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.
48 It is important to note that althought VM ioctls may only be issued from
49 the process that created the VM, a VM's lifecycle is associated with its
50 file descriptor, not its creator (process). In other words, the VM and
51 its resources, *including the associated address space*, are not freed
52 until the last reference to the VM's file descriptor has been released.
53 For example, if fork() is issued after ioctl(KVM_CREATE_VM), the VM will
54 not be freed until both the parent (original) process and its child have
55 put their references to the VM's file descriptor.
57 Because a VM's resources are not freed until the last reference to its
58 file descriptor is released, creating additional references to a VM via
59 via fork(), dup(), etc... without careful consideration is strongly
60 discouraged and may have unwanted side effects, e.g. memory allocated
61 by and on behalf of the VM's process may not be freed/unaccounted when
68 As of Linux 2.6.22, the KVM ABI has been stabilized: no backward
69 incompatible change are allowed. However, there is an extension
70 facility that allows backward-compatible extensions to the API to be
73 The extension mechanism is not based on the Linux version number.
74 Instead, kvm defines extension identifiers and a facility to query
75 whether a particular extension identifier is available. If it is, a
76 set of ioctls is available for application use.
82 This section describes ioctls that can be used to control kvm guests.
83 For each ioctl, the following information is provided along with a
86 Capability: which KVM extension provides this ioctl. Can be 'basic',
87 which means that is will be provided by any kernel that supports
88 API version 12 (see section 4.1), a KVM_CAP_xyz constant, which
89 means availability needs to be checked with KVM_CHECK_EXTENSION
90 (see section 4.4), or 'none' which means that while not all kernels
91 support this ioctl, there's no capability bit to check its
92 availability: for kernels that don't support the ioctl,
93 the ioctl returns -ENOTTY.
95 Architectures: which instruction set architectures provide this ioctl.
96 x86 includes both i386 and x86_64.
98 Type: system, vm, or vcpu.
100 Parameters: what parameters are accepted by the ioctl.
102 Returns: the return value. General error numbers (EBADF, ENOMEM, EINVAL)
103 are not detailed, but errors with specific meanings are.
106 4.1 KVM_GET_API_VERSION
112 Returns: the constant KVM_API_VERSION (=12)
114 This identifies the API version as the stable kvm API. It is not
115 expected that this number will change. However, Linux 2.6.20 and
116 2.6.21 report earlier versions; these are not documented and not
117 supported. Applications should refuse to run if KVM_GET_API_VERSION
118 returns a value other than 12. If this check passes, all ioctls
119 described as 'basic' will be available.
127 Parameters: machine type identifier (KVM_VM_*)
128 Returns: a VM fd that can be used to control the new virtual machine.
130 The new VM has no virtual cpus and no memory.
131 You probably want to use 0 as machine type.
133 In order to create user controlled virtual machines on S390, check
134 KVM_CAP_S390_UCONTROL and use the flag KVM_VM_S390_UCONTROL as
135 privileged user (CAP_SYS_ADMIN).
137 To use hardware assisted virtualization on MIPS (VZ ASE) rather than
138 the default trap & emulate implementation (which changes the virtual
139 memory layout to fit in user mode), check KVM_CAP_MIPS_VZ and use the
143 On arm64, the physical address size for a VM (IPA Size limit) is limited
144 to 40bits by default. The limit can be configured if the host supports the
145 extension KVM_CAP_ARM_VM_IPA_SIZE. When supported, use
146 KVM_VM_TYPE_ARM_IPA_SIZE(IPA_Bits) to set the size in the machine type
147 identifier, where IPA_Bits is the maximum width of any physical
148 address used by the VM. The IPA_Bits is encoded in bits[7-0] of the
149 machine type identifier.
151 e.g, to configure a guest to use 48bit physical address size :
153 vm_fd = ioctl(dev_fd, KVM_CREATE_VM, KVM_VM_TYPE_ARM_IPA_SIZE(48));
155 The requested size (IPA_Bits) must be :
156 0 - Implies default size, 40bits (for backward compatibility)
160 N - Implies N bits, where N is a positive integer such that,
161 32 <= N <= Host_IPA_Limit
163 Host_IPA_Limit is the maximum possible value for IPA_Bits on the host and
164 is dependent on the CPU capability and the kernel configuration. The limit can
165 be retrieved using KVM_CAP_ARM_VM_IPA_SIZE of the KVM_CHECK_EXTENSION
168 Please note that configuring the IPA size does not affect the capability
169 exposed by the guest CPUs in ID_AA64MMFR0_EL1[PARange]. It only affects
170 size of the address translated by the stage2 level (guest physical to
171 host physical address translations).
174 4.3 KVM_GET_MSR_INDEX_LIST, KVM_GET_MSR_FEATURE_INDEX_LIST
176 Capability: basic, KVM_CAP_GET_MSR_FEATURES for KVM_GET_MSR_FEATURE_INDEX_LIST
179 Parameters: struct kvm_msr_list (in/out)
180 Returns: 0 on success; -1 on error
182 EFAULT: the msr index list cannot be read from or written to
183 E2BIG: the msr index list is to be to fit in the array specified by
186 struct kvm_msr_list {
187 __u32 nmsrs; /* number of msrs in entries */
191 The user fills in the size of the indices array in nmsrs, and in return
192 kvm adjusts nmsrs to reflect the actual number of msrs and fills in the
193 indices array with their numbers.
195 KVM_GET_MSR_INDEX_LIST returns the guest msrs that are supported. The list
196 varies by kvm version and host processor, but does not change otherwise.
198 Note: if kvm indicates supports MCE (KVM_CAP_MCE), then the MCE bank MSRs are
199 not returned in the MSR list, as different vcpus can have a different number
200 of banks, as set via the KVM_X86_SETUP_MCE ioctl.
202 KVM_GET_MSR_FEATURE_INDEX_LIST returns the list of MSRs that can be passed
203 to the KVM_GET_MSRS system ioctl. This lets userspace probe host capabilities
204 and processor features that are exposed via MSRs (e.g., VMX capabilities).
205 This list also varies by kvm version and host processor, but does not change
209 4.4 KVM_CHECK_EXTENSION
211 Capability: basic, KVM_CAP_CHECK_EXTENSION_VM for vm ioctl
213 Type: system ioctl, vm ioctl
214 Parameters: extension identifier (KVM_CAP_*)
215 Returns: 0 if unsupported; 1 (or some other positive integer) if supported
217 The API allows the application to query about extensions to the core
218 kvm API. Userspace passes an extension identifier (an integer) and
219 receives an integer that describes the extension availability.
220 Generally 0 means no and 1 means yes, but some extensions may report
221 additional information in the integer return value.
223 Based on their initialization different VMs may have different capabilities.
224 It is thus encouraged to use the vm ioctl to query for capabilities (available
225 with KVM_CAP_CHECK_EXTENSION_VM on the vm fd)
227 4.5 KVM_GET_VCPU_MMAP_SIZE
233 Returns: size of vcpu mmap area, in bytes
235 The KVM_RUN ioctl (cf.) communicates with userspace via a shared
236 memory region. This ioctl returns the size of that region. See the
237 KVM_RUN documentation for details.
240 4.6 KVM_SET_MEMORY_REGION
245 Parameters: struct kvm_memory_region (in)
246 Returns: 0 on success, -1 on error
248 This ioctl is obsolete and has been removed.
256 Parameters: vcpu id (apic id on x86)
257 Returns: vcpu fd on success, -1 on error
259 This API adds a vcpu to a virtual machine. No more than max_vcpus may be added.
260 The vcpu id is an integer in the range [0, max_vcpu_id).
262 The recommended max_vcpus value can be retrieved using the KVM_CAP_NR_VCPUS of
263 the KVM_CHECK_EXTENSION ioctl() at run-time.
264 The maximum possible value for max_vcpus can be retrieved using the
265 KVM_CAP_MAX_VCPUS of the KVM_CHECK_EXTENSION ioctl() at run-time.
267 If the KVM_CAP_NR_VCPUS does not exist, you should assume that max_vcpus is 4
269 If the KVM_CAP_MAX_VCPUS does not exist, you should assume that max_vcpus is
270 same as the value returned from KVM_CAP_NR_VCPUS.
272 The maximum possible value for max_vcpu_id can be retrieved using the
273 KVM_CAP_MAX_VCPU_ID of the KVM_CHECK_EXTENSION ioctl() at run-time.
275 If the KVM_CAP_MAX_VCPU_ID does not exist, you should assume that max_vcpu_id
276 is the same as the value returned from KVM_CAP_MAX_VCPUS.
278 On powerpc using book3s_hv mode, the vcpus are mapped onto virtual
279 threads in one or more virtual CPU cores. (This is because the
280 hardware requires all the hardware threads in a CPU core to be in the
281 same partition.) The KVM_CAP_PPC_SMT capability indicates the number
282 of vcpus per virtual core (vcore). The vcore id is obtained by
283 dividing the vcpu id by the number of vcpus per vcore. The vcpus in a
284 given vcore will always be in the same physical core as each other
285 (though that might be a different physical core from time to time).
286 Userspace can control the threading (SMT) mode of the guest by its
287 allocation of vcpu ids. For example, if userspace wants
288 single-threaded guest vcpus, it should make all vcpu ids be a multiple
289 of the number of vcpus per vcore.
291 For virtual cpus that have been created with S390 user controlled virtual
292 machines, the resulting vcpu fd can be memory mapped at page offset
293 KVM_S390_SIE_PAGE_OFFSET in order to obtain a memory map of the virtual
294 cpu's hardware control block.
297 4.8 KVM_GET_DIRTY_LOG (vm ioctl)
302 Parameters: struct kvm_dirty_log (in/out)
303 Returns: 0 on success, -1 on error
305 /* for KVM_GET_DIRTY_LOG */
306 struct kvm_dirty_log {
310 void __user *dirty_bitmap; /* one bit per page */
315 Given a memory slot, return a bitmap containing any pages dirtied
316 since the last call to this ioctl. Bit 0 is the first page in the
317 memory slot. Ensure the entire structure is cleared to avoid padding
320 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 specifies
321 the address space for which you want to return the dirty bitmap.
322 They must be less than the value that KVM_CHECK_EXTENSION returns for
323 the KVM_CAP_MULTI_ADDRESS_SPACE capability.
325 The bits in the dirty bitmap are cleared before the ioctl returns, unless
326 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT is enabled. For more information,
327 see the description of the capability.
329 4.9 KVM_SET_MEMORY_ALIAS
334 Parameters: struct kvm_memory_alias (in)
335 Returns: 0 (success), -1 (error)
337 This ioctl is obsolete and has been removed.
346 Returns: 0 on success, -1 on error
348 EINTR: an unmasked signal is pending
350 This ioctl is used to run a guest virtual cpu. While there are no
351 explicit parameters, there is an implicit parameter block that can be
352 obtained by mmap()ing the vcpu fd at offset 0, with the size given by
353 KVM_GET_VCPU_MMAP_SIZE. The parameter block is formatted as a 'struct
354 kvm_run' (see below).
360 Architectures: all except ARM, arm64
362 Parameters: struct kvm_regs (out)
363 Returns: 0 on success, -1 on error
365 Reads the general purpose registers from the vcpu.
369 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
370 __u64 rax, rbx, rcx, rdx;
371 __u64 rsi, rdi, rsp, rbp;
372 __u64 r8, r9, r10, r11;
373 __u64 r12, r13, r14, r15;
379 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
390 Architectures: all except ARM, arm64
392 Parameters: struct kvm_regs (in)
393 Returns: 0 on success, -1 on error
395 Writes the general purpose registers into the vcpu.
397 See KVM_GET_REGS for the data structure.
403 Architectures: x86, ppc
405 Parameters: struct kvm_sregs (out)
406 Returns: 0 on success, -1 on error
408 Reads special registers from the vcpu.
412 struct kvm_segment cs, ds, es, fs, gs, ss;
413 struct kvm_segment tr, ldt;
414 struct kvm_dtable gdt, idt;
415 __u64 cr0, cr2, cr3, cr4, cr8;
418 __u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64];
421 /* ppc -- see arch/powerpc/include/uapi/asm/kvm.h */
423 interrupt_bitmap is a bitmap of pending external interrupts. At most
424 one bit may be set. This interrupt has been acknowledged by the APIC
425 but not yet injected into the cpu core.
431 Architectures: x86, ppc
433 Parameters: struct kvm_sregs (in)
434 Returns: 0 on success, -1 on error
436 Writes special registers into the vcpu. See KVM_GET_SREGS for the
445 Parameters: struct kvm_translation (in/out)
446 Returns: 0 on success, -1 on error
448 Translates a virtual address according to the vcpu's current address
451 struct kvm_translation {
453 __u64 linear_address;
456 __u64 physical_address;
467 Architectures: x86, ppc, mips
469 Parameters: struct kvm_interrupt (in)
470 Returns: 0 on success, negative on failure.
472 Queues a hardware interrupt vector to be injected.
474 /* for KVM_INTERRUPT */
475 struct kvm_interrupt {
482 Returns: 0 on success,
483 -EEXIST if an interrupt is already enqueued
484 -EINVAL the the irq number is invalid
485 -ENXIO if the PIC is in the kernel
486 -EFAULT if the pointer is invalid
488 Note 'irq' is an interrupt vector, not an interrupt pin or line. This
489 ioctl is useful if the in-kernel PIC is not used.
493 Queues an external interrupt to be injected. This ioctl is overleaded
494 with 3 different irq values:
498 This injects an edge type external interrupt into the guest once it's ready
499 to receive interrupts. When injected, the interrupt is done.
501 b) KVM_INTERRUPT_UNSET
503 This unsets any pending interrupt.
505 Only available with KVM_CAP_PPC_UNSET_IRQ.
507 c) KVM_INTERRUPT_SET_LEVEL
509 This injects a level type external interrupt into the guest context. The
510 interrupt stays pending until a specific ioctl with KVM_INTERRUPT_UNSET
513 Only available with KVM_CAP_PPC_IRQ_LEVEL.
515 Note that any value for 'irq' other than the ones stated above is invalid
516 and incurs unexpected behavior.
520 Queues an external interrupt to be injected into the virtual CPU. A negative
521 interrupt number dequeues the interrupt.
532 Support for this has been removed. Use KVM_SET_GUEST_DEBUG instead.
537 Capability: basic (vcpu), KVM_CAP_GET_MSR_FEATURES (system)
539 Type: system ioctl, vcpu ioctl
540 Parameters: struct kvm_msrs (in/out)
541 Returns: number of msrs successfully returned;
544 When used as a system ioctl:
545 Reads the values of MSR-based features that are available for the VM. This
546 is similar to KVM_GET_SUPPORTED_CPUID, but it returns MSR indices and values.
547 The list of msr-based features can be obtained using KVM_GET_MSR_FEATURE_INDEX_LIST
550 When used as a vcpu ioctl:
551 Reads model-specific registers from the vcpu. Supported msr indices can
552 be obtained using KVM_GET_MSR_INDEX_LIST in a system ioctl.
555 __u32 nmsrs; /* number of msrs in entries */
558 struct kvm_msr_entry entries[0];
561 struct kvm_msr_entry {
567 Application code should set the 'nmsrs' member (which indicates the
568 size of the entries array) and the 'index' member of each array entry.
569 kvm will fill in the 'data' member.
577 Parameters: struct kvm_msrs (in)
578 Returns: 0 on success, -1 on error
580 Writes model-specific registers to the vcpu. See KVM_GET_MSRS for the
583 Application code should set the 'nmsrs' member (which indicates the
584 size of the entries array), and the 'index' and 'data' members of each
593 Parameters: struct kvm_cpuid (in)
594 Returns: 0 on success, -1 on error
596 Defines the vcpu responses to the cpuid instruction. Applications
597 should use the KVM_SET_CPUID2 ioctl if available.
600 struct kvm_cpuid_entry {
609 /* for KVM_SET_CPUID */
613 struct kvm_cpuid_entry entries[0];
617 4.21 KVM_SET_SIGNAL_MASK
622 Parameters: struct kvm_signal_mask (in)
623 Returns: 0 on success, -1 on error
625 Defines which signals are blocked during execution of KVM_RUN. This
626 signal mask temporarily overrides the threads signal mask. Any
627 unblocked signal received (except SIGKILL and SIGSTOP, which retain
628 their traditional behaviour) will cause KVM_RUN to return with -EINTR.
630 Note the signal will only be delivered if not blocked by the original
633 /* for KVM_SET_SIGNAL_MASK */
634 struct kvm_signal_mask {
645 Parameters: struct kvm_fpu (out)
646 Returns: 0 on success, -1 on error
648 Reads the floating point state from the vcpu.
650 /* for KVM_GET_FPU and KVM_SET_FPU */
655 __u8 ftwx; /* in fxsave format */
671 Parameters: struct kvm_fpu (in)
672 Returns: 0 on success, -1 on error
674 Writes the floating point state to the vcpu.
676 /* for KVM_GET_FPU and KVM_SET_FPU */
681 __u8 ftwx; /* in fxsave format */
692 4.24 KVM_CREATE_IRQCHIP
694 Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQCHIP (s390)
695 Architectures: x86, ARM, arm64, s390
698 Returns: 0 on success, -1 on error
700 Creates an interrupt controller model in the kernel.
701 On x86, creates a virtual ioapic, a virtual PIC (two PICs, nested), and sets up
702 future vcpus to have a local APIC. IRQ routing for GSIs 0-15 is set to both
703 PIC and IOAPIC; GSI 16-23 only go to the IOAPIC.
704 On ARM/arm64, a GICv2 is created. Any other GIC versions require the usage of
705 KVM_CREATE_DEVICE, which also supports creating a GICv2. Using
706 KVM_CREATE_DEVICE is preferred over KVM_CREATE_IRQCHIP for GICv2.
707 On s390, a dummy irq routing table is created.
709 Note that on s390 the KVM_CAP_S390_IRQCHIP vm capability needs to be enabled
710 before KVM_CREATE_IRQCHIP can be used.
715 Capability: KVM_CAP_IRQCHIP
716 Architectures: x86, arm, arm64
718 Parameters: struct kvm_irq_level
719 Returns: 0 on success, -1 on error
721 Sets the level of a GSI input to the interrupt controller model in the kernel.
722 On some architectures it is required that an interrupt controller model has
723 been previously created with KVM_CREATE_IRQCHIP. Note that edge-triggered
724 interrupts require the level to be set to 1 and then back to 0.
726 On real hardware, interrupt pins can be active-low or active-high. This
727 does not matter for the level field of struct kvm_irq_level: 1 always
728 means active (asserted), 0 means inactive (deasserted).
730 x86 allows the operating system to program the interrupt polarity
731 (active-low/active-high) for level-triggered interrupts, and KVM used
732 to consider the polarity. However, due to bitrot in the handling of
733 active-low interrupts, the above convention is now valid on x86 too.
734 This is signaled by KVM_CAP_X86_IOAPIC_POLARITY_IGNORED. Userspace
735 should not present interrupts to the guest as active-low unless this
736 capability is present (or unless it is not using the in-kernel irqchip,
740 ARM/arm64 can signal an interrupt either at the CPU level, or at the
741 in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to
742 use PPIs designated for specific cpus. The irq field is interpreted
745 Â bits: | 31 ... 24 | 23 ... 16 | 15 ... 0 |
746 field: | irq_type | vcpu_index | irq_id |
748 The irq_type field has the following values:
749 - irq_type[0]: out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ
750 - irq_type[1]: in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.)
751 (the vcpu_index field is ignored)
752 - irq_type[2]: in-kernel GIC: PPI, irq_id between 16 and 31 (incl.)
754 (The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs)
756 In both cases, level is used to assert/deassert the line.
758 struct kvm_irq_level {
761 __s32 status; /* not used for KVM_IRQ_LEVEL */
763 __u32 level; /* 0 or 1 */
769 Capability: KVM_CAP_IRQCHIP
772 Parameters: struct kvm_irqchip (in/out)
773 Returns: 0 on success, -1 on error
775 Reads the state of a kernel interrupt controller created with
776 KVM_CREATE_IRQCHIP into a buffer provided by the caller.
779 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
782 char dummy[512]; /* reserving space */
783 struct kvm_pic_state pic;
784 struct kvm_ioapic_state ioapic;
791 Capability: KVM_CAP_IRQCHIP
794 Parameters: struct kvm_irqchip (in)
795 Returns: 0 on success, -1 on error
797 Sets the state of a kernel interrupt controller created with
798 KVM_CREATE_IRQCHIP from a buffer provided by the caller.
801 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
804 char dummy[512]; /* reserving space */
805 struct kvm_pic_state pic;
806 struct kvm_ioapic_state ioapic;
811 4.28 KVM_XEN_HVM_CONFIG
813 Capability: KVM_CAP_XEN_HVM
816 Parameters: struct kvm_xen_hvm_config (in)
817 Returns: 0 on success, -1 on error
819 Sets the MSR that the Xen HVM guest uses to initialize its hypercall
820 page, and provides the starting address and size of the hypercall
821 blobs in userspace. When the guest writes the MSR, kvm copies one
822 page of a blob (32- or 64-bit, depending on the vcpu mode) to guest
825 struct kvm_xen_hvm_config {
838 Capability: KVM_CAP_ADJUST_CLOCK
841 Parameters: struct kvm_clock_data (out)
842 Returns: 0 on success, -1 on error
844 Gets the current timestamp of kvmclock as seen by the current guest. In
845 conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios
848 When KVM_CAP_ADJUST_CLOCK is passed to KVM_CHECK_EXTENSION, it returns the
849 set of bits that KVM can return in struct kvm_clock_data's flag member.
