1 The Definitive KVM (Kernel-based Virtual Machine) API Documentation
2 ===================================================================
7 The kvm API is a set of ioctls that are issued to control various aspects
8 of a virtual machine. The ioctls belong to three classes
10 - System ioctls: These query and set global attributes which affect the
11 whole kvm subsystem. In addition a system ioctl is used to create
14 - VM ioctls: These query and set attributes that affect an entire virtual
15 machine, for example memory layout. In addition a VM ioctl is used to
16 create virtual cpus (vcpus).
18 Only run VM ioctls from the same process (address space) that was used
21 - vcpu ioctls: These query and set attributes that control the operation
22 of a single virtual cpu.
24 Only run vcpu ioctls from the same thread that was used to create the
31 The kvm API is centered around file descriptors. An initial
32 open("/dev/kvm") obtains a handle to the kvm subsystem; this handle
33 can be used to issue system ioctls. A KVM_CREATE_VM ioctl on this
34 handle will create a VM file descriptor which can be used to issue VM
35 ioctls. A KVM_CREATE_VCPU ioctl on a VM fd will create a virtual cpu
36 and return a file descriptor pointing to it. Finally, ioctls on a vcpu
37 fd can be used to control the vcpu, including the important task of
38 actually running guest code.
40 In general file descriptors can be migrated among processes by means
41 of fork() and the SCM_RIGHTS facility of unix domain socket. These
42 kinds of tricks are explicitly not supported by kvm. While they will
43 not cause harm to the host, their actual behavior is not guaranteed by
44 the API. The only supported use is one virtual machine per process,
45 and one vcpu per thread.
51 As of Linux 2.6.22, the KVM ABI has been stabilized: no backward
52 incompatible change are allowed. However, there is an extension
53 facility that allows backward-compatible extensions to the API to be
56 The extension mechanism is not based on the Linux version number.
57 Instead, kvm defines extension identifiers and a facility to query
58 whether a particular extension identifier is available. If it is, a
59 set of ioctls is available for application use.
65 This section describes ioctls that can be used to control kvm guests.
66 For each ioctl, the following information is provided along with a
69 Capability: which KVM extension provides this ioctl. Can be 'basic',
70 which means that is will be provided by any kernel that supports
71 API version 12 (see section 4.1), a KVM_CAP_xyz constant, which
72 means availability needs to be checked with KVM_CHECK_EXTENSION
73 (see section 4.4), or 'none' which means that while not all kernels
74 support this ioctl, there's no capability bit to check its
75 availability: for kernels that don't support the ioctl,
76 the ioctl returns -ENOTTY.
78 Architectures: which instruction set architectures provide this ioctl.
79 x86 includes both i386 and x86_64.
81 Type: system, vm, or vcpu.
83 Parameters: what parameters are accepted by the ioctl.
85 Returns: the return value. General error numbers (EBADF, ENOMEM, EINVAL)
86 are not detailed, but errors with specific meanings are.
89 4.1 KVM_GET_API_VERSION
95 Returns: the constant KVM_API_VERSION (=12)
97 This identifies the API version as the stable kvm API. It is not
98 expected that this number will change. However, Linux 2.6.20 and
99 2.6.21 report earlier versions; these are not documented and not
100 supported. Applications should refuse to run if KVM_GET_API_VERSION
101 returns a value other than 12. If this check passes, all ioctls
102 described as 'basic' will be available.
110 Parameters: machine type identifier (KVM_VM_*)
111 Returns: a VM fd that can be used to control the new virtual machine.
113 The new VM has no virtual cpus and no memory. An mmap() of a VM fd
114 will access the virtual machine's physical address space; offset zero
115 corresponds to guest physical address zero. Use of mmap() on a VM fd
116 is discouraged if userspace memory allocation (KVM_CAP_USER_MEMORY) is
118 You most certainly want to use 0 as machine type.
120 In order to create user controlled virtual machines on S390, check
121 KVM_CAP_S390_UCONTROL and use the flag KVM_VM_S390_UCONTROL as
122 privileged user (CAP_SYS_ADMIN).
125 4.3 KVM_GET_MSR_INDEX_LIST
130 Parameters: struct kvm_msr_list (in/out)
131 Returns: 0 on success; -1 on error
133 E2BIG: the msr index list is to be to fit in the array specified by
136 struct kvm_msr_list {
137 __u32 nmsrs; /* number of msrs in entries */
141 This ioctl returns the guest msrs that are supported. The list varies
142 by kvm version and host processor, but does not change otherwise. The
143 user fills in the size of the indices array in nmsrs, and in return
144 kvm adjusts nmsrs to reflect the actual number of msrs and fills in
145 the indices array with their numbers.
147 Note: if kvm indicates supports MCE (KVM_CAP_MCE), then the MCE bank MSRs are
148 not returned in the MSR list, as different vcpus can have a different number
149 of banks, as set via the KVM_X86_SETUP_MCE ioctl.
152 4.4 KVM_CHECK_EXTENSION
154 Capability: basic, KVM_CAP_CHECK_EXTENSION_VM for vm ioctl
156 Type: system ioctl, vm ioctl
157 Parameters: extension identifier (KVM_CAP_*)
158 Returns: 0 if unsupported; 1 (or some other positive integer) if supported
160 The API allows the application to query about extensions to the core
161 kvm API. Userspace passes an extension identifier (an integer) and
162 receives an integer that describes the extension availability.
163 Generally 0 means no and 1 means yes, but some extensions may report
164 additional information in the integer return value.
166 Based on their initialization different VMs may have different capabilities.
167 It is thus encouraged to use the vm ioctl to query for capabilities (available
168 with KVM_CAP_CHECK_EXTENSION_VM on the vm fd)
170 4.5 KVM_GET_VCPU_MMAP_SIZE
176 Returns: size of vcpu mmap area, in bytes
178 The KVM_RUN ioctl (cf.) communicates with userspace via a shared
179 memory region. This ioctl returns the size of that region. See the
180 KVM_RUN documentation for details.
183 4.6 KVM_SET_MEMORY_REGION
188 Parameters: struct kvm_memory_region (in)
189 Returns: 0 on success, -1 on error
191 This ioctl is obsolete and has been removed.
199 Parameters: vcpu id (apic id on x86)
200 Returns: vcpu fd on success, -1 on error
202 This API adds a vcpu to a virtual machine. No more than max_vcpus may be added.
203 The vcpu id is an integer in the range [0, max_vcpu_id).
205 The recommended max_vcpus value can be retrieved using the KVM_CAP_NR_VCPUS of
206 the KVM_CHECK_EXTENSION ioctl() at run-time.
207 The maximum possible value for max_vcpus can be retrieved using the
208 KVM_CAP_MAX_VCPUS of the KVM_CHECK_EXTENSION ioctl() at run-time.
210 If the KVM_CAP_NR_VCPUS does not exist, you should assume that max_vcpus is 4
212 If the KVM_CAP_MAX_VCPUS does not exist, you should assume that max_vcpus is
213 same as the value returned from KVM_CAP_NR_VCPUS.
215 The maximum possible value for max_vcpu_id can be retrieved using the
216 KVM_CAP_MAX_VCPU_ID of the KVM_CHECK_EXTENSION ioctl() at run-time.
218 If the KVM_CAP_MAX_VCPU_ID does not exist, you should assume that max_vcpu_id
219 is the same as the value returned from KVM_CAP_MAX_VCPUS.
221 On powerpc using book3s_hv mode, the vcpus are mapped onto virtual
222 threads in one or more virtual CPU cores. (This is because the
223 hardware requires all the hardware threads in a CPU core to be in the
224 same partition.) The KVM_CAP_PPC_SMT capability indicates the number
225 of vcpus per virtual core (vcore). The vcore id is obtained by
226 dividing the vcpu id by the number of vcpus per vcore. The vcpus in a
227 given vcore will always be in the same physical core as each other
228 (though that might be a different physical core from time to time).
229 Userspace can control the threading (SMT) mode of the guest by its
230 allocation of vcpu ids. For example, if userspace wants
231 single-threaded guest vcpus, it should make all vcpu ids be a multiple
232 of the number of vcpus per vcore.
234 For virtual cpus that have been created with S390 user controlled virtual
235 machines, the resulting vcpu fd can be memory mapped at page offset
236 KVM_S390_SIE_PAGE_OFFSET in order to obtain a memory map of the virtual
237 cpu's hardware control block.
240 4.8 KVM_GET_DIRTY_LOG (vm ioctl)
245 Parameters: struct kvm_dirty_log (in/out)
246 Returns: 0 on success, -1 on error
248 /* for KVM_GET_DIRTY_LOG */
249 struct kvm_dirty_log {
253 void __user *dirty_bitmap; /* one bit per page */
258 Given a memory slot, return a bitmap containing any pages dirtied
259 since the last call to this ioctl. Bit 0 is the first page in the
260 memory slot. Ensure the entire structure is cleared to avoid padding
263 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 specifies
264 the address space for which you want to return the dirty bitmap.
265 They must be less than the value that KVM_CHECK_EXTENSION returns for
266 the KVM_CAP_MULTI_ADDRESS_SPACE capability.
269 4.9 KVM_SET_MEMORY_ALIAS
274 Parameters: struct kvm_memory_alias (in)
275 Returns: 0 (success), -1 (error)
277 This ioctl is obsolete and has been removed.
286 Returns: 0 on success, -1 on error
288 EINTR: an unmasked signal is pending
290 This ioctl is used to run a guest virtual cpu. While there are no
291 explicit parameters, there is an implicit parameter block that can be
292 obtained by mmap()ing the vcpu fd at offset 0, with the size given by
293 KVM_GET_VCPU_MMAP_SIZE. The parameter block is formatted as a 'struct
294 kvm_run' (see below).
300 Architectures: all except ARM, arm64
302 Parameters: struct kvm_regs (out)
303 Returns: 0 on success, -1 on error
305 Reads the general purpose registers from the vcpu.
309 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
310 __u64 rax, rbx, rcx, rdx;
311 __u64 rsi, rdi, rsp, rbp;
312 __u64 r8, r9, r10, r11;
313 __u64 r12, r13, r14, r15;
319 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
330 Architectures: all except ARM, arm64
332 Parameters: struct kvm_regs (in)
333 Returns: 0 on success, -1 on error
335 Writes the general purpose registers into the vcpu.
337 See KVM_GET_REGS for the data structure.
343 Architectures: x86, ppc
345 Parameters: struct kvm_sregs (out)
346 Returns: 0 on success, -1 on error
348 Reads special registers from the vcpu.
352 struct kvm_segment cs, ds, es, fs, gs, ss;
353 struct kvm_segment tr, ldt;
354 struct kvm_dtable gdt, idt;
355 __u64 cr0, cr2, cr3, cr4, cr8;
358 __u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64];
361 /* ppc -- see arch/powerpc/include/uapi/asm/kvm.h */
363 interrupt_bitmap is a bitmap of pending external interrupts. At most
364 one bit may be set. This interrupt has been acknowledged by the APIC
365 but not yet injected into the cpu core.
371 Architectures: x86, ppc
373 Parameters: struct kvm_sregs (in)
374 Returns: 0 on success, -1 on error
376 Writes special registers into the vcpu. See KVM_GET_SREGS for the
385 Parameters: struct kvm_translation (in/out)
386 Returns: 0 on success, -1 on error
388 Translates a virtual address according to the vcpu's current address
391 struct kvm_translation {
393 __u64 linear_address;
396 __u64 physical_address;
407 Architectures: x86, ppc, mips
409 Parameters: struct kvm_interrupt (in)
410 Returns: 0 on success, negative on failure.
412 Queues a hardware interrupt vector to be injected.
414 /* for KVM_INTERRUPT */
415 struct kvm_interrupt {
422 Returns: 0 on success,
423 -EEXIST if an interrupt is already enqueued
424 -EINVAL the the irq number is invalid
425 -ENXIO if the PIC is in the kernel
426 -EFAULT if the pointer is invalid
428 Note 'irq' is an interrupt vector, not an interrupt pin or line. This
429 ioctl is useful if the in-kernel PIC is not used.
433 Queues an external interrupt to be injected. This ioctl is overleaded
434 with 3 different irq values:
438 This injects an edge type external interrupt into the guest once it's ready
439 to receive interrupts. When injected, the interrupt is done.
441 b) KVM_INTERRUPT_UNSET
443 This unsets any pending interrupt.
445 Only available with KVM_CAP_PPC_UNSET_IRQ.
447 c) KVM_INTERRUPT_SET_LEVEL
449 This injects a level type external interrupt into the guest context. The
450 interrupt stays pending until a specific ioctl with KVM_INTERRUPT_UNSET
453 Only available with KVM_CAP_PPC_IRQ_LEVEL.
455 Note that any value for 'irq' other than the ones stated above is invalid
456 and incurs unexpected behavior.
460 Queues an external interrupt to be injected into the virtual CPU. A negative
461 interrupt number dequeues the interrupt.
472 Support for this has been removed. Use KVM_SET_GUEST_DEBUG instead.
480 Parameters: struct kvm_msrs (in/out)
481 Returns: 0 on success, -1 on error
483 Reads model-specific registers from the vcpu. Supported msr indices can
484 be obtained using KVM_GET_MSR_INDEX_LIST.
487 __u32 nmsrs; /* number of msrs in entries */
490 struct kvm_msr_entry entries[0];
493 struct kvm_msr_entry {
499 Application code should set the 'nmsrs' member (which indicates the
500 size of the entries array) and the 'index' member of each array entry.
501 kvm will fill in the 'data' member.
509 Parameters: struct kvm_msrs (in)
510 Returns: 0 on success, -1 on error
512 Writes model-specific registers to the vcpu. See KVM_GET_MSRS for the
515 Application code should set the 'nmsrs' member (which indicates the
516 size of the entries array), and the 'index' and 'data' members of each
525 Parameters: struct kvm_cpuid (in)
526 Returns: 0 on success, -1 on error
528 Defines the vcpu responses to the cpuid instruction. Applications
529 should use the KVM_SET_CPUID2 ioctl if available.
532 struct kvm_cpuid_entry {
541 /* for KVM_SET_CPUID */
545 struct kvm_cpuid_entry entries[0];
549 4.21 KVM_SET_SIGNAL_MASK
554 Parameters: struct kvm_signal_mask (in)
555 Returns: 0 on success, -1 on error
557 Defines which signals are blocked during execution of KVM_RUN. This
558 signal mask temporarily overrides the threads signal mask. Any
559 unblocked signal received (except SIGKILL and SIGSTOP, which retain
560 their traditional behaviour) will cause KVM_RUN to return with -EINTR.
