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 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), or a KVM_CAP_xyz constant, which
72 means availability needs to be checked with KVM_CHECK_EXTENSION
75 Architectures: which instruction set architectures provide this ioctl.
76 x86 includes both i386 and x86_64.
78 Type: system, vm, or vcpu.
80 Parameters: what parameters are accepted by the ioctl.
82 Returns: the return value. General error numbers (EBADF, ENOMEM, EINVAL)
83 are not detailed, but errors with specific meanings are.
86 4.1 KVM_GET_API_VERSION
92 Returns: the constant KVM_API_VERSION (=12)
94 This identifies the API version as the stable kvm API. It is not
95 expected that this number will change. However, Linux 2.6.20 and
96 2.6.21 report earlier versions; these are not documented and not
97 supported. Applications should refuse to run if KVM_GET_API_VERSION
98 returns a value other than 12. If this check passes, all ioctls
99 described as 'basic' will be available.
107 Parameters: machine type identifier (KVM_VM_*)
108 Returns: a VM fd that can be used to control the new virtual machine.
110 The new VM has no virtual cpus and no memory. An mmap() of a VM fd
111 will access the virtual machine's physical address space; offset zero
112 corresponds to guest physical address zero. Use of mmap() on a VM fd
113 is discouraged if userspace memory allocation (KVM_CAP_USER_MEMORY) is
115 You most certainly want to use 0 as machine type.
117 In order to create user controlled virtual machines on S390, check
118 KVM_CAP_S390_UCONTROL and use the flag KVM_VM_S390_UCONTROL as
119 privileged user (CAP_SYS_ADMIN).
122 4.3 KVM_GET_MSR_INDEX_LIST
127 Parameters: struct kvm_msr_list (in/out)
128 Returns: 0 on success; -1 on error
130 E2BIG: the msr index list is to be to fit in the array specified by
133 struct kvm_msr_list {
134 __u32 nmsrs; /* number of msrs in entries */
138 This ioctl returns the guest msrs that are supported. The list varies
139 by kvm version and host processor, but does not change otherwise. The
140 user fills in the size of the indices array in nmsrs, and in return
141 kvm adjusts nmsrs to reflect the actual number of msrs and fills in
142 the indices array with their numbers.
144 Note: if kvm indicates supports MCE (KVM_CAP_MCE), then the MCE bank MSRs are
145 not returned in the MSR list, as different vcpus can have a different number
146 of banks, as set via the KVM_X86_SETUP_MCE ioctl.
149 4.4 KVM_CHECK_EXTENSION
151 Capability: basic, KVM_CAP_CHECK_EXTENSION_VM for vm ioctl
153 Type: system ioctl, vm ioctl
154 Parameters: extension identifier (KVM_CAP_*)
155 Returns: 0 if unsupported; 1 (or some other positive integer) if supported
157 The API allows the application to query about extensions to the core
158 kvm API. Userspace passes an extension identifier (an integer) and
159 receives an integer that describes the extension availability.
160 Generally 0 means no and 1 means yes, but some extensions may report
161 additional information in the integer return value.
163 Based on their initialization different VMs may have different capabilities.
164 It is thus encouraged to use the vm ioctl to query for capabilities (available
165 with KVM_CAP_CHECK_EXTENSION_VM on the vm fd)
167 4.5 KVM_GET_VCPU_MMAP_SIZE
173 Returns: size of vcpu mmap area, in bytes
175 The KVM_RUN ioctl (cf.) communicates with userspace via a shared
176 memory region. This ioctl returns the size of that region. See the
177 KVM_RUN documentation for details.
180 4.6 KVM_SET_MEMORY_REGION
185 Parameters: struct kvm_memory_region (in)
186 Returns: 0 on success, -1 on error
188 This ioctl is obsolete and has been removed.
196 Parameters: vcpu id (apic id on x86)
197 Returns: vcpu fd on success, -1 on error
199 This API adds a vcpu to a virtual machine. The vcpu id is a small integer
200 in the range [0, max_vcpus).
202 The recommended max_vcpus value can be retrieved using the KVM_CAP_NR_VCPUS of
203 the KVM_CHECK_EXTENSION ioctl() at run-time.
204 The maximum possible value for max_vcpus can be retrieved using the
205 KVM_CAP_MAX_VCPUS of the KVM_CHECK_EXTENSION ioctl() at run-time.
207 If the KVM_CAP_NR_VCPUS does not exist, you should assume that max_vcpus is 4
209 If the KVM_CAP_MAX_VCPUS does not exist, you should assume that max_vcpus is
210 same as the value returned from KVM_CAP_NR_VCPUS.
212 On powerpc using book3s_hv mode, the vcpus are mapped onto virtual
213 threads in one or more virtual CPU cores. (This is because the
214 hardware requires all the hardware threads in a CPU core to be in the
215 same partition.) The KVM_CAP_PPC_SMT capability indicates the number
216 of vcpus per virtual core (vcore). The vcore id is obtained by
217 dividing the vcpu id by the number of vcpus per vcore. The vcpus in a
218 given vcore will always be in the same physical core as each other
219 (though that might be a different physical core from time to time).
220 Userspace can control the threading (SMT) mode of the guest by its
221 allocation of vcpu ids. For example, if userspace wants
222 single-threaded guest vcpus, it should make all vcpu ids be a multiple
223 of the number of vcpus per vcore.
225 For virtual cpus that have been created with S390 user controlled virtual
226 machines, the resulting vcpu fd can be memory mapped at page offset
227 KVM_S390_SIE_PAGE_OFFSET in order to obtain a memory map of the virtual
228 cpu's hardware control block.
231 4.8 KVM_GET_DIRTY_LOG (vm ioctl)
236 Parameters: struct kvm_dirty_log (in/out)
237 Returns: 0 on success, -1 on error
239 /* for KVM_GET_DIRTY_LOG */
240 struct kvm_dirty_log {
244 void __user *dirty_bitmap; /* one bit per page */
249 Given a memory slot, return a bitmap containing any pages dirtied
250 since the last call to this ioctl. Bit 0 is the first page in the
251 memory slot. Ensure the entire structure is cleared to avoid padding
255 4.9 KVM_SET_MEMORY_ALIAS
260 Parameters: struct kvm_memory_alias (in)
261 Returns: 0 (success), -1 (error)
263 This ioctl is obsolete and has been removed.
272 Returns: 0 on success, -1 on error
274 EINTR: an unmasked signal is pending
276 This ioctl is used to run a guest virtual cpu. While there are no
277 explicit parameters, there is an implicit parameter block that can be
278 obtained by mmap()ing the vcpu fd at offset 0, with the size given by
279 KVM_GET_VCPU_MMAP_SIZE. The parameter block is formatted as a 'struct
280 kvm_run' (see below).
286 Architectures: all except ARM, arm64
288 Parameters: struct kvm_regs (out)
289 Returns: 0 on success, -1 on error
291 Reads the general purpose registers from the vcpu.
295 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
296 __u64 rax, rbx, rcx, rdx;
297 __u64 rsi, rdi, rsp, rbp;
298 __u64 r8, r9, r10, r11;
299 __u64 r12, r13, r14, r15;
307 Architectures: all except ARM, arm64
309 Parameters: struct kvm_regs (in)
310 Returns: 0 on success, -1 on error
312 Writes the general purpose registers into the vcpu.
314 See KVM_GET_REGS for the data structure.
320 Architectures: x86, ppc
322 Parameters: struct kvm_sregs (out)
323 Returns: 0 on success, -1 on error
325 Reads special registers from the vcpu.
329 struct kvm_segment cs, ds, es, fs, gs, ss;
330 struct kvm_segment tr, ldt;
331 struct kvm_dtable gdt, idt;
332 __u64 cr0, cr2, cr3, cr4, cr8;
335 __u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64];
338 /* ppc -- see arch/powerpc/include/uapi/asm/kvm.h */
340 interrupt_bitmap is a bitmap of pending external interrupts. At most
341 one bit may be set. This interrupt has been acknowledged by the APIC
342 but not yet injected into the cpu core.
348 Architectures: x86, ppc
350 Parameters: struct kvm_sregs (in)
351 Returns: 0 on success, -1 on error
353 Writes special registers into the vcpu. See KVM_GET_SREGS for the
362 Parameters: struct kvm_translation (in/out)
363 Returns: 0 on success, -1 on error
365 Translates a virtual address according to the vcpu's current address
368 struct kvm_translation {
370 __u64 linear_address;
373 __u64 physical_address;
384 Architectures: x86, ppc
386 Parameters: struct kvm_interrupt (in)
387 Returns: 0 on success, -1 on error
389 Queues a hardware interrupt vector to be injected. This is only
390 useful if in-kernel local APIC or equivalent is not used.
392 /* for KVM_INTERRUPT */
393 struct kvm_interrupt {
400 Note 'irq' is an interrupt vector, not an interrupt pin or line.
404 Queues an external interrupt to be injected. This ioctl is overleaded
405 with 3 different irq values:
409 This injects an edge type external interrupt into the guest once it's ready
410 to receive interrupts. When injected, the interrupt is done.
412 b) KVM_INTERRUPT_UNSET
414 This unsets any pending interrupt.
416 Only available with KVM_CAP_PPC_UNSET_IRQ.
418 c) KVM_INTERRUPT_SET_LEVEL
420 This injects a level type external interrupt into the guest context. The
421 interrupt stays pending until a specific ioctl with KVM_INTERRUPT_UNSET
424 Only available with KVM_CAP_PPC_IRQ_LEVEL.
426 Note that any value for 'irq' other than the ones stated above is invalid
427 and incurs unexpected behavior.
438 Support for this has been removed. Use KVM_SET_GUEST_DEBUG instead.
446 Parameters: struct kvm_msrs (in/out)
447 Returns: 0 on success, -1 on error
449 Reads model-specific registers from the vcpu. Supported msr indices can
450 be obtained using KVM_GET_MSR_INDEX_LIST.
453 __u32 nmsrs; /* number of msrs in entries */
456 struct kvm_msr_entry entries[0];
459 struct kvm_msr_entry {
465 Application code should set the 'nmsrs' member (which indicates the
466 size of the entries array) and the 'index' member of each array entry.
467 kvm will fill in the 'data' member.
