1 .. SPDX-License-Identifier: GPL-2.0
3 ===================================================================
4 The Definitive KVM (Kernel-based Virtual Machine) API Documentation
5 ===================================================================
10 The kvm API is a set of ioctls that are issued to control various aspects
11 of a virtual machine. The ioctls belong to the following classes:
13 - System ioctls: These query and set global attributes which affect the
14 whole kvm subsystem. In addition a system ioctl is used to create
17 - VM ioctls: These query and set attributes that affect an entire virtual
18 machine, for example memory layout. In addition a VM ioctl is used to
19 create virtual cpus (vcpus) and devices.
21 VM ioctls must be issued from the same process (address space) that was
22 used to create the VM.
24 - vcpu ioctls: These query and set attributes that control the operation
25 of a single virtual cpu.
27 vcpu ioctls should be issued from the same thread that was used to create
28 the vcpu, except for asynchronous vcpu ioctl that are marked as such in
29 the documentation. Otherwise, the first ioctl after switching threads
30 could see a performance impact.
32 - device ioctls: These query and set attributes that control the operation
35 device ioctls must be issued from the same process (address space) that
36 was used to create the VM.
41 The kvm API is centered around file descriptors. An initial
42 open("/dev/kvm") obtains a handle to the kvm subsystem; this handle
43 can be used to issue system ioctls. A KVM_CREATE_VM ioctl on this
44 handle will create a VM file descriptor which can be used to issue VM
45 ioctls. A KVM_CREATE_VCPU or KVM_CREATE_DEVICE ioctl on a VM fd will
46 create a virtual cpu or device and return a file descriptor pointing to
47 the new resource. Finally, ioctls on a vcpu or device fd can be used
48 to control the vcpu or device. For vcpus, this includes the important
49 task of actually running guest code.
51 In general file descriptors can be migrated among processes by means
52 of fork() and the SCM_RIGHTS facility of unix domain socket. These
53 kinds of tricks are explicitly not supported by kvm. While they will
54 not cause harm to the host, their actual behavior is not guaranteed by
55 the API. See "General description" for details on the ioctl usage
56 model that is supported by KVM.
58 It is important to note that althought VM ioctls may only be issued from
59 the process that created the VM, a VM's lifecycle is associated with its
60 file descriptor, not its creator (process). In other words, the VM and
61 its resources, *including the associated address space*, are not freed
62 until the last reference to the VM's file descriptor has been released.
63 For example, if fork() is issued after ioctl(KVM_CREATE_VM), the VM will
64 not be freed until both the parent (original) process and its child have
65 put their references to the VM's file descriptor.
67 Because a VM's resources are not freed until the last reference to its
68 file descriptor is released, creating additional references to a VM
69 via fork(), dup(), etc... without careful consideration is strongly
70 discouraged and may have unwanted side effects, e.g. memory allocated
71 by and on behalf of the VM's process may not be freed/unaccounted when
78 As of Linux 2.6.22, the KVM ABI has been stabilized: no backward
79 incompatible change are allowed. However, there is an extension
80 facility that allows backward-compatible extensions to the API to be
83 The extension mechanism is not based on the Linux version number.
84 Instead, kvm defines extension identifiers and a facility to query
85 whether a particular extension identifier is available. If it is, a
86 set of ioctls is available for application use.
92 This section describes ioctls that can be used to control kvm guests.
93 For each ioctl, the following information is provided along with a
97 which KVM extension provides this ioctl. Can be 'basic',
98 which means that is will be provided by any kernel that supports
99 API version 12 (see section 4.1), a KVM_CAP_xyz constant, which
100 means availability needs to be checked with KVM_CHECK_EXTENSION
101 (see section 4.4), or 'none' which means that while not all kernels
102 support this ioctl, there's no capability bit to check its
103 availability: for kernels that don't support the ioctl,
104 the ioctl returns -ENOTTY.
107 which instruction set architectures provide this ioctl.
108 x86 includes both i386 and x86_64.
114 what parameters are accepted by the ioctl.
117 the return value. General error numbers (EBADF, ENOMEM, EINVAL)
118 are not detailed, but errors with specific meanings are.
121 4.1 KVM_GET_API_VERSION
122 -----------------------
128 :Returns: the constant KVM_API_VERSION (=12)
130 This identifies the API version as the stable kvm API. It is not
131 expected that this number will change. However, Linux 2.6.20 and
132 2.6.21 report earlier versions; these are not documented and not
133 supported. Applications should refuse to run if KVM_GET_API_VERSION
134 returns a value other than 12. If this check passes, all ioctls
135 described as 'basic' will be available.
144 :Parameters: machine type identifier (KVM_VM_*)
145 :Returns: a VM fd that can be used to control the new virtual machine.
147 The new VM has no virtual cpus and no memory.
148 You probably want to use 0 as machine type.
150 In order to create user controlled virtual machines on S390, check
151 KVM_CAP_S390_UCONTROL and use the flag KVM_VM_S390_UCONTROL as
152 privileged user (CAP_SYS_ADMIN).
154 To use hardware assisted virtualization on MIPS (VZ ASE) rather than
155 the default trap & emulate implementation (which changes the virtual
156 memory layout to fit in user mode), check KVM_CAP_MIPS_VZ and use the
160 On arm64, the physical address size for a VM (IPA Size limit) is limited
161 to 40bits by default. The limit can be configured if the host supports the
162 extension KVM_CAP_ARM_VM_IPA_SIZE. When supported, use
163 KVM_VM_TYPE_ARM_IPA_SIZE(IPA_Bits) to set the size in the machine type
164 identifier, where IPA_Bits is the maximum width of any physical
165 address used by the VM. The IPA_Bits is encoded in bits[7-0] of the
166 machine type identifier.
168 e.g, to configure a guest to use 48bit physical address size::
170 vm_fd = ioctl(dev_fd, KVM_CREATE_VM, KVM_VM_TYPE_ARM_IPA_SIZE(48));
172 The requested size (IPA_Bits) must be:
174 == =========================================================
175 0 Implies default size, 40bits (for backward compatibility)
176 N Implies N bits, where N is a positive integer such that,
177 32 <= N <= Host_IPA_Limit
178 == =========================================================
180 Host_IPA_Limit is the maximum possible value for IPA_Bits on the host and
181 is dependent on the CPU capability and the kernel configuration. The limit can
182 be retrieved using KVM_CAP_ARM_VM_IPA_SIZE of the KVM_CHECK_EXTENSION
185 Please note that configuring the IPA size does not affect the capability
186 exposed by the guest CPUs in ID_AA64MMFR0_EL1[PARange]. It only affects
187 size of the address translated by the stage2 level (guest physical to
188 host physical address translations).
191 4.3 KVM_GET_MSR_INDEX_LIST, KVM_GET_MSR_FEATURE_INDEX_LIST
192 ----------------------------------------------------------
194 :Capability: basic, KVM_CAP_GET_MSR_FEATURES for KVM_GET_MSR_FEATURE_INDEX_LIST
197 :Parameters: struct kvm_msr_list (in/out)
198 :Returns: 0 on success; -1 on error
202 ====== ============================================================
203 EFAULT the msr index list cannot be read from or written to
204 E2BIG the msr index list is to be to fit in the array specified by
206 ====== ============================================================
210 struct kvm_msr_list {
211 __u32 nmsrs; /* number of msrs in entries */
215 The user fills in the size of the indices array in nmsrs, and in return
216 kvm adjusts nmsrs to reflect the actual number of msrs and fills in the
217 indices array with their numbers.
219 KVM_GET_MSR_INDEX_LIST returns the guest msrs that are supported. The list
220 varies by kvm version and host processor, but does not change otherwise.
222 Note: if kvm indicates supports MCE (KVM_CAP_MCE), then the MCE bank MSRs are
223 not returned in the MSR list, as different vcpus can have a different number
224 of banks, as set via the KVM_X86_SETUP_MCE ioctl.
226 KVM_GET_MSR_FEATURE_INDEX_LIST returns the list of MSRs that can be passed
227 to the KVM_GET_MSRS system ioctl. This lets userspace probe host capabilities
228 and processor features that are exposed via MSRs (e.g., VMX capabilities).
229 This list also varies by kvm version and host processor, but does not change
233 4.4 KVM_CHECK_EXTENSION
234 -----------------------
236 :Capability: basic, KVM_CAP_CHECK_EXTENSION_VM for vm ioctl
238 :Type: system ioctl, vm ioctl
239 :Parameters: extension identifier (KVM_CAP_*)
240 :Returns: 0 if unsupported; 1 (or some other positive integer) if supported
242 The API allows the application to query about extensions to the core
243 kvm API. Userspace passes an extension identifier (an integer) and
244 receives an integer that describes the extension availability.
245 Generally 0 means no and 1 means yes, but some extensions may report
246 additional information in the integer return value.
248 Based on their initialization different VMs may have different capabilities.
249 It is thus encouraged to use the vm ioctl to query for capabilities (available
250 with KVM_CAP_CHECK_EXTENSION_VM on the vm fd)
252 4.5 KVM_GET_VCPU_MMAP_SIZE
253 --------------------------
259 :Returns: size of vcpu mmap area, in bytes
261 The KVM_RUN ioctl (cf.) communicates with userspace via a shared
262 memory region. This ioctl returns the size of that region. See the
263 KVM_RUN documentation for details.
266 4.6 KVM_SET_MEMORY_REGION
267 -------------------------
272 :Parameters: struct kvm_memory_region (in)
273 :Returns: 0 on success, -1 on error
275 This ioctl is obsolete and has been removed.
284 :Parameters: vcpu id (apic id on x86)
285 :Returns: vcpu fd on success, -1 on error
287 This API adds a vcpu to a virtual machine. No more than max_vcpus may be added.
288 The vcpu id is an integer in the range [0, max_vcpu_id).
290 The recommended max_vcpus value can be retrieved using the KVM_CAP_NR_VCPUS of
291 the KVM_CHECK_EXTENSION ioctl() at run-time.
292 The maximum possible value for max_vcpus can be retrieved using the
293 KVM_CAP_MAX_VCPUS of the KVM_CHECK_EXTENSION ioctl() at run-time.
295 If the KVM_CAP_NR_VCPUS does not exist, you should assume that max_vcpus is 4
297 If the KVM_CAP_MAX_VCPUS does not exist, you should assume that max_vcpus is
298 same as the value returned from KVM_CAP_NR_VCPUS.
300 The maximum possible value for max_vcpu_id can be retrieved using the
301 KVM_CAP_MAX_VCPU_ID of the KVM_CHECK_EXTENSION ioctl() at run-time.
303 If the KVM_CAP_MAX_VCPU_ID does not exist, you should assume that max_vcpu_id
304 is the same as the value returned from KVM_CAP_MAX_VCPUS.
306 On powerpc using book3s_hv mode, the vcpus are mapped onto virtual
307 threads in one or more virtual CPU cores. (This is because the
308 hardware requires all the hardware threads in a CPU core to be in the
309 same partition.) The KVM_CAP_PPC_SMT capability indicates the number
310 of vcpus per virtual core (vcore). The vcore id is obtained by
311 dividing the vcpu id by the number of vcpus per vcore. The vcpus in a
312 given vcore will always be in the same physical core as each other
313 (though that might be a different physical core from time to time).
314 Userspace can control the threading (SMT) mode of the guest by its
315 allocation of vcpu ids. For example, if userspace wants
316 single-threaded guest vcpus, it should make all vcpu ids be a multiple
317 of the number of vcpus per vcore.
319 For virtual cpus that have been created with S390 user controlled virtual
320 machines, the resulting vcpu fd can be memory mapped at page offset
321 KVM_S390_SIE_PAGE_OFFSET in order to obtain a memory map of the virtual
322 cpu's hardware control block.
325 4.8 KVM_GET_DIRTY_LOG (vm ioctl)
326 --------------------------------
331 :Parameters: struct kvm_dirty_log (in/out)
332 :Returns: 0 on success, -1 on error
336 /* for KVM_GET_DIRTY_LOG */
337 struct kvm_dirty_log {
341 void __user *dirty_bitmap; /* one bit per page */
346 Given a memory slot, return a bitmap containing any pages dirtied
347 since the last call to this ioctl. Bit 0 is the first page in the
348 memory slot. Ensure the entire structure is cleared to avoid padding
351 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 specifies
352 the address space for which you want to return the dirty bitmap.
353 They must be less than the value that KVM_CHECK_EXTENSION returns for
354 the KVM_CAP_MULTI_ADDRESS_SPACE capability.
356 The bits in the dirty bitmap are cleared before the ioctl returns, unless
357 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is enabled. For more information,
358 see the description of the capability.
360 4.9 KVM_SET_MEMORY_ALIAS
361 ------------------------
366 :Parameters: struct kvm_memory_alias (in)
367 :Returns: 0 (success), -1 (error)
369 This ioctl is obsolete and has been removed.
379 :Returns: 0 on success, -1 on error
383 ===== =============================
384 EINTR an unmasked signal is pending
385 ===== =============================
387 This ioctl is used to run a guest virtual cpu. While there are no
388 explicit parameters, there is an implicit parameter block that can be
389 obtained by mmap()ing the vcpu fd at offset 0, with the size given by
390 KVM_GET_VCPU_MMAP_SIZE. The parameter block is formatted as a 'struct
391 kvm_run' (see below).
398 :Architectures: all except ARM, arm64
400 :Parameters: struct kvm_regs (out)
401 :Returns: 0 on success, -1 on error
403 Reads the general purpose registers from the vcpu.
409 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
410 __u64 rax, rbx, rcx, rdx;
411 __u64 rsi, rdi, rsp, rbp;
412 __u64 r8, r9, r10, r11;
413 __u64 r12, r13, r14, r15;
419 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
431 :Architectures: all except ARM, arm64
433 :Parameters: struct kvm_regs (in)
434 :Returns: 0 on success, -1 on error
436 Writes the general purpose registers into the vcpu.
438 See KVM_GET_REGS for the data structure.
445 :Architectures: x86, ppc
447 :Parameters: struct kvm_sregs (out)
448 :Returns: 0 on success, -1 on error
450 Reads special registers from the vcpu.
456 struct kvm_segment cs, ds, es, fs, gs, ss;
457 struct kvm_segment tr, ldt;
458 struct kvm_dtable gdt, idt;
459 __u64 cr0, cr2, cr3, cr4, cr8;
462 __u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64];
465 /* ppc -- see arch/powerpc/include/uapi/asm/kvm.h */
467 interrupt_bitmap is a bitmap of pending external interrupts. At most
468 one bit may be set. This interrupt has been acknowledged by the APIC
469 but not yet injected into the cpu core.
476 :Architectures: x86, ppc
478 :Parameters: struct kvm_sregs (in)
479 :Returns: 0 on success, -1 on error
481 Writes special registers into the vcpu. See KVM_GET_SREGS for the
491 :Parameters: struct kvm_translation (in/out)
492 :Returns: 0 on success, -1 on error
494 Translates a virtual address according to the vcpu's current address
499 struct kvm_translation {
501 __u64 linear_address;
504 __u64 physical_address;
516 :Architectures: x86, ppc, mips
518 :Parameters: struct kvm_interrupt (in)
519 :Returns: 0 on success, negative on failure.
521 Queues a hardware interrupt vector to be injected.
525 /* for KVM_INTERRUPT */
526 struct kvm_interrupt {
536 ========= ===================================
538 -EEXIST if an interrupt is already enqueued
539 -EINVAL the irq number is invalid
540 -ENXIO if the PIC is in the kernel
541 -EFAULT if the pointer is invalid
542 ========= ===================================
544 Note 'irq' is an interrupt vector, not an interrupt pin or line. This
545 ioctl is useful if the in-kernel PIC is not used.
550 Queues an external interrupt to be injected. This ioctl is overleaded
551 with 3 different irq values:
555 This injects an edge type external interrupt into the guest once it's ready
556 to receive interrupts. When injected, the interrupt is done.