851 The only flag defined now is KVM_CLOCK_TSC_STABLE. If set, the returned
852 value is the exact kvmclock value seen by all VCPUs at the instant
853 when KVM_GET_CLOCK was called. If clear, the returned value is simply
854 CLOCK_MONOTONIC plus a constant offset; the offset can be modified
855 with KVM_SET_CLOCK. KVM will try to make all VCPUs follow this clock,
856 but the exact value read by each VCPU could differ, because the host
859 struct kvm_clock_data {
860 __u64 clock; /* kvmclock current value */
868 Capability: KVM_CAP_ADJUST_CLOCK
871 Parameters: struct kvm_clock_data (in)
872 Returns: 0 on success, -1 on error
874 Sets the current timestamp of kvmclock to the value specified in its parameter.
875 In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios
878 struct kvm_clock_data {
879 __u64 clock; /* kvmclock current value */
885 4.31 KVM_GET_VCPU_EVENTS
887 Capability: KVM_CAP_VCPU_EVENTS
888 Extended by: KVM_CAP_INTR_SHADOW
889 Architectures: x86, arm, arm64
891 Parameters: struct kvm_vcpu_event (out)
892 Returns: 0 on success, -1 on error
896 Gets currently pending exceptions, interrupts, and NMIs as well as related
899 struct kvm_vcpu_events {
928 __u8 exception_has_payload;
929 __u64 exception_payload;
932 The following bits are defined in the flags field:
934 - KVM_VCPUEVENT_VALID_SHADOW may be set to signal that
935 interrupt.shadow contains a valid state.
937 - KVM_VCPUEVENT_VALID_SMM may be set to signal that smi contains a
940 - KVM_VCPUEVENT_VALID_PAYLOAD may be set to signal that the
941 exception_has_payload, exception_payload, and exception.pending
942 fields contain a valid state. This bit will be set whenever
943 KVM_CAP_EXCEPTION_PAYLOAD is enabled.
947 If the guest accesses a device that is being emulated by the host kernel in
948 such a way that a real device would generate a physical SError, KVM may make
949 a virtual SError pending for that VCPU. This system error interrupt remains
950 pending until the guest takes the exception by unmasking PSTATE.A.
952 Running the VCPU may cause it to take a pending SError, or make an access that
953 causes an SError to become pending. The event's description is only valid while
954 the VPCU is not running.
956 This API provides a way to read and write the pending 'event' state that is not
957 visible to the guest. To save, restore or migrate a VCPU the struct representing
958 the state can be read then written using this GET/SET API, along with the other
959 guest-visible registers. It is not possible to 'cancel' an SError that has been
962 A device being emulated in user-space may also wish to generate an SError. To do
963 this the events structure can be populated by user-space. The current state
964 should be read first, to ensure no existing SError is pending. If an existing
965 SError is pending, the architecture's 'Multiple SError interrupts' rules should
966 be followed. (2.5.3 of DDI0587.a "ARM Reliability, Availability, and
967 Serviceability (RAS) Specification").
969 SError exceptions always have an ESR value. Some CPUs have the ability to
970 specify what the virtual SError's ESR value should be. These systems will
971 advertise KVM_CAP_ARM_INJECT_SERROR_ESR. In this case exception.has_esr will
972 always have a non-zero value when read, and the agent making an SError pending
973 should specify the ISS field in the lower 24 bits of exception.serror_esr. If
974 the system supports KVM_CAP_ARM_INJECT_SERROR_ESR, but user-space sets the events
975 with exception.has_esr as zero, KVM will choose an ESR.
977 Specifying exception.has_esr on a system that does not support it will return
978 -EINVAL. Setting anything other than the lower 24bits of exception.serror_esr
981 struct kvm_vcpu_events {
985 /* Align it to 8 bytes */
992 4.32 KVM_SET_VCPU_EVENTS
994 Capability: KVM_CAP_VCPU_EVENTS
995 Extended by: KVM_CAP_INTR_SHADOW
996 Architectures: x86, arm, arm64
998 Parameters: struct kvm_vcpu_event (in)
999 Returns: 0 on success, -1 on error
1003 Set pending exceptions, interrupts, and NMIs as well as related states of the
1006 See KVM_GET_VCPU_EVENTS for the data structure.
1008 Fields that may be modified asynchronously by running VCPUs can be excluded
1009 from the update. These fields are nmi.pending, sipi_vector, smi.smm,
1010 smi.pending. Keep the corresponding bits in the flags field cleared to
1011 suppress overwriting the current in-kernel state. The bits are:
1013 KVM_VCPUEVENT_VALID_NMI_PENDING - transfer nmi.pending to the kernel
1014 KVM_VCPUEVENT_VALID_SIPI_VECTOR - transfer sipi_vector
1015 KVM_VCPUEVENT_VALID_SMM - transfer the smi sub-struct.
1017 If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in
1018 the flags field to signal that interrupt.shadow contains a valid state and
1019 shall be written into the VCPU.
1021 KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available.
1023 If KVM_CAP_EXCEPTION_PAYLOAD is enabled, KVM_VCPUEVENT_VALID_PAYLOAD
1024 can be set in the flags field to signal that the
1025 exception_has_payload, exception_payload, and exception.pending fields
1026 contain a valid state and shall be written into the VCPU.
1030 Set the pending SError exception state for this VCPU. It is not possible to
1031 'cancel' an Serror that has been made pending.
1033 See KVM_GET_VCPU_EVENTS for the data structure.
1036 4.33 KVM_GET_DEBUGREGS
1038 Capability: KVM_CAP_DEBUGREGS
1041 Parameters: struct kvm_debugregs (out)
1042 Returns: 0 on success, -1 on error
1044 Reads debug registers from the vcpu.
1046 struct kvm_debugregs {
1055 4.34 KVM_SET_DEBUGREGS
1057 Capability: KVM_CAP_DEBUGREGS
1060 Parameters: struct kvm_debugregs (in)
1061 Returns: 0 on success, -1 on error
1063 Writes debug registers into the vcpu.
1065 See KVM_GET_DEBUGREGS for the data structure. The flags field is unused
1066 yet and must be cleared on entry.
1069 4.35 KVM_SET_USER_MEMORY_REGION
1071 Capability: KVM_CAP_USER_MEM
1074 Parameters: struct kvm_userspace_memory_region (in)
1075 Returns: 0 on success, -1 on error
1077 struct kvm_userspace_memory_region {
1080 __u64 guest_phys_addr;
1081 __u64 memory_size; /* bytes */
1082 __u64 userspace_addr; /* start of the userspace allocated memory */
1085 /* for kvm_memory_region::flags */
1086 #define KVM_MEM_LOG_DIRTY_PAGES (1UL << 0)
1087 #define KVM_MEM_READONLY (1UL << 1)
1089 This ioctl allows the user to create or modify a guest physical memory
1090 slot. When changing an existing slot, it may be moved in the guest
1091 physical memory space, or its flags may be modified. It may not be
1092 resized. Slots may not overlap in guest physical address space.
1093 Bits 0-15 of "slot" specifies the slot id and this value should be
1094 less than the maximum number of user memory slots supported per VM.
1095 The maximum allowed slots can be queried using KVM_CAP_NR_MEMSLOTS,
1096 if this capability is supported by the architecture.
1098 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot"
1099 specifies the address space which is being modified. They must be
1100 less than the value that KVM_CHECK_EXTENSION returns for the
1101 KVM_CAP_MULTI_ADDRESS_SPACE capability. Slots in separate address spaces
1102 are unrelated; the restriction on overlapping slots only applies within
1105 Memory for the region is taken starting at the address denoted by the
1106 field userspace_addr, which must point at user addressable memory for
1107 the entire memory slot size. Any object may back this memory, including
1108 anonymous memory, ordinary files, and hugetlbfs.
1110 It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr
1111 be identical. This allows large pages in the guest to be backed by large
1114 The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and
1115 KVM_MEM_READONLY. The former can be set to instruct KVM to keep track of
1116 writes to memory within the slot. See KVM_GET_DIRTY_LOG ioctl to know how to
1117 use it. The latter can be set, if KVM_CAP_READONLY_MEM capability allows it,
1118 to make a new slot read-only. In this case, writes to this memory will be
1119 posted to userspace as KVM_EXIT_MMIO exits.
1121 When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of
1122 the memory region are automatically reflected into the guest. For example, an
1123 mmap() that affects the region will be made visible immediately. Another
1124 example is madvise(MADV_DROP).
1126 It is recommended to use this API instead of the KVM_SET_MEMORY_REGION ioctl.
1127 The KVM_SET_MEMORY_REGION does not allow fine grained control over memory
1128 allocation and is deprecated.
1131 4.36 KVM_SET_TSS_ADDR
1133 Capability: KVM_CAP_SET_TSS_ADDR
1136 Parameters: unsigned long tss_address (in)
1137 Returns: 0 on success, -1 on error
1139 This ioctl defines the physical address of a three-page region in the guest
1140 physical address space. The region must be within the first 4GB of the
1141 guest physical address space and must not conflict with any memory slot
1142 or any mmio address. The guest may malfunction if it accesses this memory
1145 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1146 because of a quirk in the virtualization implementation (see the internals
1147 documentation when it pops into existence).
1152 Capability: KVM_CAP_ENABLE_CAP
1153 Architectures: mips, ppc, s390
1155 Parameters: struct kvm_enable_cap (in)
1156 Returns: 0 on success; -1 on error
1158 Capability: KVM_CAP_ENABLE_CAP_VM
1161 Parameters: struct kvm_enable_cap (in)
1162 Returns: 0 on success; -1 on error
1164 +Not all extensions are enabled by default. Using this ioctl the application
1165 can enable an extension, making it available to the guest.
1167 On systems that do not support this ioctl, it always fails. On systems that
1168 do support it, it only works for extensions that are supported for enablement.
1170 To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should
1173 struct kvm_enable_cap {
1177 The capability that is supposed to get enabled.
1181 A bitfield indicating future enhancements. Has to be 0 for now.
1185 Arguments for enabling a feature. If a feature needs initial values to
1186 function properly, this is the place to put them.
1191 The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl
1192 for vm-wide capabilities.
1194 4.38 KVM_GET_MP_STATE
1196 Capability: KVM_CAP_MP_STATE
1197 Architectures: x86, s390, arm, arm64
1199 Parameters: struct kvm_mp_state (out)
1200 Returns: 0 on success; -1 on error
1202 struct kvm_mp_state {
1206 Returns the vcpu's current "multiprocessing state" (though also valid on
1207 uniprocessor guests).
1209 Possible values are:
1211 - KVM_MP_STATE_RUNNABLE: the vcpu is currently running [x86,arm/arm64]
1212 - KVM_MP_STATE_UNINITIALIZED: the vcpu is an application processor (AP)
1213 which has not yet received an INIT signal [x86]
1214 - KVM_MP_STATE_INIT_RECEIVED: the vcpu has received an INIT signal, and is
1215 now ready for a SIPI [x86]
1216 - KVM_MP_STATE_HALTED: the vcpu has executed a HLT instruction and
1217 is waiting for an interrupt [x86]
1218 - KVM_MP_STATE_SIPI_RECEIVED: the vcpu has just received a SIPI (vector
1219 accessible via KVM_GET_VCPU_EVENTS) [x86]
1220 - KVM_MP_STATE_STOPPED: the vcpu is stopped [s390,arm/arm64]
1221 - KVM_MP_STATE_CHECK_STOP: the vcpu is in a special error state [s390]
1222 - KVM_MP_STATE_OPERATING: the vcpu is operating (running or halted)
1224 - KVM_MP_STATE_LOAD: the vcpu is in a special load/startup state
1227 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1228 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1229 these architectures.
1233 The only states that are valid are KVM_MP_STATE_STOPPED and
1234 KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not.
1236 4.39 KVM_SET_MP_STATE
1238 Capability: KVM_CAP_MP_STATE
1239 Architectures: x86, s390, arm, arm64
1241 Parameters: struct kvm_mp_state (in)
1242 Returns: 0 on success; -1 on error
1244 Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for
1247 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1248 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1249 these architectures.
1253 The only states that are valid are KVM_MP_STATE_STOPPED and
1254 KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not.
1256 4.40 KVM_SET_IDENTITY_MAP_ADDR
1258 Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR
1261 Parameters: unsigned long identity (in)
1262 Returns: 0 on success, -1 on error
1264 This ioctl defines the physical address of a one-page region in the guest
1265 physical address space. The region must be within the first 4GB of the
1266 guest physical address space and must not conflict with any memory slot
1267 or any mmio address. The guest may malfunction if it accesses this memory
1270 Setting the address to 0 will result in resetting the address to its default
1273 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1274 because of a quirk in the virtualization implementation (see the internals
1275 documentation when it pops into existence).
1277 Fails if any VCPU has already been created.
1279 4.41 KVM_SET_BOOT_CPU_ID
1281 Capability: KVM_CAP_SET_BOOT_CPU_ID
1284 Parameters: unsigned long vcpu_id
1285 Returns: 0 on success, -1 on error
1287 Define which vcpu is the Bootstrap Processor (BSP). Values are the same
1288 as the vcpu id in KVM_CREATE_VCPU. If this ioctl is not called, the default
1294 Capability: KVM_CAP_XSAVE
1297 Parameters: struct kvm_xsave (out)
1298 Returns: 0 on success, -1 on error
1304 This ioctl would copy current vcpu's xsave struct to the userspace.
1309 Capability: KVM_CAP_XSAVE
1312 Parameters: struct kvm_xsave (in)
1313 Returns: 0 on success, -1 on error
1319 This ioctl would copy userspace's xsave struct to the kernel.
1324 Capability: KVM_CAP_XCRS
1327 Parameters: struct kvm_xcrs (out)
1328 Returns: 0 on success, -1 on error
1339 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1343 This ioctl would copy current vcpu's xcrs to the userspace.
1348 Capability: KVM_CAP_XCRS
1351 Parameters: struct kvm_xcrs (in)
1352 Returns: 0 on success, -1 on error
1363 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1367 This ioctl would set vcpu's xcr to the value userspace specified.
1370 4.46 KVM_GET_SUPPORTED_CPUID
1372 Capability: KVM_CAP_EXT_CPUID
1375 Parameters: struct kvm_cpuid2 (in/out)
1376 Returns: 0 on success, -1 on error
1381 struct kvm_cpuid_entry2 entries[0];
1384 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
1385 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1)
1386 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2)
1388 struct kvm_cpuid_entry2 {
1399 This ioctl returns x86 cpuid features which are supported by both the
1400 hardware and kvm in its default configuration. Userspace can use the
1401 information returned by this ioctl to construct cpuid information (for
1402 KVM_SET_CPUID2) that is consistent with hardware, kernel, and
1403 userspace capabilities, and with user requirements (for example, the
1404 user may wish to constrain cpuid to emulate older hardware, or for
1405 feature consistency across a cluster).
1407 Note that certain capabilities, such as KVM_CAP_X86_DISABLE_EXITS, may
1408 expose cpuid features (e.g. MONITOR) which are not supported by kvm in
1409 its default configuration. If userspace enables such capabilities, it
1410 is responsible for modifying the results of this ioctl appropriately.
1412 Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure
1413 with the 'nent' field indicating the number of entries in the variable-size
1414 array 'entries'. If the number of entries is too low to describe the cpu
1415 capabilities, an error (E2BIG) is returned. If the number is too high,
1416 the 'nent' field is adjusted and an error (ENOMEM) is returned. If the
1417 number is just right, the 'nent' field is adjusted to the number of valid
1418 entries in the 'entries' array, which is then filled.
1420 The entries returned are the host cpuid as returned by the cpuid instruction,
1421 with unknown or unsupported features masked out. Some features (for example,
1422 x2apic), may not be present in the host cpu, but are exposed by kvm if it can
1423 emulate them efficiently. The fields in each entry are defined as follows:
1425 function: the eax value used to obtain the entry
1426 index: the ecx value used to obtain the entry (for entries that are
1428 flags: an OR of zero or more of the following:
1429 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
1430 if the index field is valid
1431 KVM_CPUID_FLAG_STATEFUL_FUNC:
1432 if cpuid for this function returns different values for successive
1433 invocations; there will be several entries with the same function,
1434 all with this flag set
1435 KVM_CPUID_FLAG_STATE_READ_NEXT:
1436 for KVM_CPUID_FLAG_STATEFUL_FUNC entries, set if this entry is
1437 the first entry to be read by a cpu
1438 eax, ebx, ecx, edx: the values returned by the cpuid instruction for
1439 this function/index combination
1441 The TSC deadline timer feature (CPUID leaf 1, ecx[24]) is always returned
1442 as false, since the feature depends on KVM_CREATE_IRQCHIP for local APIC
1443 support. Instead it is reported via
1445 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER)
1447 if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the
1448 feature in userspace, then you can enable the feature for KVM_SET_CPUID2.
1451 4.47 KVM_PPC_GET_PVINFO
1453 Capability: KVM_CAP_PPC_GET_PVINFO
1456 Parameters: struct kvm_ppc_pvinfo (out)
1457 Returns: 0 on success, !0 on error
1459 struct kvm_ppc_pvinfo {
1465 This ioctl fetches PV specific information that need to be passed to the guest
1466 using the device tree or other means from vm context.
1468 The hcall array defines 4 instructions that make up a hypercall.
1470 If any additional field gets added to this structure later on, a bit for that
1471 additional piece of information will be set in the flags bitmap.
1473 The flags bitmap is defined as:
1475 /* the host supports the ePAPR idle hcall
1476 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1<<0)
1478 4.52 KVM_SET_GSI_ROUTING
1480 Capability: KVM_CAP_IRQ_ROUTING
1481 Architectures: x86 s390 arm arm64
1483 Parameters: struct kvm_irq_routing (in)
1484 Returns: 0 on success, -1 on error
1486 Sets the GSI routing table entries, overwriting any previously set entries.
1488 On arm/arm64, GSI routing has the following limitation:
1489 - GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD.
1491 struct kvm_irq_routing {
1494 struct kvm_irq_routing_entry entries[0];
1497 No flags are specified so far, the corresponding field must be set to zero.
1499 struct kvm_irq_routing_entry {
1505 struct kvm_irq_routing_irqchip irqchip;
1506 struct kvm_irq_routing_msi msi;
1507 struct kvm_irq_routing_s390_adapter adapter;
1508 struct kvm_irq_routing_hv_sint hv_sint;
1513 /* gsi routing entry types */
1514 #define KVM_IRQ_ROUTING_IRQCHIP 1
1515 #define KVM_IRQ_ROUTING_MSI 2
1516 #define KVM_IRQ_ROUTING_S390_ADAPTER 3
1517 #define KVM_IRQ_ROUTING_HV_SINT 4
1520 - KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry
1521 type, specifies that the devid field contains a valid value. The per-VM
1522 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
1523 the device ID. If this capability is not available, userspace should
1524 never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
1527 struct kvm_irq_routing_irqchip {
1532 struct kvm_irq_routing_msi {
1542 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
1543 for the device that wrote the MSI message. For PCI, this is usually a
1544 BFD identifier in the lower 16 bits.
1546 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
1547 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
1548 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
1549 address_hi must be zero.
1551 struct kvm_irq_routing_s390_adapter {
1555 __u32 summary_offset;
1559 struct kvm_irq_routing_hv_sint {
1565 4.55 KVM_SET_TSC_KHZ
1567 Capability: KVM_CAP_TSC_CONTROL
1570 Parameters: virtual tsc_khz
1571 Returns: 0 on success, -1 on error
1573 Specifies the tsc frequency for the virtual machine. The unit of the
1577 4.56 KVM_GET_TSC_KHZ
1579 Capability: KVM_CAP_GET_TSC_KHZ
1583 Returns: virtual tsc-khz on success, negative value on error
1585 Returns the tsc frequency of the guest. The unit of the return value is
1586 KHz. If the host has unstable tsc this ioctl returns -EIO instead as an
1592 Capability: KVM_CAP_IRQCHIP
1595 Parameters: struct kvm_lapic_state (out)
1596 Returns: 0 on success, -1 on error
1598 #define KVM_APIC_REG_SIZE 0x400
1599 struct kvm_lapic_state {
1600 char regs[KVM_APIC_REG_SIZE];
1603 Reads the Local APIC registers and copies them into the input argument. The
1604 data format and layout are the same as documented in the architecture manual.