562 Note the signal will only be delivered if not blocked by the original
565 /* for KVM_SET_SIGNAL_MASK */
566 struct kvm_signal_mask {
577 Parameters: struct kvm_fpu (out)
578 Returns: 0 on success, -1 on error
580 Reads the floating point state from the vcpu.
582 /* for KVM_GET_FPU and KVM_SET_FPU */
587 __u8 ftwx; /* in fxsave format */
603 Parameters: struct kvm_fpu (in)
604 Returns: 0 on success, -1 on error
606 Writes the floating point state to the vcpu.
608 /* for KVM_GET_FPU and KVM_SET_FPU */
613 __u8 ftwx; /* in fxsave format */
624 4.24 KVM_CREATE_IRQCHIP
626 Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQCHIP (s390)
627 Architectures: x86, ARM, arm64, s390
630 Returns: 0 on success, -1 on error
632 Creates an interrupt controller model in the kernel.
633 On x86, creates a virtual ioapic, a virtual PIC (two PICs, nested), and sets up
634 future vcpus to have a local APIC. IRQ routing for GSIs 0-15 is set to both
635 PIC and IOAPIC; GSI 16-23 only go to the IOAPIC.
636 On ARM/arm64, a GICv2 is created. Any other GIC versions require the usage of
637 KVM_CREATE_DEVICE, which also supports creating a GICv2. Using
638 KVM_CREATE_DEVICE is preferred over KVM_CREATE_IRQCHIP for GICv2.
639 On s390, a dummy irq routing table is created.
641 Note that on s390 the KVM_CAP_S390_IRQCHIP vm capability needs to be enabled
642 before KVM_CREATE_IRQCHIP can be used.
647 Capability: KVM_CAP_IRQCHIP
648 Architectures: x86, arm, arm64
650 Parameters: struct kvm_irq_level
651 Returns: 0 on success, -1 on error
653 Sets the level of a GSI input to the interrupt controller model in the kernel.
654 On some architectures it is required that an interrupt controller model has
655 been previously created with KVM_CREATE_IRQCHIP. Note that edge-triggered
656 interrupts require the level to be set to 1 and then back to 0.
658 On real hardware, interrupt pins can be active-low or active-high. This
659 does not matter for the level field of struct kvm_irq_level: 1 always
660 means active (asserted), 0 means inactive (deasserted).
662 x86 allows the operating system to program the interrupt polarity
663 (active-low/active-high) for level-triggered interrupts, and KVM used
664 to consider the polarity. However, due to bitrot in the handling of
665 active-low interrupts, the above convention is now valid on x86 too.
666 This is signaled by KVM_CAP_X86_IOAPIC_POLARITY_IGNORED. Userspace
667 should not present interrupts to the guest as active-low unless this
668 capability is present (or unless it is not using the in-kernel irqchip,
672 ARM/arm64 can signal an interrupt either at the CPU level, or at the
673 in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to
674 use PPIs designated for specific cpus. The irq field is interpreted
677 Â bits: | 31 ... 24 | 23 ... 16 | 15 ... 0 |
678 field: | irq_type | vcpu_index | irq_id |
680 The irq_type field has the following values:
681 - irq_type[0]: out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ
682 - irq_type[1]: in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.)
683 (the vcpu_index field is ignored)
684 - irq_type[2]: in-kernel GIC: PPI, irq_id between 16 and 31 (incl.)
686 (The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs)
688 In both cases, level is used to assert/deassert the line.
690 struct kvm_irq_level {
693 __s32 status; /* not used for KVM_IRQ_LEVEL */
695 __u32 level; /* 0 or 1 */
701 Capability: KVM_CAP_IRQCHIP
704 Parameters: struct kvm_irqchip (in/out)
705 Returns: 0 on success, -1 on error
707 Reads the state of a kernel interrupt controller created with
708 KVM_CREATE_IRQCHIP into a buffer provided by the caller.
711 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
714 char dummy[512]; /* reserving space */
715 struct kvm_pic_state pic;
716 struct kvm_ioapic_state ioapic;
723 Capability: KVM_CAP_IRQCHIP
726 Parameters: struct kvm_irqchip (in)
727 Returns: 0 on success, -1 on error
729 Sets the state of a kernel interrupt controller created with
730 KVM_CREATE_IRQCHIP from a buffer provided by the caller.
733 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
736 char dummy[512]; /* reserving space */
737 struct kvm_pic_state pic;
738 struct kvm_ioapic_state ioapic;
743 4.28 KVM_XEN_HVM_CONFIG
745 Capability: KVM_CAP_XEN_HVM
748 Parameters: struct kvm_xen_hvm_config (in)
749 Returns: 0 on success, -1 on error
751 Sets the MSR that the Xen HVM guest uses to initialize its hypercall
752 page, and provides the starting address and size of the hypercall
753 blobs in userspace. When the guest writes the MSR, kvm copies one
754 page of a blob (32- or 64-bit, depending on the vcpu mode) to guest
757 struct kvm_xen_hvm_config {
770 Capability: KVM_CAP_ADJUST_CLOCK
773 Parameters: struct kvm_clock_data (out)
774 Returns: 0 on success, -1 on error
776 Gets the current timestamp of kvmclock as seen by the current guest. In
777 conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios
780 struct kvm_clock_data {
781 __u64 clock; /* kvmclock current value */
789 Capability: KVM_CAP_ADJUST_CLOCK
792 Parameters: struct kvm_clock_data (in)
793 Returns: 0 on success, -1 on error
795 Sets the current timestamp of kvmclock to the value specified in its parameter.
796 In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios
799 struct kvm_clock_data {
800 __u64 clock; /* kvmclock current value */
806 4.31 KVM_GET_VCPU_EVENTS
808 Capability: KVM_CAP_VCPU_EVENTS
809 Extended by: KVM_CAP_INTR_SHADOW
812 Parameters: struct kvm_vcpu_event (out)
813 Returns: 0 on success, -1 on error
815 Gets currently pending exceptions, interrupts, and NMIs as well as related
818 struct kvm_vcpu_events {
848 Only two fields are defined in the flags field:
850 - KVM_VCPUEVENT_VALID_SHADOW may be set in the flags field to signal that
851 interrupt.shadow contains a valid state.
853 - KVM_VCPUEVENT_VALID_SMM may be set in the flags field to signal that
854 smi contains a valid state.
856 4.32 KVM_SET_VCPU_EVENTS
858 Capability: KVM_CAP_VCPU_EVENTS
859 Extended by: KVM_CAP_INTR_SHADOW
862 Parameters: struct kvm_vcpu_event (in)
863 Returns: 0 on success, -1 on error
865 Set pending exceptions, interrupts, and NMIs as well as related states of the
868 See KVM_GET_VCPU_EVENTS for the data structure.
870 Fields that may be modified asynchronously by running VCPUs can be excluded
871 from the update. These fields are nmi.pending, sipi_vector, smi.smm,
872 smi.pending. Keep the corresponding bits in the flags field cleared to
873 suppress overwriting the current in-kernel state. The bits are:
875 KVM_VCPUEVENT_VALID_NMI_PENDING - transfer nmi.pending to the kernel
876 KVM_VCPUEVENT_VALID_SIPI_VECTOR - transfer sipi_vector
877 KVM_VCPUEVENT_VALID_SMM - transfer the smi sub-struct.
879 If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in
880 the flags field to signal that interrupt.shadow contains a valid state and
881 shall be written into the VCPU.
883 KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available.
886 4.33 KVM_GET_DEBUGREGS
888 Capability: KVM_CAP_DEBUGREGS
891 Parameters: struct kvm_debugregs (out)
892 Returns: 0 on success, -1 on error
894 Reads debug registers from the vcpu.
896 struct kvm_debugregs {
905 4.34 KVM_SET_DEBUGREGS
907 Capability: KVM_CAP_DEBUGREGS
910 Parameters: struct kvm_debugregs (in)
911 Returns: 0 on success, -1 on error
913 Writes debug registers into the vcpu.
915 See KVM_GET_DEBUGREGS for the data structure. The flags field is unused
916 yet and must be cleared on entry.
919 4.35 KVM_SET_USER_MEMORY_REGION
921 Capability: KVM_CAP_USER_MEM
924 Parameters: struct kvm_userspace_memory_region (in)
925 Returns: 0 on success, -1 on error
927 struct kvm_userspace_memory_region {
930 __u64 guest_phys_addr;
931 __u64 memory_size; /* bytes */
932 __u64 userspace_addr; /* start of the userspace allocated memory */
935 /* for kvm_memory_region::flags */
936 #define KVM_MEM_LOG_DIRTY_PAGES (1UL << 0)
937 #define KVM_MEM_READONLY (1UL << 1)
939 This ioctl allows the user to create or modify a guest physical memory
940 slot. When changing an existing slot, it may be moved in the guest
941 physical memory space, or its flags may be modified. It may not be
942 resized. Slots may not overlap in guest physical address space.
944 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot"
945 specifies the address space which is being modified. They must be
946 less than the value that KVM_CHECK_EXTENSION returns for the
947 KVM_CAP_MULTI_ADDRESS_SPACE capability. Slots in separate address spaces
948 are unrelated; the restriction on overlapping slots only applies within
951 Memory for the region is taken starting at the address denoted by the
952 field userspace_addr, which must point at user addressable memory for
953 the entire memory slot size. Any object may back this memory, including
954 anonymous memory, ordinary files, and hugetlbfs.
956 It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr
957 be identical. This allows large pages in the guest to be backed by large
960 The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and
961 KVM_MEM_READONLY. The former can be set to instruct KVM to keep track of
962 writes to memory within the slot. See KVM_GET_DIRTY_LOG ioctl to know how to
963 use it. The latter can be set, if KVM_CAP_READONLY_MEM capability allows it,
964 to make a new slot read-only. In this case, writes to this memory will be
965 posted to userspace as KVM_EXIT_MMIO exits.
967 When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of
968 the memory region are automatically reflected into the guest. For example, an
969 mmap() that affects the region will be made visible immediately. Another
970 example is madvise(MADV_DROP).
972 It is recommended to use this API instead of the KVM_SET_MEMORY_REGION ioctl.
973 The KVM_SET_MEMORY_REGION does not allow fine grained control over memory
974 allocation and is deprecated.
977 4.36 KVM_SET_TSS_ADDR
979 Capability: KVM_CAP_SET_TSS_ADDR
982 Parameters: unsigned long tss_address (in)
983 Returns: 0 on success, -1 on error
985 This ioctl defines the physical address of a three-page region in the guest
986 physical address space. The region must be within the first 4GB of the
987 guest physical address space and must not conflict with any memory slot
988 or any mmio address. The guest may malfunction if it accesses this memory
991 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
992 because of a quirk in the virtualization implementation (see the internals
993 documentation when it pops into existence).
998 Capability: KVM_CAP_ENABLE_CAP, KVM_CAP_ENABLE_CAP_VM
999 Architectures: x86 (only KVM_CAP_ENABLE_CAP_VM),
1000 mips (only KVM_CAP_ENABLE_CAP), ppc, s390
1001 Type: vcpu ioctl, vm ioctl (with KVM_CAP_ENABLE_CAP_VM)
1002 Parameters: struct kvm_enable_cap (in)
1003 Returns: 0 on success; -1 on error
1005 +Not all extensions are enabled by default. Using this ioctl the application
1006 can enable an extension, making it available to the guest.
1008 On systems that do not support this ioctl, it always fails. On systems that
1009 do support it, it only works for extensions that are supported for enablement.
1011 To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should
1014 struct kvm_enable_cap {
1018 The capability that is supposed to get enabled.
1022 A bitfield indicating future enhancements. Has to be 0 for now.
1026 Arguments for enabling a feature. If a feature needs initial values to
1027 function properly, this is the place to put them.
1032 The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl
1033 for vm-wide capabilities.
1035 4.38 KVM_GET_MP_STATE
1037 Capability: KVM_CAP_MP_STATE
1038 Architectures: x86, s390, arm, arm64
1040 Parameters: struct kvm_mp_state (out)
1041 Returns: 0 on success; -1 on error
1043 struct kvm_mp_state {
1047 Returns the vcpu's current "multiprocessing state" (though also valid on
1048 uniprocessor guests).
1050 Possible values are:
1052 - KVM_MP_STATE_RUNNABLE: the vcpu is currently running [x86,arm/arm64]
1053 - KVM_MP_STATE_UNINITIALIZED: the vcpu is an application processor (AP)
1054 which has not yet received an INIT signal [x86]
1055 - KVM_MP_STATE_INIT_RECEIVED: the vcpu has received an INIT signal, and is
1056 now ready for a SIPI [x86]
1057 - KVM_MP_STATE_HALTED: the vcpu has executed a HLT instruction and
1058 is waiting for an interrupt [x86]
1059 - KVM_MP_STATE_SIPI_RECEIVED: the vcpu has just received a SIPI (vector
1060 accessible via KVM_GET_VCPU_EVENTS) [x86]
1061 - KVM_MP_STATE_STOPPED: the vcpu is stopped [s390,arm/arm64]
1062 - KVM_MP_STATE_CHECK_STOP: the vcpu is in a special error state [s390]
1063 - KVM_MP_STATE_OPERATING: the vcpu is operating (running or halted)
1065 - KVM_MP_STATE_LOAD: the vcpu is in a special load/startup state
1068 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1069 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1070 these architectures.
1074 The only states that are valid are KVM_MP_STATE_STOPPED and
1075 KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not.
1077 4.39 KVM_SET_MP_STATE
1079 Capability: KVM_CAP_MP_STATE
1080 Architectures: x86, s390, arm, arm64
1082 Parameters: struct kvm_mp_state (in)
1083 Returns: 0 on success; -1 on error
1085 Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for
1088 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1089 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1090 these architectures.
1094 The only states that are valid are KVM_MP_STATE_STOPPED and
1095 KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not.
1097 4.40 KVM_SET_IDENTITY_MAP_ADDR
1099 Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR
1102 Parameters: unsigned long identity (in)
1103 Returns: 0 on success, -1 on error
1105 This ioctl defines the physical address of a one-page region in the guest
1106 physical address space. The region must be within the first 4GB of the
1107 guest physical address space and must not conflict with any memory slot
1108 or any mmio address. The guest may malfunction if it accesses this memory
1111 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1112 because of a quirk in the virtualization implementation (see the internals
1113 documentation when it pops into existence).
1116 4.41 KVM_SET_BOOT_CPU_ID
1118 Capability: KVM_CAP_SET_BOOT_CPU_ID
1121 Parameters: unsigned long vcpu_id
1122 Returns: 0 on success, -1 on error
1124 Define which vcpu is the Bootstrap Processor (BSP). Values are the same
1125 as the vcpu id in KVM_CREATE_VCPU. If this ioctl is not called, the default
1131 Capability: KVM_CAP_XSAVE
1134 Parameters: struct kvm_xsave (out)
1135 Returns: 0 on success, -1 on error
1141 This ioctl would copy current vcpu's xsave struct to the userspace.
1146 Capability: KVM_CAP_XSAVE
1149 Parameters: struct kvm_xsave (in)
1150 Returns: 0 on success, -1 on error
1156 This ioctl would copy userspace's xsave struct to the kernel.
1161 Capability: KVM_CAP_XCRS
1164 Parameters: struct kvm_xcrs (out)
1165 Returns: 0 on success, -1 on error
1176 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1180 This ioctl would copy current vcpu's xcrs to the userspace.
1185 Capability: KVM_CAP_XCRS
1188 Parameters: struct kvm_xcrs (in)
1189 Returns: 0 on success, -1 on error
1200 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1204 This ioctl would set vcpu's xcr to the value userspace specified.
1207 4.46 KVM_GET_SUPPORTED_CPUID
1209 Capability: KVM_CAP_EXT_CPUID
1212 Parameters: struct kvm_cpuid2 (in/out)
1213 Returns: 0 on success, -1 on error
1218 struct kvm_cpuid_entry2 entries[0];
1221 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
1222 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1)
1223 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2)
1225 struct kvm_cpuid_entry2 {
1236 This ioctl returns x86 cpuid features which are supported by both the hardware
1237 and kvm. Userspace can use the information returned by this ioctl to
1238 construct cpuid information (for KVM_SET_CPUID2) that is consistent with
1239 hardware, kernel, and userspace capabilities, and with user requirements (for
1240 example, the user may wish to constrain cpuid to emulate older hardware,
1241 or for feature consistency across a cluster).