475 Parameters: struct kvm_msrs (in)
476 Returns: 0 on success, -1 on error
478 Writes model-specific registers to the vcpu. See KVM_GET_MSRS for the
481 Application code should set the 'nmsrs' member (which indicates the
482 size of the entries array), and the 'index' and 'data' members of each
491 Parameters: struct kvm_cpuid (in)
492 Returns: 0 on success, -1 on error
494 Defines the vcpu responses to the cpuid instruction. Applications
495 should use the KVM_SET_CPUID2 ioctl if available.
498 struct kvm_cpuid_entry {
507 /* for KVM_SET_CPUID */
511 struct kvm_cpuid_entry entries[0];
515 4.21 KVM_SET_SIGNAL_MASK
520 Parameters: struct kvm_signal_mask (in)
521 Returns: 0 on success, -1 on error
523 Defines which signals are blocked during execution of KVM_RUN. This
524 signal mask temporarily overrides the threads signal mask. Any
525 unblocked signal received (except SIGKILL and SIGSTOP, which retain
526 their traditional behaviour) will cause KVM_RUN to return with -EINTR.
528 Note the signal will only be delivered if not blocked by the original
531 /* for KVM_SET_SIGNAL_MASK */
532 struct kvm_signal_mask {
543 Parameters: struct kvm_fpu (out)
544 Returns: 0 on success, -1 on error
546 Reads the floating point state from the vcpu.
548 /* for KVM_GET_FPU and KVM_SET_FPU */
553 __u8 ftwx; /* in fxsave format */
569 Parameters: struct kvm_fpu (in)
570 Returns: 0 on success, -1 on error
572 Writes the floating point state to the vcpu.
574 /* for KVM_GET_FPU and KVM_SET_FPU */
579 __u8 ftwx; /* in fxsave format */
590 4.24 KVM_CREATE_IRQCHIP
592 Capability: KVM_CAP_IRQCHIP
593 Architectures: x86, ia64, ARM, arm64
596 Returns: 0 on success, -1 on error
598 Creates an interrupt controller model in the kernel. On x86, creates a virtual
599 ioapic, a virtual PIC (two PICs, nested), and sets up future vcpus to have a
600 local APIC. IRQ routing for GSIs 0-15 is set to both PIC and IOAPIC; GSI 16-23
601 only go to the IOAPIC. On ia64, a IOSAPIC is created. On ARM/arm64, a GIC is
607 Capability: KVM_CAP_IRQCHIP
608 Architectures: x86, ia64, arm, arm64
610 Parameters: struct kvm_irq_level
611 Returns: 0 on success, -1 on error
613 Sets the level of a GSI input to the interrupt controller model in the kernel.
614 On some architectures it is required that an interrupt controller model has
615 been previously created with KVM_CREATE_IRQCHIP. Note that edge-triggered
616 interrupts require the level to be set to 1 and then back to 0.
618 ARM/arm64 can signal an interrupt either at the CPU level, or at the
619 in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to
620 use PPIs designated for specific cpus. The irq field is interpreted
623 Â bits: | 31 ... 24 | 23 ... 16 | 15 ... 0 |
624 field: | irq_type | vcpu_index | irq_id |
626 The irq_type field has the following values:
627 - irq_type[0]: out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ
628 - irq_type[1]: in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.)
629 (the vcpu_index field is ignored)
630 - irq_type[2]: in-kernel GIC: PPI, irq_id between 16 and 31 (incl.)
632 (The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs)
634 In both cases, level is used to raise/lower the line.
636 struct kvm_irq_level {
639 __s32 status; /* not used for KVM_IRQ_LEVEL */
641 __u32 level; /* 0 or 1 */
647 Capability: KVM_CAP_IRQCHIP
648 Architectures: x86, ia64
650 Parameters: struct kvm_irqchip (in/out)
651 Returns: 0 on success, -1 on error
653 Reads the state of a kernel interrupt controller created with
654 KVM_CREATE_IRQCHIP into a buffer provided by the caller.
657 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
660 char dummy[512]; /* reserving space */
661 struct kvm_pic_state pic;
662 struct kvm_ioapic_state ioapic;
669 Capability: KVM_CAP_IRQCHIP
670 Architectures: x86, ia64
672 Parameters: struct kvm_irqchip (in)
673 Returns: 0 on success, -1 on error
675 Sets the state of a kernel interrupt controller created with
676 KVM_CREATE_IRQCHIP from a buffer provided by the caller.
679 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
682 char dummy[512]; /* reserving space */
683 struct kvm_pic_state pic;
684 struct kvm_ioapic_state ioapic;
689 4.28 KVM_XEN_HVM_CONFIG
691 Capability: KVM_CAP_XEN_HVM
694 Parameters: struct kvm_xen_hvm_config (in)
695 Returns: 0 on success, -1 on error
697 Sets the MSR that the Xen HVM guest uses to initialize its hypercall
698 page, and provides the starting address and size of the hypercall
699 blobs in userspace. When the guest writes the MSR, kvm copies one
700 page of a blob (32- or 64-bit, depending on the vcpu mode) to guest
703 struct kvm_xen_hvm_config {
716 Capability: KVM_CAP_ADJUST_CLOCK
719 Parameters: struct kvm_clock_data (out)
720 Returns: 0 on success, -1 on error
722 Gets the current timestamp of kvmclock as seen by the current guest. In
723 conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios
726 struct kvm_clock_data {
727 __u64 clock; /* kvmclock current value */
735 Capability: KVM_CAP_ADJUST_CLOCK
738 Parameters: struct kvm_clock_data (in)
739 Returns: 0 on success, -1 on error
741 Sets the current timestamp of kvmclock to the value specified in its parameter.
742 In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios
745 struct kvm_clock_data {
746 __u64 clock; /* kvmclock current value */
752 4.31 KVM_GET_VCPU_EVENTS
754 Capability: KVM_CAP_VCPU_EVENTS
755 Extended by: KVM_CAP_INTR_SHADOW
758 Parameters: struct kvm_vcpu_event (out)
759 Returns: 0 on success, -1 on error
761 Gets currently pending exceptions, interrupts, and NMIs as well as related
764 struct kvm_vcpu_events {
788 KVM_VCPUEVENT_VALID_SHADOW may be set in the flags field to signal that
789 interrupt.shadow contains a valid state. Otherwise, this field is undefined.
792 4.32 KVM_SET_VCPU_EVENTS
794 Capability: KVM_CAP_VCPU_EVENTS
795 Extended by: KVM_CAP_INTR_SHADOW
798 Parameters: struct kvm_vcpu_event (in)
799 Returns: 0 on success, -1 on error
801 Set pending exceptions, interrupts, and NMIs as well as related states of the
804 See KVM_GET_VCPU_EVENTS for the data structure.
806 Fields that may be modified asynchronously by running VCPUs can be excluded
807 from the update. These fields are nmi.pending and sipi_vector. Keep the
808 corresponding bits in the flags field cleared to suppress overwriting the
809 current in-kernel state. The bits are:
811 KVM_VCPUEVENT_VALID_NMI_PENDING - transfer nmi.pending to the kernel
812 KVM_VCPUEVENT_VALID_SIPI_VECTOR - transfer sipi_vector
814 If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in
815 the flags field to signal that interrupt.shadow contains a valid state and
816 shall be written into the VCPU.
819 4.33 KVM_GET_DEBUGREGS
821 Capability: KVM_CAP_DEBUGREGS
824 Parameters: struct kvm_debugregs (out)
825 Returns: 0 on success, -1 on error
827 Reads debug registers from the vcpu.
829 struct kvm_debugregs {
838 4.34 KVM_SET_DEBUGREGS
840 Capability: KVM_CAP_DEBUGREGS
843 Parameters: struct kvm_debugregs (in)
844 Returns: 0 on success, -1 on error
846 Writes debug registers into the vcpu.
848 See KVM_GET_DEBUGREGS for the data structure. The flags field is unused
849 yet and must be cleared on entry.
852 4.35 KVM_SET_USER_MEMORY_REGION
854 Capability: KVM_CAP_USER_MEM
857 Parameters: struct kvm_userspace_memory_region (in)
858 Returns: 0 on success, -1 on error
860 struct kvm_userspace_memory_region {
863 __u64 guest_phys_addr;
864 __u64 memory_size; /* bytes */
865 __u64 userspace_addr; /* start of the userspace allocated memory */
868 /* for kvm_memory_region::flags */
869 #define KVM_MEM_LOG_DIRTY_PAGES (1UL << 0)
870 #define KVM_MEM_READONLY (1UL << 1)
872 This ioctl allows the user to create or modify a guest physical memory
873 slot. When changing an existing slot, it may be moved in the guest
874 physical memory space, or its flags may be modified. It may not be
875 resized. Slots may not overlap in guest physical address space.
877 Memory for the region is taken starting at the address denoted by the
878 field userspace_addr, which must point at user addressable memory for
879 the entire memory slot size. Any object may back this memory, including
880 anonymous memory, ordinary files, and hugetlbfs.
882 It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr
883 be identical. This allows large pages in the guest to be backed by large
886 The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and
887 KVM_MEM_READONLY. The former can be set to instruct KVM to keep track of
888 writes to memory within the slot. See KVM_GET_DIRTY_LOG ioctl to know how to
889 use it. The latter can be set, if KVM_CAP_READONLY_MEM capability allows it,
890 to make a new slot read-only. In this case, writes to this memory will be
891 posted to userspace as KVM_EXIT_MMIO exits.
893 When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of
894 the memory region are automatically reflected into the guest. For example, an
895 mmap() that affects the region will be made visible immediately. Another
896 example is madvise(MADV_DROP).
898 It is recommended to use this API instead of the KVM_SET_MEMORY_REGION ioctl.
899 The KVM_SET_MEMORY_REGION does not allow fine grained control over memory
900 allocation and is deprecated.
903 4.36 KVM_SET_TSS_ADDR
905 Capability: KVM_CAP_SET_TSS_ADDR
908 Parameters: unsigned long tss_address (in)
909 Returns: 0 on success, -1 on error
911 This ioctl defines the physical address of a three-page region in the guest
912 physical address space. The region must be within the first 4GB of the
913 guest physical address space and must not conflict with any memory slot
914 or any mmio address. The guest may malfunction if it accesses this memory
917 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
918 because of a quirk in the virtualization implementation (see the internals
919 documentation when it pops into existence).