558 b) KVM_INTERRUPT_UNSET
560 This unsets any pending interrupt.
562 Only available with KVM_CAP_PPC_UNSET_IRQ.
564 c) KVM_INTERRUPT_SET_LEVEL
566 This injects a level type external interrupt into the guest context. The
567 interrupt stays pending until a specific ioctl with KVM_INTERRUPT_UNSET
570 Only available with KVM_CAP_PPC_IRQ_LEVEL.
572 Note that any value for 'irq' other than the ones stated above is invalid
573 and incurs unexpected behavior.
575 This is an asynchronous vcpu ioctl and can be invoked from any thread.
580 Queues an external interrupt to be injected into the virtual CPU. A negative
581 interrupt number dequeues the interrupt.
583 This is an asynchronous vcpu ioctl and can be invoked from any thread.
593 :Returns: -1 on error
595 Support for this has been removed. Use KVM_SET_GUEST_DEBUG instead.
601 :Capability: basic (vcpu), KVM_CAP_GET_MSR_FEATURES (system)
603 :Type: system ioctl, vcpu ioctl
604 :Parameters: struct kvm_msrs (in/out)
605 :Returns: number of msrs successfully returned;
608 When used as a system ioctl:
609 Reads the values of MSR-based features that are available for the VM. This
610 is similar to KVM_GET_SUPPORTED_CPUID, but it returns MSR indices and values.
611 The list of msr-based features can be obtained using KVM_GET_MSR_FEATURE_INDEX_LIST
614 When used as a vcpu ioctl:
615 Reads model-specific registers from the vcpu. Supported msr indices can
616 be obtained using KVM_GET_MSR_INDEX_LIST in a system ioctl.
621 __u32 nmsrs; /* number of msrs in entries */
624 struct kvm_msr_entry entries[0];
627 struct kvm_msr_entry {
633 Application code should set the 'nmsrs' member (which indicates the
634 size of the entries array) and the 'index' member of each array entry.
635 kvm will fill in the 'data' member.
644 :Parameters: struct kvm_msrs (in)
645 :Returns: number of msrs successfully set (see below), -1 on error
647 Writes model-specific registers to the vcpu. See KVM_GET_MSRS for the
650 Application code should set the 'nmsrs' member (which indicates the
651 size of the entries array), and the 'index' and 'data' members of each
654 It tries to set the MSRs in array entries[] one by one. If setting an MSR
655 fails, e.g., due to setting reserved bits, the MSR isn't supported/emulated
656 by KVM, etc..., it stops processing the MSR list and returns the number of
657 MSRs that have been set successfully.
666 :Parameters: struct kvm_cpuid (in)
667 :Returns: 0 on success, -1 on error
669 Defines the vcpu responses to the cpuid instruction. Applications
670 should use the KVM_SET_CPUID2 ioctl if available.
672 Note, when this IOCTL fails, KVM gives no guarantees that previous valid CPUID
673 configuration (if there is) is not corrupted. Userspace can get a copy of the
674 resulting CPUID configuration through KVM_GET_CPUID2 in case.
678 struct kvm_cpuid_entry {
687 /* for KVM_SET_CPUID */
691 struct kvm_cpuid_entry entries[0];
695 4.21 KVM_SET_SIGNAL_MASK
696 ------------------------
701 :Parameters: struct kvm_signal_mask (in)
702 :Returns: 0 on success, -1 on error
704 Defines which signals are blocked during execution of KVM_RUN. This
705 signal mask temporarily overrides the threads signal mask. Any
706 unblocked signal received (except SIGKILL and SIGSTOP, which retain
707 their traditional behaviour) will cause KVM_RUN to return with -EINTR.
709 Note the signal will only be delivered if not blocked by the original
714 /* for KVM_SET_SIGNAL_MASK */
715 struct kvm_signal_mask {
727 :Parameters: struct kvm_fpu (out)
728 :Returns: 0 on success, -1 on error
730 Reads the floating point state from the vcpu.
734 /* for KVM_GET_FPU and KVM_SET_FPU */
739 __u8 ftwx; /* in fxsave format */
756 :Parameters: struct kvm_fpu (in)
757 :Returns: 0 on success, -1 on error
759 Writes the floating point state to the vcpu.
763 /* for KVM_GET_FPU and KVM_SET_FPU */
768 __u8 ftwx; /* in fxsave format */
779 4.24 KVM_CREATE_IRQCHIP
780 -----------------------
782 :Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQCHIP (s390)
783 :Architectures: x86, ARM, arm64, s390
786 :Returns: 0 on success, -1 on error
788 Creates an interrupt controller model in the kernel.
789 On x86, creates a virtual ioapic, a virtual PIC (two PICs, nested), and sets up
790 future vcpus to have a local APIC. IRQ routing for GSIs 0-15 is set to both
791 PIC and IOAPIC; GSI 16-23 only go to the IOAPIC.
792 On ARM/arm64, a GICv2 is created. Any other GIC versions require the usage of
793 KVM_CREATE_DEVICE, which also supports creating a GICv2. Using
794 KVM_CREATE_DEVICE is preferred over KVM_CREATE_IRQCHIP for GICv2.
795 On s390, a dummy irq routing table is created.
797 Note that on s390 the KVM_CAP_S390_IRQCHIP vm capability needs to be enabled
798 before KVM_CREATE_IRQCHIP can be used.
804 :Capability: KVM_CAP_IRQCHIP
805 :Architectures: x86, arm, arm64
807 :Parameters: struct kvm_irq_level
808 :Returns: 0 on success, -1 on error
810 Sets the level of a GSI input to the interrupt controller model in the kernel.
811 On some architectures it is required that an interrupt controller model has
812 been previously created with KVM_CREATE_IRQCHIP. Note that edge-triggered
813 interrupts require the level to be set to 1 and then back to 0.
815 On real hardware, interrupt pins can be active-low or active-high. This
816 does not matter for the level field of struct kvm_irq_level: 1 always
817 means active (asserted), 0 means inactive (deasserted).
819 x86 allows the operating system to program the interrupt polarity
820 (active-low/active-high) for level-triggered interrupts, and KVM used
821 to consider the polarity. However, due to bitrot in the handling of
822 active-low interrupts, the above convention is now valid on x86 too.
823 This is signaled by KVM_CAP_X86_IOAPIC_POLARITY_IGNORED. Userspace
824 should not present interrupts to the guest as active-low unless this
825 capability is present (or unless it is not using the in-kernel irqchip,
829 ARM/arm64 can signal an interrupt either at the CPU level, or at the
830 in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to
831 use PPIs designated for specific cpus. The irq field is interpreted
834 Â bits: | 31 ... 28 | 27 ... 24 | 23 ... 16 | 15 ... 0 |
835 field: | vcpu2_index | irq_type | vcpu_index | irq_id |
837 The irq_type field has the following values:
840 out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ
842 in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.)
843 (the vcpu_index field is ignored)
845 in-kernel GIC: PPI, irq_id between 16 and 31 (incl.)
847 (The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs)
849 In both cases, level is used to assert/deassert the line.
851 When KVM_CAP_ARM_IRQ_LINE_LAYOUT_2 is supported, the target vcpu is
852 identified as (256 * vcpu2_index + vcpu_index). Otherwise, vcpu2_index
855 Note that on arm/arm64, the KVM_CAP_IRQCHIP capability only conditions
856 injection of interrupts for the in-kernel irqchip. KVM_IRQ_LINE can always
857 be used for a userspace interrupt controller.
861 struct kvm_irq_level {
864 __s32 status; /* not used for KVM_IRQ_LEVEL */
866 __u32 level; /* 0 or 1 */
873 :Capability: KVM_CAP_IRQCHIP
876 :Parameters: struct kvm_irqchip (in/out)
877 :Returns: 0 on success, -1 on error
879 Reads the state of a kernel interrupt controller created with
880 KVM_CREATE_IRQCHIP into a buffer provided by the caller.
885 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
888 char dummy[512]; /* reserving space */
889 struct kvm_pic_state pic;
890 struct kvm_ioapic_state ioapic;
898 :Capability: KVM_CAP_IRQCHIP
901 :Parameters: struct kvm_irqchip (in)
902 :Returns: 0 on success, -1 on error
904 Sets the state of a kernel interrupt controller created with
905 KVM_CREATE_IRQCHIP from a buffer provided by the caller.
910 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
913 char dummy[512]; /* reserving space */
914 struct kvm_pic_state pic;
915 struct kvm_ioapic_state ioapic;
920 4.28 KVM_XEN_HVM_CONFIG
921 -----------------------
923 :Capability: KVM_CAP_XEN_HVM
926 :Parameters: struct kvm_xen_hvm_config (in)
927 :Returns: 0 on success, -1 on error
929 Sets the MSR that the Xen HVM guest uses to initialize its hypercall
930 page, and provides the starting address and size of the hypercall
931 blobs in userspace. When the guest writes the MSR, kvm copies one
932 page of a blob (32- or 64-bit, depending on the vcpu mode) to guest
937 struct kvm_xen_hvm_config {
951 :Capability: KVM_CAP_ADJUST_CLOCK
954 :Parameters: struct kvm_clock_data (out)
955 :Returns: 0 on success, -1 on error
957 Gets the current timestamp of kvmclock as seen by the current guest. In
958 conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios
961 When KVM_CAP_ADJUST_CLOCK is passed to KVM_CHECK_EXTENSION, it returns the
962 set of bits that KVM can return in struct kvm_clock_data's flag member.
964 The only flag defined now is KVM_CLOCK_TSC_STABLE. If set, the returned
965 value is the exact kvmclock value seen by all VCPUs at the instant
966 when KVM_GET_CLOCK was called. If clear, the returned value is simply
967 CLOCK_MONOTONIC plus a constant offset; the offset can be modified
968 with KVM_SET_CLOCK. KVM will try to make all VCPUs follow this clock,
969 but the exact value read by each VCPU could differ, because the host
974 struct kvm_clock_data {
975 __u64 clock; /* kvmclock current value */
984 :Capability: KVM_CAP_ADJUST_CLOCK
987 :Parameters: struct kvm_clock_data (in)
988 :Returns: 0 on success, -1 on error
990 Sets the current timestamp of kvmclock to the value specified in its parameter.
991 In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios
996 struct kvm_clock_data {
997 __u64 clock; /* kvmclock current value */
1003 4.31 KVM_GET_VCPU_EVENTS
1004 ------------------------
1006 :Capability: KVM_CAP_VCPU_EVENTS
1007 :Extended by: KVM_CAP_INTR_SHADOW
1008 :Architectures: x86, arm, arm64
1010 :Parameters: struct kvm_vcpu_event (out)
1011 :Returns: 0 on success, -1 on error
1016 Gets currently pending exceptions, interrupts, and NMIs as well as related
1021 struct kvm_vcpu_events {
1025 __u8 has_error_code;
1046 __u8 smm_inside_nmi;
1050 __u8 exception_has_payload;
1051 __u64 exception_payload;
1054 The following bits are defined in the flags field:
1056 - KVM_VCPUEVENT_VALID_SHADOW may be set to signal that
1057 interrupt.shadow contains a valid state.
1059 - KVM_VCPUEVENT_VALID_SMM may be set to signal that smi contains a
1062 - KVM_VCPUEVENT_VALID_PAYLOAD may be set to signal that the
1063 exception_has_payload, exception_payload, and exception.pending
1064 fields contain a valid state. This bit will be set whenever
1065 KVM_CAP_EXCEPTION_PAYLOAD is enabled.
1070 If the guest accesses a device that is being emulated by the host kernel in
1071 such a way that a real device would generate a physical SError, KVM may make
1072 a virtual SError pending for that VCPU. This system error interrupt remains
1073 pending until the guest takes the exception by unmasking PSTATE.A.
1075 Running the VCPU may cause it to take a pending SError, or make an access that
1076 causes an SError to become pending. The event's description is only valid while
1077 the VPCU is not running.
1079 This API provides a way to read and write the pending 'event' state that is not
1080 visible to the guest. To save, restore or migrate a VCPU the struct representing
1081 the state can be read then written using this GET/SET API, along with the other
1082 guest-visible registers. It is not possible to 'cancel' an SError that has been
1085 A device being emulated in user-space may also wish to generate an SError. To do
1086 this the events structure can be populated by user-space. The current state
1087 should be read first, to ensure no existing SError is pending. If an existing
1088 SError is pending, the architecture's 'Multiple SError interrupts' rules should
1089 be followed. (2.5.3 of DDI0587.a "ARM Reliability, Availability, and
1090 Serviceability (RAS) Specification").
1092 SError exceptions always have an ESR value. Some CPUs have the ability to
1093 specify what the virtual SError's ESR value should be. These systems will
1094 advertise KVM_CAP_ARM_INJECT_SERROR_ESR. In this case exception.has_esr will
1095 always have a non-zero value when read, and the agent making an SError pending
1096 should specify the ISS field in the lower 24 bits of exception.serror_esr. If
1097 the system supports KVM_CAP_ARM_INJECT_SERROR_ESR, but user-space sets the events
1098 with exception.has_esr as zero, KVM will choose an ESR.
1100 Specifying exception.has_esr on a system that does not support it will return
1101 -EINVAL. Setting anything other than the lower 24bits of exception.serror_esr
1102 will return -EINVAL.
1104 It is not possible to read back a pending external abort (injected via
1105 KVM_SET_VCPU_EVENTS or otherwise) because such an exception is always delivered
1106 directly to the virtual CPU).
1110 struct kvm_vcpu_events {
1112 __u8 serror_pending;
1113 __u8 serror_has_esr;
1114 __u8 ext_dabt_pending;
1115 /* Align it to 8 bytes */
1122 4.32 KVM_SET_VCPU_EVENTS
1123 ------------------------
1125 :Capability: KVM_CAP_VCPU_EVENTS
1126 :Extended by: KVM_CAP_INTR_SHADOW
1127 :Architectures: x86, arm, arm64
1129 :Parameters: struct kvm_vcpu_event (in)
1130 :Returns: 0 on success, -1 on error
1135 Set pending exceptions, interrupts, and NMIs as well as related states of the
1138 See KVM_GET_VCPU_EVENTS for the data structure.
1140 Fields that may be modified asynchronously by running VCPUs can be excluded
1141 from the update. These fields are nmi.pending, sipi_vector, smi.smm,
1142 smi.pending. Keep the corresponding bits in the flags field cleared to
1143 suppress overwriting the current in-kernel state. The bits are:
1145 =============================== ==================================
1146 KVM_VCPUEVENT_VALID_NMI_PENDING transfer nmi.pending to the kernel
1147 KVM_VCPUEVENT_VALID_SIPI_VECTOR transfer sipi_vector
1148 KVM_VCPUEVENT_VALID_SMM transfer the smi sub-struct.
1149 =============================== ==================================
1151 If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in
1152 the flags field to signal that interrupt.shadow contains a valid state and
1153 shall be written into the VCPU.
1155 KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available.
1157 If KVM_CAP_EXCEPTION_PAYLOAD is enabled, KVM_VCPUEVENT_VALID_PAYLOAD
1158 can be set in the flags field to signal that the
1159 exception_has_payload, exception_payload, and exception.pending fields
1160 contain a valid state and shall be written into the VCPU.
1165 User space may need to inject several types of events to the guest.
1167 Set the pending SError exception state for this VCPU. It is not possible to
1168 'cancel' an Serror that has been made pending.
1170 If the guest performed an access to I/O memory which could not be handled by
1171 userspace, for example because of missing instruction syndrome decode
1172 information or because there is no device mapped at the accessed IPA, then
1173 userspace can ask the kernel to inject an external abort using the address
1174 from the exiting fault on the VCPU. It is a programming error to set
1175 ext_dabt_pending after an exit which was not either KVM_EXIT_MMIO or
1176 KVM_EXIT_ARM_NISV. This feature is only available if the system supports
1177 KVM_CAP_ARM_INJECT_EXT_DABT. This is a helper which provides commonality in
1178 how userspace reports accesses for the above cases to guests, across different
1179 userspace implementations. Nevertheless, userspace can still emulate all Arm
1180 exceptions by manipulating individual registers using the KVM_SET_ONE_REG API.
