1 The Definitive KVM (Kernel-based Virtual Machine) API Documentation
2 ===================================================================
7 The kvm API is a set of ioctls that are issued to control various aspects
8 of a virtual machine. The ioctls belong to three classes
10 - System ioctls: These query and set global attributes which affect the
11 whole kvm subsystem. In addition a system ioctl is used to create
14 - VM ioctls: These query and set attributes that affect an entire virtual
15 machine, for example memory layout. In addition a VM ioctl is used to
16 create virtual cpus (vcpus).
18 Only run VM ioctls from the same process (address space) that was used
21 - vcpu ioctls: These query and set attributes that control the operation
22 of a single virtual cpu.
24 Only run vcpu ioctls from the same thread that was used to create the
31 The kvm API is centered around file descriptors. An initial
32 open("/dev/kvm") obtains a handle to the kvm subsystem; this handle
33 can be used to issue system ioctls. A KVM_CREATE_VM ioctl on this
34 handle will create a VM file descriptor which can be used to issue VM
35 ioctls. A KVM_CREATE_VCPU ioctl on a VM fd will create a virtual cpu
36 and return a file descriptor pointing to it. Finally, ioctls on a vcpu
37 fd can be used to control the vcpu, including the important task of
38 actually running guest code.
40 In general file descriptors can be migrated among processes by means
41 of fork() and the SCM_RIGHTS facility of unix domain socket. These
42 kinds of tricks are explicitly not supported by kvm. While they will
43 not cause harm to the host, their actual behavior is not guaranteed by
44 the API. The only supported use is one virtual machine per process,
45 and one vcpu per thread.
51 As of Linux 2.6.22, the KVM ABI has been stabilized: no backward
52 incompatible change are allowed. However, there is an extension
53 facility that allows backward-compatible extensions to the API to be
56 The extension mechanism is not based on the Linux version number.
57 Instead, kvm defines extension identifiers and a facility to query
58 whether a particular extension identifier is available. If it is, a
59 set of ioctls is available for application use.
65 This section describes ioctls that can be used to control kvm guests.
66 For each ioctl, the following information is provided along with a
69 Capability: which KVM extension provides this ioctl. Can be 'basic',
70 which means that is will be provided by any kernel that supports
71 API version 12 (see section 4.1), a KVM_CAP_xyz constant, which
72 means availability needs to be checked with KVM_CHECK_EXTENSION
73 (see section 4.4), or 'none' which means that while not all kernels
74 support this ioctl, there's no capability bit to check its
75 availability: for kernels that don't support the ioctl,
76 the ioctl returns -ENOTTY.
78 Architectures: which instruction set architectures provide this ioctl.
79 x86 includes both i386 and x86_64.
81 Type: system, vm, or vcpu.
83 Parameters: what parameters are accepted by the ioctl.
85 Returns: the return value. General error numbers (EBADF, ENOMEM, EINVAL)
86 are not detailed, but errors with specific meanings are.
89 4.1 KVM_GET_API_VERSION
95 Returns: the constant KVM_API_VERSION (=12)
97 This identifies the API version as the stable kvm API. It is not
98 expected that this number will change. However, Linux 2.6.20 and
99 2.6.21 report earlier versions; these are not documented and not
100 supported. Applications should refuse to run if KVM_GET_API_VERSION
101 returns a value other than 12. If this check passes, all ioctls
102 described as 'basic' will be available.
110 Parameters: machine type identifier (KVM_VM_*)
111 Returns: a VM fd that can be used to control the new virtual machine.
113 The new VM has no virtual cpus and no memory. An mmap() of a VM fd
114 will access the virtual machine's physical address space; offset zero
115 corresponds to guest physical address zero. Use of mmap() on a VM fd
116 is discouraged if userspace memory allocation (KVM_CAP_USER_MEMORY) is
118 You probably want to use 0 as machine type.
120 In order to create user controlled virtual machines on S390, check
121 KVM_CAP_S390_UCONTROL and use the flag KVM_VM_S390_UCONTROL as
122 privileged user (CAP_SYS_ADMIN).
124 To use hardware assisted virtualization on MIPS (VZ ASE) rather than
125 the default trap & emulate implementation (which changes the virtual
126 memory layout to fit in user mode), check KVM_CAP_MIPS_VZ and use the
130 4.3 KVM_GET_MSR_INDEX_LIST
135 Parameters: struct kvm_msr_list (in/out)
136 Returns: 0 on success; -1 on error
138 E2BIG: the msr index list is to be to fit in the array specified by
141 struct kvm_msr_list {
142 __u32 nmsrs; /* number of msrs in entries */
146 This ioctl returns the guest msrs that are supported. The list varies
147 by kvm version and host processor, but does not change otherwise. The
148 user fills in the size of the indices array in nmsrs, and in return
149 kvm adjusts nmsrs to reflect the actual number of msrs and fills in
150 the indices array with their numbers.
152 Note: if kvm indicates supports MCE (KVM_CAP_MCE), then the MCE bank MSRs are
153 not returned in the MSR list, as different vcpus can have a different number
154 of banks, as set via the KVM_X86_SETUP_MCE ioctl.
157 4.4 KVM_CHECK_EXTENSION
159 Capability: basic, KVM_CAP_CHECK_EXTENSION_VM for vm ioctl
161 Type: system ioctl, vm ioctl
162 Parameters: extension identifier (KVM_CAP_*)
163 Returns: 0 if unsupported; 1 (or some other positive integer) if supported
165 The API allows the application to query about extensions to the core
166 kvm API. Userspace passes an extension identifier (an integer) and
167 receives an integer that describes the extension availability.
168 Generally 0 means no and 1 means yes, but some extensions may report
169 additional information in the integer return value.
171 Based on their initialization different VMs may have different capabilities.
172 It is thus encouraged to use the vm ioctl to query for capabilities (available
173 with KVM_CAP_CHECK_EXTENSION_VM on the vm fd)
175 4.5 KVM_GET_VCPU_MMAP_SIZE
181 Returns: size of vcpu mmap area, in bytes
183 The KVM_RUN ioctl (cf.) communicates with userspace via a shared
184 memory region. This ioctl returns the size of that region. See the
185 KVM_RUN documentation for details.
188 4.6 KVM_SET_MEMORY_REGION
193 Parameters: struct kvm_memory_region (in)
194 Returns: 0 on success, -1 on error
196 This ioctl is obsolete and has been removed.
204 Parameters: vcpu id (apic id on x86)
205 Returns: vcpu fd on success, -1 on error
207 This API adds a vcpu to a virtual machine. No more than max_vcpus may be added.
208 The vcpu id is an integer in the range [0, max_vcpu_id).
210 The recommended max_vcpus value can be retrieved using the KVM_CAP_NR_VCPUS of
211 the KVM_CHECK_EXTENSION ioctl() at run-time.
212 The maximum possible value for max_vcpus can be retrieved using the
213 KVM_CAP_MAX_VCPUS of the KVM_CHECK_EXTENSION ioctl() at run-time.
215 If the KVM_CAP_NR_VCPUS does not exist, you should assume that max_vcpus is 4
217 If the KVM_CAP_MAX_VCPUS does not exist, you should assume that max_vcpus is
218 same as the value returned from KVM_CAP_NR_VCPUS.
220 The maximum possible value for max_vcpu_id can be retrieved using the
221 KVM_CAP_MAX_VCPU_ID of the KVM_CHECK_EXTENSION ioctl() at run-time.
223 If the KVM_CAP_MAX_VCPU_ID does not exist, you should assume that max_vcpu_id
224 is the same as the value returned from KVM_CAP_MAX_VCPUS.
226 On powerpc using book3s_hv mode, the vcpus are mapped onto virtual
227 threads in one or more virtual CPU cores. (This is because the
228 hardware requires all the hardware threads in a CPU core to be in the
229 same partition.) The KVM_CAP_PPC_SMT capability indicates the number
230 of vcpus per virtual core (vcore). The vcore id is obtained by
231 dividing the vcpu id by the number of vcpus per vcore. The vcpus in a
232 given vcore will always be in the same physical core as each other
233 (though that might be a different physical core from time to time).
234 Userspace can control the threading (SMT) mode of the guest by its
235 allocation of vcpu ids. For example, if userspace wants
236 single-threaded guest vcpus, it should make all vcpu ids be a multiple
237 of the number of vcpus per vcore.
239 For virtual cpus that have been created with S390 user controlled virtual
240 machines, the resulting vcpu fd can be memory mapped at page offset
241 KVM_S390_SIE_PAGE_OFFSET in order to obtain a memory map of the virtual
242 cpu's hardware control block.
245 4.8 KVM_GET_DIRTY_LOG (vm ioctl)
250 Parameters: struct kvm_dirty_log (in/out)
251 Returns: 0 on success, -1 on error
253 /* for KVM_GET_DIRTY_LOG */
254 struct kvm_dirty_log {
258 void __user *dirty_bitmap; /* one bit per page */
263 Given a memory slot, return a bitmap containing any pages dirtied
264 since the last call to this ioctl. Bit 0 is the first page in the
265 memory slot. Ensure the entire structure is cleared to avoid padding
268 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 specifies
269 the address space for which you want to return the dirty bitmap.
270 They must be less than the value that KVM_CHECK_EXTENSION returns for
271 the KVM_CAP_MULTI_ADDRESS_SPACE capability.
274 4.9 KVM_SET_MEMORY_ALIAS
279 Parameters: struct kvm_memory_alias (in)
280 Returns: 0 (success), -1 (error)
282 This ioctl is obsolete and has been removed.
291 Returns: 0 on success, -1 on error
293 EINTR: an unmasked signal is pending
295 This ioctl is used to run a guest virtual cpu. While there are no
296 explicit parameters, there is an implicit parameter block that can be
297 obtained by mmap()ing the vcpu fd at offset 0, with the size given by
298 KVM_GET_VCPU_MMAP_SIZE. The parameter block is formatted as a 'struct
299 kvm_run' (see below).
305 Architectures: all except ARM, arm64
307 Parameters: struct kvm_regs (out)
308 Returns: 0 on success, -1 on error
310 Reads the general purpose registers from the vcpu.
314 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
315 __u64 rax, rbx, rcx, rdx;
316 __u64 rsi, rdi, rsp, rbp;
317 __u64 r8, r9, r10, r11;
318 __u64 r12, r13, r14, r15;
324 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
335 Architectures: all except ARM, arm64
337 Parameters: struct kvm_regs (in)
338 Returns: 0 on success, -1 on error
340 Writes the general purpose registers into the vcpu.
342 See KVM_GET_REGS for the data structure.
348 Architectures: x86, ppc
350 Parameters: struct kvm_sregs (out)
351 Returns: 0 on success, -1 on error
353 Reads special registers from the vcpu.
357 struct kvm_segment cs, ds, es, fs, gs, ss;
358 struct kvm_segment tr, ldt;
359 struct kvm_dtable gdt, idt;
360 __u64 cr0, cr2, cr3, cr4, cr8;
363 __u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64];
366 /* ppc -- see arch/powerpc/include/uapi/asm/kvm.h */
368 interrupt_bitmap is a bitmap of pending external interrupts. At most
369 one bit may be set. This interrupt has been acknowledged by the APIC
370 but not yet injected into the cpu core.
376 Architectures: x86, ppc
378 Parameters: struct kvm_sregs (in)
379 Returns: 0 on success, -1 on error
381 Writes special registers into the vcpu. See KVM_GET_SREGS for the
390 Parameters: struct kvm_translation (in/out)
391 Returns: 0 on success, -1 on error
393 Translates a virtual address according to the vcpu's current address
396 struct kvm_translation {
398 __u64 linear_address;
401 __u64 physical_address;
412 Architectures: x86, ppc, mips
414 Parameters: struct kvm_interrupt (in)
415 Returns: 0 on success, negative on failure.
417 Queues a hardware interrupt vector to be injected.
419 /* for KVM_INTERRUPT */
420 struct kvm_interrupt {
427 Returns: 0 on success,
428 -EEXIST if an interrupt is already enqueued
429 -EINVAL the the irq number is invalid
430 -ENXIO if the PIC is in the kernel
431 -EFAULT if the pointer is invalid
433 Note 'irq' is an interrupt vector, not an interrupt pin or line. This
434 ioctl is useful if the in-kernel PIC is not used.
438 Queues an external interrupt to be injected. This ioctl is overleaded
439 with 3 different irq values:
443 This injects an edge type external interrupt into the guest once it's ready
444 to receive interrupts. When injected, the interrupt is done.
446 b) KVM_INTERRUPT_UNSET
448 This unsets any pending interrupt.
450 Only available with KVM_CAP_PPC_UNSET_IRQ.
452 c) KVM_INTERRUPT_SET_LEVEL
454 This injects a level type external interrupt into the guest context. The
455 interrupt stays pending until a specific ioctl with KVM_INTERRUPT_UNSET
458 Only available with KVM_CAP_PPC_IRQ_LEVEL.
460 Note that any value for 'irq' other than the ones stated above is invalid
461 and incurs unexpected behavior.
465 Queues an external interrupt to be injected into the virtual CPU. A negative
466 interrupt number dequeues the interrupt.
477 Support for this has been removed. Use KVM_SET_GUEST_DEBUG instead.
485 Parameters: struct kvm_msrs (in/out)
486 Returns: 0 on success, -1 on error
488 Reads model-specific registers from the vcpu. Supported msr indices can
489 be obtained using KVM_GET_MSR_INDEX_LIST.
492 __u32 nmsrs; /* number of msrs in entries */
495 struct kvm_msr_entry entries[0];
498 struct kvm_msr_entry {
504 Application code should set the 'nmsrs' member (which indicates the
505 size of the entries array) and the 'index' member of each array entry.
506 kvm will fill in the 'data' member.
514 Parameters: struct kvm_msrs (in)
515 Returns: 0 on success, -1 on error
517 Writes model-specific registers to the vcpu. See KVM_GET_MSRS for the
520 Application code should set the 'nmsrs' member (which indicates the
521 size of the entries array), and the 'index' and 'data' members of each
530 Parameters: struct kvm_cpuid (in)
531 Returns: 0 on success, -1 on error
533 Defines the vcpu responses to the cpuid instruction. Applications
534 should use the KVM_SET_CPUID2 ioctl if available.
537 struct kvm_cpuid_entry {
546 /* for KVM_SET_CPUID */
550 struct kvm_cpuid_entry entries[0];
554 4.21 KVM_SET_SIGNAL_MASK
559 Parameters: struct kvm_signal_mask (in)
560 Returns: 0 on success, -1 on error
562 Defines which signals are blocked during execution of KVM_RUN. This
563 signal mask temporarily overrides the threads signal mask. Any
564 unblocked signal received (except SIGKILL and SIGSTOP, which retain
565 their traditional behaviour) will cause KVM_RUN to return with -EINTR.
567 Note the signal will only be delivered if not blocked by the original
570 /* for KVM_SET_SIGNAL_MASK */
571 struct kvm_signal_mask {
582 Parameters: struct kvm_fpu (out)
583 Returns: 0 on success, -1 on error
585 Reads the floating point state from the vcpu.
587 /* for KVM_GET_FPU and KVM_SET_FPU */
592 __u8 ftwx; /* in fxsave format */
608 Parameters: struct kvm_fpu (in)
609 Returns: 0 on success, -1 on error
611 Writes the floating point state to the vcpu.
613 /* for KVM_GET_FPU and KVM_SET_FPU */
618 __u8 ftwx; /* in fxsave format */
629 4.24 KVM_CREATE_IRQCHIP
631 Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQCHIP (s390)
632 Architectures: x86, ARM, arm64, s390
635 Returns: 0 on success, -1 on error
637 Creates an interrupt controller model in the kernel.
638 On x86, creates a virtual ioapic, a virtual PIC (two PICs, nested), and sets up
639 future vcpus to have a local APIC. IRQ routing for GSIs 0-15 is set to both
640 PIC and IOAPIC; GSI 16-23 only go to the IOAPIC.
641 On ARM/arm64, a GICv2 is created. Any other GIC versions require the usage of
642 KVM_CREATE_DEVICE, which also supports creating a GICv2. Using
643 KVM_CREATE_DEVICE is preferred over KVM_CREATE_IRQCHIP for GICv2.
644 On s390, a dummy irq routing table is created.
646 Note that on s390 the KVM_CAP_S390_IRQCHIP vm capability needs to be enabled
647 before KVM_CREATE_IRQCHIP can be used.
652 Capability: KVM_CAP_IRQCHIP
653 Architectures: x86, arm, arm64
655 Parameters: struct kvm_irq_level
656 Returns: 0 on success, -1 on error
658 Sets the level of a GSI input to the interrupt controller model in the kernel.
659 On some architectures it is required that an interrupt controller model has
660 been previously created with KVM_CREATE_IRQCHIP. Note that edge-triggered
661 interrupts require the level to be set to 1 and then back to 0.
