3565 For extensibility, the character_mask and behaviour_mask fields
3566 indicate which bits of character and behaviour have been filled in by
3567 the kernel. If the set of defined bits is extended in future then
3568 userspace will be able to tell whether it is running on a kernel that
3569 knows about the new bits.
3571 The character field describes attributes of the CPU which can help
3572 with preventing inadvertent information disclosure - specifically,
3573 whether there is an instruction to flash-invalidate the L1 data cache
3574 (ori 30,30,0 or mtspr SPRN_TRIG2,rN), whether the L1 data cache is set
3575 to a mode where entries can only be used by the thread that created
3576 them, whether the bcctr[l] instruction prevents speculation, and
3577 whether a speculation barrier instruction (ori 31,31,0) is provided.
3579 The behaviour field describes actions that software should take to
3580 prevent inadvertent information disclosure, and thus describes which
3581 vulnerabilities the hardware is subject to; specifically whether the
3582 L1 data cache should be flushed when returning to user mode from the
3583 kernel, and whether a speculation barrier should be placed between an
3584 array bounds check and the array access.
3586 These fields use the same bit definitions as the new
3587 H_GET_CPU_CHARACTERISTICS hypercall.
3589 4.110 KVM_MEMORY_ENCRYPT_OP
3594 Parameters: an opaque platform specific structure (in/out)
3595 Returns: 0 on success; -1 on error
3597 If the platform supports creating encrypted VMs then this ioctl can be used
3598 for issuing platform-specific memory encryption commands to manage those
3601 Currently, this ioctl is used for issuing Secure Encrypted Virtualization
3602 (SEV) commands on AMD Processors. The SEV commands are defined in
3603 Documentation/virtual/kvm/amd-memory-encryption.rst.
3605 4.111 KVM_MEMORY_ENCRYPT_REG_REGION
3610 Parameters: struct kvm_enc_region (in)
3611 Returns: 0 on success; -1 on error
3613 This ioctl can be used to register a guest memory region which may
3614 contain encrypted data (e.g. guest RAM, SMRAM etc).
3616 It is used in the SEV-enabled guest. When encryption is enabled, a guest
3617 memory region may contain encrypted data. The SEV memory encryption
3618 engine uses a tweak such that two identical plaintext pages, each at
3619 different locations will have differing ciphertexts. So swapping or
3620 moving ciphertext of those pages will not result in plaintext being
3621 swapped. So relocating (or migrating) physical backing pages for the SEV
3622 guest will require some additional steps.
3624 Note: The current SEV key management spec does not provide commands to
3625 swap or migrate (move) ciphertext pages. Hence, for now we pin the guest
3626 memory region registered with the ioctl.
3628 4.112 KVM_MEMORY_ENCRYPT_UNREG_REGION
3633 Parameters: struct kvm_enc_region (in)
3634 Returns: 0 on success; -1 on error
3636 This ioctl can be used to unregister the guest memory region registered
3637 with KVM_MEMORY_ENCRYPT_REG_REGION ioctl above.
3639 4.113 KVM_HYPERV_EVENTFD
3641 Capability: KVM_CAP_HYPERV_EVENTFD
3644 Parameters: struct kvm_hyperv_eventfd (in)
3646 This ioctl (un)registers an eventfd to receive notifications from the guest on
3647 the specified Hyper-V connection id through the SIGNAL_EVENT hypercall, without
3648 causing a user exit. SIGNAL_EVENT hypercall with non-zero event flag number
3649 (bits 24-31) still triggers a KVM_EXIT_HYPERV_HCALL user exit.
3651 struct kvm_hyperv_eventfd {
3658 The conn_id field should fit within 24 bits:
3660 #define KVM_HYPERV_CONN_ID_MASK 0x00ffffff
3662 The acceptable values for the flags field are:
3664 #define KVM_HYPERV_EVENTFD_DEASSIGN (1 << 0)
3666 Returns: 0 on success,
3667 -EINVAL if conn_id or flags is outside the allowed range
3668 -ENOENT on deassign if the conn_id isn't registered
3669 -EEXIST on assign if the conn_id is already registered
3671 4.114 KVM_GET_NESTED_STATE
3673 Capability: KVM_CAP_NESTED_STATE
3676 Parameters: struct kvm_nested_state (in/out)
3677 Returns: 0 on success, -1 on error
3679 E2BIG: the total state size (including the fixed-size part of struct
3680 kvm_nested_state) exceeds the value of 'size' specified by
3681 the user; the size required will be written into size.
3683 struct kvm_nested_state {
3688 struct kvm_vmx_nested_state vmx;
3689 struct kvm_svm_nested_state svm;
3695 #define KVM_STATE_NESTED_GUEST_MODE 0x00000001
3696 #define KVM_STATE_NESTED_RUN_PENDING 0x00000002
3698 #define KVM_STATE_NESTED_SMM_GUEST_MODE 0x00000001
3699 #define KVM_STATE_NESTED_SMM_VMXON 0x00000002
3701 struct kvm_vmx_nested_state {
3710 This ioctl copies the vcpu's nested virtualization state from the kernel to
3713 The maximum size of the state, including the fixed-size part of struct
3714 kvm_nested_state, can be retrieved by passing KVM_CAP_NESTED_STATE to
3715 the KVM_CHECK_EXTENSION ioctl().
3717 4.115 KVM_SET_NESTED_STATE
3719 Capability: KVM_CAP_NESTED_STATE
3722 Parameters: struct kvm_nested_state (in)
3723 Returns: 0 on success, -1 on error
3725 This copies the vcpu's kvm_nested_state struct from userspace to the kernel. For
3726 the definition of struct kvm_nested_state, see KVM_GET_NESTED_STATE.