1606 If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is
1607 enabled, then the format of APIC_ID register depends on the APIC mode
1608 (reported by MSR_IA32_APICBASE) of its VCPU. x2APIC stores APIC ID in
1609 the APIC_ID register (bytes 32-35). xAPIC only allows an 8-bit APIC ID
1610 which is stored in bits 31-24 of the APIC register, or equivalently in
1611 byte 35 of struct kvm_lapic_state's regs field. KVM_GET_LAPIC must then
1612 be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR.
1614 If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state
1615 always uses xAPIC format.
1620 Capability: KVM_CAP_IRQCHIP
1623 Parameters: struct kvm_lapic_state (in)
1624 Returns: 0 on success, -1 on error
1626 #define KVM_APIC_REG_SIZE 0x400
1627 struct kvm_lapic_state {
1628 char regs[KVM_APIC_REG_SIZE];
1631 Copies the input argument into the Local APIC registers. The data format
1632 and layout are the same as documented in the architecture manual.
1634 The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's
1635 regs field) depends on the state of the KVM_CAP_X2APIC_API capability.
1636 See the note in KVM_GET_LAPIC.
1641 Capability: KVM_CAP_IOEVENTFD
1644 Parameters: struct kvm_ioeventfd (in)
1645 Returns: 0 on success, !0 on error
1647 This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address
1648 within the guest. A guest write in the registered address will signal the
1649 provided event instead of triggering an exit.
1651 struct kvm_ioeventfd {
1653 __u64 addr; /* legal pio/mmio address */
1654 __u32 len; /* 0, 1, 2, 4, or 8 bytes */
1660 For the special case of virtio-ccw devices on s390, the ioevent is matched
1661 to a subchannel/virtqueue tuple instead.
1663 The following flags are defined:
1665 #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch)
1666 #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio)
1667 #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign)
1668 #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \
1669 (1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify)
1671 If datamatch flag is set, the event will be signaled only if the written value
1672 to the registered address is equal to datamatch in struct kvm_ioeventfd.
1674 For virtio-ccw devices, addr contains the subchannel id and datamatch the
1677 With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and
1678 the kernel will ignore the length of guest write and may get a faster vmexit.
1679 The speedup may only apply to specific architectures, but the ioeventfd will
1684 Capability: KVM_CAP_SW_TLB
1687 Parameters: struct kvm_dirty_tlb (in)
1688 Returns: 0 on success, -1 on error
1690 struct kvm_dirty_tlb {
1695 This must be called whenever userspace has changed an entry in the shared
1696 TLB, prior to calling KVM_RUN on the associated vcpu.
1698 The "bitmap" field is the userspace address of an array. This array
1699 consists of a number of bits, equal to the total number of TLB entries as
1700 determined by the last successful call to KVM_CONFIG_TLB, rounded up to the
1701 nearest multiple of 64.
1703 Each bit corresponds to one TLB entry, ordered the same as in the shared TLB
1706 The array is little-endian: the bit 0 is the least significant bit of the
1707 first byte, bit 8 is the least significant bit of the second byte, etc.
1708 This avoids any complications with differing word sizes.
1710 The "num_dirty" field is a performance hint for KVM to determine whether it
1711 should skip processing the bitmap and just invalidate everything. It must
1712 be set to the number of set bits in the bitmap.
1715 4.62 KVM_CREATE_SPAPR_TCE
1717 Capability: KVM_CAP_SPAPR_TCE
1718 Architectures: powerpc
1720 Parameters: struct kvm_create_spapr_tce (in)
1721 Returns: file descriptor for manipulating the created TCE table
1723 This creates a virtual TCE (translation control entry) table, which
1724 is an IOMMU for PAPR-style virtual I/O. It is used to translate
1725 logical addresses used in virtual I/O into guest physical addresses,
1726 and provides a scatter/gather capability for PAPR virtual I/O.
1728 /* for KVM_CAP_SPAPR_TCE */
1729 struct kvm_create_spapr_tce {
1734 The liobn field gives the logical IO bus number for which to create a
1735 TCE table. The window_size field specifies the size of the DMA window
1736 which this TCE table will translate - the table will contain one 64
1737 bit TCE entry for every 4kiB of the DMA window.
1739 When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE
1740 table has been created using this ioctl(), the kernel will handle it
1741 in real mode, updating the TCE table. H_PUT_TCE calls for other
1742 liobns will cause a vm exit and must be handled by userspace.
1744 The return value is a file descriptor which can be passed to mmap(2)
1745 to map the created TCE table into userspace. This lets userspace read
1746 the entries written by kernel-handled H_PUT_TCE calls, and also lets
1747 userspace update the TCE table directly which is useful in some
1751 4.63 KVM_ALLOCATE_RMA
1753 Capability: KVM_CAP_PPC_RMA
1754 Architectures: powerpc
1756 Parameters: struct kvm_allocate_rma (out)
1757 Returns: file descriptor for mapping the allocated RMA
1759 This allocates a Real Mode Area (RMA) from the pool allocated at boot
1760 time by the kernel. An RMA is a physically-contiguous, aligned region
1761 of memory used on older POWER processors to provide the memory which
1762 will be accessed by real-mode (MMU off) accesses in a KVM guest.
1763 POWER processors support a set of sizes for the RMA that usually
1764 includes 64MB, 128MB, 256MB and some larger powers of two.
1766 /* for KVM_ALLOCATE_RMA */
1767 struct kvm_allocate_rma {
1771 The return value is a file descriptor which can be passed to mmap(2)
1772 to map the allocated RMA into userspace. The mapped area can then be
1773 passed to the KVM_SET_USER_MEMORY_REGION ioctl to establish it as the
1774 RMA for a virtual machine. The size of the RMA in bytes (which is
1775 fixed at host kernel boot time) is returned in the rma_size field of
1776 the argument structure.
1778 The KVM_CAP_PPC_RMA capability is 1 or 2 if the KVM_ALLOCATE_RMA ioctl
1779 is supported; 2 if the processor requires all virtual machines to have
1780 an RMA, or 1 if the processor can use an RMA but doesn't require it,
1781 because it supports the Virtual RMA (VRMA) facility.
1786 Capability: KVM_CAP_USER_NMI
1790 Returns: 0 on success, -1 on error
1792 Queues an NMI on the thread's vcpu. Note this is well defined only
1793 when KVM_CREATE_IRQCHIP has not been called, since this is an interface
1794 between the virtual cpu core and virtual local APIC. After KVM_CREATE_IRQCHIP
1795 has been called, this interface is completely emulated within the kernel.
1797 To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the
1798 following algorithm:
1801 - read the local APIC's state (KVM_GET_LAPIC)
1802 - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1)
1803 - if so, issue KVM_NMI
1806 Some guests configure the LINT1 NMI input to cause a panic, aiding in
1810 4.65 KVM_S390_UCAS_MAP
1812 Capability: KVM_CAP_S390_UCONTROL
1815 Parameters: struct kvm_s390_ucas_mapping (in)
1816 Returns: 0 in case of success
1818 The parameter is defined like this:
1819 struct kvm_s390_ucas_mapping {
1825 This ioctl maps the memory at "user_addr" with the length "length" to
1826 the vcpu's address space starting at "vcpu_addr". All parameters need to
1827 be aligned by 1 megabyte.
1830 4.66 KVM_S390_UCAS_UNMAP
1832 Capability: KVM_CAP_S390_UCONTROL
1835 Parameters: struct kvm_s390_ucas_mapping (in)
1836 Returns: 0 in case of success
1838 The parameter is defined like this:
1839 struct kvm_s390_ucas_mapping {
1845 This ioctl unmaps the memory in the vcpu's address space starting at
1846 "vcpu_addr" with the length "length". The field "user_addr" is ignored.
1847 All parameters need to be aligned by 1 megabyte.
1850 4.67 KVM_S390_VCPU_FAULT
1852 Capability: KVM_CAP_S390_UCONTROL
1855 Parameters: vcpu absolute address (in)
1856 Returns: 0 in case of success
1858 This call creates a page table entry on the virtual cpu's address space
1859 (for user controlled virtual machines) or the virtual machine's address
1860 space (for regular virtual machines). This only works for minor faults,
1861 thus it's recommended to access subject memory page via the user page
1862 table upfront. This is useful to handle validity intercepts for user
1863 controlled virtual machines to fault in the virtual cpu's lowcore pages
1864 prior to calling the KVM_RUN ioctl.
1867 4.68 KVM_SET_ONE_REG
1869 Capability: KVM_CAP_ONE_REG
1872 Parameters: struct kvm_one_reg (in)
1873 Returns: 0 on success, negative value on failure
1875 Â ENOENT: Â Â no such register
1876 Â EINVAL: Â Â other errors, such as bad size encoding for a known register
1878 struct kvm_one_reg {
1883 Using this ioctl, a single vcpu register can be set to a specific value
1884 defined by user space with the passed in struct kvm_one_reg, where id
1885 refers to the register identifier as described below and addr is a pointer
1886 to a variable with the respective size. There can be architecture agnostic
1887 and architecture specific registers. Each have their own range of operation
1888 and their own constants and width. To keep track of the implemented
1889 registers, find a list below:
1891 Arch | Register | Width (bits)
1893 PPC | KVM_REG_PPC_HIOR | 64
1894 PPC | KVM_REG_PPC_IAC1 | 64
1895 PPC | KVM_REG_PPC_IAC2 | 64
1896 PPC | KVM_REG_PPC_IAC3 | 64
1897 PPC | KVM_REG_PPC_IAC4 | 64
1898 PPC | KVM_REG_PPC_DAC1 | 64
1899 PPC | KVM_REG_PPC_DAC2 | 64
1900 PPC | KVM_REG_PPC_DABR | 64
1901 PPC | KVM_REG_PPC_DSCR | 64
1902 PPC | KVM_REG_PPC_PURR | 64
1903 PPC | KVM_REG_PPC_SPURR | 64
1904 PPC | KVM_REG_PPC_DAR | 64
1905 PPC | KVM_REG_PPC_DSISR | 32
1906 PPC | KVM_REG_PPC_AMR | 64
1907 PPC | KVM_REG_PPC_UAMOR | 64
1908 PPC | KVM_REG_PPC_MMCR0 | 64
1909 PPC | KVM_REG_PPC_MMCR1 | 64
1910 PPC | KVM_REG_PPC_MMCRA | 64
1911 PPC | KVM_REG_PPC_MMCR2 | 64
1912 PPC | KVM_REG_PPC_MMCRS | 64
1913 PPC | KVM_REG_PPC_SIAR | 64
1914 PPC | KVM_REG_PPC_SDAR | 64
1915 PPC | KVM_REG_PPC_SIER | 64
1916 PPC | KVM_REG_PPC_PMC1 | 32
1917 PPC | KVM_REG_PPC_PMC2 | 32
1918 PPC | KVM_REG_PPC_PMC3 | 32
1919 PPC | KVM_REG_PPC_PMC4 | 32
1920 PPC | KVM_REG_PPC_PMC5 | 32
1921 PPC | KVM_REG_PPC_PMC6 | 32
1922 PPC | KVM_REG_PPC_PMC7 | 32
1923 PPC | KVM_REG_PPC_PMC8 | 32
1924 PPC | KVM_REG_PPC_FPR0 | 64
1926 PPC | KVM_REG_PPC_FPR31 | 64
1927 PPC | KVM_REG_PPC_VR0 | 128
1929 PPC | KVM_REG_PPC_VR31 | 128
1930 PPC | KVM_REG_PPC_VSR0 | 128
1932 PPC | KVM_REG_PPC_VSR31 | 128
1933 PPC | KVM_REG_PPC_FPSCR | 64
1934 PPC | KVM_REG_PPC_VSCR | 32
1935 PPC | KVM_REG_PPC_VPA_ADDR | 64
1936 PPC | KVM_REG_PPC_VPA_SLB | 128
1937 PPC | KVM_REG_PPC_VPA_DTL | 128
1938 PPC | KVM_REG_PPC_EPCR | 32
1939 PPC | KVM_REG_PPC_EPR | 32
1940 PPC | KVM_REG_PPC_TCR | 32
1941 PPC | KVM_REG_PPC_TSR | 32
1942 PPC | KVM_REG_PPC_OR_TSR | 32
1943 PPC | KVM_REG_PPC_CLEAR_TSR | 32
1944 PPC | KVM_REG_PPC_MAS0 | 32
1945 PPC | KVM_REG_PPC_MAS1 | 32
1946 PPC | KVM_REG_PPC_MAS2 | 64
1947 PPC | KVM_REG_PPC_MAS7_3 | 64
1948 PPC | KVM_REG_PPC_MAS4 | 32
1949 PPC | KVM_REG_PPC_MAS6 | 32
1950 PPC | KVM_REG_PPC_MMUCFG | 32
1951 PPC | KVM_REG_PPC_TLB0CFG | 32
1952 PPC | KVM_REG_PPC_TLB1CFG | 32
1953 PPC | KVM_REG_PPC_TLB2CFG | 32
1954 PPC | KVM_REG_PPC_TLB3CFG | 32
1955 PPC | KVM_REG_PPC_TLB0PS | 32
1956 PPC | KVM_REG_PPC_TLB1PS | 32
1957 PPC | KVM_REG_PPC_TLB2PS | 32
1958 PPC | KVM_REG_PPC_TLB3PS | 32
1959 PPC | KVM_REG_PPC_EPTCFG | 32
1960 PPC | KVM_REG_PPC_ICP_STATE | 64
1961 PPC | KVM_REG_PPC_TB_OFFSET | 64
1962 PPC | KVM_REG_PPC_SPMC1 | 32
1963 PPC | KVM_REG_PPC_SPMC2 | 32
1964 PPC | KVM_REG_PPC_IAMR | 64
1965 PPC | KVM_REG_PPC_TFHAR | 64
1966 PPC | KVM_REG_PPC_TFIAR | 64
1967 PPC | KVM_REG_PPC_TEXASR | 64
1968 PPC | KVM_REG_PPC_FSCR | 64
1969 PPC | KVM_REG_PPC_PSPB | 32
1970 PPC | KVM_REG_PPC_EBBHR | 64
1971 PPC | KVM_REG_PPC_EBBRR | 64
1972 PPC | KVM_REG_PPC_BESCR | 64
1973 PPC | KVM_REG_PPC_TAR | 64
1974 PPC | KVM_REG_PPC_DPDES | 64
1975 PPC | KVM_REG_PPC_DAWR | 64
1976 PPC | KVM_REG_PPC_DAWRX | 64
1977 PPC | KVM_REG_PPC_CIABR | 64
1978 PPC | KVM_REG_PPC_IC | 64
1979 PPC | KVM_REG_PPC_VTB | 64
1980 PPC | KVM_REG_PPC_CSIGR | 64
1981 PPC | KVM_REG_PPC_TACR | 64
1982 PPC | KVM_REG_PPC_TCSCR | 64
1983 PPC | KVM_REG_PPC_PID | 64
1984 PPC | KVM_REG_PPC_ACOP | 64
1985 PPC | KVM_REG_PPC_VRSAVE | 32
1986 PPC | KVM_REG_PPC_LPCR | 32
1987 PPC | KVM_REG_PPC_LPCR_64 | 64
1988 PPC | KVM_REG_PPC_PPR | 64
1989 PPC | KVM_REG_PPC_ARCH_COMPAT | 32
1990 PPC | KVM_REG_PPC_DABRX | 32
1991 PPC | KVM_REG_PPC_WORT | 64
1992 PPC | KVM_REG_PPC_SPRG9 | 64
1993 PPC | KVM_REG_PPC_DBSR | 32
1994 PPC | KVM_REG_PPC_TIDR | 64
1995 PPC | KVM_REG_PPC_PSSCR | 64
1996 PPC | KVM_REG_PPC_DEC_EXPIRY | 64
1997 PPC | KVM_REG_PPC_PTCR | 64
1998 PPC | KVM_REG_PPC_TM_GPR0 | 64
2000 PPC | KVM_REG_PPC_TM_GPR31 | 64
2001 PPC | KVM_REG_PPC_TM_VSR0 | 128
2003 PPC | KVM_REG_PPC_TM_VSR63 | 128
2004 PPC | KVM_REG_PPC_TM_CR | 64
2005 PPC | KVM_REG_PPC_TM_LR | 64
2006 PPC | KVM_REG_PPC_TM_CTR | 64
2007 PPC | KVM_REG_PPC_TM_FPSCR | 64
2008 PPC | KVM_REG_PPC_TM_AMR | 64
2009 PPC | KVM_REG_PPC_TM_PPR | 64
2010 PPC | KVM_REG_PPC_TM_VRSAVE | 64
2011 PPC | KVM_REG_PPC_TM_VSCR | 32
2012 PPC | KVM_REG_PPC_TM_DSCR | 64
2013 PPC | KVM_REG_PPC_TM_TAR | 64
2014 PPC | KVM_REG_PPC_TM_XER | 64
2016 MIPS | KVM_REG_MIPS_R0 | 64
2018 MIPS | KVM_REG_MIPS_R31 | 64
2019 MIPS | KVM_REG_MIPS_HI | 64
2020 MIPS | KVM_REG_MIPS_LO | 64
2021 MIPS | KVM_REG_MIPS_PC | 64
2022 MIPS | KVM_REG_MIPS_CP0_INDEX | 32
2023 MIPS | KVM_REG_MIPS_CP0_ENTRYLO0 | 64
2024 MIPS | KVM_REG_MIPS_CP0_ENTRYLO1 | 64
2025 MIPS | KVM_REG_MIPS_CP0_CONTEXT | 64
2026 MIPS | KVM_REG_MIPS_CP0_CONTEXTCONFIG| 32
2027 MIPS | KVM_REG_MIPS_CP0_USERLOCAL | 64
2028 MIPS | KVM_REG_MIPS_CP0_XCONTEXTCONFIG| 64
2029 MIPS | KVM_REG_MIPS_CP0_PAGEMASK | 32
2030 MIPS | KVM_REG_MIPS_CP0_PAGEGRAIN | 32
2031 MIPS | KVM_REG_MIPS_CP0_SEGCTL0 | 64
2032 MIPS | KVM_REG_MIPS_CP0_SEGCTL1 | 64
2033 MIPS | KVM_REG_MIPS_CP0_SEGCTL2 | 64
2034 MIPS | KVM_REG_MIPS_CP0_PWBASE | 64
2035 MIPS | KVM_REG_MIPS_CP0_PWFIELD | 64
2036 MIPS | KVM_REG_MIPS_CP0_PWSIZE | 64
2037 MIPS | KVM_REG_MIPS_CP0_WIRED | 32
2038 MIPS | KVM_REG_MIPS_CP0_PWCTL | 32
2039 MIPS | KVM_REG_MIPS_CP0_HWRENA | 32
2040 MIPS | KVM_REG_MIPS_CP0_BADVADDR | 64
2041 MIPS | KVM_REG_MIPS_CP0_BADINSTR | 32
2042 MIPS | KVM_REG_MIPS_CP0_BADINSTRP | 32
2043 MIPS | KVM_REG_MIPS_CP0_COUNT | 32
2044 MIPS | KVM_REG_MIPS_CP0_ENTRYHI | 64
2045 MIPS | KVM_REG_MIPS_CP0_COMPARE | 32
2046 MIPS | KVM_REG_MIPS_CP0_STATUS | 32
2047 MIPS | KVM_REG_MIPS_CP0_INTCTL | 32
2048 MIPS | KVM_REG_MIPS_CP0_CAUSE | 32
2049 MIPS | KVM_REG_MIPS_CP0_EPC | 64
2050 MIPS | KVM_REG_MIPS_CP0_PRID | 32
2051 MIPS | KVM_REG_MIPS_CP0_EBASE | 64
2052 MIPS | KVM_REG_MIPS_CP0_CONFIG | 32
2053 MIPS | KVM_REG_MIPS_CP0_CONFIG1 | 32
2054 MIPS | KVM_REG_MIPS_CP0_CONFIG2 | 32
2055 MIPS | KVM_REG_MIPS_CP0_CONFIG3 | 32
2056 MIPS | KVM_REG_MIPS_CP0_CONFIG4 | 32
2057 MIPS | KVM_REG_MIPS_CP0_CONFIG5 | 32
2058 MIPS | KVM_REG_MIPS_CP0_CONFIG7 | 32
2059 MIPS | KVM_REG_MIPS_CP0_XCONTEXT | 64
2060 MIPS | KVM_REG_MIPS_CP0_ERROREPC | 64
2061 MIPS | KVM_REG_MIPS_CP0_KSCRATCH1 | 64
2062 MIPS | KVM_REG_MIPS_CP0_KSCRATCH2 | 64
2063 MIPS | KVM_REG_MIPS_CP0_KSCRATCH3 | 64
2064 MIPS | KVM_REG_MIPS_CP0_KSCRATCH4 | 64
2065 MIPS | KVM_REG_MIPS_CP0_KSCRATCH5 | 64
2066 MIPS | KVM_REG_MIPS_CP0_KSCRATCH6 | 64
2067 MIPS | KVM_REG_MIPS_CP0_MAAR(0..63) | 64
2068 MIPS | KVM_REG_MIPS_COUNT_CTL | 64
2069 MIPS | KVM_REG_MIPS_COUNT_RESUME | 64
2070 MIPS | KVM_REG_MIPS_COUNT_HZ | 64
2071 MIPS | KVM_REG_MIPS_FPR_32(0..31) | 32
2072 MIPS | KVM_REG_MIPS_FPR_64(0..31) | 64
2073 MIPS | KVM_REG_MIPS_VEC_128(0..31) | 128
2074 MIPS | KVM_REG_MIPS_FCR_IR | 32
2075 MIPS | KVM_REG_MIPS_FCR_CSR | 32
2076 MIPS | KVM_REG_MIPS_MSA_IR | 32
2077 MIPS | KVM_REG_MIPS_MSA_CSR | 32
2079 ARM registers are mapped using the lower 32 bits. The upper 16 of that
2080 is the register group type, or coprocessor number:
2082 ARM core registers have the following id bit patterns:
2083 0x4020 0000 0010 <index into the kvm_regs struct:16>
2085 ARM 32-bit CP15 registers have the following id bit patterns:
2086 0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3>
2088 ARM 64-bit CP15 registers have the following id bit patterns:
2089 0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3>
2091 ARM CCSIDR registers are demultiplexed by CSSELR value:
2092 0x4020 0000 0011 00 <csselr:8>
2094 ARM 32-bit VFP control registers have the following id bit patterns:
2095 0x4020 0000 0012 1 <regno:12>
2097 ARM 64-bit FP registers have the following id bit patterns:
2098 0x4030 0000 0012 0 <regno:12>
2100 ARM firmware pseudo-registers have the following bit pattern:
2101 0x4030 0000 0014 <regno:16>
2104 arm64 registers are mapped using the lower 32 bits. The upper 16 of
2105 that is the register group type, or coprocessor number:
2107 arm64 core/FP-SIMD registers have the following id bit patterns. Note
2108 that the size of the access is variable, as the kvm_regs structure
2109 contains elements ranging from 32 to 128 bits. The index is a 32bit
2110 value in the kvm_regs structure seen as a 32bit array.
