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
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.
164 4.5 KVM_GET_VCPU_MMAP_SIZE
170 Returns: size of vcpu mmap area, in bytes
172 The KVM_RUN ioctl (cf.) communicates with userspace via a shared
173 memory region. This ioctl returns the size of that region. See the
174 KVM_RUN documentation for details.
177 4.6 KVM_SET_MEMORY_REGION
182 Parameters: struct kvm_memory_region (in)
183 Returns: 0 on success, -1 on error
185 This ioctl is obsolete and has been removed.
193 Parameters: vcpu id (apic id on x86)
194 Returns: vcpu fd on success, -1 on error
196 This API adds a vcpu to a virtual machine. The vcpu id is a small integer
197 in the range [0, max_vcpus).
199 The recommended max_vcpus value can be retrieved using the KVM_CAP_NR_VCPUS of
200 the KVM_CHECK_EXTENSION ioctl() at run-time.
201 The maximum possible value for max_vcpus can be retrieved using the
202 KVM_CAP_MAX_VCPUS of the KVM_CHECK_EXTENSION ioctl() at run-time.
204 If the KVM_CAP_NR_VCPUS does not exist, you should assume that max_vcpus is 4
206 If the KVM_CAP_MAX_VCPUS does not exist, you should assume that max_vcpus is
207 same as the value returned from KVM_CAP_NR_VCPUS.
209 On powerpc using book3s_hv mode, the vcpus are mapped onto virtual
210 threads in one or more virtual CPU cores. (This is because the
211 hardware requires all the hardware threads in a CPU core to be in the
212 same partition.) The KVM_CAP_PPC_SMT capability indicates the number
213 of vcpus per virtual core (vcore). The vcore id is obtained by
214 dividing the vcpu id by the number of vcpus per vcore. The vcpus in a
215 given vcore will always be in the same physical core as each other
216 (though that might be a different physical core from time to time).
217 Userspace can control the threading (SMT) mode of the guest by its
218 allocation of vcpu ids. For example, if userspace wants
219 single-threaded guest vcpus, it should make all vcpu ids be a multiple
220 of the number of vcpus per vcore.
222 For virtual cpus that have been created with S390 user controlled virtual
223 machines, the resulting vcpu fd can be memory mapped at page offset
224 KVM_S390_SIE_PAGE_OFFSET in order to obtain a memory map of the virtual
225 cpu's hardware control block.
228 4.8 KVM_GET_DIRTY_LOG (vm ioctl)
233 Parameters: struct kvm_dirty_log (in/out)
234 Returns: 0 on success, -1 on error
236 /* for KVM_GET_DIRTY_LOG */
237 struct kvm_dirty_log {
241 void __user *dirty_bitmap; /* one bit per page */
246 Given a memory slot, return a bitmap containing any pages dirtied
247 since the last call to this ioctl. Bit 0 is the first page in the
248 memory slot. Ensure the entire structure is cleared to avoid padding
252 4.9 KVM_SET_MEMORY_ALIAS
257 Parameters: struct kvm_memory_alias (in)
258 Returns: 0 (success), -1 (error)
260 This ioctl is obsolete and has been removed.
269 Returns: 0 on success, -1 on error
271 EINTR: an unmasked signal is pending
273 This ioctl is used to run a guest virtual cpu. While there are no
274 explicit parameters, there is an implicit parameter block that can be
275 obtained by mmap()ing the vcpu fd at offset 0, with the size given by
276 KVM_GET_VCPU_MMAP_SIZE. The parameter block is formatted as a 'struct
277 kvm_run' (see below).
283 Architectures: all except ARM
285 Parameters: struct kvm_regs (out)
286 Returns: 0 on success, -1 on error
288 Reads the general purpose registers from the vcpu.
292 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
293 __u64 rax, rbx, rcx, rdx;
294 __u64 rsi, rdi, rsp, rbp;
295 __u64 r8, r9, r10, r11;
296 __u64 r12, r13, r14, r15;
304 Architectures: all except ARM
306 Parameters: struct kvm_regs (in)
307 Returns: 0 on success, -1 on error
309 Writes the general purpose registers into the vcpu.
311 See KVM_GET_REGS for the data structure.
317 Architectures: x86, ppc
319 Parameters: struct kvm_sregs (out)
320 Returns: 0 on success, -1 on error
322 Reads special registers from the vcpu.
326 struct kvm_segment cs, ds, es, fs, gs, ss;
327 struct kvm_segment tr, ldt;
328 struct kvm_dtable gdt, idt;
329 __u64 cr0, cr2, cr3, cr4, cr8;
332 __u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64];
335 /* ppc -- see arch/powerpc/include/uapi/asm/kvm.h */
337 interrupt_bitmap is a bitmap of pending external interrupts. At most
338 one bit may be set. This interrupt has been acknowledged by the APIC
339 but not yet injected into the cpu core.
345 Architectures: x86, ppc
347 Parameters: struct kvm_sregs (in)
348 Returns: 0 on success, -1 on error
350 Writes special registers into the vcpu. See KVM_GET_SREGS for the
359 Parameters: struct kvm_translation (in/out)
360 Returns: 0 on success, -1 on error
362 Translates a virtual address according to the vcpu's current address
365 struct kvm_translation {
367 __u64 linear_address;
370 __u64 physical_address;
381 Architectures: x86, ppc
383 Parameters: struct kvm_interrupt (in)
384 Returns: 0 on success, -1 on error
386 Queues a hardware interrupt vector to be injected. This is only
387 useful if in-kernel local APIC or equivalent is not used.
389 /* for KVM_INTERRUPT */
390 struct kvm_interrupt {
397 Note 'irq' is an interrupt vector, not an interrupt pin or line.
401 Queues an external interrupt to be injected. This ioctl is overleaded
402 with 3 different irq values:
406 This injects an edge type external interrupt into the guest once it's ready
407 to receive interrupts. When injected, the interrupt is done.
409 b) KVM_INTERRUPT_UNSET
411 This unsets any pending interrupt.
413 Only available with KVM_CAP_PPC_UNSET_IRQ.
415 c) KVM_INTERRUPT_SET_LEVEL
417 This injects a level type external interrupt into the guest context. The
418 interrupt stays pending until a specific ioctl with KVM_INTERRUPT_UNSET
421 Only available with KVM_CAP_PPC_IRQ_LEVEL.
423 Note that any value for 'irq' other than the ones stated above is invalid
424 and incurs unexpected behavior.
435 Support for this has been removed. Use KVM_SET_GUEST_DEBUG instead.
443 Parameters: struct kvm_msrs (in/out)
444 Returns: 0 on success, -1 on error
446 Reads model-specific registers from the vcpu. Supported msr indices can
447 be obtained using KVM_GET_MSR_INDEX_LIST.
450 __u32 nmsrs; /* number of msrs in entries */
453 struct kvm_msr_entry entries[0];
456 struct kvm_msr_entry {
462 Application code should set the 'nmsrs' member (which indicates the
463 size of the entries array) and the 'index' member of each array entry.
464 kvm will fill in the 'data' member.
472 Parameters: struct kvm_msrs (in)
473 Returns: 0 on success, -1 on error
475 Writes model-specific registers to the vcpu. See KVM_GET_MSRS for the
478 Application code should set the 'nmsrs' member (which indicates the
479 size of the entries array), and the 'index' and 'data' members of each
488 Parameters: struct kvm_cpuid (in)
489 Returns: 0 on success, -1 on error
491 Defines the vcpu responses to the cpuid instruction. Applications
492 should use the KVM_SET_CPUID2 ioctl if available.
495 struct kvm_cpuid_entry {
504 /* for KVM_SET_CPUID */
508 struct kvm_cpuid_entry entries[0];
512 4.21 KVM_SET_SIGNAL_MASK
517 Parameters: struct kvm_signal_mask (in)
518 Returns: 0 on success, -1 on error
520 Defines which signals are blocked during execution of KVM_RUN. This
521 signal mask temporarily overrides the threads signal mask. Any
522 unblocked signal received (except SIGKILL and SIGSTOP, which retain
523 their traditional behaviour) will cause KVM_RUN to return with -EINTR.
525 Note the signal will only be delivered if not blocked by the original
528 /* for KVM_SET_SIGNAL_MASK */
529 struct kvm_signal_mask {
540 Parameters: struct kvm_fpu (out)
541 Returns: 0 on success, -1 on error
543 Reads the floating point state from the vcpu.
545 /* for KVM_GET_FPU and KVM_SET_FPU */
550 __u8 ftwx; /* in fxsave format */
566 Parameters: struct kvm_fpu (in)
567 Returns: 0 on success, -1 on error
569 Writes the floating point state to the vcpu.
571 /* for KVM_GET_FPU and KVM_SET_FPU */
576 __u8 ftwx; /* in fxsave format */
587 4.24 KVM_CREATE_IRQCHIP
589 Capability: KVM_CAP_IRQCHIP
590 Architectures: x86, ia64, ARM
593 Returns: 0 on success, -1 on error
595 Creates an interrupt controller model in the kernel. On x86, creates a virtual
596 ioapic, a virtual PIC (two PICs, nested), and sets up future vcpus to have a
597 local APIC. IRQ routing for GSIs 0-15 is set to both PIC and IOAPIC; GSI 16-23
598 only go to the IOAPIC. On ia64, a IOSAPIC is created. On ARM, a GIC is
604 Capability: KVM_CAP_IRQCHIP
605 Architectures: x86, ia64, arm
607 Parameters: struct kvm_irq_level
608 Returns: 0 on success, -1 on error
610 Sets the level of a GSI input to the interrupt controller model in the kernel.
611 On some architectures it is required that an interrupt controller model has
612 been previously created with KVM_CREATE_IRQCHIP. Note that edge-triggered
613 interrupts require the level to be set to 1 and then back to 0.