3728 4.116 KVM_(UN)REGISTER_COALESCED_MMIO
3730 Capability: KVM_CAP_COALESCED_MMIO (for coalesced mmio)
3731 KVM_CAP_COALESCED_PIO (for coalesced pio)
3734 Parameters: struct kvm_coalesced_mmio_zone
3735 Returns: 0 on success, < 0 on error
3737 Coalesced I/O is a performance optimization that defers hardware
3738 register write emulation so that userspace exits are avoided. It is
3739 typically used to reduce the overhead of emulating frequently accessed
3742 When a hardware register is configured for coalesced I/O, write accesses
3743 do not exit to userspace and their value is recorded in a ring buffer
3744 that is shared between kernel and userspace.
3746 Coalesced I/O is used if one or more write accesses to a hardware
3747 register can be deferred until a read or a write to another hardware
3748 register on the same device. This last access will cause a vmexit and
3749 userspace will process accesses from the ring buffer before emulating
3750 it. That will avoid exiting to userspace on repeated writes.
3752 Coalesced pio is based on coalesced mmio. There is little difference
3753 between coalesced mmio and pio except that coalesced pio records accesses
3756 5. The kvm_run structure
3757 ------------------------
3759 Application code obtains a pointer to the kvm_run structure by
3760 mmap()ing a vcpu fd. From that point, application code can control
3761 execution by changing fields in kvm_run prior to calling the KVM_RUN
3762 ioctl, and obtain information about the reason KVM_RUN returned by
3763 looking up structure members.
3767 __u8 request_interrupt_window;
3769 Request that KVM_RUN return when it becomes possible to inject external
3770 interrupts into the guest. Useful in conjunction with KVM_INTERRUPT.
3772 __u8 immediate_exit;
3774 This field is polled once when KVM_RUN starts; if non-zero, KVM_RUN
3775 exits immediately, returning -EINTR. In the common scenario where a
3776 signal is used to "kick" a VCPU out of KVM_RUN, this field can be used
3777 to avoid usage of KVM_SET_SIGNAL_MASK, which has worse scalability.
3778 Rather than blocking the signal outside KVM_RUN, userspace can set up
3779 a signal handler that sets run->immediate_exit to a non-zero value.
3781 This field is ignored if KVM_CAP_IMMEDIATE_EXIT is not available.
3788 When KVM_RUN has returned successfully (return value 0), this informs
3789 application code why KVM_RUN has returned. Allowable values for this
3790 field are detailed below.
3792 __u8 ready_for_interrupt_injection;
3794 If request_interrupt_window has been specified, this field indicates
3795 an interrupt can be injected now with KVM_INTERRUPT.
3799 The value of the current interrupt flag. Only valid if in-kernel
3800 local APIC is not used.
3804 More architecture-specific flags detailing state of the VCPU that may
3805 affect the device's behavior. The only currently defined flag is
3806 KVM_RUN_X86_SMM, which is valid on x86 machines and is set if the
3807 VCPU is in system management mode.
3809 /* in (pre_kvm_run), out (post_kvm_run) */
3812 The value of the cr8 register. Only valid if in-kernel local APIC is
3813 not used. Both input and output.
3817 The value of the APIC BASE msr. Only valid if in-kernel local
3818 APIC is not used. Both input and output.
3821 /* KVM_EXIT_UNKNOWN */
3823 __u64 hardware_exit_reason;
3826 If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown
3827 reasons. Further architecture-specific information is available in
3828 hardware_exit_reason.
3830 /* KVM_EXIT_FAIL_ENTRY */
3832 __u64 hardware_entry_failure_reason;
3835 If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due
3836 to unknown reasons. Further architecture-specific information is
3837 available in hardware_entry_failure_reason.
3839 /* KVM_EXIT_EXCEPTION */
3849 #define KVM_EXIT_IO_IN 0
3850 #define KVM_EXIT_IO_OUT 1
3852 __u8 size; /* bytes */
3855 __u64 data_offset; /* relative to kvm_run start */
3858 If exit_reason is KVM_EXIT_IO, then the vcpu has
3859 executed a port I/O instruction which could not be satisfied by kvm.
3860 data_offset describes where the data is located (KVM_EXIT_IO_OUT) or
3861 where kvm expects application code to place the data for the next
3862 KVM_RUN invocation (KVM_EXIT_IO_IN). Data format is a packed array.
3864 /* KVM_EXIT_DEBUG */
3866 struct kvm_debug_exit_arch arch;
3869 If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event
3870 for which architecture specific information is returned.
3880 If exit_reason is KVM_EXIT_MMIO, then the vcpu has
3881 executed a memory-mapped I/O instruction which could not be satisfied
3882 by kvm. The 'data' member contains the written data if 'is_write' is
3883 true, and should be filled by application code otherwise.
3885 The 'data' member contains, in its first 'len' bytes, the value as it would
3886 appear if the VCPU performed a load or store of the appropriate width directly
3889 NOTE: For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR and
3890 KVM_EXIT_EPR the corresponding
3891 operations are complete (and guest state is consistent) only after userspace
3892 has re-entered the kernel with KVM_RUN. The kernel side will first finish
3893 incomplete operations and then check for pending signals. Userspace
3894 can re-enter the guest with an unmasked signal pending to complete
3897 /* KVM_EXIT_HYPERCALL */
3906 Unused. This was once used for 'hypercall to userspace'. To implement
3907 such functionality, use KVM_EXIT_IO (x86) or KVM_EXIT_MMIO (all except s390).