924 Capability: KVM_CAP_ENABLE_CAP
925 Architectures: ppc, s390
927 Parameters: struct kvm_enable_cap (in)
928 Returns: 0 on success; -1 on error
930 +Not all extensions are enabled by default. Using this ioctl the application
931 can enable an extension, making it available to the guest.
933 On systems that do not support this ioctl, it always fails. On systems that
934 do support it, it only works for extensions that are supported for enablement.
936 To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should
939 struct kvm_enable_cap {
943 The capability that is supposed to get enabled.
947 A bitfield indicating future enhancements. Has to be 0 for now.
951 Arguments for enabling a feature. If a feature needs initial values to
952 function properly, this is the place to put them.
958 4.38 KVM_GET_MP_STATE
960 Capability: KVM_CAP_MP_STATE
961 Architectures: x86, ia64
963 Parameters: struct kvm_mp_state (out)
964 Returns: 0 on success; -1 on error
966 struct kvm_mp_state {
970 Returns the vcpu's current "multiprocessing state" (though also valid on
971 uniprocessor guests).
975 - KVM_MP_STATE_RUNNABLE: the vcpu is currently running [x86, ia64]
976 - KVM_MP_STATE_UNINITIALIZED: the vcpu is an application processor (AP)
977 which has not yet received an INIT signal [x86,
979 - KVM_MP_STATE_INIT_RECEIVED: the vcpu has received an INIT signal, and is
980 now ready for a SIPI [x86, ia64]
981 - KVM_MP_STATE_HALTED: the vcpu has executed a HLT instruction and
982 is waiting for an interrupt [x86, ia64]
983 - KVM_MP_STATE_SIPI_RECEIVED: the vcpu has just received a SIPI (vector
984 accessible via KVM_GET_VCPU_EVENTS) [x86, ia64]
986 On x86 and ia64, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
987 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
991 4.39 KVM_SET_MP_STATE
993 Capability: KVM_CAP_MP_STATE
994 Architectures: x86, ia64
996 Parameters: struct kvm_mp_state (in)
997 Returns: 0 on success; -1 on error
999 Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for
1002 On x86 and ia64, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1003 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1004 these architectures.
1007 4.40 KVM_SET_IDENTITY_MAP_ADDR
1009 Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR
1012 Parameters: unsigned long identity (in)
1013 Returns: 0 on success, -1 on error
1015 This ioctl defines the physical address of a one-page region in the guest
1016 physical address space. The region must be within the first 4GB of the
1017 guest physical address space and must not conflict with any memory slot
1018 or any mmio address. The guest may malfunction if it accesses this memory
1021 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1022 because of a quirk in the virtualization implementation (see the internals
1023 documentation when it pops into existence).
1026 4.41 KVM_SET_BOOT_CPU_ID
1028 Capability: KVM_CAP_SET_BOOT_CPU_ID
1029 Architectures: x86, ia64
1031 Parameters: unsigned long vcpu_id
1032 Returns: 0 on success, -1 on error
1034 Define which vcpu is the Bootstrap Processor (BSP). Values are the same
1035 as the vcpu id in KVM_CREATE_VCPU. If this ioctl is not called, the default
1041 Capability: KVM_CAP_XSAVE
1044 Parameters: struct kvm_xsave (out)
1045 Returns: 0 on success, -1 on error
1051 This ioctl would copy current vcpu's xsave struct to the userspace.
1056 Capability: KVM_CAP_XSAVE
1059 Parameters: struct kvm_xsave (in)
1060 Returns: 0 on success, -1 on error
1066 This ioctl would copy userspace's xsave struct to the kernel.
1071 Capability: KVM_CAP_XCRS
1074 Parameters: struct kvm_xcrs (out)
1075 Returns: 0 on success, -1 on error
1086 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1090 This ioctl would copy current vcpu's xcrs to the userspace.
1095 Capability: KVM_CAP_XCRS
1098 Parameters: struct kvm_xcrs (in)
1099 Returns: 0 on success, -1 on error
1110 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1114 This ioctl would set vcpu's xcr to the value userspace specified.
1117 4.46 KVM_GET_SUPPORTED_CPUID
1119 Capability: KVM_CAP_EXT_CPUID
1122 Parameters: struct kvm_cpuid2 (in/out)
1123 Returns: 0 on success, -1 on error
1128 struct kvm_cpuid_entry2 entries[0];
1131 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
1132 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1)
1133 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2)
1135 struct kvm_cpuid_entry2 {
1146 This ioctl returns x86 cpuid features which are supported by both the hardware
1147 and kvm. Userspace can use the information returned by this ioctl to
1148 construct cpuid information (for KVM_SET_CPUID2) that is consistent with
1149 hardware, kernel, and userspace capabilities, and with user requirements (for
1150 example, the user may wish to constrain cpuid to emulate older hardware,
1151 or for feature consistency across a cluster).
1153 Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure
1154 with the 'nent' field indicating the number of entries in the variable-size
1155 array 'entries'. If the number of entries is too low to describe the cpu
1156 capabilities, an error (E2BIG) is returned. If the number is too high,
1157 the 'nent' field is adjusted and an error (ENOMEM) is returned. If the
1158 number is just right, the 'nent' field is adjusted to the number of valid
1159 entries in the 'entries' array, which is then filled.
1161 The entries returned are the host cpuid as returned by the cpuid instruction,
1162 with unknown or unsupported features masked out. Some features (for example,
1163 x2apic), may not be present in the host cpu, but are exposed by kvm if it can
1164 emulate them efficiently. The fields in each entry are defined as follows:
1166 function: the eax value used to obtain the entry
1167 index: the ecx value used to obtain the entry (for entries that are
1169 flags: an OR of zero or more of the following:
1170 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
1171 if the index field is valid
1172 KVM_CPUID_FLAG_STATEFUL_FUNC:
1173 if cpuid for this function returns different values for successive
1174 invocations; there will be several entries with the same function,
1175 all with this flag set
1176 KVM_CPUID_FLAG_STATE_READ_NEXT:
1177 for KVM_CPUID_FLAG_STATEFUL_FUNC entries, set if this entry is
1178 the first entry to be read by a cpu
1179 eax, ebx, ecx, edx: the values returned by the cpuid instruction for
1180 this function/index combination
1182 The TSC deadline timer feature (CPUID leaf 1, ecx[24]) is always returned
1183 as false, since the feature depends on KVM_CREATE_IRQCHIP for local APIC
1184 support. Instead it is reported via
1186 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER)
1188 if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the
1189 feature in userspace, then you can enable the feature for KVM_SET_CPUID2.
1192 4.47 KVM_PPC_GET_PVINFO
1194 Capability: KVM_CAP_PPC_GET_PVINFO
1197 Parameters: struct kvm_ppc_pvinfo (out)
1198 Returns: 0 on success, !0 on error
1200 struct kvm_ppc_pvinfo {
1206 This ioctl fetches PV specific information that need to be passed to the guest
1207 using the device tree or other means from vm context.
1209 The hcall array defines 4 instructions that make up a hypercall.
1211 If any additional field gets added to this structure later on, a bit for that
1212 additional piece of information will be set in the flags bitmap.
1214 The flags bitmap is defined as:
1216 /* the host supports the ePAPR idle hcall
1217 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1<<0)
1219 4.48 KVM_ASSIGN_PCI_DEVICE
1221 Capability: KVM_CAP_DEVICE_ASSIGNMENT
1222 Architectures: x86 ia64
1224 Parameters: struct kvm_assigned_pci_dev (in)
1225 Returns: 0 on success, -1 on error
1227 Assigns a host PCI device to the VM.
1229 struct kvm_assigned_pci_dev {
1230 __u32 assigned_dev_id;
1240 The PCI device is specified by the triple segnr, busnr, and devfn.
1241 Identification in succeeding service requests is done via assigned_dev_id. The
1242 following flags are specified:
1244 /* Depends on KVM_CAP_IOMMU */
1245 #define KVM_DEV_ASSIGN_ENABLE_IOMMU (1 << 0)
1246 /* The following two depend on KVM_CAP_PCI_2_3 */
1247 #define KVM_DEV_ASSIGN_PCI_2_3 (1 << 1)
1248 #define KVM_DEV_ASSIGN_MASK_INTX (1 << 2)
1250 If KVM_DEV_ASSIGN_PCI_2_3 is set, the kernel will manage legacy INTx interrupts
1251 via the PCI-2.3-compliant device-level mask, thus enable IRQ sharing with other
1252 assigned devices or host devices. KVM_DEV_ASSIGN_MASK_INTX specifies the
1253 guest's view on the INTx mask, see KVM_ASSIGN_SET_INTX_MASK for details.
1255 The KVM_DEV_ASSIGN_ENABLE_IOMMU flag is a mandatory option to ensure
1256 isolation of the device. Usages not specifying this flag are deprecated.
1258 Only PCI header type 0 devices with PCI BAR resources are supported by
1259 device assignment. The user requesting this ioctl must have read/write
1260 access to the PCI sysfs resource files associated with the device.
1263 4.49 KVM_DEASSIGN_PCI_DEVICE
1265 Capability: KVM_CAP_DEVICE_DEASSIGNMENT
1266 Architectures: x86 ia64
1268 Parameters: struct kvm_assigned_pci_dev (in)
1269 Returns: 0 on success, -1 on error
1271 Ends PCI device assignment, releasing all associated resources.
1273 See KVM_CAP_DEVICE_ASSIGNMENT for the data structure. Only assigned_dev_id is
1274 used in kvm_assigned_pci_dev to identify the device.