1243 Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure
1244 with the 'nent' field indicating the number of entries in the variable-size
1245 array 'entries'. If the number of entries is too low to describe the cpu
1246 capabilities, an error (E2BIG) is returned. If the number is too high,
1247 the 'nent' field is adjusted and an error (ENOMEM) is returned. If the
1248 number is just right, the 'nent' field is adjusted to the number of valid
1249 entries in the 'entries' array, which is then filled.
1251 The entries returned are the host cpuid as returned by the cpuid instruction,
1252 with unknown or unsupported features masked out. Some features (for example,
1253 x2apic), may not be present in the host cpu, but are exposed by kvm if it can
1254 emulate them efficiently. The fields in each entry are defined as follows:
1256 function: the eax value used to obtain the entry
1257 index: the ecx value used to obtain the entry (for entries that are
1259 flags: an OR of zero or more of the following:
1260 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
1261 if the index field is valid
1262 KVM_CPUID_FLAG_STATEFUL_FUNC:
1263 if cpuid for this function returns different values for successive
1264 invocations; there will be several entries with the same function,
1265 all with this flag set
1266 KVM_CPUID_FLAG_STATE_READ_NEXT:
1267 for KVM_CPUID_FLAG_STATEFUL_FUNC entries, set if this entry is
1268 the first entry to be read by a cpu
1269 eax, ebx, ecx, edx: the values returned by the cpuid instruction for
1270 this function/index combination
1272 The TSC deadline timer feature (CPUID leaf 1, ecx[24]) is always returned
1273 as false, since the feature depends on KVM_CREATE_IRQCHIP for local APIC
1274 support. Instead it is reported via
1276 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER)
1278 if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the
1279 feature in userspace, then you can enable the feature for KVM_SET_CPUID2.
1282 4.47 KVM_PPC_GET_PVINFO
1284 Capability: KVM_CAP_PPC_GET_PVINFO
1287 Parameters: struct kvm_ppc_pvinfo (out)
1288 Returns: 0 on success, !0 on error
1290 struct kvm_ppc_pvinfo {
1296 This ioctl fetches PV specific information that need to be passed to the guest
1297 using the device tree or other means from vm context.
1299 The hcall array defines 4 instructions that make up a hypercall.
1301 If any additional field gets added to this structure later on, a bit for that
1302 additional piece of information will be set in the flags bitmap.
1304 The flags bitmap is defined as:
1306 /* the host supports the ePAPR idle hcall
1307 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1<<0)
1309 4.48 KVM_ASSIGN_PCI_DEVICE (deprecated)
1314 Parameters: struct kvm_assigned_pci_dev (in)
1315 Returns: 0 on success, -1 on error
1317 Assigns a host PCI device to the VM.
1319 struct kvm_assigned_pci_dev {
1320 __u32 assigned_dev_id;
1330 The PCI device is specified by the triple segnr, busnr, and devfn.
1331 Identification in succeeding service requests is done via assigned_dev_id. The
1332 following flags are specified:
1334 /* Depends on KVM_CAP_IOMMU */
1335 #define KVM_DEV_ASSIGN_ENABLE_IOMMU (1 << 0)
1336 /* The following two depend on KVM_CAP_PCI_2_3 */
1337 #define KVM_DEV_ASSIGN_PCI_2_3 (1 << 1)
1338 #define KVM_DEV_ASSIGN_MASK_INTX (1 << 2)
1340 If KVM_DEV_ASSIGN_PCI_2_3 is set, the kernel will manage legacy INTx interrupts
1341 via the PCI-2.3-compliant device-level mask, thus enable IRQ sharing with other
1342 assigned devices or host devices. KVM_DEV_ASSIGN_MASK_INTX specifies the
1343 guest's view on the INTx mask, see KVM_ASSIGN_SET_INTX_MASK for details.
1345 The KVM_DEV_ASSIGN_ENABLE_IOMMU flag is a mandatory option to ensure
1346 isolation of the device. Usages not specifying this flag are deprecated.
1348 Only PCI header type 0 devices with PCI BAR resources are supported by
1349 device assignment. The user requesting this ioctl must have read/write
1350 access to the PCI sysfs resource files associated with the device.
1353 ENOTTY: kernel does not support this ioctl
1355 Other error conditions may be defined by individual device types or
1356 have their standard meanings.
1359 4.49 KVM_DEASSIGN_PCI_DEVICE (deprecated)
1364 Parameters: struct kvm_assigned_pci_dev (in)
1365 Returns: 0 on success, -1 on error
1367 Ends PCI device assignment, releasing all associated resources.
1369 See KVM_ASSIGN_PCI_DEVICE for the data structure. Only assigned_dev_id is
1370 used in kvm_assigned_pci_dev to identify the device.
1373 ENOTTY: kernel does not support this ioctl
1375 Other error conditions may be defined by individual device types or
1376 have their standard meanings.
1378 4.50 KVM_ASSIGN_DEV_IRQ (deprecated)
1380 Capability: KVM_CAP_ASSIGN_DEV_IRQ
1383 Parameters: struct kvm_assigned_irq (in)
1384 Returns: 0 on success, -1 on error
1386 Assigns an IRQ to a passed-through device.
1388 struct kvm_assigned_irq {
1389 __u32 assigned_dev_id;
1390 __u32 host_irq; /* ignored (legacy field) */
1398 The following flags are defined:
1400 #define KVM_DEV_IRQ_HOST_INTX (1 << 0)
1401 #define KVM_DEV_IRQ_HOST_MSI (1 << 1)
1402 #define KVM_DEV_IRQ_HOST_MSIX (1 << 2)
1404 #define KVM_DEV_IRQ_GUEST_INTX (1 << 8)
1405 #define KVM_DEV_IRQ_GUEST_MSI (1 << 9)
1406 #define KVM_DEV_IRQ_GUEST_MSIX (1 << 10)
1408 It is not valid to specify multiple types per host or guest IRQ. However, the
1409 IRQ type of host and guest can differ or can even be null.
1412 ENOTTY: kernel does not support this ioctl
1414 Other error conditions may be defined by individual device types or
1415 have their standard meanings.
1418 4.51 KVM_DEASSIGN_DEV_IRQ (deprecated)
1420 Capability: KVM_CAP_ASSIGN_DEV_IRQ
1423 Parameters: struct kvm_assigned_irq (in)
1424 Returns: 0 on success, -1 on error
1426 Ends an IRQ assignment to a passed-through device.
1428 See KVM_ASSIGN_DEV_IRQ for the data structure. The target device is specified
1429 by assigned_dev_id, flags must correspond to the IRQ type specified on
1430 KVM_ASSIGN_DEV_IRQ. Partial deassignment of host or guest IRQ is allowed.
1433 4.52 KVM_SET_GSI_ROUTING
1435 Capability: KVM_CAP_IRQ_ROUTING
1436 Architectures: x86 s390 arm arm64
1438 Parameters: struct kvm_irq_routing (in)
1439 Returns: 0 on success, -1 on error
1441 Sets the GSI routing table entries, overwriting any previously set entries.
1443 On arm/arm64, GSI routing has the following limitation:
1444 - GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD.
1446 struct kvm_irq_routing {
1449 struct kvm_irq_routing_entry entries[0];
1452 No flags are specified so far, the corresponding field must be set to zero.
1454 struct kvm_irq_routing_entry {
1460 struct kvm_irq_routing_irqchip irqchip;
1461 struct kvm_irq_routing_msi msi;
1462 struct kvm_irq_routing_s390_adapter adapter;
1463 struct kvm_irq_routing_hv_sint hv_sint;
1468 /* gsi routing entry types */
1469 #define KVM_IRQ_ROUTING_IRQCHIP 1
1470 #define KVM_IRQ_ROUTING_MSI 2
1471 #define KVM_IRQ_ROUTING_S390_ADAPTER 3
1472 #define KVM_IRQ_ROUTING_HV_SINT 4
1475 - KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry
1476 type, specifies that the devid field contains a valid value. The per-VM
1477 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
1478 the device ID. If this capability is not available, userspace should
1479 never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
1482 struct kvm_irq_routing_irqchip {
1487 struct kvm_irq_routing_msi {
1497 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
1498 for the device that wrote the MSI message. For PCI, this is usually a
1499 BFD identifier in the lower 16 bits.
1501 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
1502 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
1503 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
1504 address_hi must be zero.
1506 struct kvm_irq_routing_s390_adapter {
1510 __u32 summary_offset;
1514 struct kvm_irq_routing_hv_sint {
1519 4.53 KVM_ASSIGN_SET_MSIX_NR (deprecated)
1524 Parameters: struct kvm_assigned_msix_nr (in)
1525 Returns: 0 on success, -1 on error
1527 Set the number of MSI-X interrupts for an assigned device. The number is
1528 reset again by terminating the MSI-X assignment of the device via
1529 KVM_DEASSIGN_DEV_IRQ. Calling this service more than once at any earlier
1532 struct kvm_assigned_msix_nr {
1533 __u32 assigned_dev_id;
1538 #define KVM_MAX_MSIX_PER_DEV 256
1541 4.54 KVM_ASSIGN_SET_MSIX_ENTRY (deprecated)
1546 Parameters: struct kvm_assigned_msix_entry (in)
1547 Returns: 0 on success, -1 on error
1549 Specifies the routing of an MSI-X assigned device interrupt to a GSI. Setting
1550 the GSI vector to zero means disabling the interrupt.
1552 struct kvm_assigned_msix_entry {
1553 __u32 assigned_dev_id;
1555 __u16 entry; /* The index of entry in the MSI-X table */
1560 ENOTTY: kernel does not support this ioctl
1562 Other error conditions may be defined by individual device types or
1563 have their standard meanings.
1566 4.55 KVM_SET_TSC_KHZ
1568 Capability: KVM_CAP_TSC_CONTROL
1571 Parameters: virtual tsc_khz
1572 Returns: 0 on success, -1 on error
1574 Specifies the tsc frequency for the virtual machine. The unit of the
1578 4.56 KVM_GET_TSC_KHZ
1580 Capability: KVM_CAP_GET_TSC_KHZ
1584 Returns: virtual tsc-khz on success, negative value on error
1586 Returns the tsc frequency of the guest. The unit of the return value is
1587 KHz. If the host has unstable tsc this ioctl returns -EIO instead as an
1593 Capability: KVM_CAP_IRQCHIP
1596 Parameters: struct kvm_lapic_state (out)
1597 Returns: 0 on success, -1 on error
1599 #define KVM_APIC_REG_SIZE 0x400
1600 struct kvm_lapic_state {
1601 char regs[KVM_APIC_REG_SIZE];
1604 Reads the Local APIC registers and copies them into the input argument. The
1605 data format and layout are the same as documented in the architecture manual.
1607 If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is
1608 enabled, then the format of APIC_ID register depends on the APIC mode
1609 (reported by MSR_IA32_APICBASE) of its VCPU. x2APIC stores APIC ID in
1610 the APIC_ID register (bytes 32-35). xAPIC only allows an 8-bit APIC ID
1611 which is stored in bits 31-24 of the APIC register, or equivalently in
1612 byte 35 of struct kvm_lapic_state's regs field. KVM_GET_LAPIC must then
1613 be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR.
1615 If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state
1616 always uses xAPIC format.
1621 Capability: KVM_CAP_IRQCHIP
1624 Parameters: struct kvm_lapic_state (in)
1625 Returns: 0 on success, -1 on error
1627 #define KVM_APIC_REG_SIZE 0x400
1628 struct kvm_lapic_state {
1629 char regs[KVM_APIC_REG_SIZE];
1632 Copies the input argument into the Local APIC registers. The data format
1633 and layout are the same as documented in the architecture manual.
1635 The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's
1636 regs field) depends on the state of the KVM_CAP_X2APIC_API capability.
1637 See the note in KVM_GET_LAPIC.
1642 Capability: KVM_CAP_IOEVENTFD
1645 Parameters: struct kvm_ioeventfd (in)
1646 Returns: 0 on success, !0 on error
1648 This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address
1649 within the guest. A guest write in the registered address will signal the
1650 provided event instead of triggering an exit.
1652 struct kvm_ioeventfd {
1654 __u64 addr; /* legal pio/mmio address */
1655 __u32 len; /* 0, 1, 2, 4, or 8 bytes */
1661 For the special case of virtio-ccw devices on s390, the ioevent is matched
1662 to a subchannel/virtqueue tuple instead.
1664 The following flags are defined:
1666 #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch)
1667 #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio)
1668 #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign)
1669 #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \
1670 (1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify)
1672 If datamatch flag is set, the event will be signaled only if the written value
1673 to the registered address is equal to datamatch in struct kvm_ioeventfd.
1675 For virtio-ccw devices, addr contains the subchannel id and datamatch the
1678 With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and
1679 the kernel will ignore the length of guest write and may get a faster vmexit.
1680 The speedup may only apply to specific architectures, but the ioeventfd will
1685 Capability: KVM_CAP_SW_TLB
1688 Parameters: struct kvm_dirty_tlb (in)
1689 Returns: 0 on success, -1 on error
1691 struct kvm_dirty_tlb {
1696 This must be called whenever userspace has changed an entry in the shared
1697 TLB, prior to calling KVM_RUN on the associated vcpu.
1699 The "bitmap" field is the userspace address of an array. This array
1700 consists of a number of bits, equal to the total number of TLB entries as
1701 determined by the last successful call to KVM_CONFIG_TLB, rounded up to the
1702 nearest multiple of 64.
1704 Each bit corresponds to one TLB entry, ordered the same as in the shared TLB
1707 The array is little-endian: the bit 0 is the least significant bit of the
1708 first byte, bit 8 is the least significant bit of the second byte, etc.
1709 This avoids any complications with differing word sizes.
1711 The "num_dirty" field is a performance hint for KVM to determine whether it
1712 should skip processing the bitmap and just invalidate everything. It must
1713 be set to the number of set bits in the bitmap.
1716 4.61 KVM_ASSIGN_SET_INTX_MASK (deprecated)
1718 Capability: KVM_CAP_PCI_2_3
1721 Parameters: struct kvm_assigned_pci_dev (in)
1722 Returns: 0 on success, -1 on error
1724 Allows userspace to mask PCI INTx interrupts from the assigned device. The
1725 kernel will not deliver INTx interrupts to the guest between setting and
1726 clearing of KVM_ASSIGN_SET_INTX_MASK via this interface. This enables use of
1727 and emulation of PCI 2.3 INTx disable command register behavior.
1729 This may be used for both PCI 2.3 devices supporting INTx disable natively and
1730 older devices lacking this support. Userspace is responsible for emulating the
1731 read value of the INTx disable bit in the guest visible PCI command register.
1732 When modifying the INTx disable state, userspace should precede updating the
1733 physical device command register by calling this ioctl to inform the kernel of
1734 the new intended INTx mask state.