1182 See KVM_GET_VCPU_EVENTS for the data structure.
1185 4.33 KVM_GET_DEBUGREGS
1186 ----------------------
1188 :Capability: KVM_CAP_DEBUGREGS
1191 :Parameters: struct kvm_debugregs (out)
1192 :Returns: 0 on success, -1 on error
1194 Reads debug registers from the vcpu.
1198 struct kvm_debugregs {
1207 4.34 KVM_SET_DEBUGREGS
1208 ----------------------
1210 :Capability: KVM_CAP_DEBUGREGS
1213 :Parameters: struct kvm_debugregs (in)
1214 :Returns: 0 on success, -1 on error
1216 Writes debug registers into the vcpu.
1218 See KVM_GET_DEBUGREGS for the data structure. The flags field is unused
1219 yet and must be cleared on entry.
1222 4.35 KVM_SET_USER_MEMORY_REGION
1223 -------------------------------
1225 :Capability: KVM_CAP_USER_MEMORY
1228 :Parameters: struct kvm_userspace_memory_region (in)
1229 :Returns: 0 on success, -1 on error
1233 struct kvm_userspace_memory_region {
1236 __u64 guest_phys_addr;
1237 __u64 memory_size; /* bytes */
1238 __u64 userspace_addr; /* start of the userspace allocated memory */
1241 /* for kvm_memory_region::flags */
1242 #define KVM_MEM_LOG_DIRTY_PAGES (1UL << 0)
1243 #define KVM_MEM_READONLY (1UL << 1)
1245 This ioctl allows the user to create, modify or delete a guest physical
1246 memory slot. Bits 0-15 of "slot" specify the slot id and this value
1247 should be less than the maximum number of user memory slots supported per
1248 VM. The maximum allowed slots can be queried using KVM_CAP_NR_MEMSLOTS.
1249 Slots may not overlap in guest physical address space.
1251 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot"
1252 specifies the address space which is being modified. They must be
1253 less than the value that KVM_CHECK_EXTENSION returns for the
1254 KVM_CAP_MULTI_ADDRESS_SPACE capability. Slots in separate address spaces
1255 are unrelated; the restriction on overlapping slots only applies within
1258 Deleting a slot is done by passing zero for memory_size. When changing
1259 an existing slot, it may be moved in the guest physical memory space,
1260 or its flags may be modified, but it may not be resized.
1262 Memory for the region is taken starting at the address denoted by the
1263 field userspace_addr, which must point at user addressable memory for
1264 the entire memory slot size. Any object may back this memory, including
1265 anonymous memory, ordinary files, and hugetlbfs.
1267 It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr
1268 be identical. This allows large pages in the guest to be backed by large
1271 The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and
1272 KVM_MEM_READONLY. The former can be set to instruct KVM to keep track of
1273 writes to memory within the slot. See KVM_GET_DIRTY_LOG ioctl to know how to
1274 use it. The latter can be set, if KVM_CAP_READONLY_MEM capability allows it,
1275 to make a new slot read-only. In this case, writes to this memory will be
1276 posted to userspace as KVM_EXIT_MMIO exits.
1278 When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of
1279 the memory region are automatically reflected into the guest. For example, an
1280 mmap() that affects the region will be made visible immediately. Another
1281 example is madvise(MADV_DROP).
1283 It is recommended to use this API instead of the KVM_SET_MEMORY_REGION ioctl.
1284 The KVM_SET_MEMORY_REGION does not allow fine grained control over memory
1285 allocation and is deprecated.
1288 4.36 KVM_SET_TSS_ADDR
1289 ---------------------
1291 :Capability: KVM_CAP_SET_TSS_ADDR
1294 :Parameters: unsigned long tss_address (in)
1295 :Returns: 0 on success, -1 on error
1297 This ioctl defines the physical address of a three-page region in the guest
1298 physical address space. The region must be within the first 4GB of the
1299 guest physical address space and must not conflict with any memory slot
1300 or any mmio address. The guest may malfunction if it accesses this memory
1303 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1304 because of a quirk in the virtualization implementation (see the internals
1305 documentation when it pops into existence).
1311 :Capability: KVM_CAP_ENABLE_CAP
1312 :Architectures: mips, ppc, s390
1314 :Parameters: struct kvm_enable_cap (in)
1315 :Returns: 0 on success; -1 on error
1317 :Capability: KVM_CAP_ENABLE_CAP_VM
1320 :Parameters: struct kvm_enable_cap (in)
1321 :Returns: 0 on success; -1 on error
1325 Not all extensions are enabled by default. Using this ioctl the application
1326 can enable an extension, making it available to the guest.
1328 On systems that do not support this ioctl, it always fails. On systems that
1329 do support it, it only works for extensions that are supported for enablement.
1331 To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should
1336 struct kvm_enable_cap {
1340 The capability that is supposed to get enabled.
1346 A bitfield indicating future enhancements. Has to be 0 for now.
1352 Arguments for enabling a feature. If a feature needs initial values to
1353 function properly, this is the place to put them.
1360 The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl
1361 for vm-wide capabilities.
1363 4.38 KVM_GET_MP_STATE
1364 ---------------------
1366 :Capability: KVM_CAP_MP_STATE
1367 :Architectures: x86, s390, arm, arm64
1369 :Parameters: struct kvm_mp_state (out)
1370 :Returns: 0 on success; -1 on error
1374 struct kvm_mp_state {
1378 Returns the vcpu's current "multiprocessing state" (though also valid on
1379 uniprocessor guests).
1381 Possible values are:
1383 ========================== ===============================================
1384 KVM_MP_STATE_RUNNABLE the vcpu is currently running [x86,arm/arm64]
1385 KVM_MP_STATE_UNINITIALIZED the vcpu is an application processor (AP)
1386 which has not yet received an INIT signal [x86]
1387 KVM_MP_STATE_INIT_RECEIVED the vcpu has received an INIT signal, and is
1388 now ready for a SIPI [x86]
1389 KVM_MP_STATE_HALTED the vcpu has executed a HLT instruction and
1390 is waiting for an interrupt [x86]
1391 KVM_MP_STATE_SIPI_RECEIVED the vcpu has just received a SIPI (vector
1392 accessible via KVM_GET_VCPU_EVENTS) [x86]
1393 KVM_MP_STATE_STOPPED the vcpu is stopped [s390,arm/arm64]
1394 KVM_MP_STATE_CHECK_STOP the vcpu is in a special error state [s390]
1395 KVM_MP_STATE_OPERATING the vcpu is operating (running or halted)
1397 KVM_MP_STATE_LOAD the vcpu is in a special load/startup state
1399 ========================== ===============================================
1401 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1402 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1403 these architectures.
1408 The only states that are valid are KVM_MP_STATE_STOPPED and
1409 KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not.
1411 4.39 KVM_SET_MP_STATE
1412 ---------------------
1414 :Capability: KVM_CAP_MP_STATE
1415 :Architectures: x86, s390, arm, arm64
1417 :Parameters: struct kvm_mp_state (in)
1418 :Returns: 0 on success; -1 on error
1420 Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for
1423 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1424 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1425 these architectures.
1430 The only states that are valid are KVM_MP_STATE_STOPPED and
1431 KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not.
1433 4.40 KVM_SET_IDENTITY_MAP_ADDR
1434 ------------------------------
1436 :Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR
1439 :Parameters: unsigned long identity (in)
1440 :Returns: 0 on success, -1 on error
1442 This ioctl defines the physical address of a one-page region in the guest
1443 physical address space. The region must be within the first 4GB of the
1444 guest physical address space and must not conflict with any memory slot
1445 or any mmio address. The guest may malfunction if it accesses this memory
1448 Setting the address to 0 will result in resetting the address to its default
1451 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1452 because of a quirk in the virtualization implementation (see the internals
1453 documentation when it pops into existence).
1455 Fails if any VCPU has already been created.
1457 4.41 KVM_SET_BOOT_CPU_ID
1458 ------------------------
1460 :Capability: KVM_CAP_SET_BOOT_CPU_ID
1463 :Parameters: unsigned long vcpu_id
1464 :Returns: 0 on success, -1 on error
1466 Define which vcpu is the Bootstrap Processor (BSP). Values are the same
1467 as the vcpu id in KVM_CREATE_VCPU. If this ioctl is not called, the default
1474 :Capability: KVM_CAP_XSAVE
1477 :Parameters: struct kvm_xsave (out)
1478 :Returns: 0 on success, -1 on error
1487 This ioctl would copy current vcpu's xsave struct to the userspace.
1493 :Capability: KVM_CAP_XSAVE
1496 :Parameters: struct kvm_xsave (in)
1497 :Returns: 0 on success, -1 on error
1506 This ioctl would copy userspace's xsave struct to the kernel.
1512 :Capability: KVM_CAP_XCRS
1515 :Parameters: struct kvm_xcrs (out)
1516 :Returns: 0 on success, -1 on error
1529 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1533 This ioctl would copy current vcpu's xcrs to the userspace.
1539 :Capability: KVM_CAP_XCRS
1542 :Parameters: struct kvm_xcrs (in)
1543 :Returns: 0 on success, -1 on error
1556 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1560 This ioctl would set vcpu's xcr to the value userspace specified.
1563 4.46 KVM_GET_SUPPORTED_CPUID
1564 ----------------------------
1566 :Capability: KVM_CAP_EXT_CPUID
1569 :Parameters: struct kvm_cpuid2 (in/out)
1570 :Returns: 0 on success, -1 on error
1577 struct kvm_cpuid_entry2 entries[0];
1580 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
1581 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */
1582 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */
1584 struct kvm_cpuid_entry2 {
1595 This ioctl returns x86 cpuid features which are supported by both the
1596 hardware and kvm in its default configuration. Userspace can use the
1597 information returned by this ioctl to construct cpuid information (for
1598 KVM_SET_CPUID2) that is consistent with hardware, kernel, and
1599 userspace capabilities, and with user requirements (for example, the
1600 user may wish to constrain cpuid to emulate older hardware, or for
1601 feature consistency across a cluster).
1603 Note that certain capabilities, such as KVM_CAP_X86_DISABLE_EXITS, may
1604 expose cpuid features (e.g. MONITOR) which are not supported by kvm in
1605 its default configuration. If userspace enables such capabilities, it
1606 is responsible for modifying the results of this ioctl appropriately.
1608 Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure
1609 with the 'nent' field indicating the number of entries in the variable-size
1610 array 'entries'. If the number of entries is too low to describe the cpu
1611 capabilities, an error (E2BIG) is returned. If the number is too high,
1612 the 'nent' field is adjusted and an error (ENOMEM) is returned. If the
1613 number is just right, the 'nent' field is adjusted to the number of valid
1614 entries in the 'entries' array, which is then filled.
1616 The entries returned are the host cpuid as returned by the cpuid instruction,
1617 with unknown or unsupported features masked out. Some features (for example,
1618 x2apic), may not be present in the host cpu, but are exposed by kvm if it can
1619 emulate them efficiently. The fields in each entry are defined as follows:
1622 the eax value used to obtain the entry
1625 the ecx value used to obtain the entry (for entries that are
1629 an OR of zero or more of the following:
1631 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
1632 if the index field is valid
1635 the values returned by the cpuid instruction for
1636 this function/index combination
1638 The TSC deadline timer feature (CPUID leaf 1, ecx[24]) is always returned
1639 as false, since the feature depends on KVM_CREATE_IRQCHIP for local APIC
1640 support. Instead it is reported via::
1642 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER)
1644 if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the
1645 feature in userspace, then you can enable the feature for KVM_SET_CPUID2.
1648 4.47 KVM_PPC_GET_PVINFO
1649 -----------------------
1651 :Capability: KVM_CAP_PPC_GET_PVINFO
1654 :Parameters: struct kvm_ppc_pvinfo (out)
1655 :Returns: 0 on success, !0 on error
1659 struct kvm_ppc_pvinfo {
1665 This ioctl fetches PV specific information that need to be passed to the guest
1666 using the device tree or other means from vm context.
1668 The hcall array defines 4 instructions that make up a hypercall.
1670 If any additional field gets added to this structure later on, a bit for that
1671 additional piece of information will be set in the flags bitmap.
1673 The flags bitmap is defined as::
1675 /* the host supports the ePAPR idle hcall
1676 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1<<0)
1678 4.52 KVM_SET_GSI_ROUTING
1679 ------------------------
1681 :Capability: KVM_CAP_IRQ_ROUTING
1682 :Architectures: x86 s390 arm arm64
1684 :Parameters: struct kvm_irq_routing (in)
1685 :Returns: 0 on success, -1 on error
1687 Sets the GSI routing table entries, overwriting any previously set entries.
1689 On arm/arm64, GSI routing has the following limitation:
1691 - GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD.
1695 struct kvm_irq_routing {
1698 struct kvm_irq_routing_entry entries[0];
1701 No flags are specified so far, the corresponding field must be set to zero.
1705 struct kvm_irq_routing_entry {
1711 struct kvm_irq_routing_irqchip irqchip;
1712 struct kvm_irq_routing_msi msi;
1713 struct kvm_irq_routing_s390_adapter adapter;
1714 struct kvm_irq_routing_hv_sint hv_sint;
1719 /* gsi routing entry types */
1720 #define KVM_IRQ_ROUTING_IRQCHIP 1
1721 #define KVM_IRQ_ROUTING_MSI 2
1722 #define KVM_IRQ_ROUTING_S390_ADAPTER 3
1723 #define KVM_IRQ_ROUTING_HV_SINT 4
1727 - KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry
1728 type, specifies that the devid field contains a valid value. The per-VM
1729 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
1730 the device ID. If this capability is not available, userspace should
1731 never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
1736 struct kvm_irq_routing_irqchip {
1741 struct kvm_irq_routing_msi {
1751 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
1752 for the device that wrote the MSI message. For PCI, this is usually a
1753 BFD identifier in the lower 16 bits.
1755 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
1756 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
1757 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
1758 address_hi must be zero.
1762 struct kvm_irq_routing_s390_adapter {
1766 __u32 summary_offset;
1770 struct kvm_irq_routing_hv_sint {
1776 4.55 KVM_SET_TSC_KHZ
1777 --------------------
1779 :Capability: KVM_CAP_TSC_CONTROL
1782 :Parameters: virtual tsc_khz
1783 :Returns: 0 on success, -1 on error
1785 Specifies the tsc frequency for the virtual machine. The unit of the
1789 4.56 KVM_GET_TSC_KHZ
1790 --------------------
1792 :Capability: KVM_CAP_GET_TSC_KHZ
1796 :Returns: virtual tsc-khz on success, negative value on error
1798 Returns the tsc frequency of the guest. The unit of the return value is
1799 KHz. If the host has unstable tsc this ioctl returns -EIO instead as an
1806 :Capability: KVM_CAP_IRQCHIP
1809 :Parameters: struct kvm_lapic_state (out)
1810 :Returns: 0 on success, -1 on error
1814 #define KVM_APIC_REG_SIZE 0x400
1815 struct kvm_lapic_state {
1816 char regs[KVM_APIC_REG_SIZE];
1819 Reads the Local APIC registers and copies them into the input argument. The
1820 data format and layout are the same as documented in the architecture manual.
1822 If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is
1823 enabled, then the format of APIC_ID register depends on the APIC mode
1824 (reported by MSR_IA32_APICBASE) of its VCPU. x2APIC stores APIC ID in
1825 the APIC_ID register (bytes 32-35). xAPIC only allows an 8-bit APIC ID
1826 which is stored in bits 31-24 of the APIC register, or equivalently in
1827 byte 35 of struct kvm_lapic_state's regs field. KVM_GET_LAPIC must then
1828 be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR.
1830 If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state
1831 always uses xAPIC format.
1837 :Capability: KVM_CAP_IRQCHIP
1840 :Parameters: struct kvm_lapic_state (in)
1841 :Returns: 0 on success, -1 on error
1845 #define KVM_APIC_REG_SIZE 0x400
1846 struct kvm_lapic_state {
1847 char regs[KVM_APIC_REG_SIZE];
1850 Copies the input argument into the Local APIC registers. The data format
1851 and layout are the same as documented in the architecture manual.