1277 4.50 KVM_ASSIGN_DEV_IRQ
1279 Capability: KVM_CAP_ASSIGN_DEV_IRQ
1280 Architectures: x86 ia64
1282 Parameters: struct kvm_assigned_irq (in)
1283 Returns: 0 on success, -1 on error
1285 Assigns an IRQ to a passed-through device.
1287 struct kvm_assigned_irq {
1288 __u32 assigned_dev_id;
1289 __u32 host_irq; /* ignored (legacy field) */
1297 The following flags are defined:
1299 #define KVM_DEV_IRQ_HOST_INTX (1 << 0)
1300 #define KVM_DEV_IRQ_HOST_MSI (1 << 1)
1301 #define KVM_DEV_IRQ_HOST_MSIX (1 << 2)
1303 #define KVM_DEV_IRQ_GUEST_INTX (1 << 8)
1304 #define KVM_DEV_IRQ_GUEST_MSI (1 << 9)
1305 #define KVM_DEV_IRQ_GUEST_MSIX (1 << 10)
1307 It is not valid to specify multiple types per host or guest IRQ. However, the
1308 IRQ type of host and guest can differ or can even be null.
1311 4.51 KVM_DEASSIGN_DEV_IRQ
1313 Capability: KVM_CAP_ASSIGN_DEV_IRQ
1314 Architectures: x86 ia64
1316 Parameters: struct kvm_assigned_irq (in)
1317 Returns: 0 on success, -1 on error
1319 Ends an IRQ assignment to a passed-through device.
1321 See KVM_ASSIGN_DEV_IRQ for the data structure. The target device is specified
1322 by assigned_dev_id, flags must correspond to the IRQ type specified on
1323 KVM_ASSIGN_DEV_IRQ. Partial deassignment of host or guest IRQ is allowed.
1326 4.52 KVM_SET_GSI_ROUTING
1328 Capability: KVM_CAP_IRQ_ROUTING
1329 Architectures: x86 ia64
1331 Parameters: struct kvm_irq_routing (in)
1332 Returns: 0 on success, -1 on error
1334 Sets the GSI routing table entries, overwriting any previously set entries.
1336 struct kvm_irq_routing {
1339 struct kvm_irq_routing_entry entries[0];
1342 No flags are specified so far, the corresponding field must be set to zero.
1344 struct kvm_irq_routing_entry {
1350 struct kvm_irq_routing_irqchip irqchip;
1351 struct kvm_irq_routing_msi msi;
1356 /* gsi routing entry types */
1357 #define KVM_IRQ_ROUTING_IRQCHIP 1
1358 #define KVM_IRQ_ROUTING_MSI 2
1360 No flags are specified so far, the corresponding field must be set to zero.
1362 struct kvm_irq_routing_irqchip {
1367 struct kvm_irq_routing_msi {
1375 4.53 KVM_ASSIGN_SET_MSIX_NR
1377 Capability: KVM_CAP_DEVICE_MSIX
1378 Architectures: x86 ia64
1380 Parameters: struct kvm_assigned_msix_nr (in)
1381 Returns: 0 on success, -1 on error
1383 Set the number of MSI-X interrupts for an assigned device. The number is
1384 reset again by terminating the MSI-X assignment of the device via
1385 KVM_DEASSIGN_DEV_IRQ. Calling this service more than once at any earlier
1388 struct kvm_assigned_msix_nr {
1389 __u32 assigned_dev_id;
1394 #define KVM_MAX_MSIX_PER_DEV 256
1397 4.54 KVM_ASSIGN_SET_MSIX_ENTRY
1399 Capability: KVM_CAP_DEVICE_MSIX
1400 Architectures: x86 ia64
1402 Parameters: struct kvm_assigned_msix_entry (in)
1403 Returns: 0 on success, -1 on error
1405 Specifies the routing of an MSI-X assigned device interrupt to a GSI. Setting
1406 the GSI vector to zero means disabling the interrupt.
1408 struct kvm_assigned_msix_entry {
1409 __u32 assigned_dev_id;
1411 __u16 entry; /* The index of entry in the MSI-X table */
1416 4.55 KVM_SET_TSC_KHZ
1418 Capability: KVM_CAP_TSC_CONTROL
1421 Parameters: virtual tsc_khz
1422 Returns: 0 on success, -1 on error
1424 Specifies the tsc frequency for the virtual machine. The unit of the
1428 4.56 KVM_GET_TSC_KHZ
1430 Capability: KVM_CAP_GET_TSC_KHZ
1434 Returns: virtual tsc-khz on success, negative value on error
1436 Returns the tsc frequency of the guest. The unit of the return value is
1437 KHz. If the host has unstable tsc this ioctl returns -EIO instead as an
1443 Capability: KVM_CAP_IRQCHIP
1446 Parameters: struct kvm_lapic_state (out)
1447 Returns: 0 on success, -1 on error
1449 #define KVM_APIC_REG_SIZE 0x400
1450 struct kvm_lapic_state {
1451 char regs[KVM_APIC_REG_SIZE];
1454 Reads the Local APIC registers and copies them into the input argument. The
1455 data format and layout are the same as documented in the architecture manual.
1460 Capability: KVM_CAP_IRQCHIP
1463 Parameters: struct kvm_lapic_state (in)
1464 Returns: 0 on success, -1 on error
1466 #define KVM_APIC_REG_SIZE 0x400
1467 struct kvm_lapic_state {
1468 char regs[KVM_APIC_REG_SIZE];
1471 Copies the input argument into the the Local APIC registers. The data format
1472 and layout are the same as documented in the architecture manual.
1477 Capability: KVM_CAP_IOEVENTFD
1480 Parameters: struct kvm_ioeventfd (in)
1481 Returns: 0 on success, !0 on error
1483 This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address
1484 within the guest. A guest write in the registered address will signal the
1485 provided event instead of triggering an exit.
1487 struct kvm_ioeventfd {
1489 __u64 addr; /* legal pio/mmio address */
1490 __u32 len; /* 1, 2, 4, or 8 bytes */
1496 For the special case of virtio-ccw devices on s390, the ioevent is matched
1497 to a subchannel/virtqueue tuple instead.
1499 The following flags are defined:
1501 #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch)
1502 #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio)
1503 #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign)
1504 #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \
1505 (1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify)
1507 If datamatch flag is set, the event will be signaled only if the written value
1508 to the registered address is equal to datamatch in struct kvm_ioeventfd.
1510 For virtio-ccw devices, addr contains the subchannel id and datamatch the
1516 Capability: KVM_CAP_SW_TLB
1519 Parameters: struct kvm_dirty_tlb (in)
1520 Returns: 0 on success, -1 on error
1522 struct kvm_dirty_tlb {
1527 This must be called whenever userspace has changed an entry in the shared
1528 TLB, prior to calling KVM_RUN on the associated vcpu.
1530 The "bitmap" field is the userspace address of an array. This array
1531 consists of a number of bits, equal to the total number of TLB entries as
1532 determined by the last successful call to KVM_CONFIG_TLB, rounded up to the
1533 nearest multiple of 64.
1535 Each bit corresponds to one TLB entry, ordered the same as in the shared TLB
1538 The array is little-endian: the bit 0 is the least significant bit of the
1539 first byte, bit 8 is the least significant bit of the second byte, etc.
1540 This avoids any complications with differing word sizes.
1542 The "num_dirty" field is a performance hint for KVM to determine whether it
1543 should skip processing the bitmap and just invalidate everything. It must
1544 be set to the number of set bits in the bitmap.
1547 4.61 KVM_ASSIGN_SET_INTX_MASK
1549 Capability: KVM_CAP_PCI_2_3
1552 Parameters: struct kvm_assigned_pci_dev (in)
1553 Returns: 0 on success, -1 on error
1555 Allows userspace to mask PCI INTx interrupts from the assigned device. The
1556 kernel will not deliver INTx interrupts to the guest between setting and
1557 clearing of KVM_ASSIGN_SET_INTX_MASK via this interface. This enables use of
1558 and emulation of PCI 2.3 INTx disable command register behavior.
1560 This may be used for both PCI 2.3 devices supporting INTx disable natively and
1561 older devices lacking this support. Userspace is responsible for emulating the
1562 read value of the INTx disable bit in the guest visible PCI command register.
1563 When modifying the INTx disable state, userspace should precede updating the
1564 physical device command register by calling this ioctl to inform the kernel of
1565 the new intended INTx mask state.
1567 Note that the kernel uses the device INTx disable bit to internally manage the
1568 device interrupt state for PCI 2.3 devices. Reads of this register may
1569 therefore not match the expected value. Writes should always use the guest
1570 intended INTx disable value rather than attempting to read-copy-update the
1571 current physical device state. Races between user and kernel updates to the
1572 INTx disable bit are handled lazily in the kernel. It's possible the device
1573 may generate unintended interrupts, but they will not be injected into the
1576 See KVM_ASSIGN_DEV_IRQ for the data structure. The target device is specified
1577 by assigned_dev_id. In the flags field, only KVM_DEV_ASSIGN_MASK_INTX is
1581 4.62 KVM_CREATE_SPAPR_TCE
1583 Capability: KVM_CAP_SPAPR_TCE
1584 Architectures: powerpc
1586 Parameters: struct kvm_create_spapr_tce (in)
1587 Returns: file descriptor for manipulating the created TCE table
1589 This creates a virtual TCE (translation control entry) table, which
1590 is an IOMMU for PAPR-style virtual I/O. It is used to translate
1591 logical addresses used in virtual I/O into guest physical addresses,
1592 and provides a scatter/gather capability for PAPR virtual I/O.
1594 /* for KVM_CAP_SPAPR_TCE */
1595 struct kvm_create_spapr_tce {
1600 The liobn field gives the logical IO bus number for which to create a
1601 TCE table. The window_size field specifies the size of the DMA window
1602 which this TCE table will translate - the table will contain one 64
1603 bit TCE entry for every 4kiB of the DMA window.
1605 When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE
1606 table has been created using this ioctl(), the kernel will handle it
1607 in real mode, updating the TCE table. H_PUT_TCE calls for other
1608 liobns will cause a vm exit and must be handled by userspace.
1610 The return value is a file descriptor which can be passed to mmap(2)
1611 to map the created TCE table into userspace. This lets userspace read
1612 the entries written by kernel-handled H_PUT_TCE calls, and also lets
1613 userspace update the TCE table directly which is useful in some
1617 4.63 KVM_ALLOCATE_RMA
1619 Capability: KVM_CAP_PPC_RMA
1620 Architectures: powerpc
1622 Parameters: struct kvm_allocate_rma (out)
1623 Returns: file descriptor for mapping the allocated RMA
1625 This allocates a Real Mode Area (RMA) from the pool allocated at boot
1626 time by the kernel. An RMA is a physically-contiguous, aligned region
1627 of memory used on older POWER processors to provide the memory which
1628 will be accessed by real-mode (MMU off) accesses in a KVM guest.
1629 POWER processors support a set of sizes for the RMA that usually
1630 includes 64MB, 128MB, 256MB and some larger powers of two.
1632 /* for KVM_ALLOCATE_RMA */
1633 struct kvm_allocate_rma {
1637 The return value is a file descriptor which can be passed to mmap(2)
1638 to map the allocated RMA into userspace. The mapped area can then be
1639 passed to the KVM_SET_USER_MEMORY_REGION ioctl to establish it as the
1640 RMA for a virtual machine. The size of the RMA in bytes (which is
1641 fixed at host kernel boot time) is returned in the rma_size field of
1642 the argument structure.