3908 Note KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO.
3910 /* KVM_EXIT_TPR_ACCESS */
3917 To be documented (KVM_TPR_ACCESS_REPORTING).
3919 /* KVM_EXIT_S390_SIEIC */
3922 __u64 mask; /* psw upper half */
3923 __u64 addr; /* psw lower half */
3930 /* KVM_EXIT_S390_RESET */
3931 #define KVM_S390_RESET_POR 1
3932 #define KVM_S390_RESET_CLEAR 2
3933 #define KVM_S390_RESET_SUBSYSTEM 4
3934 #define KVM_S390_RESET_CPU_INIT 8
3935 #define KVM_S390_RESET_IPL 16
3936 __u64 s390_reset_flags;
3940 /* KVM_EXIT_S390_UCONTROL */
3942 __u64 trans_exc_code;
3946 s390 specific. A page fault has occurred for a user controlled virtual
3947 machine (KVM_VM_S390_UNCONTROL) on it's host page table that cannot be
3948 resolved by the kernel.
3949 The program code and the translation exception code that were placed
3950 in the cpu's lowcore are presented here as defined by the z Architecture
3951 Principles of Operation Book in the Chapter for Dynamic Address Translation
3961 Deprecated - was used for 440 KVM.
3968 MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch
3969 hypercalls and exit with this exit struct that contains all the guest gprs.
3971 If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall.
3972 Userspace can now handle the hypercall and when it's done modify the gprs as
3973 necessary. Upon guest entry all guest GPRs will then be replaced by the values
3976 /* KVM_EXIT_PAPR_HCALL */
3983 This is used on 64-bit PowerPC when emulating a pSeries partition,
3984 e.g. with the 'pseries' machine type in qemu. It occurs when the
3985 guest does a hypercall using the 'sc 1' instruction. The 'nr' field
3986 contains the hypercall number (from the guest R3), and 'args' contains
3987 the arguments (from the guest R4 - R12). Userspace should put the
3988 return code in 'ret' and any extra returned values in args[].
3989 The possible hypercalls are defined in the Power Architecture Platform
3990 Requirements (PAPR) document available from www.power.org (free
3991 developer registration required to access it).
3993 /* KVM_EXIT_S390_TSCH */
3995 __u16 subchannel_id;
3996 __u16 subchannel_nr;
4003 s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled
4004 and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O
4005 interrupt for the target subchannel has been dequeued and subchannel_id,
4006 subchannel_nr, io_int_parm and io_int_word contain the parameters for that
4007 interrupt. ipb is needed for instruction parameter decoding.
4014 On FSL BookE PowerPC chips, the interrupt controller has a fast patch
4015 interrupt acknowledge path to the core. When the core successfully
4016 delivers an interrupt, it automatically populates the EPR register with
4017 the interrupt vector number and acknowledges the interrupt inside
4018 the interrupt controller.
4020 In case the interrupt controller lives in user space, we need to do
4021 the interrupt acknowledge cycle through it to fetch the next to be
4022 delivered interrupt vector using this exit.
4024 It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an
4025 external interrupt has just been delivered into the guest. User space
4026 should put the acknowledged interrupt vector into the 'epr' field.
4028 /* KVM_EXIT_SYSTEM_EVENT */
4030 #define KVM_SYSTEM_EVENT_SHUTDOWN 1
4031 #define KVM_SYSTEM_EVENT_RESET 2
4032 #define KVM_SYSTEM_EVENT_CRASH 3
4037 If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered
4038 a system-level event using some architecture specific mechanism (hypercall
4039 or some special instruction). In case of ARM/ARM64, this is triggered using
4040 HVC instruction based PSCI call from the vcpu. The 'type' field describes
4041 the system-level event type. The 'flags' field describes architecture
4042 specific flags for the system-level event.
4044 Valid values for 'type' are:
4045 KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the
4046 VM. Userspace is not obliged to honour this, and if it does honour
4047 this does not need to destroy the VM synchronously (ie it may call
4048 KVM_RUN again before shutdown finally occurs).
4049 KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM.
4050 As with SHUTDOWN, userspace can choose to ignore the request, or
4051 to schedule the reset to occur in the future and may call KVM_RUN again.
4052 KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest
4053 has requested a crash condition maintenance. Userspace can choose
4054 to ignore the request, or to gather VM memory core dump and/or
4055 reset/shutdown of the VM.
4057 /* KVM_EXIT_IOAPIC_EOI */
4062 Indicates that the VCPU's in-kernel local APIC received an EOI for a
4063 level-triggered IOAPIC interrupt. This exit only triggers when the
4064 IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled);
4065 the userspace IOAPIC should process the EOI and retrigger the interrupt if
4066 it is still asserted. Vector is the LAPIC interrupt vector for which the
4069 struct kvm_hyperv_exit {
4070 #define KVM_EXIT_HYPERV_SYNIC 1
4071 #define KVM_EXIT_HYPERV_HCALL 2
4087 /* KVM_EXIT_HYPERV */
4088 struct kvm_hyperv_exit hyperv;
4089 Indicates that the VCPU exits into userspace to process some tasks
4090 related to Hyper-V emulation.
4091 Valid values for 'type' are:
4092 KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about
4093 Hyper-V SynIC state change. Notification is used to remap SynIC
4094 event/message pages and to enable/disable SynIC messages/events processing
4097 /* Fix the size of the union. */
4102 * shared registers between kvm and userspace.