615 ARM can signal an interrupt either at the CPU level, or at the in-kernel irqchip
616 (GIC), and for in-kernel irqchip can tell the GIC to use PPIs designated for
617 specific cpus. The irq field is interpreted like this:
619 Â bits: | 31 ... 24 | 23 ... 16 | 15 ... 0 |
620 field: | irq_type | vcpu_index | irq_id |
622 The irq_type field has the following values:
623 - irq_type[0]: out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ
624 - irq_type[1]: in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.)
625 (the vcpu_index field is ignored)
626 - irq_type[2]: in-kernel GIC: PPI, irq_id between 16 and 31 (incl.)
628 (The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs)
630 In both cases, level is used to raise/lower the line.
632 struct kvm_irq_level {
635 __s32 status; /* not used for KVM_IRQ_LEVEL */
637 __u32 level; /* 0 or 1 */
643 Capability: KVM_CAP_IRQCHIP
644 Architectures: x86, ia64
646 Parameters: struct kvm_irqchip (in/out)
647 Returns: 0 on success, -1 on error
649 Reads the state of a kernel interrupt controller created with
650 KVM_CREATE_IRQCHIP into a buffer provided by the caller.
653 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
656 char dummy[512]; /* reserving space */
657 struct kvm_pic_state pic;
658 struct kvm_ioapic_state ioapic;
665 Capability: KVM_CAP_IRQCHIP
666 Architectures: x86, ia64
668 Parameters: struct kvm_irqchip (in)
669 Returns: 0 on success, -1 on error
671 Sets the state of a kernel interrupt controller created with
672 KVM_CREATE_IRQCHIP from a buffer provided by the caller.
675 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
678 char dummy[512]; /* reserving space */
679 struct kvm_pic_state pic;
680 struct kvm_ioapic_state ioapic;
685 4.28 KVM_XEN_HVM_CONFIG
687 Capability: KVM_CAP_XEN_HVM
690 Parameters: struct kvm_xen_hvm_config (in)
691 Returns: 0 on success, -1 on error
693 Sets the MSR that the Xen HVM guest uses to initialize its hypercall
694 page, and provides the starting address and size of the hypercall
695 blobs in userspace. When the guest writes the MSR, kvm copies one
696 page of a blob (32- or 64-bit, depending on the vcpu mode) to guest
699 struct kvm_xen_hvm_config {
712 Capability: KVM_CAP_ADJUST_CLOCK
715 Parameters: struct kvm_clock_data (out)
716 Returns: 0 on success, -1 on error
718 Gets the current timestamp of kvmclock as seen by the current guest. In
719 conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios
722 struct kvm_clock_data {
723 __u64 clock; /* kvmclock current value */
731 Capability: KVM_CAP_ADJUST_CLOCK
734 Parameters: struct kvm_clock_data (in)
735 Returns: 0 on success, -1 on error
737 Sets the current timestamp of kvmclock to the value specified in its parameter.
738 In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios
741 struct kvm_clock_data {
742 __u64 clock; /* kvmclock current value */
748 4.31 KVM_GET_VCPU_EVENTS
750 Capability: KVM_CAP_VCPU_EVENTS
751 Extended by: KVM_CAP_INTR_SHADOW
754 Parameters: struct kvm_vcpu_event (out)
755 Returns: 0 on success, -1 on error
757 Gets currently pending exceptions, interrupts, and NMIs as well as related
760 struct kvm_vcpu_events {
784 KVM_VCPUEVENT_VALID_SHADOW may be set in the flags field to signal that
785 interrupt.shadow contains a valid state. Otherwise, this field is undefined.
788 4.32 KVM_SET_VCPU_EVENTS
790 Capability: KVM_CAP_VCPU_EVENTS
791 Extended by: KVM_CAP_INTR_SHADOW
794 Parameters: struct kvm_vcpu_event (in)
795 Returns: 0 on success, -1 on error
797 Set pending exceptions, interrupts, and NMIs as well as related states of the
800 See KVM_GET_VCPU_EVENTS for the data structure.
802 Fields that may be modified asynchronously by running VCPUs can be excluded
803 from the update. These fields are nmi.pending and sipi_vector. Keep the
804 corresponding bits in the flags field cleared to suppress overwriting the
805 current in-kernel state. The bits are:
807 KVM_VCPUEVENT_VALID_NMI_PENDING - transfer nmi.pending to the kernel
808 KVM_VCPUEVENT_VALID_SIPI_VECTOR - transfer sipi_vector
810 If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in
811 the flags field to signal that interrupt.shadow contains a valid state and
812 shall be written into the VCPU.
815 4.33 KVM_GET_DEBUGREGS
817 Capability: KVM_CAP_DEBUGREGS
820 Parameters: struct kvm_debugregs (out)
821 Returns: 0 on success, -1 on error
823 Reads debug registers from the vcpu.
825 struct kvm_debugregs {
834 4.34 KVM_SET_DEBUGREGS
836 Capability: KVM_CAP_DEBUGREGS
839 Parameters: struct kvm_debugregs (in)
840 Returns: 0 on success, -1 on error
842 Writes debug registers into the vcpu.
844 See KVM_GET_DEBUGREGS for the data structure. The flags field is unused
845 yet and must be cleared on entry.
848 4.35 KVM_SET_USER_MEMORY_REGION
850 Capability: KVM_CAP_USER_MEM
853 Parameters: struct kvm_userspace_memory_region (in)
854 Returns: 0 on success, -1 on error
856 struct kvm_userspace_memory_region {
859 __u64 guest_phys_addr;
860 __u64 memory_size; /* bytes */
861 __u64 userspace_addr; /* start of the userspace allocated memory */
864 /* for kvm_memory_region::flags */
865 #define KVM_MEM_LOG_DIRTY_PAGES (1UL << 0)
866 #define KVM_MEM_READONLY (1UL << 1)
868 This ioctl allows the user to create or modify a guest physical memory
869 slot. When changing an existing slot, it may be moved in the guest
870 physical memory space, or its flags may be modified. It may not be
871 resized. Slots may not overlap in guest physical address space.
873 Memory for the region is taken starting at the address denoted by the
874 field userspace_addr, which must point at user addressable memory for
875 the entire memory slot size. Any object may back this memory, including
876 anonymous memory, ordinary files, and hugetlbfs.
878 It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr
879 be identical. This allows large pages in the guest to be backed by large
882 The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and
883 KVM_MEM_READONLY. The former can be set to instruct KVM to keep track of
884 writes to memory within the slot. See KVM_GET_DIRTY_LOG ioctl to know how to
885 use it. The latter can be set, if KVM_CAP_READONLY_MEM capability allows it,
886 to make a new slot read-only. In this case, writes to this memory will be
887 posted to userspace as KVM_EXIT_MMIO exits.
889 When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of
890 the memory region are automatically reflected into the guest. For example, an
891 mmap() that affects the region will be made visible immediately. Another
892 example is madvise(MADV_DROP).
894 It is recommended to use this API instead of the KVM_SET_MEMORY_REGION ioctl.
895 The KVM_SET_MEMORY_REGION does not allow fine grained control over memory
896 allocation and is deprecated.
899 4.36 KVM_SET_TSS_ADDR
901 Capability: KVM_CAP_SET_TSS_ADDR
904 Parameters: unsigned long tss_address (in)
905 Returns: 0 on success, -1 on error
907 This ioctl defines the physical address of a three-page region in the guest
908 physical address space. The region must be within the first 4GB of the
909 guest physical address space and must not conflict with any memory slot
910 or any mmio address. The guest may malfunction if it accesses this memory
913 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
914 because of a quirk in the virtualization implementation (see the internals
915 documentation when it pops into existence).
920 Capability: KVM_CAP_ENABLE_CAP
921 Architectures: ppc, s390
923 Parameters: struct kvm_enable_cap (in)
924 Returns: 0 on success; -1 on error
926 +Not all extensions are enabled by default. Using this ioctl the application
927 can enable an extension, making it available to the guest.
929 On systems that do not support this ioctl, it always fails. On systems that
930 do support it, it only works for extensions that are supported for enablement.
932 To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should
935 struct kvm_enable_cap {
939 The capability that is supposed to get enabled.
943 A bitfield indicating future enhancements. Has to be 0 for now.
947 Arguments for enabling a feature. If a feature needs initial values to
948 function properly, this is the place to put them.
954 4.38 KVM_GET_MP_STATE
956 Capability: KVM_CAP_MP_STATE
957 Architectures: x86, ia64
959 Parameters: struct kvm_mp_state (out)
960 Returns: 0 on success; -1 on error
962 struct kvm_mp_state {
966 Returns the vcpu's current "multiprocessing state" (though also valid on
967 uniprocessor guests).
971 - KVM_MP_STATE_RUNNABLE: the vcpu is currently running
972 - KVM_MP_STATE_UNINITIALIZED: the vcpu is an application processor (AP)
973 which has not yet received an INIT signal
974 - KVM_MP_STATE_INIT_RECEIVED: the vcpu has received an INIT signal, and is
976 - KVM_MP_STATE_HALTED: the vcpu has executed a HLT instruction and
977 is waiting for an interrupt
978 - KVM_MP_STATE_SIPI_RECEIVED: the vcpu has just received a SIPI (vector
979 accessible via KVM_GET_VCPU_EVENTS)
981 This ioctl is only useful after KVM_CREATE_IRQCHIP. Without an in-kernel
982 irqchip, the multiprocessing state must be maintained by userspace.