1644 The KVM_CAP_PPC_RMA capability is 1 or 2 if the KVM_ALLOCATE_RMA ioctl
1645 is supported; 2 if the processor requires all virtual machines to have
1646 an RMA, or 1 if the processor can use an RMA but doesn't require it,
1647 because it supports the Virtual RMA (VRMA) facility.
1652 Capability: KVM_CAP_USER_NMI
1656 Returns: 0 on success, -1 on error
1658 Queues an NMI on the thread's vcpu. Note this is well defined only
1659 when KVM_CREATE_IRQCHIP has not been called, since this is an interface
1660 between the virtual cpu core and virtual local APIC. After KVM_CREATE_IRQCHIP
1661 has been called, this interface is completely emulated within the kernel.
1663 To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the
1664 following algorithm:
1667 - read the local APIC's state (KVM_GET_LAPIC)
1668 - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1)
1669 - if so, issue KVM_NMI
1672 Some guests configure the LINT1 NMI input to cause a panic, aiding in
1676 4.65 KVM_S390_UCAS_MAP
1678 Capability: KVM_CAP_S390_UCONTROL
1681 Parameters: struct kvm_s390_ucas_mapping (in)
1682 Returns: 0 in case of success
1684 The parameter is defined like this:
1685 struct kvm_s390_ucas_mapping {
1691 This ioctl maps the memory at "user_addr" with the length "length" to
1692 the vcpu's address space starting at "vcpu_addr". All parameters need to
1693 be alligned by 1 megabyte.
1696 4.66 KVM_S390_UCAS_UNMAP
1698 Capability: KVM_CAP_S390_UCONTROL
1701 Parameters: struct kvm_s390_ucas_mapping (in)
1702 Returns: 0 in case of success
1704 The parameter is defined like this:
1705 struct kvm_s390_ucas_mapping {
1711 This ioctl unmaps the memory in the vcpu's address space starting at
1712 "vcpu_addr" with the length "length". The field "user_addr" is ignored.
1713 All parameters need to be alligned by 1 megabyte.
1716 4.67 KVM_S390_VCPU_FAULT
1718 Capability: KVM_CAP_S390_UCONTROL
1721 Parameters: vcpu absolute address (in)
1722 Returns: 0 in case of success
1724 This call creates a page table entry on the virtual cpu's address space
1725 (for user controlled virtual machines) or the virtual machine's address
1726 space (for regular virtual machines). This only works for minor faults,
1727 thus it's recommended to access subject memory page via the user page
1728 table upfront. This is useful to handle validity intercepts for user
1729 controlled virtual machines to fault in the virtual cpu's lowcore pages
1730 prior to calling the KVM_RUN ioctl.
1733 4.68 KVM_SET_ONE_REG
1735 Capability: KVM_CAP_ONE_REG
1738 Parameters: struct kvm_one_reg (in)
1739 Returns: 0 on success, negative value on failure
1741 struct kvm_one_reg {
1746 Using this ioctl, a single vcpu register can be set to a specific value
1747 defined by user space with the passed in struct kvm_one_reg, where id
1748 refers to the register identifier as described below and addr is a pointer
1749 to a variable with the respective size. There can be architecture agnostic
1750 and architecture specific registers. Each have their own range of operation
1751 and their own constants and width. To keep track of the implemented
1752 registers, find a list below:
1754 Arch | Register | Width (bits)
1756 PPC | KVM_REG_PPC_HIOR | 64
1757 PPC | KVM_REG_PPC_IAC1 | 64
1758 PPC | KVM_REG_PPC_IAC2 | 64
1759 PPC | KVM_REG_PPC_IAC3 | 64
1760 PPC | KVM_REG_PPC_IAC4 | 64
1761 PPC | KVM_REG_PPC_DAC1 | 64
1762 PPC | KVM_REG_PPC_DAC2 | 64
1763 PPC | KVM_REG_PPC_DABR | 64
1764 PPC | KVM_REG_PPC_DSCR | 64
1765 PPC | KVM_REG_PPC_PURR | 64
1766 PPC | KVM_REG_PPC_SPURR | 64
1767 PPC | KVM_REG_PPC_DAR | 64
1768 PPC | KVM_REG_PPC_DSISR | 32
1769 PPC | KVM_REG_PPC_AMR | 64
1770 PPC | KVM_REG_PPC_UAMOR | 64
1771 PPC | KVM_REG_PPC_MMCR0 | 64
1772 PPC | KVM_REG_PPC_MMCR1 | 64
1773 PPC | KVM_REG_PPC_MMCRA | 64
1774 PPC | KVM_REG_PPC_PMC1 | 32
1775 PPC | KVM_REG_PPC_PMC2 | 32
1776 PPC | KVM_REG_PPC_PMC3 | 32
1777 PPC | KVM_REG_PPC_PMC4 | 32
1778 PPC | KVM_REG_PPC_PMC5 | 32
1779 PPC | KVM_REG_PPC_PMC6 | 32
1780 PPC | KVM_REG_PPC_PMC7 | 32
1781 PPC | KVM_REG_PPC_PMC8 | 32
1782 PPC | KVM_REG_PPC_FPR0 | 64
1784 PPC | KVM_REG_PPC_FPR31 | 64
1785 PPC | KVM_REG_PPC_VR0 | 128
1787 PPC | KVM_REG_PPC_VR31 | 128
1788 PPC | KVM_REG_PPC_VSR0 | 128
1790 PPC | KVM_REG_PPC_VSR31 | 128
1791 PPC | KVM_REG_PPC_FPSCR | 64
1792 PPC | KVM_REG_PPC_VSCR | 32
1793 PPC | KVM_REG_PPC_VPA_ADDR | 64
1794 PPC | KVM_REG_PPC_VPA_SLB | 128
1795 PPC | KVM_REG_PPC_VPA_DTL | 128
1796 PPC | KVM_REG_PPC_EPCR | 32
1797 PPC | KVM_REG_PPC_EPR | 32
1798 PPC | KVM_REG_PPC_TCR | 32
1799 PPC | KVM_REG_PPC_TSR | 32
1800 PPC | KVM_REG_PPC_OR_TSR | 32
1801 PPC | KVM_REG_PPC_CLEAR_TSR | 32
1802 PPC | KVM_REG_PPC_MAS0 | 32
1803 PPC | KVM_REG_PPC_MAS1 | 32
1804 PPC | KVM_REG_PPC_MAS2 | 64
1805 PPC | KVM_REG_PPC_MAS7_3 | 64
1806 PPC | KVM_REG_PPC_MAS4 | 32
1807 PPC | KVM_REG_PPC_MAS6 | 32
1808 PPC | KVM_REG_PPC_MMUCFG | 32
1809 PPC | KVM_REG_PPC_TLB0CFG | 32
1810 PPC | KVM_REG_PPC_TLB1CFG | 32
1811 PPC | KVM_REG_PPC_TLB2CFG | 32
1812 PPC | KVM_REG_PPC_TLB3CFG | 32
1813 PPC | KVM_REG_PPC_TLB0PS | 32
1814 PPC | KVM_REG_PPC_TLB1PS | 32
1815 PPC | KVM_REG_PPC_TLB2PS | 32
1816 PPC | KVM_REG_PPC_TLB3PS | 32
1817 PPC | KVM_REG_PPC_EPTCFG | 32
1818 PPC | KVM_REG_PPC_ICP_STATE | 64
1820 ARM registers are mapped using the lower 32 bits. The upper 16 of that
1821 is the register group type, or coprocessor number:
1823 ARM core registers have the following id bit patterns:
1824 0x4020 0000 0010 <index into the kvm_regs struct:16>
1826 ARM 32-bit CP15 registers have the following id bit patterns:
1827 0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3>
1829 ARM 64-bit CP15 registers have the following id bit patterns:
1830 0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3>
1832 ARM CCSIDR registers are demultiplexed by CSSELR value:
1833 0x4020 0000 0011 00 <csselr:8>
1835 ARM 32-bit VFP control registers have the following id bit patterns:
1836 0x4020 0000 0012 1 <regno:12>
1838 ARM 64-bit FP registers have the following id bit patterns:
1839 0x4030 0000 0012 0 <regno:12>
1842 arm64 registers are mapped using the lower 32 bits. The upper 16 of
1843 that is the register group type, or coprocessor number:
1845 arm64 core/FP-SIMD registers have the following id bit patterns. Note
1846 that the size of the access is variable, as the kvm_regs structure
1847 contains elements ranging from 32 to 128 bits. The index is a 32bit
1848 value in the kvm_regs structure seen as a 32bit array.
1849 0x60x0 0000 0010 <index into the kvm_regs struct:16>
1851 arm64 CCSIDR registers are demultiplexed by CSSELR value:
1852 0x6020 0000 0011 00 <csselr:8>
1854 arm64 system registers have the following id bit patterns:
1855 0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3>
1857 4.69 KVM_GET_ONE_REG
1859 Capability: KVM_CAP_ONE_REG
1862 Parameters: struct kvm_one_reg (in and out)
1863 Returns: 0 on success, negative value on failure
1865 This ioctl allows to receive the value of a single register implemented
1866 in a vcpu. The register to read is indicated by the "id" field of the
1867 kvm_one_reg struct passed in. On success, the register value can be found
1868 at the memory location pointed to by "addr".
1870 The list of registers accessible using this interface is identical to the
1874 4.70 KVM_KVMCLOCK_CTRL
1876 Capability: KVM_CAP_KVMCLOCK_CTRL
1877 Architectures: Any that implement pvclocks (currently x86 only)
1880 Returns: 0 on success, -1 on error
1882 This signals to the host kernel that the specified guest is being paused by
1883 userspace. The host will set a flag in the pvclock structure that is checked
1884 from the soft lockup watchdog. The flag is part of the pvclock structure that
1885 is shared between guest and host, specifically the second bit of the flags
1886 field of the pvclock_vcpu_time_info structure. It will be set exclusively by
1887 the host and read/cleared exclusively by the guest. The guest operation of
1888 checking and clearing the flag must an atomic operation so
1889 load-link/store-conditional, or equivalent must be used. There are two cases
1890 where the guest will clear the flag: when the soft lockup watchdog timer resets
1891 itself or when a soft lockup is detected. This ioctl can be called any time
1892 after pausing the vcpu, but before it is resumed.