663 On real hardware, interrupt pins can be active-low or active-high. This
664 does not matter for the level field of struct kvm_irq_level: 1 always
665 means active (asserted), 0 means inactive (deasserted).
667 x86 allows the operating system to program the interrupt polarity
668 (active-low/active-high) for level-triggered interrupts, and KVM used
669 to consider the polarity. However, due to bitrot in the handling of
670 active-low interrupts, the above convention is now valid on x86 too.
671 This is signaled by KVM_CAP_X86_IOAPIC_POLARITY_IGNORED. Userspace
672 should not present interrupts to the guest as active-low unless this
673 capability is present (or unless it is not using the in-kernel irqchip,
677 ARM/arm64 can signal an interrupt either at the CPU level, or at the
678 in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to
679 use PPIs designated for specific cpus. The irq field is interpreted
682 Â bits: | 31 ... 24 | 23 ... 16 | 15 ... 0 |
683 field: | irq_type | vcpu_index | irq_id |
685 The irq_type field has the following values:
686 - irq_type[0]: out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ
687 - irq_type[1]: in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.)
688 (the vcpu_index field is ignored)
689 - irq_type[2]: in-kernel GIC: PPI, irq_id between 16 and 31 (incl.)
691 (The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs)
693 In both cases, level is used to assert/deassert the line.
695 struct kvm_irq_level {
698 __s32 status; /* not used for KVM_IRQ_LEVEL */
700 __u32 level; /* 0 or 1 */
706 Capability: KVM_CAP_IRQCHIP
709 Parameters: struct kvm_irqchip (in/out)
710 Returns: 0 on success, -1 on error
712 Reads the state of a kernel interrupt controller created with
713 KVM_CREATE_IRQCHIP into a buffer provided by the caller.
716 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
719 char dummy[512]; /* reserving space */
720 struct kvm_pic_state pic;
721 struct kvm_ioapic_state ioapic;
728 Capability: KVM_CAP_IRQCHIP
731 Parameters: struct kvm_irqchip (in)
732 Returns: 0 on success, -1 on error
734 Sets the state of a kernel interrupt controller created with
735 KVM_CREATE_IRQCHIP from a buffer provided by the caller.
738 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
741 char dummy[512]; /* reserving space */
742 struct kvm_pic_state pic;
743 struct kvm_ioapic_state ioapic;
748 4.28 KVM_XEN_HVM_CONFIG
750 Capability: KVM_CAP_XEN_HVM
753 Parameters: struct kvm_xen_hvm_config (in)
754 Returns: 0 on success, -1 on error
756 Sets the MSR that the Xen HVM guest uses to initialize its hypercall
757 page, and provides the starting address and size of the hypercall
758 blobs in userspace. When the guest writes the MSR, kvm copies one
759 page of a blob (32- or 64-bit, depending on the vcpu mode) to guest
762 struct kvm_xen_hvm_config {
775 Capability: KVM_CAP_ADJUST_CLOCK
778 Parameters: struct kvm_clock_data (out)
779 Returns: 0 on success, -1 on error
781 Gets the current timestamp of kvmclock as seen by the current guest. In
782 conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios
785 When KVM_CAP_ADJUST_CLOCK is passed to KVM_CHECK_EXTENSION, it returns the
786 set of bits that KVM can return in struct kvm_clock_data's flag member.
788 The only flag defined now is KVM_CLOCK_TSC_STABLE. If set, the returned
789 value is the exact kvmclock value seen by all VCPUs at the instant
790 when KVM_GET_CLOCK was called. If clear, the returned value is simply
791 CLOCK_MONOTONIC plus a constant offset; the offset can be modified
792 with KVM_SET_CLOCK. KVM will try to make all VCPUs follow this clock,
793 but the exact value read by each VCPU could differ, because the host
796 struct kvm_clock_data {
797 __u64 clock; /* kvmclock current value */
805 Capability: KVM_CAP_ADJUST_CLOCK
808 Parameters: struct kvm_clock_data (in)
809 Returns: 0 on success, -1 on error
811 Sets the current timestamp of kvmclock to the value specified in its parameter.
812 In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios
815 struct kvm_clock_data {
816 __u64 clock; /* kvmclock current value */
822 4.31 KVM_GET_VCPU_EVENTS
824 Capability: KVM_CAP_VCPU_EVENTS
825 Extended by: KVM_CAP_INTR_SHADOW
828 Parameters: struct kvm_vcpu_event (out)
829 Returns: 0 on success, -1 on error
831 Gets currently pending exceptions, interrupts, and NMIs as well as related
834 struct kvm_vcpu_events {
864 Only two fields are defined in the flags field:
866 - KVM_VCPUEVENT_VALID_SHADOW may be set in the flags field to signal that
867 interrupt.shadow contains a valid state.
869 - KVM_VCPUEVENT_VALID_SMM may be set in the flags field to signal that
870 smi contains a valid state.
872 4.32 KVM_SET_VCPU_EVENTS
874 Capability: KVM_CAP_VCPU_EVENTS
875 Extended by: KVM_CAP_INTR_SHADOW
878 Parameters: struct kvm_vcpu_event (in)
879 Returns: 0 on success, -1 on error
881 Set pending exceptions, interrupts, and NMIs as well as related states of the
884 See KVM_GET_VCPU_EVENTS for the data structure.
886 Fields that may be modified asynchronously by running VCPUs can be excluded
887 from the update. These fields are nmi.pending, sipi_vector, smi.smm,
888 smi.pending. Keep the corresponding bits in the flags field cleared to
889 suppress overwriting the current in-kernel state. The bits are:
891 KVM_VCPUEVENT_VALID_NMI_PENDING - transfer nmi.pending to the kernel
892 KVM_VCPUEVENT_VALID_SIPI_VECTOR - transfer sipi_vector
893 KVM_VCPUEVENT_VALID_SMM - transfer the smi sub-struct.
895 If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in
896 the flags field to signal that interrupt.shadow contains a valid state and
897 shall be written into the VCPU.
899 KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available.
902 4.33 KVM_GET_DEBUGREGS
904 Capability: KVM_CAP_DEBUGREGS
907 Parameters: struct kvm_debugregs (out)
908 Returns: 0 on success, -1 on error
910 Reads debug registers from the vcpu.
912 struct kvm_debugregs {
921 4.34 KVM_SET_DEBUGREGS
923 Capability: KVM_CAP_DEBUGREGS
926 Parameters: struct kvm_debugregs (in)
927 Returns: 0 on success, -1 on error
929 Writes debug registers into the vcpu.
931 See KVM_GET_DEBUGREGS for the data structure. The flags field is unused
932 yet and must be cleared on entry.
935 4.35 KVM_SET_USER_MEMORY_REGION
937 Capability: KVM_CAP_USER_MEM
940 Parameters: struct kvm_userspace_memory_region (in)
941 Returns: 0 on success, -1 on error
943 struct kvm_userspace_memory_region {
946 __u64 guest_phys_addr;
947 __u64 memory_size; /* bytes */
948 __u64 userspace_addr; /* start of the userspace allocated memory */
951 /* for kvm_memory_region::flags */
952 #define KVM_MEM_LOG_DIRTY_PAGES (1UL << 0)
953 #define KVM_MEM_READONLY (1UL << 1)
955 This ioctl allows the user to create or modify a guest physical memory
956 slot. When changing an existing slot, it may be moved in the guest
957 physical memory space, or its flags may be modified. It may not be
958 resized. Slots may not overlap in guest physical address space.
959 Bits 0-15 of "slot" specifies the slot id and this value should be
960 less than the maximum number of user memory slots supported per VM.
961 The maximum allowed slots can be queried using KVM_CAP_NR_MEMSLOTS,
962 if this capability is supported by the architecture.
964 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot"
965 specifies the address space which is being modified. They must be
966 less than the value that KVM_CHECK_EXTENSION returns for the
967 KVM_CAP_MULTI_ADDRESS_SPACE capability. Slots in separate address spaces
968 are unrelated; the restriction on overlapping slots only applies within
971 Memory for the region is taken starting at the address denoted by the
972 field userspace_addr, which must point at user addressable memory for
973 the entire memory slot size. Any object may back this memory, including
974 anonymous memory, ordinary files, and hugetlbfs.
976 It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr
977 be identical. This allows large pages in the guest to be backed by large
980 The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and
981 KVM_MEM_READONLY. The former can be set to instruct KVM to keep track of
982 writes to memory within the slot. See KVM_GET_DIRTY_LOG ioctl to know how to
983 use it. The latter can be set, if KVM_CAP_READONLY_MEM capability allows it,
984 to make a new slot read-only. In this case, writes to this memory will be
985 posted to userspace as KVM_EXIT_MMIO exits.
987 When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of
988 the memory region are automatically reflected into the guest. For example, an
989 mmap() that affects the region will be made visible immediately. Another
990 example is madvise(MADV_DROP).
992 It is recommended to use this API instead of the KVM_SET_MEMORY_REGION ioctl.
993 The KVM_SET_MEMORY_REGION does not allow fine grained control over memory
994 allocation and is deprecated.
997 4.36 KVM_SET_TSS_ADDR
999 Capability: KVM_CAP_SET_TSS_ADDR
1002 Parameters: unsigned long tss_address (in)
1003 Returns: 0 on success, -1 on error
1005 This ioctl defines the physical address of a three-page region in the guest
1006 physical address space. The region must be within the first 4GB of the
1007 guest physical address space and must not conflict with any memory slot
1008 or any mmio address. The guest may malfunction if it accesses this memory
1011 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1012 because of a quirk in the virtualization implementation (see the internals
1013 documentation when it pops into existence).
1018 Capability: KVM_CAP_ENABLE_CAP, KVM_CAP_ENABLE_CAP_VM
1019 Architectures: x86 (only KVM_CAP_ENABLE_CAP_VM),
1020 mips (only KVM_CAP_ENABLE_CAP), ppc, s390
1021 Type: vcpu ioctl, vm ioctl (with KVM_CAP_ENABLE_CAP_VM)
1022 Parameters: struct kvm_enable_cap (in)
1023 Returns: 0 on success; -1 on error
1025 +Not all extensions are enabled by default. Using this ioctl the application
1026 can enable an extension, making it available to the guest.
1028 On systems that do not support this ioctl, it always fails. On systems that
1029 do support it, it only works for extensions that are supported for enablement.
1031 To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should
1034 struct kvm_enable_cap {
1038 The capability that is supposed to get enabled.
1042 A bitfield indicating future enhancements. Has to be 0 for now.
1046 Arguments for enabling a feature. If a feature needs initial values to
1047 function properly, this is the place to put them.
1052 The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl
1053 for vm-wide capabilities.
1055 4.38 KVM_GET_MP_STATE
1057 Capability: KVM_CAP_MP_STATE
1058 Architectures: x86, s390, arm, arm64
1060 Parameters: struct kvm_mp_state (out)
1061 Returns: 0 on success; -1 on error
1063 struct kvm_mp_state {
1067 Returns the vcpu's current "multiprocessing state" (though also valid on
1068 uniprocessor guests).
1070 Possible values are:
1072 - KVM_MP_STATE_RUNNABLE: the vcpu is currently running [x86,arm/arm64]
1073 - KVM_MP_STATE_UNINITIALIZED: the vcpu is an application processor (AP)
1074 which has not yet received an INIT signal [x86]
1075 - KVM_MP_STATE_INIT_RECEIVED: the vcpu has received an INIT signal, and is
1076 now ready for a SIPI [x86]
1077 - KVM_MP_STATE_HALTED: the vcpu has executed a HLT instruction and
1078 is waiting for an interrupt [x86]
1079 - KVM_MP_STATE_SIPI_RECEIVED: the vcpu has just received a SIPI (vector
1080 accessible via KVM_GET_VCPU_EVENTS) [x86]
1081 - KVM_MP_STATE_STOPPED: the vcpu is stopped [s390,arm/arm64]
1082 - KVM_MP_STATE_CHECK_STOP: the vcpu is in a special error state [s390]
1083 - KVM_MP_STATE_OPERATING: the vcpu is operating (running or halted)
1085 - KVM_MP_STATE_LOAD: the vcpu is in a special load/startup state
1088 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1089 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1090 these architectures.
1094 The only states that are valid are KVM_MP_STATE_STOPPED and
1095 KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not.
1097 4.39 KVM_SET_MP_STATE
1099 Capability: KVM_CAP_MP_STATE
1100 Architectures: x86, s390, arm, arm64
1102 Parameters: struct kvm_mp_state (in)
1103 Returns: 0 on success; -1 on error
1105 Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for
1108 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1109 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1110 these architectures.
1114 The only states that are valid are KVM_MP_STATE_STOPPED and
1115 KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not.
1117 4.40 KVM_SET_IDENTITY_MAP_ADDR
1119 Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR
1122 Parameters: unsigned long identity (in)
1123 Returns: 0 on success, -1 on error
1125 This ioctl defines the physical address of a one-page region in the guest
1126 physical address space. The region must be within the first 4GB of the
1127 guest physical address space and must not conflict with any memory slot
1128 or any mmio address. The guest may malfunction if it accesses this memory
1131 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1132 because of a quirk in the virtualization implementation (see the internals
1133 documentation when it pops into existence).
1136 4.41 KVM_SET_BOOT_CPU_ID
1138 Capability: KVM_CAP_SET_BOOT_CPU_ID
1141 Parameters: unsigned long vcpu_id
1142 Returns: 0 on success, -1 on error
1144 Define which vcpu is the Bootstrap Processor (BSP). Values are the same
1145 as the vcpu id in KVM_CREATE_VCPU. If this ioctl is not called, the default
1151 Capability: KVM_CAP_XSAVE
1154 Parameters: struct kvm_xsave (out)
1155 Returns: 0 on success, -1 on error
1161 This ioctl would copy current vcpu's xsave struct to the userspace.
1166 Capability: KVM_CAP_XSAVE
1169 Parameters: struct kvm_xsave (in)
1170 Returns: 0 on success, -1 on error
1176 This ioctl would copy userspace's xsave struct to the kernel.
1181 Capability: KVM_CAP_XCRS
1184 Parameters: struct kvm_xcrs (out)
1185 Returns: 0 on success, -1 on error
1196 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1200 This ioctl would copy current vcpu's xcrs to the userspace.
1205 Capability: KVM_CAP_XCRS
1208 Parameters: struct kvm_xcrs (in)
1209 Returns: 0 on success, -1 on error
1220 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1224 This ioctl would set vcpu's xcr to the value userspace specified.
1227 4.46 KVM_GET_SUPPORTED_CPUID
1229 Capability: KVM_CAP_EXT_CPUID
1232 Parameters: struct kvm_cpuid2 (in/out)
1233 Returns: 0 on success, -1 on error
1238 struct kvm_cpuid_entry2 entries[0];
1241 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
1242 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1)
1243 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2)
1245 struct kvm_cpuid_entry2 {
1256 This ioctl returns x86 cpuid features which are supported by both the hardware
1257 and kvm. Userspace can use the information returned by this ioctl to
1258 construct cpuid information (for KVM_SET_CPUID2) that is consistent with
1259 hardware, kernel, and userspace capabilities, and with user requirements (for
1260 example, the user may wish to constrain cpuid to emulate older hardware,
1261 or for feature consistency across a cluster).
1263 Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure
1264 with the 'nent' field indicating the number of entries in the variable-size
1265 array 'entries'. If the number of entries is too low to describe the cpu
1266 capabilities, an error (E2BIG) is returned. If the number is too high,
1267 the 'nent' field is adjusted and an error (ENOMEM) is returned. If the
1268 number is just right, the 'nent' field is adjusted to the number of valid
1269 entries in the 'entries' array, which is then filled.
1271 The entries returned are the host cpuid as returned by the cpuid instruction,
1272 with unknown or unsupported features masked out. Some features (for example,
1273 x2apic), may not be present in the host cpu, but are exposed by kvm if it can
1274 emulate them efficiently. The fields in each entry are defined as follows:
1276 function: the eax value used to obtain the entry
1277 index: the ecx value used to obtain the entry (for entries that are
1279 flags: an OR of zero or more of the following:
1280 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
1281 if the index field is valid
1282 KVM_CPUID_FLAG_STATEFUL_FUNC:
1283 if cpuid for this function returns different values for successive
1284 invocations; there will be several entries with the same function,
1285 all with this flag set
1286 KVM_CPUID_FLAG_STATE_READ_NEXT:
1287 for KVM_CPUID_FLAG_STATEFUL_FUNC entries, set if this entry is
1288 the first entry to be read by a cpu
1289 eax, ebx, ecx, edx: the values returned by the cpuid instruction for
1290 this function/index combination
1292 The TSC deadline timer feature (CPUID leaf 1, ecx[24]) is always returned
1293 as false, since the feature depends on KVM_CREATE_IRQCHIP for local APIC
1294 support. Instead it is reported via
1296 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER)
1298 if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the
1299 feature in userspace, then you can enable the feature for KVM_SET_CPUID2.