1736 Note that the kernel uses the device INTx disable bit to internally manage the
1737 device interrupt state for PCI 2.3 devices. Reads of this register may
1738 therefore not match the expected value. Writes should always use the guest
1739 intended INTx disable value rather than attempting to read-copy-update the
1740 current physical device state. Races between user and kernel updates to the
1741 INTx disable bit are handled lazily in the kernel. It's possible the device
1742 may generate unintended interrupts, but they will not be injected into the
1745 See KVM_ASSIGN_DEV_IRQ for the data structure. The target device is specified
1746 by assigned_dev_id. In the flags field, only KVM_DEV_ASSIGN_MASK_INTX is
1750 4.62 KVM_CREATE_SPAPR_TCE
1752 Capability: KVM_CAP_SPAPR_TCE
1753 Architectures: powerpc
1755 Parameters: struct kvm_create_spapr_tce (in)
1756 Returns: file descriptor for manipulating the created TCE table
1758 This creates a virtual TCE (translation control entry) table, which
1759 is an IOMMU for PAPR-style virtual I/O. It is used to translate
1760 logical addresses used in virtual I/O into guest physical addresses,
1761 and provides a scatter/gather capability for PAPR virtual I/O.
1763 /* for KVM_CAP_SPAPR_TCE */
1764 struct kvm_create_spapr_tce {
1769 The liobn field gives the logical IO bus number for which to create a
1770 TCE table. The window_size field specifies the size of the DMA window
1771 which this TCE table will translate - the table will contain one 64
1772 bit TCE entry for every 4kiB of the DMA window.
1774 When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE
1775 table has been created using this ioctl(), the kernel will handle it
1776 in real mode, updating the TCE table. H_PUT_TCE calls for other
1777 liobns will cause a vm exit and must be handled by userspace.
1779 The return value is a file descriptor which can be passed to mmap(2)
1780 to map the created TCE table into userspace. This lets userspace read
1781 the entries written by kernel-handled H_PUT_TCE calls, and also lets
1782 userspace update the TCE table directly which is useful in some
1786 4.63 KVM_ALLOCATE_RMA
1788 Capability: KVM_CAP_PPC_RMA
1789 Architectures: powerpc
1791 Parameters: struct kvm_allocate_rma (out)
1792 Returns: file descriptor for mapping the allocated RMA
1794 This allocates a Real Mode Area (RMA) from the pool allocated at boot
1795 time by the kernel. An RMA is a physically-contiguous, aligned region
1796 of memory used on older POWER processors to provide the memory which
1797 will be accessed by real-mode (MMU off) accesses in a KVM guest.
1798 POWER processors support a set of sizes for the RMA that usually
1799 includes 64MB, 128MB, 256MB and some larger powers of two.
1801 /* for KVM_ALLOCATE_RMA */
1802 struct kvm_allocate_rma {
1806 The return value is a file descriptor which can be passed to mmap(2)
1807 to map the allocated RMA into userspace. The mapped area can then be
1808 passed to the KVM_SET_USER_MEMORY_REGION ioctl to establish it as the
1809 RMA for a virtual machine. The size of the RMA in bytes (which is
1810 fixed at host kernel boot time) is returned in the rma_size field of
1811 the argument structure.
1813 The KVM_CAP_PPC_RMA capability is 1 or 2 if the KVM_ALLOCATE_RMA ioctl
1814 is supported; 2 if the processor requires all virtual machines to have
1815 an RMA, or 1 if the processor can use an RMA but doesn't require it,
1816 because it supports the Virtual RMA (VRMA) facility.
1821 Capability: KVM_CAP_USER_NMI
1825 Returns: 0 on success, -1 on error
1827 Queues an NMI on the thread's vcpu. Note this is well defined only
1828 when KVM_CREATE_IRQCHIP has not been called, since this is an interface
1829 between the virtual cpu core and virtual local APIC. After KVM_CREATE_IRQCHIP
1830 has been called, this interface is completely emulated within the kernel.
1832 To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the
1833 following algorithm:
1836 - read the local APIC's state (KVM_GET_LAPIC)
1837 - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1)
1838 - if so, issue KVM_NMI
1841 Some guests configure the LINT1 NMI input to cause a panic, aiding in
1845 4.65 KVM_S390_UCAS_MAP
1847 Capability: KVM_CAP_S390_UCONTROL
1850 Parameters: struct kvm_s390_ucas_mapping (in)
1851 Returns: 0 in case of success
1853 The parameter is defined like this:
1854 struct kvm_s390_ucas_mapping {
1860 This ioctl maps the memory at "user_addr" with the length "length" to
1861 the vcpu's address space starting at "vcpu_addr". All parameters need to
1862 be aligned by 1 megabyte.
1865 4.66 KVM_S390_UCAS_UNMAP
1867 Capability: KVM_CAP_S390_UCONTROL
1870 Parameters: struct kvm_s390_ucas_mapping (in)
1871 Returns: 0 in case of success
1873 The parameter is defined like this:
1874 struct kvm_s390_ucas_mapping {
1880 This ioctl unmaps the memory in the vcpu's address space starting at
1881 "vcpu_addr" with the length "length". The field "user_addr" is ignored.
1882 All parameters need to be aligned by 1 megabyte.
1885 4.67 KVM_S390_VCPU_FAULT
1887 Capability: KVM_CAP_S390_UCONTROL
1890 Parameters: vcpu absolute address (in)
1891 Returns: 0 in case of success
1893 This call creates a page table entry on the virtual cpu's address space
1894 (for user controlled virtual machines) or the virtual machine's address
1895 space (for regular virtual machines). This only works for minor faults,
1896 thus it's recommended to access subject memory page via the user page
1897 table upfront. This is useful to handle validity intercepts for user
1898 controlled virtual machines to fault in the virtual cpu's lowcore pages
1899 prior to calling the KVM_RUN ioctl.
1902 4.68 KVM_SET_ONE_REG
1904 Capability: KVM_CAP_ONE_REG
1907 Parameters: struct kvm_one_reg (in)
1908 Returns: 0 on success, negative value on failure
1910 struct kvm_one_reg {
1915 Using this ioctl, a single vcpu register can be set to a specific value
1916 defined by user space with the passed in struct kvm_one_reg, where id
1917 refers to the register identifier as described below and addr is a pointer
1918 to a variable with the respective size. There can be architecture agnostic
1919 and architecture specific registers. Each have their own range of operation
1920 and their own constants and width. To keep track of the implemented
1921 registers, find a list below:
1923 Arch | Register | Width (bits)
1925 PPC | KVM_REG_PPC_HIOR | 64
1926 PPC | KVM_REG_PPC_IAC1 | 64
1927 PPC | KVM_REG_PPC_IAC2 | 64
1928 PPC | KVM_REG_PPC_IAC3 | 64
1929 PPC | KVM_REG_PPC_IAC4 | 64
1930 PPC | KVM_REG_PPC_DAC1 | 64
1931 PPC | KVM_REG_PPC_DAC2 | 64
1932 PPC | KVM_REG_PPC_DABR | 64
1933 PPC | KVM_REG_PPC_DSCR | 64
1934 PPC | KVM_REG_PPC_PURR | 64
1935 PPC | KVM_REG_PPC_SPURR | 64
1936 PPC | KVM_REG_PPC_DAR | 64
1937 PPC | KVM_REG_PPC_DSISR | 32
1938 PPC | KVM_REG_PPC_AMR | 64
1939 PPC | KVM_REG_PPC_UAMOR | 64
1940 PPC | KVM_REG_PPC_MMCR0 | 64
1941 PPC | KVM_REG_PPC_MMCR1 | 64
1942 PPC | KVM_REG_PPC_MMCRA | 64
1943 PPC | KVM_REG_PPC_MMCR2 | 64
1944 PPC | KVM_REG_PPC_MMCRS | 64
1945 PPC | KVM_REG_PPC_SIAR | 64
1946 PPC | KVM_REG_PPC_SDAR | 64
1947 PPC | KVM_REG_PPC_SIER | 64
1948 PPC | KVM_REG_PPC_PMC1 | 32
1949 PPC | KVM_REG_PPC_PMC2 | 32
1950 PPC | KVM_REG_PPC_PMC3 | 32
1951 PPC | KVM_REG_PPC_PMC4 | 32
1952 PPC | KVM_REG_PPC_PMC5 | 32
1953 PPC | KVM_REG_PPC_PMC6 | 32
1954 PPC | KVM_REG_PPC_PMC7 | 32
1955 PPC | KVM_REG_PPC_PMC8 | 32
1956 PPC | KVM_REG_PPC_FPR0 | 64
1958 PPC | KVM_REG_PPC_FPR31 | 64
1959 PPC | KVM_REG_PPC_VR0 | 128
1961 PPC | KVM_REG_PPC_VR31 | 128
1962 PPC | KVM_REG_PPC_VSR0 | 128
1964 PPC | KVM_REG_PPC_VSR31 | 128
1965 PPC | KVM_REG_PPC_FPSCR | 64
1966 PPC | KVM_REG_PPC_VSCR | 32
1967 PPC | KVM_REG_PPC_VPA_ADDR | 64
1968 PPC | KVM_REG_PPC_VPA_SLB | 128
1969 PPC | KVM_REG_PPC_VPA_DTL | 128
1970 PPC | KVM_REG_PPC_EPCR | 32
1971 PPC | KVM_REG_PPC_EPR | 32
1972 PPC | KVM_REG_PPC_TCR | 32
1973 PPC | KVM_REG_PPC_TSR | 32
1974 PPC | KVM_REG_PPC_OR_TSR | 32
1975 PPC | KVM_REG_PPC_CLEAR_TSR | 32
1976 PPC | KVM_REG_PPC_MAS0 | 32
1977 PPC | KVM_REG_PPC_MAS1 | 32
1978 PPC | KVM_REG_PPC_MAS2 | 64
1979 PPC | KVM_REG_PPC_MAS7_3 | 64
1980 PPC | KVM_REG_PPC_MAS4 | 32
1981 PPC | KVM_REG_PPC_MAS6 | 32
1982 PPC | KVM_REG_PPC_MMUCFG | 32
1983 PPC | KVM_REG_PPC_TLB0CFG | 32
1984 PPC | KVM_REG_PPC_TLB1CFG | 32
1985 PPC | KVM_REG_PPC_TLB2CFG | 32
1986 PPC | KVM_REG_PPC_TLB3CFG | 32
1987 PPC | KVM_REG_PPC_TLB0PS | 32
1988 PPC | KVM_REG_PPC_TLB1PS | 32
1989 PPC | KVM_REG_PPC_TLB2PS | 32
1990 PPC | KVM_REG_PPC_TLB3PS | 32
1991 PPC | KVM_REG_PPC_EPTCFG | 32
1992 PPC | KVM_REG_PPC_ICP_STATE | 64
1993 PPC | KVM_REG_PPC_TB_OFFSET | 64
1994 PPC | KVM_REG_PPC_SPMC1 | 32
1995 PPC | KVM_REG_PPC_SPMC2 | 32
1996 PPC | KVM_REG_PPC_IAMR | 64
1997 PPC | KVM_REG_PPC_TFHAR | 64
1998 PPC | KVM_REG_PPC_TFIAR | 64
1999 PPC | KVM_REG_PPC_TEXASR | 64
2000 PPC | KVM_REG_PPC_FSCR | 64
2001 PPC | KVM_REG_PPC_PSPB | 32
2002 PPC | KVM_REG_PPC_EBBHR | 64
2003 PPC | KVM_REG_PPC_EBBRR | 64
2004 PPC | KVM_REG_PPC_BESCR | 64
2005 PPC | KVM_REG_PPC_TAR | 64
2006 PPC | KVM_REG_PPC_DPDES | 64
2007 PPC | KVM_REG_PPC_DAWR | 64
2008 PPC | KVM_REG_PPC_DAWRX | 64
2009 PPC | KVM_REG_PPC_CIABR | 64
2010 PPC | KVM_REG_PPC_IC | 64
2011 PPC | KVM_REG_PPC_VTB | 64
2012 PPC | KVM_REG_PPC_CSIGR | 64
2013 PPC | KVM_REG_PPC_TACR | 64
2014 PPC | KVM_REG_PPC_TCSCR | 64
2015 PPC | KVM_REG_PPC_PID | 64
2016 PPC | KVM_REG_PPC_ACOP | 64
2017 PPC | KVM_REG_PPC_VRSAVE | 32
2018 PPC | KVM_REG_PPC_LPCR | 32
2019 PPC | KVM_REG_PPC_LPCR_64 | 64
2020 PPC | KVM_REG_PPC_PPR | 64
2021 PPC | KVM_REG_PPC_ARCH_COMPAT | 32
2022 PPC | KVM_REG_PPC_DABRX | 32
2023 PPC | KVM_REG_PPC_WORT | 64
2024 PPC | KVM_REG_PPC_SPRG9 | 64
2025 PPC | KVM_REG_PPC_DBSR | 32
2026 PPC | KVM_REG_PPC_TIDR | 64
2027 PPC | KVM_REG_PPC_PSSCR | 64
2028 PPC | KVM_REG_PPC_TM_GPR0 | 64
2030 PPC | KVM_REG_PPC_TM_GPR31 | 64
2031 PPC | KVM_REG_PPC_TM_VSR0 | 128
2033 PPC | KVM_REG_PPC_TM_VSR63 | 128
2034 PPC | KVM_REG_PPC_TM_CR | 64
2035 PPC | KVM_REG_PPC_TM_LR | 64
2036 PPC | KVM_REG_PPC_TM_CTR | 64
2037 PPC | KVM_REG_PPC_TM_FPSCR | 64
2038 PPC | KVM_REG_PPC_TM_AMR | 64
2039 PPC | KVM_REG_PPC_TM_PPR | 64
2040 PPC | KVM_REG_PPC_TM_VRSAVE | 64
2041 PPC | KVM_REG_PPC_TM_VSCR | 32
2042 PPC | KVM_REG_PPC_TM_DSCR | 64
2043 PPC | KVM_REG_PPC_TM_TAR | 64
2044 PPC | KVM_REG_PPC_TM_XER | 64
2046 MIPS | KVM_REG_MIPS_R0 | 64
2048 MIPS | KVM_REG_MIPS_R31 | 64
2049 MIPS | KVM_REG_MIPS_HI | 64
2050 MIPS | KVM_REG_MIPS_LO | 64
2051 MIPS | KVM_REG_MIPS_PC | 64
2052 MIPS | KVM_REG_MIPS_CP0_INDEX | 32
2053 MIPS | KVM_REG_MIPS_CP0_CONTEXT | 64
2054 MIPS | KVM_REG_MIPS_CP0_USERLOCAL | 64
2055 MIPS | KVM_REG_MIPS_CP0_PAGEMASK | 32
2056 MIPS | KVM_REG_MIPS_CP0_WIRED | 32
2057 MIPS | KVM_REG_MIPS_CP0_HWRENA | 32
2058 MIPS | KVM_REG_MIPS_CP0_BADVADDR | 64
2059 MIPS | KVM_REG_MIPS_CP0_COUNT | 32
2060 MIPS | KVM_REG_MIPS_CP0_ENTRYHI | 64
2061 MIPS | KVM_REG_MIPS_CP0_COMPARE | 32
2062 MIPS | KVM_REG_MIPS_CP0_STATUS | 32
2063 MIPS | KVM_REG_MIPS_CP0_CAUSE | 32
2064 MIPS | KVM_REG_MIPS_CP0_EPC | 64
2065 MIPS | KVM_REG_MIPS_CP0_PRID | 32
2066 MIPS | KVM_REG_MIPS_CP0_CONFIG | 32
2067 MIPS | KVM_REG_MIPS_CP0_CONFIG1 | 32
2068 MIPS | KVM_REG_MIPS_CP0_CONFIG2 | 32
2069 MIPS | KVM_REG_MIPS_CP0_CONFIG3 | 32
2070 MIPS | KVM_REG_MIPS_CP0_CONFIG4 | 32
2071 MIPS | KVM_REG_MIPS_CP0_CONFIG5 | 32
2072 MIPS | KVM_REG_MIPS_CP0_CONFIG7 | 32
2073 MIPS | KVM_REG_MIPS_CP0_ERROREPC | 64
2074 MIPS | KVM_REG_MIPS_CP0_KSCRATCH1 | 64
2075 MIPS | KVM_REG_MIPS_CP0_KSCRATCH2 | 64
2076 MIPS | KVM_REG_MIPS_CP0_KSCRATCH3 | 64
2077 MIPS | KVM_REG_MIPS_CP0_KSCRATCH4 | 64
2078 MIPS | KVM_REG_MIPS_CP0_KSCRATCH5 | 64
2079 MIPS | KVM_REG_MIPS_CP0_KSCRATCH6 | 64
2080 MIPS | KVM_REG_MIPS_COUNT_CTL | 64
2081 MIPS | KVM_REG_MIPS_COUNT_RESUME | 64
2082 MIPS | KVM_REG_MIPS_COUNT_HZ | 64
2083 MIPS | KVM_REG_MIPS_FPR_32(0..31) | 32
2084 MIPS | KVM_REG_MIPS_FPR_64(0..31) | 64
2085 MIPS | KVM_REG_MIPS_VEC_128(0..31) | 128
2086 MIPS | KVM_REG_MIPS_FCR_IR | 32
2087 MIPS | KVM_REG_MIPS_FCR_CSR | 32
2088 MIPS | KVM_REG_MIPS_MSA_IR | 32
2089 MIPS | KVM_REG_MIPS_MSA_CSR | 32
2091 ARM registers are mapped using the lower 32 bits. The upper 16 of that
2092 is the register group type, or coprocessor number:
2094 ARM core registers have the following id bit patterns:
2095 0x4020 0000 0010 <index into the kvm_regs struct:16>
2097 ARM 32-bit CP15 registers have the following id bit patterns:
2098 0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3>
2100 ARM 64-bit CP15 registers have the following id bit patterns:
2101 0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3>
2103 ARM CCSIDR registers are demultiplexed by CSSELR value:
2104 0x4020 0000 0011 00 <csselr:8>
2106 ARM 32-bit VFP control registers have the following id bit patterns:
2107 0x4020 0000 0012 1 <regno:12>
2109 ARM 64-bit FP registers have the following id bit patterns:
2110 0x4030 0000 0012 0 <regno:12>
2113 arm64 registers are mapped using the lower 32 bits. The upper 16 of
2114 that is the register group type, or coprocessor number:
2116 arm64 core/FP-SIMD registers have the following id bit patterns. Note
2117 that the size of the access is variable, as the kvm_regs structure
2118 contains elements ranging from 32 to 128 bits. The index is a 32bit
2119 value in the kvm_regs structure seen as a 32bit array.