1897 Capability: KVM_CAP_SIGNAL_MSI
1900 Parameters: struct kvm_msi (in)
1901 Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error
1903 Directly inject a MSI message. Only valid with in-kernel irqchip that handles
1914 No flags are defined so far. The corresponding field must be 0.
1917 4.71 KVM_CREATE_PIT2
1919 Capability: KVM_CAP_PIT2
1922 Parameters: struct kvm_pit_config (in)
1923 Returns: 0 on success, -1 on error
1925 Creates an in-kernel device model for the i8254 PIT. This call is only valid
1926 after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following
1927 parameters have to be passed:
1929 struct kvm_pit_config {
1936 #define KVM_PIT_SPEAKER_DUMMY 1 /* emulate speaker port stub */
1938 PIT timer interrupts may use a per-VM kernel thread for injection. If it
1939 exists, this thread will have a name of the following pattern:
1941 kvm-pit/<owner-process-pid>
1943 When running a guest with elevated priorities, the scheduling parameters of
1944 this thread may have to be adjusted accordingly.
1946 This IOCTL replaces the obsolete KVM_CREATE_PIT.
1951 Capability: KVM_CAP_PIT_STATE2
1954 Parameters: struct kvm_pit_state2 (out)
1955 Returns: 0 on success, -1 on error
1957 Retrieves the state of the in-kernel PIT model. Only valid after
1958 KVM_CREATE_PIT2. The state is returned in the following structure:
1960 struct kvm_pit_state2 {
1961 struct kvm_pit_channel_state channels[3];
1968 /* disable PIT in HPET legacy mode */
1969 #define KVM_PIT_FLAGS_HPET_LEGACY 0x00000001
1971 This IOCTL replaces the obsolete KVM_GET_PIT.
1976 Capability: KVM_CAP_PIT_STATE2
1979 Parameters: struct kvm_pit_state2 (in)
1980 Returns: 0 on success, -1 on error
1982 Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2.
1983 See KVM_GET_PIT2 for details on struct kvm_pit_state2.
1985 This IOCTL replaces the obsolete KVM_SET_PIT.
1988 4.74 KVM_PPC_GET_SMMU_INFO
1990 Capability: KVM_CAP_PPC_GET_SMMU_INFO
1991 Architectures: powerpc
1994 Returns: 0 on success, -1 on error
1996 This populates and returns a structure describing the features of
1997 the "Server" class MMU emulation supported by KVM.
1998 This can in turn be used by userspace to generate the appropariate
1999 device-tree properties for the guest operating system.
2001 The structure contains some global informations, followed by an
2002 array of supported segment page sizes:
2004 struct kvm_ppc_smmu_info {
2008 struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ];
2011 The supported flags are:
2013 - KVM_PPC_PAGE_SIZES_REAL:
2014 When that flag is set, guest page sizes must "fit" the backing
2015 store page sizes. When not set, any page size in the list can
2016 be used regardless of how they are backed by userspace.
2018 - KVM_PPC_1T_SEGMENTS
2019 The emulated MMU supports 1T segments in addition to the
2022 The "slb_size" field indicates how many SLB entries are supported
2024 The "sps" array contains 8 entries indicating the supported base
2025 page sizes for a segment in increasing order. Each entry is defined
2028 struct kvm_ppc_one_seg_page_size {
2029 __u32 page_shift; /* Base page shift of segment (or 0) */
2030 __u32 slb_enc; /* SLB encoding for BookS */
2031 struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ];
2034 An entry with a "page_shift" of 0 is unused. Because the array is
2035 organized in increasing order, a lookup can stop when encoutering
2038 The "slb_enc" field provides the encoding to use in the SLB for the
2039 page size. The bits are in positions such as the value can directly
2040 be OR'ed into the "vsid" argument of the slbmte instruction.
2042 The "enc" array is a list which for each of those segment base page
2043 size provides the list of supported actual page sizes (which can be
2044 only larger or equal to the base page size), along with the
2045 corresponding encoding in the hash PTE. Similarily, the array is
2046 8 entries sorted by increasing sizes and an entry with a "0" shift
2047 is an empty entry and a terminator:
2049 struct kvm_ppc_one_page_size {
2050 __u32 page_shift; /* Page shift (or 0) */
2051 __u32 pte_enc; /* Encoding in the HPTE (>>12) */
2054 The "pte_enc" field provides a value that can OR'ed into the hash
2055 PTE's RPN field (ie, it needs to be shifted left by 12 to OR it
2056 into the hash PTE second double word).
2060 Capability: KVM_CAP_IRQFD
2063 Parameters: struct kvm_irqfd (in)
2064 Returns: 0 on success, -1 on error
2066 Allows setting an eventfd to directly trigger a guest interrupt.
2067 kvm_irqfd.fd specifies the file descriptor to use as the eventfd and
2068 kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When
2069 an event is tiggered on the eventfd, an interrupt is injected into
2070 the guest using the specified gsi pin. The irqfd is removed using
2071 the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd
2074 With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify
2075 mechanism allowing emulation of level-triggered, irqfd-based
2076 interrupts. When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an
2077 additional eventfd in the kvm_irqfd.resamplefd field. When operating
2078 in resample mode, posting of an interrupt through kvm_irq.fd asserts
2079 the specified gsi in the irqchip. When the irqchip is resampled, such
2080 as from an EOI, the gsi is de-asserted and the user is notifed via
2081 kvm_irqfd.resamplefd. It is the user's responsibility to re-queue
2082 the interrupt if the device making use of it still requires service.
2083 Note that closing the resamplefd is not sufficient to disable the
2084 irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment
2085 and need not be specified with KVM_IRQFD_FLAG_DEASSIGN.
2087 4.76 KVM_PPC_ALLOCATE_HTAB
2089 Capability: KVM_CAP_PPC_ALLOC_HTAB
2090 Architectures: powerpc
2092 Parameters: Pointer to u32 containing hash table order (in/out)
2093 Returns: 0 on success, -1 on error
2095 This requests the host kernel to allocate an MMU hash table for a
2096 guest using the PAPR paravirtualization interface. This only does
2097 anything if the kernel is configured to use the Book 3S HV style of
2098 virtualization. Otherwise the capability doesn't exist and the ioctl
2099 returns an ENOTTY error. The rest of this description assumes Book 3S
2102 There must be no vcpus running when this ioctl is called; if there
2103 are, it will do nothing and return an EBUSY error.
2105 The parameter is a pointer to a 32-bit unsigned integer variable
2106 containing the order (log base 2) of the desired size of the hash
2107 table, which must be between 18 and 46. On successful return from the
2108 ioctl, it will have been updated with the order of the hash table that
2111 If no hash table has been allocated when any vcpu is asked to run
2112 (with the KVM_RUN ioctl), the host kernel will allocate a
2113 default-sized hash table (16 MB).
2115 If this ioctl is called when a hash table has already been allocated,
2116 the kernel will clear out the existing hash table (zero all HPTEs) and
2117 return the hash table order in the parameter. (If the guest is using
2118 the virtualized real-mode area (VRMA) facility, the kernel will
2119 re-create the VMRA HPTEs on the next KVM_RUN of any vcpu.)
2121 4.77 KVM_S390_INTERRUPT
2125 Type: vm ioctl, vcpu ioctl
2126 Parameters: struct kvm_s390_interrupt (in)
2127 Returns: 0 on success, -1 on error
2129 Allows to inject an interrupt to the guest. Interrupts can be floating
2130 (vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type.
2132 Interrupt parameters are passed via kvm_s390_interrupt:
2134 struct kvm_s390_interrupt {
2140 type can be one of the following:
2142 KVM_S390_SIGP_STOP (vcpu) - sigp restart
2143 KVM_S390_PROGRAM_INT (vcpu) - program check; code in parm
2144 KVM_S390_SIGP_SET_PREFIX (vcpu) - sigp set prefix; prefix address in parm
2145 KVM_S390_RESTART (vcpu) - restart
2146 KVM_S390_INT_VIRTIO (vm) - virtio external interrupt; external interrupt
2147 parameters in parm and parm64
2148 KVM_S390_INT_SERVICE (vm) - sclp external interrupt; sclp parameter in parm
2149 KVM_S390_INT_EMERGENCY (vcpu) - sigp emergency; source cpu in parm
2150 KVM_S390_INT_EXTERNAL_CALL (vcpu) - sigp external call; source cpu in parm
2151 KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm) - compound value to indicate an
2152 I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel);
2153 I/O interruption parameters in parm (subchannel) and parm64 (intparm,
2154 interruption subclass)
2155 KVM_S390_MCHK (vm, vcpu) - machine check interrupt; cr 14 bits in parm,
2156 machine check interrupt code in parm64 (note that
2157 machine checks needing further payload are not
2158 supported by this ioctl)
2160 Note that the vcpu ioctl is asynchronous to vcpu execution.
2162 4.78 KVM_PPC_GET_HTAB_FD
2164 Capability: KVM_CAP_PPC_HTAB_FD
2165 Architectures: powerpc
2167 Parameters: Pointer to struct kvm_get_htab_fd (in)
2168 Returns: file descriptor number (>= 0) on success, -1 on error
2170 This returns a file descriptor that can be used either to read out the
2171 entries in the guest's hashed page table (HPT), or to write entries to
2172 initialize the HPT. The returned fd can only be written to if the
2173 KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and
2174 can only be read if that bit is clear. The argument struct looks like
2177 /* For KVM_PPC_GET_HTAB_FD */
2178 struct kvm_get_htab_fd {
2184 /* Values for kvm_get_htab_fd.flags */
2185 #define KVM_GET_HTAB_BOLTED_ONLY ((__u64)0x1)
2186 #define KVM_GET_HTAB_WRITE ((__u64)0x2)
2188 The `start_index' field gives the index in the HPT of the entry at
2189 which to start reading. It is ignored when writing.
2191 Reads on the fd will initially supply information about all
2192 "interesting" HPT entries. Interesting entries are those with the
2193 bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise
2194 all entries. When the end of the HPT is reached, the read() will
2195 return. If read() is called again on the fd, it will start again from
2196 the beginning of the HPT, but will only return HPT entries that have
2197 changed since they were last read.