1302 4.47 KVM_PPC_GET_PVINFO
1304 Capability: KVM_CAP_PPC_GET_PVINFO
1307 Parameters: struct kvm_ppc_pvinfo (out)
1308 Returns: 0 on success, !0 on error
1310 struct kvm_ppc_pvinfo {
1316 This ioctl fetches PV specific information that need to be passed to the guest
1317 using the device tree or other means from vm context.
1319 The hcall array defines 4 instructions that make up a hypercall.
1321 If any additional field gets added to this structure later on, a bit for that
1322 additional piece of information will be set in the flags bitmap.
1324 The flags bitmap is defined as:
1326 /* the host supports the ePAPR idle hcall
1327 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1<<0)
1329 4.48 KVM_ASSIGN_PCI_DEVICE (deprecated)
1334 Parameters: struct kvm_assigned_pci_dev (in)
1335 Returns: 0 on success, -1 on error
1337 Assigns a host PCI device to the VM.
1339 struct kvm_assigned_pci_dev {
1340 __u32 assigned_dev_id;
1350 The PCI device is specified by the triple segnr, busnr, and devfn.
1351 Identification in succeeding service requests is done via assigned_dev_id. The
1352 following flags are specified:
1354 /* Depends on KVM_CAP_IOMMU */
1355 #define KVM_DEV_ASSIGN_ENABLE_IOMMU (1 << 0)
1356 /* The following two depend on KVM_CAP_PCI_2_3 */
1357 #define KVM_DEV_ASSIGN_PCI_2_3 (1 << 1)
1358 #define KVM_DEV_ASSIGN_MASK_INTX (1 << 2)
1360 If KVM_DEV_ASSIGN_PCI_2_3 is set, the kernel will manage legacy INTx interrupts
1361 via the PCI-2.3-compliant device-level mask, thus enable IRQ sharing with other
1362 assigned devices or host devices. KVM_DEV_ASSIGN_MASK_INTX specifies the
1363 guest's view on the INTx mask, see KVM_ASSIGN_SET_INTX_MASK for details.
1365 The KVM_DEV_ASSIGN_ENABLE_IOMMU flag is a mandatory option to ensure
1366 isolation of the device. Usages not specifying this flag are deprecated.
1368 Only PCI header type 0 devices with PCI BAR resources are supported by
1369 device assignment. The user requesting this ioctl must have read/write
1370 access to the PCI sysfs resource files associated with the device.
1373 ENOTTY: kernel does not support this ioctl
1375 Other error conditions may be defined by individual device types or
1376 have their standard meanings.
1379 4.49 KVM_DEASSIGN_PCI_DEVICE (deprecated)
1384 Parameters: struct kvm_assigned_pci_dev (in)
1385 Returns: 0 on success, -1 on error
1387 Ends PCI device assignment, releasing all associated resources.
1389 See KVM_ASSIGN_PCI_DEVICE for the data structure. Only assigned_dev_id is
1390 used in kvm_assigned_pci_dev to identify the device.
1393 ENOTTY: kernel does not support this ioctl
1395 Other error conditions may be defined by individual device types or
1396 have their standard meanings.
1398 4.50 KVM_ASSIGN_DEV_IRQ (deprecated)
1400 Capability: KVM_CAP_ASSIGN_DEV_IRQ
1403 Parameters: struct kvm_assigned_irq (in)
1404 Returns: 0 on success, -1 on error
1406 Assigns an IRQ to a passed-through device.
1408 struct kvm_assigned_irq {
1409 __u32 assigned_dev_id;
1410 __u32 host_irq; /* ignored (legacy field) */
1418 The following flags are defined:
1420 #define KVM_DEV_IRQ_HOST_INTX (1 << 0)
1421 #define KVM_DEV_IRQ_HOST_MSI (1 << 1)
1422 #define KVM_DEV_IRQ_HOST_MSIX (1 << 2)
1424 #define KVM_DEV_IRQ_GUEST_INTX (1 << 8)
1425 #define KVM_DEV_IRQ_GUEST_MSI (1 << 9)
1426 #define KVM_DEV_IRQ_GUEST_MSIX (1 << 10)
1428 It is not valid to specify multiple types per host or guest IRQ. However, the
1429 IRQ type of host and guest can differ or can even be null.
1432 ENOTTY: kernel does not support this ioctl
1434 Other error conditions may be defined by individual device types or
1435 have their standard meanings.
1438 4.51 KVM_DEASSIGN_DEV_IRQ (deprecated)
1440 Capability: KVM_CAP_ASSIGN_DEV_IRQ
1443 Parameters: struct kvm_assigned_irq (in)
1444 Returns: 0 on success, -1 on error
1446 Ends an IRQ assignment to a passed-through device.
1448 See KVM_ASSIGN_DEV_IRQ for the data structure. The target device is specified
1449 by assigned_dev_id, flags must correspond to the IRQ type specified on
1450 KVM_ASSIGN_DEV_IRQ. Partial deassignment of host or guest IRQ is allowed.
1453 4.52 KVM_SET_GSI_ROUTING
1455 Capability: KVM_CAP_IRQ_ROUTING
1456 Architectures: x86 s390 arm arm64
1458 Parameters: struct kvm_irq_routing (in)
1459 Returns: 0 on success, -1 on error
1461 Sets the GSI routing table entries, overwriting any previously set entries.
1463 On arm/arm64, GSI routing has the following limitation:
1464 - GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD.
1466 struct kvm_irq_routing {
1469 struct kvm_irq_routing_entry entries[0];
1472 No flags are specified so far, the corresponding field must be set to zero.
1474 struct kvm_irq_routing_entry {
1480 struct kvm_irq_routing_irqchip irqchip;
1481 struct kvm_irq_routing_msi msi;
1482 struct kvm_irq_routing_s390_adapter adapter;
1483 struct kvm_irq_routing_hv_sint hv_sint;
1488 /* gsi routing entry types */
1489 #define KVM_IRQ_ROUTING_IRQCHIP 1
1490 #define KVM_IRQ_ROUTING_MSI 2
1491 #define KVM_IRQ_ROUTING_S390_ADAPTER 3
1492 #define KVM_IRQ_ROUTING_HV_SINT 4
1495 - KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry
1496 type, specifies that the devid field contains a valid value. The per-VM
1497 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
1498 the device ID. If this capability is not available, userspace should
1499 never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
1502 struct kvm_irq_routing_irqchip {
1507 struct kvm_irq_routing_msi {
1517 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
1518 for the device that wrote the MSI message. For PCI, this is usually a
1519 BFD identifier in the lower 16 bits.
1521 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
1522 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
1523 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
1524 address_hi must be zero.
1526 struct kvm_irq_routing_s390_adapter {
1530 __u32 summary_offset;
1534 struct kvm_irq_routing_hv_sint {
1539 4.53 KVM_ASSIGN_SET_MSIX_NR (deprecated)
1544 Parameters: struct kvm_assigned_msix_nr (in)
1545 Returns: 0 on success, -1 on error
1547 Set the number of MSI-X interrupts for an assigned device. The number is
1548 reset again by terminating the MSI-X assignment of the device via
1549 KVM_DEASSIGN_DEV_IRQ. Calling this service more than once at any earlier
1552 struct kvm_assigned_msix_nr {
1553 __u32 assigned_dev_id;
1558 #define KVM_MAX_MSIX_PER_DEV 256
1561 4.54 KVM_ASSIGN_SET_MSIX_ENTRY (deprecated)
1566 Parameters: struct kvm_assigned_msix_entry (in)
1567 Returns: 0 on success, -1 on error
1569 Specifies the routing of an MSI-X assigned device interrupt to a GSI. Setting
1570 the GSI vector to zero means disabling the interrupt.
1572 struct kvm_assigned_msix_entry {
1573 __u32 assigned_dev_id;
1575 __u16 entry; /* The index of entry in the MSI-X table */
1580 ENOTTY: kernel does not support this ioctl
1582 Other error conditions may be defined by individual device types or
1583 have their standard meanings.
1586 4.55 KVM_SET_TSC_KHZ
1588 Capability: KVM_CAP_TSC_CONTROL
1591 Parameters: virtual tsc_khz
1592 Returns: 0 on success, -1 on error
1594 Specifies the tsc frequency for the virtual machine. The unit of the
1598 4.56 KVM_GET_TSC_KHZ
1600 Capability: KVM_CAP_GET_TSC_KHZ
1604 Returns: virtual tsc-khz on success, negative value on error
1606 Returns the tsc frequency of the guest. The unit of the return value is
1607 KHz. If the host has unstable tsc this ioctl returns -EIO instead as an
1613 Capability: KVM_CAP_IRQCHIP
1616 Parameters: struct kvm_lapic_state (out)
1617 Returns: 0 on success, -1 on error
1619 #define KVM_APIC_REG_SIZE 0x400
1620 struct kvm_lapic_state {
1621 char regs[KVM_APIC_REG_SIZE];
1624 Reads the Local APIC registers and copies them into the input argument. The
1625 data format and layout are the same as documented in the architecture manual.
1627 If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is
1628 enabled, then the format of APIC_ID register depends on the APIC mode
1629 (reported by MSR_IA32_APICBASE) of its VCPU. x2APIC stores APIC ID in
1630 the APIC_ID register (bytes 32-35). xAPIC only allows an 8-bit APIC ID
1631 which is stored in bits 31-24 of the APIC register, or equivalently in
1632 byte 35 of struct kvm_lapic_state's regs field. KVM_GET_LAPIC must then
1633 be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR.
1635 If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state
1636 always uses xAPIC format.
1641 Capability: KVM_CAP_IRQCHIP
1644 Parameters: struct kvm_lapic_state (in)
1645 Returns: 0 on success, -1 on error
1647 #define KVM_APIC_REG_SIZE 0x400
1648 struct kvm_lapic_state {
1649 char regs[KVM_APIC_REG_SIZE];
1652 Copies the input argument into the Local APIC registers. The data format
1653 and layout are the same as documented in the architecture manual.
1655 The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's
1656 regs field) depends on the state of the KVM_CAP_X2APIC_API capability.
1657 See the note in KVM_GET_LAPIC.
1662 Capability: KVM_CAP_IOEVENTFD
1665 Parameters: struct kvm_ioeventfd (in)
1666 Returns: 0 on success, !0 on error
1668 This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address
1669 within the guest. A guest write in the registered address will signal the
1670 provided event instead of triggering an exit.
1672 struct kvm_ioeventfd {
1674 __u64 addr; /* legal pio/mmio address */
1675 __u32 len; /* 0, 1, 2, 4, or 8 bytes */
1681 For the special case of virtio-ccw devices on s390, the ioevent is matched
1682 to a subchannel/virtqueue tuple instead.
1684 The following flags are defined:
1686 #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch)
1687 #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio)
1688 #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign)
1689 #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \
1690 (1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify)
1692 If datamatch flag is set, the event will be signaled only if the written value
1693 to the registered address is equal to datamatch in struct kvm_ioeventfd.
1695 For virtio-ccw devices, addr contains the subchannel id and datamatch the
1698 With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and
1699 the kernel will ignore the length of guest write and may get a faster vmexit.
1700 The speedup may only apply to specific architectures, but the ioeventfd will
1705 Capability: KVM_CAP_SW_TLB
1708 Parameters: struct kvm_dirty_tlb (in)
1709 Returns: 0 on success, -1 on error
1711 struct kvm_dirty_tlb {
1716 This must be called whenever userspace has changed an entry in the shared
1717 TLB, prior to calling KVM_RUN on the associated vcpu.
1719 The "bitmap" field is the userspace address of an array. This array
1720 consists of a number of bits, equal to the total number of TLB entries as
1721 determined by the last successful call to KVM_CONFIG_TLB, rounded up to the
1722 nearest multiple of 64.
1724 Each bit corresponds to one TLB entry, ordered the same as in the shared TLB
1727 The array is little-endian: the bit 0 is the least significant bit of the
1728 first byte, bit 8 is the least significant bit of the second byte, etc.
1729 This avoids any complications with differing word sizes.
1731 The "num_dirty" field is a performance hint for KVM to determine whether it
1732 should skip processing the bitmap and just invalidate everything. It must
1733 be set to the number of set bits in the bitmap.
1736 4.61 KVM_ASSIGN_SET_INTX_MASK (deprecated)
1738 Capability: KVM_CAP_PCI_2_3
1741 Parameters: struct kvm_assigned_pci_dev (in)
1742 Returns: 0 on success, -1 on error
1744 Allows userspace to mask PCI INTx interrupts from the assigned device. The
1745 kernel will not deliver INTx interrupts to the guest between setting and
1746 clearing of KVM_ASSIGN_SET_INTX_MASK via this interface. This enables use of
1747 and emulation of PCI 2.3 INTx disable command register behavior.
1749 This may be used for both PCI 2.3 devices supporting INTx disable natively and
1750 older devices lacking this support. Userspace is responsible for emulating the
1751 read value of the INTx disable bit in the guest visible PCI command register.
1752 When modifying the INTx disable state, userspace should precede updating the
1753 physical device command register by calling this ioctl to inform the kernel of
1754 the new intended INTx mask state.
1756 Note that the kernel uses the device INTx disable bit to internally manage the
1757 device interrupt state for PCI 2.3 devices. Reads of this register may
1758 therefore not match the expected value. Writes should always use the guest
1759 intended INTx disable value rather than attempting to read-copy-update the
1760 current physical device state. Races between user and kernel updates to the
1761 INTx disable bit are handled lazily in the kernel. It's possible the device
1762 may generate unintended interrupts, but they will not be injected into the
1765 See KVM_ASSIGN_DEV_IRQ for the data structure. The target device is specified
1766 by assigned_dev_id. In the flags field, only KVM_DEV_ASSIGN_MASK_INTX is
1770 4.62 KVM_CREATE_SPAPR_TCE
1772 Capability: KVM_CAP_SPAPR_TCE
1773 Architectures: powerpc
1775 Parameters: struct kvm_create_spapr_tce (in)
1776 Returns: file descriptor for manipulating the created TCE table
1778 This creates a virtual TCE (translation control entry) table, which
1779 is an IOMMU for PAPR-style virtual I/O. It is used to translate
1780 logical addresses used in virtual I/O into guest physical addresses,
1781 and provides a scatter/gather capability for PAPR virtual I/O.
1783 /* for KVM_CAP_SPAPR_TCE */
1784 struct kvm_create_spapr_tce {
1789 The liobn field gives the logical IO bus number for which to create a
1790 TCE table. The window_size field specifies the size of the DMA window
1791 which this TCE table will translate - the table will contain one 64
1792 bit TCE entry for every 4kiB of the DMA window.
1794 When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE
1795 table has been created using this ioctl(), the kernel will handle it
1796 in real mode, updating the TCE table. H_PUT_TCE calls for other
1797 liobns will cause a vm exit and must be handled by userspace.
1799 The return value is a file descriptor which can be passed to mmap(2)
1800 to map the created TCE table into userspace. This lets userspace read
1801 the entries written by kernel-handled H_PUT_TCE calls, and also lets
1802 userspace update the TCE table directly which is useful in some
1806 4.63 KVM_ALLOCATE_RMA
1808 Capability: KVM_CAP_PPC_RMA
1809 Architectures: powerpc
1811 Parameters: struct kvm_allocate_rma (out)
1812 Returns: file descriptor for mapping the allocated RMA
1814 This allocates a Real Mode Area (RMA) from the pool allocated at boot
1815 time by the kernel. An RMA is a physically-contiguous, aligned region
1816 of memory used on older POWER processors to provide the memory which
1817 will be accessed by real-mode (MMU off) accesses in a KVM guest.
1818 POWER processors support a set of sizes for the RMA that usually
1819 includes 64MB, 128MB, 256MB and some larger powers of two.
1821 /* for KVM_ALLOCATE_RMA */
1822 struct kvm_allocate_rma {
1826 The return value is a file descriptor which can be passed to mmap(2)
1827 to map the allocated RMA into userspace. The mapped area can then be
1828 passed to the KVM_SET_USER_MEMORY_REGION ioctl to establish it as the
1829 RMA for a virtual machine. The size of the RMA in bytes (which is
1830 fixed at host kernel boot time) is returned in the rma_size field of
1831 the argument structure.
1833 The KVM_CAP_PPC_RMA capability is 1 or 2 if the KVM_ALLOCATE_RMA ioctl
1834 is supported; 2 if the processor requires all virtual machines to have
1835 an RMA, or 1 if the processor can use an RMA but doesn't require it,
1836 because it supports the Virtual RMA (VRMA) facility.
1841 Capability: KVM_CAP_USER_NMI
1845 Returns: 0 on success, -1 on error
1847 Queues an NMI on the thread's vcpu. Note this is well defined only
1848 when KVM_CREATE_IRQCHIP has not been called, since this is an interface
1849 between the virtual cpu core and virtual local APIC. After KVM_CREATE_IRQCHIP
1850 has been called, this interface is completely emulated within the kernel.