2120 0x60x0 0000 0010 <index into the kvm_regs struct:16>
2122 arm64 CCSIDR registers are demultiplexed by CSSELR value:
2123 0x6020 0000 0011 00 <csselr:8>
2125 arm64 system registers have the following id bit patterns:
2126 0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3>
2129 MIPS registers are mapped using the lower 32 bits. The upper 16 of that is
2130 the register group type:
2132 MIPS core registers (see above) have the following id bit patterns:
2133 0x7030 0000 0000 <reg:16>
2135 MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit
2136 patterns depending on whether they're 32-bit or 64-bit registers:
2137 0x7020 0000 0001 00 <reg:5> <sel:3> (32-bit)
2138 0x7030 0000 0001 00 <reg:5> <sel:3> (64-bit)
2140 MIPS KVM control registers (see above) have the following id bit patterns:
2141 0x7030 0000 0002 <reg:16>
2143 MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following
2144 id bit patterns depending on the size of the register being accessed. They are
2145 always accessed according to the current guest FPU mode (Status.FR and
2146 Config5.FRE), i.e. as the guest would see them, and they become unpredictable
2147 if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector
2148 registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they
2149 overlap the FPU registers:
2150 0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers)
2151 0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers)
2152 0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers)
2154 MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the
2155 following id bit patterns:
2156 0x7020 0000 0003 01 <0:3> <reg:5>
2158 MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the
2159 following id bit patterns:
2160 0x7020 0000 0003 02 <0:3> <reg:5>
2163 4.69 KVM_GET_ONE_REG
2165 Capability: KVM_CAP_ONE_REG
2168 Parameters: struct kvm_one_reg (in and out)
2169 Returns: 0 on success, negative value on failure
2171 This ioctl allows to receive the value of a single register implemented
2172 in a vcpu. The register to read is indicated by the "id" field of the
2173 kvm_one_reg struct passed in. On success, the register value can be found
2174 at the memory location pointed to by "addr".
2176 The list of registers accessible using this interface is identical to the
2180 4.70 KVM_KVMCLOCK_CTRL
2182 Capability: KVM_CAP_KVMCLOCK_CTRL
2183 Architectures: Any that implement pvclocks (currently x86 only)
2186 Returns: 0 on success, -1 on error
2188 This signals to the host kernel that the specified guest is being paused by
2189 userspace. The host will set a flag in the pvclock structure that is checked
2190 from the soft lockup watchdog. The flag is part of the pvclock structure that
2191 is shared between guest and host, specifically the second bit of the flags
2192 field of the pvclock_vcpu_time_info structure. It will be set exclusively by
2193 the host and read/cleared exclusively by the guest. The guest operation of
2194 checking and clearing the flag must an atomic operation so
2195 load-link/store-conditional, or equivalent must be used. There are two cases
2196 where the guest will clear the flag: when the soft lockup watchdog timer resets
2197 itself or when a soft lockup is detected. This ioctl can be called any time
2198 after pausing the vcpu, but before it is resumed.
2203 Capability: KVM_CAP_SIGNAL_MSI
2204 Architectures: x86 arm arm64
2206 Parameters: struct kvm_msi (in)
2207 Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error
2209 Directly inject a MSI message. Only valid with in-kernel irqchip that handles
2221 flags: KVM_MSI_VALID_DEVID: devid contains a valid value. The per-VM
2222 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
2223 the device ID. If this capability is not available, userspace
2224 should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
2226 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
2227 for the device that wrote the MSI message. For PCI, this is usually a
2228 BFD identifier in the lower 16 bits.
2230 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
2231 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
2232 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
2233 address_hi must be zero.
2236 4.71 KVM_CREATE_PIT2
2238 Capability: KVM_CAP_PIT2
2241 Parameters: struct kvm_pit_config (in)
2242 Returns: 0 on success, -1 on error
2244 Creates an in-kernel device model for the i8254 PIT. This call is only valid
2245 after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following
2246 parameters have to be passed:
2248 struct kvm_pit_config {
2255 #define KVM_PIT_SPEAKER_DUMMY 1 /* emulate speaker port stub */
2257 PIT timer interrupts may use a per-VM kernel thread for injection. If it
2258 exists, this thread will have a name of the following pattern:
2260 kvm-pit/<owner-process-pid>
2262 When running a guest with elevated priorities, the scheduling parameters of
2263 this thread may have to be adjusted accordingly.
2265 This IOCTL replaces the obsolete KVM_CREATE_PIT.
2270 Capability: KVM_CAP_PIT_STATE2
2273 Parameters: struct kvm_pit_state2 (out)
2274 Returns: 0 on success, -1 on error
2276 Retrieves the state of the in-kernel PIT model. Only valid after
2277 KVM_CREATE_PIT2. The state is returned in the following structure:
2279 struct kvm_pit_state2 {
2280 struct kvm_pit_channel_state channels[3];
2287 /* disable PIT in HPET legacy mode */
2288 #define KVM_PIT_FLAGS_HPET_LEGACY 0x00000001
2290 This IOCTL replaces the obsolete KVM_GET_PIT.
2295 Capability: KVM_CAP_PIT_STATE2
2298 Parameters: struct kvm_pit_state2 (in)
2299 Returns: 0 on success, -1 on error
2301 Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2.
2302 See KVM_GET_PIT2 for details on struct kvm_pit_state2.
2304 This IOCTL replaces the obsolete KVM_SET_PIT.
2307 4.74 KVM_PPC_GET_SMMU_INFO
2309 Capability: KVM_CAP_PPC_GET_SMMU_INFO
2310 Architectures: powerpc
2313 Returns: 0 on success, -1 on error
2315 This populates and returns a structure describing the features of
2316 the "Server" class MMU emulation supported by KVM.
2317 This can in turn be used by userspace to generate the appropriate
2318 device-tree properties for the guest operating system.
2320 The structure contains some global information, followed by an
2321 array of supported segment page sizes:
2323 struct kvm_ppc_smmu_info {
2327 struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ];
2330 The supported flags are:
2332 - KVM_PPC_PAGE_SIZES_REAL:
2333 When that flag is set, guest page sizes must "fit" the backing
2334 store page sizes. When not set, any page size in the list can
2335 be used regardless of how they are backed by userspace.
2337 - KVM_PPC_1T_SEGMENTS
2338 The emulated MMU supports 1T segments in addition to the
2341 The "slb_size" field indicates how many SLB entries are supported
2343 The "sps" array contains 8 entries indicating the supported base
2344 page sizes for a segment in increasing order. Each entry is defined
2347 struct kvm_ppc_one_seg_page_size {
2348 __u32 page_shift; /* Base page shift of segment (or 0) */
2349 __u32 slb_enc; /* SLB encoding for BookS */
2350 struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ];
2353 An entry with a "page_shift" of 0 is unused. Because the array is
2354 organized in increasing order, a lookup can stop when encoutering
2357 The "slb_enc" field provides the encoding to use in the SLB for the
2358 page size. The bits are in positions such as the value can directly
2359 be OR'ed into the "vsid" argument of the slbmte instruction.
2361 The "enc" array is a list which for each of those segment base page
2362 size provides the list of supported actual page sizes (which can be
2363 only larger or equal to the base page size), along with the
2364 corresponding encoding in the hash PTE. Similarly, the array is
2365 8 entries sorted by increasing sizes and an entry with a "0" shift
2366 is an empty entry and a terminator:
2368 struct kvm_ppc_one_page_size {
2369 __u32 page_shift; /* Page shift (or 0) */
2370 __u32 pte_enc; /* Encoding in the HPTE (>>12) */
2373 The "pte_enc" field provides a value that can OR'ed into the hash
2374 PTE's RPN field (ie, it needs to be shifted left by 12 to OR it
2375 into the hash PTE second double word).
2379 Capability: KVM_CAP_IRQFD
2380 Architectures: x86 s390 arm arm64
2382 Parameters: struct kvm_irqfd (in)
2383 Returns: 0 on success, -1 on error
2385 Allows setting an eventfd to directly trigger a guest interrupt.
2386 kvm_irqfd.fd specifies the file descriptor to use as the eventfd and
2387 kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When
2388 an event is triggered on the eventfd, an interrupt is injected into
2389 the guest using the specified gsi pin. The irqfd is removed using
2390 the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd
2393 With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify
2394 mechanism allowing emulation of level-triggered, irqfd-based
2395 interrupts. When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an
2396 additional eventfd in the kvm_irqfd.resamplefd field. When operating
2397 in resample mode, posting of an interrupt through kvm_irq.fd asserts
2398 the specified gsi in the irqchip. When the irqchip is resampled, such
2399 as from an EOI, the gsi is de-asserted and the user is notified via
2400 kvm_irqfd.resamplefd. It is the user's responsibility to re-queue
2401 the interrupt if the device making use of it still requires service.
2402 Note that closing the resamplefd is not sufficient to disable the
2403 irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment
2404 and need not be specified with KVM_IRQFD_FLAG_DEASSIGN.
2406 On arm/arm64, gsi routing being supported, the following can happen:
2407 - in case no routing entry is associated to this gsi, injection fails
2408 - in case the gsi is associated to an irqchip routing entry,
2409 irqchip.pin + 32 corresponds to the injected SPI ID.
2410 - in case the gsi is associated to an MSI routing entry, the MSI
2411 message and device ID are translated into an LPI (support restricted
2412 to GICv3 ITS in-kernel emulation).
2414 4.76 KVM_PPC_ALLOCATE_HTAB
2416 Capability: KVM_CAP_PPC_ALLOC_HTAB
2417 Architectures: powerpc
2419 Parameters: Pointer to u32 containing hash table order (in/out)
2420 Returns: 0 on success, -1 on error
2422 This requests the host kernel to allocate an MMU hash table for a
2423 guest using the PAPR paravirtualization interface. This only does
2424 anything if the kernel is configured to use the Book 3S HV style of
2425 virtualization. Otherwise the capability doesn't exist and the ioctl
2426 returns an ENOTTY error. The rest of this description assumes Book 3S
2429 There must be no vcpus running when this ioctl is called; if there
2430 are, it will do nothing and return an EBUSY error.
2432 The parameter is a pointer to a 32-bit unsigned integer variable
2433 containing the order (log base 2) of the desired size of the hash
2434 table, which must be between 18 and 46. On successful return from the
2435 ioctl, it will have been updated with the order of the hash table that
2438 If no hash table has been allocated when any vcpu is asked to run
2439 (with the KVM_RUN ioctl), the host kernel will allocate a
2440 default-sized hash table (16 MB).
2442 If this ioctl is called when a hash table has already been allocated,
2443 the kernel will clear out the existing hash table (zero all HPTEs) and
2444 return the hash table order in the parameter. (If the guest is using
2445 the virtualized real-mode area (VRMA) facility, the kernel will
2446 re-create the VMRA HPTEs on the next KVM_RUN of any vcpu.)
2448 4.77 KVM_S390_INTERRUPT
2452 Type: vm ioctl, vcpu ioctl
2453 Parameters: struct kvm_s390_interrupt (in)
2454 Returns: 0 on success, -1 on error
2456 Allows to inject an interrupt to the guest. Interrupts can be floating
2457 (vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type.