1853 The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's
1854 regs field) depends on the state of the KVM_CAP_X2APIC_API capability.
1855 See the note in KVM_GET_LAPIC.
1861 :Capability: KVM_CAP_IOEVENTFD
1864 :Parameters: struct kvm_ioeventfd (in)
1865 :Returns: 0 on success, !0 on error
1867 This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address
1868 within the guest. A guest write in the registered address will signal the
1869 provided event instead of triggering an exit.
1873 struct kvm_ioeventfd {
1875 __u64 addr; /* legal pio/mmio address */
1876 __u32 len; /* 0, 1, 2, 4, or 8 bytes */
1882 For the special case of virtio-ccw devices on s390, the ioevent is matched
1883 to a subchannel/virtqueue tuple instead.
1885 The following flags are defined::
1887 #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch)
1888 #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio)
1889 #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign)
1890 #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \
1891 (1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify)
1893 If datamatch flag is set, the event will be signaled only if the written value
1894 to the registered address is equal to datamatch in struct kvm_ioeventfd.
1896 For virtio-ccw devices, addr contains the subchannel id and datamatch the
1899 With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and
1900 the kernel will ignore the length of guest write and may get a faster vmexit.
1901 The speedup may only apply to specific architectures, but the ioeventfd will
1907 :Capability: KVM_CAP_SW_TLB
1910 :Parameters: struct kvm_dirty_tlb (in)
1911 :Returns: 0 on success, -1 on error
1915 struct kvm_dirty_tlb {
1920 This must be called whenever userspace has changed an entry in the shared
1921 TLB, prior to calling KVM_RUN on the associated vcpu.
1923 The "bitmap" field is the userspace address of an array. This array
1924 consists of a number of bits, equal to the total number of TLB entries as
1925 determined by the last successful call to KVM_CONFIG_TLB, rounded up to the
1926 nearest multiple of 64.
1928 Each bit corresponds to one TLB entry, ordered the same as in the shared TLB
1931 The array is little-endian: the bit 0 is the least significant bit of the
1932 first byte, bit 8 is the least significant bit of the second byte, etc.
1933 This avoids any complications with differing word sizes.
1935 The "num_dirty" field is a performance hint for KVM to determine whether it
1936 should skip processing the bitmap and just invalidate everything. It must
1937 be set to the number of set bits in the bitmap.
1940 4.62 KVM_CREATE_SPAPR_TCE
1941 -------------------------
1943 :Capability: KVM_CAP_SPAPR_TCE
1944 :Architectures: powerpc
1946 :Parameters: struct kvm_create_spapr_tce (in)
1947 :Returns: file descriptor for manipulating the created TCE table
1949 This creates a virtual TCE (translation control entry) table, which
1950 is an IOMMU for PAPR-style virtual I/O. It is used to translate
1951 logical addresses used in virtual I/O into guest physical addresses,
1952 and provides a scatter/gather capability for PAPR virtual I/O.
1956 /* for KVM_CAP_SPAPR_TCE */
1957 struct kvm_create_spapr_tce {
1962 The liobn field gives the logical IO bus number for which to create a
1963 TCE table. The window_size field specifies the size of the DMA window
1964 which this TCE table will translate - the table will contain one 64
1965 bit TCE entry for every 4kiB of the DMA window.
1967 When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE
1968 table has been created using this ioctl(), the kernel will handle it
1969 in real mode, updating the TCE table. H_PUT_TCE calls for other
1970 liobns will cause a vm exit and must be handled by userspace.
1972 The return value is a file descriptor which can be passed to mmap(2)
1973 to map the created TCE table into userspace. This lets userspace read
1974 the entries written by kernel-handled H_PUT_TCE calls, and also lets
1975 userspace update the TCE table directly which is useful in some
1979 4.63 KVM_ALLOCATE_RMA
1980 ---------------------
1982 :Capability: KVM_CAP_PPC_RMA
1983 :Architectures: powerpc
1985 :Parameters: struct kvm_allocate_rma (out)
1986 :Returns: file descriptor for mapping the allocated RMA
1988 This allocates a Real Mode Area (RMA) from the pool allocated at boot
1989 time by the kernel. An RMA is a physically-contiguous, aligned region
1990 of memory used on older POWER processors to provide the memory which
1991 will be accessed by real-mode (MMU off) accesses in a KVM guest.
1992 POWER processors support a set of sizes for the RMA that usually
1993 includes 64MB, 128MB, 256MB and some larger powers of two.
1997 /* for KVM_ALLOCATE_RMA */
1998 struct kvm_allocate_rma {
2002 The return value is a file descriptor which can be passed to mmap(2)
2003 to map the allocated RMA into userspace. The mapped area can then be
2004 passed to the KVM_SET_USER_MEMORY_REGION ioctl to establish it as the
2005 RMA for a virtual machine. The size of the RMA in bytes (which is
2006 fixed at host kernel boot time) is returned in the rma_size field of
2007 the argument structure.
2009 The KVM_CAP_PPC_RMA capability is 1 or 2 if the KVM_ALLOCATE_RMA ioctl
2010 is supported; 2 if the processor requires all virtual machines to have
2011 an RMA, or 1 if the processor can use an RMA but doesn't require it,
2012 because it supports the Virtual RMA (VRMA) facility.
2018 :Capability: KVM_CAP_USER_NMI
2022 :Returns: 0 on success, -1 on error
2024 Queues an NMI on the thread's vcpu. Note this is well defined only
2025 when KVM_CREATE_IRQCHIP has not been called, since this is an interface
2026 between the virtual cpu core and virtual local APIC. After KVM_CREATE_IRQCHIP
2027 has been called, this interface is completely emulated within the kernel.
2029 To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the
2030 following algorithm:
2033 - read the local APIC's state (KVM_GET_LAPIC)
2034 - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1)
2035 - if so, issue KVM_NMI
2038 Some guests configure the LINT1 NMI input to cause a panic, aiding in
2042 4.65 KVM_S390_UCAS_MAP
2043 ----------------------
2045 :Capability: KVM_CAP_S390_UCONTROL
2046 :Architectures: s390
2048 :Parameters: struct kvm_s390_ucas_mapping (in)
2049 :Returns: 0 in case of success
2051 The parameter is defined like this::
2053 struct kvm_s390_ucas_mapping {
2059 This ioctl maps the memory at "user_addr" with the length "length" to
2060 the vcpu's address space starting at "vcpu_addr". All parameters need to
2061 be aligned by 1 megabyte.
2064 4.66 KVM_S390_UCAS_UNMAP
2065 ------------------------
2067 :Capability: KVM_CAP_S390_UCONTROL
2068 :Architectures: s390
2070 :Parameters: struct kvm_s390_ucas_mapping (in)
2071 :Returns: 0 in case of success
2073 The parameter is defined like this::
2075 struct kvm_s390_ucas_mapping {
2081 This ioctl unmaps the memory in the vcpu's address space starting at
2082 "vcpu_addr" with the length "length". The field "user_addr" is ignored.
2083 All parameters need to be aligned by 1 megabyte.
2086 4.67 KVM_S390_VCPU_FAULT
2087 ------------------------
2089 :Capability: KVM_CAP_S390_UCONTROL
2090 :Architectures: s390
2092 :Parameters: vcpu absolute address (in)
2093 :Returns: 0 in case of success
2095 This call creates a page table entry on the virtual cpu's address space
2096 (for user controlled virtual machines) or the virtual machine's address
2097 space (for regular virtual machines). This only works for minor faults,
2098 thus it's recommended to access subject memory page via the user page
2099 table upfront. This is useful to handle validity intercepts for user
2100 controlled virtual machines to fault in the virtual cpu's lowcore pages
2101 prior to calling the KVM_RUN ioctl.
2104 4.68 KVM_SET_ONE_REG
2105 --------------------
2107 :Capability: KVM_CAP_ONE_REG
2110 :Parameters: struct kvm_one_reg (in)
2111 :Returns: 0 on success, negative value on failure
2115 ====== ============================================================
2116 Â ENOENT Â Â no such register
2117 Â EINVAL Â Â invalid register ID, or no such register or used with VMs in
2118 protected virtualization mode on s390
2119 Â EPERM Â Â Â (arm64) register access not allowed before vcpu finalization
2120 ====== ============================================================
2122 (These error codes are indicative only: do not rely on a specific error
2123 code being returned in a specific situation.)
2127 struct kvm_one_reg {
2132 Using this ioctl, a single vcpu register can be set to a specific value
2133 defined by user space with the passed in struct kvm_one_reg, where id
2134 refers to the register identifier as described below and addr is a pointer
2135 to a variable with the respective size. There can be architecture agnostic
2136 and architecture specific registers. Each have their own range of operation
2137 and their own constants and width. To keep track of the implemented
2138 registers, find a list below:
2140 ======= =============================== ============
2141 Arch Register Width (bits)
2142 ======= =============================== ============
2143 PPC KVM_REG_PPC_HIOR 64
2144 PPC KVM_REG_PPC_IAC1 64
2145 PPC KVM_REG_PPC_IAC2 64
2146 PPC KVM_REG_PPC_IAC3 64
2147 PPC KVM_REG_PPC_IAC4 64
2148 PPC KVM_REG_PPC_DAC1 64
2149 PPC KVM_REG_PPC_DAC2 64
2150 PPC KVM_REG_PPC_DABR 64
2151 PPC KVM_REG_PPC_DSCR 64
2152 PPC KVM_REG_PPC_PURR 64
2153 PPC KVM_REG_PPC_SPURR 64
2154 PPC KVM_REG_PPC_DAR 64
2155 PPC KVM_REG_PPC_DSISR 32
2156 PPC KVM_REG_PPC_AMR 64
2157 PPC KVM_REG_PPC_UAMOR 64
2158 PPC KVM_REG_PPC_MMCR0 64
2159 PPC KVM_REG_PPC_MMCR1 64
2160 PPC KVM_REG_PPC_MMCRA 64
2161 PPC KVM_REG_PPC_MMCR2 64
2162 PPC KVM_REG_PPC_MMCRS 64
2163 PPC KVM_REG_PPC_MMCR3 64
2164 PPC KVM_REG_PPC_SIAR 64
2165 PPC KVM_REG_PPC_SDAR 64
2166 PPC KVM_REG_PPC_SIER 64
2167 PPC KVM_REG_PPC_SIER2 64
2168 PPC KVM_REG_PPC_SIER3 64
2169 PPC KVM_REG_PPC_PMC1 32
2170 PPC KVM_REG_PPC_PMC2 32
2171 PPC KVM_REG_PPC_PMC3 32
2172 PPC KVM_REG_PPC_PMC4 32
2173 PPC KVM_REG_PPC_PMC5 32
2174 PPC KVM_REG_PPC_PMC6 32
2175 PPC KVM_REG_PPC_PMC7 32
2176 PPC KVM_REG_PPC_PMC8 32
2177 PPC KVM_REG_PPC_FPR0 64
2179 PPC KVM_REG_PPC_FPR31 64
2180 PPC KVM_REG_PPC_VR0 128
2182 PPC KVM_REG_PPC_VR31 128
2183 PPC KVM_REG_PPC_VSR0 128
2185 PPC KVM_REG_PPC_VSR31 128
2186 PPC KVM_REG_PPC_FPSCR 64
2187 PPC KVM_REG_PPC_VSCR 32
2188 PPC KVM_REG_PPC_VPA_ADDR 64
2189 PPC KVM_REG_PPC_VPA_SLB 128
2190 PPC KVM_REG_PPC_VPA_DTL 128
2191 PPC KVM_REG_PPC_EPCR 32
2192 PPC KVM_REG_PPC_EPR 32
2193 PPC KVM_REG_PPC_TCR 32
2194 PPC KVM_REG_PPC_TSR 32
2195 PPC KVM_REG_PPC_OR_TSR 32
2196 PPC KVM_REG_PPC_CLEAR_TSR 32
2197 PPC KVM_REG_PPC_MAS0 32
2198 PPC KVM_REG_PPC_MAS1 32
2199 PPC KVM_REG_PPC_MAS2 64
2200 PPC KVM_REG_PPC_MAS7_3 64
2201 PPC KVM_REG_PPC_MAS4 32
2202 PPC KVM_REG_PPC_MAS6 32
2203 PPC KVM_REG_PPC_MMUCFG 32
2204 PPC KVM_REG_PPC_TLB0CFG 32
2205 PPC KVM_REG_PPC_TLB1CFG 32
2206 PPC KVM_REG_PPC_TLB2CFG 32
2207 PPC KVM_REG_PPC_TLB3CFG 32
2208 PPC KVM_REG_PPC_TLB0PS 32
2209 PPC KVM_REG_PPC_TLB1PS 32
2210 PPC KVM_REG_PPC_TLB2PS 32
2211 PPC KVM_REG_PPC_TLB3PS 32
2212 PPC KVM_REG_PPC_EPTCFG 32
2213 PPC KVM_REG_PPC_ICP_STATE 64
2214 PPC KVM_REG_PPC_VP_STATE 128
2215 PPC KVM_REG_PPC_TB_OFFSET 64
2216 PPC KVM_REG_PPC_SPMC1 32
2217 PPC KVM_REG_PPC_SPMC2 32
2218 PPC KVM_REG_PPC_IAMR 64
2219 PPC KVM_REG_PPC_TFHAR 64
2220 PPC KVM_REG_PPC_TFIAR 64
2221 PPC KVM_REG_PPC_TEXASR 64
2222 PPC KVM_REG_PPC_FSCR 64
2223 PPC KVM_REG_PPC_PSPB 32
2224 PPC KVM_REG_PPC_EBBHR 64
2225 PPC KVM_REG_PPC_EBBRR 64
2226 PPC KVM_REG_PPC_BESCR 64
2227 PPC KVM_REG_PPC_TAR 64
2228 PPC KVM_REG_PPC_DPDES 64
2229 PPC KVM_REG_PPC_DAWR 64
2230 PPC KVM_REG_PPC_DAWRX 64
2231 PPC KVM_REG_PPC_CIABR 64
2232 PPC KVM_REG_PPC_IC 64
2233 PPC KVM_REG_PPC_VTB 64
2234 PPC KVM_REG_PPC_CSIGR 64
2235 PPC KVM_REG_PPC_TACR 64
2236 PPC KVM_REG_PPC_TCSCR 64
2237 PPC KVM_REG_PPC_PID 64
2238 PPC KVM_REG_PPC_ACOP 64
2239 PPC KVM_REG_PPC_VRSAVE 32
2240 PPC KVM_REG_PPC_LPCR 32
2241 PPC KVM_REG_PPC_LPCR_64 64
2242 PPC KVM_REG_PPC_PPR 64
2243 PPC KVM_REG_PPC_ARCH_COMPAT 32
2244 PPC KVM_REG_PPC_DABRX 32
2245 PPC KVM_REG_PPC_WORT 64
2246 PPC KVM_REG_PPC_SPRG9 64
2247 PPC KVM_REG_PPC_DBSR 32
2248 PPC KVM_REG_PPC_TIDR 64
2249 PPC KVM_REG_PPC_PSSCR 64
2250 PPC KVM_REG_PPC_DEC_EXPIRY 64
2251 PPC KVM_REG_PPC_PTCR 64
2252 PPC KVM_REG_PPC_TM_GPR0 64
2254 PPC KVM_REG_PPC_TM_GPR31 64
2255 PPC KVM_REG_PPC_TM_VSR0 128
2257 PPC KVM_REG_PPC_TM_VSR63 128
2258 PPC KVM_REG_PPC_TM_CR 64
2259 PPC KVM_REG_PPC_TM_LR 64
2260 PPC KVM_REG_PPC_TM_CTR 64
2261 PPC KVM_REG_PPC_TM_FPSCR 64
2262 PPC KVM_REG_PPC_TM_AMR 64
2263 PPC KVM_REG_PPC_TM_PPR 64
2264 PPC KVM_REG_PPC_TM_VRSAVE 64
2265 PPC KVM_REG_PPC_TM_VSCR 32
2266 PPC KVM_REG_PPC_TM_DSCR 64
2267 PPC KVM_REG_PPC_TM_TAR 64
2268 PPC KVM_REG_PPC_TM_XER 64