1852 To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the
1853 following algorithm:
1856 - read the local APIC's state (KVM_GET_LAPIC)
1857 - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1)
1858 - if so, issue KVM_NMI
1861 Some guests configure the LINT1 NMI input to cause a panic, aiding in
1865 4.65 KVM_S390_UCAS_MAP
1867 Capability: KVM_CAP_S390_UCONTROL
1870 Parameters: struct kvm_s390_ucas_mapping (in)
1871 Returns: 0 in case of success
1873 The parameter is defined like this:
1874 struct kvm_s390_ucas_mapping {
1880 This ioctl maps the memory at "user_addr" with the length "length" to
1881 the vcpu's address space starting at "vcpu_addr". All parameters need to
1882 be aligned by 1 megabyte.
1885 4.66 KVM_S390_UCAS_UNMAP
1887 Capability: KVM_CAP_S390_UCONTROL
1890 Parameters: struct kvm_s390_ucas_mapping (in)
1891 Returns: 0 in case of success
1893 The parameter is defined like this:
1894 struct kvm_s390_ucas_mapping {
1900 This ioctl unmaps the memory in the vcpu's address space starting at
1901 "vcpu_addr" with the length "length". The field "user_addr" is ignored.
1902 All parameters need to be aligned by 1 megabyte.
1905 4.67 KVM_S390_VCPU_FAULT
1907 Capability: KVM_CAP_S390_UCONTROL
1910 Parameters: vcpu absolute address (in)
1911 Returns: 0 in case of success
1913 This call creates a page table entry on the virtual cpu's address space
1914 (for user controlled virtual machines) or the virtual machine's address
1915 space (for regular virtual machines). This only works for minor faults,
1916 thus it's recommended to access subject memory page via the user page
1917 table upfront. This is useful to handle validity intercepts for user
1918 controlled virtual machines to fault in the virtual cpu's lowcore pages
1919 prior to calling the KVM_RUN ioctl.
1922 4.68 KVM_SET_ONE_REG
1924 Capability: KVM_CAP_ONE_REG
1927 Parameters: struct kvm_one_reg (in)
1928 Returns: 0 on success, negative value on failure
1930 struct kvm_one_reg {
1935 Using this ioctl, a single vcpu register can be set to a specific value
1936 defined by user space with the passed in struct kvm_one_reg, where id
1937 refers to the register identifier as described below and addr is a pointer
1938 to a variable with the respective size. There can be architecture agnostic
1939 and architecture specific registers. Each have their own range of operation
1940 and their own constants and width. To keep track of the implemented
1941 registers, find a list below:
1943 Arch | Register | Width (bits)
1945 PPC | KVM_REG_PPC_HIOR | 64
1946 PPC | KVM_REG_PPC_IAC1 | 64
1947 PPC | KVM_REG_PPC_IAC2 | 64
1948 PPC | KVM_REG_PPC_IAC3 | 64
1949 PPC | KVM_REG_PPC_IAC4 | 64
1950 PPC | KVM_REG_PPC_DAC1 | 64
1951 PPC | KVM_REG_PPC_DAC2 | 64
1952 PPC | KVM_REG_PPC_DABR | 64
1953 PPC | KVM_REG_PPC_DSCR | 64
1954 PPC | KVM_REG_PPC_PURR | 64
1955 PPC | KVM_REG_PPC_SPURR | 64
1956 PPC | KVM_REG_PPC_DAR | 64
1957 PPC | KVM_REG_PPC_DSISR | 32
1958 PPC | KVM_REG_PPC_AMR | 64
1959 PPC | KVM_REG_PPC_UAMOR | 64
1960 PPC | KVM_REG_PPC_MMCR0 | 64
1961 PPC | KVM_REG_PPC_MMCR1 | 64
1962 PPC | KVM_REG_PPC_MMCRA | 64
1963 PPC | KVM_REG_PPC_MMCR2 | 64
1964 PPC | KVM_REG_PPC_MMCRS | 64
1965 PPC | KVM_REG_PPC_SIAR | 64
1966 PPC | KVM_REG_PPC_SDAR | 64
1967 PPC | KVM_REG_PPC_SIER | 64
1968 PPC | KVM_REG_PPC_PMC1 | 32
1969 PPC | KVM_REG_PPC_PMC2 | 32
1970 PPC | KVM_REG_PPC_PMC3 | 32
1971 PPC | KVM_REG_PPC_PMC4 | 32
1972 PPC | KVM_REG_PPC_PMC5 | 32
1973 PPC | KVM_REG_PPC_PMC6 | 32
1974 PPC | KVM_REG_PPC_PMC7 | 32
1975 PPC | KVM_REG_PPC_PMC8 | 32
1976 PPC | KVM_REG_PPC_FPR0 | 64
1978 PPC | KVM_REG_PPC_FPR31 | 64
1979 PPC | KVM_REG_PPC_VR0 | 128
1981 PPC | KVM_REG_PPC_VR31 | 128
1982 PPC | KVM_REG_PPC_VSR0 | 128
1984 PPC | KVM_REG_PPC_VSR31 | 128
1985 PPC | KVM_REG_PPC_FPSCR | 64
1986 PPC | KVM_REG_PPC_VSCR | 32
1987 PPC | KVM_REG_PPC_VPA_ADDR | 64
1988 PPC | KVM_REG_PPC_VPA_SLB | 128
1989 PPC | KVM_REG_PPC_VPA_DTL | 128
1990 PPC | KVM_REG_PPC_EPCR | 32
1991 PPC | KVM_REG_PPC_EPR | 32
1992 PPC | KVM_REG_PPC_TCR | 32
1993 PPC | KVM_REG_PPC_TSR | 32
1994 PPC | KVM_REG_PPC_OR_TSR | 32
1995 PPC | KVM_REG_PPC_CLEAR_TSR | 32
1996 PPC | KVM_REG_PPC_MAS0 | 32
1997 PPC | KVM_REG_PPC_MAS1 | 32
1998 PPC | KVM_REG_PPC_MAS2 | 64
1999 PPC | KVM_REG_PPC_MAS7_3 | 64
2000 PPC | KVM_REG_PPC_MAS4 | 32
2001 PPC | KVM_REG_PPC_MAS6 | 32
2002 PPC | KVM_REG_PPC_MMUCFG | 32
2003 PPC | KVM_REG_PPC_TLB0CFG | 32
2004 PPC | KVM_REG_PPC_TLB1CFG | 32
2005 PPC | KVM_REG_PPC_TLB2CFG | 32
2006 PPC | KVM_REG_PPC_TLB3CFG | 32
2007 PPC | KVM_REG_PPC_TLB0PS | 32
2008 PPC | KVM_REG_PPC_TLB1PS | 32
2009 PPC | KVM_REG_PPC_TLB2PS | 32
2010 PPC | KVM_REG_PPC_TLB3PS | 32
2011 PPC | KVM_REG_PPC_EPTCFG | 32
2012 PPC | KVM_REG_PPC_ICP_STATE | 64
2013 PPC | KVM_REG_PPC_TB_OFFSET | 64
2014 PPC | KVM_REG_PPC_SPMC1 | 32
2015 PPC | KVM_REG_PPC_SPMC2 | 32
2016 PPC | KVM_REG_PPC_IAMR | 64
2017 PPC | KVM_REG_PPC_TFHAR | 64
2018 PPC | KVM_REG_PPC_TFIAR | 64
2019 PPC | KVM_REG_PPC_TEXASR | 64
2020 PPC | KVM_REG_PPC_FSCR | 64
2021 PPC | KVM_REG_PPC_PSPB | 32
2022 PPC | KVM_REG_PPC_EBBHR | 64
2023 PPC | KVM_REG_PPC_EBBRR | 64
2024 PPC | KVM_REG_PPC_BESCR | 64
2025 PPC | KVM_REG_PPC_TAR | 64
2026 PPC | KVM_REG_PPC_DPDES | 64
2027 PPC | KVM_REG_PPC_DAWR | 64
2028 PPC | KVM_REG_PPC_DAWRX | 64
2029 PPC | KVM_REG_PPC_CIABR | 64
2030 PPC | KVM_REG_PPC_IC | 64
2031 PPC | KVM_REG_PPC_VTB | 64
2032 PPC | KVM_REG_PPC_CSIGR | 64
2033 PPC | KVM_REG_PPC_TACR | 64
2034 PPC | KVM_REG_PPC_TCSCR | 64
2035 PPC | KVM_REG_PPC_PID | 64
2036 PPC | KVM_REG_PPC_ACOP | 64
2037 PPC | KVM_REG_PPC_VRSAVE | 32
2038 PPC | KVM_REG_PPC_LPCR | 32
2039 PPC | KVM_REG_PPC_LPCR_64 | 64
2040 PPC | KVM_REG_PPC_PPR | 64
2041 PPC | KVM_REG_PPC_ARCH_COMPAT | 32
2042 PPC | KVM_REG_PPC_DABRX | 32
2043 PPC | KVM_REG_PPC_WORT | 64
2044 PPC | KVM_REG_PPC_SPRG9 | 64
2045 PPC | KVM_REG_PPC_DBSR | 32
2046 PPC | KVM_REG_PPC_TIDR | 64
2047 PPC | KVM_REG_PPC_PSSCR | 64
2048 PPC | KVM_REG_PPC_TM_GPR0 | 64
2050 PPC | KVM_REG_PPC_TM_GPR31 | 64
2051 PPC | KVM_REG_PPC_TM_VSR0 | 128
2053 PPC | KVM_REG_PPC_TM_VSR63 | 128
2054 PPC | KVM_REG_PPC_TM_CR | 64
2055 PPC | KVM_REG_PPC_TM_LR | 64
2056 PPC | KVM_REG_PPC_TM_CTR | 64
2057 PPC | KVM_REG_PPC_TM_FPSCR | 64
2058 PPC | KVM_REG_PPC_TM_AMR | 64
2059 PPC | KVM_REG_PPC_TM_PPR | 64
2060 PPC | KVM_REG_PPC_TM_VRSAVE | 64
2061 PPC | KVM_REG_PPC_TM_VSCR | 32
2062 PPC | KVM_REG_PPC_TM_DSCR | 64
2063 PPC | KVM_REG_PPC_TM_TAR | 64
2064 PPC | KVM_REG_PPC_TM_XER | 64
2066 MIPS | KVM_REG_MIPS_R0 | 64
2068 MIPS | KVM_REG_MIPS_R31 | 64
2069 MIPS | KVM_REG_MIPS_HI | 64
2070 MIPS | KVM_REG_MIPS_LO | 64
2071 MIPS | KVM_REG_MIPS_PC | 64
2072 MIPS | KVM_REG_MIPS_CP0_INDEX | 32
2073 MIPS | KVM_REG_MIPS_CP0_ENTRYLO0 | 64
2074 MIPS | KVM_REG_MIPS_CP0_ENTRYLO1 | 64
2075 MIPS | KVM_REG_MIPS_CP0_CONTEXT | 64
2076 MIPS | KVM_REG_MIPS_CP0_CONTEXTCONFIG| 32
2077 MIPS | KVM_REG_MIPS_CP0_USERLOCAL | 64
2078 MIPS | KVM_REG_MIPS_CP0_XCONTEXTCONFIG| 64
2079 MIPS | KVM_REG_MIPS_CP0_PAGEMASK | 32
2080 MIPS | KVM_REG_MIPS_CP0_PAGEGRAIN | 32
2081 MIPS | KVM_REG_MIPS_CP0_SEGCTL0 | 64
2082 MIPS | KVM_REG_MIPS_CP0_SEGCTL1 | 64
2083 MIPS | KVM_REG_MIPS_CP0_SEGCTL2 | 64
2084 MIPS | KVM_REG_MIPS_CP0_PWBASE | 64
2085 MIPS | KVM_REG_MIPS_CP0_PWFIELD | 64
2086 MIPS | KVM_REG_MIPS_CP0_PWSIZE | 64
2087 MIPS | KVM_REG_MIPS_CP0_WIRED | 32
2088 MIPS | KVM_REG_MIPS_CP0_PWCTL | 32
2089 MIPS | KVM_REG_MIPS_CP0_HWRENA | 32
2090 MIPS | KVM_REG_MIPS_CP0_BADVADDR | 64
2091 MIPS | KVM_REG_MIPS_CP0_BADINSTR | 32
2092 MIPS | KVM_REG_MIPS_CP0_BADINSTRP | 32
2093 MIPS | KVM_REG_MIPS_CP0_COUNT | 32
2094 MIPS | KVM_REG_MIPS_CP0_ENTRYHI | 64
2095 MIPS | KVM_REG_MIPS_CP0_COMPARE | 32
2096 MIPS | KVM_REG_MIPS_CP0_STATUS | 32
2097 MIPS | KVM_REG_MIPS_CP0_INTCTL | 32
2098 MIPS | KVM_REG_MIPS_CP0_CAUSE | 32
2099 MIPS | KVM_REG_MIPS_CP0_EPC | 64
2100 MIPS | KVM_REG_MIPS_CP0_PRID | 32
2101 MIPS | KVM_REG_MIPS_CP0_EBASE | 64
2102 MIPS | KVM_REG_MIPS_CP0_CONFIG | 32
2103 MIPS | KVM_REG_MIPS_CP0_CONFIG1 | 32
2104 MIPS | KVM_REG_MIPS_CP0_CONFIG2 | 32
2105 MIPS | KVM_REG_MIPS_CP0_CONFIG3 | 32
2106 MIPS | KVM_REG_MIPS_CP0_CONFIG4 | 32
2107 MIPS | KVM_REG_MIPS_CP0_CONFIG5 | 32
2108 MIPS | KVM_REG_MIPS_CP0_CONFIG7 | 32
2109 MIPS | KVM_REG_MIPS_CP0_XCONTEXT | 64
2110 MIPS | KVM_REG_MIPS_CP0_ERROREPC | 64
2111 MIPS | KVM_REG_MIPS_CP0_KSCRATCH1 | 64
2112 MIPS | KVM_REG_MIPS_CP0_KSCRATCH2 | 64
2113 MIPS | KVM_REG_MIPS_CP0_KSCRATCH3 | 64
2114 MIPS | KVM_REG_MIPS_CP0_KSCRATCH4 | 64
2115 MIPS | KVM_REG_MIPS_CP0_KSCRATCH5 | 64
2116 MIPS | KVM_REG_MIPS_CP0_KSCRATCH6 | 64
2117 MIPS | KVM_REG_MIPS_CP0_MAAR(0..63) | 64
2118 MIPS | KVM_REG_MIPS_COUNT_CTL | 64
2119 MIPS | KVM_REG_MIPS_COUNT_RESUME | 64
2120 MIPS | KVM_REG_MIPS_COUNT_HZ | 64
2121 MIPS | KVM_REG_MIPS_FPR_32(0..31) | 32
2122 MIPS | KVM_REG_MIPS_FPR_64(0..31) | 64
2123 MIPS | KVM_REG_MIPS_VEC_128(0..31) | 128
2124 MIPS | KVM_REG_MIPS_FCR_IR | 32
2125 MIPS | KVM_REG_MIPS_FCR_CSR | 32
2126 MIPS | KVM_REG_MIPS_MSA_IR | 32
2127 MIPS | KVM_REG_MIPS_MSA_CSR | 32
2129 ARM registers are mapped using the lower 32 bits. The upper 16 of that
2130 is the register group type, or coprocessor number:
2132 ARM core registers have the following id bit patterns:
2133 0x4020 0000 0010 <index into the kvm_regs struct:16>
2135 ARM 32-bit CP15 registers have the following id bit patterns:
2136 0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3>
2138 ARM 64-bit CP15 registers have the following id bit patterns:
2139 0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3>
2141 ARM CCSIDR registers are demultiplexed by CSSELR value:
2142 0x4020 0000 0011 00 <csselr:8>
2144 ARM 32-bit VFP control registers have the following id bit patterns:
2145 0x4020 0000 0012 1 <regno:12>
2147 ARM 64-bit FP registers have the following id bit patterns:
2148 0x4030 0000 0012 0 <regno:12>
2151 arm64 registers are mapped using the lower 32 bits. The upper 16 of
2152 that is the register group type, or coprocessor number:
2154 arm64 core/FP-SIMD registers have the following id bit patterns. Note
2155 that the size of the access is variable, as the kvm_regs structure
2156 contains elements ranging from 32 to 128 bits. The index is a 32bit
2157 value in the kvm_regs structure seen as a 32bit array.