2199 Data read or written is structured as a header (8 bytes) followed by a
2200 series of valid HPT entries (16 bytes) each. The header indicates how
2201 many valid HPT entries there are and how many invalid entries follow
2202 the valid entries. The invalid entries are not represented explicitly
2203 in the stream. The header format is:
2205 struct kvm_get_htab_header {
2211 Writes to the fd create HPT entries starting at the index given in the
2212 header; first `n_valid' valid entries with contents from the data
2213 written, then `n_invalid' invalid entries, invalidating any previously
2214 valid entries found.
2216 4.79 KVM_CREATE_DEVICE
2218 Capability: KVM_CAP_DEVICE_CTRL
2220 Parameters: struct kvm_create_device (in/out)
2221 Returns: 0 on success, -1 on error
2223 ENODEV: The device type is unknown or unsupported
2224 EEXIST: Device already created, and this type of device may not
2225 be instantiated multiple times
2227 Other error conditions may be defined by individual device types or
2228 have their standard meanings.
2230 Creates an emulated device in the kernel. The file descriptor returned
2231 in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR.
2233 If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the
2234 device type is supported (not necessarily whether it can be created
2237 Individual devices should not define flags. Attributes should be used
2238 for specifying any behavior that is not implied by the device type
2241 struct kvm_create_device {
2242 __u32 type; /* in: KVM_DEV_TYPE_xxx */
2243 __u32 fd; /* out: device handle */
2244 __u32 flags; /* in: KVM_CREATE_DEVICE_xxx */
2247 4.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR
2249 Capability: KVM_CAP_DEVICE_CTRL
2251 Parameters: struct kvm_device_attr
2252 Returns: 0 on success, -1 on error
2254 ENXIO: The group or attribute is unknown/unsupported for this device
2255 EPERM: The attribute cannot (currently) be accessed this way
2256 (e.g. read-only attribute, or attribute that only makes
2257 sense when the device is in a different state)
2259 Other error conditions may be defined by individual device types.
2261 Gets/sets a specified piece of device configuration and/or state. The
2262 semantics are device-specific. See individual device documentation in
2263 the "devices" directory. As with ONE_REG, the size of the data
2264 transferred is defined by the particular attribute.
2266 struct kvm_device_attr {
2267 __u32 flags; /* no flags currently defined */
2268 __u32 group; /* device-defined */
2269 __u64 attr; /* group-defined */
2270 __u64 addr; /* userspace address of attr data */
2273 4.81 KVM_HAS_DEVICE_ATTR
2275 Capability: KVM_CAP_DEVICE_CTRL
2277 Parameters: struct kvm_device_attr
2278 Returns: 0 on success, -1 on error
2280 ENXIO: The group or attribute is unknown/unsupported for this device
2282 Tests whether a device supports a particular attribute. A successful
2283 return indicates the attribute is implemented. It does not necessarily
2284 indicate that the attribute can be read or written in the device's
2285 current state. "addr" is ignored.
2287 4.77 KVM_ARM_VCPU_INIT
2290 Architectures: arm, arm64
2292 Parameters: struct struct kvm_vcpu_init (in)
2293 Returns: 0 on success; -1 on error
2295 Â EINVAL: Â Â Â the target is unknown, or the combination of features is invalid.
2296 Â ENOENT: Â Â Â a features bit specified is unknown.
2298 This tells KVM what type of CPU to present to the guest, and what
2299 optional features it should have. Â This will cause a reset of the cpu
2300 registers to their initial values. Â If this is not called, KVM_RUN will
2301 return ENOEXEC for that vcpu.
2303 Note that because some registers reflect machine topology, all vcpus
2304 should be created before this ioctl is invoked.
2307 - KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state.
2308 Depends on KVM_CAP_ARM_PSCI.
2309 - KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode.
2310 Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only).
2313 4.78 KVM_GET_REG_LIST
2316 Architectures: arm, arm64
2318 Parameters: struct kvm_reg_list (in/out)
2319 Returns: 0 on success; -1 on error
2321 Â E2BIG: Â Â Â Â the reg index list is too big to fit in the array specified by
2322 Â Â Â Â Â Â Â Â Â Â Â Â the user (the number required will be written into n).
2324 struct kvm_reg_list {
2325 __u64 n; /* number of registers in reg[] */
2329 This ioctl returns the guest registers that are supported for the
2330 KVM_GET_ONE_REG/KVM_SET_ONE_REG calls.
2333 4.85 KVM_ARM_SET_DEVICE_ADDR (deprecated)
2335 Capability: KVM_CAP_ARM_SET_DEVICE_ADDR
2336 Architectures: arm, arm64
2338 Parameters: struct kvm_arm_device_address (in)
2339 Returns: 0 on success, -1 on error
2341 ENODEV: The device id is unknown
2342 ENXIO: Device not supported on current system
2343 EEXIST: Address already set
2344 E2BIG: Address outside guest physical address space
2345 EBUSY: Address overlaps with other device range
2347 struct kvm_arm_device_addr {
2352 Specify a device address in the guest's physical address space where guests
2353 can access emulated or directly exposed devices, which the host kernel needs
2354 to know about. The id field is an architecture specific identifier for a
2357 ARM/arm64 divides the id field into two parts, a device id and an
2358 address type id specific to the individual device.
2360 Â bits: | 63 ... 32 | 31 ... 16 | 15 ... 0 |
2361 field: | 0x00000000 | device id | addr type id |
2363 ARM/arm64 currently only require this when using the in-kernel GIC
2364 support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2
2365 as the device id. When setting the base address for the guest's
2366 mapping of the VGIC virtual CPU and distributor interface, the ioctl
2367 must be called after calling KVM_CREATE_IRQCHIP, but before calling
2368 KVM_RUN on any of the VCPUs. Calling this ioctl twice for any of the
2369 base addresses will return -EEXIST.
2371 Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API
2372 should be used instead.
2375 4.86 KVM_PPC_RTAS_DEFINE_TOKEN
2377 Capability: KVM_CAP_PPC_RTAS
2380 Parameters: struct kvm_rtas_token_args
2381 Returns: 0 on success, -1 on error
2383 Defines a token value for a RTAS (Run Time Abstraction Services)
2384 service in order to allow it to be handled in the kernel. The
2385 argument struct gives the name of the service, which must be the name
2386 of a service that has a kernel-side implementation. If the token
2387 value is non-zero, it will be associated with that service, and
2388 subsequent RTAS calls by the guest specifying that token will be
2389 handled by the kernel. If the token value is 0, then any token
2390 associated with the service will be forgotten, and subsequent RTAS
2391 calls by the guest for that service will be passed to userspace to be
2395 5. The kvm_run structure
2396 ------------------------
2398 Application code obtains a pointer to the kvm_run structure by
2399 mmap()ing a vcpu fd. From that point, application code can control
2400 execution by changing fields in kvm_run prior to calling the KVM_RUN
2401 ioctl, and obtain information about the reason KVM_RUN returned by
2402 looking up structure members.
2406 __u8 request_interrupt_window;
2408 Request that KVM_RUN return when it becomes possible to inject external
2409 interrupts into the guest. Useful in conjunction with KVM_INTERRUPT.
2416 When KVM_RUN has returned successfully (return value 0), this informs
2417 application code why KVM_RUN has returned. Allowable values for this
2418 field are detailed below.
2420 __u8 ready_for_interrupt_injection;
2422 If request_interrupt_window has been specified, this field indicates
2423 an interrupt can be injected now with KVM_INTERRUPT.
2427 The value of the current interrupt flag. Only valid if in-kernel
2428 local APIC is not used.
2432 /* in (pre_kvm_run), out (post_kvm_run) */
2435 The value of the cr8 register. Only valid if in-kernel local APIC is
2436 not used. Both input and output.
2440 The value of the APIC BASE msr. Only valid if in-kernel local
2441 APIC is not used. Both input and output.
2444 /* KVM_EXIT_UNKNOWN */
2446 __u64 hardware_exit_reason;
2449 If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown
2450 reasons. Further architecture-specific information is available in
2451 hardware_exit_reason.
2453 /* KVM_EXIT_FAIL_ENTRY */
2455 __u64 hardware_entry_failure_reason;
2458 If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due
2459 to unknown reasons. Further architecture-specific information is
2460 available in hardware_entry_failure_reason.
2462 /* KVM_EXIT_EXCEPTION */
2472 #define KVM_EXIT_IO_IN 0
2473 #define KVM_EXIT_IO_OUT 1
2475 __u8 size; /* bytes */
2478 __u64 data_offset; /* relative to kvm_run start */
2481 If exit_reason is KVM_EXIT_IO, then the vcpu has
2482 executed a port I/O instruction which could not be satisfied by kvm.
2483 data_offset describes where the data is located (KVM_EXIT_IO_OUT) or
2484 where kvm expects application code to place the data for the next
2485 KVM_RUN invocation (KVM_EXIT_IO_IN). Data format is a packed array.
2488 struct kvm_debug_exit_arch arch;
2501 If exit_reason is KVM_EXIT_MMIO, then the vcpu has
2502 executed a memory-mapped I/O instruction which could not be satisfied
2503 by kvm. The 'data' member contains the written data if 'is_write' is
2504 true, and should be filled by application code otherwise.
2506 NOTE: For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_DCR,
2507 KVM_EXIT_PAPR and KVM_EXIT_EPR the corresponding
2508 operations are complete (and guest state is consistent) only after userspace
2509 has re-entered the kernel with KVM_RUN. The kernel side will first finish
2510 incomplete operations and then check for pending signals. Userspace
2511 can re-enter the guest with an unmasked signal pending to complete
2514 /* KVM_EXIT_HYPERCALL */
2523 Unused. This was once used for 'hypercall to userspace'. To implement
2524 such functionality, use KVM_EXIT_IO (x86) or KVM_EXIT_MMIO (all except s390).
2525 Note KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO.
2527 /* KVM_EXIT_TPR_ACCESS */
2534 To be documented (KVM_TPR_ACCESS_REPORTING).
2536 /* KVM_EXIT_S390_SIEIC */
2539 __u64 mask; /* psw upper half */
2540 __u64 addr; /* psw lower half */
2547 /* KVM_EXIT_S390_RESET */
2548 #define KVM_S390_RESET_POR 1
2549 #define KVM_S390_RESET_CLEAR 2
2550 #define KVM_S390_RESET_SUBSYSTEM 4
2551 #define KVM_S390_RESET_CPU_INIT 8
2552 #define KVM_S390_RESET_IPL 16
2553 __u64 s390_reset_flags;
2557 /* KVM_EXIT_S390_UCONTROL */
2559 __u64 trans_exc_code;
2563 s390 specific. A page fault has occurred for a user controlled virtual
2564 machine (KVM_VM_S390_UNCONTROL) on it's host page table that cannot be
2565 resolved by the kernel.