2111 0x60x0 0000 0010 <index into the kvm_regs struct:16>
2114 Encoding Register Bits kvm_regs member
2115 ----------------------------------------------------------------
2116 0x6030 0000 0010 0000 X0 64 regs.regs[0]
2117 0x6030 0000 0010 0002 X1 64 regs.regs[1]
2119 0x6030 0000 0010 003c X30 64 regs.regs[30]
2120 0x6030 0000 0010 003e SP 64 regs.sp
2121 0x6030 0000 0010 0040 PC 64 regs.pc
2122 0x6030 0000 0010 0042 PSTATE 64 regs.pstate
2123 0x6030 0000 0010 0044 SP_EL1 64 sp_el1
2124 0x6030 0000 0010 0046 ELR_EL1 64 elr_el1
2125 0x6030 0000 0010 0048 SPSR_EL1 64 spsr[KVM_SPSR_EL1] (alias SPSR_SVC)
2126 0x6030 0000 0010 004a SPSR_ABT 64 spsr[KVM_SPSR_ABT]
2127 0x6030 0000 0010 004c SPSR_UND 64 spsr[KVM_SPSR_UND]
2128 0x6030 0000 0010 004e SPSR_IRQ 64 spsr[KVM_SPSR_IRQ]
2129 0x6060 0000 0010 0050 SPSR_FIQ 64 spsr[KVM_SPSR_FIQ]
2130 0x6040 0000 0010 0054 V0 128 fp_regs.vregs[0]
2131 0x6040 0000 0010 0058 V1 128 fp_regs.vregs[1]
2133 0x6040 0000 0010 00d0 V31 128 fp_regs.vregs[31]
2134 0x6020 0000 0010 00d4 FPSR 32 fp_regs.fpsr
2135 0x6020 0000 0010 00d5 FPCR 32 fp_regs.fpcr
2137 arm64 CCSIDR registers are demultiplexed by CSSELR value:
2138 0x6020 0000 0011 00 <csselr:8>
2140 arm64 system registers have the following id bit patterns:
2141 0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3>
2143 arm64 firmware pseudo-registers have the following bit pattern:
2144 0x6030 0000 0014 <regno:16>
2147 MIPS registers are mapped using the lower 32 bits. The upper 16 of that is
2148 the register group type:
2150 MIPS core registers (see above) have the following id bit patterns:
2151 0x7030 0000 0000 <reg:16>
2153 MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit
2154 patterns depending on whether they're 32-bit or 64-bit registers:
2155 0x7020 0000 0001 00 <reg:5> <sel:3> (32-bit)
2156 0x7030 0000 0001 00 <reg:5> <sel:3> (64-bit)
2158 Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_MIPS_CP0_ENTRYLO1 are the MIPS64
2159 versions of the EntryLo registers regardless of the word size of the host
2160 hardware, host kernel, guest, and whether XPA is present in the guest, i.e.
2161 with the RI and XI bits (if they exist) in bits 63 and 62 respectively, and
2162 the PFNX field starting at bit 30.
2164 MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) above) have the following id bit
2166 0x7030 0000 0001 01 <reg:8>
2168 MIPS KVM control registers (see above) have the following id bit patterns:
2169 0x7030 0000 0002 <reg:16>
2171 MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following
2172 id bit patterns depending on the size of the register being accessed. They are
2173 always accessed according to the current guest FPU mode (Status.FR and
2174 Config5.FRE), i.e. as the guest would see them, and they become unpredictable
2175 if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector
2176 registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they
2177 overlap the FPU registers:
2178 0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers)
2179 0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers)
2180 0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers)
2182 MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the
2183 following id bit patterns:
2184 0x7020 0000 0003 01 <0:3> <reg:5>
2186 MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the
2187 following id bit patterns:
2188 0x7020 0000 0003 02 <0:3> <reg:5>
2191 4.69 KVM_GET_ONE_REG
2193 Capability: KVM_CAP_ONE_REG
2196 Parameters: struct kvm_one_reg (in and out)
2197 Returns: 0 on success, negative value on failure
2199 Â ENOENT: Â Â no such register
2200 Â EINVAL: Â Â other errors, such as bad size encoding for a known register
2202 This ioctl allows to receive the value of a single register implemented
2203 in a vcpu. The register to read is indicated by the "id" field of the
2204 kvm_one_reg struct passed in. On success, the register value can be found
2205 at the memory location pointed to by "addr".
2207 The list of registers accessible using this interface is identical to the
2211 4.70 KVM_KVMCLOCK_CTRL
2213 Capability: KVM_CAP_KVMCLOCK_CTRL
2214 Architectures: Any that implement pvclocks (currently x86 only)
2217 Returns: 0 on success, -1 on error
2219 This signals to the host kernel that the specified guest is being paused by
2220 userspace. The host will set a flag in the pvclock structure that is checked
2221 from the soft lockup watchdog. The flag is part of the pvclock structure that
2222 is shared between guest and host, specifically the second bit of the flags
2223 field of the pvclock_vcpu_time_info structure. It will be set exclusively by
2224 the host and read/cleared exclusively by the guest. The guest operation of
2225 checking and clearing the flag must an atomic operation so
2226 load-link/store-conditional, or equivalent must be used. There are two cases
2227 where the guest will clear the flag: when the soft lockup watchdog timer resets
2228 itself or when a soft lockup is detected. This ioctl can be called any time
2229 after pausing the vcpu, but before it is resumed.
2234 Capability: KVM_CAP_SIGNAL_MSI
2235 Architectures: x86 arm arm64
2237 Parameters: struct kvm_msi (in)
2238 Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error
2240 Directly inject a MSI message. Only valid with in-kernel irqchip that handles
2252 flags: KVM_MSI_VALID_DEVID: devid contains a valid value. The per-VM
2253 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
2254 the device ID. If this capability is not available, userspace
2255 should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
2257 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
2258 for the device that wrote the MSI message. For PCI, this is usually a
2259 BFD identifier in the lower 16 bits.
2261 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
2262 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
2263 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
2264 address_hi must be zero.
2267 4.71 KVM_CREATE_PIT2
2269 Capability: KVM_CAP_PIT2
2272 Parameters: struct kvm_pit_config (in)
2273 Returns: 0 on success, -1 on error
2275 Creates an in-kernel device model for the i8254 PIT. This call is only valid
2276 after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following
2277 parameters have to be passed:
2279 struct kvm_pit_config {
2286 #define KVM_PIT_SPEAKER_DUMMY 1 /* emulate speaker port stub */
2288 PIT timer interrupts may use a per-VM kernel thread for injection. If it
2289 exists, this thread will have a name of the following pattern:
2291 kvm-pit/<owner-process-pid>
2293 When running a guest with elevated priorities, the scheduling parameters of
2294 this thread may have to be adjusted accordingly.
2296 This IOCTL replaces the obsolete KVM_CREATE_PIT.
2301 Capability: KVM_CAP_PIT_STATE2
2304 Parameters: struct kvm_pit_state2 (out)
2305 Returns: 0 on success, -1 on error
2307 Retrieves the state of the in-kernel PIT model. Only valid after
2308 KVM_CREATE_PIT2. The state is returned in the following structure:
2310 struct kvm_pit_state2 {
2311 struct kvm_pit_channel_state channels[3];
2318 /* disable PIT in HPET legacy mode */
2319 #define KVM_PIT_FLAGS_HPET_LEGACY 0x00000001
2321 This IOCTL replaces the obsolete KVM_GET_PIT.
2326 Capability: KVM_CAP_PIT_STATE2
2329 Parameters: struct kvm_pit_state2 (in)
2330 Returns: 0 on success, -1 on error
2332 Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2.
2333 See KVM_GET_PIT2 for details on struct kvm_pit_state2.
2335 This IOCTL replaces the obsolete KVM_SET_PIT.
2338 4.74 KVM_PPC_GET_SMMU_INFO
2340 Capability: KVM_CAP_PPC_GET_SMMU_INFO
2341 Architectures: powerpc
2344 Returns: 0 on success, -1 on error
2346 This populates and returns a structure describing the features of
2347 the "Server" class MMU emulation supported by KVM.
2348 This can in turn be used by userspace to generate the appropriate
2349 device-tree properties for the guest operating system.
2351 The structure contains some global information, followed by an
2352 array of supported segment page sizes:
2354 struct kvm_ppc_smmu_info {
2358 struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ];
2361 The supported flags are:
2363 - KVM_PPC_PAGE_SIZES_REAL:
2364 When that flag is set, guest page sizes must "fit" the backing
2365 store page sizes. When not set, any page size in the list can
2366 be used regardless of how they are backed by userspace.
2368 - KVM_PPC_1T_SEGMENTS
2369 The emulated MMU supports 1T segments in addition to the
2373 This flag indicates that HPT guests are not supported by KVM,
2374 thus all guests must use radix MMU mode.
2376 The "slb_size" field indicates how many SLB entries are supported
2378 The "sps" array contains 8 entries indicating the supported base
2379 page sizes for a segment in increasing order. Each entry is defined
2382 struct kvm_ppc_one_seg_page_size {
2383 __u32 page_shift; /* Base page shift of segment (or 0) */
2384 __u32 slb_enc; /* SLB encoding for BookS */
2385 struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ];
2388 An entry with a "page_shift" of 0 is unused. Because the array is
2389 organized in increasing order, a lookup can stop when encoutering
2392 The "slb_enc" field provides the encoding to use in the SLB for the
2393 page size. The bits are in positions such as the value can directly
2394 be OR'ed into the "vsid" argument of the slbmte instruction.
2396 The "enc" array is a list which for each of those segment base page
2397 size provides the list of supported actual page sizes (which can be
2398 only larger or equal to the base page size), along with the
2399 corresponding encoding in the hash PTE. Similarly, the array is
2400 8 entries sorted by increasing sizes and an entry with a "0" shift
2401 is an empty entry and a terminator:
2403 struct kvm_ppc_one_page_size {
2404 __u32 page_shift; /* Page shift (or 0) */
2405 __u32 pte_enc; /* Encoding in the HPTE (>>12) */
2408 The "pte_enc" field provides a value that can OR'ed into the hash
2409 PTE's RPN field (ie, it needs to be shifted left by 12 to OR it
2410 into the hash PTE second double word).
2414 Capability: KVM_CAP_IRQFD
2415 Architectures: x86 s390 arm arm64
2417 Parameters: struct kvm_irqfd (in)
2418 Returns: 0 on success, -1 on error
2420 Allows setting an eventfd to directly trigger a guest interrupt.
2421 kvm_irqfd.fd specifies the file descriptor to use as the eventfd and
2422 kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When
2423 an event is triggered on the eventfd, an interrupt is injected into
2424 the guest using the specified gsi pin. The irqfd is removed using
2425 the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd
2428 With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify
2429 mechanism allowing emulation of level-triggered, irqfd-based
2430 interrupts. When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an
2431 additional eventfd in the kvm_irqfd.resamplefd field. When operating
2432 in resample mode, posting of an interrupt through kvm_irq.fd asserts
2433 the specified gsi in the irqchip. When the irqchip is resampled, such
2434 as from an EOI, the gsi is de-asserted and the user is notified via
2435 kvm_irqfd.resamplefd. It is the user's responsibility to re-queue
2436 the interrupt if the device making use of it still requires service.
2437 Note that closing the resamplefd is not sufficient to disable the
2438 irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment
2439 and need not be specified with KVM_IRQFD_FLAG_DEASSIGN.
2441 On arm/arm64, gsi routing being supported, the following can happen:
2442 - in case no routing entry is associated to this gsi, injection fails
2443 - in case the gsi is associated to an irqchip routing entry,
2444 irqchip.pin + 32 corresponds to the injected SPI ID.
2445 - in case the gsi is associated to an MSI routing entry, the MSI
2446 message and device ID are translated into an LPI (support restricted
2447 to GICv3 ITS in-kernel emulation).
2449 4.76 KVM_PPC_ALLOCATE_HTAB
2451 Capability: KVM_CAP_PPC_ALLOC_HTAB
2452 Architectures: powerpc
2454 Parameters: Pointer to u32 containing hash table order (in/out)
2455 Returns: 0 on success, -1 on error
2457 This requests the host kernel to allocate an MMU hash table for a
2458 guest using the PAPR paravirtualization interface. This only does
2459 anything if the kernel is configured to use the Book 3S HV style of
2460 virtualization. Otherwise the capability doesn't exist and the ioctl
2461 returns an ENOTTY error. The rest of this description assumes Book 3S
2464 There must be no vcpus running when this ioctl is called; if there
2465 are, it will do nothing and return an EBUSY error.
2467 The parameter is a pointer to a 32-bit unsigned integer variable
2468 containing the order (log base 2) of the desired size of the hash
2469 table, which must be between 18 and 46. On successful return from the
2470 ioctl, the value will not be changed by the kernel.
2472 If no hash table has been allocated when any vcpu is asked to run
2473 (with the KVM_RUN ioctl), the host kernel will allocate a
2474 default-sized hash table (16 MB).
2476 If this ioctl is called when a hash table has already been allocated,
2477 with a different order from the existing hash table, the existing hash
2478 table will be freed and a new one allocated. If this is ioctl is
2479 called when a hash table has already been allocated of the same order
2480 as specified, the kernel will clear out the existing hash table (zero
2481 all HPTEs). In either case, if the guest is using the virtualized
2482 real-mode area (VRMA) facility, the kernel will re-create the VMRA
2483 HPTEs on the next KVM_RUN of any vcpu.
2485 4.77 KVM_S390_INTERRUPT
2489 Type: vm ioctl, vcpu ioctl
2490 Parameters: struct kvm_s390_interrupt (in)
2491 Returns: 0 on success, -1 on error
2493 Allows to inject an interrupt to the guest. Interrupts can be floating
2494 (vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type.
2496 Interrupt parameters are passed via kvm_s390_interrupt:
2498 struct kvm_s390_interrupt {
2504 type can be one of the following:
2506 KVM_S390_SIGP_STOP (vcpu) - sigp stop; optional flags in parm
2507 KVM_S390_PROGRAM_INT (vcpu) - program check; code in parm
2508 KVM_S390_SIGP_SET_PREFIX (vcpu) - sigp set prefix; prefix address in parm
2509 KVM_S390_RESTART (vcpu) - restart
2510 KVM_S390_INT_CLOCK_COMP (vcpu) - clock comparator interrupt
2511 KVM_S390_INT_CPU_TIMER (vcpu) - CPU timer interrupt
2512 KVM_S390_INT_VIRTIO (vm) - virtio external interrupt; external interrupt
2513 parameters in parm and parm64
2514 KVM_S390_INT_SERVICE (vm) - sclp external interrupt; sclp parameter in parm
2515 KVM_S390_INT_EMERGENCY (vcpu) - sigp emergency; source cpu in parm
2516 KVM_S390_INT_EXTERNAL_CALL (vcpu) - sigp external call; source cpu in parm
2517 KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm) - compound value to indicate an
2518 I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel);
2519 I/O interruption parameters in parm (subchannel) and parm64 (intparm,
2520 interruption subclass)
2521 KVM_S390_MCHK (vm, vcpu) - machine check interrupt; cr 14 bits in parm,
2522 machine check interrupt code in parm64 (note that
2523 machine checks needing further payload are not
2524 supported by this ioctl)
2526 Note that the vcpu ioctl is asynchronous to vcpu execution.
2528 4.78 KVM_PPC_GET_HTAB_FD
2530 Capability: KVM_CAP_PPC_HTAB_FD
2531 Architectures: powerpc
2533 Parameters: Pointer to struct kvm_get_htab_fd (in)
2534 Returns: file descriptor number (>= 0) on success, -1 on error
2536 This returns a file descriptor that can be used either to read out the
2537 entries in the guest's hashed page table (HPT), or to write entries to
2538 initialize the HPT. The returned fd can only be written to if the
2539 KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and
2540 can only be read if that bit is clear. The argument struct looks like
2543 /* For KVM_PPC_GET_HTAB_FD */
2544 struct kvm_get_htab_fd {
2550 /* Values for kvm_get_htab_fd.flags */
2551 #define KVM_GET_HTAB_BOLTED_ONLY ((__u64)0x1)
2552 #define KVM_GET_HTAB_WRITE ((__u64)0x2)
2554 The `start_index' field gives the index in the HPT of the entry at
2555 which to start reading. It is ignored when writing.
2557 Reads on the fd will initially supply information about all
2558 "interesting" HPT entries. Interesting entries are those with the
2559 bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise
2560 all entries. When the end of the HPT is reached, the read() will
2561 return. If read() is called again on the fd, it will start again from
2562 the beginning of the HPT, but will only return HPT entries that have
2563 changed since they were last read.
2565 Data read or written is structured as a header (8 bytes) followed by a
2566 series of valid HPT entries (16 bytes) each. The header indicates how
2567 many valid HPT entries there are and how many invalid entries follow
2568 the valid entries. The invalid entries are not represented explicitly
2569 in the stream. The header format is:
2571 struct kvm_get_htab_header {
2577 Writes to the fd create HPT entries starting at the index given in the
2578 header; first `n_valid' valid entries with contents from the data
2579 written, then `n_invalid' invalid entries, invalidating any previously
2580 valid entries found.
2582 4.79 KVM_CREATE_DEVICE
2584 Capability: KVM_CAP_DEVICE_CTRL
2586 Parameters: struct kvm_create_device (in/out)
2587 Returns: 0 on success, -1 on error
2589 ENODEV: The device type is unknown or unsupported
2590 EEXIST: Device already created, and this type of device may not
2591 be instantiated multiple times
2593 Other error conditions may be defined by individual device types or
2594 have their standard meanings.
2596 Creates an emulated device in the kernel. The file descriptor returned
2597 in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR.
2599 If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the
2600 device type is supported (not necessarily whether it can be created
2603 Individual devices should not define flags. Attributes should be used
2604 for specifying any behavior that is not implied by the device type
2607 struct kvm_create_device {
2608 __u32 type; /* in: KVM_DEV_TYPE_xxx */
2609 __u32 fd; /* out: device handle */
2610 __u32 flags; /* in: KVM_CREATE_DEVICE_xxx */
2613 4.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR
2615 Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
2616 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
2617 Type: device ioctl, vm ioctl, vcpu ioctl
2618 Parameters: struct kvm_device_attr
2619 Returns: 0 on success, -1 on error
2621 ENXIO: The group or attribute is unknown/unsupported for this device
2622 or hardware support is missing.