985 4.39 KVM_SET_MP_STATE
987 Capability: KVM_CAP_MP_STATE
988 Architectures: x86, ia64
990 Parameters: struct kvm_mp_state (in)
991 Returns: 0 on success; -1 on error
993 Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for
996 This ioctl is only useful after KVM_CREATE_IRQCHIP. Without an in-kernel
997 irqchip, the multiprocessing state must be maintained by userspace.
1000 4.40 KVM_SET_IDENTITY_MAP_ADDR
1002 Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR
1005 Parameters: unsigned long identity (in)
1006 Returns: 0 on success, -1 on error
1008 This ioctl defines the physical address of a one-page region in the guest
1009 physical address space. The region must be within the first 4GB of the
1010 guest physical address space and must not conflict with any memory slot
1011 or any mmio address. The guest may malfunction if it accesses this memory
1014 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1015 because of a quirk in the virtualization implementation (see the internals
1016 documentation when it pops into existence).
1019 4.41 KVM_SET_BOOT_CPU_ID
1021 Capability: KVM_CAP_SET_BOOT_CPU_ID
1022 Architectures: x86, ia64
1024 Parameters: unsigned long vcpu_id
1025 Returns: 0 on success, -1 on error
1027 Define which vcpu is the Bootstrap Processor (BSP). Values are the same
1028 as the vcpu id in KVM_CREATE_VCPU. If this ioctl is not called, the default
1034 Capability: KVM_CAP_XSAVE
1037 Parameters: struct kvm_xsave (out)
1038 Returns: 0 on success, -1 on error
1044 This ioctl would copy current vcpu's xsave struct to the userspace.
1049 Capability: KVM_CAP_XSAVE
1052 Parameters: struct kvm_xsave (in)
1053 Returns: 0 on success, -1 on error
1059 This ioctl would copy userspace's xsave struct to the kernel.
1064 Capability: KVM_CAP_XCRS
1067 Parameters: struct kvm_xcrs (out)
1068 Returns: 0 on success, -1 on error
1079 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1083 This ioctl would copy current vcpu's xcrs to the userspace.
1088 Capability: KVM_CAP_XCRS
1091 Parameters: struct kvm_xcrs (in)
1092 Returns: 0 on success, -1 on error
1103 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1107 This ioctl would set vcpu's xcr to the value userspace specified.
1110 4.46 KVM_GET_SUPPORTED_CPUID
1112 Capability: KVM_CAP_EXT_CPUID
1115 Parameters: struct kvm_cpuid2 (in/out)
1116 Returns: 0 on success, -1 on error
1121 struct kvm_cpuid_entry2 entries[0];
1124 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX 1
1125 #define KVM_CPUID_FLAG_STATEFUL_FUNC 2
1126 #define KVM_CPUID_FLAG_STATE_READ_NEXT 4
1128 struct kvm_cpuid_entry2 {
1139 This ioctl returns x86 cpuid features which are supported by both the hardware
1140 and kvm. Userspace can use the information returned by this ioctl to
1141 construct cpuid information (for KVM_SET_CPUID2) that is consistent with
1142 hardware, kernel, and userspace capabilities, and with user requirements (for
1143 example, the user may wish to constrain cpuid to emulate older hardware,
1144 or for feature consistency across a cluster).
1146 Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure
1147 with the 'nent' field indicating the number of entries in the variable-size
1148 array 'entries'. If the number of entries is too low to describe the cpu
1149 capabilities, an error (E2BIG) is returned. If the number is too high,
1150 the 'nent' field is adjusted and an error (ENOMEM) is returned. If the
1151 number is just right, the 'nent' field is adjusted to the number of valid
1152 entries in the 'entries' array, which is then filled.
1154 The entries returned are the host cpuid as returned by the cpuid instruction,
1155 with unknown or unsupported features masked out. Some features (for example,
1156 x2apic), may not be present in the host cpu, but are exposed by kvm if it can
1157 emulate them efficiently. The fields in each entry are defined as follows:
1159 function: the eax value used to obtain the entry
1160 index: the ecx value used to obtain the entry (for entries that are
1162 flags: an OR of zero or more of the following:
1163 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
1164 if the index field is valid
1165 KVM_CPUID_FLAG_STATEFUL_FUNC:
1166 if cpuid for this function returns different values for successive
1167 invocations; there will be several entries with the same function,
1168 all with this flag set
1169 KVM_CPUID_FLAG_STATE_READ_NEXT:
1170 for KVM_CPUID_FLAG_STATEFUL_FUNC entries, set if this entry is
1171 the first entry to be read by a cpu
1172 eax, ebx, ecx, edx: the values returned by the cpuid instruction for
1173 this function/index combination
1175 The TSC deadline timer feature (CPUID leaf 1, ecx[24]) is always returned
1176 as false, since the feature depends on KVM_CREATE_IRQCHIP for local APIC
1177 support. Instead it is reported via
1179 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER)
1181 if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the
1182 feature in userspace, then you can enable the feature for KVM_SET_CPUID2.
1185 4.47 KVM_PPC_GET_PVINFO
1187 Capability: KVM_CAP_PPC_GET_PVINFO
1190 Parameters: struct kvm_ppc_pvinfo (out)
1191 Returns: 0 on success, !0 on error
1193 struct kvm_ppc_pvinfo {
1199 This ioctl fetches PV specific information that need to be passed to the guest
1200 using the device tree or other means from vm context.
1202 The hcall array defines 4 instructions that make up a hypercall.
1204 If any additional field gets added to this structure later on, a bit for that
1205 additional piece of information will be set in the flags bitmap.
1207 The flags bitmap is defined as:
1209 /* the host supports the ePAPR idle hcall
1210 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1<<0)
1212 4.48 KVM_ASSIGN_PCI_DEVICE
1214 Capability: KVM_CAP_DEVICE_ASSIGNMENT
1215 Architectures: x86 ia64
1217 Parameters: struct kvm_assigned_pci_dev (in)
1218 Returns: 0 on success, -1 on error
1220 Assigns a host PCI device to the VM.
1222 struct kvm_assigned_pci_dev {
1223 __u32 assigned_dev_id;
1233 The PCI device is specified by the triple segnr, busnr, and devfn.
1234 Identification in succeeding service requests is done via assigned_dev_id. The
1235 following flags are specified:
1237 /* Depends on KVM_CAP_IOMMU */
1238 #define KVM_DEV_ASSIGN_ENABLE_IOMMU (1 << 0)
1239 /* The following two depend on KVM_CAP_PCI_2_3 */
1240 #define KVM_DEV_ASSIGN_PCI_2_3 (1 << 1)
1241 #define KVM_DEV_ASSIGN_MASK_INTX (1 << 2)
1243 If KVM_DEV_ASSIGN_PCI_2_3 is set, the kernel will manage legacy INTx interrupts
1244 via the PCI-2.3-compliant device-level mask, thus enable IRQ sharing with other
1245 assigned devices or host devices. KVM_DEV_ASSIGN_MASK_INTX specifies the
1246 guest's view on the INTx mask, see KVM_ASSIGN_SET_INTX_MASK for details.
1248 The KVM_DEV_ASSIGN_ENABLE_IOMMU flag is a mandatory option to ensure
1249 isolation of the device. Usages not specifying this flag are deprecated.
1251 Only PCI header type 0 devices with PCI BAR resources are supported by
1252 device assignment. The user requesting this ioctl must have read/write
1253 access to the PCI sysfs resource files associated with the device.
1256 4.49 KVM_DEASSIGN_PCI_DEVICE
1258 Capability: KVM_CAP_DEVICE_DEASSIGNMENT
1259 Architectures: x86 ia64
1261 Parameters: struct kvm_assigned_pci_dev (in)
1262 Returns: 0 on success, -1 on error
1264 Ends PCI device assignment, releasing all associated resources.
1266 See KVM_CAP_DEVICE_ASSIGNMENT for the data structure. Only assigned_dev_id is
1267 used in kvm_assigned_pci_dev to identify the device.
1270 4.50 KVM_ASSIGN_DEV_IRQ
1272 Capability: KVM_CAP_ASSIGN_DEV_IRQ
1273 Architectures: x86 ia64
1275 Parameters: struct kvm_assigned_irq (in)
1276 Returns: 0 on success, -1 on error
1278 Assigns an IRQ to a passed-through device.
1280 struct kvm_assigned_irq {
1281 __u32 assigned_dev_id;
1282 __u32 host_irq; /* ignored (legacy field) */
1290 The following flags are defined:
1292 #define KVM_DEV_IRQ_HOST_INTX (1 << 0)
1293 #define KVM_DEV_IRQ_HOST_MSI (1 << 1)
1294 #define KVM_DEV_IRQ_HOST_MSIX (1 << 2)
1296 #define KVM_DEV_IRQ_GUEST_INTX (1 << 8)
1297 #define KVM_DEV_IRQ_GUEST_MSI (1 << 9)
1298 #define KVM_DEV_IRQ_GUEST_MSIX (1 << 10)