2158 0x60x0 0000 0010 <index into the kvm_regs struct:16>
2160 arm64 CCSIDR registers are demultiplexed by CSSELR value:
2161 0x6020 0000 0011 00 <csselr:8>
2163 arm64 system registers have the following id bit patterns:
2164 0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3>
2167 MIPS registers are mapped using the lower 32 bits. The upper 16 of that is
2168 the register group type:
2170 MIPS core registers (see above) have the following id bit patterns:
2171 0x7030 0000 0000 <reg:16>
2173 MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit
2174 patterns depending on whether they're 32-bit or 64-bit registers:
2175 0x7020 0000 0001 00 <reg:5> <sel:3> (32-bit)
2176 0x7030 0000 0001 00 <reg:5> <sel:3> (64-bit)
2178 Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_MIPS_CP0_ENTRYLO1 are the MIPS64
2179 versions of the EntryLo registers regardless of the word size of the host
2180 hardware, host kernel, guest, and whether XPA is present in the guest, i.e.
2181 with the RI and XI bits (if they exist) in bits 63 and 62 respectively, and
2182 the PFNX field starting at bit 30.
2184 MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) above) have the following id bit
2186 0x7030 0000 0001 01 <reg:8>
2188 MIPS KVM control registers (see above) have the following id bit patterns:
2189 0x7030 0000 0002 <reg:16>
2191 MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following
2192 id bit patterns depending on the size of the register being accessed. They are
2193 always accessed according to the current guest FPU mode (Status.FR and
2194 Config5.FRE), i.e. as the guest would see them, and they become unpredictable
2195 if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector
2196 registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they
2197 overlap the FPU registers:
2198 0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers)
2199 0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers)
2200 0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers)
2202 MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the
2203 following id bit patterns:
2204 0x7020 0000 0003 01 <0:3> <reg:5>
2206 MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the
2207 following id bit patterns:
2208 0x7020 0000 0003 02 <0:3> <reg:5>
2211 4.69 KVM_GET_ONE_REG
2213 Capability: KVM_CAP_ONE_REG
2216 Parameters: struct kvm_one_reg (in and out)
2217 Returns: 0 on success, negative value on failure
2219 This ioctl allows to receive the value of a single register implemented
2220 in a vcpu. The register to read is indicated by the "id" field of the
2221 kvm_one_reg struct passed in. On success, the register value can be found
2222 at the memory location pointed to by "addr".
2224 The list of registers accessible using this interface is identical to the
2228 4.70 KVM_KVMCLOCK_CTRL
2230 Capability: KVM_CAP_KVMCLOCK_CTRL
2231 Architectures: Any that implement pvclocks (currently x86 only)
2234 Returns: 0 on success, -1 on error
2236 This signals to the host kernel that the specified guest is being paused by
2237 userspace. The host will set a flag in the pvclock structure that is checked
2238 from the soft lockup watchdog. The flag is part of the pvclock structure that
2239 is shared between guest and host, specifically the second bit of the flags
2240 field of the pvclock_vcpu_time_info structure. It will be set exclusively by
2241 the host and read/cleared exclusively by the guest. The guest operation of
2242 checking and clearing the flag must an atomic operation so
2243 load-link/store-conditional, or equivalent must be used. There are two cases
2244 where the guest will clear the flag: when the soft lockup watchdog timer resets
2245 itself or when a soft lockup is detected. This ioctl can be called any time
2246 after pausing the vcpu, but before it is resumed.
2251 Capability: KVM_CAP_SIGNAL_MSI
2252 Architectures: x86 arm arm64
2254 Parameters: struct kvm_msi (in)
2255 Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error
2257 Directly inject a MSI message. Only valid with in-kernel irqchip that handles
2269 flags: KVM_MSI_VALID_DEVID: devid contains a valid value. The per-VM
2270 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
2271 the device ID. If this capability is not available, userspace
2272 should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
2274 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
2275 for the device that wrote the MSI message. For PCI, this is usually a
2276 BFD identifier in the lower 16 bits.
2278 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
2279 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
2280 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
2281 address_hi must be zero.
2284 4.71 KVM_CREATE_PIT2
2286 Capability: KVM_CAP_PIT2
2289 Parameters: struct kvm_pit_config (in)
2290 Returns: 0 on success, -1 on error
2292 Creates an in-kernel device model for the i8254 PIT. This call is only valid
2293 after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following
2294 parameters have to be passed:
2296 struct kvm_pit_config {
2303 #define KVM_PIT_SPEAKER_DUMMY 1 /* emulate speaker port stub */
2305 PIT timer interrupts may use a per-VM kernel thread for injection. If it
2306 exists, this thread will have a name of the following pattern:
2308 kvm-pit/<owner-process-pid>
2310 When running a guest with elevated priorities, the scheduling parameters of
2311 this thread may have to be adjusted accordingly.
2313 This IOCTL replaces the obsolete KVM_CREATE_PIT.
2318 Capability: KVM_CAP_PIT_STATE2
2321 Parameters: struct kvm_pit_state2 (out)
2322 Returns: 0 on success, -1 on error
2324 Retrieves the state of the in-kernel PIT model. Only valid after
2325 KVM_CREATE_PIT2. The state is returned in the following structure:
2327 struct kvm_pit_state2 {
2328 struct kvm_pit_channel_state channels[3];
2335 /* disable PIT in HPET legacy mode */
2336 #define KVM_PIT_FLAGS_HPET_LEGACY 0x00000001
2338 This IOCTL replaces the obsolete KVM_GET_PIT.
2343 Capability: KVM_CAP_PIT_STATE2
2346 Parameters: struct kvm_pit_state2 (in)
2347 Returns: 0 on success, -1 on error
2349 Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2.
2350 See KVM_GET_PIT2 for details on struct kvm_pit_state2.
2352 This IOCTL replaces the obsolete KVM_SET_PIT.
2355 4.74 KVM_PPC_GET_SMMU_INFO
2357 Capability: KVM_CAP_PPC_GET_SMMU_INFO
2358 Architectures: powerpc
2361 Returns: 0 on success, -1 on error
2363 This populates and returns a structure describing the features of
2364 the "Server" class MMU emulation supported by KVM.
2365 This can in turn be used by userspace to generate the appropriate
2366 device-tree properties for the guest operating system.
2368 The structure contains some global information, followed by an
2369 array of supported segment page sizes:
2371 struct kvm_ppc_smmu_info {
2375 struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ];
2378 The supported flags are:
2380 - KVM_PPC_PAGE_SIZES_REAL:
2381 When that flag is set, guest page sizes must "fit" the backing
2382 store page sizes. When not set, any page size in the list can
2383 be used regardless of how they are backed by userspace.
2385 - KVM_PPC_1T_SEGMENTS
2386 The emulated MMU supports 1T segments in addition to the
2389 The "slb_size" field indicates how many SLB entries are supported
2391 The "sps" array contains 8 entries indicating the supported base
2392 page sizes for a segment in increasing order. Each entry is defined
2395 struct kvm_ppc_one_seg_page_size {
2396 __u32 page_shift; /* Base page shift of segment (or 0) */
2397 __u32 slb_enc; /* SLB encoding for BookS */
2398 struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ];
2401 An entry with a "page_shift" of 0 is unused. Because the array is
2402 organized in increasing order, a lookup can stop when encoutering
2405 The "slb_enc" field provides the encoding to use in the SLB for the
2406 page size. The bits are in positions such as the value can directly
2407 be OR'ed into the "vsid" argument of the slbmte instruction.
2409 The "enc" array is a list which for each of those segment base page
2410 size provides the list of supported actual page sizes (which can be
2411 only larger or equal to the base page size), along with the
2412 corresponding encoding in the hash PTE. Similarly, the array is
2413 8 entries sorted by increasing sizes and an entry with a "0" shift
2414 is an empty entry and a terminator:
2416 struct kvm_ppc_one_page_size {
2417 __u32 page_shift; /* Page shift (or 0) */
2418 __u32 pte_enc; /* Encoding in the HPTE (>>12) */
2421 The "pte_enc" field provides a value that can OR'ed into the hash
2422 PTE's RPN field (ie, it needs to be shifted left by 12 to OR it
2423 into the hash PTE second double word).
2427 Capability: KVM_CAP_IRQFD
2428 Architectures: x86 s390 arm arm64
2430 Parameters: struct kvm_irqfd (in)
2431 Returns: 0 on success, -1 on error
2433 Allows setting an eventfd to directly trigger a guest interrupt.
2434 kvm_irqfd.fd specifies the file descriptor to use as the eventfd and
2435 kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When
2436 an event is triggered on the eventfd, an interrupt is injected into
2437 the guest using the specified gsi pin. The irqfd is removed using
2438 the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd
2441 With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify
2442 mechanism allowing emulation of level-triggered, irqfd-based
2443 interrupts. When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an
2444 additional eventfd in the kvm_irqfd.resamplefd field. When operating
2445 in resample mode, posting of an interrupt through kvm_irq.fd asserts
2446 the specified gsi in the irqchip. When the irqchip is resampled, such
2447 as from an EOI, the gsi is de-asserted and the user is notified via
2448 kvm_irqfd.resamplefd. It is the user's responsibility to re-queue
2449 the interrupt if the device making use of it still requires service.
2450 Note that closing the resamplefd is not sufficient to disable the
2451 irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment
2452 and need not be specified with KVM_IRQFD_FLAG_DEASSIGN.
2454 On arm/arm64, gsi routing being supported, the following can happen:
2455 - in case no routing entry is associated to this gsi, injection fails
2456 - in case the gsi is associated to an irqchip routing entry,
2457 irqchip.pin + 32 corresponds to the injected SPI ID.
2458 - in case the gsi is associated to an MSI routing entry, the MSI
2459 message and device ID are translated into an LPI (support restricted
2460 to GICv3 ITS in-kernel emulation).
2462 4.76 KVM_PPC_ALLOCATE_HTAB
2464 Capability: KVM_CAP_PPC_ALLOC_HTAB
2465 Architectures: powerpc
2467 Parameters: Pointer to u32 containing hash table order (in/out)
2468 Returns: 0 on success, -1 on error
2470 This requests the host kernel to allocate an MMU hash table for a
2471 guest using the PAPR paravirtualization interface. This only does
2472 anything if the kernel is configured to use the Book 3S HV style of
2473 virtualization. Otherwise the capability doesn't exist and the ioctl
2474 returns an ENOTTY error. The rest of this description assumes Book 3S
2477 There must be no vcpus running when this ioctl is called; if there
2478 are, it will do nothing and return an EBUSY error.
2480 The parameter is a pointer to a 32-bit unsigned integer variable
2481 containing the order (log base 2) of the desired size of the hash
2482 table, which must be between 18 and 46. On successful return from the
2483 ioctl, the value will not be changed by the kernel.
2485 If no hash table has been allocated when any vcpu is asked to run
2486 (with the KVM_RUN ioctl), the host kernel will allocate a
2487 default-sized hash table (16 MB).
2489 If this ioctl is called when a hash table has already been allocated,
2490 with a different order from the existing hash table, the existing hash
2491 table will be freed and a new one allocated. If this is ioctl is
2492 called when a hash table has already been allocated of the same order
2493 as specified, the kernel will clear out the existing hash table (zero
2494 all HPTEs). In either case, if the guest is using the virtualized
2495 real-mode area (VRMA) facility, the kernel will re-create the VMRA
2496 HPTEs on the next KVM_RUN of any vcpu.
2498 4.77 KVM_S390_INTERRUPT
2502 Type: vm ioctl, vcpu ioctl
2503 Parameters: struct kvm_s390_interrupt (in)
2504 Returns: 0 on success, -1 on error
2506 Allows to inject an interrupt to the guest. Interrupts can be floating
2507 (vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type.
2509 Interrupt parameters are passed via kvm_s390_interrupt:
2511 struct kvm_s390_interrupt {
2517 type can be one of the following:
2519 KVM_S390_SIGP_STOP (vcpu) - sigp stop; optional flags in parm
2520 KVM_S390_PROGRAM_INT (vcpu) - program check; code in parm
2521 KVM_S390_SIGP_SET_PREFIX (vcpu) - sigp set prefix; prefix address in parm
2522 KVM_S390_RESTART (vcpu) - restart
2523 KVM_S390_INT_CLOCK_COMP (vcpu) - clock comparator interrupt
2524 KVM_S390_INT_CPU_TIMER (vcpu) - CPU timer interrupt
2525 KVM_S390_INT_VIRTIO (vm) - virtio external interrupt; external interrupt
2526 parameters in parm and parm64
2527 KVM_S390_INT_SERVICE (vm) - sclp external interrupt; sclp parameter in parm
2528 KVM_S390_INT_EMERGENCY (vcpu) - sigp emergency; source cpu in parm
2529 KVM_S390_INT_EXTERNAL_CALL (vcpu) - sigp external call; source cpu in parm
2530 KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm) - compound value to indicate an
2531 I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel);
2532 I/O interruption parameters in parm (subchannel) and parm64 (intparm,
2533 interruption subclass)
2534 KVM_S390_MCHK (vm, vcpu) - machine check interrupt; cr 14 bits in parm,
2535 machine check interrupt code in parm64 (note that
2536 machine checks needing further payload are not
2537 supported by this ioctl)
2539 Note that the vcpu ioctl is asynchronous to vcpu execution.
2541 4.78 KVM_PPC_GET_HTAB_FD
2543 Capability: KVM_CAP_PPC_HTAB_FD
2544 Architectures: powerpc
2546 Parameters: Pointer to struct kvm_get_htab_fd (in)
2547 Returns: file descriptor number (>= 0) on success, -1 on error
2549 This returns a file descriptor that can be used either to read out the
2550 entries in the guest's hashed page table (HPT), or to write entries to
2551 initialize the HPT. The returned fd can only be written to if the
2552 KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and
2553 can only be read if that bit is clear. The argument struct looks like
2556 /* For KVM_PPC_GET_HTAB_FD */
2557 struct kvm_get_htab_fd {
2563 /* Values for kvm_get_htab_fd.flags */
2564 #define KVM_GET_HTAB_BOLTED_ONLY ((__u64)0x1)
2565 #define KVM_GET_HTAB_WRITE ((__u64)0x2)
2567 The `start_index' field gives the index in the HPT of the entry at
2568 which to start reading. It is ignored when writing.
2570 Reads on the fd will initially supply information about all
2571 "interesting" HPT entries. Interesting entries are those with the
2572 bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise
2573 all entries. When the end of the HPT is reached, the read() will
2574 return. If read() is called again on the fd, it will start again from
2575 the beginning of the HPT, but will only return HPT entries that have
2576 changed since they were last read.
2578 Data read or written is structured as a header (8 bytes) followed by a
2579 series of valid HPT entries (16 bytes) each. The header indicates how
2580 many valid HPT entries there are and how many invalid entries follow
2581 the valid entries. The invalid entries are not represented explicitly
2582 in the stream. The header format is:
2584 struct kvm_get_htab_header {
2590 Writes to the fd create HPT entries starting at the index given in the
2591 header; first `n_valid' valid entries with contents from the data
2592 written, then `n_invalid' invalid entries, invalidating any previously
2593 valid entries found.
2595 4.79 KVM_CREATE_DEVICE
2597 Capability: KVM_CAP_DEVICE_CTRL
2599 Parameters: struct kvm_create_device (in/out)
2600 Returns: 0 on success, -1 on error
2602 ENODEV: The device type is unknown or unsupported
2603 EEXIST: Device already created, and this type of device may not
2604 be instantiated multiple times
2606 Other error conditions may be defined by individual device types or
2607 have their standard meanings.
2609 Creates an emulated device in the kernel. The file descriptor returned
2610 in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR.
2612 If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the
2613 device type is supported (not necessarily whether it can be created
2616 Individual devices should not define flags. Attributes should be used
2617 for specifying any behavior that is not implied by the device type
2620 struct kvm_create_device {
2621 __u32 type; /* in: KVM_DEV_TYPE_xxx */
2622 __u32 fd; /* out: device handle */
2623 __u32 flags; /* in: KVM_CREATE_DEVICE_xxx */
2626 4.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR
2628 Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
2629 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
2630 Type: device ioctl, vm ioctl, vcpu ioctl
2631 Parameters: struct kvm_device_attr
2632 Returns: 0 on success, -1 on error
2634 ENXIO: The group or attribute is unknown/unsupported for this device
2635 or hardware support is missing.
2636 EPERM: The attribute cannot (currently) be accessed this way
2637 (e.g. read-only attribute, or attribute that only makes
2638 sense when the device is in a different state)
2640 Other error conditions may be defined by individual device types.
2642 Gets/sets a specified piece of device configuration and/or state. The
2643 semantics are device-specific. See individual device documentation in
2644 the "devices" directory. As with ONE_REG, the size of the data
2645 transferred is defined by the particular attribute.
2647 struct kvm_device_attr {
2648 __u32 flags; /* no flags currently defined */
2649 __u32 group; /* device-defined */
2650 __u64 attr; /* group-defined */
2651 __u64 addr; /* userspace address of attr data */
2654 4.81 KVM_HAS_DEVICE_ATTR
2656 Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
2657 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
2658 Type: device ioctl, vm ioctl, vcpu ioctl
2659 Parameters: struct kvm_device_attr
2660 Returns: 0 on success, -1 on error
2662 ENXIO: The group or attribute is unknown/unsupported for this device
2663 or hardware support is missing.