2623 EPERM: The attribute cannot (currently) be accessed this way
2624 (e.g. read-only attribute, or attribute that only makes
2625 sense when the device is in a different state)
2627 Other error conditions may be defined by individual device types.
2629 Gets/sets a specified piece of device configuration and/or state. The
2630 semantics are device-specific. See individual device documentation in
2631 the "devices" directory. As with ONE_REG, the size of the data
2632 transferred is defined by the particular attribute.
2634 struct kvm_device_attr {
2635 __u32 flags; /* no flags currently defined */
2636 __u32 group; /* device-defined */
2637 __u64 attr; /* group-defined */
2638 __u64 addr; /* userspace address of attr data */
2641 4.81 KVM_HAS_DEVICE_ATTR
2643 Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
2644 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
2645 Type: device ioctl, vm ioctl, vcpu ioctl
2646 Parameters: struct kvm_device_attr
2647 Returns: 0 on success, -1 on error
2649 ENXIO: The group or attribute is unknown/unsupported for this device
2650 or hardware support is missing.
2652 Tests whether a device supports a particular attribute. A successful
2653 return indicates the attribute is implemented. It does not necessarily
2654 indicate that the attribute can be read or written in the device's
2655 current state. "addr" is ignored.
2657 4.82 KVM_ARM_VCPU_INIT
2660 Architectures: arm, arm64
2662 Parameters: struct kvm_vcpu_init (in)
2663 Returns: 0 on success; -1 on error
2665 Â EINVAL: Â Â Â the target is unknown, or the combination of features is invalid.
2666 Â ENOENT: Â Â Â a features bit specified is unknown.
2668 This tells KVM what type of CPU to present to the guest, and what
2669 optional features it should have. Â This will cause a reset of the cpu
2670 registers to their initial values. Â If this is not called, KVM_RUN will
2671 return ENOEXEC for that vcpu.
2673 Note that because some registers reflect machine topology, all vcpus
2674 should be created before this ioctl is invoked.
2676 Userspace can call this function multiple times for a given vcpu, including
2677 after the vcpu has been run. This will reset the vcpu to its initial
2678 state. All calls to this function after the initial call must use the same
2679 target and same set of feature flags, otherwise EINVAL will be returned.
2682 - KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state.
2683 Depends on KVM_CAP_ARM_PSCI. If not set, the CPU will be powered on
2684 and execute guest code when KVM_RUN is called.
2685 - KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode.
2686 Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only).
2687 - KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 (or a future revision
2688 backward compatible with v0.2) for the CPU.
2689 Depends on KVM_CAP_ARM_PSCI_0_2.
2690 - KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU.
2691 Depends on KVM_CAP_ARM_PMU_V3.
2694 4.83 KVM_ARM_PREFERRED_TARGET
2697 Architectures: arm, arm64
2699 Parameters: struct struct kvm_vcpu_init (out)
2700 Returns: 0 on success; -1 on error
2702 ENODEV: no preferred target available for the host
2704 This queries KVM for preferred CPU target type which can be emulated
2705 by KVM on underlying host.
2707 The ioctl returns struct kvm_vcpu_init instance containing information
2708 about preferred CPU target type and recommended features for it. The
2709 kvm_vcpu_init->features bitmap returned will have feature bits set if
2710 the preferred target recommends setting these features, but this is
2713 The information returned by this ioctl can be used to prepare an instance
2714 of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in
2715 in VCPU matching underlying host.
2718 4.84 KVM_GET_REG_LIST
2721 Architectures: arm, arm64, mips
2723 Parameters: struct kvm_reg_list (in/out)
2724 Returns: 0 on success; -1 on error
2726 Â E2BIG: Â Â Â Â the reg index list is too big to fit in the array specified by
2727 Â Â Â Â Â Â Â Â Â Â Â Â the user (the number required will be written into n).
2729 struct kvm_reg_list {
2730 __u64 n; /* number of registers in reg[] */
2734 This ioctl returns the guest registers that are supported for the
2735 KVM_GET_ONE_REG/KVM_SET_ONE_REG calls.
2738 4.85 KVM_ARM_SET_DEVICE_ADDR (deprecated)
2740 Capability: KVM_CAP_ARM_SET_DEVICE_ADDR
2741 Architectures: arm, arm64
2743 Parameters: struct kvm_arm_device_address (in)
2744 Returns: 0 on success, -1 on error
2746 ENODEV: The device id is unknown
2747 ENXIO: Device not supported on current system
2748 EEXIST: Address already set
2749 E2BIG: Address outside guest physical address space
2750 EBUSY: Address overlaps with other device range
2752 struct kvm_arm_device_addr {
2757 Specify a device address in the guest's physical address space where guests
2758 can access emulated or directly exposed devices, which the host kernel needs
2759 to know about. The id field is an architecture specific identifier for a
2762 ARM/arm64 divides the id field into two parts, a device id and an
2763 address type id specific to the individual device.
2765 Â bits: | 63 ... 32 | 31 ... 16 | 15 ... 0 |
2766 field: | 0x00000000 | device id | addr type id |
2768 ARM/arm64 currently only require this when using the in-kernel GIC
2769 support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2
2770 as the device id. When setting the base address for the guest's
2771 mapping of the VGIC virtual CPU and distributor interface, the ioctl
2772 must be called after calling KVM_CREATE_IRQCHIP, but before calling
2773 KVM_RUN on any of the VCPUs. Calling this ioctl twice for any of the
2774 base addresses will return -EEXIST.
2776 Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API
2777 should be used instead.
2780 4.86 KVM_PPC_RTAS_DEFINE_TOKEN
2782 Capability: KVM_CAP_PPC_RTAS
2785 Parameters: struct kvm_rtas_token_args
2786 Returns: 0 on success, -1 on error
2788 Defines a token value for a RTAS (Run Time Abstraction Services)
2789 service in order to allow it to be handled in the kernel. The
2790 argument struct gives the name of the service, which must be the name
2791 of a service that has a kernel-side implementation. If the token
2792 value is non-zero, it will be associated with that service, and
2793 subsequent RTAS calls by the guest specifying that token will be
2794 handled by the kernel. If the token value is 0, then any token
2795 associated with the service will be forgotten, and subsequent RTAS
2796 calls by the guest for that service will be passed to userspace to be
2799 4.87 KVM_SET_GUEST_DEBUG
2801 Capability: KVM_CAP_SET_GUEST_DEBUG
2802 Architectures: x86, s390, ppc, arm64
2804 Parameters: struct kvm_guest_debug (in)
2805 Returns: 0 on success; -1 on error
2807 struct kvm_guest_debug {
2810 struct kvm_guest_debug_arch arch;
2813 Set up the processor specific debug registers and configure vcpu for
2814 handling guest debug events. There are two parts to the structure, the
2815 first a control bitfield indicates the type of debug events to handle
2816 when running. Common control bits are:
2818 - KVM_GUESTDBG_ENABLE: guest debugging is enabled
2819 - KVM_GUESTDBG_SINGLESTEP: the next run should single-step
2821 The top 16 bits of the control field are architecture specific control
2822 flags which can include the following:
2824 - KVM_GUESTDBG_USE_SW_BP: using software breakpoints [x86, arm64]
2825 - KVM_GUESTDBG_USE_HW_BP: using hardware breakpoints [x86, s390, arm64]
2826 - KVM_GUESTDBG_INJECT_DB: inject DB type exception [x86]
2827 - KVM_GUESTDBG_INJECT_BP: inject BP type exception [x86]
2828 - KVM_GUESTDBG_EXIT_PENDING: trigger an immediate guest exit [s390]
2830 For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints
2831 are enabled in memory so we need to ensure breakpoint exceptions are
2832 correctly trapped and the KVM run loop exits at the breakpoint and not
2833 running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP
2834 we need to ensure the guest vCPUs architecture specific registers are
2835 updated to the correct (supplied) values.
2837 The second part of the structure is architecture specific and
2838 typically contains a set of debug registers.
2840 For arm64 the number of debug registers is implementation defined and
2841 can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and
2842 KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number
2843 indicating the number of supported registers.
2845 When debug events exit the main run loop with the reason
2846 KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run
2847 structure containing architecture specific debug information.
2849 4.88 KVM_GET_EMULATED_CPUID
2851 Capability: KVM_CAP_EXT_EMUL_CPUID
2854 Parameters: struct kvm_cpuid2 (in/out)
2855 Returns: 0 on success, -1 on error
2860 struct kvm_cpuid_entry2 entries[0];
2863 The member 'flags' is used for passing flags from userspace.
2865 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
2866 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1)
2867 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2)
2869 struct kvm_cpuid_entry2 {
2880 This ioctl returns x86 cpuid features which are emulated by
2881 kvm.Userspace can use the information returned by this ioctl to query
2882 which features are emulated by kvm instead of being present natively.
2884 Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2
2885 structure with the 'nent' field indicating the number of entries in
2886 the variable-size array 'entries'. If the number of entries is too low
2887 to describe the cpu capabilities, an error (E2BIG) is returned. If the
2888 number is too high, the 'nent' field is adjusted and an error (ENOMEM)
2889 is returned. If the number is just right, the 'nent' field is adjusted
2890 to the number of valid entries in the 'entries' array, which is then
2893 The entries returned are the set CPUID bits of the respective features
2894 which kvm emulates, as returned by the CPUID instruction, with unknown
2895 or unsupported feature bits cleared.
2897 Features like x2apic, for example, may not be present in the host cpu
2898 but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be
2899 emulated efficiently and thus not included here.
2901 The fields in each entry are defined as follows:
2903 function: the eax value used to obtain the entry
2904 index: the ecx value used to obtain the entry (for entries that are
2906 flags: an OR of zero or more of the following:
2907 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
2908 if the index field is valid
2909 KVM_CPUID_FLAG_STATEFUL_FUNC:
2910 if cpuid for this function returns different values for successive
2911 invocations; there will be several entries with the same function,
2912 all with this flag set
2913 KVM_CPUID_FLAG_STATE_READ_NEXT:
2914 for KVM_CPUID_FLAG_STATEFUL_FUNC entries, set if this entry is
2915 the first entry to be read by a cpu
2916 eax, ebx, ecx, edx: the values returned by the cpuid instruction for
2917 this function/index combination
2919 4.89 KVM_S390_MEM_OP
2921 Capability: KVM_CAP_S390_MEM_OP
2924 Parameters: struct kvm_s390_mem_op (in)
2925 Returns: = 0 on success,
2926 < 0 on generic error (e.g. -EFAULT or -ENOMEM),
2927 > 0 if an exception occurred while walking the page tables
2929 Read or write data from/to the logical (virtual) memory of a VCPU.
2931 Parameters are specified via the following structure:
2933 struct kvm_s390_mem_op {
2934 __u64 gaddr; /* the guest address */
2935 __u64 flags; /* flags */
2936 __u32 size; /* amount of bytes */
2937 __u32 op; /* type of operation */
2938 __u64 buf; /* buffer in userspace */
2939 __u8 ar; /* the access register number */
2940 __u8 reserved[31]; /* should be set to 0 */
2943 The type of operation is specified in the "op" field. It is either
2944 KVM_S390_MEMOP_LOGICAL_READ for reading from logical memory space or
2945 KVM_S390_MEMOP_LOGICAL_WRITE for writing to logical memory space. The
2946 KVM_S390_MEMOP_F_CHECK_ONLY flag can be set in the "flags" field to check
2947 whether the corresponding memory access would create an access exception
2948 (without touching the data in the memory at the destination). In case an
2949 access exception occurred while walking the MMU tables of the guest, the
2950 ioctl returns a positive error number to indicate the type of exception.
2951 This exception is also raised directly at the corresponding VCPU if the
2952 flag KVM_S390_MEMOP_F_INJECT_EXCEPTION is set in the "flags" field.
2954 The start address of the memory region has to be specified in the "gaddr"
2955 field, and the length of the region in the "size" field. "buf" is the buffer
2956 supplied by the userspace application where the read data should be written
2957 to for KVM_S390_MEMOP_LOGICAL_READ, or where the data that should be written
2958 is stored for a KVM_S390_MEMOP_LOGICAL_WRITE. "buf" is unused and can be NULL
2959 when KVM_S390_MEMOP_F_CHECK_ONLY is specified. "ar" designates the access
2960 register number to be used.
2962 The "reserved" field is meant for future extensions. It is not used by
2963 KVM with the currently defined set of flags.
2965 4.90 KVM_S390_GET_SKEYS
2967 Capability: KVM_CAP_S390_SKEYS
2970 Parameters: struct kvm_s390_skeys
2971 Returns: 0 on success, KVM_S390_GET_KEYS_NONE if guest is not using storage
2972 keys, negative value on error
2974 This ioctl is used to get guest storage key values on the s390
2975 architecture. The ioctl takes parameters via the kvm_s390_skeys struct.
2977 struct kvm_s390_skeys {
2980 __u64 skeydata_addr;
2985 The start_gfn field is the number of the first guest frame whose storage keys
2988 The count field is the number of consecutive frames (starting from start_gfn)
2989 whose storage keys to get. The count field must be at least 1 and the maximum
2990 allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range
2991 will cause the ioctl to return -EINVAL.
2993 The skeydata_addr field is the address to a buffer large enough to hold count
2994 bytes. This buffer will be filled with storage key data by the ioctl.
2996 4.91 KVM_S390_SET_SKEYS
2998 Capability: KVM_CAP_S390_SKEYS
3001 Parameters: struct kvm_s390_skeys
3002 Returns: 0 on success, negative value on error
3004 This ioctl is used to set guest storage key values on the s390
3005 architecture. The ioctl takes parameters via the kvm_s390_skeys struct.
3006 See section on KVM_S390_GET_SKEYS for struct definition.
3008 The start_gfn field is the number of the first guest frame whose storage keys
3011 The count field is the number of consecutive frames (starting from start_gfn)
3012 whose storage keys to get. The count field must be at least 1 and the maximum
3013 allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range
3014 will cause the ioctl to return -EINVAL.
3016 The skeydata_addr field is the address to a buffer containing count bytes of
3017 storage keys. Each byte in the buffer will be set as the storage key for a
3018 single frame starting at start_gfn for count frames.
3020 Note: If any architecturally invalid key value is found in the given data then
3021 the ioctl will return -EINVAL.
3025 Capability: KVM_CAP_S390_INJECT_IRQ
3028 Parameters: struct kvm_s390_irq (in)
3029 Returns: 0 on success, -1 on error
3031 EINVAL: interrupt type is invalid
3032 type is KVM_S390_SIGP_STOP and flag parameter is invalid value
3033 type is KVM_S390_INT_EXTERNAL_CALL and code is bigger
3034 than the maximum of VCPUs
3035 EBUSY: type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped
3036 type is KVM_S390_SIGP_STOP and a stop irq is already pending
3037 type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt
3040 Allows to inject an interrupt to the guest.
3042 Using struct kvm_s390_irq as a parameter allows
3043 to inject additional payload which is not
3044 possible via KVM_S390_INTERRUPT.
3046 Interrupt parameters are passed via kvm_s390_irq:
3048 struct kvm_s390_irq {
3051 struct kvm_s390_io_info io;
3052 struct kvm_s390_ext_info ext;
3053 struct kvm_s390_pgm_info pgm;
3054 struct kvm_s390_emerg_info emerg;
3055 struct kvm_s390_extcall_info extcall;
3056 struct kvm_s390_prefix_info prefix;
3057 struct kvm_s390_stop_info stop;
3058 struct kvm_s390_mchk_info mchk;
3063 type can be one of the following:
3065 KVM_S390_SIGP_STOP - sigp stop; parameter in .stop
3066 KVM_S390_PROGRAM_INT - program check; parameters in .pgm
3067 KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix
3068 KVM_S390_RESTART - restart; no parameters
3069 KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters
3070 KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters
3071 KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg
3072 KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall
3073 KVM_S390_MCHK - machine check interrupt; parameters in .mchk
3076 Note that the vcpu ioctl is asynchronous to vcpu execution.
3078 4.94 KVM_S390_GET_IRQ_STATE
3080 Capability: KVM_CAP_S390_IRQ_STATE
3083 Parameters: struct kvm_s390_irq_state (out)
3084 Returns: >= number of bytes copied into buffer,
3085 -EINVAL if buffer size is 0,
3086 -ENOBUFS if buffer size is too small to fit all pending interrupts,
3087 -EFAULT if the buffer address was invalid
3089 This ioctl allows userspace to retrieve the complete state of all currently
3090 pending interrupts in a single buffer. Use cases include migration
3091 and introspection. The parameter structure contains the address of a
3092 userspace buffer and its length:
3094 struct kvm_s390_irq_state {
3096 __u32 flags; /* will stay unused for compatibility reasons */
3098 __u32 reserved[4]; /* will stay unused for compatibility reasons */
3101 Userspace passes in the above struct and for each pending interrupt a
3102 struct kvm_s390_irq is copied to the provided buffer.
3104 The structure contains a flags and a reserved field for future extensions. As
3105 the kernel never checked for flags == 0 and QEMU never pre-zeroed flags and
3106 reserved, these fields can not be used in the future without breaking
3109 If -ENOBUFS is returned the buffer provided was too small and userspace
3110 may retry with a bigger buffer.
3112 4.95 KVM_S390_SET_IRQ_STATE
3114 Capability: KVM_CAP_S390_IRQ_STATE
3117 Parameters: struct kvm_s390_irq_state (in)
3118 Returns: 0 on success,
3119 -EFAULT if the buffer address was invalid,
3120 -EINVAL for an invalid buffer length (see below),
3121 -EBUSY if there were already interrupts pending,
3122 errors occurring when actually injecting the
3123 interrupt. See KVM_S390_IRQ.
3125 This ioctl allows userspace to set the complete state of all cpu-local
3126 interrupts currently pending for the vcpu. It is intended for restoring
3127 interrupt state after a migration. The input parameter is a userspace buffer
3128 containing a struct kvm_s390_irq_state:
3130 struct kvm_s390_irq_state {
3132 __u32 flags; /* will stay unused for compatibility reasons */
3134 __u32 reserved[4]; /* will stay unused for compatibility reasons */
3137 The restrictions for flags and reserved apply as well.
3138 (see KVM_S390_GET_IRQ_STATE)
3140 The userspace memory referenced by buf contains a struct kvm_s390_irq
3141 for each interrupt to be injected into the guest.
3142 If one of the interrupts could not be injected for some reason the
3145 len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0
3146 and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq),
3147 which is the maximum number of possibly pending cpu-local interrupts.
3151 Capability: KVM_CAP_X86_SMM
3155 Returns: 0 on success, -1 on error
3157 Queues an SMI on the thread's vcpu.
3159 4.97 KVM_CAP_PPC_MULTITCE
3161 Capability: KVM_CAP_PPC_MULTITCE
3165 This capability means the kernel is capable of handling hypercalls
3166 H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user
3167 space. This significantly accelerates DMA operations for PPC KVM guests.
3168 User space should expect that its handlers for these hypercalls
3169 are not going to be called if user space previously registered LIOBN
3170 in KVM (via KVM_CREATE_SPAPR_TCE or similar calls).
3172 In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest,
3173 user space might have to advertise it for the guest. For example,
3174 IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is
3175 present in the "ibm,hypertas-functions" device-tree property.
3177 The hypercalls mentioned above may or may not be processed successfully
3178 in the kernel based fast path. If they can not be handled by the kernel,
3179 they will get passed on to user space. So user space still has to have
3180 an implementation for these despite the in kernel acceleration.
3182 This capability is always enabled.
3184 4.98 KVM_CREATE_SPAPR_TCE_64
3186 Capability: KVM_CAP_SPAPR_TCE_64
3187 Architectures: powerpc
3189 Parameters: struct kvm_create_spapr_tce_64 (in)
3190 Returns: file descriptor for manipulating the created TCE table
3192 This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit
3193 windows, described in 4.62 KVM_CREATE_SPAPR_TCE
3195 This capability uses extended struct in ioctl interface:
3197 /* for KVM_CAP_SPAPR_TCE_64 */
3198 struct kvm_create_spapr_tce_64 {
3202 __u64 offset; /* in pages */
3203 __u64 size; /* in pages */
3206 The aim of extension is to support an additional bigger DMA window with
3207 a variable page size.
3208 KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and
3209 a bus offset of the corresponding DMA window, @size and @offset are numbers
3212 @flags are not used at the moment.
3214 The rest of functionality is identical to KVM_CREATE_SPAPR_TCE.
3216 4.99 KVM_REINJECT_CONTROL
3218 Capability: KVM_CAP_REINJECT_CONTROL
3221 Parameters: struct kvm_reinject_control (in)
3222 Returns: 0 on success,
3223 -EFAULT if struct kvm_reinject_control cannot be read,
3224 -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier.