4103 * kvm_valid_regs specifies the register classes set by the host
4104 * kvm_dirty_regs specified the register classes dirtied by userspace
4105 * struct kvm_sync_regs is architecture specific, as well as the
4106 * bits for kvm_valid_regs and kvm_dirty_regs
4108 __u64 kvm_valid_regs;
4109 __u64 kvm_dirty_regs;
4111 struct kvm_sync_regs regs;
4112 char padding[SYNC_REGS_SIZE_BYTES];
4115 If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access
4116 certain guest registers without having to call SET/GET_*REGS. Thus we can
4117 avoid some system call overhead if userspace has to handle the exit.
4118 Userspace can query the validity of the structure by checking
4119 kvm_valid_regs for specific bits. These bits are architecture specific
4120 and usually define the validity of a groups of registers. (e.g. one bit
4121 for general purpose registers)
4123 Please note that the kernel is allowed to use the kvm_run structure as the
4124 primary storage for certain register types. Therefore, the kernel may use the
4125 values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set.
4131 6. Capabilities that can be enabled on vCPUs
4132 --------------------------------------------
4134 There are certain capabilities that change the behavior of the virtual CPU or
4135 the virtual machine when enabled. To enable them, please see section 4.37.
4136 Below you can find a list of capabilities and what their effect on the vCPU or
4137 the virtual machine is when enabling them.
4139 The following information is provided along with the description:
4141 Architectures: which instruction set architectures provide this ioctl.
4142 x86 includes both i386 and x86_64.
4144 Target: whether this is a per-vcpu or per-vm capability.
4146 Parameters: what parameters are accepted by the capability.
4148 Returns: the return value. General error numbers (EBADF, ENOMEM, EINVAL)
4149 are not detailed, but errors with specific meanings are.
4157 Returns: 0 on success; -1 on error
4159 This capability enables interception of OSI hypercalls that otherwise would
4160 be treated as normal system calls to be injected into the guest. OSI hypercalls
4161 were invented by Mac-on-Linux to have a standardized communication mechanism
4162 between the guest and the host.
4164 When this capability is enabled, KVM_EXIT_OSI can occur.
4167 6.2 KVM_CAP_PPC_PAPR
4172 Returns: 0 on success; -1 on error
4174 This capability enables interception of PAPR hypercalls. PAPR hypercalls are
4175 done using the hypercall instruction "sc 1".
4177 It also sets the guest privilege level to "supervisor" mode. Usually the guest
4178 runs in "hypervisor" privilege mode with a few missing features.
4180 In addition to the above, it changes the semantics of SDR1. In this mode, the
4181 HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the
4182 HTAB invisible to the guest.
4184 When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur.
4191 Parameters: args[0] is the address of a struct kvm_config_tlb
4192 Returns: 0 on success; -1 on error
4194 struct kvm_config_tlb {
4201 Configures the virtual CPU's TLB array, establishing a shared memory area
4202 between userspace and KVM. The "params" and "array" fields are userspace
4203 addresses of mmu-type-specific data structures. The "array_len" field is an
4204 safety mechanism, and should be set to the size in bytes of the memory that
4205 userspace has reserved for the array. It must be at least the size dictated
4206 by "mmu_type" and "params".
4208 While KVM_RUN is active, the shared region is under control of KVM. Its
4209 contents are undefined, and any modification by userspace results in
4210 boundedly undefined behavior.
4212 On return from KVM_RUN, the shared region will reflect the current state of
4213 the guest's TLB. If userspace makes any changes, it must call KVM_DIRTY_TLB
4214 to tell KVM which entries have been changed, prior to calling KVM_RUN again
4217 For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV:
4218 - The "params" field is of type "struct kvm_book3e_206_tlb_params".
4219 - The "array" field points to an array of type "struct
4220 kvm_book3e_206_tlb_entry".
4221 - The array consists of all entries in the first TLB, followed by all
4222 entries in the second TLB.
4223 - Within a TLB, entries are ordered first by increasing set number. Within a
4224 set, entries are ordered by way (increasing ESEL).
4225 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1)
4226 where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value.
4227 - The tsize field of mas1 shall be set to 4K on TLB0, even though the
4228 hardware ignores this value for TLB0.
4230 6.4 KVM_CAP_S390_CSS_SUPPORT
4235 Returns: 0 on success; -1 on error
4237 This capability enables support for handling of channel I/O instructions.
4239 TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are
4240 handled in-kernel, while the other I/O instructions are passed to userspace.
4242 When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST
4243 SUBCHANNEL intercepts.
4245 Note that even though this capability is enabled per-vcpu, the complete
4246 virtual machine is affected.
4252 Parameters: args[0] defines whether the proxy facility is active
4253 Returns: 0 on success; -1 on error
4255 This capability enables or disables the delivery of interrupts through the
4256 external proxy facility.
4258 When enabled (args[0] != 0), every time the guest gets an external interrupt
4259 delivered, it automatically exits into user space with a KVM_EXIT_EPR exit
4260 to receive the topmost interrupt vector.
4262 When disabled (args[0] == 0), behavior is as if this facility is unsupported.
4264 When this capability is enabled, KVM_EXIT_EPR can occur.
4266 6.6 KVM_CAP_IRQ_MPIC
4269 Parameters: args[0] is the MPIC device fd
4270 args[1] is the MPIC CPU number for this vcpu
4272 This capability connects the vcpu to an in-kernel MPIC device.
4274 6.7 KVM_CAP_IRQ_XICS
4278 Parameters: args[0] is the XICS device fd
4279 args[1] is the XICS CPU number (server ID) for this vcpu
4281 This capability connects the vcpu to an in-kernel XICS device.
4283 6.8 KVM_CAP_S390_IRQCHIP
4289 This capability enables the in-kernel irqchip for s390. Please refer to
4290 "4.24 KVM_CREATE_IRQCHIP" for details.