2459 Interrupt parameters are passed via kvm_s390_interrupt:
2461 struct kvm_s390_interrupt {
2467 type can be one of the following:
2469 KVM_S390_SIGP_STOP (vcpu) - sigp stop; optional flags in parm
2470 KVM_S390_PROGRAM_INT (vcpu) - program check; code in parm
2471 KVM_S390_SIGP_SET_PREFIX (vcpu) - sigp set prefix; prefix address in parm
2472 KVM_S390_RESTART (vcpu) - restart
2473 KVM_S390_INT_CLOCK_COMP (vcpu) - clock comparator interrupt
2474 KVM_S390_INT_CPU_TIMER (vcpu) - CPU timer interrupt
2475 KVM_S390_INT_VIRTIO (vm) - virtio external interrupt; external interrupt
2476 parameters in parm and parm64
2477 KVM_S390_INT_SERVICE (vm) - sclp external interrupt; sclp parameter in parm
2478 KVM_S390_INT_EMERGENCY (vcpu) - sigp emergency; source cpu in parm
2479 KVM_S390_INT_EXTERNAL_CALL (vcpu) - sigp external call; source cpu in parm
2480 KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm) - compound value to indicate an
2481 I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel);
2482 I/O interruption parameters in parm (subchannel) and parm64 (intparm,
2483 interruption subclass)
2484 KVM_S390_MCHK (vm, vcpu) - machine check interrupt; cr 14 bits in parm,
2485 machine check interrupt code in parm64 (note that
2486 machine checks needing further payload are not
2487 supported by this ioctl)
2489 Note that the vcpu ioctl is asynchronous to vcpu execution.
2491 4.78 KVM_PPC_GET_HTAB_FD
2493 Capability: KVM_CAP_PPC_HTAB_FD
2494 Architectures: powerpc
2496 Parameters: Pointer to struct kvm_get_htab_fd (in)
2497 Returns: file descriptor number (>= 0) on success, -1 on error
2499 This returns a file descriptor that can be used either to read out the
2500 entries in the guest's hashed page table (HPT), or to write entries to
2501 initialize the HPT. The returned fd can only be written to if the
2502 KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and
2503 can only be read if that bit is clear. The argument struct looks like
2506 /* For KVM_PPC_GET_HTAB_FD */
2507 struct kvm_get_htab_fd {
2513 /* Values for kvm_get_htab_fd.flags */
2514 #define KVM_GET_HTAB_BOLTED_ONLY ((__u64)0x1)
2515 #define KVM_GET_HTAB_WRITE ((__u64)0x2)
2517 The `start_index' field gives the index in the HPT of the entry at
2518 which to start reading. It is ignored when writing.
2520 Reads on the fd will initially supply information about all
2521 "interesting" HPT entries. Interesting entries are those with the
2522 bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise
2523 all entries. When the end of the HPT is reached, the read() will
2524 return. If read() is called again on the fd, it will start again from
2525 the beginning of the HPT, but will only return HPT entries that have
2526 changed since they were last read.
2528 Data read or written is structured as a header (8 bytes) followed by a
2529 series of valid HPT entries (16 bytes) each. The header indicates how
2530 many valid HPT entries there are and how many invalid entries follow
2531 the valid entries. The invalid entries are not represented explicitly
2532 in the stream. The header format is:
2534 struct kvm_get_htab_header {
2540 Writes to the fd create HPT entries starting at the index given in the
2541 header; first `n_valid' valid entries with contents from the data
2542 written, then `n_invalid' invalid entries, invalidating any previously
2543 valid entries found.
2545 4.79 KVM_CREATE_DEVICE
2547 Capability: KVM_CAP_DEVICE_CTRL
2549 Parameters: struct kvm_create_device (in/out)
2550 Returns: 0 on success, -1 on error
2552 ENODEV: The device type is unknown or unsupported
2553 EEXIST: Device already created, and this type of device may not
2554 be instantiated multiple times
2556 Other error conditions may be defined by individual device types or
2557 have their standard meanings.
2559 Creates an emulated device in the kernel. The file descriptor returned
2560 in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR.
2562 If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the
2563 device type is supported (not necessarily whether it can be created
2566 Individual devices should not define flags. Attributes should be used
2567 for specifying any behavior that is not implied by the device type
2570 struct kvm_create_device {
2571 __u32 type; /* in: KVM_DEV_TYPE_xxx */
2572 __u32 fd; /* out: device handle */
2573 __u32 flags; /* in: KVM_CREATE_DEVICE_xxx */
2576 4.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR
2578 Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
2579 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
2580 Type: device ioctl, vm ioctl, vcpu ioctl
2581 Parameters: struct kvm_device_attr
2582 Returns: 0 on success, -1 on error
2584 ENXIO: The group or attribute is unknown/unsupported for this device
2585 or hardware support is missing.
2586 EPERM: The attribute cannot (currently) be accessed this way
2587 (e.g. read-only attribute, or attribute that only makes
2588 sense when the device is in a different state)
2590 Other error conditions may be defined by individual device types.
2592 Gets/sets a specified piece of device configuration and/or state. The
2593 semantics are device-specific. See individual device documentation in
2594 the "devices" directory. As with ONE_REG, the size of the data
2595 transferred is defined by the particular attribute.
2597 struct kvm_device_attr {
2598 __u32 flags; /* no flags currently defined */
2599 __u32 group; /* device-defined */
2600 __u64 attr; /* group-defined */
2601 __u64 addr; /* userspace address of attr data */
2604 4.81 KVM_HAS_DEVICE_ATTR
2606 Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
2607 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
2608 Type: device ioctl, vm ioctl, vcpu ioctl
2609 Parameters: struct kvm_device_attr
2610 Returns: 0 on success, -1 on error
2612 ENXIO: The group or attribute is unknown/unsupported for this device
2613 or hardware support is missing.
2615 Tests whether a device supports a particular attribute. A successful
2616 return indicates the attribute is implemented. It does not necessarily
2617 indicate that the attribute can be read or written in the device's
2618 current state. "addr" is ignored.
2620 4.82 KVM_ARM_VCPU_INIT
2623 Architectures: arm, arm64
2625 Parameters: struct kvm_vcpu_init (in)
2626 Returns: 0 on success; -1 on error
2628 Â EINVAL: Â Â Â the target is unknown, or the combination of features is invalid.
2629 Â ENOENT: Â Â Â a features bit specified is unknown.
2631 This tells KVM what type of CPU to present to the guest, and what
2632 optional features it should have. Â This will cause a reset of the cpu
2633 registers to their initial values. Â If this is not called, KVM_RUN will
2634 return ENOEXEC for that vcpu.
2636 Note that because some registers reflect machine topology, all vcpus
2637 should be created before this ioctl is invoked.
2639 Userspace can call this function multiple times for a given vcpu, including
2640 after the vcpu has been run. This will reset the vcpu to its initial
2641 state. All calls to this function after the initial call must use the same
2642 target and same set of feature flags, otherwise EINVAL will be returned.
2645 - KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state.
2646 Depends on KVM_CAP_ARM_PSCI. If not set, the CPU will be powered on
2647 and execute guest code when KVM_RUN is called.
2648 - KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode.
2649 Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only).
2650 - KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 for the CPU.
2651 Depends on KVM_CAP_ARM_PSCI_0_2.
2652 - KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU.
2653 Depends on KVM_CAP_ARM_PMU_V3.
2656 4.83 KVM_ARM_PREFERRED_TARGET
2659 Architectures: arm, arm64
2661 Parameters: struct struct kvm_vcpu_init (out)
2662 Returns: 0 on success; -1 on error
2664 ENODEV: no preferred target available for the host
2666 This queries KVM for preferred CPU target type which can be emulated
2667 by KVM on underlying host.
2669 The ioctl returns struct kvm_vcpu_init instance containing information
2670 about preferred CPU target type and recommended features for it. The
2671 kvm_vcpu_init->features bitmap returned will have feature bits set if
2672 the preferred target recommends setting these features, but this is
2675 The information returned by this ioctl can be used to prepare an instance
2676 of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in
2677 in VCPU matching underlying host.
2680 4.84 KVM_GET_REG_LIST
2683 Architectures: arm, arm64, mips
2685 Parameters: struct kvm_reg_list (in/out)
2686 Returns: 0 on success; -1 on error
2688 Â E2BIG: Â Â Â Â the reg index list is too big to fit in the array specified by
2689 Â Â Â Â Â Â Â Â Â Â Â Â the user (the number required will be written into n).
2691 struct kvm_reg_list {
2692 __u64 n; /* number of registers in reg[] */
2696 This ioctl returns the guest registers that are supported for the
2697 KVM_GET_ONE_REG/KVM_SET_ONE_REG calls.
2700 4.85 KVM_ARM_SET_DEVICE_ADDR (deprecated)
2702 Capability: KVM_CAP_ARM_SET_DEVICE_ADDR
2703 Architectures: arm, arm64
2705 Parameters: struct kvm_arm_device_address (in)
2706 Returns: 0 on success, -1 on error
2708 ENODEV: The device id is unknown
2709 ENXIO: Device not supported on current system
2710 EEXIST: Address already set
2711 E2BIG: Address outside guest physical address space
2712 EBUSY: Address overlaps with other device range
2714 struct kvm_arm_device_addr {
2719 Specify a device address in the guest's physical address space where guests
2720 can access emulated or directly exposed devices, which the host kernel needs
2721 to know about. The id field is an architecture specific identifier for a
2724 ARM/arm64 divides the id field into two parts, a device id and an
2725 address type id specific to the individual device.
2727 Â bits: | 63 ... 32 | 31 ... 16 | 15 ... 0 |
2728 field: | 0x00000000 | device id | addr type id |
2730 ARM/arm64 currently only require this when using the in-kernel GIC
2731 support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2
2732 as the device id. When setting the base address for the guest's
2733 mapping of the VGIC virtual CPU and distributor interface, the ioctl
2734 must be called after calling KVM_CREATE_IRQCHIP, but before calling
2735 KVM_RUN on any of the VCPUs. Calling this ioctl twice for any of the
2736 base addresses will return -EEXIST.
2738 Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API
2739 should be used instead.
2742 4.86 KVM_PPC_RTAS_DEFINE_TOKEN
2744 Capability: KVM_CAP_PPC_RTAS
2747 Parameters: struct kvm_rtas_token_args
2748 Returns: 0 on success, -1 on error
2750 Defines a token value for a RTAS (Run Time Abstraction Services)
2751 service in order to allow it to be handled in the kernel. The
2752 argument struct gives the name of the service, which must be the name
2753 of a service that has a kernel-side implementation. If the token
2754 value is non-zero, it will be associated with that service, and
2755 subsequent RTAS calls by the guest specifying that token will be
2756 handled by the kernel. If the token value is 0, then any token
2757 associated with the service will be forgotten, and subsequent RTAS
2758 calls by the guest for that service will be passed to userspace to be
2761 4.87 KVM_SET_GUEST_DEBUG
2763 Capability: KVM_CAP_SET_GUEST_DEBUG
2764 Architectures: x86, s390, ppc, arm64
2766 Parameters: struct kvm_guest_debug (in)
2767 Returns: 0 on success; -1 on error
2769 struct kvm_guest_debug {
2772 struct kvm_guest_debug_arch arch;
2775 Set up the processor specific debug registers and configure vcpu for
2776 handling guest debug events. There are two parts to the structure, the
2777 first a control bitfield indicates the type of debug events to handle
2778 when running. Common control bits are:
2780 - KVM_GUESTDBG_ENABLE: guest debugging is enabled
2781 - KVM_GUESTDBG_SINGLESTEP: the next run should single-step
2783 The top 16 bits of the control field are architecture specific control
2784 flags which can include the following:
2786 - KVM_GUESTDBG_USE_SW_BP: using software breakpoints [x86, arm64]
2787 - KVM_GUESTDBG_USE_HW_BP: using hardware breakpoints [x86, s390, arm64]
2788 - KVM_GUESTDBG_INJECT_DB: inject DB type exception [x86]
2789 - KVM_GUESTDBG_INJECT_BP: inject BP type exception [x86]
2790 - KVM_GUESTDBG_EXIT_PENDING: trigger an immediate guest exit [s390]
2792 For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints
2793 are enabled in memory so we need to ensure breakpoint exceptions are
2794 correctly trapped and the KVM run loop exits at the breakpoint and not
2795 running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP
2796 we need to ensure the guest vCPUs architecture specific registers are
2797 updated to the correct (supplied) values.
2799 The second part of the structure is architecture specific and
2800 typically contains a set of debug registers.
2802 For arm64 the number of debug registers is implementation defined and
2803 can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and
2804 KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number
2805 indicating the number of supported registers.
2807 When debug events exit the main run loop with the reason
2808 KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run
2809 structure containing architecture specific debug information.
2811 4.88 KVM_GET_EMULATED_CPUID
2813 Capability: KVM_CAP_EXT_EMUL_CPUID
2816 Parameters: struct kvm_cpuid2 (in/out)
2817 Returns: 0 on success, -1 on error
2822 struct kvm_cpuid_entry2 entries[0];
2825 The member 'flags' is used for passing flags from userspace.
2827 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
2828 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1)
2829 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2)
2831 struct kvm_cpuid_entry2 {
2842 This ioctl returns x86 cpuid features which are emulated by
2843 kvm.Userspace can use the information returned by this ioctl to query
2844 which features are emulated by kvm instead of being present natively.
2846 Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2
2847 structure with the 'nent' field indicating the number of entries in
2848 the variable-size array 'entries'. If the number of entries is too low
2849 to describe the cpu capabilities, an error (E2BIG) is returned. If the
2850 number is too high, the 'nent' field is adjusted and an error (ENOMEM)
2851 is returned. If the number is just right, the 'nent' field is adjusted
2852 to the number of valid entries in the 'entries' array, which is then
2855 The entries returned are the set CPUID bits of the respective features
2856 which kvm emulates, as returned by the CPUID instruction, with unknown
2857 or unsupported feature bits cleared.
2859 Features like x2apic, for example, may not be present in the host cpu
2860 but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be
2861 emulated efficiently and thus not included here.
2863 The fields in each entry are defined as follows:
2865 function: the eax value used to obtain the entry
2866 index: the ecx value used to obtain the entry (for entries that are
2868 flags: an OR of zero or more of the following:
2869 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
2870 if the index field is valid
2871 KVM_CPUID_FLAG_STATEFUL_FUNC:
2872 if cpuid for this function returns different values for successive
2873 invocations; there will be several entries with the same function,
2874 all with this flag set
2875 KVM_CPUID_FLAG_STATE_READ_NEXT:
2876 for KVM_CPUID_FLAG_STATEFUL_FUNC entries, set if this entry is
2877 the first entry to be read by a cpu
2878 eax, ebx, ecx, edx: the values returned by the cpuid instruction for
2879 this function/index combination
2881 4.89 KVM_S390_MEM_OP
2883 Capability: KVM_CAP_S390_MEM_OP
2886 Parameters: struct kvm_s390_mem_op (in)
2887 Returns: = 0 on success,
2888 < 0 on generic error (e.g. -EFAULT or -ENOMEM),
2889 > 0 if an exception occurred while walking the page tables
2891 Read or write data from/to the logical (virtual) memory of a VCPU.
2893 Parameters are specified via the following structure:
2895 struct kvm_s390_mem_op {
2896 __u64 gaddr; /* the guest address */
2897 __u64 flags; /* flags */
2898 __u32 size; /* amount of bytes */
2899 __u32 op; /* type of operation */
2900 __u64 buf; /* buffer in userspace */
2901 __u8 ar; /* the access register number */
2902 __u8 reserved[31]; /* should be set to 0 */
2905 The type of operation is specified in the "op" field. It is either
2906 KVM_S390_MEMOP_LOGICAL_READ for reading from logical memory space or
2907 KVM_S390_MEMOP_LOGICAL_WRITE for writing to logical memory space. The
2908 KVM_S390_MEMOP_F_CHECK_ONLY flag can be set in the "flags" field to check
2909 whether the corresponding memory access would create an access exception
2910 (without touching the data in the memory at the destination). In case an
2911 access exception occurred while walking the MMU tables of the guest, the
2912 ioctl returns a positive error number to indicate the type of exception.
2913 This exception is also raised directly at the corresponding VCPU if the
2914 flag KVM_S390_MEMOP_F_INJECT_EXCEPTION is set in the "flags" field.