3226 i8254 (PIT) has two modes, reinject and !reinject. The default is reinject,
3227 where KVM queues elapsed i8254 ticks and monitors completion of interrupt from
3228 vector(s) that i8254 injects. Reinject mode dequeues a tick and injects its
3229 interrupt whenever there isn't a pending interrupt from i8254.
3230 !reinject mode injects an interrupt as soon as a tick arrives.
3232 struct kvm_reinject_control {
3237 pit_reinject = 0 (!reinject mode) is recommended, unless running an old
3238 operating system that uses the PIT for timing (e.g. Linux 2.4.x).
3240 4.100 KVM_PPC_CONFIGURE_V3_MMU
3242 Capability: KVM_CAP_PPC_RADIX_MMU or KVM_CAP_PPC_HASH_MMU_V3
3245 Parameters: struct kvm_ppc_mmuv3_cfg (in)
3246 Returns: 0 on success,
3247 -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read,
3248 -EINVAL if the configuration is invalid
3250 This ioctl controls whether the guest will use radix or HPT (hashed
3251 page table) translation, and sets the pointer to the process table for
3254 struct kvm_ppc_mmuv3_cfg {
3256 __u64 process_table;
3259 There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and
3260 KVM_PPC_MMUV3_GTSE. KVM_PPC_MMUV3_RADIX, if set, configures the guest
3261 to use radix tree translation, and if clear, to use HPT translation.
3262 KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest
3263 to be able to use the global TLB and SLB invalidation instructions;
3264 if clear, the guest may not use these instructions.
3266 The process_table field specifies the address and size of the guest
3267 process table, which is in the guest's space. This field is formatted
3268 as the second doubleword of the partition table entry, as defined in
3269 the Power ISA V3.00, Book III section 5.7.6.1.
3271 4.101 KVM_PPC_GET_RMMU_INFO
3273 Capability: KVM_CAP_PPC_RADIX_MMU
3276 Parameters: struct kvm_ppc_rmmu_info (out)
3277 Returns: 0 on success,
3278 -EFAULT if struct kvm_ppc_rmmu_info cannot be written,
3279 -EINVAL if no useful information can be returned
3281 This ioctl returns a structure containing two things: (a) a list
3282 containing supported radix tree geometries, and (b) a list that maps
3283 page sizes to put in the "AP" (actual page size) field for the tlbie
3284 (TLB invalidate entry) instruction.
3286 struct kvm_ppc_rmmu_info {
3287 struct kvm_ppc_radix_geom {
3292 __u32 ap_encodings[8];
3295 The geometries[] field gives up to 8 supported geometries for the
3296 radix page table, in terms of the log base 2 of the smallest page
3297 size, and the number of bits indexed at each level of the tree, from
3298 the PTE level up to the PGD level in that order. Any unused entries
3299 will have 0 in the page_shift field.
3301 The ap_encodings gives the supported page sizes and their AP field
3302 encodings, encoded with the AP value in the top 3 bits and the log
3303 base 2 of the page size in the bottom 6 bits.
3305 4.102 KVM_PPC_RESIZE_HPT_PREPARE
3307 Capability: KVM_CAP_SPAPR_RESIZE_HPT
3308 Architectures: powerpc
3310 Parameters: struct kvm_ppc_resize_hpt (in)
3311 Returns: 0 on successful completion,
3312 >0 if a new HPT is being prepared, the value is an estimated
3313 number of milliseconds until preparation is complete
3314 -EFAULT if struct kvm_reinject_control cannot be read,
3315 -EINVAL if the supplied shift or flags are invalid
3316 -ENOMEM if unable to allocate the new HPT
3317 -ENOSPC if there was a hash collision when moving existing
3318 HPT entries to the new HPT
3319 -EIO on other error conditions
3321 Used to implement the PAPR extension for runtime resizing of a guest's
3322 Hashed Page Table (HPT). Specifically this starts, stops or monitors
3323 the preparation of a new potential HPT for the guest, essentially
3324 implementing the H_RESIZE_HPT_PREPARE hypercall.
3326 If called with shift > 0 when there is no pending HPT for the guest,
3327 this begins preparation of a new pending HPT of size 2^(shift) bytes.
3328 It then returns a positive integer with the estimated number of
3329 milliseconds until preparation is complete.
3331 If called when there is a pending HPT whose size does not match that
3332 requested in the parameters, discards the existing pending HPT and
3333 creates a new one as above.
3335 If called when there is a pending HPT of the size requested, will:
3336 * If preparation of the pending HPT is already complete, return 0
3337 * If preparation of the pending HPT has failed, return an error
3338 code, then discard the pending HPT.
3339 * If preparation of the pending HPT is still in progress, return an
3340 estimated number of milliseconds until preparation is complete.
3342 If called with shift == 0, discards any currently pending HPT and
3343 returns 0 (i.e. cancels any in-progress preparation).
3345 flags is reserved for future expansion, currently setting any bits in
3346 flags will result in an -EINVAL.
3348 Normally this will be called repeatedly with the same parameters until
3349 it returns <= 0. The first call will initiate preparation, subsequent
3350 ones will monitor preparation until it completes or fails.
3352 struct kvm_ppc_resize_hpt {
3358 4.103 KVM_PPC_RESIZE_HPT_COMMIT
3360 Capability: KVM_CAP_SPAPR_RESIZE_HPT
3361 Architectures: powerpc
3363 Parameters: struct kvm_ppc_resize_hpt (in)
3364 Returns: 0 on successful completion,
3365 -EFAULT if struct kvm_reinject_control cannot be read,
3366 -EINVAL if the supplied shift or flags are invalid
3367 -ENXIO is there is no pending HPT, or the pending HPT doesn't
3368 have the requested size
3369 -EBUSY if the pending HPT is not fully prepared
3370 -ENOSPC if there was a hash collision when moving existing
3371 HPT entries to the new HPT
3372 -EIO on other error conditions
3374 Used to implement the PAPR extension for runtime resizing of a guest's
3375 Hashed Page Table (HPT). Specifically this requests that the guest be
3376 transferred to working with the new HPT, essentially implementing the
3377 H_RESIZE_HPT_COMMIT hypercall.
3379 This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has
3380 returned 0 with the same parameters. In other cases
3381 KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or
3382 -EBUSY, though others may be possible if the preparation was started,
3385 This will have undefined effects on the guest if it has not already
3386 placed itself in a quiescent state where no vcpu will make MMU enabled
3389 On succsful completion, the pending HPT will become the guest's active
3390 HPT and the previous HPT will be discarded.
3392 On failure, the guest will still be operating on its previous HPT.
3394 struct kvm_ppc_resize_hpt {
3400 4.104 KVM_X86_GET_MCE_CAP_SUPPORTED
3402 Capability: KVM_CAP_MCE
3405 Parameters: u64 mce_cap (out)
3406 Returns: 0 on success, -1 on error
3408 Returns supported MCE capabilities. The u64 mce_cap parameter
3409 has the same format as the MSR_IA32_MCG_CAP register. Supported
3410 capabilities will have the corresponding bits set.
3412 4.105 KVM_X86_SETUP_MCE
3414 Capability: KVM_CAP_MCE
3417 Parameters: u64 mcg_cap (in)
3418 Returns: 0 on success,
3419 -EFAULT if u64 mcg_cap cannot be read,
3420 -EINVAL if the requested number of banks is invalid,
3421 -EINVAL if requested MCE capability is not supported.
3423 Initializes MCE support for use. The u64 mcg_cap parameter
3424 has the same format as the MSR_IA32_MCG_CAP register and
3425 specifies which capabilities should be enabled. The maximum
3426 supported number of error-reporting banks can be retrieved when
3427 checking for KVM_CAP_MCE. The supported capabilities can be
3428 retrieved with KVM_X86_GET_MCE_CAP_SUPPORTED.
3430 4.106 KVM_X86_SET_MCE
3432 Capability: KVM_CAP_MCE
3435 Parameters: struct kvm_x86_mce (in)
3436 Returns: 0 on success,
3437 -EFAULT if struct kvm_x86_mce cannot be read,
3438 -EINVAL if the bank number is invalid,
3439 -EINVAL if VAL bit is not set in status field.
3441 Inject a machine check error (MCE) into the guest. The input
3444 struct kvm_x86_mce {
3454 If the MCE being reported is an uncorrected error, KVM will
3455 inject it as an MCE exception into the guest. If the guest
3456 MCG_STATUS register reports that an MCE is in progress, KVM
3457 causes an KVM_EXIT_SHUTDOWN vmexit.
3459 Otherwise, if the MCE is a corrected error, KVM will just
3460 store it in the corresponding bank (provided this bank is
3461 not holding a previously reported uncorrected error).
3463 4.107 KVM_S390_GET_CMMA_BITS
3465 Capability: KVM_CAP_S390_CMMA_MIGRATION
3468 Parameters: struct kvm_s390_cmma_log (in, out)
3469 Returns: 0 on success, a negative value on error
3471 This ioctl is used to get the values of the CMMA bits on the s390
3472 architecture. It is meant to be used in two scenarios:
3473 - During live migration to save the CMMA values. Live migration needs
3474 to be enabled via the KVM_REQ_START_MIGRATION VM property.
3475 - To non-destructively peek at the CMMA values, with the flag
3476 KVM_S390_CMMA_PEEK set.
3478 The ioctl takes parameters via the kvm_s390_cmma_log struct. The desired
3479 values are written to a buffer whose location is indicated via the "values"
3480 member in the kvm_s390_cmma_log struct. The values in the input struct are
3481 also updated as needed.
3482 Each CMMA value takes up one byte.
3484 struct kvm_s390_cmma_log {
3495 start_gfn is the number of the first guest frame whose CMMA values are
3498 count is the length of the buffer in bytes,
3500 values points to the buffer where the result will be written to.
3502 If count is greater than KVM_S390_SKEYS_MAX, then it is considered to be
3503 KVM_S390_SKEYS_MAX. KVM_S390_SKEYS_MAX is re-used for consistency with
3506 The result is written in the buffer pointed to by the field values, and
3507 the values of the input parameter are updated as follows.
3509 Depending on the flags, different actions are performed. The only
3510 supported flag so far is KVM_S390_CMMA_PEEK.
3512 The default behaviour if KVM_S390_CMMA_PEEK is not set is:
3513 start_gfn will indicate the first page frame whose CMMA bits were dirty.
3514 It is not necessarily the same as the one passed as input, as clean pages
3517 count will indicate the number of bytes actually written in the buffer.
3518 It can (and very often will) be smaller than the input value, since the
3519 buffer is only filled until 16 bytes of clean values are found (which
3520 are then not copied in the buffer). Since a CMMA migration block needs
3521 the base address and the length, for a total of 16 bytes, we will send
3522 back some clean data if there is some dirty data afterwards, as long as
3523 the size of the clean data does not exceed the size of the header. This
3524 allows to minimize the amount of data to be saved or transferred over
3525 the network at the expense of more roundtrips to userspace. The next
3526 invocation of the ioctl will skip over all the clean values, saving
3527 potentially more than just the 16 bytes we found.
3529 If KVM_S390_CMMA_PEEK is set:
3530 the existing storage attributes are read even when not in migration
3531 mode, and no other action is performed;
3533 the output start_gfn will be equal to the input start_gfn,
3535 the output count will be equal to the input count, except if the end of
3536 memory has been reached.
3539 the field "remaining" will indicate the total number of dirty CMMA values
3540 still remaining, or 0 if KVM_S390_CMMA_PEEK is set and migration mode is
3545 values points to the userspace buffer where the result will be stored.
3547 This ioctl can fail with -ENOMEM if not enough memory can be allocated to
3548 complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
3549 KVM_S390_CMMA_PEEK is not set but migration mode was not enabled, with
3550 -EFAULT if the userspace address is invalid or if no page table is
3551 present for the addresses (e.g. when using hugepages).
3553 4.108 KVM_S390_SET_CMMA_BITS
3555 Capability: KVM_CAP_S390_CMMA_MIGRATION
3558 Parameters: struct kvm_s390_cmma_log (in)
3559 Returns: 0 on success, a negative value on error
3561 This ioctl is used to set the values of the CMMA bits on the s390
3562 architecture. It is meant to be used during live migration to restore
3563 the CMMA values, but there are no restrictions on its use.
3564 The ioctl takes parameters via the kvm_s390_cmma_values struct.
3565 Each CMMA value takes up one byte.
3567 struct kvm_s390_cmma_log {
3578 start_gfn indicates the starting guest frame number,
3580 count indicates how many values are to be considered in the buffer,
3582 flags is not used and must be 0.
3584 mask indicates which PGSTE bits are to be considered.
3586 remaining is not used.
3588 values points to the buffer in userspace where to store the values.
3590 This ioctl can fail with -ENOMEM if not enough memory can be allocated to
3591 complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
3592 the count field is too large (e.g. more than KVM_S390_CMMA_SIZE_MAX) or
3593 if the flags field was not 0, with -EFAULT if the userspace address is
3594 invalid, if invalid pages are written to (e.g. after the end of memory)
3595 or if no page table is present for the addresses (e.g. when using
3598 4.109 KVM_PPC_GET_CPU_CHAR
3600 Capability: KVM_CAP_PPC_GET_CPU_CHAR
3601 Architectures: powerpc
3603 Parameters: struct kvm_ppc_cpu_char (out)
3604 Returns: 0 on successful completion
3605 -EFAULT if struct kvm_ppc_cpu_char cannot be written
3607 This ioctl gives userspace information about certain characteristics
3608 of the CPU relating to speculative execution of instructions and
3609 possible information leakage resulting from speculative execution (see
3610 CVE-2017-5715, CVE-2017-5753 and CVE-2017-5754). The information is
3611 returned in struct kvm_ppc_cpu_char, which looks like this:
3613 struct kvm_ppc_cpu_char {
3614 __u64 character; /* characteristics of the CPU */
3615 __u64 behaviour; /* recommended software behaviour */
3616 __u64 character_mask; /* valid bits in character */
3617 __u64 behaviour_mask; /* valid bits in behaviour */
3620 For extensibility, the character_mask and behaviour_mask fields
3621 indicate which bits of character and behaviour have been filled in by
3622 the kernel. If the set of defined bits is extended in future then
3623 userspace will be able to tell whether it is running on a kernel that
3624 knows about the new bits.
3626 The character field describes attributes of the CPU which can help
3627 with preventing inadvertent information disclosure - specifically,
3628 whether there is an instruction to flash-invalidate the L1 data cache
3629 (ori 30,30,0 or mtspr SPRN_TRIG2,rN), whether the L1 data cache is set
3630 to a mode where entries can only be used by the thread that created
3631 them, whether the bcctr[l] instruction prevents speculation, and
3632 whether a speculation barrier instruction (ori 31,31,0) is provided.
3634 The behaviour field describes actions that software should take to
3635 prevent inadvertent information disclosure, and thus describes which
3636 vulnerabilities the hardware is subject to; specifically whether the
3637 L1 data cache should be flushed when returning to user mode from the
3638 kernel, and whether a speculation barrier should be placed between an
3639 array bounds check and the array access.
3641 These fields use the same bit definitions as the new
3642 H_GET_CPU_CHARACTERISTICS hypercall.
3644 4.110 KVM_MEMORY_ENCRYPT_OP
3649 Parameters: an opaque platform specific structure (in/out)
3650 Returns: 0 on success; -1 on error
3652 If the platform supports creating encrypted VMs then this ioctl can be used
3653 for issuing platform-specific memory encryption commands to manage those
3656 Currently, this ioctl is used for issuing Secure Encrypted Virtualization
3657 (SEV) commands on AMD Processors. The SEV commands are defined in
3658 Documentation/virtual/kvm/amd-memory-encryption.rst.
3660 4.111 KVM_MEMORY_ENCRYPT_REG_REGION
3665 Parameters: struct kvm_enc_region (in)
3666 Returns: 0 on success; -1 on error
3668 This ioctl can be used to register a guest memory region which may
3669 contain encrypted data (e.g. guest RAM, SMRAM etc).
3671 It is used in the SEV-enabled guest. When encryption is enabled, a guest
3672 memory region may contain encrypted data. The SEV memory encryption
3673 engine uses a tweak such that two identical plaintext pages, each at
3674 different locations will have differing ciphertexts. So swapping or
3675 moving ciphertext of those pages will not result in plaintext being
3676 swapped. So relocating (or migrating) physical backing pages for the SEV
3677 guest will require some additional steps.
3679 Note: The current SEV key management spec does not provide commands to
3680 swap or migrate (move) ciphertext pages. Hence, for now we pin the guest
3681 memory region registered with the ioctl.
3683 4.112 KVM_MEMORY_ENCRYPT_UNREG_REGION
3688 Parameters: struct kvm_enc_region (in)
3689 Returns: 0 on success; -1 on error
3691 This ioctl can be used to unregister the guest memory region registered
3692 with KVM_MEMORY_ENCRYPT_REG_REGION ioctl above.
3694 4.113 KVM_HYPERV_EVENTFD
3696 Capability: KVM_CAP_HYPERV_EVENTFD
3699 Parameters: struct kvm_hyperv_eventfd (in)
3701 This ioctl (un)registers an eventfd to receive notifications from the guest on
3702 the specified Hyper-V connection id through the SIGNAL_EVENT hypercall, without
3703 causing a user exit. SIGNAL_EVENT hypercall with non-zero event flag number
3704 (bits 24-31) still triggers a KVM_EXIT_HYPERV_HCALL user exit.
3706 struct kvm_hyperv_eventfd {
3713 The conn_id field should fit within 24 bits:
3715 #define KVM_HYPERV_CONN_ID_MASK 0x00ffffff
3717 The acceptable values for the flags field are:
3719 #define KVM_HYPERV_EVENTFD_DEASSIGN (1 << 0)
3721 Returns: 0 on success,
3722 -EINVAL if conn_id or flags is outside the allowed range
3723 -ENOENT on deassign if the conn_id isn't registered
3724 -EEXIST on assign if the conn_id is already registered
3726 4.114 KVM_GET_NESTED_STATE
3728 Capability: KVM_CAP_NESTED_STATE
3731 Parameters: struct kvm_nested_state (in/out)
3732 Returns: 0 on success, -1 on error
3734 E2BIG: the total state size (including the fixed-size part of struct
3735 kvm_nested_state) exceeds the value of 'size' specified by
3736 the user; the size required will be written into size.
3738 struct kvm_nested_state {
3743 struct kvm_vmx_nested_state vmx;
3744 struct kvm_svm_nested_state svm;
3750 #define KVM_STATE_NESTED_GUEST_MODE 0x00000001
3751 #define KVM_STATE_NESTED_RUN_PENDING 0x00000002
3753 #define KVM_STATE_NESTED_SMM_GUEST_MODE 0x00000001
3754 #define KVM_STATE_NESTED_SMM_VMXON 0x00000002
3756 struct kvm_vmx_nested_state {
3765 This ioctl copies the vcpu's nested virtualization state from the kernel to
3768 The maximum size of the state, including the fixed-size part of struct
3769 kvm_nested_state, can be retrieved by passing KVM_CAP_NESTED_STATE to
3770 the KVM_CHECK_EXTENSION ioctl().
3772 4.115 KVM_SET_NESTED_STATE
3774 Capability: KVM_CAP_NESTED_STATE
3777 Parameters: struct kvm_nested_state (in)
3778 Returns: 0 on success, -1 on error
3780 This copies the vcpu's kvm_nested_state struct from userspace to the kernel. For
3781 the definition of struct kvm_nested_state, see KVM_GET_NESTED_STATE.
3783 4.116 KVM_(UN)REGISTER_COALESCED_MMIO
3785 Capability: KVM_CAP_COALESCED_MMIO (for coalesced mmio)
3786 KVM_CAP_COALESCED_PIO (for coalesced pio)
3789 Parameters: struct kvm_coalesced_mmio_zone
3790 Returns: 0 on success, < 0 on error
3792 Coalesced I/O is a performance optimization that defers hardware
3793 register write emulation so that userspace exits are avoided. It is
3794 typically used to reduce the overhead of emulating frequently accessed
3797 When a hardware register is configured for coalesced I/O, write accesses
3798 do not exit to userspace and their value is recorded in a ring buffer
3799 that is shared between kernel and userspace.
3801 Coalesced I/O is used if one or more write accesses to a hardware
3802 register can be deferred until a read or a write to another hardware
3803 register on the same device. This last access will cause a vmexit and
3804 userspace will process accesses from the ring buffer before emulating
3805 it. That will avoid exiting to userspace on repeated writes.