4292 6.9 KVM_CAP_MIPS_FPU
4296 Parameters: args[0] is reserved for future use (should be 0).
4298 This capability allows the use of the host Floating Point Unit by the guest. It
4299 allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is
4300 done the KVM_REG_MIPS_FPR_* and KVM_REG_MIPS_FCR_* registers can be accessed
4301 (depending on the current guest FPU register mode), and the Status.FR,
4302 Config5.FRE bits are accessible via the KVM API and also from the guest,
4303 depending on them being supported by the FPU.
4305 6.10 KVM_CAP_MIPS_MSA
4309 Parameters: args[0] is reserved for future use (should be 0).
4311 This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest.
4312 It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest.
4313 Once this is done the KVM_REG_MIPS_VEC_* and KVM_REG_MIPS_MSA_* registers can be
4314 accessed, and the Config5.MSAEn bit is accessible via the KVM API and also from
4317 6.74 KVM_CAP_SYNC_REGS
4318 Architectures: s390, x86
4319 Target: s390: always enabled, x86: vcpu
4321 Returns: x86: KVM_CHECK_EXTENSION returns a bit-array indicating which register
4322 sets are supported (bitfields defined in arch/x86/include/uapi/asm/kvm.h).
4324 As described above in the kvm_sync_regs struct info in section 5 (kvm_run):
4325 KVM_CAP_SYNC_REGS "allow[s] userspace to access certain guest registers
4326 without having to call SET/GET_*REGS". This reduces overhead by eliminating
4327 repeated ioctl calls for setting and/or getting register values. This is
4328 particularly important when userspace is making synchronous guest state
4329 modifications, e.g. when emulating and/or intercepting instructions in
4332 For s390 specifics, please refer to the source code.
4335 - the register sets to be copied out to kvm_run are selectable
4336 by userspace (rather that all sets being copied out for every exit).
4337 - vcpu_events are available in addition to regs and sregs.
4339 For x86, the 'kvm_valid_regs' field of struct kvm_run is overloaded to
4340 function as an input bit-array field set by userspace to indicate the
4341 specific register sets to be copied out on the next exit.
4343 To indicate when userspace has modified values that should be copied into
4344 the vCPU, the all architecture bitarray field, 'kvm_dirty_regs' must be set.
4345 This is done using the same bitflags as for the 'kvm_valid_regs' field.
4346 If the dirty bit is not set, then the register set values will not be copied
4347 into the vCPU even if they've been modified.
4349 Unused bitfields in the bitarrays must be set to zero.
4351 struct kvm_sync_regs {
4352 struct kvm_regs regs;
4353 struct kvm_sregs sregs;
4354 struct kvm_vcpu_events events;
4357 7. Capabilities that can be enabled on VMs
4358 ------------------------------------------
4360 There are certain capabilities that change the behavior of the virtual
4361 machine when enabled. To enable them, please see section 4.37. Below
4362 you can find a list of capabilities and what their effect on the VM
4363 is when enabling them.
4365 The following information is provided along with the description:
4367 Architectures: which instruction set architectures provide this ioctl.
4368 x86 includes both i386 and x86_64.
4370 Parameters: what parameters are accepted by the capability.
4372 Returns: the return value. General error numbers (EBADF, ENOMEM, EINVAL)
4373 are not detailed, but errors with specific meanings are.
4376 7.1 KVM_CAP_PPC_ENABLE_HCALL
4379 Parameters: args[0] is the sPAPR hcall number
4380 args[1] is 0 to disable, 1 to enable in-kernel handling
4382 This capability controls whether individual sPAPR hypercalls (hcalls)
4383 get handled by the kernel or not. Enabling or disabling in-kernel
4384 handling of an hcall is effective across the VM. On creation, an
4385 initial set of hcalls are enabled for in-kernel handling, which
4386 consists of those hcalls for which in-kernel handlers were implemented
4387 before this capability was implemented. If disabled, the kernel will
4388 not to attempt to handle the hcall, but will always exit to userspace
4389 to handle it. Note that it may not make sense to enable some and
4390 disable others of a group of related hcalls, but KVM does not prevent
4391 userspace from doing that.
4393 If the hcall number specified is not one that has an in-kernel
4394 implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL
4397 7.2 KVM_CAP_S390_USER_SIGP
4402 This capability controls which SIGP orders will be handled completely in user
4403 space. With this capability enabled, all fast orders will be handled completely
4409 - CONDITIONAL EMERGENCY SIGNAL
4411 All other orders will be handled completely in user space.
4413 Only privileged operation exceptions will be checked for in the kernel (or even
4414 in the hardware prior to interception). If this capability is not enabled, the
4415 old way of handling SIGP orders is used (partially in kernel and user space).
4417 7.3 KVM_CAP_S390_VECTOR_REGISTERS
4421 Returns: 0 on success, negative value on error
4423 Allows use of the vector registers introduced with z13 processor, and
4424 provides for the synchronization between host and user space. Will
4425 return -EINVAL if the machine does not support vectors.
4427 7.4 KVM_CAP_S390_USER_STSI
4432 This capability allows post-handlers for the STSI instruction. After
4433 initial handling in the kernel, KVM exits to user space with
4434 KVM_EXIT_S390_STSI to allow user space to insert further data.
4436 Before exiting to userspace, kvm handlers should fill in s390_stsi field of
4447 @addr - guest address of STSI SYSIB
4451 @ar - access register number
4453 KVM handlers should exit to userspace with rc = -EREMOTE.
4455 7.5 KVM_CAP_SPLIT_IRQCHIP
4458 Parameters: args[0] - number of routes reserved for userspace IOAPICs
4459 Returns: 0 on success, -1 on error
4461 Create a local apic for each processor in the kernel. This can be used
4462 instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the
4463 IOAPIC and PIC (and also the PIT, even though this has to be enabled
4466 This capability also enables in kernel routing of interrupt requests;
4467 when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are
4468 used in the IRQ routing table. The first args[0] MSI routes are reserved
4469 for the IOAPIC pins. Whenever the LAPIC receives an EOI for these routes,
4470 a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace.