1300 It is not valid to specify multiple types per host or guest IRQ. However, the
1301 IRQ type of host and guest can differ or can even be null.
1304 4.51 KVM_DEASSIGN_DEV_IRQ
1306 Capability: KVM_CAP_ASSIGN_DEV_IRQ
1307 Architectures: x86 ia64
1309 Parameters: struct kvm_assigned_irq (in)
1310 Returns: 0 on success, -1 on error
1312 Ends an IRQ assignment to a passed-through device.
1314 See KVM_ASSIGN_DEV_IRQ for the data structure. The target device is specified
1315 by assigned_dev_id, flags must correspond to the IRQ type specified on
1316 KVM_ASSIGN_DEV_IRQ. Partial deassignment of host or guest IRQ is allowed.
1319 4.52 KVM_SET_GSI_ROUTING
1321 Capability: KVM_CAP_IRQ_ROUTING
1322 Architectures: x86 ia64
1324 Parameters: struct kvm_irq_routing (in)
1325 Returns: 0 on success, -1 on error
1327 Sets the GSI routing table entries, overwriting any previously set entries.
1329 struct kvm_irq_routing {
1332 struct kvm_irq_routing_entry entries[0];
1335 No flags are specified so far, the corresponding field must be set to zero.
1337 struct kvm_irq_routing_entry {
1343 struct kvm_irq_routing_irqchip irqchip;
1344 struct kvm_irq_routing_msi msi;
1349 /* gsi routing entry types */
1350 #define KVM_IRQ_ROUTING_IRQCHIP 1
1351 #define KVM_IRQ_ROUTING_MSI 2
1353 No flags are specified so far, the corresponding field must be set to zero.
1355 struct kvm_irq_routing_irqchip {
1360 struct kvm_irq_routing_msi {
1368 4.53 KVM_ASSIGN_SET_MSIX_NR
1370 Capability: KVM_CAP_DEVICE_MSIX
1371 Architectures: x86 ia64
1373 Parameters: struct kvm_assigned_msix_nr (in)
1374 Returns: 0 on success, -1 on error
1376 Set the number of MSI-X interrupts for an assigned device. The number is
1377 reset again by terminating the MSI-X assignment of the device via
1378 KVM_DEASSIGN_DEV_IRQ. Calling this service more than once at any earlier
1381 struct kvm_assigned_msix_nr {
1382 __u32 assigned_dev_id;
1387 #define KVM_MAX_MSIX_PER_DEV 256
1390 4.54 KVM_ASSIGN_SET_MSIX_ENTRY
1392 Capability: KVM_CAP_DEVICE_MSIX
1393 Architectures: x86 ia64
1395 Parameters: struct kvm_assigned_msix_entry (in)
1396 Returns: 0 on success, -1 on error
1398 Specifies the routing of an MSI-X assigned device interrupt to a GSI. Setting
1399 the GSI vector to zero means disabling the interrupt.
1401 struct kvm_assigned_msix_entry {
1402 __u32 assigned_dev_id;
1404 __u16 entry; /* The index of entry in the MSI-X table */
1409 4.55 KVM_SET_TSC_KHZ
1411 Capability: KVM_CAP_TSC_CONTROL
1414 Parameters: virtual tsc_khz
1415 Returns: 0 on success, -1 on error
1417 Specifies the tsc frequency for the virtual machine. The unit of the
1421 4.56 KVM_GET_TSC_KHZ
1423 Capability: KVM_CAP_GET_TSC_KHZ
1427 Returns: virtual tsc-khz on success, negative value on error
1429 Returns the tsc frequency of the guest. The unit of the return value is
1430 KHz. If the host has unstable tsc this ioctl returns -EIO instead as an
1436 Capability: KVM_CAP_IRQCHIP
1439 Parameters: struct kvm_lapic_state (out)
1440 Returns: 0 on success, -1 on error
1442 #define KVM_APIC_REG_SIZE 0x400
1443 struct kvm_lapic_state {
1444 char regs[KVM_APIC_REG_SIZE];
1447 Reads the Local APIC registers and copies them into the input argument. The
1448 data format and layout are the same as documented in the architecture manual.
1453 Capability: KVM_CAP_IRQCHIP
1456 Parameters: struct kvm_lapic_state (in)
1457 Returns: 0 on success, -1 on error
1459 #define KVM_APIC_REG_SIZE 0x400
1460 struct kvm_lapic_state {
1461 char regs[KVM_APIC_REG_SIZE];
1464 Copies the input argument into the the Local APIC registers. The data format
1465 and layout are the same as documented in the architecture manual.
1470 Capability: KVM_CAP_IOEVENTFD
1473 Parameters: struct kvm_ioeventfd (in)
1474 Returns: 0 on success, !0 on error
1476 This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address
1477 within the guest. A guest write in the registered address will signal the
1478 provided event instead of triggering an exit.
1480 struct kvm_ioeventfd {
1482 __u64 addr; /* legal pio/mmio address */
1483 __u32 len; /* 1, 2, 4, or 8 bytes */
1489 For the special case of virtio-ccw devices on s390, the ioevent is matched
1490 to a subchannel/virtqueue tuple instead.
1492 The following flags are defined:
1494 #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch)
1495 #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio)
1496 #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign)
1497 #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \
1498 (1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify)
1500 If datamatch flag is set, the event will be signaled only if the written value
1501 to the registered address is equal to datamatch in struct kvm_ioeventfd.
1503 For virtio-ccw devices, addr contains the subchannel id and datamatch the
1509 Capability: KVM_CAP_SW_TLB
1512 Parameters: struct kvm_dirty_tlb (in)
1513 Returns: 0 on success, -1 on error
1515 struct kvm_dirty_tlb {
1520 This must be called whenever userspace has changed an entry in the shared
1521 TLB, prior to calling KVM_RUN on the associated vcpu.
1523 The "bitmap" field is the userspace address of an array. This array
1524 consists of a number of bits, equal to the total number of TLB entries as
1525 determined by the last successful call to KVM_CONFIG_TLB, rounded up to the
1526 nearest multiple of 64.
1528 Each bit corresponds to one TLB entry, ordered the same as in the shared TLB
1531 The array is little-endian: the bit 0 is the least significant bit of the
1532 first byte, bit 8 is the least significant bit of the second byte, etc.
1533 This avoids any complications with differing word sizes.
1535 The "num_dirty" field is a performance hint for KVM to determine whether it
1536 should skip processing the bitmap and just invalidate everything. It must
1537 be set to the number of set bits in the bitmap.
1540 4.61 KVM_ASSIGN_SET_INTX_MASK
1542 Capability: KVM_CAP_PCI_2_3
1545 Parameters: struct kvm_assigned_pci_dev (in)
1546 Returns: 0 on success, -1 on error
1548 Allows userspace to mask PCI INTx interrupts from the assigned device. The
1549 kernel will not deliver INTx interrupts to the guest between setting and
1550 clearing of KVM_ASSIGN_SET_INTX_MASK via this interface. This enables use of
1551 and emulation of PCI 2.3 INTx disable command register behavior.
1553 This may be used for both PCI 2.3 devices supporting INTx disable natively and
1554 older devices lacking this support. Userspace is responsible for emulating the
1555 read value of the INTx disable bit in the guest visible PCI command register.
1556 When modifying the INTx disable state, userspace should precede updating the
1557 physical device command register by calling this ioctl to inform the kernel of
1558 the new intended INTx mask state.
1560 Note that the kernel uses the device INTx disable bit to internally manage the
1561 device interrupt state for PCI 2.3 devices. Reads of this register may
1562 therefore not match the expected value. Writes should always use the guest
1563 intended INTx disable value rather than attempting to read-copy-update the
1564 current physical device state. Races between user and kernel updates to the
1565 INTx disable bit are handled lazily in the kernel. It's possible the device
1566 may generate unintended interrupts, but they will not be injected into the
1569 See KVM_ASSIGN_DEV_IRQ for the data structure. The target device is specified
1570 by assigned_dev_id. In the flags field, only KVM_DEV_ASSIGN_MASK_INTX is
1574 4.62 KVM_CREATE_SPAPR_TCE
1576 Capability: KVM_CAP_SPAPR_TCE
1577 Architectures: powerpc
1579 Parameters: struct kvm_create_spapr_tce (in)
1580 Returns: file descriptor for manipulating the created TCE table
1582 This creates a virtual TCE (translation control entry) table, which
1583 is an IOMMU for PAPR-style virtual I/O. It is used to translate
1584 logical addresses used in virtual I/O into guest physical addresses,
1585 and provides a scatter/gather capability for PAPR virtual I/O.
1587 /* for KVM_CAP_SPAPR_TCE */
1588 struct kvm_create_spapr_tce {
1593 The liobn field gives the logical IO bus number for which to create a
1594 TCE table. The window_size field specifies the size of the DMA window
1595 which this TCE table will translate - the table will contain one 64
1596 bit TCE entry for every 4kiB of the DMA window.
1598 When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE
1599 table has been created using this ioctl(), the kernel will handle it
1600 in real mode, updating the TCE table. H_PUT_TCE calls for other
1601 liobns will cause a vm exit and must be handled by userspace.
1603 The return value is a file descriptor which can be passed to mmap(2)
1604 to map the created TCE table into userspace. This lets userspace read
1605 the entries written by kernel-handled H_PUT_TCE calls, and also lets
1606 userspace update the TCE table directly which is useful in some
1610 4.63 KVM_ALLOCATE_RMA
1612 Capability: KVM_CAP_PPC_RMA
1613 Architectures: powerpc
1615 Parameters: struct kvm_allocate_rma (out)
1616 Returns: file descriptor for mapping the allocated RMA
1618 This allocates a Real Mode Area (RMA) from the pool allocated at boot
1619 time by the kernel. An RMA is a physically-contiguous, aligned region
1620 of memory used on older POWER processors to provide the memory which
1621 will be accessed by real-mode (MMU off) accesses in a KVM guest.
1622 POWER processors support a set of sizes for the RMA that usually
1623 includes 64MB, 128MB, 256MB and some larger powers of two.