3807 Coalesced pio is based on coalesced mmio. There is little difference
3808 between coalesced mmio and pio except that coalesced pio records accesses
3811 4.117 KVM_CLEAR_DIRTY_LOG (vm ioctl)
3813 Capability: KVM_CAP_MANUAL_DIRTY_LOG_PROTECT
3816 Parameters: struct kvm_dirty_log (in)
3817 Returns: 0 on success, -1 on error
3819 /* for KVM_CLEAR_DIRTY_LOG */
3820 struct kvm_clear_dirty_log {
3825 void __user *dirty_bitmap; /* one bit per page */
3830 The ioctl clears the dirty status of pages in a memory slot, according to
3831 the bitmap that is passed in struct kvm_clear_dirty_log's dirty_bitmap
3832 field. Bit 0 of the bitmap corresponds to page "first_page" in the
3833 memory slot, and num_pages is the size in bits of the input bitmap.
3834 Both first_page and num_pages must be a multiple of 64. For each bit
3835 that is set in the input bitmap, the corresponding page is marked "clean"
3836 in KVM's dirty bitmap, and dirty tracking is re-enabled for that page
3837 (for example via write-protection, or by clearing the dirty bit in
3838 a page table entry).
3840 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 specifies
3841 the address space for which you want to return the dirty bitmap.
3842 They must be less than the value that KVM_CHECK_EXTENSION returns for
3843 the KVM_CAP_MULTI_ADDRESS_SPACE capability.
3845 This ioctl is mostly useful when KVM_CAP_MANUAL_DIRTY_LOG_PROTECT
3846 is enabled; for more information, see the description of the capability.
3847 However, it can always be used as long as KVM_CHECK_EXTENSION confirms
3848 that KVM_CAP_MANUAL_DIRTY_LOG_PROTECT is present.
3850 4.118 KVM_GET_SUPPORTED_HV_CPUID
3852 Capability: KVM_CAP_HYPERV_CPUID
3855 Parameters: struct kvm_cpuid2 (in/out)
3856 Returns: 0 on success, -1 on error
3861 struct kvm_cpuid_entry2 entries[0];
3864 struct kvm_cpuid_entry2 {
3875 This ioctl returns x86 cpuid features leaves related to Hyper-V emulation in
3876 KVM. Userspace can use the information returned by this ioctl to construct
3877 cpuid information presented to guests consuming Hyper-V enlightenments (e.g.
3878 Windows or Hyper-V guests).
3880 CPUID feature leaves returned by this ioctl are defined by Hyper-V Top Level
3881 Functional Specification (TLFS). These leaves can't be obtained with
3882 KVM_GET_SUPPORTED_CPUID ioctl because some of them intersect with KVM feature
3883 leaves (0x40000000, 0x40000001).
3885 Currently, the following list of CPUID leaves are returned:
3886 HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS
3887 HYPERV_CPUID_INTERFACE
3888 HYPERV_CPUID_VERSION
3889 HYPERV_CPUID_FEATURES
3890 HYPERV_CPUID_ENLIGHTMENT_INFO
3891 HYPERV_CPUID_IMPLEMENT_LIMITS
3892 HYPERV_CPUID_NESTED_FEATURES
3894 HYPERV_CPUID_NESTED_FEATURES leaf is only exposed when Enlightened VMCS was
3895 enabled on the corresponding vCPU (KVM_CAP_HYPERV_ENLIGHTENED_VMCS).
3897 Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure
3898 with the 'nent' field indicating the number of entries in the variable-size
3899 array 'entries'. If the number of entries is too low to describe all Hyper-V
3900 feature leaves, an error (E2BIG) is returned. If the number is more or equal
3901 to the number of Hyper-V feature leaves, the 'nent' field is adjusted to the
3902 number of valid entries in the 'entries' array, which is then filled.
3904 'index' and 'flags' fields in 'struct kvm_cpuid_entry2' are currently reserved,
3905 userspace should not expect to get any particular value there.
3907 5. The kvm_run structure
3908 ------------------------
3910 Application code obtains a pointer to the kvm_run structure by
3911 mmap()ing a vcpu fd. From that point, application code can control
3912 execution by changing fields in kvm_run prior to calling the KVM_RUN
3913 ioctl, and obtain information about the reason KVM_RUN returned by
3914 looking up structure members.
3918 __u8 request_interrupt_window;
3920 Request that KVM_RUN return when it becomes possible to inject external
3921 interrupts into the guest. Useful in conjunction with KVM_INTERRUPT.
3923 __u8 immediate_exit;
3925 This field is polled once when KVM_RUN starts; if non-zero, KVM_RUN
3926 exits immediately, returning -EINTR. In the common scenario where a
3927 signal is used to "kick" a VCPU out of KVM_RUN, this field can be used
3928 to avoid usage of KVM_SET_SIGNAL_MASK, which has worse scalability.
3929 Rather than blocking the signal outside KVM_RUN, userspace can set up
3930 a signal handler that sets run->immediate_exit to a non-zero value.
3932 This field is ignored if KVM_CAP_IMMEDIATE_EXIT is not available.
3939 When KVM_RUN has returned successfully (return value 0), this informs
3940 application code why KVM_RUN has returned. Allowable values for this
3941 field are detailed below.
3943 __u8 ready_for_interrupt_injection;
3945 If request_interrupt_window has been specified, this field indicates
3946 an interrupt can be injected now with KVM_INTERRUPT.
3950 The value of the current interrupt flag. Only valid if in-kernel
3951 local APIC is not used.
3955 More architecture-specific flags detailing state of the VCPU that may
3956 affect the device's behavior. The only currently defined flag is
3957 KVM_RUN_X86_SMM, which is valid on x86 machines and is set if the
3958 VCPU is in system management mode.
3960 /* in (pre_kvm_run), out (post_kvm_run) */
3963 The value of the cr8 register. Only valid if in-kernel local APIC is
3964 not used. Both input and output.
3968 The value of the APIC BASE msr. Only valid if in-kernel local
3969 APIC is not used. Both input and output.
3972 /* KVM_EXIT_UNKNOWN */
3974 __u64 hardware_exit_reason;
3977 If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown
3978 reasons. Further architecture-specific information is available in
3979 hardware_exit_reason.
3981 /* KVM_EXIT_FAIL_ENTRY */
3983 __u64 hardware_entry_failure_reason;
3986 If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due
3987 to unknown reasons. Further architecture-specific information is
3988 available in hardware_entry_failure_reason.
3990 /* KVM_EXIT_EXCEPTION */
4000 #define KVM_EXIT_IO_IN 0
4001 #define KVM_EXIT_IO_OUT 1
4003 __u8 size; /* bytes */
4006 __u64 data_offset; /* relative to kvm_run start */
4009 If exit_reason is KVM_EXIT_IO, then the vcpu has
4010 executed a port I/O instruction which could not be satisfied by kvm.
4011 data_offset describes where the data is located (KVM_EXIT_IO_OUT) or
4012 where kvm expects application code to place the data for the next
4013 KVM_RUN invocation (KVM_EXIT_IO_IN). Data format is a packed array.
4015 /* KVM_EXIT_DEBUG */
4017 struct kvm_debug_exit_arch arch;
4020 If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event
4021 for which architecture specific information is returned.
4031 If exit_reason is KVM_EXIT_MMIO, then the vcpu has
4032 executed a memory-mapped I/O instruction which could not be satisfied
4033 by kvm. The 'data' member contains the written data if 'is_write' is
4034 true, and should be filled by application code otherwise.
4036 The 'data' member contains, in its first 'len' bytes, the value as it would
4037 appear if the VCPU performed a load or store of the appropriate width directly
4040 NOTE: For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR and
4041 KVM_EXIT_EPR the corresponding
4042 operations are complete (and guest state is consistent) only after userspace
4043 has re-entered the kernel with KVM_RUN. The kernel side will first finish
4044 incomplete operations and then check for pending signals. Userspace
4045 can re-enter the guest with an unmasked signal pending to complete
4048 /* KVM_EXIT_HYPERCALL */
4057 Unused. This was once used for 'hypercall to userspace'. To implement
4058 such functionality, use KVM_EXIT_IO (x86) or KVM_EXIT_MMIO (all except s390).
4059 Note KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO.
4061 /* KVM_EXIT_TPR_ACCESS */
4068 To be documented (KVM_TPR_ACCESS_REPORTING).
4070 /* KVM_EXIT_S390_SIEIC */
4073 __u64 mask; /* psw upper half */
4074 __u64 addr; /* psw lower half */
4081 /* KVM_EXIT_S390_RESET */
4082 #define KVM_S390_RESET_POR 1
4083 #define KVM_S390_RESET_CLEAR 2
4084 #define KVM_S390_RESET_SUBSYSTEM 4
4085 #define KVM_S390_RESET_CPU_INIT 8
4086 #define KVM_S390_RESET_IPL 16
4087 __u64 s390_reset_flags;
4091 /* KVM_EXIT_S390_UCONTROL */
4093 __u64 trans_exc_code;
4097 s390 specific. A page fault has occurred for a user controlled virtual
4098 machine (KVM_VM_S390_UNCONTROL) on it's host page table that cannot be
4099 resolved by the kernel.
4100 The program code and the translation exception code that were placed
4101 in the cpu's lowcore are presented here as defined by the z Architecture
4102 Principles of Operation Book in the Chapter for Dynamic Address Translation
4112 Deprecated - was used for 440 KVM.
4119 MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch
4120 hypercalls and exit with this exit struct that contains all the guest gprs.
4122 If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall.
4123 Userspace can now handle the hypercall and when it's done modify the gprs as
4124 necessary. Upon guest entry all guest GPRs will then be replaced by the values
4127 /* KVM_EXIT_PAPR_HCALL */
4134 This is used on 64-bit PowerPC when emulating a pSeries partition,
4135 e.g. with the 'pseries' machine type in qemu. It occurs when the
4136 guest does a hypercall using the 'sc 1' instruction. The 'nr' field
4137 contains the hypercall number (from the guest R3), and 'args' contains
4138 the arguments (from the guest R4 - R12). Userspace should put the
4139 return code in 'ret' and any extra returned values in args[].
4140 The possible hypercalls are defined in the Power Architecture Platform
4141 Requirements (PAPR) document available from www.power.org (free
4142 developer registration required to access it).
4144 /* KVM_EXIT_S390_TSCH */
4146 __u16 subchannel_id;
4147 __u16 subchannel_nr;
4154 s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled
4155 and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O
4156 interrupt for the target subchannel has been dequeued and subchannel_id,
4157 subchannel_nr, io_int_parm and io_int_word contain the parameters for that
4158 interrupt. ipb is needed for instruction parameter decoding.
4165 On FSL BookE PowerPC chips, the interrupt controller has a fast patch
4166 interrupt acknowledge path to the core. When the core successfully
4167 delivers an interrupt, it automatically populates the EPR register with
4168 the interrupt vector number and acknowledges the interrupt inside
4169 the interrupt controller.
4171 In case the interrupt controller lives in user space, we need to do
4172 the interrupt acknowledge cycle through it to fetch the next to be
4173 delivered interrupt vector using this exit.
4175 It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an
4176 external interrupt has just been delivered into the guest. User space
4177 should put the acknowledged interrupt vector into the 'epr' field.
4179 /* KVM_EXIT_SYSTEM_EVENT */
4181 #define KVM_SYSTEM_EVENT_SHUTDOWN 1
4182 #define KVM_SYSTEM_EVENT_RESET 2
4183 #define KVM_SYSTEM_EVENT_CRASH 3
4188 If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered
4189 a system-level event using some architecture specific mechanism (hypercall
4190 or some special instruction). In case of ARM/ARM64, this is triggered using
4191 HVC instruction based PSCI call from the vcpu. The 'type' field describes
4192 the system-level event type. The 'flags' field describes architecture
4193 specific flags for the system-level event.
4195 Valid values for 'type' are:
4196 KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the
4197 VM. Userspace is not obliged to honour this, and if it does honour
4198 this does not need to destroy the VM synchronously (ie it may call
4199 KVM_RUN again before shutdown finally occurs).
4200 KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM.
4201 As with SHUTDOWN, userspace can choose to ignore the request, or
4202 to schedule the reset to occur in the future and may call KVM_RUN again.
4203 KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest
4204 has requested a crash condition maintenance. Userspace can choose
4205 to ignore the request, or to gather VM memory core dump and/or
4206 reset/shutdown of the VM.
4208 /* KVM_EXIT_IOAPIC_EOI */
4213 Indicates that the VCPU's in-kernel local APIC received an EOI for a
4214 level-triggered IOAPIC interrupt. This exit only triggers when the
4215 IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled);
4216 the userspace IOAPIC should process the EOI and retrigger the interrupt if
4217 it is still asserted. Vector is the LAPIC interrupt vector for which the
4220 struct kvm_hyperv_exit {
4221 #define KVM_EXIT_HYPERV_SYNIC 1
4222 #define KVM_EXIT_HYPERV_HCALL 2
4238 /* KVM_EXIT_HYPERV */
4239 struct kvm_hyperv_exit hyperv;
4240 Indicates that the VCPU exits into userspace to process some tasks
4241 related to Hyper-V emulation.
4242 Valid values for 'type' are:
4243 KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about
4244 Hyper-V SynIC state change. Notification is used to remap SynIC
4245 event/message pages and to enable/disable SynIC messages/events processing
4248 /* Fix the size of the union. */
4253 * shared registers between kvm and userspace.
4254 * kvm_valid_regs specifies the register classes set by the host
4255 * kvm_dirty_regs specified the register classes dirtied by userspace
4256 * struct kvm_sync_regs is architecture specific, as well as the
4257 * bits for kvm_valid_regs and kvm_dirty_regs
4259 __u64 kvm_valid_regs;
4260 __u64 kvm_dirty_regs;
4262 struct kvm_sync_regs regs;
4263 char padding[SYNC_REGS_SIZE_BYTES];
4266 If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access
4267 certain guest registers without having to call SET/GET_*REGS. Thus we can
4268 avoid some system call overhead if userspace has to handle the exit.
4269 Userspace can query the validity of the structure by checking
4270 kvm_valid_regs for specific bits. These bits are architecture specific
4271 and usually define the validity of a groups of registers. (e.g. one bit
4272 for general purpose registers)
4274 Please note that the kernel is allowed to use the kvm_run structure as the
4275 primary storage for certain register types. Therefore, the kernel may use the
4276 values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set.
4282 6. Capabilities that can be enabled on vCPUs
4283 --------------------------------------------
4285 There are certain capabilities that change the behavior of the virtual CPU or
4286 the virtual machine when enabled. To enable them, please see section 4.37.
4287 Below you can find a list of capabilities and what their effect on the vCPU or
4288 the virtual machine is when enabling them.
4290 The following information is provided along with the description:
4292 Architectures: which instruction set architectures provide this ioctl.
4293 x86 includes both i386 and x86_64.
4295 Target: whether this is a per-vcpu or per-vm capability.
4297 Parameters: what parameters are accepted by the capability.
4299 Returns: the return value. General error numbers (EBADF, ENOMEM, EINVAL)
4300 are not detailed, but errors with specific meanings are.
4308 Returns: 0 on success; -1 on error
4310 This capability enables interception of OSI hypercalls that otherwise would
4311 be treated as normal system calls to be injected into the guest. OSI hypercalls
4312 were invented by Mac-on-Linux to have a standardized communication mechanism
4313 between the guest and the host.
4315 When this capability is enabled, KVM_EXIT_OSI can occur.
4318 6.2 KVM_CAP_PPC_PAPR
4323 Returns: 0 on success; -1 on error
4325 This capability enables interception of PAPR hypercalls. PAPR hypercalls are
4326 done using the hypercall instruction "sc 1".
4328 It also sets the guest privilege level to "supervisor" mode. Usually the guest
4329 runs in "hypervisor" privilege mode with a few missing features.
4331 In addition to the above, it changes the semantics of SDR1. In this mode, the
4332 HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the
4333 HTAB invisible to the guest.
4335 When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur.
4342 Parameters: args[0] is the address of a struct kvm_config_tlb
4343 Returns: 0 on success; -1 on error
4345 struct kvm_config_tlb {
4352 Configures the virtual CPU's TLB array, establishing a shared memory area
4353 between userspace and KVM. The "params" and "array" fields are userspace
4354 addresses of mmu-type-specific data structures. The "array_len" field is an
4355 safety mechanism, and should be set to the size in bytes of the memory that
4356 userspace has reserved for the array. It must be at least the size dictated
4357 by "mmu_type" and "params".
4359 While KVM_RUN is active, the shared region is under control of KVM. Its
4360 contents are undefined, and any modification by userspace results in
4361 boundedly undefined behavior.
4363 On return from KVM_RUN, the shared region will reflect the current state of
4364 the guest's TLB. If userspace makes any changes, it must call KVM_DIRTY_TLB
4365 to tell KVM which entries have been changed, prior to calling KVM_RUN again
4368 For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV:
4369 - The "params" field is of type "struct kvm_book3e_206_tlb_params".
4370 - The "array" field points to an array of type "struct
4371 kvm_book3e_206_tlb_entry".
4372 - The array consists of all entries in the first TLB, followed by all
4373 entries in the second TLB.
4374 - Within a TLB, entries are ordered first by increasing set number. Within a
4375 set, entries are ordered by way (increasing ESEL).
4376 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1)
4377 where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value.
4378 - The tsize field of mas1 shall be set to 4K on TLB0, even though the
4379 hardware ignores this value for TLB0.
4381 6.4 KVM_CAP_S390_CSS_SUPPORT
4386 Returns: 0 on success; -1 on error
4388 This capability enables support for handling of channel I/O instructions.
4390 TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are
4391 handled in-kernel, while the other I/O instructions are passed to userspace.
4393 When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST
4394 SUBCHANNEL intercepts.
4396 Note that even though this capability is enabled per-vcpu, the complete
4397 virtual machine is affected.
4403 Parameters: args[0] defines whether the proxy facility is active
4404 Returns: 0 on success; -1 on error
4406 This capability enables or disables the delivery of interrupts through the
4407 external proxy facility.
4409 When enabled (args[0] != 0), every time the guest gets an external interrupt
4410 delivered, it automatically exits into user space with a KVM_EXIT_EPR exit
4411 to receive the topmost interrupt vector.
4413 When disabled (args[0] == 0), behavior is as if this facility is unsupported.
4415 When this capability is enabled, KVM_EXIT_EPR can occur.
4417 6.6 KVM_CAP_IRQ_MPIC
4420 Parameters: args[0] is the MPIC device fd
4421 args[1] is the MPIC CPU number for this vcpu
4423 This capability connects the vcpu to an in-kernel MPIC device.
4425 6.7 KVM_CAP_IRQ_XICS
4429 Parameters: args[0] is the XICS device fd
4430 args[1] is the XICS CPU number (server ID) for this vcpu
4432 This capability connects the vcpu to an in-kernel XICS device.
4434 6.8 KVM_CAP_S390_IRQCHIP
4440 This capability enables the in-kernel irqchip for s390. Please refer to
4441 "4.24 KVM_CREATE_IRQCHIP" for details.
4443 6.9 KVM_CAP_MIPS_FPU
4447 Parameters: args[0] is reserved for future use (should be 0).
4449 This capability allows the use of the host Floating Point Unit by the guest. It
4450 allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is
4451 done the KVM_REG_MIPS_FPR_* and KVM_REG_MIPS_FCR_* registers can be accessed
4452 (depending on the current guest FPU register mode), and the Status.FR,
4453 Config5.FRE bits are accessible via the KVM API and also from the guest,
4454 depending on them being supported by the FPU.
4456 6.10 KVM_CAP_MIPS_MSA
4460 Parameters: args[0] is reserved for future use (should be 0).
4462 This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest.
4463 It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest.
4464 Once this is done the KVM_REG_MIPS_VEC_* and KVM_REG_MIPS_MSA_* registers can be
4465 accessed, and the Config5.MSAEn bit is accessible via the KVM API and also from
4468 6.74 KVM_CAP_SYNC_REGS
4469 Architectures: s390, x86
4470 Target: s390: always enabled, x86: vcpu
4472 Returns: x86: KVM_CHECK_EXTENSION returns a bit-array indicating which register
4473 sets are supported (bitfields defined in arch/x86/include/uapi/asm/kvm.h).
4475 As described above in the kvm_sync_regs struct info in section 5 (kvm_run):
4476 KVM_CAP_SYNC_REGS "allow[s] userspace to access certain guest registers
4477 without having to call SET/GET_*REGS". This reduces overhead by eliminating
4478 repeated ioctl calls for setting and/or getting register values. This is
4479 particularly important when userspace is making synchronous guest state
4480 modifications, e.g. when emulating and/or intercepting instructions in
4483 For s390 specifics, please refer to the source code.
4486 - the register sets to be copied out to kvm_run are selectable
4487 by userspace (rather that all sets being copied out for every exit).
4488 - vcpu_events are available in addition to regs and sregs.
4490 For x86, the 'kvm_valid_regs' field of struct kvm_run is overloaded to
4491 function as an input bit-array field set by userspace to indicate the
4492 specific register sets to be copied out on the next exit.
4494 To indicate when userspace has modified values that should be copied into
4495 the vCPU, the all architecture bitarray field, 'kvm_dirty_regs' must be set.