4472 Fails if VCPU has already been created, or if the irqchip is already in the
4473 kernel (i.e. KVM_CREATE_IRQCHIP has already been called).
4480 Allows use of runtime-instrumentation introduced with zEC12 processor.
4481 Will return -EINVAL if the machine does not support runtime-instrumentation.
4482 Will return -EBUSY if a VCPU has already been created.
4484 7.7 KVM_CAP_X2APIC_API
4487 Parameters: args[0] - features that should be enabled
4488 Returns: 0 on success, -EINVAL when args[0] contains invalid features
4490 Valid feature flags in args[0] are
4492 #define KVM_X2APIC_API_USE_32BIT_IDS (1ULL << 0)
4493 #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK (1ULL << 1)
4495 Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of
4496 KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC,
4497 allowing the use of 32-bit APIC IDs. See KVM_CAP_X2APIC_API in their
4498 respective sections.
4500 KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work
4501 in logical mode or with more than 255 VCPUs. Otherwise, KVM treats 0xff
4502 as a broadcast even in x2APIC mode in order to support physical x2APIC
4503 without interrupt remapping. This is undesirable in logical mode,
4504 where 0xff represents CPUs 0-7 in cluster 0.
4506 7.8 KVM_CAP_S390_USER_INSTR0
4511 With this capability enabled, all illegal instructions 0x0000 (2 bytes) will
4512 be intercepted and forwarded to user space. User space can use this
4513 mechanism e.g. to realize 2-byte software breakpoints. The kernel will
4514 not inject an operating exception for these instructions, user space has
4515 to take care of that.
4517 This capability can be enabled dynamically even if VCPUs were already
4518 created and are running.
4524 Returns: 0 on success; -EINVAL if the machine does not support
4525 guarded storage; -EBUSY if a VCPU has already been created.
4527 Allows use of guarded storage for the KVM guest.
4529 7.10 KVM_CAP_S390_AIS
4534 Allow use of adapter-interruption suppression.
4535 Returns: 0 on success; -EBUSY if a VCPU has already been created.
4537 7.11 KVM_CAP_PPC_SMT
4540 Parameters: vsmt_mode, flags
4542 Enabling this capability on a VM provides userspace with a way to set
4543 the desired virtual SMT mode (i.e. the number of virtual CPUs per
4544 virtual core). The virtual SMT mode, vsmt_mode, must be a power of 2
4545 between 1 and 8. On POWER8, vsmt_mode must also be no greater than
4546 the number of threads per subcore for the host. Currently flags must
4547 be 0. A successful call to enable this capability will result in
4548 vsmt_mode being returned when the KVM_CAP_PPC_SMT capability is
4549 subsequently queried for the VM. This capability is only supported by
4550 HV KVM, and can only be set before any VCPUs have been created.
4551 The KVM_CAP_PPC_SMT_POSSIBLE capability indicates which virtual SMT
4552 modes are available.
4554 7.12 KVM_CAP_PPC_FWNMI
4559 With this capability a machine check exception in the guest address
4560 space will cause KVM to exit the guest with NMI exit reason. This
4561 enables QEMU to build error log and branch to guest kernel registered
4562 machine check handling routine. Without this capability KVM will
4563 branch to guests' 0x200 interrupt vector.
4565 7.13 KVM_CAP_X86_DISABLE_EXITS
4568 Parameters: args[0] defines which exits are disabled
4569 Returns: 0 on success, -EINVAL when args[0] contains invalid exits
4571 Valid bits in args[0] are
4573 #define KVM_X86_DISABLE_EXITS_MWAIT (1 << 0)
4574 #define KVM_X86_DISABLE_EXITS_HLT (1 << 1)
4576 Enabling this capability on a VM provides userspace with a way to no
4577 longer intercept some instructions for improved latency in some
4578 workloads, and is suggested when vCPUs are associated to dedicated
4579 physical CPUs. More bits can be added in the future; userspace can
4580 just pass the KVM_CHECK_EXTENSION result to KVM_ENABLE_CAP to disable
4583 Do not enable KVM_FEATURE_PV_UNHALT if you disable HLT exits.
4585 7.14 KVM_CAP_S390_HPAGE_1M
4589 Returns: 0 on success, -EINVAL if hpage module parameter was not set
4590 or cmma is enabled, or the VM has the KVM_VM_S390_UCONTROL
4593 With this capability the KVM support for memory backing with 1m pages
4594 through hugetlbfs can be enabled for a VM. After the capability is
4595 enabled, cmma can't be enabled anymore and pfmfi and the storage key
4596 interpretation are disabled. If cmma has already been enabled or the
4597 hpage module parameter is not set to 1, -EINVAL is returned.
4599 While it is generally possible to create a huge page backed VM without
4600 this capability, the VM will not be able to run.
4602 7.15 KVM_CAP_MSR_PLATFORM_INFO
4605 Parameters: args[0] whether feature should be enabled or not
4607 With this capability, a guest may read the MSR_PLATFORM_INFO MSR. Otherwise,
4608 a #GP would be raised when the guest tries to access. Currently, this
4609 capability does not enable write permissions of this MSR for the guest.