2270 MIPS KVM_REG_MIPS_R0 64
2272 MIPS KVM_REG_MIPS_R31 64
2273 MIPS KVM_REG_MIPS_HI 64
2274 MIPS KVM_REG_MIPS_LO 64
2275 MIPS KVM_REG_MIPS_PC 64
2276 MIPS KVM_REG_MIPS_CP0_INDEX 32
2277 MIPS KVM_REG_MIPS_CP0_ENTRYLO0 64
2278 MIPS KVM_REG_MIPS_CP0_ENTRYLO1 64
2279 MIPS KVM_REG_MIPS_CP0_CONTEXT 64
2280 MIPS KVM_REG_MIPS_CP0_CONTEXTCONFIG 32
2281 MIPS KVM_REG_MIPS_CP0_USERLOCAL 64
2282 MIPS KVM_REG_MIPS_CP0_XCONTEXTCONFIG 64
2283 MIPS KVM_REG_MIPS_CP0_PAGEMASK 32
2284 MIPS KVM_REG_MIPS_CP0_PAGEGRAIN 32
2285 MIPS KVM_REG_MIPS_CP0_SEGCTL0 64
2286 MIPS KVM_REG_MIPS_CP0_SEGCTL1 64
2287 MIPS KVM_REG_MIPS_CP0_SEGCTL2 64
2288 MIPS KVM_REG_MIPS_CP0_PWBASE 64
2289 MIPS KVM_REG_MIPS_CP0_PWFIELD 64
2290 MIPS KVM_REG_MIPS_CP0_PWSIZE 64
2291 MIPS KVM_REG_MIPS_CP0_WIRED 32
2292 MIPS KVM_REG_MIPS_CP0_PWCTL 32
2293 MIPS KVM_REG_MIPS_CP0_HWRENA 32
2294 MIPS KVM_REG_MIPS_CP0_BADVADDR 64
2295 MIPS KVM_REG_MIPS_CP0_BADINSTR 32
2296 MIPS KVM_REG_MIPS_CP0_BADINSTRP 32
2297 MIPS KVM_REG_MIPS_CP0_COUNT 32
2298 MIPS KVM_REG_MIPS_CP0_ENTRYHI 64
2299 MIPS KVM_REG_MIPS_CP0_COMPARE 32
2300 MIPS KVM_REG_MIPS_CP0_STATUS 32
2301 MIPS KVM_REG_MIPS_CP0_INTCTL 32
2302 MIPS KVM_REG_MIPS_CP0_CAUSE 32
2303 MIPS KVM_REG_MIPS_CP0_EPC 64
2304 MIPS KVM_REG_MIPS_CP0_PRID 32
2305 MIPS KVM_REG_MIPS_CP0_EBASE 64
2306 MIPS KVM_REG_MIPS_CP0_CONFIG 32
2307 MIPS KVM_REG_MIPS_CP0_CONFIG1 32
2308 MIPS KVM_REG_MIPS_CP0_CONFIG2 32
2309 MIPS KVM_REG_MIPS_CP0_CONFIG3 32
2310 MIPS KVM_REG_MIPS_CP0_CONFIG4 32
2311 MIPS KVM_REG_MIPS_CP0_CONFIG5 32
2312 MIPS KVM_REG_MIPS_CP0_CONFIG7 32
2313 MIPS KVM_REG_MIPS_CP0_XCONTEXT 64
2314 MIPS KVM_REG_MIPS_CP0_ERROREPC 64
2315 MIPS KVM_REG_MIPS_CP0_KSCRATCH1 64
2316 MIPS KVM_REG_MIPS_CP0_KSCRATCH2 64
2317 MIPS KVM_REG_MIPS_CP0_KSCRATCH3 64
2318 MIPS KVM_REG_MIPS_CP0_KSCRATCH4 64
2319 MIPS KVM_REG_MIPS_CP0_KSCRATCH5 64
2320 MIPS KVM_REG_MIPS_CP0_KSCRATCH6 64
2321 MIPS KVM_REG_MIPS_CP0_MAAR(0..63) 64
2322 MIPS KVM_REG_MIPS_COUNT_CTL 64
2323 MIPS KVM_REG_MIPS_COUNT_RESUME 64
2324 MIPS KVM_REG_MIPS_COUNT_HZ 64
2325 MIPS KVM_REG_MIPS_FPR_32(0..31) 32
2326 MIPS KVM_REG_MIPS_FPR_64(0..31) 64
2327 MIPS KVM_REG_MIPS_VEC_128(0..31) 128
2328 MIPS KVM_REG_MIPS_FCR_IR 32
2329 MIPS KVM_REG_MIPS_FCR_CSR 32
2330 MIPS KVM_REG_MIPS_MSA_IR 32
2331 MIPS KVM_REG_MIPS_MSA_CSR 32
2332 ======= =============================== ============
2334 ARM registers are mapped using the lower 32 bits. The upper 16 of that
2335 is the register group type, or coprocessor number:
2337 ARM core registers have the following id bit patterns::
2339 0x4020 0000 0010 <index into the kvm_regs struct:16>
2341 ARM 32-bit CP15 registers have the following id bit patterns::
2343 0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3>
2345 ARM 64-bit CP15 registers have the following id bit patterns::
2347 0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3>
2349 ARM CCSIDR registers are demultiplexed by CSSELR value::
2351 0x4020 0000 0011 00 <csselr:8>
2353 ARM 32-bit VFP control registers have the following id bit patterns::
2355 0x4020 0000 0012 1 <regno:12>
2357 ARM 64-bit FP registers have the following id bit patterns::
2359 0x4030 0000 0012 0 <regno:12>
2361 ARM firmware pseudo-registers have the following bit pattern::
2363 0x4030 0000 0014 <regno:16>
2366 arm64 registers are mapped using the lower 32 bits. The upper 16 of
2367 that is the register group type, or coprocessor number:
2369 arm64 core/FP-SIMD registers have the following id bit patterns. Note
2370 that the size of the access is variable, as the kvm_regs structure
2371 contains elements ranging from 32 to 128 bits. The index is a 32bit
2372 value in the kvm_regs structure seen as a 32bit array::
2374 0x60x0 0000 0010 <index into the kvm_regs struct:16>
2378 ======================= ========= ===== =======================================
2379 Encoding Register Bits kvm_regs member
2380 ======================= ========= ===== =======================================
2381 0x6030 0000 0010 0000 X0 64 regs.regs[0]
2382 0x6030 0000 0010 0002 X1 64 regs.regs[1]
2384 0x6030 0000 0010 003c X30 64 regs.regs[30]
2385 0x6030 0000 0010 003e SP 64 regs.sp
2386 0x6030 0000 0010 0040 PC 64 regs.pc
2387 0x6030 0000 0010 0042 PSTATE 64 regs.pstate
2388 0x6030 0000 0010 0044 SP_EL1 64 sp_el1
2389 0x6030 0000 0010 0046 ELR_EL1 64 elr_el1
2390 0x6030 0000 0010 0048 SPSR_EL1 64 spsr[KVM_SPSR_EL1] (alias SPSR_SVC)
2391 0x6030 0000 0010 004a SPSR_ABT 64 spsr[KVM_SPSR_ABT]
2392 0x6030 0000 0010 004c SPSR_UND 64 spsr[KVM_SPSR_UND]
2393 0x6030 0000 0010 004e SPSR_IRQ 64 spsr[KVM_SPSR_IRQ]
2394 0x6060 0000 0010 0050 SPSR_FIQ 64 spsr[KVM_SPSR_FIQ]
2395 0x6040 0000 0010 0054 V0 128 fp_regs.vregs[0] [1]_
2396 0x6040 0000 0010 0058 V1 128 fp_regs.vregs[1] [1]_
2398 0x6040 0000 0010 00d0 V31 128 fp_regs.vregs[31] [1]_
2399 0x6020 0000 0010 00d4 FPSR 32 fp_regs.fpsr
2400 0x6020 0000 0010 00d5 FPCR 32 fp_regs.fpcr
2401 ======================= ========= ===== =======================================
2403 .. [1] These encodings are not accepted for SVE-enabled vcpus. See
2406 The equivalent register content can be accessed via bits [127:0] of
2407 the corresponding SVE Zn registers instead for vcpus that have SVE
2408 enabled (see below).
2410 arm64 CCSIDR registers are demultiplexed by CSSELR value::
2412 0x6020 0000 0011 00 <csselr:8>
2414 arm64 system registers have the following id bit patterns::
2416 0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3>
2420 Two system register IDs do not follow the specified pattern. These
2421 are KVM_REG_ARM_TIMER_CVAL and KVM_REG_ARM_TIMER_CNT, which map to
2422 system registers CNTV_CVAL_EL0 and CNTVCT_EL0 respectively. These
2423 two had their values accidentally swapped, which means TIMER_CVAL is
2424 derived from the register encoding for CNTVCT_EL0 and TIMER_CNT is
2425 derived from the register encoding for CNTV_CVAL_EL0. As this is
2426 API, it must remain this way.
2428 arm64 firmware pseudo-registers have the following bit pattern::
2430 0x6030 0000 0014 <regno:16>
2432 arm64 SVE registers have the following bit patterns::
2434 0x6080 0000 0015 00 <n:5> <slice:5> Zn bits[2048*slice + 2047 : 2048*slice]
2435 0x6050 0000 0015 04 <n:4> <slice:5> Pn bits[256*slice + 255 : 256*slice]
2436 0x6050 0000 0015 060 <slice:5> FFR bits[256*slice + 255 : 256*slice]
2437 0x6060 0000 0015 ffff KVM_REG_ARM64_SVE_VLS pseudo-register
2439 Access to register IDs where 2048 * slice >= 128 * max_vq will fail with
2440 ENOENT. max_vq is the vcpu's maximum supported vector length in 128-bit
2441 quadwords: see [2]_ below.
2443 These registers are only accessible on vcpus for which SVE is enabled.
2444 See KVM_ARM_VCPU_INIT for details.
2446 In addition, except for KVM_REG_ARM64_SVE_VLS, these registers are not
2447 accessible until the vcpu's SVE configuration has been finalized
2448 using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). See KVM_ARM_VCPU_INIT
2449 and KVM_ARM_VCPU_FINALIZE for more information about this procedure.
2451 KVM_REG_ARM64_SVE_VLS is a pseudo-register that allows the set of vector
2452 lengths supported by the vcpu to be discovered and configured by
2453 userspace. When transferred to or from user memory via KVM_GET_ONE_REG
2454 or KVM_SET_ONE_REG, the value of this register is of type
2455 __u64[KVM_ARM64_SVE_VLS_WORDS], and encodes the set of vector lengths as
2458 __u64 vector_lengths[KVM_ARM64_SVE_VLS_WORDS];
2460 if (vq >= SVE_VQ_MIN && vq <= SVE_VQ_MAX &&
2461 ((vector_lengths[(vq - KVM_ARM64_SVE_VQ_MIN) / 64] >>
2462 ((vq - KVM_ARM64_SVE_VQ_MIN) % 64)) & 1))
2463 /* Vector length vq * 16 bytes supported */
2465 /* Vector length vq * 16 bytes not supported */
2467 .. [2] The maximum value vq for which the above condition is true is
2468 max_vq. This is the maximum vector length available to the guest on
2469 this vcpu, and determines which register slices are visible through
2470 this ioctl interface.
2472 (See Documentation/arm64/sve.rst for an explanation of the "vq"
2475 KVM_REG_ARM64_SVE_VLS is only accessible after KVM_ARM_VCPU_INIT.
2476 KVM_ARM_VCPU_INIT initialises it to the best set of vector lengths that
2479 Userspace may subsequently modify it if desired until the vcpu's SVE
2480 configuration is finalized using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE).
2482 Apart from simply removing all vector lengths from the host set that
2483 exceed some value, support for arbitrarily chosen sets of vector lengths
2484 is hardware-dependent and may not be available. Attempting to configure
2485 an invalid set of vector lengths via KVM_SET_ONE_REG will fail with
2488 After the vcpu's SVE configuration is finalized, further attempts to
2489 write this register will fail with EPERM.
2492 MIPS registers are mapped using the lower 32 bits. The upper 16 of that is
2493 the register group type:
2495 MIPS core registers (see above) have the following id bit patterns::
2497 0x7030 0000 0000 <reg:16>
2499 MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit
2500 patterns depending on whether they're 32-bit or 64-bit registers::
2502 0x7020 0000 0001 00 <reg:5> <sel:3> (32-bit)
2503 0x7030 0000 0001 00 <reg:5> <sel:3> (64-bit)
2505 Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_MIPS_CP0_ENTRYLO1 are the MIPS64
2506 versions of the EntryLo registers regardless of the word size of the host
2507 hardware, host kernel, guest, and whether XPA is present in the guest, i.e.
2508 with the RI and XI bits (if they exist) in bits 63 and 62 respectively, and
2509 the PFNX field starting at bit 30.
2511 MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) above) have the following id bit
2514 0x7030 0000 0001 01 <reg:8>
2516 MIPS KVM control registers (see above) have the following id bit patterns::
2518 0x7030 0000 0002 <reg:16>
2520 MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following
2521 id bit patterns depending on the size of the register being accessed. They are
2522 always accessed according to the current guest FPU mode (Status.FR and
2523 Config5.FRE), i.e. as the guest would see them, and they become unpredictable
2524 if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector
2525 registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they
2526 overlap the FPU registers::
2528 0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers)
2529 0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers)
2530 0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers)
2532 MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the
2533 following id bit patterns::
2535 0x7020 0000 0003 01 <0:3> <reg:5>
2537 MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the
2538 following id bit patterns::
2540 0x7020 0000 0003 02 <0:3> <reg:5>
2543 4.69 KVM_GET_ONE_REG
2544 --------------------
2546 :Capability: KVM_CAP_ONE_REG
2549 :Parameters: struct kvm_one_reg (in and out)
2550 :Returns: 0 on success, negative value on failure
2554 ======== ============================================================
2555 Â ENOENT Â Â no such register
2556 Â EINVAL Â Â invalid register ID, or no such register or used with VMs in
2557 protected virtualization mode on s390
2558 Â EPERM Â Â Â (arm64) register access not allowed before vcpu finalization
2559 ======== ============================================================
2561 (These error codes are indicative only: do not rely on a specific error
2562 code being returned in a specific situation.)
2564 This ioctl allows to receive the value of a single register implemented
2565 in a vcpu. The register to read is indicated by the "id" field of the
2566 kvm_one_reg struct passed in. On success, the register value can be found
2567 at the memory location pointed to by "addr".
2569 The list of registers accessible using this interface is identical to the
2573 4.70 KVM_KVMCLOCK_CTRL
2574 ----------------------
2576 :Capability: KVM_CAP_KVMCLOCK_CTRL
2577 :Architectures: Any that implement pvclocks (currently x86 only)
2580 :Returns: 0 on success, -1 on error
2582 This ioctl sets a flag accessible to the guest indicating that the specified
2583 vCPU has been paused by the host userspace.
2585 The host will set a flag in the pvclock structure that is checked from the
2586 soft lockup watchdog. The flag is part of the pvclock structure that is
2587 shared between guest and host, specifically the second bit of the flags
2588 field of the pvclock_vcpu_time_info structure. It will be set exclusively by
2589 the host and read/cleared exclusively by the guest. The guest operation of
2590 checking and clearing the flag must be an atomic operation so
2591 load-link/store-conditional, or equivalent must be used. There are two cases
2592 where the guest will clear the flag: when the soft lockup watchdog timer resets
2593 itself or when a soft lockup is detected. This ioctl can be called any time
2594 after pausing the vcpu, but before it is resumed.