2665 Tests whether a device supports a particular attribute. A successful
2666 return indicates the attribute is implemented. It does not necessarily
2667 indicate that the attribute can be read or written in the device's
2668 current state. "addr" is ignored.
2670 4.82 KVM_ARM_VCPU_INIT
2673 Architectures: arm, arm64
2675 Parameters: struct kvm_vcpu_init (in)
2676 Returns: 0 on success; -1 on error
2678 Â EINVAL: Â Â Â the target is unknown, or the combination of features is invalid.
2679 Â ENOENT: Â Â Â a features bit specified is unknown.
2681 This tells KVM what type of CPU to present to the guest, and what
2682 optional features it should have. Â This will cause a reset of the cpu
2683 registers to their initial values. Â If this is not called, KVM_RUN will
2684 return ENOEXEC for that vcpu.
2686 Note that because some registers reflect machine topology, all vcpus
2687 should be created before this ioctl is invoked.
2689 Userspace can call this function multiple times for a given vcpu, including
2690 after the vcpu has been run. This will reset the vcpu to its initial
2691 state. All calls to this function after the initial call must use the same
2692 target and same set of feature flags, otherwise EINVAL will be returned.
2695 - KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state.
2696 Depends on KVM_CAP_ARM_PSCI. If not set, the CPU will be powered on
2697 and execute guest code when KVM_RUN is called.
2698 - KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode.
2699 Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only).
2700 - KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 for the CPU.
2701 Depends on KVM_CAP_ARM_PSCI_0_2.
2702 - KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU.
2703 Depends on KVM_CAP_ARM_PMU_V3.
2706 4.83 KVM_ARM_PREFERRED_TARGET
2709 Architectures: arm, arm64
2711 Parameters: struct struct kvm_vcpu_init (out)
2712 Returns: 0 on success; -1 on error
2714 ENODEV: no preferred target available for the host
2716 This queries KVM for preferred CPU target type which can be emulated
2717 by KVM on underlying host.
2719 The ioctl returns struct kvm_vcpu_init instance containing information
2720 about preferred CPU target type and recommended features for it. The
2721 kvm_vcpu_init->features bitmap returned will have feature bits set if
2722 the preferred target recommends setting these features, but this is
2725 The information returned by this ioctl can be used to prepare an instance
2726 of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in
2727 in VCPU matching underlying host.
2730 4.84 KVM_GET_REG_LIST
2733 Architectures: arm, arm64, mips
2735 Parameters: struct kvm_reg_list (in/out)
2736 Returns: 0 on success; -1 on error
2738 Â E2BIG: Â Â Â Â the reg index list is too big to fit in the array specified by
2739 Â Â Â Â Â Â Â Â Â Â Â Â the user (the number required will be written into n).
2741 struct kvm_reg_list {
2742 __u64 n; /* number of registers in reg[] */
2746 This ioctl returns the guest registers that are supported for the
2747 KVM_GET_ONE_REG/KVM_SET_ONE_REG calls.
2750 4.85 KVM_ARM_SET_DEVICE_ADDR (deprecated)
2752 Capability: KVM_CAP_ARM_SET_DEVICE_ADDR
2753 Architectures: arm, arm64
2755 Parameters: struct kvm_arm_device_address (in)
2756 Returns: 0 on success, -1 on error
2758 ENODEV: The device id is unknown
2759 ENXIO: Device not supported on current system
2760 EEXIST: Address already set
2761 E2BIG: Address outside guest physical address space
2762 EBUSY: Address overlaps with other device range
2764 struct kvm_arm_device_addr {
2769 Specify a device address in the guest's physical address space where guests
2770 can access emulated or directly exposed devices, which the host kernel needs
2771 to know about. The id field is an architecture specific identifier for a
2774 ARM/arm64 divides the id field into two parts, a device id and an
2775 address type id specific to the individual device.
2777 Â bits: | 63 ... 32 | 31 ... 16 | 15 ... 0 |
2778 field: | 0x00000000 | device id | addr type id |
2780 ARM/arm64 currently only require this when using the in-kernel GIC
2781 support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2
2782 as the device id. When setting the base address for the guest's
2783 mapping of the VGIC virtual CPU and distributor interface, the ioctl
2784 must be called after calling KVM_CREATE_IRQCHIP, but before calling
2785 KVM_RUN on any of the VCPUs. Calling this ioctl twice for any of the
2786 base addresses will return -EEXIST.
2788 Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API
2789 should be used instead.
2792 4.86 KVM_PPC_RTAS_DEFINE_TOKEN
2794 Capability: KVM_CAP_PPC_RTAS
2797 Parameters: struct kvm_rtas_token_args
2798 Returns: 0 on success, -1 on error
2800 Defines a token value for a RTAS (Run Time Abstraction Services)
2801 service in order to allow it to be handled in the kernel. The
2802 argument struct gives the name of the service, which must be the name
2803 of a service that has a kernel-side implementation. If the token
2804 value is non-zero, it will be associated with that service, and
2805 subsequent RTAS calls by the guest specifying that token will be
2806 handled by the kernel. If the token value is 0, then any token
2807 associated with the service will be forgotten, and subsequent RTAS
2808 calls by the guest for that service will be passed to userspace to be
2811 4.87 KVM_SET_GUEST_DEBUG
2813 Capability: KVM_CAP_SET_GUEST_DEBUG
2814 Architectures: x86, s390, ppc, arm64
2816 Parameters: struct kvm_guest_debug (in)
2817 Returns: 0 on success; -1 on error
2819 struct kvm_guest_debug {
2822 struct kvm_guest_debug_arch arch;
2825 Set up the processor specific debug registers and configure vcpu for
2826 handling guest debug events. There are two parts to the structure, the
2827 first a control bitfield indicates the type of debug events to handle
2828 when running. Common control bits are:
2830 - KVM_GUESTDBG_ENABLE: guest debugging is enabled
2831 - KVM_GUESTDBG_SINGLESTEP: the next run should single-step
2833 The top 16 bits of the control field are architecture specific control
2834 flags which can include the following:
2836 - KVM_GUESTDBG_USE_SW_BP: using software breakpoints [x86, arm64]
2837 - KVM_GUESTDBG_USE_HW_BP: using hardware breakpoints [x86, s390, arm64]
2838 - KVM_GUESTDBG_INJECT_DB: inject DB type exception [x86]
2839 - KVM_GUESTDBG_INJECT_BP: inject BP type exception [x86]
2840 - KVM_GUESTDBG_EXIT_PENDING: trigger an immediate guest exit [s390]
2842 For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints
2843 are enabled in memory so we need to ensure breakpoint exceptions are
2844 correctly trapped and the KVM run loop exits at the breakpoint and not
2845 running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP
2846 we need to ensure the guest vCPUs architecture specific registers are
2847 updated to the correct (supplied) values.
2849 The second part of the structure is architecture specific and
2850 typically contains a set of debug registers.
2852 For arm64 the number of debug registers is implementation defined and
2853 can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and
2854 KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number
2855 indicating the number of supported registers.
2857 When debug events exit the main run loop with the reason
2858 KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run
2859 structure containing architecture specific debug information.
2861 4.88 KVM_GET_EMULATED_CPUID
2863 Capability: KVM_CAP_EXT_EMUL_CPUID
2866 Parameters: struct kvm_cpuid2 (in/out)
2867 Returns: 0 on success, -1 on error
2872 struct kvm_cpuid_entry2 entries[0];
2875 The member 'flags' is used for passing flags from userspace.
2877 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
2878 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1)
2879 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2)
2881 struct kvm_cpuid_entry2 {
2892 This ioctl returns x86 cpuid features which are emulated by
2893 kvm.Userspace can use the information returned by this ioctl to query
2894 which features are emulated by kvm instead of being present natively.
2896 Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2
2897 structure with the 'nent' field indicating the number of entries in
2898 the variable-size array 'entries'. If the number of entries is too low
2899 to describe the cpu capabilities, an error (E2BIG) is returned. If the
2900 number is too high, the 'nent' field is adjusted and an error (ENOMEM)
2901 is returned. If the number is just right, the 'nent' field is adjusted
2902 to the number of valid entries in the 'entries' array, which is then
2905 The entries returned are the set CPUID bits of the respective features
2906 which kvm emulates, as returned by the CPUID instruction, with unknown
2907 or unsupported feature bits cleared.
2909 Features like x2apic, for example, may not be present in the host cpu
2910 but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be
2911 emulated efficiently and thus not included here.
2913 The fields in each entry are defined as follows:
2915 function: the eax value used to obtain the entry
2916 index: the ecx value used to obtain the entry (for entries that are
2918 flags: an OR of zero or more of the following:
2919 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
2920 if the index field is valid
2921 KVM_CPUID_FLAG_STATEFUL_FUNC:
2922 if cpuid for this function returns different values for successive
2923 invocations; there will be several entries with the same function,
2924 all with this flag set
2925 KVM_CPUID_FLAG_STATE_READ_NEXT:
2926 for KVM_CPUID_FLAG_STATEFUL_FUNC entries, set if this entry is
2927 the first entry to be read by a cpu
2928 eax, ebx, ecx, edx: the values returned by the cpuid instruction for
2929 this function/index combination
2931 4.89 KVM_S390_MEM_OP
2933 Capability: KVM_CAP_S390_MEM_OP
2936 Parameters: struct kvm_s390_mem_op (in)
2937 Returns: = 0 on success,
2938 < 0 on generic error (e.g. -EFAULT or -ENOMEM),
2939 > 0 if an exception occurred while walking the page tables
2941 Read or write data from/to the logical (virtual) memory of a VCPU.
2943 Parameters are specified via the following structure:
2945 struct kvm_s390_mem_op {
2946 __u64 gaddr; /* the guest address */
2947 __u64 flags; /* flags */
2948 __u32 size; /* amount of bytes */
2949 __u32 op; /* type of operation */
2950 __u64 buf; /* buffer in userspace */
2951 __u8 ar; /* the access register number */
2952 __u8 reserved[31]; /* should be set to 0 */
2955 The type of operation is specified in the "op" field. It is either
2956 KVM_S390_MEMOP_LOGICAL_READ for reading from logical memory space or
2957 KVM_S390_MEMOP_LOGICAL_WRITE for writing to logical memory space. The
2958 KVM_S390_MEMOP_F_CHECK_ONLY flag can be set in the "flags" field to check
2959 whether the corresponding memory access would create an access exception
2960 (without touching the data in the memory at the destination). In case an
2961 access exception occurred while walking the MMU tables of the guest, the
2962 ioctl returns a positive error number to indicate the type of exception.
2963 This exception is also raised directly at the corresponding VCPU if the
2964 flag KVM_S390_MEMOP_F_INJECT_EXCEPTION is set in the "flags" field.
2966 The start address of the memory region has to be specified in the "gaddr"
2967 field, and the length of the region in the "size" field. "buf" is the buffer
2968 supplied by the userspace application where the read data should be written
2969 to for KVM_S390_MEMOP_LOGICAL_READ, or where the data that should be written
2970 is stored for a KVM_S390_MEMOP_LOGICAL_WRITE. "buf" is unused and can be NULL
2971 when KVM_S390_MEMOP_F_CHECK_ONLY is specified. "ar" designates the access
2972 register number to be used.
2974 The "reserved" field is meant for future extensions. It is not used by
2975 KVM with the currently defined set of flags.
2977 4.90 KVM_S390_GET_SKEYS
2979 Capability: KVM_CAP_S390_SKEYS
2982 Parameters: struct kvm_s390_skeys
2983 Returns: 0 on success, KVM_S390_GET_KEYS_NONE if guest is not using storage
2984 keys, negative value on error
2986 This ioctl is used to get guest storage key values on the s390
2987 architecture. The ioctl takes parameters via the kvm_s390_skeys struct.
2989 struct kvm_s390_skeys {
2992 __u64 skeydata_addr;
2997 The start_gfn field is the number of the first guest frame whose storage keys
3000 The count field is the number of consecutive frames (starting from start_gfn)
3001 whose storage keys to get. The count field must be at least 1 and the maximum
3002 allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range
3003 will cause the ioctl to return -EINVAL.
3005 The skeydata_addr field is the address to a buffer large enough to hold count
3006 bytes. This buffer will be filled with storage key data by the ioctl.
3008 4.91 KVM_S390_SET_SKEYS
3010 Capability: KVM_CAP_S390_SKEYS
3013 Parameters: struct kvm_s390_skeys
3014 Returns: 0 on success, negative value on error
3016 This ioctl is used to set guest storage key values on the s390
3017 architecture. The ioctl takes parameters via the kvm_s390_skeys struct.
3018 See section on KVM_S390_GET_SKEYS for struct definition.
3020 The start_gfn field is the number of the first guest frame whose storage keys
3023 The count field is the number of consecutive frames (starting from start_gfn)
3024 whose storage keys to get. The count field must be at least 1 and the maximum
3025 allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range
3026 will cause the ioctl to return -EINVAL.
3028 The skeydata_addr field is the address to a buffer containing count bytes of
3029 storage keys. Each byte in the buffer will be set as the storage key for a
3030 single frame starting at start_gfn for count frames.
3032 Note: If any architecturally invalid key value is found in the given data then
3033 the ioctl will return -EINVAL.
3037 Capability: KVM_CAP_S390_INJECT_IRQ
3040 Parameters: struct kvm_s390_irq (in)
3041 Returns: 0 on success, -1 on error
3043 EINVAL: interrupt type is invalid
3044 type is KVM_S390_SIGP_STOP and flag parameter is invalid value
3045 type is KVM_S390_INT_EXTERNAL_CALL and code is bigger
3046 than the maximum of VCPUs
3047 EBUSY: type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped
3048 type is KVM_S390_SIGP_STOP and a stop irq is already pending
3049 type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt
3052 Allows to inject an interrupt to the guest.
3054 Using struct kvm_s390_irq as a parameter allows
3055 to inject additional payload which is not
3056 possible via KVM_S390_INTERRUPT.
3058 Interrupt parameters are passed via kvm_s390_irq:
3060 struct kvm_s390_irq {
3063 struct kvm_s390_io_info io;
3064 struct kvm_s390_ext_info ext;
3065 struct kvm_s390_pgm_info pgm;
3066 struct kvm_s390_emerg_info emerg;
3067 struct kvm_s390_extcall_info extcall;
3068 struct kvm_s390_prefix_info prefix;
3069 struct kvm_s390_stop_info stop;
3070 struct kvm_s390_mchk_info mchk;
3075 type can be one of the following:
3077 KVM_S390_SIGP_STOP - sigp stop; parameter in .stop
3078 KVM_S390_PROGRAM_INT - program check; parameters in .pgm
3079 KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix
3080 KVM_S390_RESTART - restart; no parameters
3081 KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters
3082 KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters
3083 KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg
3084 KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall
3085 KVM_S390_MCHK - machine check interrupt; parameters in .mchk
3088 Note that the vcpu ioctl is asynchronous to vcpu execution.
3090 4.94 KVM_S390_GET_IRQ_STATE
3092 Capability: KVM_CAP_S390_IRQ_STATE
3095 Parameters: struct kvm_s390_irq_state (out)
3096 Returns: >= number of bytes copied into buffer,
3097 -EINVAL if buffer size is 0,
3098 -ENOBUFS if buffer size is too small to fit all pending interrupts,
3099 -EFAULT if the buffer address was invalid
3101 This ioctl allows userspace to retrieve the complete state of all currently
3102 pending interrupts in a single buffer. Use cases include migration
3103 and introspection. The parameter structure contains the address of a
3104 userspace buffer and its length:
3106 struct kvm_s390_irq_state {
3113 Userspace passes in the above struct and for each pending interrupt a
3114 struct kvm_s390_irq is copied to the provided buffer.
3116 If -ENOBUFS is returned the buffer provided was too small and userspace
3117 may retry with a bigger buffer.
3119 4.95 KVM_S390_SET_IRQ_STATE
3121 Capability: KVM_CAP_S390_IRQ_STATE
3124 Parameters: struct kvm_s390_irq_state (in)
3125 Returns: 0 on success,
3126 -EFAULT if the buffer address was invalid,
3127 -EINVAL for an invalid buffer length (see below),
3128 -EBUSY if there were already interrupts pending,
3129 errors occurring when actually injecting the
3130 interrupt. See KVM_S390_IRQ.
3132 This ioctl allows userspace to set the complete state of all cpu-local
3133 interrupts currently pending for the vcpu. It is intended for restoring
3134 interrupt state after a migration. The input parameter is a userspace buffer
3135 containing a struct kvm_s390_irq_state:
3137 struct kvm_s390_irq_state {
3143 The userspace memory referenced by buf contains a struct kvm_s390_irq
3144 for each interrupt to be injected into the guest.
3145 If one of the interrupts could not be injected for some reason the
3148 len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0
3149 and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq),
3150 which is the maximum number of possibly pending cpu-local interrupts.