4611 7.16 KVM_CAP_PPC_NESTED_HV
4615 Returns: 0 on success, -EINVAL when the implementation doesn't support
4616 nested-HV virtualization.
4618 HV-KVM on POWER9 and later systems allows for "nested-HV"
4619 virtualization, which provides a way for a guest VM to run guests that
4620 can run using the CPU's supervisor mode (privileged non-hypervisor
4621 state). Enabling this capability on a VM depends on the CPU having
4622 the necessary functionality and on the facility being enabled with a
4623 kvm-hv module parameter.
4625 7.17 KVM_CAP_EXCEPTION_PAYLOAD
4628 Parameters: args[0] whether feature should be enabled or not
4630 With this capability enabled, CR2 will not be modified prior to the
4631 emulated VM-exit when L1 intercepts a #PF exception that occurs in
4632 L2. Similarly, for kvm-intel only, DR6 will not be modified prior to
4633 the emulated VM-exit when L1 intercepts a #DB exception that occurs in
4634 L2. As a result, when KVM_GET_VCPU_EVENTS reports a pending #PF (or
4635 #DB) exception for L2, exception.has_payload will be set and the
4636 faulting address (or the new DR6 bits*) will be reported in the
4637 exception_payload field. Similarly, when userspace injects a #PF (or
4638 #DB) into L2 using KVM_SET_VCPU_EVENTS, it is expected to set
4639 exception.has_payload and to put the faulting address (or the new DR6
4640 bits*) in the exception_payload field.
4642 This capability also enables exception.pending in struct
4643 kvm_vcpu_events, which allows userspace to distinguish between pending
4644 and injected exceptions.
4647 * For the new DR6 bits, note that bit 16 is set iff the #DB exception
4650 8. Other capabilities.
4651 ----------------------
4653 This section lists capabilities that give information about other
4654 features of the KVM implementation.
4656 8.1 KVM_CAP_PPC_HWRNG
4660 This capability, if KVM_CHECK_EXTENSION indicates that it is
4661 available, means that that the kernel has an implementation of the
4662 H_RANDOM hypercall backed by a hardware random-number generator.
4663 If present, the kernel H_RANDOM handler can be enabled for guest use
4664 with the KVM_CAP_PPC_ENABLE_HCALL capability.
4666 8.2 KVM_CAP_HYPERV_SYNIC
4669 This capability, if KVM_CHECK_EXTENSION indicates that it is
4670 available, means that that the kernel has an implementation of the
4671 Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is
4672 used to support Windows Hyper-V based guest paravirt drivers(VMBus).
4674 In order to use SynIC, it has to be activated by setting this
4675 capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this
4676 will disable the use of APIC hardware virtualization even if supported
4677 by the CPU, as it's incompatible with SynIC auto-EOI behavior.
4679 8.3 KVM_CAP_PPC_RADIX_MMU
4683 This capability, if KVM_CHECK_EXTENSION indicates that it is
4684 available, means that that the kernel can support guests using the
4685 radix MMU defined in Power ISA V3.00 (as implemented in the POWER9
4688 8.4 KVM_CAP_PPC_HASH_MMU_V3
4692 This capability, if KVM_CHECK_EXTENSION indicates that it is
4693 available, means that that the kernel can support guests using the
4694 hashed page table MMU defined in Power ISA V3.00 (as implemented in
4695 the POWER9 processor), including in-memory segment tables.
4701 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
4702 it is available, means that full hardware assisted virtualization capabilities
4703 of the hardware are available for use through KVM. An appropriate
4704 KVM_VM_MIPS_* type must be passed to KVM_CREATE_VM to create a VM which
4707 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
4708 available, it means that the VM is using full hardware assisted virtualization
4709 capabilities of the hardware. This is useful to check after creating a VM with
4710 KVM_VM_MIPS_DEFAULT.
4712 The value returned by KVM_CHECK_EXTENSION should be compared against known
4713 values (see below). All other values are reserved. This is to allow for the
4714 possibility of other hardware assisted virtualization implementations which
4715 may be incompatible with the MIPS VZ ASE.
4717 0: The trap & emulate implementation is in use to run guest code in user
4718 mode. Guest virtual memory segments are rearranged to fit the guest in the
4719 user mode address space.
4721 1: The MIPS VZ ASE is in use, providing full hardware assisted
4722 virtualization, including standard guest virtual memory segments.
4728 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
4729 it is available, means that the trap & emulate implementation is available to
4730 run guest code in user mode, even if KVM_CAP_MIPS_VZ indicates that hardware
4731 assisted virtualisation is also available. KVM_VM_MIPS_TE (0) must be passed
4732 to KVM_CREATE_VM to create a VM which utilises it.
4734 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
4735 available, it means that the VM is using trap & emulate.
4737 8.7 KVM_CAP_MIPS_64BIT
4741 This capability indicates the supported architecture type of the guest, i.e. the
4742 supported register and address width.
4744 The values returned when this capability is checked by KVM_CHECK_EXTENSION on a
4745 kvm VM handle correspond roughly to the CP0_Config.AT register field, and should
4746 be checked specifically against known values (see below). All other values are
4749 0: MIPS32 or microMIPS32.
4750 Both registers and addresses are 32-bits wide.
4751 It will only be possible to run 32-bit guest code.
4753 1: MIPS64 or microMIPS64 with access only to 32-bit compatibility segments.
4754 Registers are 64-bits wide, but addresses are 32-bits wide.
4755 64-bit guest code may run but cannot access MIPS64 memory segments.
4756 It will also be possible to run 32-bit guest code.
4758 2: MIPS64 or microMIPS64 with access to all address segments.