1625 /* for KVM_ALLOCATE_RMA */
1626 struct kvm_allocate_rma {
1630 The return value is a file descriptor which can be passed to mmap(2)
1631 to map the allocated RMA into userspace. The mapped area can then be
1632 passed to the KVM_SET_USER_MEMORY_REGION ioctl to establish it as the
1633 RMA for a virtual machine. The size of the RMA in bytes (which is
1634 fixed at host kernel boot time) is returned in the rma_size field of
1635 the argument structure.
1637 The KVM_CAP_PPC_RMA capability is 1 or 2 if the KVM_ALLOCATE_RMA ioctl
1638 is supported; 2 if the processor requires all virtual machines to have
1639 an RMA, or 1 if the processor can use an RMA but doesn't require it,
1640 because it supports the Virtual RMA (VRMA) facility.
1645 Capability: KVM_CAP_USER_NMI
1649 Returns: 0 on success, -1 on error
1651 Queues an NMI on the thread's vcpu. Note this is well defined only
1652 when KVM_CREATE_IRQCHIP has not been called, since this is an interface
1653 between the virtual cpu core and virtual local APIC. After KVM_CREATE_IRQCHIP
1654 has been called, this interface is completely emulated within the kernel.
1656 To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the
1657 following algorithm:
1660 - read the local APIC's state (KVM_GET_LAPIC)
1661 - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1)
1662 - if so, issue KVM_NMI
1665 Some guests configure the LINT1 NMI input to cause a panic, aiding in
1669 4.65 KVM_S390_UCAS_MAP
1671 Capability: KVM_CAP_S390_UCONTROL
1674 Parameters: struct kvm_s390_ucas_mapping (in)
1675 Returns: 0 in case of success
1677 The parameter is defined like this:
1678 struct kvm_s390_ucas_mapping {
1684 This ioctl maps the memory at "user_addr" with the length "length" to
1685 the vcpu's address space starting at "vcpu_addr". All parameters need to
1686 be alligned by 1 megabyte.
1689 4.66 KVM_S390_UCAS_UNMAP
1691 Capability: KVM_CAP_S390_UCONTROL
1694 Parameters: struct kvm_s390_ucas_mapping (in)
1695 Returns: 0 in case of success
1697 The parameter is defined like this:
1698 struct kvm_s390_ucas_mapping {
1704 This ioctl unmaps the memory in the vcpu's address space starting at
1705 "vcpu_addr" with the length "length". The field "user_addr" is ignored.
1706 All parameters need to be alligned by 1 megabyte.
1709 4.67 KVM_S390_VCPU_FAULT
1711 Capability: KVM_CAP_S390_UCONTROL
1714 Parameters: vcpu absolute address (in)
1715 Returns: 0 in case of success
1717 This call creates a page table entry on the virtual cpu's address space
1718 (for user controlled virtual machines) or the virtual machine's address
1719 space (for regular virtual machines). This only works for minor faults,
1720 thus it's recommended to access subject memory page via the user page
1721 table upfront. This is useful to handle validity intercepts for user
1722 controlled virtual machines to fault in the virtual cpu's lowcore pages
1723 prior to calling the KVM_RUN ioctl.
1726 4.68 KVM_SET_ONE_REG
1728 Capability: KVM_CAP_ONE_REG
1731 Parameters: struct kvm_one_reg (in)
1732 Returns: 0 on success, negative value on failure
1734 struct kvm_one_reg {
1739 Using this ioctl, a single vcpu register can be set to a specific value
1740 defined by user space with the passed in struct kvm_one_reg, where id
1741 refers to the register identifier as described below and addr is a pointer
1742 to a variable with the respective size. There can be architecture agnostic
1743 and architecture specific registers. Each have their own range of operation
1744 and their own constants and width. To keep track of the implemented
1745 registers, find a list below:
1747 Arch | Register | Width (bits)
1749 PPC | KVM_REG_PPC_HIOR | 64
1750 PPC | KVM_REG_PPC_IAC1 | 64
1751 PPC | KVM_REG_PPC_IAC2 | 64
1752 PPC | KVM_REG_PPC_IAC3 | 64
1753 PPC | KVM_REG_PPC_IAC4 | 64
1754 PPC | KVM_REG_PPC_DAC1 | 64
1755 PPC | KVM_REG_PPC_DAC2 | 64
1756 PPC | KVM_REG_PPC_DABR | 64
1757 PPC | KVM_REG_PPC_DSCR | 64
1758 PPC | KVM_REG_PPC_PURR | 64
1759 PPC | KVM_REG_PPC_SPURR | 64
1760 PPC | KVM_REG_PPC_DAR | 64
1761 PPC | KVM_REG_PPC_DSISR | 32
1762 PPC | KVM_REG_PPC_AMR | 64
1763 PPC | KVM_REG_PPC_UAMOR | 64
1764 PPC | KVM_REG_PPC_MMCR0 | 64
1765 PPC | KVM_REG_PPC_MMCR1 | 64
1766 PPC | KVM_REG_PPC_MMCRA | 64
1767 PPC | KVM_REG_PPC_PMC1 | 32
1768 PPC | KVM_REG_PPC_PMC2 | 32
1769 PPC | KVM_REG_PPC_PMC3 | 32
1770 PPC | KVM_REG_PPC_PMC4 | 32
1771 PPC | KVM_REG_PPC_PMC5 | 32
1772 PPC | KVM_REG_PPC_PMC6 | 32
1773 PPC | KVM_REG_PPC_PMC7 | 32
1774 PPC | KVM_REG_PPC_PMC8 | 32
1775 PPC | KVM_REG_PPC_FPR0 | 64
1777 PPC | KVM_REG_PPC_FPR31 | 64
1778 PPC | KVM_REG_PPC_VR0 | 128
1780 PPC | KVM_REG_PPC_VR31 | 128
1781 PPC | KVM_REG_PPC_VSR0 | 128
1783 PPC | KVM_REG_PPC_VSR31 | 128
1784 PPC | KVM_REG_PPC_FPSCR | 64
1785 PPC | KVM_REG_PPC_VSCR | 32
1786 PPC | KVM_REG_PPC_VPA_ADDR | 64
1787 PPC | KVM_REG_PPC_VPA_SLB | 128
1788 PPC | KVM_REG_PPC_VPA_DTL | 128
1789 PPC | KVM_REG_PPC_EPCR | 32
1790 PPC | KVM_REG_PPC_EPR | 32
1791 PPC | KVM_REG_PPC_TCR | 32
1792 PPC | KVM_REG_PPC_TSR | 32
1793 PPC | KVM_REG_PPC_OR_TSR | 32
1794 PPC | KVM_REG_PPC_CLEAR_TSR | 32
1795 PPC | KVM_REG_PPC_MAS0 | 32
1796 PPC | KVM_REG_PPC_MAS1 | 32
1797 PPC | KVM_REG_PPC_MAS2 | 64
1798 PPC | KVM_REG_PPC_MAS7_3 | 64
1799 PPC | KVM_REG_PPC_MAS4 | 32
1800 PPC | KVM_REG_PPC_MAS6 | 32
1801 PPC | KVM_REG_PPC_MMUCFG | 32
1802 PPC | KVM_REG_PPC_TLB0CFG | 32
1803 PPC | KVM_REG_PPC_TLB1CFG | 32
1804 PPC | KVM_REG_PPC_TLB2CFG | 32
1805 PPC | KVM_REG_PPC_TLB3CFG | 32
1806 PPC | KVM_REG_PPC_TLB0PS | 32
1807 PPC | KVM_REG_PPC_TLB1PS | 32
1808 PPC | KVM_REG_PPC_TLB2PS | 32
1809 PPC | KVM_REG_PPC_TLB3PS | 32
1810 PPC | KVM_REG_PPC_EPTCFG | 32
1811 PPC | KVM_REG_PPC_ICP_STATE | 64
1813 ARM registers are mapped using the lower 32 bits. The upper 16 of that
1814 is the register group type, or coprocessor number:
1816 ARM core registers have the following id bit patterns:
1817 0x4020 0000 0010 <index into the kvm_regs struct:16>
1819 ARM 32-bit CP15 registers have the following id bit patterns:
1820 0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3>
1822 ARM 64-bit CP15 registers have the following id bit patterns:
1823 0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3>
1825 ARM CCSIDR registers are demultiplexed by CSSELR value:
1826 0x4020 0000 0011 00 <csselr:8>
1828 ARM 32-bit VFP control registers have the following id bit patterns:
1829 0x4020 0000 0012 1 <regno:12>
1831 ARM 64-bit FP registers have the following id bit patterns:
1832 0x4030 0000 0012 0 <regno:12>
1834 4.69 KVM_GET_ONE_REG
1836 Capability: KVM_CAP_ONE_REG
1839 Parameters: struct kvm_one_reg (in and out)
1840 Returns: 0 on success, negative value on failure
1842 This ioctl allows to receive the value of a single register implemented
1843 in a vcpu. The register to read is indicated by the "id" field of the
1844 kvm_one_reg struct passed in. On success, the register value can be found
1845 at the memory location pointed to by "addr".