2600 :Capability: KVM_CAP_SIGNAL_MSI
2601 :Architectures: x86 arm arm64
2603 :Parameters: struct kvm_msi (in)
2604 :Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error
2606 Directly inject a MSI message. Only valid with in-kernel irqchip that handles
2621 KVM_MSI_VALID_DEVID: devid contains a valid value. The per-VM
2622 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
2623 the device ID. If this capability is not available, userspace
2624 should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
2626 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
2627 for the device that wrote the MSI message. For PCI, this is usually a
2628 BFD identifier in the lower 16 bits.
2630 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
2631 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
2632 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
2633 address_hi must be zero.
2636 4.71 KVM_CREATE_PIT2
2637 --------------------
2639 :Capability: KVM_CAP_PIT2
2642 :Parameters: struct kvm_pit_config (in)
2643 :Returns: 0 on success, -1 on error
2645 Creates an in-kernel device model for the i8254 PIT. This call is only valid
2646 after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following
2647 parameters have to be passed::
2649 struct kvm_pit_config {
2656 #define KVM_PIT_SPEAKER_DUMMY 1 /* emulate speaker port stub */
2658 PIT timer interrupts may use a per-VM kernel thread for injection. If it
2659 exists, this thread will have a name of the following pattern::
2661 kvm-pit/<owner-process-pid>
2663 When running a guest with elevated priorities, the scheduling parameters of
2664 this thread may have to be adjusted accordingly.
2666 This IOCTL replaces the obsolete KVM_CREATE_PIT.
2672 :Capability: KVM_CAP_PIT_STATE2
2675 :Parameters: struct kvm_pit_state2 (out)
2676 :Returns: 0 on success, -1 on error
2678 Retrieves the state of the in-kernel PIT model. Only valid after
2679 KVM_CREATE_PIT2. The state is returned in the following structure::
2681 struct kvm_pit_state2 {
2682 struct kvm_pit_channel_state channels[3];
2689 /* disable PIT in HPET legacy mode */
2690 #define KVM_PIT_FLAGS_HPET_LEGACY 0x00000001
2692 This IOCTL replaces the obsolete KVM_GET_PIT.
2698 :Capability: KVM_CAP_PIT_STATE2
2701 :Parameters: struct kvm_pit_state2 (in)
2702 :Returns: 0 on success, -1 on error
2704 Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2.
2705 See KVM_GET_PIT2 for details on struct kvm_pit_state2.
2707 This IOCTL replaces the obsolete KVM_SET_PIT.
2710 4.74 KVM_PPC_GET_SMMU_INFO
2711 --------------------------
2713 :Capability: KVM_CAP_PPC_GET_SMMU_INFO
2714 :Architectures: powerpc
2717 :Returns: 0 on success, -1 on error
2719 This populates and returns a structure describing the features of
2720 the "Server" class MMU emulation supported by KVM.
2721 This can in turn be used by userspace to generate the appropriate
2722 device-tree properties for the guest operating system.
2724 The structure contains some global information, followed by an
2725 array of supported segment page sizes::
2727 struct kvm_ppc_smmu_info {
2731 struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ];
2734 The supported flags are:
2736 - KVM_PPC_PAGE_SIZES_REAL:
2737 When that flag is set, guest page sizes must "fit" the backing
2738 store page sizes. When not set, any page size in the list can
2739 be used regardless of how they are backed by userspace.
2741 - KVM_PPC_1T_SEGMENTS
2742 The emulated MMU supports 1T segments in addition to the
2746 This flag indicates that HPT guests are not supported by KVM,
2747 thus all guests must use radix MMU mode.
2749 The "slb_size" field indicates how many SLB entries are supported
2751 The "sps" array contains 8 entries indicating the supported base
2752 page sizes for a segment in increasing order. Each entry is defined
2755 struct kvm_ppc_one_seg_page_size {
2756 __u32 page_shift; /* Base page shift of segment (or 0) */
2757 __u32 slb_enc; /* SLB encoding for BookS */
2758 struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ];
2761 An entry with a "page_shift" of 0 is unused. Because the array is
2762 organized in increasing order, a lookup can stop when encoutering
2765 The "slb_enc" field provides the encoding to use in the SLB for the
2766 page size. The bits are in positions such as the value can directly
2767 be OR'ed into the "vsid" argument of the slbmte instruction.
2769 The "enc" array is a list which for each of those segment base page
2770 size provides the list of supported actual page sizes (which can be
2771 only larger or equal to the base page size), along with the
2772 corresponding encoding in the hash PTE. Similarly, the array is
2773 8 entries sorted by increasing sizes and an entry with a "0" shift
2774 is an empty entry and a terminator::
2776 struct kvm_ppc_one_page_size {
2777 __u32 page_shift; /* Page shift (or 0) */
2778 __u32 pte_enc; /* Encoding in the HPTE (>>12) */
2781 The "pte_enc" field provides a value that can OR'ed into the hash
2782 PTE's RPN field (ie, it needs to be shifted left by 12 to OR it
2783 into the hash PTE second double word).
2788 :Capability: KVM_CAP_IRQFD
2789 :Architectures: x86 s390 arm arm64
2791 :Parameters: struct kvm_irqfd (in)
2792 :Returns: 0 on success, -1 on error
2794 Allows setting an eventfd to directly trigger a guest interrupt.
2795 kvm_irqfd.fd specifies the file descriptor to use as the eventfd and
2796 kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When
2797 an event is triggered on the eventfd, an interrupt is injected into
2798 the guest using the specified gsi pin. The irqfd is removed using
2799 the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd
2802 With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify
2803 mechanism allowing emulation of level-triggered, irqfd-based
2804 interrupts. When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an
2805 additional eventfd in the kvm_irqfd.resamplefd field. When operating
2806 in resample mode, posting of an interrupt through kvm_irq.fd asserts
2807 the specified gsi in the irqchip. When the irqchip is resampled, such
2808 as from an EOI, the gsi is de-asserted and the user is notified via
2809 kvm_irqfd.resamplefd. It is the user's responsibility to re-queue
2810 the interrupt if the device making use of it still requires service.
2811 Note that closing the resamplefd is not sufficient to disable the
2812 irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment
2813 and need not be specified with KVM_IRQFD_FLAG_DEASSIGN.
2815 On arm/arm64, gsi routing being supported, the following can happen:
2817 - in case no routing entry is associated to this gsi, injection fails
2818 - in case the gsi is associated to an irqchip routing entry,
2819 irqchip.pin + 32 corresponds to the injected SPI ID.
2820 - in case the gsi is associated to an MSI routing entry, the MSI
2821 message and device ID are translated into an LPI (support restricted
2822 to GICv3 ITS in-kernel emulation).
2824 4.76 KVM_PPC_ALLOCATE_HTAB
2825 --------------------------
2827 :Capability: KVM_CAP_PPC_ALLOC_HTAB
2828 :Architectures: powerpc
2830 :Parameters: Pointer to u32 containing hash table order (in/out)
2831 :Returns: 0 on success, -1 on error
2833 This requests the host kernel to allocate an MMU hash table for a
2834 guest using the PAPR paravirtualization interface. This only does
2835 anything if the kernel is configured to use the Book 3S HV style of
2836 virtualization. Otherwise the capability doesn't exist and the ioctl
2837 returns an ENOTTY error. The rest of this description assumes Book 3S
2840 There must be no vcpus running when this ioctl is called; if there
2841 are, it will do nothing and return an EBUSY error.
2843 The parameter is a pointer to a 32-bit unsigned integer variable
2844 containing the order (log base 2) of the desired size of the hash
2845 table, which must be between 18 and 46. On successful return from the
2846 ioctl, the value will not be changed by the kernel.
2848 If no hash table has been allocated when any vcpu is asked to run
2849 (with the KVM_RUN ioctl), the host kernel will allocate a
2850 default-sized hash table (16 MB).
2852 If this ioctl is called when a hash table has already been allocated,
2853 with a different order from the existing hash table, the existing hash
2854 table will be freed and a new one allocated. If this is ioctl is
2855 called when a hash table has already been allocated of the same order
2856 as specified, the kernel will clear out the existing hash table (zero
2857 all HPTEs). In either case, if the guest is using the virtualized
2858 real-mode area (VRMA) facility, the kernel will re-create the VMRA
2859 HPTEs on the next KVM_RUN of any vcpu.
2861 4.77 KVM_S390_INTERRUPT
2862 -----------------------
2865 :Architectures: s390
2866 :Type: vm ioctl, vcpu ioctl
2867 :Parameters: struct kvm_s390_interrupt (in)
2868 :Returns: 0 on success, -1 on error
2870 Allows to inject an interrupt to the guest. Interrupts can be floating
2871 (vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type.
2873 Interrupt parameters are passed via kvm_s390_interrupt::
2875 struct kvm_s390_interrupt {
2881 type can be one of the following:
2883 KVM_S390_SIGP_STOP (vcpu)
2884 - sigp stop; optional flags in parm
2885 KVM_S390_PROGRAM_INT (vcpu)
2886 - program check; code in parm
2887 KVM_S390_SIGP_SET_PREFIX (vcpu)
2888 - sigp set prefix; prefix address in parm
2889 KVM_S390_RESTART (vcpu)
2891 KVM_S390_INT_CLOCK_COMP (vcpu)
2892 - clock comparator interrupt
2893 KVM_S390_INT_CPU_TIMER (vcpu)
2894 - CPU timer interrupt
2895 KVM_S390_INT_VIRTIO (vm)
2896 - virtio external interrupt; external interrupt
2897 parameters in parm and parm64
2898 KVM_S390_INT_SERVICE (vm)
2899 - sclp external interrupt; sclp parameter in parm
2900 KVM_S390_INT_EMERGENCY (vcpu)
2901 - sigp emergency; source cpu in parm
2902 KVM_S390_INT_EXTERNAL_CALL (vcpu)
2903 - sigp external call; source cpu in parm
2904 KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm)
2905 - compound value to indicate an
2906 I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel);
2907 I/O interruption parameters in parm (subchannel) and parm64 (intparm,
2908 interruption subclass)
2909 KVM_S390_MCHK (vm, vcpu)
2910 - machine check interrupt; cr 14 bits in parm, machine check interrupt
2911 code in parm64 (note that machine checks needing further payload are not
2912 supported by this ioctl)
2914 This is an asynchronous vcpu ioctl and can be invoked from any thread.
2916 4.78 KVM_PPC_GET_HTAB_FD
2917 ------------------------
2919 :Capability: KVM_CAP_PPC_HTAB_FD
2920 :Architectures: powerpc
2922 :Parameters: Pointer to struct kvm_get_htab_fd (in)
2923 :Returns: file descriptor number (>= 0) on success, -1 on error
2925 This returns a file descriptor that can be used either to read out the
2926 entries in the guest's hashed page table (HPT), or to write entries to
2927 initialize the HPT. The returned fd can only be written to if the
2928 KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and
2929 can only be read if that bit is clear. The argument struct looks like
2932 /* For KVM_PPC_GET_HTAB_FD */
2933 struct kvm_get_htab_fd {
2939 /* Values for kvm_get_htab_fd.flags */
2940 #define KVM_GET_HTAB_BOLTED_ONLY ((__u64)0x1)
2941 #define KVM_GET_HTAB_WRITE ((__u64)0x2)
2943 The 'start_index' field gives the index in the HPT of the entry at
2944 which to start reading. It is ignored when writing.
2946 Reads on the fd will initially supply information about all
2947 "interesting" HPT entries. Interesting entries are those with the
2948 bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise
2949 all entries. When the end of the HPT is reached, the read() will
2950 return. If read() is called again on the fd, it will start again from
2951 the beginning of the HPT, but will only return HPT entries that have
2952 changed since they were last read.
2954 Data read or written is structured as a header (8 bytes) followed by a
2955 series of valid HPT entries (16 bytes) each. The header indicates how
2956 many valid HPT entries there are and how many invalid entries follow
2957 the valid entries. The invalid entries are not represented explicitly
2958 in the stream. The header format is::
2960 struct kvm_get_htab_header {
2966 Writes to the fd create HPT entries starting at the index given in the
2967 header; first 'n_valid' valid entries with contents from the data
2968 written, then 'n_invalid' invalid entries, invalidating any previously
2969 valid entries found.
2971 4.79 KVM_CREATE_DEVICE
2972 ----------------------
2974 :Capability: KVM_CAP_DEVICE_CTRL
2976 :Parameters: struct kvm_create_device (in/out)
2977 :Returns: 0 on success, -1 on error
2981 ====== =======================================================
2982 ENODEV The device type is unknown or unsupported
2983 EEXIST Device already created, and this type of device may not
2984 be instantiated multiple times
2985 ====== =======================================================
2987 Other error conditions may be defined by individual device types or
2988 have their standard meanings.
2990 Creates an emulated device in the kernel. The file descriptor returned
2991 in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR.
2993 If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the
2994 device type is supported (not necessarily whether it can be created
2997 Individual devices should not define flags. Attributes should be used
2998 for specifying any behavior that is not implied by the device type
3003 struct kvm_create_device {
3004 __u32 type; /* in: KVM_DEV_TYPE_xxx */
3005 __u32 fd; /* out: device handle */
3006 __u32 flags; /* in: KVM_CREATE_DEVICE_xxx */
3009 4.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR
3010 --------------------------------------------
3012 :Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
3013 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
3014 :Type: device ioctl, vm ioctl, vcpu ioctl
3015 :Parameters: struct kvm_device_attr
3016 :Returns: 0 on success, -1 on error
3020 ===== =============================================================
3021 ENXIO The group or attribute is unknown/unsupported for this device
3022 or hardware support is missing.
3023 EPERM The attribute cannot (currently) be accessed this way
3024 (e.g. read-only attribute, or attribute that only makes
3025 sense when the device is in a different state)
3026 ===== =============================================================
3028 Other error conditions may be defined by individual device types.
3030 Gets/sets a specified piece of device configuration and/or state. The
3031 semantics are device-specific. See individual device documentation in
3032 the "devices" directory. As with ONE_REG, the size of the data
3033 transferred is defined by the particular attribute.
3037 struct kvm_device_attr {
3038 __u32 flags; /* no flags currently defined */
3039 __u32 group; /* device-defined */
3040 __u64 attr; /* group-defined */
3041 __u64 addr; /* userspace address of attr data */
3044 4.81 KVM_HAS_DEVICE_ATTR
3045 ------------------------
3047 :Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
3048 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
3049 :Type: device ioctl, vm ioctl, vcpu ioctl
3050 :Parameters: struct kvm_device_attr
3051 :Returns: 0 on success, -1 on error
3055 ===== =============================================================
3056 ENXIO The group or attribute is unknown/unsupported for this device
3057 or hardware support is missing.
3058 ===== =============================================================
3060 Tests whether a device supports a particular attribute. A successful
3061 return indicates the attribute is implemented. It does not necessarily
3062 indicate that the attribute can be read or written in the device's
3063 current state. "addr" is ignored.
3065 4.82 KVM_ARM_VCPU_INIT
3066 ----------------------
3069 :Architectures: arm, arm64
3071 :Parameters: struct kvm_vcpu_init (in)
3072 :Returns: 0 on success; -1 on error
3076 ====== =================================================================
3077 Â EINVAL Â Â Â the target is unknown, or the combination of features is invalid.
3078 Â ENOENT Â Â Â a features bit specified is unknown.
3079 ====== =================================================================
3081 This tells KVM what type of CPU to present to the guest, and what
3082 optional features it should have. Â This will cause a reset of the cpu
3083 registers to their initial values. Â If this is not called, KVM_RUN will
3084 return ENOEXEC for that vcpu.
3086 Note that because some registers reflect machine topology, all vcpus
3087 should be created before this ioctl is invoked.
3089 Userspace can call this function multiple times for a given vcpu, including
3090 after the vcpu has been run. This will reset the vcpu to its initial
3091 state. All calls to this function after the initial call must use the same
3092 target and same set of feature flags, otherwise EINVAL will be returned.
3096 - KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state.
3097 Depends on KVM_CAP_ARM_PSCI. If not set, the CPU will be powered on
3098 and execute guest code when KVM_RUN is called.
3099 - KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode.
3100 Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only).