2916 The start address of the memory region has to be specified in the "gaddr"
2917 field, and the length of the region in the "size" field. "buf" is the buffer
2918 supplied by the userspace application where the read data should be written
2919 to for KVM_S390_MEMOP_LOGICAL_READ, or where the data that should be written
2920 is stored for a KVM_S390_MEMOP_LOGICAL_WRITE. "buf" is unused and can be NULL
2921 when KVM_S390_MEMOP_F_CHECK_ONLY is specified. "ar" designates the access
2922 register number to be used.
2924 The "reserved" field is meant for future extensions. It is not used by
2925 KVM with the currently defined set of flags.
2927 4.90 KVM_S390_GET_SKEYS
2929 Capability: KVM_CAP_S390_SKEYS
2932 Parameters: struct kvm_s390_skeys
2933 Returns: 0 on success, KVM_S390_GET_KEYS_NONE if guest is not using storage
2934 keys, negative value on error
2936 This ioctl is used to get guest storage key values on the s390
2937 architecture. The ioctl takes parameters via the kvm_s390_skeys struct.
2939 struct kvm_s390_skeys {
2942 __u64 skeydata_addr;
2947 The start_gfn field is the number of the first guest frame whose storage keys
2950 The count field is the number of consecutive frames (starting from start_gfn)
2951 whose storage keys to get. The count field must be at least 1 and the maximum
2952 allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range
2953 will cause the ioctl to return -EINVAL.
2955 The skeydata_addr field is the address to a buffer large enough to hold count
2956 bytes. This buffer will be filled with storage key data by the ioctl.
2958 4.91 KVM_S390_SET_SKEYS
2960 Capability: KVM_CAP_S390_SKEYS
2963 Parameters: struct kvm_s390_skeys
2964 Returns: 0 on success, negative value on error
2966 This ioctl is used to set guest storage key values on the s390
2967 architecture. The ioctl takes parameters via the kvm_s390_skeys struct.
2968 See section on KVM_S390_GET_SKEYS for struct definition.
2970 The start_gfn field is the number of the first guest frame whose storage keys
2973 The count field is the number of consecutive frames (starting from start_gfn)
2974 whose storage keys to get. The count field must be at least 1 and the maximum
2975 allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range
2976 will cause the ioctl to return -EINVAL.
2978 The skeydata_addr field is the address to a buffer containing count bytes of
2979 storage keys. Each byte in the buffer will be set as the storage key for a
2980 single frame starting at start_gfn for count frames.
2982 Note: If any architecturally invalid key value is found in the given data then
2983 the ioctl will return -EINVAL.
2987 Capability: KVM_CAP_S390_INJECT_IRQ
2990 Parameters: struct kvm_s390_irq (in)
2991 Returns: 0 on success, -1 on error
2993 EINVAL: interrupt type is invalid
2994 type is KVM_S390_SIGP_STOP and flag parameter is invalid value
2995 type is KVM_S390_INT_EXTERNAL_CALL and code is bigger
2996 than the maximum of VCPUs
2997 EBUSY: type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped
2998 type is KVM_S390_SIGP_STOP and a stop irq is already pending
2999 type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt
3002 Allows to inject an interrupt to the guest.
3004 Using struct kvm_s390_irq as a parameter allows
3005 to inject additional payload which is not
3006 possible via KVM_S390_INTERRUPT.
3008 Interrupt parameters are passed via kvm_s390_irq:
3010 struct kvm_s390_irq {
3013 struct kvm_s390_io_info io;
3014 struct kvm_s390_ext_info ext;
3015 struct kvm_s390_pgm_info pgm;
3016 struct kvm_s390_emerg_info emerg;
3017 struct kvm_s390_extcall_info extcall;
3018 struct kvm_s390_prefix_info prefix;
3019 struct kvm_s390_stop_info stop;
3020 struct kvm_s390_mchk_info mchk;
3025 type can be one of the following:
3027 KVM_S390_SIGP_STOP - sigp stop; parameter in .stop
3028 KVM_S390_PROGRAM_INT - program check; parameters in .pgm
3029 KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix
3030 KVM_S390_RESTART - restart; no parameters
3031 KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters
3032 KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters
3033 KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg
3034 KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall
3035 KVM_S390_MCHK - machine check interrupt; parameters in .mchk
3038 Note that the vcpu ioctl is asynchronous to vcpu execution.
3040 4.94 KVM_S390_GET_IRQ_STATE
3042 Capability: KVM_CAP_S390_IRQ_STATE
3045 Parameters: struct kvm_s390_irq_state (out)
3046 Returns: >= number of bytes copied into buffer,
3047 -EINVAL if buffer size is 0,
3048 -ENOBUFS if buffer size is too small to fit all pending interrupts,
3049 -EFAULT if the buffer address was invalid
3051 This ioctl allows userspace to retrieve the complete state of all currently
3052 pending interrupts in a single buffer. Use cases include migration
3053 and introspection. The parameter structure contains the address of a
3054 userspace buffer and its length:
3056 struct kvm_s390_irq_state {
3063 Userspace passes in the above struct and for each pending interrupt a
3064 struct kvm_s390_irq is copied to the provided buffer.
3066 If -ENOBUFS is returned the buffer provided was too small and userspace
3067 may retry with a bigger buffer.
3069 4.95 KVM_S390_SET_IRQ_STATE
3071 Capability: KVM_CAP_S390_IRQ_STATE
3074 Parameters: struct kvm_s390_irq_state (in)
3075 Returns: 0 on success,
3076 -EFAULT if the buffer address was invalid,
3077 -EINVAL for an invalid buffer length (see below),
3078 -EBUSY if there were already interrupts pending,
3079 errors occurring when actually injecting the
3080 interrupt. See KVM_S390_IRQ.
3082 This ioctl allows userspace to set the complete state of all cpu-local
3083 interrupts currently pending for the vcpu. It is intended for restoring
3084 interrupt state after a migration. The input parameter is a userspace buffer
3085 containing a struct kvm_s390_irq_state:
3087 struct kvm_s390_irq_state {
3093 The userspace memory referenced by buf contains a struct kvm_s390_irq
3094 for each interrupt to be injected into the guest.
3095 If one of the interrupts could not be injected for some reason the
3098 len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0
3099 and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq),
3100 which is the maximum number of possibly pending cpu-local interrupts.
3104 Capability: KVM_CAP_X86_SMM
3108 Returns: 0 on success, -1 on error
3110 Queues an SMI on the thread's vcpu.
3112 4.97 KVM_CAP_PPC_MULTITCE
3114 Capability: KVM_CAP_PPC_MULTITCE
3118 This capability means the kernel is capable of handling hypercalls
3119 H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user
3120 space. This significantly accelerates DMA operations for PPC KVM guests.
3121 User space should expect that its handlers for these hypercalls
3122 are not going to be called if user space previously registered LIOBN
3123 in KVM (via KVM_CREATE_SPAPR_TCE or similar calls).
3125 In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest,
3126 user space might have to advertise it for the guest. For example,
3127 IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is
3128 present in the "ibm,hypertas-functions" device-tree property.
3130 The hypercalls mentioned above may or may not be processed successfully
3131 in the kernel based fast path. If they can not be handled by the kernel,
3132 they will get passed on to user space. So user space still has to have
3133 an implementation for these despite the in kernel acceleration.
3135 This capability is always enabled.
3137 4.98 KVM_CREATE_SPAPR_TCE_64
3139 Capability: KVM_CAP_SPAPR_TCE_64
3140 Architectures: powerpc
3142 Parameters: struct kvm_create_spapr_tce_64 (in)
3143 Returns: file descriptor for manipulating the created TCE table
3145 This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit
3146 windows, described in 4.62 KVM_CREATE_SPAPR_TCE
3148 This capability uses extended struct in ioctl interface:
3150 /* for KVM_CAP_SPAPR_TCE_64 */
3151 struct kvm_create_spapr_tce_64 {
3155 __u64 offset; /* in pages */
3156 __u64 size; /* in pages */
3159 The aim of extension is to support an additional bigger DMA window with
3160 a variable page size.
3161 KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and
3162 a bus offset of the corresponding DMA window, @size and @offset are numbers
3165 @flags are not used at the moment.
3167 The rest of functionality is identical to KVM_CREATE_SPAPR_TCE.
3169 4.98 KVM_REINJECT_CONTROL
3171 Capability: KVM_CAP_REINJECT_CONTROL
3174 Parameters: struct kvm_reinject_control (in)
3175 Returns: 0 on success,
3176 -EFAULT if struct kvm_reinject_control cannot be read,
3177 -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier.
3179 i8254 (PIT) has two modes, reinject and !reinject. The default is reinject,
3180 where KVM queues elapsed i8254 ticks and monitors completion of interrupt from
3181 vector(s) that i8254 injects. Reinject mode dequeues a tick and injects its
3182 interrupt whenever there isn't a pending interrupt from i8254.
3183 !reinject mode injects an interrupt as soon as a tick arrives.
3185 struct kvm_reinject_control {
3190 pit_reinject = 0 (!reinject mode) is recommended, unless running an old
3191 operating system that uses the PIT for timing (e.g. Linux 2.4.x).
3193 5. The kvm_run structure
3194 ------------------------
3196 Application code obtains a pointer to the kvm_run structure by
3197 mmap()ing a vcpu fd. From that point, application code can control
3198 execution by changing fields in kvm_run prior to calling the KVM_RUN
3199 ioctl, and obtain information about the reason KVM_RUN returned by
3200 looking up structure members.
3204 __u8 request_interrupt_window;
3206 Request that KVM_RUN return when it becomes possible to inject external
3207 interrupts into the guest. Useful in conjunction with KVM_INTERRUPT.
3214 When KVM_RUN has returned successfully (return value 0), this informs
3215 application code why KVM_RUN has returned. Allowable values for this
3216 field are detailed below.
3218 __u8 ready_for_interrupt_injection;
3220 If request_interrupt_window has been specified, this field indicates
3221 an interrupt can be injected now with KVM_INTERRUPT.
3225 The value of the current interrupt flag. Only valid if in-kernel
3226 local APIC is not used.
3230 More architecture-specific flags detailing state of the VCPU that may
3231 affect the device's behavior. The only currently defined flag is
3232 KVM_RUN_X86_SMM, which is valid on x86 machines and is set if the
3233 VCPU is in system management mode.
3235 /* in (pre_kvm_run), out (post_kvm_run) */
3238 The value of the cr8 register. Only valid if in-kernel local APIC is
3239 not used. Both input and output.
3243 The value of the APIC BASE msr. Only valid if in-kernel local
3244 APIC is not used. Both input and output.
3247 /* KVM_EXIT_UNKNOWN */
3249 __u64 hardware_exit_reason;
3252 If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown
3253 reasons. Further architecture-specific information is available in
3254 hardware_exit_reason.
3256 /* KVM_EXIT_FAIL_ENTRY */
3258 __u64 hardware_entry_failure_reason;
3261 If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due
3262 to unknown reasons. Further architecture-specific information is
3263 available in hardware_entry_failure_reason.
3265 /* KVM_EXIT_EXCEPTION */
3275 #define KVM_EXIT_IO_IN 0
3276 #define KVM_EXIT_IO_OUT 1
3278 __u8 size; /* bytes */
3281 __u64 data_offset; /* relative to kvm_run start */
3284 If exit_reason is KVM_EXIT_IO, then the vcpu has
3285 executed a port I/O instruction which could not be satisfied by kvm.
3286 data_offset describes where the data is located (KVM_EXIT_IO_OUT) or
3287 where kvm expects application code to place the data for the next
3288 KVM_RUN invocation (KVM_EXIT_IO_IN). Data format is a packed array.
3290 /* KVM_EXIT_DEBUG */
3292 struct kvm_debug_exit_arch arch;
3295 If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event
3296 for which architecture specific information is returned.
3306 If exit_reason is KVM_EXIT_MMIO, then the vcpu has
3307 executed a memory-mapped I/O instruction which could not be satisfied
3308 by kvm. The 'data' member contains the written data if 'is_write' is
3309 true, and should be filled by application code otherwise.
3311 The 'data' member contains, in its first 'len' bytes, the value as it would
3312 appear if the VCPU performed a load or store of the appropriate width directly
3315 NOTE: For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR and
3316 KVM_EXIT_EPR the corresponding
3317 operations are complete (and guest state is consistent) only after userspace
3318 has re-entered the kernel with KVM_RUN. The kernel side will first finish
3319 incomplete operations and then check for pending signals. Userspace
3320 can re-enter the guest with an unmasked signal pending to complete
3323 /* KVM_EXIT_HYPERCALL */
3332 Unused. This was once used for 'hypercall to userspace'. To implement
3333 such functionality, use KVM_EXIT_IO (x86) or KVM_EXIT_MMIO (all except s390).
3334 Note KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO.
3336 /* KVM_EXIT_TPR_ACCESS */
3343 To be documented (KVM_TPR_ACCESS_REPORTING).
3345 /* KVM_EXIT_S390_SIEIC */
3348 __u64 mask; /* psw upper half */
3349 __u64 addr; /* psw lower half */
3356 /* KVM_EXIT_S390_RESET */
3357 #define KVM_S390_RESET_POR 1
3358 #define KVM_S390_RESET_CLEAR 2
3359 #define KVM_S390_RESET_SUBSYSTEM 4
3360 #define KVM_S390_RESET_CPU_INIT 8
3361 #define KVM_S390_RESET_IPL 16
3362 __u64 s390_reset_flags;
3366 /* KVM_EXIT_S390_UCONTROL */
3368 __u64 trans_exc_code;
3372 s390 specific. A page fault has occurred for a user controlled virtual
3373 machine (KVM_VM_S390_UNCONTROL) on it's host page table that cannot be
3374 resolved by the kernel.
3375 The program code and the translation exception code that were placed
3376 in the cpu's lowcore are presented here as defined by the z Architecture
3377 Principles of Operation Book in the Chapter for Dynamic Address Translation
3387 Deprecated - was used for 440 KVM.
3394 MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch
3395 hypercalls and exit with this exit struct that contains all the guest gprs.
3397 If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall.
3398 Userspace can now handle the hypercall and when it's done modify the gprs as
3399 necessary. Upon guest entry all guest GPRs will then be replaced by the values
3402 /* KVM_EXIT_PAPR_HCALL */
3409 This is used on 64-bit PowerPC when emulating a pSeries partition,
3410 e.g. with the 'pseries' machine type in qemu. It occurs when the
3411 guest does a hypercall using the 'sc 1' instruction. The 'nr' field
3412 contains the hypercall number (from the guest R3), and 'args' contains
3413 the arguments (from the guest R4 - R12). Userspace should put the
3414 return code in 'ret' and any extra returned values in args[].
3415 The possible hypercalls are defined in the Power Architecture Platform
3416 Requirements (PAPR) document available from www.power.org (free
3417 developer registration required to access it).
3419 /* KVM_EXIT_S390_TSCH */
3421 __u16 subchannel_id;
3422 __u16 subchannel_nr;
3429 s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled
3430 and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O
3431 interrupt for the target subchannel has been dequeued and subchannel_id,
3432 subchannel_nr, io_int_parm and io_int_word contain the parameters for that
3433 interrupt. ipb is needed for instruction parameter decoding.