1847 The list of registers accessible using this interface is identical to the
1851 4.70 KVM_KVMCLOCK_CTRL
1853 Capability: KVM_CAP_KVMCLOCK_CTRL
1854 Architectures: Any that implement pvclocks (currently x86 only)
1857 Returns: 0 on success, -1 on error
1859 This signals to the host kernel that the specified guest is being paused by
1860 userspace. The host will set a flag in the pvclock structure that is checked
1861 from the soft lockup watchdog. The flag is part of the pvclock structure that
1862 is shared between guest and host, specifically the second bit of the flags
1863 field of the pvclock_vcpu_time_info structure. It will be set exclusively by
1864 the host and read/cleared exclusively by the guest. The guest operation of
1865 checking and clearing the flag must an atomic operation so
1866 load-link/store-conditional, or equivalent must be used. There are two cases
1867 where the guest will clear the flag: when the soft lockup watchdog timer resets
1868 itself or when a soft lockup is detected. This ioctl can be called any time
1869 after pausing the vcpu, but before it is resumed.
1874 Capability: KVM_CAP_SIGNAL_MSI
1877 Parameters: struct kvm_msi (in)
1878 Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error
1880 Directly inject a MSI message. Only valid with in-kernel irqchip that handles
1891 No flags are defined so far. The corresponding field must be 0.
1894 4.71 KVM_CREATE_PIT2
1896 Capability: KVM_CAP_PIT2
1899 Parameters: struct kvm_pit_config (in)
1900 Returns: 0 on success, -1 on error
1902 Creates an in-kernel device model for the i8254 PIT. This call is only valid
1903 after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following
1904 parameters have to be passed:
1906 struct kvm_pit_config {
1913 #define KVM_PIT_SPEAKER_DUMMY 1 /* emulate speaker port stub */
1915 PIT timer interrupts may use a per-VM kernel thread for injection. If it
1916 exists, this thread will have a name of the following pattern:
1918 kvm-pit/<owner-process-pid>
1920 When running a guest with elevated priorities, the scheduling parameters of
1921 this thread may have to be adjusted accordingly.
1923 This IOCTL replaces the obsolete KVM_CREATE_PIT.
1928 Capability: KVM_CAP_PIT_STATE2
1931 Parameters: struct kvm_pit_state2 (out)
1932 Returns: 0 on success, -1 on error
1934 Retrieves the state of the in-kernel PIT model. Only valid after
1935 KVM_CREATE_PIT2. The state is returned in the following structure:
1937 struct kvm_pit_state2 {
1938 struct kvm_pit_channel_state channels[3];
1945 /* disable PIT in HPET legacy mode */
1946 #define KVM_PIT_FLAGS_HPET_LEGACY 0x00000001
1948 This IOCTL replaces the obsolete KVM_GET_PIT.
1953 Capability: KVM_CAP_PIT_STATE2
1956 Parameters: struct kvm_pit_state2 (in)
1957 Returns: 0 on success, -1 on error
1959 Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2.
1960 See KVM_GET_PIT2 for details on struct kvm_pit_state2.
1962 This IOCTL replaces the obsolete KVM_SET_PIT.
1965 4.74 KVM_PPC_GET_SMMU_INFO
1967 Capability: KVM_CAP_PPC_GET_SMMU_INFO
1968 Architectures: powerpc
1971 Returns: 0 on success, -1 on error
1973 This populates and returns a structure describing the features of
1974 the "Server" class MMU emulation supported by KVM.
1975 This can in turn be used by userspace to generate the appropariate
1976 device-tree properties for the guest operating system.
1978 The structure contains some global informations, followed by an
1979 array of supported segment page sizes:
1981 struct kvm_ppc_smmu_info {
1985 struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ];
1988 The supported flags are:
1990 - KVM_PPC_PAGE_SIZES_REAL:
1991 When that flag is set, guest page sizes must "fit" the backing
1992 store page sizes. When not set, any page size in the list can
1993 be used regardless of how they are backed by userspace.
1995 - KVM_PPC_1T_SEGMENTS
1996 The emulated MMU supports 1T segments in addition to the
1999 The "slb_size" field indicates how many SLB entries are supported
2001 The "sps" array contains 8 entries indicating the supported base
2002 page sizes for a segment in increasing order. Each entry is defined
2005 struct kvm_ppc_one_seg_page_size {
2006 __u32 page_shift; /* Base page shift of segment (or 0) */
2007 __u32 slb_enc; /* SLB encoding for BookS */
2008 struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ];
2011 An entry with a "page_shift" of 0 is unused. Because the array is
2012 organized in increasing order, a lookup can stop when encoutering
2015 The "slb_enc" field provides the encoding to use in the SLB for the
2016 page size. The bits are in positions such as the value can directly
2017 be OR'ed into the "vsid" argument of the slbmte instruction.
2019 The "enc" array is a list which for each of those segment base page
2020 size provides the list of supported actual page sizes (which can be
2021 only larger or equal to the base page size), along with the
2022 corresponding encoding in the hash PTE. Similarily, the array is
2023 8 entries sorted by increasing sizes and an entry with a "0" shift
2024 is an empty entry and a terminator:
2026 struct kvm_ppc_one_page_size {
2027 __u32 page_shift; /* Page shift (or 0) */
2028 __u32 pte_enc; /* Encoding in the HPTE (>>12) */
2031 The "pte_enc" field provides a value that can OR'ed into the hash
2032 PTE's RPN field (ie, it needs to be shifted left by 12 to OR it
2033 into the hash PTE second double word).
2037 Capability: KVM_CAP_IRQFD
2040 Parameters: struct kvm_irqfd (in)
2041 Returns: 0 on success, -1 on error
2043 Allows setting an eventfd to directly trigger a guest interrupt.
2044 kvm_irqfd.fd specifies the file descriptor to use as the eventfd and
2045 kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When
2046 an event is tiggered on the eventfd, an interrupt is injected into
2047 the guest using the specified gsi pin. The irqfd is removed using
2048 the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd
2051 With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify
2052 mechanism allowing emulation of level-triggered, irqfd-based
2053 interrupts. When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an
2054 additional eventfd in the kvm_irqfd.resamplefd field. When operating
2055 in resample mode, posting of an interrupt through kvm_irq.fd asserts
2056 the specified gsi in the irqchip. When the irqchip is resampled, such
2057 as from an EOI, the gsi is de-asserted and the user is notifed via
2058 kvm_irqfd.resamplefd. It is the user's responsibility to re-queue
2059 the interrupt if the device making use of it still requires service.
2060 Note that closing the resamplefd is not sufficient to disable the
2061 irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment
2062 and need not be specified with KVM_IRQFD_FLAG_DEASSIGN.
2064 4.76 KVM_PPC_ALLOCATE_HTAB
2066 Capability: KVM_CAP_PPC_ALLOC_HTAB
2067 Architectures: powerpc
2069 Parameters: Pointer to u32 containing hash table order (in/out)
2070 Returns: 0 on success, -1 on error
2072 This requests the host kernel to allocate an MMU hash table for a
2073 guest using the PAPR paravirtualization interface. This only does
2074 anything if the kernel is configured to use the Book 3S HV style of
2075 virtualization. Otherwise the capability doesn't exist and the ioctl
2076 returns an ENOTTY error. The rest of this description assumes Book 3S
2079 There must be no vcpus running when this ioctl is called; if there
2080 are, it will do nothing and return an EBUSY error.
2082 The parameter is a pointer to a 32-bit unsigned integer variable
2083 containing the order (log base 2) of the desired size of the hash
2084 table, which must be between 18 and 46. On successful return from the
2085 ioctl, it will have been updated with the order of the hash table that
2088 If no hash table has been allocated when any vcpu is asked to run
2089 (with the KVM_RUN ioctl), the host kernel will allocate a
2090 default-sized hash table (16 MB).
2092 If this ioctl is called when a hash table has already been allocated,
2093 the kernel will clear out the existing hash table (zero all HPTEs) and
2094 return the hash table order in the parameter. (If the guest is using
2095 the virtualized real-mode area (VRMA) facility, the kernel will
2096 re-create the VMRA HPTEs on the next KVM_RUN of any vcpu.)
2098 4.77 KVM_S390_INTERRUPT
2102 Type: vm ioctl, vcpu ioctl
2103 Parameters: struct kvm_s390_interrupt (in)
2104 Returns: 0 on success, -1 on error
2106 Allows to inject an interrupt to the guest. Interrupts can be floating
2107 (vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type.
2109 Interrupt parameters are passed via kvm_s390_interrupt:
2111 struct kvm_s390_interrupt {
2117 type can be one of the following:
2119 KVM_S390_SIGP_STOP (vcpu) - sigp restart
2120 KVM_S390_PROGRAM_INT (vcpu) - program check; code in parm
2121 KVM_S390_SIGP_SET_PREFIX (vcpu) - sigp set prefix; prefix address in parm
2122 KVM_S390_RESTART (vcpu) - restart
2123 KVM_S390_INT_VIRTIO (vm) - virtio external interrupt; external interrupt
2124 parameters in parm and parm64
2125 KVM_S390_INT_SERVICE (vm) - sclp external interrupt; sclp parameter in parm
2126 KVM_S390_INT_EMERGENCY (vcpu) - sigp emergency; source cpu in parm
2127 KVM_S390_INT_EXTERNAL_CALL (vcpu) - sigp external call; source cpu in parm
2128 KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm) - compound value to indicate an
2129 I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel);
2130 I/O interruption parameters in parm (subchannel) and parm64 (intparm,
2131 interruption subclass)
2132 KVM_S390_MCHK (vm, vcpu) - machine check interrupt; cr 14 bits in parm,
2133 machine check interrupt code in parm64 (note that
2134 machine checks needing further payload are not
2135 supported by this ioctl)
2137 Note that the vcpu ioctl is asynchronous to vcpu execution.
2139 4.78 KVM_PPC_GET_HTAB_FD
2141 Capability: KVM_CAP_PPC_HTAB_FD
2142 Architectures: powerpc
2144 Parameters: Pointer to struct kvm_get_htab_fd (in)
2145 Returns: file descriptor number (>= 0) on success, -1 on error
2147 This returns a file descriptor that can be used either to read out the
2148 entries in the guest's hashed page table (HPT), or to write entries to
2149 initialize the HPT. The returned fd can only be written to if the
2150 KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and
2151 can only be read if that bit is clear. The argument struct looks like
2154 /* For KVM_PPC_GET_HTAB_FD */
2155 struct kvm_get_htab_fd {
2161 /* Values for kvm_get_htab_fd.flags */
2162 #define KVM_GET_HTAB_BOLTED_ONLY ((__u64)0x1)
2163 #define KVM_GET_HTAB_WRITE ((__u64)0x2)
2165 The `start_index' field gives the index in the HPT of the entry at
2166 which to start reading. It is ignored when writing.