4759 Both registers and addresses are 64-bits wide.
4760 It will be possible to run 64-bit or 32-bit guest code.
4762 8.9 KVM_CAP_ARM_USER_IRQ
4764 Architectures: arm, arm64
4765 This capability, if KVM_CHECK_EXTENSION indicates that it is available, means
4766 that if userspace creates a VM without an in-kernel interrupt controller, it
4767 will be notified of changes to the output level of in-kernel emulated devices,
4768 which can generate virtual interrupts, presented to the VM.
4769 For such VMs, on every return to userspace, the kernel
4770 updates the vcpu's run->s.regs.device_irq_level field to represent the actual
4771 output level of the device.
4773 Whenever kvm detects a change in the device output level, kvm guarantees at
4774 least one return to userspace before running the VM. This exit could either
4775 be a KVM_EXIT_INTR or any other exit event, like KVM_EXIT_MMIO. This way,
4776 userspace can always sample the device output level and re-compute the state of
4777 the userspace interrupt controller. Userspace should always check the state
4778 of run->s.regs.device_irq_level on every kvm exit.
4779 The value in run->s.regs.device_irq_level can represent both level and edge
4780 triggered interrupt signals, depending on the device. Edge triggered interrupt
4781 signals will exit to userspace with the bit in run->s.regs.device_irq_level
4782 set exactly once per edge signal.
4784 The field run->s.regs.device_irq_level is available independent of
4785 run->kvm_valid_regs or run->kvm_dirty_regs bits.
4787 If KVM_CAP_ARM_USER_IRQ is supported, the KVM_CHECK_EXTENSION ioctl returns a
4788 number larger than 0 indicating the version of this capability is implemented
4789 and thereby which bits in in run->s.regs.device_irq_level can signal values.
4791 Currently the following bits are defined for the device_irq_level bitmap:
4793 KVM_CAP_ARM_USER_IRQ >= 1:
4795 KVM_ARM_DEV_EL1_VTIMER - EL1 virtual timer
4796 KVM_ARM_DEV_EL1_PTIMER - EL1 physical timer
4797 KVM_ARM_DEV_PMU - ARM PMU overflow interrupt signal
4799 Future versions of kvm may implement additional events. These will get
4800 indicated by returning a higher number from KVM_CHECK_EXTENSION and will be
4803 8.10 KVM_CAP_PPC_SMT_POSSIBLE
4807 Querying this capability returns a bitmap indicating the possible
4808 virtual SMT modes that can be set using KVM_CAP_PPC_SMT. If bit N
4809 (counting from the right) is set, then a virtual SMT mode of 2^N is
4812 8.11 KVM_CAP_HYPERV_SYNIC2
4816 This capability enables a newer version of Hyper-V Synthetic interrupt
4817 controller (SynIC). The only difference with KVM_CAP_HYPERV_SYNIC is that KVM
4818 doesn't clear SynIC message and event flags pages when they are enabled by
4819 writing to the respective MSRs.
4821 8.12 KVM_CAP_HYPERV_VP_INDEX
4825 This capability indicates that userspace can load HV_X64_MSR_VP_INDEX msr. Its
4826 value is used to denote the target vcpu for a SynIC interrupt. For
4827 compatibilty, KVM initializes this msr to KVM's internal vcpu index. When this
4828 capability is absent, userspace can still query this msr's value.
4830 8.13 KVM_CAP_S390_AIS_MIGRATION
4835 This capability indicates if the flic device will be able to get/set the
4836 AIS states for migration via the KVM_DEV_FLIC_AISM_ALL attribute and allows
4837 to discover this without having to create a flic device.
4839 8.14 KVM_CAP_S390_PSW
4843 This capability indicates that the PSW is exposed via the kvm_run structure.
4845 8.15 KVM_CAP_S390_GMAP
4849 This capability indicates that the user space memory used as guest mapping can
4850 be anywhere in the user memory address space, as long as the memory slots are
4851 aligned and sized to a segment (1MB) boundary.
4853 8.16 KVM_CAP_S390_COW
4857 This capability indicates that the user space memory used as guest mapping can
4858 use copy-on-write semantics as well as dirty pages tracking via read-only page
4861 8.17 KVM_CAP_S390_BPB
4865 This capability indicates that kvm will implement the interfaces to handle
4866 reset, migration and nested KVM for branch prediction blocking. The stfle
4867 facility 82 should not be provided to the guest without this capability.
4869 8.18 KVM_CAP_HYPERV_TLBFLUSH
4873 This capability indicates that KVM supports paravirtualized Hyper-V TLB Flush
4875 HvFlushVirtualAddressSpace, HvFlushVirtualAddressSpaceEx,
4876 HvFlushVirtualAddressList, HvFlushVirtualAddressListEx.
4878 8.19 KVM_CAP_ARM_INJECT_SERROR_ESR
4880 Architectures: arm, arm64
4882 This capability indicates that userspace can specify (via the
4883 KVM_SET_VCPU_EVENTS ioctl) the syndrome value reported to the guest when it
4884 takes a virtual SError interrupt exception.
4885 If KVM advertises this capability, userspace can only specify the ISS field for
4886 the ESR syndrome. Other parts of the ESR, such as the EC are generated by the
4887 CPU when the exception is taken. If this virtual SError is taken to EL1 using
4888 AArch64, this value will be reported in the ISS field of ESR_ELx.
4890 See KVM_CAP_VCPU_EVENTS for more details.
4891 8.20 KVM_CAP_HYPERV_SEND_IPI
4895 This capability indicates that KVM supports paravirtualized Hyper-V IPI send
4897 HvCallSendSyntheticClusterIpi, HvCallSendSyntheticClusterIpiEx.