3154 Capability: KVM_CAP_X86_SMM
3158 Returns: 0 on success, -1 on error
3160 Queues an SMI on the thread's vcpu.
3162 4.97 KVM_CAP_PPC_MULTITCE
3164 Capability: KVM_CAP_PPC_MULTITCE
3168 This capability means the kernel is capable of handling hypercalls
3169 H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user
3170 space. This significantly accelerates DMA operations for PPC KVM guests.
3171 User space should expect that its handlers for these hypercalls
3172 are not going to be called if user space previously registered LIOBN
3173 in KVM (via KVM_CREATE_SPAPR_TCE or similar calls).
3175 In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest,
3176 user space might have to advertise it for the guest. For example,
3177 IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is
3178 present in the "ibm,hypertas-functions" device-tree property.
3180 The hypercalls mentioned above may or may not be processed successfully
3181 in the kernel based fast path. If they can not be handled by the kernel,
3182 they will get passed on to user space. So user space still has to have
3183 an implementation for these despite the in kernel acceleration.
3185 This capability is always enabled.
3187 4.98 KVM_CREATE_SPAPR_TCE_64
3189 Capability: KVM_CAP_SPAPR_TCE_64
3190 Architectures: powerpc
3192 Parameters: struct kvm_create_spapr_tce_64 (in)
3193 Returns: file descriptor for manipulating the created TCE table
3195 This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit
3196 windows, described in 4.62 KVM_CREATE_SPAPR_TCE
3198 This capability uses extended struct in ioctl interface:
3200 /* for KVM_CAP_SPAPR_TCE_64 */
3201 struct kvm_create_spapr_tce_64 {
3205 __u64 offset; /* in pages */
3206 __u64 size; /* in pages */
3209 The aim of extension is to support an additional bigger DMA window with
3210 a variable page size.
3211 KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and
3212 a bus offset of the corresponding DMA window, @size and @offset are numbers
3215 @flags are not used at the moment.
3217 The rest of functionality is identical to KVM_CREATE_SPAPR_TCE.
3219 4.99 KVM_REINJECT_CONTROL
3221 Capability: KVM_CAP_REINJECT_CONTROL
3224 Parameters: struct kvm_reinject_control (in)
3225 Returns: 0 on success,
3226 -EFAULT if struct kvm_reinject_control cannot be read,
3227 -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier.
3229 i8254 (PIT) has two modes, reinject and !reinject. The default is reinject,
3230 where KVM queues elapsed i8254 ticks and monitors completion of interrupt from
3231 vector(s) that i8254 injects. Reinject mode dequeues a tick and injects its
3232 interrupt whenever there isn't a pending interrupt from i8254.
3233 !reinject mode injects an interrupt as soon as a tick arrives.
3235 struct kvm_reinject_control {
3240 pit_reinject = 0 (!reinject mode) is recommended, unless running an old
3241 operating system that uses the PIT for timing (e.g. Linux 2.4.x).
3243 4.100 KVM_PPC_CONFIGURE_V3_MMU
3245 Capability: KVM_CAP_PPC_RADIX_MMU or KVM_CAP_PPC_HASH_MMU_V3
3248 Parameters: struct kvm_ppc_mmuv3_cfg (in)
3249 Returns: 0 on success,
3250 -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read,
3251 -EINVAL if the configuration is invalid
3253 This ioctl controls whether the guest will use radix or HPT (hashed
3254 page table) translation, and sets the pointer to the process table for
3257 struct kvm_ppc_mmuv3_cfg {
3259 __u64 process_table;
3262 There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and
3263 KVM_PPC_MMUV3_GTSE. KVM_PPC_MMUV3_RADIX, if set, configures the guest
3264 to use radix tree translation, and if clear, to use HPT translation.
3265 KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest
3266 to be able to use the global TLB and SLB invalidation instructions;
3267 if clear, the guest may not use these instructions.
3269 The process_table field specifies the address and size of the guest
3270 process table, which is in the guest's space. This field is formatted
3271 as the second doubleword of the partition table entry, as defined in
3272 the Power ISA V3.00, Book III section 5.7.6.1.
3274 4.101 KVM_PPC_GET_RMMU_INFO
3276 Capability: KVM_CAP_PPC_RADIX_MMU
3279 Parameters: struct kvm_ppc_rmmu_info (out)
3280 Returns: 0 on success,
3281 -EFAULT if struct kvm_ppc_rmmu_info cannot be written,
3282 -EINVAL if no useful information can be returned
3284 This ioctl returns a structure containing two things: (a) a list
3285 containing supported radix tree geometries, and (b) a list that maps
3286 page sizes to put in the "AP" (actual page size) field for the tlbie
3287 (TLB invalidate entry) instruction.
3289 struct kvm_ppc_rmmu_info {
3290 struct kvm_ppc_radix_geom {
3295 __u32 ap_encodings[8];
3298 The geometries[] field gives up to 8 supported geometries for the
3299 radix page table, in terms of the log base 2 of the smallest page
3300 size, and the number of bits indexed at each level of the tree, from
3301 the PTE level up to the PGD level in that order. Any unused entries
3302 will have 0 in the page_shift field.
3304 The ap_encodings gives the supported page sizes and their AP field
3305 encodings, encoded with the AP value in the top 3 bits and the log
3306 base 2 of the page size in the bottom 6 bits.
3308 4.102 KVM_PPC_RESIZE_HPT_PREPARE
3310 Capability: KVM_CAP_SPAPR_RESIZE_HPT
3311 Architectures: powerpc
3313 Parameters: struct kvm_ppc_resize_hpt (in)
3314 Returns: 0 on successful completion,
3315 >0 if a new HPT is being prepared, the value is an estimated
3316 number of milliseconds until preparation is complete
3317 -EFAULT if struct kvm_reinject_control cannot be read,
3318 -EINVAL if the supplied shift or flags are invalid
3319 -ENOMEM if unable to allocate the new HPT
3320 -ENOSPC if there was a hash collision when moving existing
3321 HPT entries to the new HPT
3322 -EIO on other error conditions
3324 Used to implement the PAPR extension for runtime resizing of a guest's
3325 Hashed Page Table (HPT). Specifically this starts, stops or monitors
3326 the preparation of a new potential HPT for the guest, essentially
3327 implementing the H_RESIZE_HPT_PREPARE hypercall.
3329 If called with shift > 0 when there is no pending HPT for the guest,
3330 this begins preparation of a new pending HPT of size 2^(shift) bytes.
3331 It then returns a positive integer with the estimated number of
3332 milliseconds until preparation is complete.
3334 If called when there is a pending HPT whose size does not match that
3335 requested in the parameters, discards the existing pending HPT and
3336 creates a new one as above.
3338 If called when there is a pending HPT of the size requested, will:
3339 * If preparation of the pending HPT is already complete, return 0
3340 * If preparation of the pending HPT has failed, return an error
3341 code, then discard the pending HPT.
3342 * If preparation of the pending HPT is still in progress, return an
3343 estimated number of milliseconds until preparation is complete.
3345 If called with shift == 0, discards any currently pending HPT and
3346 returns 0 (i.e. cancels any in-progress preparation).
3348 flags is reserved for future expansion, currently setting any bits in
3349 flags will result in an -EINVAL.
3351 Normally this will be called repeatedly with the same parameters until
3352 it returns <= 0. The first call will initiate preparation, subsequent
3353 ones will monitor preparation until it completes or fails.
3355 struct kvm_ppc_resize_hpt {
3361 4.103 KVM_PPC_RESIZE_HPT_COMMIT
3363 Capability: KVM_CAP_SPAPR_RESIZE_HPT
3364 Architectures: powerpc
3366 Parameters: struct kvm_ppc_resize_hpt (in)
3367 Returns: 0 on successful completion,
3368 -EFAULT if struct kvm_reinject_control cannot be read,
3369 -EINVAL if the supplied shift or flags are invalid
3370 -ENXIO is there is no pending HPT, or the pending HPT doesn't
3371 have the requested size
3372 -EBUSY if the pending HPT is not fully prepared
3373 -ENOSPC if there was a hash collision when moving existing
3374 HPT entries to the new HPT
3375 -EIO on other error conditions
3377 Used to implement the PAPR extension for runtime resizing of a guest's
3378 Hashed Page Table (HPT). Specifically this requests that the guest be
3379 transferred to working with the new HPT, essentially implementing the
3380 H_RESIZE_HPT_COMMIT hypercall.
3382 This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has
3383 returned 0 with the same parameters. In other cases
3384 KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or
3385 -EBUSY, though others may be possible if the preparation was started,
3388 This will have undefined effects on the guest if it has not already
3389 placed itself in a quiescent state where no vcpu will make MMU enabled
3392 On succsful completion, the pending HPT will become the guest's active
3393 HPT and the previous HPT will be discarded.
3395 On failure, the guest will still be operating on its previous HPT.
3397 struct kvm_ppc_resize_hpt {
3403 5. The kvm_run structure
3404 ------------------------
3406 Application code obtains a pointer to the kvm_run structure by
3407 mmap()ing a vcpu fd. From that point, application code can control
3408 execution by changing fields in kvm_run prior to calling the KVM_RUN
3409 ioctl, and obtain information about the reason KVM_RUN returned by
3410 looking up structure members.
3414 __u8 request_interrupt_window;
3416 Request that KVM_RUN return when it becomes possible to inject external
3417 interrupts into the guest. Useful in conjunction with KVM_INTERRUPT.
3419 __u8 immediate_exit;
3421 This field is polled once when KVM_RUN starts; if non-zero, KVM_RUN
3422 exits immediately, returning -EINTR. In the common scenario where a
3423 signal is used to "kick" a VCPU out of KVM_RUN, this field can be used
3424 to avoid usage of KVM_SET_SIGNAL_MASK, which has worse scalability.
3425 Rather than blocking the signal outside KVM_RUN, userspace can set up
3426 a signal handler that sets run->immediate_exit to a non-zero value.
3428 This field is ignored if KVM_CAP_IMMEDIATE_EXIT is not available.
3435 When KVM_RUN has returned successfully (return value 0), this informs
3436 application code why KVM_RUN has returned. Allowable values for this
3437 field are detailed below.
3439 __u8 ready_for_interrupt_injection;
3441 If request_interrupt_window has been specified, this field indicates
3442 an interrupt can be injected now with KVM_INTERRUPT.
3446 The value of the current interrupt flag. Only valid if in-kernel
3447 local APIC is not used.
3451 More architecture-specific flags detailing state of the VCPU that may
3452 affect the device's behavior. The only currently defined flag is
3453 KVM_RUN_X86_SMM, which is valid on x86 machines and is set if the
3454 VCPU is in system management mode.
3456 /* in (pre_kvm_run), out (post_kvm_run) */
3459 The value of the cr8 register. Only valid if in-kernel local APIC is
3460 not used. Both input and output.
3464 The value of the APIC BASE msr. Only valid if in-kernel local
3465 APIC is not used. Both input and output.
3468 /* KVM_EXIT_UNKNOWN */
3470 __u64 hardware_exit_reason;
3473 If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown
3474 reasons. Further architecture-specific information is available in
3475 hardware_exit_reason.
3477 /* KVM_EXIT_FAIL_ENTRY */
3479 __u64 hardware_entry_failure_reason;
3482 If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due
3483 to unknown reasons. Further architecture-specific information is
3484 available in hardware_entry_failure_reason.
3486 /* KVM_EXIT_EXCEPTION */
3496 #define KVM_EXIT_IO_IN 0
3497 #define KVM_EXIT_IO_OUT 1
3499 __u8 size; /* bytes */
3502 __u64 data_offset; /* relative to kvm_run start */
3505 If exit_reason is KVM_EXIT_IO, then the vcpu has
3506 executed a port I/O instruction which could not be satisfied by kvm.
3507 data_offset describes where the data is located (KVM_EXIT_IO_OUT) or
3508 where kvm expects application code to place the data for the next
3509 KVM_RUN invocation (KVM_EXIT_IO_IN). Data format is a packed array.
3511 /* KVM_EXIT_DEBUG */
3513 struct kvm_debug_exit_arch arch;
3516 If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event
3517 for which architecture specific information is returned.
3527 If exit_reason is KVM_EXIT_MMIO, then the vcpu has
3528 executed a memory-mapped I/O instruction which could not be satisfied
3529 by kvm. The 'data' member contains the written data if 'is_write' is
3530 true, and should be filled by application code otherwise.
3532 The 'data' member contains, in its first 'len' bytes, the value as it would
3533 appear if the VCPU performed a load or store of the appropriate width directly
3536 NOTE: For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR and
3537 KVM_EXIT_EPR the corresponding
3538 operations are complete (and guest state is consistent) only after userspace
3539 has re-entered the kernel with KVM_RUN. The kernel side will first finish
3540 incomplete operations and then check for pending signals. Userspace
3541 can re-enter the guest with an unmasked signal pending to complete
3544 /* KVM_EXIT_HYPERCALL */
3553 Unused. This was once used for 'hypercall to userspace'. To implement
3554 such functionality, use KVM_EXIT_IO (x86) or KVM_EXIT_MMIO (all except s390).
3555 Note KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO.
3557 /* KVM_EXIT_TPR_ACCESS */
3564 To be documented (KVM_TPR_ACCESS_REPORTING).
3566 /* KVM_EXIT_S390_SIEIC */
3569 __u64 mask; /* psw upper half */
3570 __u64 addr; /* psw lower half */
3577 /* KVM_EXIT_S390_RESET */
3578 #define KVM_S390_RESET_POR 1
3579 #define KVM_S390_RESET_CLEAR 2
3580 #define KVM_S390_RESET_SUBSYSTEM 4
3581 #define KVM_S390_RESET_CPU_INIT 8
3582 #define KVM_S390_RESET_IPL 16
3583 __u64 s390_reset_flags;
3587 /* KVM_EXIT_S390_UCONTROL */
3589 __u64 trans_exc_code;
3593 s390 specific. A page fault has occurred for a user controlled virtual
3594 machine (KVM_VM_S390_UNCONTROL) on it's host page table that cannot be
3595 resolved by the kernel.
3596 The program code and the translation exception code that were placed
3597 in the cpu's lowcore are presented here as defined by the z Architecture
3598 Principles of Operation Book in the Chapter for Dynamic Address Translation
3608 Deprecated - was used for 440 KVM.
3615 MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch
3616 hypercalls and exit with this exit struct that contains all the guest gprs.
3618 If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall.
3619 Userspace can now handle the hypercall and when it's done modify the gprs as
3620 necessary. Upon guest entry all guest GPRs will then be replaced by the values
3623 /* KVM_EXIT_PAPR_HCALL */
3630 This is used on 64-bit PowerPC when emulating a pSeries partition,
3631 e.g. with the 'pseries' machine type in qemu. It occurs when the
3632 guest does a hypercall using the 'sc 1' instruction. The 'nr' field
3633 contains the hypercall number (from the guest R3), and 'args' contains
3634 the arguments (from the guest R4 - R12). Userspace should put the
3635 return code in 'ret' and any extra returned values in args[].
3636 The possible hypercalls are defined in the Power Architecture Platform
3637 Requirements (PAPR) document available from www.power.org (free
3638 developer registration required to access it).
3640 /* KVM_EXIT_S390_TSCH */
3642 __u16 subchannel_id;
3643 __u16 subchannel_nr;
3650 s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled
3651 and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O
3652 interrupt for the target subchannel has been dequeued and subchannel_id,
3653 subchannel_nr, io_int_parm and io_int_word contain the parameters for that
3654 interrupt. ipb is needed for instruction parameter decoding.
3661 On FSL BookE PowerPC chips, the interrupt controller has a fast patch
3662 interrupt acknowledge path to the core. When the core successfully
3663 delivers an interrupt, it automatically populates the EPR register with
3664 the interrupt vector number and acknowledges the interrupt inside
3665 the interrupt controller.
3667 In case the interrupt controller lives in user space, we need to do
3668 the interrupt acknowledge cycle through it to fetch the next to be
3669 delivered interrupt vector using this exit.
3671 It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an
3672 external interrupt has just been delivered into the guest. User space
3673 should put the acknowledged interrupt vector into the 'epr' field.
3675 /* KVM_EXIT_SYSTEM_EVENT */
3677 #define KVM_SYSTEM_EVENT_SHUTDOWN 1
3678 #define KVM_SYSTEM_EVENT_RESET 2
3679 #define KVM_SYSTEM_EVENT_CRASH 3
3684 If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered
3685 a system-level event using some architecture specific mechanism (hypercall
3686 or some special instruction). In case of ARM/ARM64, this is triggered using
3687 HVC instruction based PSCI call from the vcpu. The 'type' field describes
3688 the system-level event type. The 'flags' field describes architecture
3689 specific flags for the system-level event.