3440 On FSL BookE PowerPC chips, the interrupt controller has a fast patch
3441 interrupt acknowledge path to the core. When the core successfully
3442 delivers an interrupt, it automatically populates the EPR register with
3443 the interrupt vector number and acknowledges the interrupt inside
3444 the interrupt controller.
3446 In case the interrupt controller lives in user space, we need to do
3447 the interrupt acknowledge cycle through it to fetch the next to be
3448 delivered interrupt vector using this exit.
3450 It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an
3451 external interrupt has just been delivered into the guest. User space
3452 should put the acknowledged interrupt vector into the 'epr' field.
3454 /* KVM_EXIT_SYSTEM_EVENT */
3456 #define KVM_SYSTEM_EVENT_SHUTDOWN 1
3457 #define KVM_SYSTEM_EVENT_RESET 2
3458 #define KVM_SYSTEM_EVENT_CRASH 3
3463 If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered
3464 a system-level event using some architecture specific mechanism (hypercall
3465 or some special instruction). In case of ARM/ARM64, this is triggered using
3466 HVC instruction based PSCI call from the vcpu. The 'type' field describes
3467 the system-level event type. The 'flags' field describes architecture
3468 specific flags for the system-level event.
3470 Valid values for 'type' are:
3471 KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the
3472 VM. Userspace is not obliged to honour this, and if it does honour
3473 this does not need to destroy the VM synchronously (ie it may call
3474 KVM_RUN again before shutdown finally occurs).
3475 KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM.
3476 As with SHUTDOWN, userspace can choose to ignore the request, or
3477 to schedule the reset to occur in the future and may call KVM_RUN again.
3478 KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest
3479 has requested a crash condition maintenance. Userspace can choose
3480 to ignore the request, or to gather VM memory core dump and/or
3481 reset/shutdown of the VM.
3483 /* KVM_EXIT_IOAPIC_EOI */
3488 Indicates that the VCPU's in-kernel local APIC received an EOI for a
3489 level-triggered IOAPIC interrupt. This exit only triggers when the
3490 IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled);
3491 the userspace IOAPIC should process the EOI and retrigger the interrupt if
3492 it is still asserted. Vector is the LAPIC interrupt vector for which the
3495 struct kvm_hyperv_exit {
3496 #define KVM_EXIT_HYPERV_SYNIC 1
3497 #define KVM_EXIT_HYPERV_HCALL 2
3513 /* KVM_EXIT_HYPERV */
3514 struct kvm_hyperv_exit hyperv;
3515 Indicates that the VCPU exits into userspace to process some tasks
3516 related to Hyper-V emulation.
3517 Valid values for 'type' are:
3518 KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about
3519 Hyper-V SynIC state change. Notification is used to remap SynIC
3520 event/message pages and to enable/disable SynIC messages/events processing
3523 /* Fix the size of the union. */
3528 * shared registers between kvm and userspace.
3529 * kvm_valid_regs specifies the register classes set by the host
3530 * kvm_dirty_regs specified the register classes dirtied by userspace
3531 * struct kvm_sync_regs is architecture specific, as well as the
3532 * bits for kvm_valid_regs and kvm_dirty_regs
3534 __u64 kvm_valid_regs;
3535 __u64 kvm_dirty_regs;
3537 struct kvm_sync_regs regs;
3541 If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access
3542 certain guest registers without having to call SET/GET_*REGS. Thus we can
3543 avoid some system call overhead if userspace has to handle the exit.
3544 Userspace can query the validity of the structure by checking
3545 kvm_valid_regs for specific bits. These bits are architecture specific
3546 and usually define the validity of a groups of registers. (e.g. one bit
3547 for general purpose registers)
3549 Please note that the kernel is allowed to use the kvm_run structure as the
3550 primary storage for certain register types. Therefore, the kernel may use the
3551 values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set.
3557 6. Capabilities that can be enabled on vCPUs
3558 --------------------------------------------
3560 There are certain capabilities that change the behavior of the virtual CPU or
3561 the virtual machine when enabled. To enable them, please see section 4.37.
3562 Below you can find a list of capabilities and what their effect on the vCPU or
3563 the virtual machine is when enabling them.
3565 The following information is provided along with the description:
3567 Architectures: which instruction set architectures provide this ioctl.
3568 x86 includes both i386 and x86_64.
3570 Target: whether this is a per-vcpu or per-vm capability.
3572 Parameters: what parameters are accepted by the capability.
3574 Returns: the return value. General error numbers (EBADF, ENOMEM, EINVAL)
3575 are not detailed, but errors with specific meanings are.
3583 Returns: 0 on success; -1 on error
3585 This capability enables interception of OSI hypercalls that otherwise would
3586 be treated as normal system calls to be injected into the guest. OSI hypercalls
3587 were invented by Mac-on-Linux to have a standardized communication mechanism
3588 between the guest and the host.
3590 When this capability is enabled, KVM_EXIT_OSI can occur.
3593 6.2 KVM_CAP_PPC_PAPR
3598 Returns: 0 on success; -1 on error
3600 This capability enables interception of PAPR hypercalls. PAPR hypercalls are
3601 done using the hypercall instruction "sc 1".
3603 It also sets the guest privilege level to "supervisor" mode. Usually the guest
3604 runs in "hypervisor" privilege mode with a few missing features.
3606 In addition to the above, it changes the semantics of SDR1. In this mode, the
3607 HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the
3608 HTAB invisible to the guest.
3610 When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur.
3617 Parameters: args[0] is the address of a struct kvm_config_tlb
3618 Returns: 0 on success; -1 on error
3620 struct kvm_config_tlb {
3627 Configures the virtual CPU's TLB array, establishing a shared memory area
3628 between userspace and KVM. The "params" and "array" fields are userspace
3629 addresses of mmu-type-specific data structures. The "array_len" field is an
3630 safety mechanism, and should be set to the size in bytes of the memory that
3631 userspace has reserved for the array. It must be at least the size dictated
3632 by "mmu_type" and "params".
3634 While KVM_RUN is active, the shared region is under control of KVM. Its
3635 contents are undefined, and any modification by userspace results in
3636 boundedly undefined behavior.
3638 On return from KVM_RUN, the shared region will reflect the current state of
3639 the guest's TLB. If userspace makes any changes, it must call KVM_DIRTY_TLB
3640 to tell KVM which entries have been changed, prior to calling KVM_RUN again
3643 For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV:
3644 - The "params" field is of type "struct kvm_book3e_206_tlb_params".
3645 - The "array" field points to an array of type "struct
3646 kvm_book3e_206_tlb_entry".
3647 - The array consists of all entries in the first TLB, followed by all
3648 entries in the second TLB.
3649 - Within a TLB, entries are ordered first by increasing set number. Within a
3650 set, entries are ordered by way (increasing ESEL).
3651 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1)
3652 where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value.
3653 - The tsize field of mas1 shall be set to 4K on TLB0, even though the
3654 hardware ignores this value for TLB0.
3656 6.4 KVM_CAP_S390_CSS_SUPPORT
3661 Returns: 0 on success; -1 on error
3663 This capability enables support for handling of channel I/O instructions.
3665 TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are
3666 handled in-kernel, while the other I/O instructions are passed to userspace.
3668 When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST
3669 SUBCHANNEL intercepts.
3671 Note that even though this capability is enabled per-vcpu, the complete
3672 virtual machine is affected.
3678 Parameters: args[0] defines whether the proxy facility is active
3679 Returns: 0 on success; -1 on error
3681 This capability enables or disables the delivery of interrupts through the
3682 external proxy facility.
3684 When enabled (args[0] != 0), every time the guest gets an external interrupt
3685 delivered, it automatically exits into user space with a KVM_EXIT_EPR exit
3686 to receive the topmost interrupt vector.
3688 When disabled (args[0] == 0), behavior is as if this facility is unsupported.
3690 When this capability is enabled, KVM_EXIT_EPR can occur.
3692 6.6 KVM_CAP_IRQ_MPIC
3695 Parameters: args[0] is the MPIC device fd
3696 args[1] is the MPIC CPU number for this vcpu
3698 This capability connects the vcpu to an in-kernel MPIC device.
3700 6.7 KVM_CAP_IRQ_XICS
3704 Parameters: args[0] is the XICS device fd
3705 args[1] is the XICS CPU number (server ID) for this vcpu
3707 This capability connects the vcpu to an in-kernel XICS device.
3709 6.8 KVM_CAP_S390_IRQCHIP
3715 This capability enables the in-kernel irqchip for s390. Please refer to
3716 "4.24 KVM_CREATE_IRQCHIP" for details.
3718 6.9 KVM_CAP_MIPS_FPU
3722 Parameters: args[0] is reserved for future use (should be 0).
3724 This capability allows the use of the host Floating Point Unit by the guest. It
3725 allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is
3726 done the KVM_REG_MIPS_FPR_* and KVM_REG_MIPS_FCR_* registers can be accessed
3727 (depending on the current guest FPU register mode), and the Status.FR,
3728 Config5.FRE bits are accessible via the KVM API and also from the guest,
3729 depending on them being supported by the FPU.
3731 6.10 KVM_CAP_MIPS_MSA
3735 Parameters: args[0] is reserved for future use (should be 0).
3737 This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest.
3738 It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest.
3739 Once this is done the KVM_REG_MIPS_VEC_* and KVM_REG_MIPS_MSA_* registers can be
3740 accessed, and the Config5.MSAEn bit is accessible via the KVM API and also from
3743 7. Capabilities that can be enabled on VMs
3744 ------------------------------------------
3746 There are certain capabilities that change the behavior of the virtual
3747 machine when enabled. To enable them, please see section 4.37. Below
3748 you can find a list of capabilities and what their effect on the VM
3749 is when enabling them.
3751 The following information is provided along with the description:
3753 Architectures: which instruction set architectures provide this ioctl.
3754 x86 includes both i386 and x86_64.
3756 Parameters: what parameters are accepted by the capability.
3758 Returns: the return value. General error numbers (EBADF, ENOMEM, EINVAL)
3759 are not detailed, but errors with specific meanings are.
3762 7.1 KVM_CAP_PPC_ENABLE_HCALL
3765 Parameters: args[0] is the sPAPR hcall number
3766 args[1] is 0 to disable, 1 to enable in-kernel handling
3768 This capability controls whether individual sPAPR hypercalls (hcalls)
3769 get handled by the kernel or not. Enabling or disabling in-kernel
3770 handling of an hcall is effective across the VM. On creation, an
3771 initial set of hcalls are enabled for in-kernel handling, which
3772 consists of those hcalls for which in-kernel handlers were implemented
3773 before this capability was implemented. If disabled, the kernel will
3774 not to attempt to handle the hcall, but will always exit to userspace
3775 to handle it. Note that it may not make sense to enable some and
3776 disable others of a group of related hcalls, but KVM does not prevent
3777 userspace from doing that.
3779 If the hcall number specified is not one that has an in-kernel
3780 implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL
3783 7.2 KVM_CAP_S390_USER_SIGP
3788 This capability controls which SIGP orders will be handled completely in user
3789 space. With this capability enabled, all fast orders will be handled completely
3795 - CONDITIONAL EMERGENCY SIGNAL
3797 All other orders will be handled completely in user space.
3799 Only privileged operation exceptions will be checked for in the kernel (or even
3800 in the hardware prior to interception). If this capability is not enabled, the
3801 old way of handling SIGP orders is used (partially in kernel and user space).
3803 7.3 KVM_CAP_S390_VECTOR_REGISTERS
3807 Returns: 0 on success, negative value on error
3809 Allows use of the vector registers introduced with z13 processor, and
3810 provides for the synchronization between host and user space. Will
3811 return -EINVAL if the machine does not support vectors.
3813 7.4 KVM_CAP_S390_USER_STSI
3818 This capability allows post-handlers for the STSI instruction. After
3819 initial handling in the kernel, KVM exits to user space with
3820 KVM_EXIT_S390_STSI to allow user space to insert further data.
3822 Before exiting to userspace, kvm handlers should fill in s390_stsi field of
3833 @addr - guest address of STSI SYSIB
3837 @ar - access register number
3839 KVM handlers should exit to userspace with rc = -EREMOTE.
3841 7.5 KVM_CAP_SPLIT_IRQCHIP
3844 Parameters: args[0] - number of routes reserved for userspace IOAPICs
3845 Returns: 0 on success, -1 on error
3847 Create a local apic for each processor in the kernel. This can be used
3848 instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the
3849 IOAPIC and PIC (and also the PIT, even though this has to be enabled
3852 This capability also enables in kernel routing of interrupt requests;
3853 when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are
3854 used in the IRQ routing table. The first args[0] MSI routes are reserved
3855 for the IOAPIC pins. Whenever the LAPIC receives an EOI for these routes,
3856 a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace.
3858 Fails if VCPU has already been created, or if the irqchip is already in the
3859 kernel (i.e. KVM_CREATE_IRQCHIP has already been called).
3866 Allows use of runtime-instrumentation introduced with zEC12 processor.
3867 Will return -EINVAL if the machine does not support runtime-instrumentation.
3868 Will return -EBUSY if a VCPU has already been created.
3870 7.7 KVM_CAP_X2APIC_API
3873 Parameters: args[0] - features that should be enabled
3874 Returns: 0 on success, -EINVAL when args[0] contains invalid features
3876 Valid feature flags in args[0] are
3878 #define KVM_X2APIC_API_USE_32BIT_IDS (1ULL << 0)
3879 #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK (1ULL << 1)
3881 Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of
3882 KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC,
3883 allowing the use of 32-bit APIC IDs. See KVM_CAP_X2APIC_API in their
3884 respective sections.
3886 KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work
3887 in logical mode or with more than 255 VCPUs. Otherwise, KVM treats 0xff
3888 as a broadcast even in x2APIC mode in order to support physical x2APIC
3889 without interrupt remapping. This is undesirable in logical mode,
3890 where 0xff represents CPUs 0-7 in cluster 0.
3892 7.8 KVM_CAP_S390_USER_INSTR0
3897 With this capability enabled, all illegal instructions 0x0000 (2 bytes) will
3898 be intercepted and forwarded to user space. User space can use this
3899 mechanism e.g. to realize 2-byte software breakpoints. The kernel will
3900 not inject an operating exception for these instructions, user space has
3901 to take care of that.
3903 This capability can be enabled dynamically even if VCPUs were already
3904 created and are running.
3906 8. Other capabilities.
3907 ----------------------
3909 This section lists capabilities that give information about other
3910 features of the KVM implementation.
3912 8.1 KVM_CAP_PPC_HWRNG
3916 This capability, if KVM_CHECK_EXTENSION indicates that it is
3917 available, means that that the kernel has an implementation of the
3918 H_RANDOM hypercall backed by a hardware random-number generator.
3919 If present, the kernel H_RANDOM handler can be enabled for guest use
3920 with the KVM_CAP_PPC_ENABLE_HCALL capability.
3922 8.2 KVM_CAP_HYPERV_SYNIC
3925 This capability, if KVM_CHECK_EXTENSION indicates that it is
3926 available, means that that the kernel has an implementation of the
3927 Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is
3928 used to support Windows Hyper-V based guest paravirt drivers(VMBus).
3930 In order to use SynIC, it has to be activated by setting this
3931 capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this
3932 will disable the use of APIC hardware virtualization even if supported
3933 by the CPU, as it's incompatible with SynIC auto-EOI behavior.