2566 The program code and the translation exception code that were placed
2567 in the cpu's lowcore are presented here as defined by the z Architecture
2568 Principles of Operation Book in the Chapter for Dynamic Address Translation
2585 MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch
2586 hypercalls and exit with this exit struct that contains all the guest gprs.
2588 If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall.
2589 Userspace can now handle the hypercall and when it's done modify the gprs as
2590 necessary. Upon guest entry all guest GPRs will then be replaced by the values
2593 /* KVM_EXIT_PAPR_HCALL */
2600 This is used on 64-bit PowerPC when emulating a pSeries partition,
2601 e.g. with the 'pseries' machine type in qemu. It occurs when the
2602 guest does a hypercall using the 'sc 1' instruction. The 'nr' field
2603 contains the hypercall number (from the guest R3), and 'args' contains
2604 the arguments (from the guest R4 - R12). Userspace should put the
2605 return code in 'ret' and any extra returned values in args[].
2606 The possible hypercalls are defined in the Power Architecture Platform
2607 Requirements (PAPR) document available from www.power.org (free
2608 developer registration required to access it).
2610 /* KVM_EXIT_S390_TSCH */
2612 __u16 subchannel_id;
2613 __u16 subchannel_nr;
2620 s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled
2621 and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O
2622 interrupt for the target subchannel has been dequeued and subchannel_id,
2623 subchannel_nr, io_int_parm and io_int_word contain the parameters for that
2624 interrupt. ipb is needed for instruction parameter decoding.
2631 On FSL BookE PowerPC chips, the interrupt controller has a fast patch
2632 interrupt acknowledge path to the core. When the core successfully
2633 delivers an interrupt, it automatically populates the EPR register with
2634 the interrupt vector number and acknowledges the interrupt inside
2635 the interrupt controller.
2637 In case the interrupt controller lives in user space, we need to do
2638 the interrupt acknowledge cycle through it to fetch the next to be
2639 delivered interrupt vector using this exit.
2641 It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an
2642 external interrupt has just been delivered into the guest. User space
2643 should put the acknowledged interrupt vector into the 'epr' field.
2645 /* KVM_EXIT_SYSTEM_EVENT */
2647 #define KVM_SYSTEM_EVENT_SHUTDOWN 1
2648 #define KVM_SYSTEM_EVENT_RESET 2
2653 If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered
2654 a system-level event using some architecture specific mechanism (hypercall
2655 or some special instruction). In case of ARM/ARM64, this is triggered using
2656 HVC instruction based PSCI call from the vcpu. The 'type' field describes
2657 the system-level event type. The 'flags' field describes architecture
2658 specific flags for the system-level event.
2660 /* Fix the size of the union. */
2665 * shared registers between kvm and userspace.
2666 * kvm_valid_regs specifies the register classes set by the host
2667 * kvm_dirty_regs specified the register classes dirtied by userspace
2668 * struct kvm_sync_regs is architecture specific, as well as the
2669 * bits for kvm_valid_regs and kvm_dirty_regs
2671 __u64 kvm_valid_regs;
2672 __u64 kvm_dirty_regs;
2674 struct kvm_sync_regs regs;
2678 If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access
2679 certain guest registers without having to call SET/GET_*REGS. Thus we can
2680 avoid some system call overhead if userspace has to handle the exit.
2681 Userspace can query the validity of the structure by checking
2682 kvm_valid_regs for specific bits. These bits are architecture specific
2683 and usually define the validity of a groups of registers. (e.g. one bit
2684 for general purpose registers)
2689 4.81 KVM_GET_EMULATED_CPUID
2691 Capability: KVM_CAP_EXT_EMUL_CPUID
2694 Parameters: struct kvm_cpuid2 (in/out)
2695 Returns: 0 on success, -1 on error
2700 struct kvm_cpuid_entry2 entries[0];
2703 The member 'flags' is used for passing flags from userspace.
2705 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
2706 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1)
2707 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2)
2709 struct kvm_cpuid_entry2 {
2720 This ioctl returns x86 cpuid features which are emulated by
2721 kvm.Userspace can use the information returned by this ioctl to query
2722 which features are emulated by kvm instead of being present natively.
2724 Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2
2725 structure with the 'nent' field indicating the number of entries in
2726 the variable-size array 'entries'. If the number of entries is too low
2727 to describe the cpu capabilities, an error (E2BIG) is returned. If the
2728 number is too high, the 'nent' field is adjusted and an error (ENOMEM)
2729 is returned. If the number is just right, the 'nent' field is adjusted
2730 to the number of valid entries in the 'entries' array, which is then
2733 The entries returned are the set CPUID bits of the respective features
2734 which kvm emulates, as returned by the CPUID instruction, with unknown
2735 or unsupported feature bits cleared.
2737 Features like x2apic, for example, may not be present in the host cpu
2738 but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be
2739 emulated efficiently and thus not included here.
2741 The fields in each entry are defined as follows:
2743 function: the eax value used to obtain the entry
2744 index: the ecx value used to obtain the entry (for entries that are
2746 flags: an OR of zero or more of the following:
2747 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
2748 if the index field is valid
2749 KVM_CPUID_FLAG_STATEFUL_FUNC:
2750 if cpuid for this function returns different values for successive
2751 invocations; there will be several entries with the same function,
2752 all with this flag set
2753 KVM_CPUID_FLAG_STATE_READ_NEXT:
2754 for KVM_CPUID_FLAG_STATEFUL_FUNC entries, set if this entry is
2755 the first entry to be read by a cpu
2756 eax, ebx, ecx, edx: the values returned by the cpuid instruction for
2757 this function/index combination
2760 6. Capabilities that can be enabled
2761 -----------------------------------
2763 There are certain capabilities that change the behavior of the virtual CPU when
2764 enabled. To enable them, please see section 4.37. Below you can find a list of
2765 capabilities and what their effect on the vCPU is when enabling them.
2767 The following information is provided along with the description:
2769 Architectures: which instruction set architectures provide this ioctl.
2770 x86 includes both i386 and x86_64.
2772 Parameters: what parameters are accepted by the capability.
2774 Returns: the return value. General error numbers (EBADF, ENOMEM, EINVAL)
2775 are not detailed, but errors with specific meanings are.
2782 Returns: 0 on success; -1 on error
2784 This capability enables interception of OSI hypercalls that otherwise would
2785 be treated as normal system calls to be injected into the guest. OSI hypercalls
2786 were invented by Mac-on-Linux to have a standardized communication mechanism
2787 between the guest and the host.
2789 When this capability is enabled, KVM_EXIT_OSI can occur.
2792 6.2 KVM_CAP_PPC_PAPR
2796 Returns: 0 on success; -1 on error
2798 This capability enables interception of PAPR hypercalls. PAPR hypercalls are
2799 done using the hypercall instruction "sc 1".
2801 It also sets the guest privilege level to "supervisor" mode. Usually the guest
2802 runs in "hypervisor" privilege mode with a few missing features.
2804 In addition to the above, it changes the semantics of SDR1. In this mode, the
2805 HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the
2806 HTAB invisible to the guest.
2808 When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur.
2814 Parameters: args[0] is the address of a struct kvm_config_tlb
2815 Returns: 0 on success; -1 on error
2817 struct kvm_config_tlb {
2824 Configures the virtual CPU's TLB array, establishing a shared memory area
2825 between userspace and KVM. The "params" and "array" fields are userspace
2826 addresses of mmu-type-specific data structures. The "array_len" field is an
2827 safety mechanism, and should be set to the size in bytes of the memory that
2828 userspace has reserved for the array. It must be at least the size dictated
2829 by "mmu_type" and "params".
2831 While KVM_RUN is active, the shared region is under control of KVM. Its
2832 contents are undefined, and any modification by userspace results in
2833 boundedly undefined behavior.
2835 On return from KVM_RUN, the shared region will reflect the current state of
2836 the guest's TLB. If userspace makes any changes, it must call KVM_DIRTY_TLB
2837 to tell KVM which entries have been changed, prior to calling KVM_RUN again
2840 For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV:
2841 - The "params" field is of type "struct kvm_book3e_206_tlb_params".
2842 - The "array" field points to an array of type "struct
2843 kvm_book3e_206_tlb_entry".
2844 - The array consists of all entries in the first TLB, followed by all
2845 entries in the second TLB.
2846 - Within a TLB, entries are ordered first by increasing set number. Within a
2847 set, entries are ordered by way (increasing ESEL).
2848 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1)
2849 where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value.
2850 - The tsize field of mas1 shall be set to 4K on TLB0, even though the
2851 hardware ignores this value for TLB0.
2853 6.4 KVM_CAP_S390_CSS_SUPPORT
2857 Returns: 0 on success; -1 on error
2859 This capability enables support for handling of channel I/O instructions.
2861 TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are
2862 handled in-kernel, while the other I/O instructions are passed to userspace.
2864 When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST
2865 SUBCHANNEL intercepts.
2870 Parameters: args[0] defines whether the proxy facility is active
2871 Returns: 0 on success; -1 on error
2873 This capability enables or disables the delivery of interrupts through the
2874 external proxy facility.
2876 When enabled (args[0] != 0), every time the guest gets an external interrupt
2877 delivered, it automatically exits into user space with a KVM_EXIT_EPR exit
2878 to receive the topmost interrupt vector.
2880 When disabled (args[0] == 0), behavior is as if this facility is unsupported.
2882 When this capability is enabled, KVM_EXIT_EPR can occur.
2884 6.6 KVM_CAP_IRQ_MPIC
2887 Parameters: args[0] is the MPIC device fd
2888 args[1] is the MPIC CPU number for this vcpu
2890 This capability connects the vcpu to an in-kernel MPIC device.
2892 6.7 KVM_CAP_IRQ_XICS
2895 Parameters: args[0] is the XICS device fd
2896 args[1] is the XICS CPU number (server ID) for this vcpu
2898 This capability connects the vcpu to an in-kernel XICS device.