2168 Reads on the fd will initially supply information about all
2169 "interesting" HPT entries. Interesting entries are those with the
2170 bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise
2171 all entries. When the end of the HPT is reached, the read() will
2172 return. If read() is called again on the fd, it will start again from
2173 the beginning of the HPT, but will only return HPT entries that have
2174 changed since they were last read.
2176 Data read or written is structured as a header (8 bytes) followed by a
2177 series of valid HPT entries (16 bytes) each. The header indicates how
2178 many valid HPT entries there are and how many invalid entries follow
2179 the valid entries. The invalid entries are not represented explicitly
2180 in the stream. The header format is:
2182 struct kvm_get_htab_header {
2188 Writes to the fd create HPT entries starting at the index given in the
2189 header; first `n_valid' valid entries with contents from the data
2190 written, then `n_invalid' invalid entries, invalidating any previously
2191 valid entries found.
2193 4.79 KVM_CREATE_DEVICE
2195 Capability: KVM_CAP_DEVICE_CTRL
2197 Parameters: struct kvm_create_device (in/out)
2198 Returns: 0 on success, -1 on error
2200 ENODEV: The device type is unknown or unsupported
2201 EEXIST: Device already created, and this type of device may not
2202 be instantiated multiple times
2204 Other error conditions may be defined by individual device types or
2205 have their standard meanings.
2207 Creates an emulated device in the kernel. The file descriptor returned
2208 in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR.
2210 If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the
2211 device type is supported (not necessarily whether it can be created
2214 Individual devices should not define flags. Attributes should be used
2215 for specifying any behavior that is not implied by the device type
2218 struct kvm_create_device {
2219 __u32 type; /* in: KVM_DEV_TYPE_xxx */
2220 __u32 fd; /* out: device handle */
2221 __u32 flags; /* in: KVM_CREATE_DEVICE_xxx */
2224 4.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR
2226 Capability: KVM_CAP_DEVICE_CTRL
2228 Parameters: struct kvm_device_attr
2229 Returns: 0 on success, -1 on error
2231 ENXIO: The group or attribute is unknown/unsupported for this device
2232 EPERM: The attribute cannot (currently) be accessed this way
2233 (e.g. read-only attribute, or attribute that only makes
2234 sense when the device is in a different state)
2236 Other error conditions may be defined by individual device types.
2238 Gets/sets a specified piece of device configuration and/or state. The
2239 semantics are device-specific. See individual device documentation in
2240 the "devices" directory. As with ONE_REG, the size of the data
2241 transferred is defined by the particular attribute.
2243 struct kvm_device_attr {
2244 __u32 flags; /* no flags currently defined */
2245 __u32 group; /* device-defined */
2246 __u64 attr; /* group-defined */
2247 __u64 addr; /* userspace address of attr data */
2250 4.81 KVM_HAS_DEVICE_ATTR
2252 Capability: KVM_CAP_DEVICE_CTRL
2254 Parameters: struct kvm_device_attr
2255 Returns: 0 on success, -1 on error
2257 ENXIO: The group or attribute is unknown/unsupported for this device
2259 Tests whether a device supports a particular attribute. A successful
2260 return indicates the attribute is implemented. It does not necessarily
2261 indicate that the attribute can be read or written in the device's
2262 current state. "addr" is ignored.
2264 4.77 KVM_ARM_VCPU_INIT
2269 Parameters: struct struct kvm_vcpu_init (in)
2270 Returns: 0 on success; -1 on error
2272 Â EINVAL: Â Â Â the target is unknown, or the combination of features is invalid.
2273 Â ENOENT: Â Â Â a features bit specified is unknown.
2275 This tells KVM what type of CPU to present to the guest, and what
2276 optional features it should have. Â This will cause a reset of the cpu
2277 registers to their initial values. Â If this is not called, KVM_RUN will
2278 return ENOEXEC for that vcpu.
2280 Note that because some registers reflect machine topology, all vcpus
2281 should be created before this ioctl is invoked.
2284 - KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state.
2285 Depends on KVM_CAP_ARM_PSCI.
2288 4.78 KVM_GET_REG_LIST
2293 Parameters: struct kvm_reg_list (in/out)
2294 Returns: 0 on success; -1 on error
2296 Â E2BIG: Â Â Â Â the reg index list is too big to fit in the array specified by
2297 Â Â Â Â Â Â Â Â Â Â Â Â the user (the number required will be written into n).
2299 struct kvm_reg_list {
2300 __u64 n; /* number of registers in reg[] */
2304 This ioctl returns the guest registers that are supported for the
2305 KVM_GET_ONE_REG/KVM_SET_ONE_REG calls.
2308 4.80 KVM_ARM_SET_DEVICE_ADDR
2310 Capability: KVM_CAP_ARM_SET_DEVICE_ADDR
2313 Parameters: struct kvm_arm_device_address (in)
2314 Returns: 0 on success, -1 on error
2316 ENODEV: The device id is unknown
2317 ENXIO: Device not supported on current system
2318 EEXIST: Address already set
2319 E2BIG: Address outside guest physical address space
2320 EBUSY: Address overlaps with other device range
2322 struct kvm_arm_device_addr {
2327 Specify a device address in the guest's physical address space where guests
2328 can access emulated or directly exposed devices, which the host kernel needs
2329 to know about. The id field is an architecture specific identifier for a
2332 ARM divides the id field into two parts, a device id and an address type id
2333 specific to the individual device.
2335 Â bits: | 63 ... 32 | 31 ... 16 | 15 ... 0 |
2336 field: | 0x00000000 | device id | addr type id |
2338 ARM currently only require this when using the in-kernel GIC support for the
2339 hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2 as the device id. When
2340 setting the base address for the guest's mapping of the VGIC virtual CPU
2341 and distributor interface, the ioctl must be called after calling
2342 KVM_CREATE_IRQCHIP, but before calling KVM_RUN on any of the VCPUs. Calling
2343 this ioctl twice for any of the base addresses will return -EEXIST.
2345 4.82 KVM_PPC_RTAS_DEFINE_TOKEN
2347 Capability: KVM_CAP_PPC_RTAS
2350 Parameters: struct kvm_rtas_token_args
2351 Returns: 0 on success, -1 on error
2353 Defines a token value for a RTAS (Run Time Abstraction Services)
2354 service in order to allow it to be handled in the kernel. The
2355 argument struct gives the name of the service, which must be the name
2356 of a service that has a kernel-side implementation. If the token
2357 value is non-zero, it will be associated with that service, and
2358 subsequent RTAS calls by the guest specifying that token will be
2359 handled by the kernel. If the token value is 0, then any token
2360 associated with the service will be forgotten, and subsequent RTAS
2361 calls by the guest for that service will be passed to userspace to be
2365 5. The kvm_run structure
2366 ------------------------
2368 Application code obtains a pointer to the kvm_run structure by
2369 mmap()ing a vcpu fd. From that point, application code can control
2370 execution by changing fields in kvm_run prior to calling the KVM_RUN
2371 ioctl, and obtain information about the reason KVM_RUN returned by
2372 looking up structure members.
2376 __u8 request_interrupt_window;
2378 Request that KVM_RUN return when it becomes possible to inject external
2379 interrupts into the guest. Useful in conjunction with KVM_INTERRUPT.
2386 When KVM_RUN has returned successfully (return value 0), this informs
2387 application code why KVM_RUN has returned. Allowable values for this
2388 field are detailed below.
2390 __u8 ready_for_interrupt_injection;
2392 If request_interrupt_window has been specified, this field indicates
2393 an interrupt can be injected now with KVM_INTERRUPT.
2397 The value of the current interrupt flag. Only valid if in-kernel
2398 local APIC is not used.
2402 /* in (pre_kvm_run), out (post_kvm_run) */
2405 The value of the cr8 register. Only valid if in-kernel local APIC is
2406 not used. Both input and output.
2410 The value of the APIC BASE msr. Only valid if in-kernel local
2411 APIC is not used. Both input and output.
2414 /* KVM_EXIT_UNKNOWN */
2416 __u64 hardware_exit_reason;
2419 If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown
2420 reasons. Further architecture-specific information is available in
2421 hardware_exit_reason.
2423 /* KVM_EXIT_FAIL_ENTRY */
2425 __u64 hardware_entry_failure_reason;
2428 If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due
2429 to unknown reasons. Further architecture-specific information is
2430 available in hardware_entry_failure_reason.
2432 /* KVM_EXIT_EXCEPTION */
2442 #define KVM_EXIT_IO_IN 0
2443 #define KVM_EXIT_IO_OUT 1
2445 __u8 size; /* bytes */
2448 __u64 data_offset; /* relative to kvm_run start */
2451 If exit_reason is KVM_EXIT_IO, then the vcpu has
2452 executed a port I/O instruction which could not be satisfied by kvm.