3101 - KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 (or a future revision
3102 backward compatible with v0.2) for the CPU.
3103 Depends on KVM_CAP_ARM_PSCI_0_2.
3104 - KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU.
3105 Depends on KVM_CAP_ARM_PMU_V3.
3107 - KVM_ARM_VCPU_PTRAUTH_ADDRESS: Enables Address Pointer authentication
3109 Depends on KVM_CAP_ARM_PTRAUTH_ADDRESS.
3110 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are
3111 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and
3112 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be
3115 - KVM_ARM_VCPU_PTRAUTH_GENERIC: Enables Generic Pointer authentication
3117 Depends on KVM_CAP_ARM_PTRAUTH_GENERIC.
3118 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are
3119 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and
3120 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be
3123 - KVM_ARM_VCPU_SVE: Enables SVE for the CPU (arm64 only).
3124 Depends on KVM_CAP_ARM_SVE.
3125 Requires KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3127 * After KVM_ARM_VCPU_INIT:
3129 - KVM_REG_ARM64_SVE_VLS may be read using KVM_GET_ONE_REG: the
3130 initial value of this pseudo-register indicates the best set of
3131 vector lengths possible for a vcpu on this host.
3133 * Before KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3135 - KVM_RUN and KVM_GET_REG_LIST are not available;
3137 - KVM_GET_ONE_REG and KVM_SET_ONE_REG cannot be used to access
3138 the scalable archietctural SVE registers
3139 KVM_REG_ARM64_SVE_ZREG(), KVM_REG_ARM64_SVE_PREG() or
3140 KVM_REG_ARM64_SVE_FFR;
3142 - KVM_REG_ARM64_SVE_VLS may optionally be written using
3143 KVM_SET_ONE_REG, to modify the set of vector lengths available
3146 * After KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3148 - the KVM_REG_ARM64_SVE_VLS pseudo-register is immutable, and can
3149 no longer be written using KVM_SET_ONE_REG.
3151 4.83 KVM_ARM_PREFERRED_TARGET
3152 -----------------------------
3155 :Architectures: arm, arm64
3157 :Parameters: struct kvm_vcpu_init (out)
3158 :Returns: 0 on success; -1 on error
3162 ====== ==========================================
3163 ENODEV no preferred target available for the host
3164 ====== ==========================================
3166 This queries KVM for preferred CPU target type which can be emulated
3167 by KVM on underlying host.
3169 The ioctl returns struct kvm_vcpu_init instance containing information
3170 about preferred CPU target type and recommended features for it. The
3171 kvm_vcpu_init->features bitmap returned will have feature bits set if
3172 the preferred target recommends setting these features, but this is
3175 The information returned by this ioctl can be used to prepare an instance
3176 of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in
3177 VCPU matching underlying host.
3180 4.84 KVM_GET_REG_LIST
3181 ---------------------
3184 :Architectures: arm, arm64, mips
3186 :Parameters: struct kvm_reg_list (in/out)
3187 :Returns: 0 on success; -1 on error
3191 ===== ==============================================================
3192 Â E2BIG Â Â Â Â the reg index list is too big to fit in the array specified by
3193 Â Â Â Â Â Â Â Â Â Â Â Â the user (the number required will be written into n).
3194 ===== ==============================================================
3198 struct kvm_reg_list {
3199 __u64 n; /* number of registers in reg[] */
3203 This ioctl returns the guest registers that are supported for the
3204 KVM_GET_ONE_REG/KVM_SET_ONE_REG calls.
3207 4.85 KVM_ARM_SET_DEVICE_ADDR (deprecated)
3208 -----------------------------------------
3210 :Capability: KVM_CAP_ARM_SET_DEVICE_ADDR
3211 :Architectures: arm, arm64
3213 :Parameters: struct kvm_arm_device_address (in)
3214 :Returns: 0 on success, -1 on error
3218 ====== ============================================
3219 ENODEV The device id is unknown
3220 ENXIO Device not supported on current system
3221 EEXIST Address already set
3222 E2BIG Address outside guest physical address space
3223 EBUSY Address overlaps with other device range
3224 ====== ============================================
3228 struct kvm_arm_device_addr {
3233 Specify a device address in the guest's physical address space where guests
3234 can access emulated or directly exposed devices, which the host kernel needs
3235 to know about. The id field is an architecture specific identifier for a
3238 ARM/arm64 divides the id field into two parts, a device id and an
3239 address type id specific to the individual device::
3241 Â bits: | 63 ... 32 | 31 ... 16 | 15 ... 0 |
3242 field: | 0x00000000 | device id | addr type id |
3244 ARM/arm64 currently only require this when using the in-kernel GIC
3245 support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2
3246 as the device id. When setting the base address for the guest's
3247 mapping of the VGIC virtual CPU and distributor interface, the ioctl
3248 must be called after calling KVM_CREATE_IRQCHIP, but before calling
3249 KVM_RUN on any of the VCPUs. Calling this ioctl twice for any of the
3250 base addresses will return -EEXIST.
3252 Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API
3253 should be used instead.
3256 4.86 KVM_PPC_RTAS_DEFINE_TOKEN
3257 ------------------------------
3259 :Capability: KVM_CAP_PPC_RTAS
3262 :Parameters: struct kvm_rtas_token_args
3263 :Returns: 0 on success, -1 on error
3265 Defines a token value for a RTAS (Run Time Abstraction Services)
3266 service in order to allow it to be handled in the kernel. The
3267 argument struct gives the name of the service, which must be the name
3268 of a service that has a kernel-side implementation. If the token
3269 value is non-zero, it will be associated with that service, and
3270 subsequent RTAS calls by the guest specifying that token will be
3271 handled by the kernel. If the token value is 0, then any token
3272 associated with the service will be forgotten, and subsequent RTAS
3273 calls by the guest for that service will be passed to userspace to be
3276 4.87 KVM_SET_GUEST_DEBUG
3277 ------------------------
3279 :Capability: KVM_CAP_SET_GUEST_DEBUG
3280 :Architectures: x86, s390, ppc, arm64
3282 :Parameters: struct kvm_guest_debug (in)
3283 :Returns: 0 on success; -1 on error
3287 struct kvm_guest_debug {
3290 struct kvm_guest_debug_arch arch;
3293 Set up the processor specific debug registers and configure vcpu for
3294 handling guest debug events. There are two parts to the structure, the
3295 first a control bitfield indicates the type of debug events to handle
3296 when running. Common control bits are:
3298 - KVM_GUESTDBG_ENABLE: guest debugging is enabled
3299 - KVM_GUESTDBG_SINGLESTEP: the next run should single-step
3301 The top 16 bits of the control field are architecture specific control
3302 flags which can include the following:
3304 - KVM_GUESTDBG_USE_SW_BP: using software breakpoints [x86, arm64]
3305 - KVM_GUESTDBG_USE_HW_BP: using hardware breakpoints [x86, s390, arm64]
3306 - KVM_GUESTDBG_INJECT_DB: inject DB type exception [x86]
3307 - KVM_GUESTDBG_INJECT_BP: inject BP type exception [x86]
3308 - KVM_GUESTDBG_EXIT_PENDING: trigger an immediate guest exit [s390]
3310 For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints
3311 are enabled in memory so we need to ensure breakpoint exceptions are
3312 correctly trapped and the KVM run loop exits at the breakpoint and not
3313 running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP
3314 we need to ensure the guest vCPUs architecture specific registers are
3315 updated to the correct (supplied) values.
3317 The second part of the structure is architecture specific and
3318 typically contains a set of debug registers.
3320 For arm64 the number of debug registers is implementation defined and
3321 can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and
3322 KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number
3323 indicating the number of supported registers.
3325 For ppc, the KVM_CAP_PPC_GUEST_DEBUG_SSTEP capability indicates whether
3326 the single-step debug event (KVM_GUESTDBG_SINGLESTEP) is supported.
3328 When debug events exit the main run loop with the reason
3329 KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run
3330 structure containing architecture specific debug information.
3332 4.88 KVM_GET_EMULATED_CPUID
3333 ---------------------------
3335 :Capability: KVM_CAP_EXT_EMUL_CPUID
3338 :Parameters: struct kvm_cpuid2 (in/out)
3339 :Returns: 0 on success, -1 on error
3346 struct kvm_cpuid_entry2 entries[0];
3349 The member 'flags' is used for passing flags from userspace.
3353 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
3354 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */
3355 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */
3357 struct kvm_cpuid_entry2 {
3368 This ioctl returns x86 cpuid features which are emulated by
3369 kvm.Userspace can use the information returned by this ioctl to query
3370 which features are emulated by kvm instead of being present natively.
3372 Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2
3373 structure with the 'nent' field indicating the number of entries in
3374 the variable-size array 'entries'. If the number of entries is too low
3375 to describe the cpu capabilities, an error (E2BIG) is returned. If the
3376 number is too high, the 'nent' field is adjusted and an error (ENOMEM)
3377 is returned. If the number is just right, the 'nent' field is adjusted
3378 to the number of valid entries in the 'entries' array, which is then
3381 The entries returned are the set CPUID bits of the respective features
3382 which kvm emulates, as returned by the CPUID instruction, with unknown
3383 or unsupported feature bits cleared.
3385 Features like x2apic, for example, may not be present in the host cpu
3386 but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be
3387 emulated efficiently and thus not included here.
3389 The fields in each entry are defined as follows:
3392 the eax value used to obtain the entry
3394 the ecx value used to obtain the entry (for entries that are
3397 an OR of zero or more of the following:
3399 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
3400 if the index field is valid
3404 the values returned by the cpuid instruction for
3405 this function/index combination
3407 4.89 KVM_S390_MEM_OP
3408 --------------------
3410 :Capability: KVM_CAP_S390_MEM_OP
3411 :Architectures: s390
3413 :Parameters: struct kvm_s390_mem_op (in)
3414 :Returns: = 0 on success,
3415 < 0 on generic error (e.g. -EFAULT or -ENOMEM),
3416 > 0 if an exception occurred while walking the page tables
3418 Read or write data from/to the logical (virtual) memory of a VCPU.
3420 Parameters are specified via the following structure::
3422 struct kvm_s390_mem_op {
3423 __u64 gaddr; /* the guest address */
3424 __u64 flags; /* flags */
3425 __u32 size; /* amount of bytes */
3426 __u32 op; /* type of operation */
3427 __u64 buf; /* buffer in userspace */
3428 __u8 ar; /* the access register number */
3429 __u8 reserved[31]; /* should be set to 0 */
3432 The type of operation is specified in the "op" field. It is either
3433 KVM_S390_MEMOP_LOGICAL_READ for reading from logical memory space or
3434 KVM_S390_MEMOP_LOGICAL_WRITE for writing to logical memory space. The
3435 KVM_S390_MEMOP_F_CHECK_ONLY flag can be set in the "flags" field to check
3436 whether the corresponding memory access would create an access exception
3437 (without touching the data in the memory at the destination). In case an
3438 access exception occurred while walking the MMU tables of the guest, the
3439 ioctl returns a positive error number to indicate the type of exception.
3440 This exception is also raised directly at the corresponding VCPU if the
3441 flag KVM_S390_MEMOP_F_INJECT_EXCEPTION is set in the "flags" field.
3443 The start address of the memory region has to be specified in the "gaddr"
3444 field, and the length of the region in the "size" field (which must not
3445 be 0). The maximum value for "size" can be obtained by checking the
3446 KVM_CAP_S390_MEM_OP capability. "buf" is the buffer supplied by the
3447 userspace application where the read data should be written to for
3448 KVM_S390_MEMOP_LOGICAL_READ, or where the data that should be written is
3449 stored for a KVM_S390_MEMOP_LOGICAL_WRITE. When KVM_S390_MEMOP_F_CHECK_ONLY
3450 is specified, "buf" is unused and can be NULL. "ar" designates the access
3451 register number to be used; the valid range is 0..15.
3453 The "reserved" field is meant for future extensions. It is not used by
3454 KVM with the currently defined set of flags.
3456 4.90 KVM_S390_GET_SKEYS
3457 -----------------------
3459 :Capability: KVM_CAP_S390_SKEYS
3460 :Architectures: s390
3462 :Parameters: struct kvm_s390_skeys
3463 :Returns: 0 on success, KVM_S390_GET_KEYS_NONE if guest is not using storage
3464 keys, negative value on error
3466 This ioctl is used to get guest storage key values on the s390
3467 architecture. The ioctl takes parameters via the kvm_s390_skeys struct::
3469 struct kvm_s390_skeys {
3472 __u64 skeydata_addr;
3477 The start_gfn field is the number of the first guest frame whose storage keys
3480 The count field is the number of consecutive frames (starting from start_gfn)
3481 whose storage keys to get. The count field must be at least 1 and the maximum
3482 allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range
3483 will cause the ioctl to return -EINVAL.
3485 The skeydata_addr field is the address to a buffer large enough to hold count
3486 bytes. This buffer will be filled with storage key data by the ioctl.
3488 4.91 KVM_S390_SET_SKEYS
3489 -----------------------
3491 :Capability: KVM_CAP_S390_SKEYS
3492 :Architectures: s390
3494 :Parameters: struct kvm_s390_skeys
3495 :Returns: 0 on success, negative value on error
3497 This ioctl is used to set guest storage key values on the s390
3498 architecture. The ioctl takes parameters via the kvm_s390_skeys struct.
3499 See section on KVM_S390_GET_SKEYS for struct definition.
3501 The start_gfn field is the number of the first guest frame whose storage keys
3504 The count field is the number of consecutive frames (starting from start_gfn)
3505 whose storage keys to get. The count field must be at least 1 and the maximum
3506 allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range
3507 will cause the ioctl to return -EINVAL.
3509 The skeydata_addr field is the address to a buffer containing count bytes of
3510 storage keys. Each byte in the buffer will be set as the storage key for a
3511 single frame starting at start_gfn for count frames.
3513 Note: If any architecturally invalid key value is found in the given data then
3514 the ioctl will return -EINVAL.
3519 :Capability: KVM_CAP_S390_INJECT_IRQ
3520 :Architectures: s390
3522 :Parameters: struct kvm_s390_irq (in)
3523 :Returns: 0 on success, -1 on error
3528 ====== =================================================================
3529 EINVAL interrupt type is invalid
3530 type is KVM_S390_SIGP_STOP and flag parameter is invalid value,
3531 type is KVM_S390_INT_EXTERNAL_CALL and code is bigger
3532 than the maximum of VCPUs
3533 EBUSY type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped,
3534 type is KVM_S390_SIGP_STOP and a stop irq is already pending,
3535 type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt
3537 ====== =================================================================
3539 Allows to inject an interrupt to the guest.
3541 Using struct kvm_s390_irq as a parameter allows
3542 to inject additional payload which is not
3543 possible via KVM_S390_INTERRUPT.
3545 Interrupt parameters are passed via kvm_s390_irq::
3547 struct kvm_s390_irq {
3550 struct kvm_s390_io_info io;
3551 struct kvm_s390_ext_info ext;
3552 struct kvm_s390_pgm_info pgm;
3553 struct kvm_s390_emerg_info emerg;
3554 struct kvm_s390_extcall_info extcall;
3555 struct kvm_s390_prefix_info prefix;
3556 struct kvm_s390_stop_info stop;
3557 struct kvm_s390_mchk_info mchk;
3562 type can be one of the following:
3564 - KVM_S390_SIGP_STOP - sigp stop; parameter in .stop
3565 - KVM_S390_PROGRAM_INT - program check; parameters in .pgm
3566 - KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix
3567 - KVM_S390_RESTART - restart; no parameters
3568 - KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters
3569 - KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters
3570 - KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg
3571 - KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall
3572 - KVM_S390_MCHK - machine check interrupt; parameters in .mchk
3574 This is an asynchronous vcpu ioctl and can be invoked from any thread.
3576 4.94 KVM_S390_GET_IRQ_STATE
3577 ---------------------------
3579 :Capability: KVM_CAP_S390_IRQ_STATE
3580 :Architectures: s390
3582 :Parameters: struct kvm_s390_irq_state (out)
3583 :Returns: >= number of bytes copied into buffer,
3584 -EINVAL if buffer size is 0,
3585 -ENOBUFS if buffer size is too small to fit all pending interrupts,
3586 -EFAULT if the buffer address was invalid
3588 This ioctl allows userspace to retrieve the complete state of all currently
3589 pending interrupts in a single buffer. Use cases include migration
3590 and introspection. The parameter structure contains the address of a
3591 userspace buffer and its length::
3593 struct kvm_s390_irq_state {
3595 __u32 flags; /* will stay unused for compatibility reasons */
3597 __u32 reserved[4]; /* will stay unused for compatibility reasons */
3600 Userspace passes in the above struct and for each pending interrupt a
3601 struct kvm_s390_irq is copied to the provided buffer.