3691 Valid values for 'type' are:
3692 KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the
3693 VM. Userspace is not obliged to honour this, and if it does honour
3694 this does not need to destroy the VM synchronously (ie it may call
3695 KVM_RUN again before shutdown finally occurs).
3696 KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM.
3697 As with SHUTDOWN, userspace can choose to ignore the request, or
3698 to schedule the reset to occur in the future and may call KVM_RUN again.
3699 KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest
3700 has requested a crash condition maintenance. Userspace can choose
3701 to ignore the request, or to gather VM memory core dump and/or
3702 reset/shutdown of the VM.
3704 /* KVM_EXIT_IOAPIC_EOI */
3709 Indicates that the VCPU's in-kernel local APIC received an EOI for a
3710 level-triggered IOAPIC interrupt. This exit only triggers when the
3711 IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled);
3712 the userspace IOAPIC should process the EOI and retrigger the interrupt if
3713 it is still asserted. Vector is the LAPIC interrupt vector for which the
3716 struct kvm_hyperv_exit {
3717 #define KVM_EXIT_HYPERV_SYNIC 1
3718 #define KVM_EXIT_HYPERV_HCALL 2
3734 /* KVM_EXIT_HYPERV */
3735 struct kvm_hyperv_exit hyperv;
3736 Indicates that the VCPU exits into userspace to process some tasks
3737 related to Hyper-V emulation.
3738 Valid values for 'type' are:
3739 KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about
3740 Hyper-V SynIC state change. Notification is used to remap SynIC
3741 event/message pages and to enable/disable SynIC messages/events processing
3744 /* Fix the size of the union. */
3749 * shared registers between kvm and userspace.
3750 * kvm_valid_regs specifies the register classes set by the host
3751 * kvm_dirty_regs specified the register classes dirtied by userspace
3752 * struct kvm_sync_regs is architecture specific, as well as the
3753 * bits for kvm_valid_regs and kvm_dirty_regs
3755 __u64 kvm_valid_regs;
3756 __u64 kvm_dirty_regs;
3758 struct kvm_sync_regs regs;
3762 If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access
3763 certain guest registers without having to call SET/GET_*REGS. Thus we can
3764 avoid some system call overhead if userspace has to handle the exit.
3765 Userspace can query the validity of the structure by checking
3766 kvm_valid_regs for specific bits. These bits are architecture specific
3767 and usually define the validity of a groups of registers. (e.g. one bit
3768 for general purpose registers)
3770 Please note that the kernel is allowed to use the kvm_run structure as the
3771 primary storage for certain register types. Therefore, the kernel may use the
3772 values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set.
3778 6. Capabilities that can be enabled on vCPUs
3779 --------------------------------------------
3781 There are certain capabilities that change the behavior of the virtual CPU or
3782 the virtual machine when enabled. To enable them, please see section 4.37.
3783 Below you can find a list of capabilities and what their effect on the vCPU or
3784 the virtual machine is when enabling them.
3786 The following information is provided along with the description:
3788 Architectures: which instruction set architectures provide this ioctl.
3789 x86 includes both i386 and x86_64.
3791 Target: whether this is a per-vcpu or per-vm capability.
3793 Parameters: what parameters are accepted by the capability.
3795 Returns: the return value. General error numbers (EBADF, ENOMEM, EINVAL)
3796 are not detailed, but errors with specific meanings are.
3804 Returns: 0 on success; -1 on error
3806 This capability enables interception of OSI hypercalls that otherwise would
3807 be treated as normal system calls to be injected into the guest. OSI hypercalls
3808 were invented by Mac-on-Linux to have a standardized communication mechanism
3809 between the guest and the host.
3811 When this capability is enabled, KVM_EXIT_OSI can occur.
3814 6.2 KVM_CAP_PPC_PAPR
3819 Returns: 0 on success; -1 on error
3821 This capability enables interception of PAPR hypercalls. PAPR hypercalls are
3822 done using the hypercall instruction "sc 1".
3824 It also sets the guest privilege level to "supervisor" mode. Usually the guest
3825 runs in "hypervisor" privilege mode with a few missing features.
3827 In addition to the above, it changes the semantics of SDR1. In this mode, the
3828 HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the
3829 HTAB invisible to the guest.
3831 When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur.
3838 Parameters: args[0] is the address of a struct kvm_config_tlb
3839 Returns: 0 on success; -1 on error
3841 struct kvm_config_tlb {
3848 Configures the virtual CPU's TLB array, establishing a shared memory area
3849 between userspace and KVM. The "params" and "array" fields are userspace
3850 addresses of mmu-type-specific data structures. The "array_len" field is an
3851 safety mechanism, and should be set to the size in bytes of the memory that
3852 userspace has reserved for the array. It must be at least the size dictated
3853 by "mmu_type" and "params".
3855 While KVM_RUN is active, the shared region is under control of KVM. Its
3856 contents are undefined, and any modification by userspace results in
3857 boundedly undefined behavior.
3859 On return from KVM_RUN, the shared region will reflect the current state of
3860 the guest's TLB. If userspace makes any changes, it must call KVM_DIRTY_TLB
3861 to tell KVM which entries have been changed, prior to calling KVM_RUN again
3864 For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV:
3865 - The "params" field is of type "struct kvm_book3e_206_tlb_params".
3866 - The "array" field points to an array of type "struct
3867 kvm_book3e_206_tlb_entry".
3868 - The array consists of all entries in the first TLB, followed by all
3869 entries in the second TLB.
3870 - Within a TLB, entries are ordered first by increasing set number. Within a
3871 set, entries are ordered by way (increasing ESEL).
3872 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1)
3873 where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value.
3874 - The tsize field of mas1 shall be set to 4K on TLB0, even though the
3875 hardware ignores this value for TLB0.
3877 6.4 KVM_CAP_S390_CSS_SUPPORT
3882 Returns: 0 on success; -1 on error
3884 This capability enables support for handling of channel I/O instructions.
3886 TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are
3887 handled in-kernel, while the other I/O instructions are passed to userspace.
3889 When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST
3890 SUBCHANNEL intercepts.
3892 Note that even though this capability is enabled per-vcpu, the complete
3893 virtual machine is affected.
3899 Parameters: args[0] defines whether the proxy facility is active
3900 Returns: 0 on success; -1 on error
3902 This capability enables or disables the delivery of interrupts through the
3903 external proxy facility.
3905 When enabled (args[0] != 0), every time the guest gets an external interrupt
3906 delivered, it automatically exits into user space with a KVM_EXIT_EPR exit
3907 to receive the topmost interrupt vector.
3909 When disabled (args[0] == 0), behavior is as if this facility is unsupported.
3911 When this capability is enabled, KVM_EXIT_EPR can occur.
3913 6.6 KVM_CAP_IRQ_MPIC
3916 Parameters: args[0] is the MPIC device fd
3917 args[1] is the MPIC CPU number for this vcpu
3919 This capability connects the vcpu to an in-kernel MPIC device.
3921 6.7 KVM_CAP_IRQ_XICS
3925 Parameters: args[0] is the XICS device fd
3926 args[1] is the XICS CPU number (server ID) for this vcpu
3928 This capability connects the vcpu to an in-kernel XICS device.
3930 6.8 KVM_CAP_S390_IRQCHIP
3936 This capability enables the in-kernel irqchip for s390. Please refer to
3937 "4.24 KVM_CREATE_IRQCHIP" for details.
3939 6.9 KVM_CAP_MIPS_FPU
3943 Parameters: args[0] is reserved for future use (should be 0).
3945 This capability allows the use of the host Floating Point Unit by the guest. It
3946 allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is
3947 done the KVM_REG_MIPS_FPR_* and KVM_REG_MIPS_FCR_* registers can be accessed
3948 (depending on the current guest FPU register mode), and the Status.FR,
3949 Config5.FRE bits are accessible via the KVM API and also from the guest,
3950 depending on them being supported by the FPU.
3952 6.10 KVM_CAP_MIPS_MSA
3956 Parameters: args[0] is reserved for future use (should be 0).
3958 This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest.
3959 It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest.
3960 Once this is done the KVM_REG_MIPS_VEC_* and KVM_REG_MIPS_MSA_* registers can be
3961 accessed, and the Config5.MSAEn bit is accessible via the KVM API and also from
3964 7. Capabilities that can be enabled on VMs
3965 ------------------------------------------
3967 There are certain capabilities that change the behavior of the virtual
3968 machine when enabled. To enable them, please see section 4.37. Below
3969 you can find a list of capabilities and what their effect on the VM
3970 is when enabling them.
3972 The following information is provided along with the description:
3974 Architectures: which instruction set architectures provide this ioctl.
3975 x86 includes both i386 and x86_64.
3977 Parameters: what parameters are accepted by the capability.
3979 Returns: the return value. General error numbers (EBADF, ENOMEM, EINVAL)
3980 are not detailed, but errors with specific meanings are.
3983 7.1 KVM_CAP_PPC_ENABLE_HCALL
3986 Parameters: args[0] is the sPAPR hcall number
3987 args[1] is 0 to disable, 1 to enable in-kernel handling
3989 This capability controls whether individual sPAPR hypercalls (hcalls)
3990 get handled by the kernel or not. Enabling or disabling in-kernel
3991 handling of an hcall is effective across the VM. On creation, an
3992 initial set of hcalls are enabled for in-kernel handling, which
3993 consists of those hcalls for which in-kernel handlers were implemented
3994 before this capability was implemented. If disabled, the kernel will
3995 not to attempt to handle the hcall, but will always exit to userspace
3996 to handle it. Note that it may not make sense to enable some and
3997 disable others of a group of related hcalls, but KVM does not prevent
3998 userspace from doing that.
4000 If the hcall number specified is not one that has an in-kernel
4001 implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL
4004 7.2 KVM_CAP_S390_USER_SIGP
4009 This capability controls which SIGP orders will be handled completely in user
4010 space. With this capability enabled, all fast orders will be handled completely
4016 - CONDITIONAL EMERGENCY SIGNAL
4018 All other orders will be handled completely in user space.
4020 Only privileged operation exceptions will be checked for in the kernel (or even
4021 in the hardware prior to interception). If this capability is not enabled, the
4022 old way of handling SIGP orders is used (partially in kernel and user space).
4024 7.3 KVM_CAP_S390_VECTOR_REGISTERS
4028 Returns: 0 on success, negative value on error
4030 Allows use of the vector registers introduced with z13 processor, and
4031 provides for the synchronization between host and user space. Will
4032 return -EINVAL if the machine does not support vectors.
4034 7.4 KVM_CAP_S390_USER_STSI
4039 This capability allows post-handlers for the STSI instruction. After
4040 initial handling in the kernel, KVM exits to user space with
4041 KVM_EXIT_S390_STSI to allow user space to insert further data.
4043 Before exiting to userspace, kvm handlers should fill in s390_stsi field of
4054 @addr - guest address of STSI SYSIB
4058 @ar - access register number
4060 KVM handlers should exit to userspace with rc = -EREMOTE.
4062 7.5 KVM_CAP_SPLIT_IRQCHIP
4065 Parameters: args[0] - number of routes reserved for userspace IOAPICs
4066 Returns: 0 on success, -1 on error
4068 Create a local apic for each processor in the kernel. This can be used
4069 instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the
4070 IOAPIC and PIC (and also the PIT, even though this has to be enabled
4073 This capability also enables in kernel routing of interrupt requests;
4074 when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are
4075 used in the IRQ routing table. The first args[0] MSI routes are reserved
4076 for the IOAPIC pins. Whenever the LAPIC receives an EOI for these routes,
4077 a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace.
4079 Fails if VCPU has already been created, or if the irqchip is already in the
4080 kernel (i.e. KVM_CREATE_IRQCHIP has already been called).
4087 Allows use of runtime-instrumentation introduced with zEC12 processor.
4088 Will return -EINVAL if the machine does not support runtime-instrumentation.
4089 Will return -EBUSY if a VCPU has already been created.
4091 7.7 KVM_CAP_X2APIC_API
4094 Parameters: args[0] - features that should be enabled
4095 Returns: 0 on success, -EINVAL when args[0] contains invalid features
4097 Valid feature flags in args[0] are
4099 #define KVM_X2APIC_API_USE_32BIT_IDS (1ULL << 0)
4100 #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK (1ULL << 1)
4102 Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of
4103 KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC,
4104 allowing the use of 32-bit APIC IDs. See KVM_CAP_X2APIC_API in their
4105 respective sections.
4107 KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work
4108 in logical mode or with more than 255 VCPUs. Otherwise, KVM treats 0xff
4109 as a broadcast even in x2APIC mode in order to support physical x2APIC
4110 without interrupt remapping. This is undesirable in logical mode,
4111 where 0xff represents CPUs 0-7 in cluster 0.
4113 7.8 KVM_CAP_S390_USER_INSTR0
4118 With this capability enabled, all illegal instructions 0x0000 (2 bytes) will
4119 be intercepted and forwarded to user space. User space can use this
4120 mechanism e.g. to realize 2-byte software breakpoints. The kernel will
4121 not inject an operating exception for these instructions, user space has
4122 to take care of that.
4124 This capability can be enabled dynamically even if VCPUs were already
4125 created and are running.
4127 8. Other capabilities.
4128 ----------------------
4130 This section lists capabilities that give information about other
4131 features of the KVM implementation.
4133 8.1 KVM_CAP_PPC_HWRNG
4137 This capability, if KVM_CHECK_EXTENSION indicates that it is
4138 available, means that that the kernel has an implementation of the
4139 H_RANDOM hypercall backed by a hardware random-number generator.
4140 If present, the kernel H_RANDOM handler can be enabled for guest use
4141 with the KVM_CAP_PPC_ENABLE_HCALL capability.
4143 8.2 KVM_CAP_HYPERV_SYNIC
4146 This capability, if KVM_CHECK_EXTENSION indicates that it is
4147 available, means that that the kernel has an implementation of the
4148 Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is
4149 used to support Windows Hyper-V based guest paravirt drivers(VMBus).
4151 In order to use SynIC, it has to be activated by setting this
4152 capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this
4153 will disable the use of APIC hardware virtualization even if supported
4154 by the CPU, as it's incompatible with SynIC auto-EOI behavior.
4156 8.3 KVM_CAP_PPC_RADIX_MMU
4160 This capability, if KVM_CHECK_EXTENSION indicates that it is
4161 available, means that that the kernel can support guests using the
4162 radix MMU defined in Power ISA V3.00 (as implemented in the POWER9
4165 8.4 KVM_CAP_PPC_HASH_MMU_V3
4169 This capability, if KVM_CHECK_EXTENSION indicates that it is
4170 available, means that that the kernel can support guests using the
4171 hashed page table MMU defined in Power ISA V3.00 (as implemented in
4172 the POWER9 processor), including in-memory segment tables.
4178 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
4179 it is available, means that full hardware assisted virtualization capabilities
4180 of the hardware are available for use through KVM. An appropriate
4181 KVM_VM_MIPS_* type must be passed to KVM_CREATE_VM to create a VM which
4184 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
4185 available, it means that the VM is using full hardware assisted virtualization
4186 capabilities of the hardware. This is useful to check after creating a VM with
4187 KVM_VM_MIPS_DEFAULT.
4189 The value returned by KVM_CHECK_EXTENSION should be compared against known
4190 values (see below). All other values are reserved. This is to allow for the
4191 possibility of other hardware assisted virtualization implementations which
4192 may be incompatible with the MIPS VZ ASE.
4194 0: The trap & emulate implementation is in use to run guest code in user
4195 mode. Guest virtual memory segments are rearranged to fit the guest in the
4196 user mode address space.
4198 1: The MIPS VZ ASE is in use, providing full hardware assisted
4199 virtualization, including standard guest virtual memory segments.
4205 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
4206 it is available, means that the trap & emulate implementation is available to
4207 run guest code in user mode, even if KVM_CAP_MIPS_VZ indicates that hardware
4208 assisted virtualisation is also available. KVM_VM_MIPS_TE (0) must be passed
4209 to KVM_CREATE_VM to create a VM which utilises it.
4211 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
4212 available, it means that the VM is using trap & emulate.
4214 8.7 KVM_CAP_MIPS_64BIT
4218 This capability indicates the supported architecture type of the guest, i.e. the
4219 supported register and address width.
4221 The values returned when this capability is checked by KVM_CHECK_EXTENSION on a
4222 kvm VM handle correspond roughly to the CP0_Config.AT register field, and should
4223 be checked specifically against known values (see below). All other values are
4226 0: MIPS32 or microMIPS32.
4227 Both registers and addresses are 32-bits wide.
4228 It will only be possible to run 32-bit guest code.
4230 1: MIPS64 or microMIPS64 with access only to 32-bit compatibility segments.
4231 Registers are 64-bits wide, but addresses are 32-bits wide.
4232 64-bit guest code may run but cannot access MIPS64 memory segments.
4233 It will also be possible to run 32-bit guest code.
4235 2: MIPS64 or microMIPS64 with access to all address segments.
4236 Both registers and addresses are 64-bits wide.
4237 It will be possible to run 64-bit or 32-bit guest code.