2453 data_offset describes where the data is located (KVM_EXIT_IO_OUT) or
2454 where kvm expects application code to place the data for the next
2455 KVM_RUN invocation (KVM_EXIT_IO_IN). Data format is a packed array.
2458 struct kvm_debug_exit_arch arch;
2471 If exit_reason is KVM_EXIT_MMIO, then the vcpu has
2472 executed a memory-mapped I/O instruction which could not be satisfied
2473 by kvm. The 'data' member contains the written data if 'is_write' is
2474 true, and should be filled by application code otherwise.
2476 NOTE: For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_DCR,
2477 KVM_EXIT_PAPR and KVM_EXIT_EPR the corresponding
2478 operations are complete (and guest state is consistent) only after userspace
2479 has re-entered the kernel with KVM_RUN. The kernel side will first finish
2480 incomplete operations and then check for pending signals. Userspace
2481 can re-enter the guest with an unmasked signal pending to complete
2484 /* KVM_EXIT_HYPERCALL */
2493 Unused. This was once used for 'hypercall to userspace'. To implement
2494 such functionality, use KVM_EXIT_IO (x86) or KVM_EXIT_MMIO (all except s390).
2495 Note KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO.
2497 /* KVM_EXIT_TPR_ACCESS */
2504 To be documented (KVM_TPR_ACCESS_REPORTING).
2506 /* KVM_EXIT_S390_SIEIC */
2509 __u64 mask; /* psw upper half */
2510 __u64 addr; /* psw lower half */
2517 /* KVM_EXIT_S390_RESET */
2518 #define KVM_S390_RESET_POR 1
2519 #define KVM_S390_RESET_CLEAR 2
2520 #define KVM_S390_RESET_SUBSYSTEM 4
2521 #define KVM_S390_RESET_CPU_INIT 8
2522 #define KVM_S390_RESET_IPL 16
2523 __u64 s390_reset_flags;
2527 /* KVM_EXIT_S390_UCONTROL */
2529 __u64 trans_exc_code;
2533 s390 specific. A page fault has occurred for a user controlled virtual
2534 machine (KVM_VM_S390_UNCONTROL) on it's host page table that cannot be
2535 resolved by the kernel.
2536 The program code and the translation exception code that were placed
2537 in the cpu's lowcore are presented here as defined by the z Architecture
2538 Principles of Operation Book in the Chapter for Dynamic Address Translation
2555 MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch
2556 hypercalls and exit with this exit struct that contains all the guest gprs.
2558 If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall.
2559 Userspace can now handle the hypercall and when it's done modify the gprs as
2560 necessary. Upon guest entry all guest GPRs will then be replaced by the values
2563 /* KVM_EXIT_PAPR_HCALL */
2570 This is used on 64-bit PowerPC when emulating a pSeries partition,
2571 e.g. with the 'pseries' machine type in qemu. It occurs when the
2572 guest does a hypercall using the 'sc 1' instruction. The 'nr' field
2573 contains the hypercall number (from the guest R3), and 'args' contains
2574 the arguments (from the guest R4 - R12). Userspace should put the
2575 return code in 'ret' and any extra returned values in args[].
2576 The possible hypercalls are defined in the Power Architecture Platform
2577 Requirements (PAPR) document available from www.power.org (free
2578 developer registration required to access it).
2580 /* KVM_EXIT_S390_TSCH */
2582 __u16 subchannel_id;
2583 __u16 subchannel_nr;
2590 s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled
2591 and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O
2592 interrupt for the target subchannel has been dequeued and subchannel_id,
2593 subchannel_nr, io_int_parm and io_int_word contain the parameters for that
2594 interrupt. ipb is needed for instruction parameter decoding.
2601 On FSL BookE PowerPC chips, the interrupt controller has a fast patch
2602 interrupt acknowledge path to the core. When the core successfully
2603 delivers an interrupt, it automatically populates the EPR register with
2604 the interrupt vector number and acknowledges the interrupt inside
2605 the interrupt controller.
2607 In case the interrupt controller lives in user space, we need to do
2608 the interrupt acknowledge cycle through it to fetch the next to be
2609 delivered interrupt vector using this exit.
2611 It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an
2612 external interrupt has just been delivered into the guest. User space
2613 should put the acknowledged interrupt vector into the 'epr' field.
2615 /* Fix the size of the union. */
2620 * shared registers between kvm and userspace.
2621 * kvm_valid_regs specifies the register classes set by the host
2622 * kvm_dirty_regs specified the register classes dirtied by userspace
2623 * struct kvm_sync_regs is architecture specific, as well as the
2624 * bits for kvm_valid_regs and kvm_dirty_regs
2626 __u64 kvm_valid_regs;
2627 __u64 kvm_dirty_regs;
2629 struct kvm_sync_regs regs;
2633 If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access
2634 certain guest registers without having to call SET/GET_*REGS. Thus we can
2635 avoid some system call overhead if userspace has to handle the exit.
2636 Userspace can query the validity of the structure by checking
2637 kvm_valid_regs for specific bits. These bits are architecture specific
2638 and usually define the validity of a groups of registers. (e.g. one bit
2639 for general purpose registers)
2644 6. Capabilities that can be enabled
2645 -----------------------------------
2647 There are certain capabilities that change the behavior of the virtual CPU when
2648 enabled. To enable them, please see section 4.37. Below you can find a list of
2649 capabilities and what their effect on the vCPU is when enabling them.
2651 The following information is provided along with the description:
2653 Architectures: which instruction set architectures provide this ioctl.
2654 x86 includes both i386 and x86_64.
2656 Parameters: what parameters are accepted by the capability.
2658 Returns: the return value. General error numbers (EBADF, ENOMEM, EINVAL)
2659 are not detailed, but errors with specific meanings are.
2666 Returns: 0 on success; -1 on error
2668 This capability enables interception of OSI hypercalls that otherwise would
2669 be treated as normal system calls to be injected into the guest. OSI hypercalls
2670 were invented by Mac-on-Linux to have a standardized communication mechanism
2671 between the guest and the host.
2673 When this capability is enabled, KVM_EXIT_OSI can occur.
2676 6.2 KVM_CAP_PPC_PAPR
2680 Returns: 0 on success; -1 on error
2682 This capability enables interception of PAPR hypercalls. PAPR hypercalls are
2683 done using the hypercall instruction "sc 1".
2685 It also sets the guest privilege level to "supervisor" mode. Usually the guest
2686 runs in "hypervisor" privilege mode with a few missing features.
2688 In addition to the above, it changes the semantics of SDR1. In this mode, the
2689 HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the
2690 HTAB invisible to the guest.
2692 When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur.
2698 Parameters: args[0] is the address of a struct kvm_config_tlb
2699 Returns: 0 on success; -1 on error
2701 struct kvm_config_tlb {
2708 Configures the virtual CPU's TLB array, establishing a shared memory area
2709 between userspace and KVM. The "params" and "array" fields are userspace
2710 addresses of mmu-type-specific data structures. The "array_len" field is an
2711 safety mechanism, and should be set to the size in bytes of the memory that
2712 userspace has reserved for the array. It must be at least the size dictated
2713 by "mmu_type" and "params".
2715 While KVM_RUN is active, the shared region is under control of KVM. Its
2716 contents are undefined, and any modification by userspace results in
2717 boundedly undefined behavior.
2719 On return from KVM_RUN, the shared region will reflect the current state of
2720 the guest's TLB. If userspace makes any changes, it must call KVM_DIRTY_TLB
2721 to tell KVM which entries have been changed, prior to calling KVM_RUN again
2724 For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV:
2725 - The "params" field is of type "struct kvm_book3e_206_tlb_params".
2726 - The "array" field points to an array of type "struct
2727 kvm_book3e_206_tlb_entry".
2728 - The array consists of all entries in the first TLB, followed by all
2729 entries in the second TLB.
2730 - Within a TLB, entries are ordered first by increasing set number. Within a
2731 set, entries are ordered by way (increasing ESEL).
2732 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1)
2733 where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value.
2734 - The tsize field of mas1 shall be set to 4K on TLB0, even though the
2735 hardware ignores this value for TLB0.
2737 6.4 KVM_CAP_S390_CSS_SUPPORT
2741 Returns: 0 on success; -1 on error
2743 This capability enables support for handling of channel I/O instructions.
2745 TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are
2746 handled in-kernel, while the other I/O instructions are passed to userspace.
2748 When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST
2749 SUBCHANNEL intercepts.
2754 Parameters: args[0] defines whether the proxy facility is active
2755 Returns: 0 on success; -1 on error
2757 This capability enables or disables the delivery of interrupts through the
2758 external proxy facility.
2760 When enabled (args[0] != 0), every time the guest gets an external interrupt
2761 delivered, it automatically exits into user space with a KVM_EXIT_EPR exit
2762 to receive the topmost interrupt vector.
2764 When disabled (args[0] == 0), behavior is as if this facility is unsupported.
2766 When this capability is enabled, KVM_EXIT_EPR can occur.
2768 6.6 KVM_CAP_IRQ_MPIC
2771 Parameters: args[0] is the MPIC device fd
2772 args[1] is the MPIC CPU number for this vcpu
2774 This capability connects the vcpu to an in-kernel MPIC device.
2776 6.7 KVM_CAP_IRQ_XICS
2779 Parameters: args[0] is the XICS device fd
2780 args[1] is the XICS CPU number (server ID) for this vcpu
2782 This capability connects the vcpu to an in-kernel XICS device.