4496 This is done using the same bitflags as for the 'kvm_valid_regs' field.
4497 If the dirty bit is not set, then the register set values will not be copied
4498 into the vCPU even if they've been modified.
4500 Unused bitfields in the bitarrays must be set to zero.
4502 struct kvm_sync_regs {
4503 struct kvm_regs regs;
4504 struct kvm_sregs sregs;
4505 struct kvm_vcpu_events events;
4508 7. Capabilities that can be enabled on VMs
4509 ------------------------------------------
4511 There are certain capabilities that change the behavior of the virtual
4512 machine when enabled. To enable them, please see section 4.37. Below
4513 you can find a list of capabilities and what their effect on the VM
4514 is when enabling them.
4516 The following information is provided along with the description:
4518 Architectures: which instruction set architectures provide this ioctl.
4519 x86 includes both i386 and x86_64.
4521 Parameters: what parameters are accepted by the capability.
4523 Returns: the return value. General error numbers (EBADF, ENOMEM, EINVAL)
4524 are not detailed, but errors with specific meanings are.
4527 7.1 KVM_CAP_PPC_ENABLE_HCALL
4530 Parameters: args[0] is the sPAPR hcall number
4531 args[1] is 0 to disable, 1 to enable in-kernel handling
4533 This capability controls whether individual sPAPR hypercalls (hcalls)
4534 get handled by the kernel or not. Enabling or disabling in-kernel
4535 handling of an hcall is effective across the VM. On creation, an
4536 initial set of hcalls are enabled for in-kernel handling, which
4537 consists of those hcalls for which in-kernel handlers were implemented
4538 before this capability was implemented. If disabled, the kernel will
4539 not to attempt to handle the hcall, but will always exit to userspace
4540 to handle it. Note that it may not make sense to enable some and
4541 disable others of a group of related hcalls, but KVM does not prevent
4542 userspace from doing that.
4544 If the hcall number specified is not one that has an in-kernel
4545 implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL
4548 7.2 KVM_CAP_S390_USER_SIGP
4553 This capability controls which SIGP orders will be handled completely in user
4554 space. With this capability enabled, all fast orders will be handled completely
4560 - CONDITIONAL EMERGENCY SIGNAL
4562 All other orders will be handled completely in user space.
4564 Only privileged operation exceptions will be checked for in the kernel (or even
4565 in the hardware prior to interception). If this capability is not enabled, the
4566 old way of handling SIGP orders is used (partially in kernel and user space).
4568 7.3 KVM_CAP_S390_VECTOR_REGISTERS
4572 Returns: 0 on success, negative value on error
4574 Allows use of the vector registers introduced with z13 processor, and
4575 provides for the synchronization between host and user space. Will
4576 return -EINVAL if the machine does not support vectors.
4578 7.4 KVM_CAP_S390_USER_STSI
4583 This capability allows post-handlers for the STSI instruction. After
4584 initial handling in the kernel, KVM exits to user space with
4585 KVM_EXIT_S390_STSI to allow user space to insert further data.
4587 Before exiting to userspace, kvm handlers should fill in s390_stsi field of
4598 @addr - guest address of STSI SYSIB
4602 @ar - access register number
4604 KVM handlers should exit to userspace with rc = -EREMOTE.
4606 7.5 KVM_CAP_SPLIT_IRQCHIP
4609 Parameters: args[0] - number of routes reserved for userspace IOAPICs
4610 Returns: 0 on success, -1 on error
4612 Create a local apic for each processor in the kernel. This can be used
4613 instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the
4614 IOAPIC and PIC (and also the PIT, even though this has to be enabled
4617 This capability also enables in kernel routing of interrupt requests;
4618 when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are
4619 used in the IRQ routing table. The first args[0] MSI routes are reserved
4620 for the IOAPIC pins. Whenever the LAPIC receives an EOI for these routes,
4621 a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace.
4623 Fails if VCPU has already been created, or if the irqchip is already in the
4624 kernel (i.e. KVM_CREATE_IRQCHIP has already been called).
4631 Allows use of runtime-instrumentation introduced with zEC12 processor.
4632 Will return -EINVAL if the machine does not support runtime-instrumentation.
4633 Will return -EBUSY if a VCPU has already been created.
4635 7.7 KVM_CAP_X2APIC_API
4638 Parameters: args[0] - features that should be enabled
4639 Returns: 0 on success, -EINVAL when args[0] contains invalid features
4641 Valid feature flags in args[0] are
4643 #define KVM_X2APIC_API_USE_32BIT_IDS (1ULL << 0)
4644 #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK (1ULL << 1)
4646 Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of
4647 KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC,
4648 allowing the use of 32-bit APIC IDs. See KVM_CAP_X2APIC_API in their
4649 respective sections.
4651 KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work
4652 in logical mode or with more than 255 VCPUs. Otherwise, KVM treats 0xff
4653 as a broadcast even in x2APIC mode in order to support physical x2APIC
4654 without interrupt remapping. This is undesirable in logical mode,
4655 where 0xff represents CPUs 0-7 in cluster 0.
4657 7.8 KVM_CAP_S390_USER_INSTR0
4662 With this capability enabled, all illegal instructions 0x0000 (2 bytes) will
4663 be intercepted and forwarded to user space. User space can use this
4664 mechanism e.g. to realize 2-byte software breakpoints. The kernel will
4665 not inject an operating exception for these instructions, user space has
4666 to take care of that.
4668 This capability can be enabled dynamically even if VCPUs were already
4669 created and are running.
4675 Returns: 0 on success; -EINVAL if the machine does not support
4676 guarded storage; -EBUSY if a VCPU has already been created.
4678 Allows use of guarded storage for the KVM guest.
4680 7.10 KVM_CAP_S390_AIS
4685 Allow use of adapter-interruption suppression.
4686 Returns: 0 on success; -EBUSY if a VCPU has already been created.
4688 7.11 KVM_CAP_PPC_SMT
4691 Parameters: vsmt_mode, flags
4693 Enabling this capability on a VM provides userspace with a way to set
4694 the desired virtual SMT mode (i.e. the number of virtual CPUs per
4695 virtual core). The virtual SMT mode, vsmt_mode, must be a power of 2
4696 between 1 and 8. On POWER8, vsmt_mode must also be no greater than
4697 the number of threads per subcore for the host. Currently flags must
4698 be 0. A successful call to enable this capability will result in
4699 vsmt_mode being returned when the KVM_CAP_PPC_SMT capability is
4700 subsequently queried for the VM. This capability is only supported by
4701 HV KVM, and can only be set before any VCPUs have been created.
4702 The KVM_CAP_PPC_SMT_POSSIBLE capability indicates which virtual SMT
4703 modes are available.
4705 7.12 KVM_CAP_PPC_FWNMI
4710 With this capability a machine check exception in the guest address
4711 space will cause KVM to exit the guest with NMI exit reason. This
4712 enables QEMU to build error log and branch to guest kernel registered
4713 machine check handling routine. Without this capability KVM will
4714 branch to guests' 0x200 interrupt vector.
4716 7.13 KVM_CAP_X86_DISABLE_EXITS
4719 Parameters: args[0] defines which exits are disabled
4720 Returns: 0 on success, -EINVAL when args[0] contains invalid exits
4722 Valid bits in args[0] are
4724 #define KVM_X86_DISABLE_EXITS_MWAIT (1 << 0)
4725 #define KVM_X86_DISABLE_EXITS_HLT (1 << 1)
4727 Enabling this capability on a VM provides userspace with a way to no
4728 longer intercept some instructions for improved latency in some
4729 workloads, and is suggested when vCPUs are associated to dedicated
4730 physical CPUs. More bits can be added in the future; userspace can
4731 just pass the KVM_CHECK_EXTENSION result to KVM_ENABLE_CAP to disable
4734 Do not enable KVM_FEATURE_PV_UNHALT if you disable HLT exits.
4736 7.14 KVM_CAP_S390_HPAGE_1M
4740 Returns: 0 on success, -EINVAL if hpage module parameter was not set
4741 or cmma is enabled, or the VM has the KVM_VM_S390_UCONTROL
4744 With this capability the KVM support for memory backing with 1m pages
4745 through hugetlbfs can be enabled for a VM. After the capability is
4746 enabled, cmma can't be enabled anymore and pfmfi and the storage key
4747 interpretation are disabled. If cmma has already been enabled or the
4748 hpage module parameter is not set to 1, -EINVAL is returned.
4750 While it is generally possible to create a huge page backed VM without
4751 this capability, the VM will not be able to run.
4753 7.15 KVM_CAP_MSR_PLATFORM_INFO
4756 Parameters: args[0] whether feature should be enabled or not
4758 With this capability, a guest may read the MSR_PLATFORM_INFO MSR. Otherwise,
4759 a #GP would be raised when the guest tries to access. Currently, this
4760 capability does not enable write permissions of this MSR for the guest.
4762 7.16 KVM_CAP_PPC_NESTED_HV
4766 Returns: 0 on success, -EINVAL when the implementation doesn't support
4767 nested-HV virtualization.
4769 HV-KVM on POWER9 and later systems allows for "nested-HV"
4770 virtualization, which provides a way for a guest VM to run guests that
4771 can run using the CPU's supervisor mode (privileged non-hypervisor
4772 state). Enabling this capability on a VM depends on the CPU having
4773 the necessary functionality and on the facility being enabled with a
4774 kvm-hv module parameter.
4776 7.17 KVM_CAP_EXCEPTION_PAYLOAD
4779 Parameters: args[0] whether feature should be enabled or not
4781 With this capability enabled, CR2 will not be modified prior to the
4782 emulated VM-exit when L1 intercepts a #PF exception that occurs in
4783 L2. Similarly, for kvm-intel only, DR6 will not be modified prior to
4784 the emulated VM-exit when L1 intercepts a #DB exception that occurs in
4785 L2. As a result, when KVM_GET_VCPU_EVENTS reports a pending #PF (or
4786 #DB) exception for L2, exception.has_payload will be set and the
4787 faulting address (or the new DR6 bits*) will be reported in the
4788 exception_payload field. Similarly, when userspace injects a #PF (or
4789 #DB) into L2 using KVM_SET_VCPU_EVENTS, it is expected to set
4790 exception.has_payload and to put the faulting address (or the new DR6
4791 bits*) in the exception_payload field.
4793 This capability also enables exception.pending in struct
4794 kvm_vcpu_events, which allows userspace to distinguish between pending
4795 and injected exceptions.
4798 * For the new DR6 bits, note that bit 16 is set iff the #DB exception
4801 7.18 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT
4804 Parameters: args[0] whether feature should be enabled or not
4806 With this capability enabled, KVM_GET_DIRTY_LOG will not automatically
4807 clear and write-protect all pages that are returned as dirty.
4808 Rather, userspace will have to do this operation separately using
4809 KVM_CLEAR_DIRTY_LOG.
4811 At the cost of a slightly more complicated operation, this provides better
4812 scalability and responsiveness for two reasons. First,
4813 KVM_CLEAR_DIRTY_LOG ioctl can operate on a 64-page granularity rather
4814 than requiring to sync a full memslot; this ensures that KVM does not
4815 take spinlocks for an extended period of time. Second, in some cases a
4816 large amount of time can pass between a call to KVM_GET_DIRTY_LOG and
4817 userspace actually using the data in the page. Pages can be modified
4818 during this time, which is inefficint for both the guest and userspace:
4819 the guest will incur a higher penalty due to write protection faults,
4820 while userspace can see false reports of dirty pages. Manual reprotection
4821 helps reducing this time, improving guest performance and reducing the
4822 number of dirty log false positives.
4825 8. Other capabilities.
4826 ----------------------
4828 This section lists capabilities that give information about other
4829 features of the KVM implementation.
4831 8.1 KVM_CAP_PPC_HWRNG
4835 This capability, if KVM_CHECK_EXTENSION indicates that it is
4836 available, means that that the kernel has an implementation of the
4837 H_RANDOM hypercall backed by a hardware random-number generator.
4838 If present, the kernel H_RANDOM handler can be enabled for guest use
4839 with the KVM_CAP_PPC_ENABLE_HCALL capability.
4841 8.2 KVM_CAP_HYPERV_SYNIC
4844 This capability, if KVM_CHECK_EXTENSION indicates that it is
4845 available, means that that the kernel has an implementation of the
4846 Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is
4847 used to support Windows Hyper-V based guest paravirt drivers(VMBus).
4849 In order to use SynIC, it has to be activated by setting this
4850 capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this
4851 will disable the use of APIC hardware virtualization even if supported
4852 by the CPU, as it's incompatible with SynIC auto-EOI behavior.
4854 8.3 KVM_CAP_PPC_RADIX_MMU
4858 This capability, if KVM_CHECK_EXTENSION indicates that it is
4859 available, means that that the kernel can support guests using the
4860 radix MMU defined in Power ISA V3.00 (as implemented in the POWER9
4863 8.4 KVM_CAP_PPC_HASH_MMU_V3
4867 This capability, if KVM_CHECK_EXTENSION indicates that it is
4868 available, means that that the kernel can support guests using the
4869 hashed page table MMU defined in Power ISA V3.00 (as implemented in
4870 the POWER9 processor), including in-memory segment tables.
4876 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
4877 it is available, means that full hardware assisted virtualization capabilities
4878 of the hardware are available for use through KVM. An appropriate
4879 KVM_VM_MIPS_* type must be passed to KVM_CREATE_VM to create a VM which
4882 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
4883 available, it means that the VM is using full hardware assisted virtualization
4884 capabilities of the hardware. This is useful to check after creating a VM with
4885 KVM_VM_MIPS_DEFAULT.
4887 The value returned by KVM_CHECK_EXTENSION should be compared against known
4888 values (see below). All other values are reserved. This is to allow for the
4889 possibility of other hardware assisted virtualization implementations which
4890 may be incompatible with the MIPS VZ ASE.
4892 0: The trap & emulate implementation is in use to run guest code in user
4893 mode. Guest virtual memory segments are rearranged to fit the guest in the
4894 user mode address space.
4896 1: The MIPS VZ ASE is in use, providing full hardware assisted
4897 virtualization, including standard guest virtual memory segments.
4903 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
4904 it is available, means that the trap & emulate implementation is available to
4905 run guest code in user mode, even if KVM_CAP_MIPS_VZ indicates that hardware
4906 assisted virtualisation is also available. KVM_VM_MIPS_TE (0) must be passed
4907 to KVM_CREATE_VM to create a VM which utilises it.
4909 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
4910 available, it means that the VM is using trap & emulate.
4912 8.7 KVM_CAP_MIPS_64BIT
4916 This capability indicates the supported architecture type of the guest, i.e. the
4917 supported register and address width.
4919 The values returned when this capability is checked by KVM_CHECK_EXTENSION on a
4920 kvm VM handle correspond roughly to the CP0_Config.AT register field, and should
4921 be checked specifically against known values (see below). All other values are
4924 0: MIPS32 or microMIPS32.
4925 Both registers and addresses are 32-bits wide.
4926 It will only be possible to run 32-bit guest code.
4928 1: MIPS64 or microMIPS64 with access only to 32-bit compatibility segments.
4929 Registers are 64-bits wide, but addresses are 32-bits wide.
4930 64-bit guest code may run but cannot access MIPS64 memory segments.
4931 It will also be possible to run 32-bit guest code.
4933 2: MIPS64 or microMIPS64 with access to all address segments.
4934 Both registers and addresses are 64-bits wide.
4935 It will be possible to run 64-bit or 32-bit guest code.
4937 8.9 KVM_CAP_ARM_USER_IRQ
4939 Architectures: arm, arm64
4940 This capability, if KVM_CHECK_EXTENSION indicates that it is available, means
4941 that if userspace creates a VM without an in-kernel interrupt controller, it
4942 will be notified of changes to the output level of in-kernel emulated devices,
4943 which can generate virtual interrupts, presented to the VM.
4944 For such VMs, on every return to userspace, the kernel
4945 updates the vcpu's run->s.regs.device_irq_level field to represent the actual
4946 output level of the device.
4948 Whenever kvm detects a change in the device output level, kvm guarantees at
4949 least one return to userspace before running the VM. This exit could either
4950 be a KVM_EXIT_INTR or any other exit event, like KVM_EXIT_MMIO. This way,
4951 userspace can always sample the device output level and re-compute the state of
4952 the userspace interrupt controller. Userspace should always check the state
4953 of run->s.regs.device_irq_level on every kvm exit.
4954 The value in run->s.regs.device_irq_level can represent both level and edge
4955 triggered interrupt signals, depending on the device. Edge triggered interrupt
4956 signals will exit to userspace with the bit in run->s.regs.device_irq_level
4957 set exactly once per edge signal.
4959 The field run->s.regs.device_irq_level is available independent of
4960 run->kvm_valid_regs or run->kvm_dirty_regs bits.
4962 If KVM_CAP_ARM_USER_IRQ is supported, the KVM_CHECK_EXTENSION ioctl returns a
4963 number larger than 0 indicating the version of this capability is implemented
4964 and thereby which bits in in run->s.regs.device_irq_level can signal values.
4966 Currently the following bits are defined for the device_irq_level bitmap:
4968 KVM_CAP_ARM_USER_IRQ >= 1:
4970 KVM_ARM_DEV_EL1_VTIMER - EL1 virtual timer
4971 KVM_ARM_DEV_EL1_PTIMER - EL1 physical timer
4972 KVM_ARM_DEV_PMU - ARM PMU overflow interrupt signal
4974 Future versions of kvm may implement additional events. These will get
4975 indicated by returning a higher number from KVM_CHECK_EXTENSION and will be
4978 8.10 KVM_CAP_PPC_SMT_POSSIBLE
4982 Querying this capability returns a bitmap indicating the possible
4983 virtual SMT modes that can be set using KVM_CAP_PPC_SMT. If bit N
4984 (counting from the right) is set, then a virtual SMT mode of 2^N is
4987 8.11 KVM_CAP_HYPERV_SYNIC2
4991 This capability enables a newer version of Hyper-V Synthetic interrupt
4992 controller (SynIC). The only difference with KVM_CAP_HYPERV_SYNIC is that KVM
4993 doesn't clear SynIC message and event flags pages when they are enabled by
4994 writing to the respective MSRs.
4996 8.12 KVM_CAP_HYPERV_VP_INDEX
5000 This capability indicates that userspace can load HV_X64_MSR_VP_INDEX msr. Its
5001 value is used to denote the target vcpu for a SynIC interrupt. For
5002 compatibilty, KVM initializes this msr to KVM's internal vcpu index. When this
5003 capability is absent, userspace can still query this msr's value.
5005 8.13 KVM_CAP_S390_AIS_MIGRATION
5010 This capability indicates if the flic device will be able to get/set the
5011 AIS states for migration via the KVM_DEV_FLIC_AISM_ALL attribute and allows
5012 to discover this without having to create a flic device.
5014 8.14 KVM_CAP_S390_PSW
5018 This capability indicates that the PSW is exposed via the kvm_run structure.
5020 8.15 KVM_CAP_S390_GMAP
5024 This capability indicates that the user space memory used as guest mapping can
5025 be anywhere in the user memory address space, as long as the memory slots are
5026 aligned and sized to a segment (1MB) boundary.
5028 8.16 KVM_CAP_S390_COW
5032 This capability indicates that the user space memory used as guest mapping can
5033 use copy-on-write semantics as well as dirty pages tracking via read-only page
5036 8.17 KVM_CAP_S390_BPB
5040 This capability indicates that kvm will implement the interfaces to handle
5041 reset, migration and nested KVM for branch prediction blocking. The stfle
5042 facility 82 should not be provided to the guest without this capability.
5044 8.18 KVM_CAP_HYPERV_TLBFLUSH
5048 This capability indicates that KVM supports paravirtualized Hyper-V TLB Flush
5050 HvFlushVirtualAddressSpace, HvFlushVirtualAddressSpaceEx,
5051 HvFlushVirtualAddressList, HvFlushVirtualAddressListEx.
5053 8.19 KVM_CAP_ARM_INJECT_SERROR_ESR
5055 Architectures: arm, arm64
5057 This capability indicates that userspace can specify (via the
5058 KVM_SET_VCPU_EVENTS ioctl) the syndrome value reported to the guest when it
5059 takes a virtual SError interrupt exception.
5060 If KVM advertises this capability, userspace can only specify the ISS field for
5061 the ESR syndrome. Other parts of the ESR, such as the EC are generated by the
5062 CPU when the exception is taken. If this virtual SError is taken to EL1 using
5063 AArch64, this value will be reported in the ISS field of ESR_ELx.
5065 See KVM_CAP_VCPU_EVENTS for more details.
5066 8.20 KVM_CAP_HYPERV_SEND_IPI
5070 This capability indicates that KVM supports paravirtualized Hyper-V IPI send
5072 HvCallSendSyntheticClusterIpi, HvCallSendSyntheticClusterIpiEx.