3603 The structure contains a flags and a reserved field for future extensions. As
3604 the kernel never checked for flags == 0 and QEMU never pre-zeroed flags and
3605 reserved, these fields can not be used in the future without breaking
3608 If -ENOBUFS is returned the buffer provided was too small and userspace
3609 may retry with a bigger buffer.
3611 4.95 KVM_S390_SET_IRQ_STATE
3612 ---------------------------
3614 :Capability: KVM_CAP_S390_IRQ_STATE
3615 :Architectures: s390
3617 :Parameters: struct kvm_s390_irq_state (in)
3618 :Returns: 0 on success,
3619 -EFAULT if the buffer address was invalid,
3620 -EINVAL for an invalid buffer length (see below),
3621 -EBUSY if there were already interrupts pending,
3622 errors occurring when actually injecting the
3623 interrupt. See KVM_S390_IRQ.
3625 This ioctl allows userspace to set the complete state of all cpu-local
3626 interrupts currently pending for the vcpu. It is intended for restoring
3627 interrupt state after a migration. The input parameter is a userspace buffer
3628 containing a struct kvm_s390_irq_state::
3630 struct kvm_s390_irq_state {
3632 __u32 flags; /* will stay unused for compatibility reasons */
3634 __u32 reserved[4]; /* will stay unused for compatibility reasons */
3637 The restrictions for flags and reserved apply as well.
3638 (see KVM_S390_GET_IRQ_STATE)
3640 The userspace memory referenced by buf contains a struct kvm_s390_irq
3641 for each interrupt to be injected into the guest.
3642 If one of the interrupts could not be injected for some reason the
3645 len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0
3646 and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq),
3647 which is the maximum number of possibly pending cpu-local interrupts.
3652 :Capability: KVM_CAP_X86_SMM
3656 :Returns: 0 on success, -1 on error
3658 Queues an SMI on the thread's vcpu.
3660 4.97 KVM_CAP_PPC_MULTITCE
3661 -------------------------
3663 :Capability: KVM_CAP_PPC_MULTITCE
3667 This capability means the kernel is capable of handling hypercalls
3668 H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user
3669 space. This significantly accelerates DMA operations for PPC KVM guests.
3670 User space should expect that its handlers for these hypercalls
3671 are not going to be called if user space previously registered LIOBN
3672 in KVM (via KVM_CREATE_SPAPR_TCE or similar calls).
3674 In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest,
3675 user space might have to advertise it for the guest. For example,
3676 IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is
3677 present in the "ibm,hypertas-functions" device-tree property.
3679 The hypercalls mentioned above may or may not be processed successfully
3680 in the kernel based fast path. If they can not be handled by the kernel,
3681 they will get passed on to user space. So user space still has to have
3682 an implementation for these despite the in kernel acceleration.
3684 This capability is always enabled.
3686 4.98 KVM_CREATE_SPAPR_TCE_64
3687 ----------------------------
3689 :Capability: KVM_CAP_SPAPR_TCE_64
3690 :Architectures: powerpc
3692 :Parameters: struct kvm_create_spapr_tce_64 (in)
3693 :Returns: file descriptor for manipulating the created TCE table
3695 This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit
3696 windows, described in 4.62 KVM_CREATE_SPAPR_TCE
3698 This capability uses extended struct in ioctl interface::
3700 /* for KVM_CAP_SPAPR_TCE_64 */
3701 struct kvm_create_spapr_tce_64 {
3705 __u64 offset; /* in pages */
3706 __u64 size; /* in pages */
3709 The aim of extension is to support an additional bigger DMA window with
3710 a variable page size.
3711 KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and
3712 a bus offset of the corresponding DMA window, @size and @offset are numbers
3715 @flags are not used at the moment.
3717 The rest of functionality is identical to KVM_CREATE_SPAPR_TCE.
3719 4.99 KVM_REINJECT_CONTROL
3720 -------------------------
3722 :Capability: KVM_CAP_REINJECT_CONTROL
3725 :Parameters: struct kvm_reinject_control (in)
3726 :Returns: 0 on success,
3727 -EFAULT if struct kvm_reinject_control cannot be read,
3728 -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier.
3730 i8254 (PIT) has two modes, reinject and !reinject. The default is reinject,
3731 where KVM queues elapsed i8254 ticks and monitors completion of interrupt from
3732 vector(s) that i8254 injects. Reinject mode dequeues a tick and injects its
3733 interrupt whenever there isn't a pending interrupt from i8254.
3734 !reinject mode injects an interrupt as soon as a tick arrives.
3738 struct kvm_reinject_control {
3743 pit_reinject = 0 (!reinject mode) is recommended, unless running an old
3744 operating system that uses the PIT for timing (e.g. Linux 2.4.x).
3746 4.100 KVM_PPC_CONFIGURE_V3_MMU
3747 ------------------------------
3749 :Capability: KVM_CAP_PPC_RADIX_MMU or KVM_CAP_PPC_HASH_MMU_V3
3752 :Parameters: struct kvm_ppc_mmuv3_cfg (in)
3753 :Returns: 0 on success,
3754 -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read,
3755 -EINVAL if the configuration is invalid
3757 This ioctl controls whether the guest will use radix or HPT (hashed
3758 page table) translation, and sets the pointer to the process table for
3763 struct kvm_ppc_mmuv3_cfg {
3765 __u64 process_table;
3768 There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and
3769 KVM_PPC_MMUV3_GTSE. KVM_PPC_MMUV3_RADIX, if set, configures the guest
3770 to use radix tree translation, and if clear, to use HPT translation.
3771 KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest
3772 to be able to use the global TLB and SLB invalidation instructions;
3773 if clear, the guest may not use these instructions.
3775 The process_table field specifies the address and size of the guest
3776 process table, which is in the guest's space. This field is formatted
3777 as the second doubleword of the partition table entry, as defined in
3778 the Power ISA V3.00, Book III section 5.7.6.1.
3780 4.101 KVM_PPC_GET_RMMU_INFO
3781 ---------------------------
3783 :Capability: KVM_CAP_PPC_RADIX_MMU
3786 :Parameters: struct kvm_ppc_rmmu_info (out)
3787 :Returns: 0 on success,
3788 -EFAULT if struct kvm_ppc_rmmu_info cannot be written,
3789 -EINVAL if no useful information can be returned
3791 This ioctl returns a structure containing two things: (a) a list
3792 containing supported radix tree geometries, and (b) a list that maps
3793 page sizes to put in the "AP" (actual page size) field for the tlbie
3794 (TLB invalidate entry) instruction.
3798 struct kvm_ppc_rmmu_info {
3799 struct kvm_ppc_radix_geom {
3804 __u32 ap_encodings[8];
3807 The geometries[] field gives up to 8 supported geometries for the
3808 radix page table, in terms of the log base 2 of the smallest page
3809 size, and the number of bits indexed at each level of the tree, from
3810 the PTE level up to the PGD level in that order. Any unused entries
3811 will have 0 in the page_shift field.
3813 The ap_encodings gives the supported page sizes and their AP field
3814 encodings, encoded with the AP value in the top 3 bits and the log
3815 base 2 of the page size in the bottom 6 bits.
3817 4.102 KVM_PPC_RESIZE_HPT_PREPARE
3818 --------------------------------
3820 :Capability: KVM_CAP_SPAPR_RESIZE_HPT
3821 :Architectures: powerpc
3823 :Parameters: struct kvm_ppc_resize_hpt (in)
3824 :Returns: 0 on successful completion,
3825 >0 if a new HPT is being prepared, the value is an estimated
3826 number of milliseconds until preparation is complete,
3827 -EFAULT if struct kvm_reinject_control cannot be read,
3828 -EINVAL if the supplied shift or flags are invalid,
3829 -ENOMEM if unable to allocate the new HPT,
3830 -ENOSPC if there was a hash collision
3834 struct kvm_ppc_rmmu_info {
3835 struct kvm_ppc_radix_geom {
3840 __u32 ap_encodings[8];
3843 The geometries[] field gives up to 8 supported geometries for the
3844 radix page table, in terms of the log base 2 of the smallest page
3845 size, and the number of bits indexed at each level of the tree, from
3846 the PTE level up to the PGD level in that order. Any unused entries
3847 will have 0 in the page_shift field.
3849 The ap_encodings gives the supported page sizes and their AP field
3850 encodings, encoded with the AP value in the top 3 bits and the log
3851 base 2 of the page size in the bottom 6 bits.
3853 4.102 KVM_PPC_RESIZE_HPT_PREPARE
3854 --------------------------------
3856 :Capability: KVM_CAP_SPAPR_RESIZE_HPT
3857 :Architectures: powerpc
3859 :Parameters: struct kvm_ppc_resize_hpt (in)
3860 :Returns: 0 on successful completion,
3861 >0 if a new HPT is being prepared, the value is an estimated
3862 number of milliseconds until preparation is complete,
3863 -EFAULT if struct kvm_reinject_control cannot be read,
3864 -EINVAL if the supplied shift or flags are invalid,when moving existing
3865 HPT entries to the new HPT,
3866 -EIO on other error conditions
3868 Used to implement the PAPR extension for runtime resizing of a guest's
3869 Hashed Page Table (HPT). Specifically this starts, stops or monitors
3870 the preparation of a new potential HPT for the guest, essentially
3871 implementing the H_RESIZE_HPT_PREPARE hypercall.
3873 If called with shift > 0 when there is no pending HPT for the guest,
3874 this begins preparation of a new pending HPT of size 2^(shift) bytes.
3875 It then returns a positive integer with the estimated number of
3876 milliseconds until preparation is complete.
3878 If called when there is a pending HPT whose size does not match that
3879 requested in the parameters, discards the existing pending HPT and
3880 creates a new one as above.
3882 If called when there is a pending HPT of the size requested, will:
3884 * If preparation of the pending HPT is already complete, return 0
3885 * If preparation of the pending HPT has failed, return an error
3886 code, then discard the pending HPT.
3887 * If preparation of the pending HPT is still in progress, return an
3888 estimated number of milliseconds until preparation is complete.
3890 If called with shift == 0, discards any currently pending HPT and
3891 returns 0 (i.e. cancels any in-progress preparation).
3893 flags is reserved for future expansion, currently setting any bits in
3894 flags will result in an -EINVAL.
3896 Normally this will be called repeatedly with the same parameters until
3897 it returns <= 0. The first call will initiate preparation, subsequent
3898 ones will monitor preparation until it completes or fails.
3902 struct kvm_ppc_resize_hpt {
3908 4.103 KVM_PPC_RESIZE_HPT_COMMIT
3909 -------------------------------
3911 :Capability: KVM_CAP_SPAPR_RESIZE_HPT
3912 :Architectures: powerpc
3914 :Parameters: struct kvm_ppc_resize_hpt (in)
3915 :Returns: 0 on successful completion,
3916 -EFAULT if struct kvm_reinject_control cannot be read,
3917 -EINVAL if the supplied shift or flags are invalid,
3918 -ENXIO is there is no pending HPT, or the pending HPT doesn't
3919 have the requested size,
3920 -EBUSY if the pending HPT is not fully prepared,
3921 -ENOSPC if there was a hash collision when moving existing
3922 HPT entries to the new HPT,
3923 -EIO on other error conditions
3925 Used to implement the PAPR extension for runtime resizing of a guest's
3926 Hashed Page Table (HPT). Specifically this requests that the guest be
3927 transferred to working with the new HPT, essentially implementing the
3928 H_RESIZE_HPT_COMMIT hypercall.
3930 This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has
3931 returned 0 with the same parameters. In other cases
3932 KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or
3933 -EBUSY, though others may be possible if the preparation was started,
3936 This will have undefined effects on the guest if it has not already
3937 placed itself in a quiescent state where no vcpu will make MMU enabled
3940 On succsful completion, the pending HPT will become the guest's active
3941 HPT and the previous HPT will be discarded.
3943 On failure, the guest will still be operating on its previous HPT.
3947 struct kvm_ppc_resize_hpt {
3953 4.104 KVM_X86_GET_MCE_CAP_SUPPORTED
3954 -----------------------------------
3956 :Capability: KVM_CAP_MCE
3959 :Parameters: u64 mce_cap (out)
3960 :Returns: 0 on success, -1 on error
3962 Returns supported MCE capabilities. The u64 mce_cap parameter
3963 has the same format as the MSR_IA32_MCG_CAP register. Supported
3964 capabilities will have the corresponding bits set.
3966 4.105 KVM_X86_SETUP_MCE
3967 -----------------------
3969 :Capability: KVM_CAP_MCE
3972 :Parameters: u64 mcg_cap (in)
3973 :Returns: 0 on success,
3974 -EFAULT if u64 mcg_cap cannot be read,
3975 -EINVAL if the requested number of banks is invalid,
3976 -EINVAL if requested MCE capability is not supported.
3978 Initializes MCE support for use. The u64 mcg_cap parameter
3979 has the same format as the MSR_IA32_MCG_CAP register and
3980 specifies which capabilities should be enabled. The maximum
3981 supported number of error-reporting banks can be retrieved when
3982 checking for KVM_CAP_MCE. The supported capabilities can be
3983 retrieved with KVM_X86_GET_MCE_CAP_SUPPORTED.
3985 4.106 KVM_X86_SET_MCE
3986 ---------------------
3988 :Capability: KVM_CAP_MCE
3991 :Parameters: struct kvm_x86_mce (in)
3992 :Returns: 0 on success,
3993 -EFAULT if struct kvm_x86_mce cannot be read,
3994 -EINVAL if the bank number is invalid,
3995 -EINVAL if VAL bit is not set in status field.
3997 Inject a machine check error (MCE) into the guest. The input
4000 struct kvm_x86_mce {
4010 If the MCE being reported is an uncorrected error, KVM will
4011 inject it as an MCE exception into the guest. If the guest
4012 MCG_STATUS register reports that an MCE is in progress, KVM
4013 causes an KVM_EXIT_SHUTDOWN vmexit.
4015 Otherwise, if the MCE is a corrected error, KVM will just
4016 store it in the corresponding bank (provided this bank is
4017 not holding a previously reported uncorrected error).
4019 4.107 KVM_S390_GET_CMMA_BITS
4020 ----------------------------
4022 :Capability: KVM_CAP_S390_CMMA_MIGRATION
4023 :Architectures: s390
4025 :Parameters: struct kvm_s390_cmma_log (in, out)
4026 :Returns: 0 on success, a negative value on error
4028 This ioctl is used to get the values of the CMMA bits on the s390
4029 architecture. It is meant to be used in two scenarios:
4031 - During live migration to save the CMMA values. Live migration needs
4032 to be enabled via the KVM_REQ_START_MIGRATION VM property.
4033 - To non-destructively peek at the CMMA values, with the flag
4034 KVM_S390_CMMA_PEEK set.
4036 The ioctl takes parameters via the kvm_s390_cmma_log struct. The desired
4037 values are written to a buffer whose location is indicated via the "values"
4038 member in the kvm_s390_cmma_log struct. The values in the input struct are
4039 also updated as needed.
4041 Each CMMA value takes up one byte.
4045 struct kvm_s390_cmma_log {
4056 start_gfn is the number of the first guest frame whose CMMA values are
4059 count is the length of the buffer in bytes,
4061 values points to the buffer where the result will be written to.
4063 If count is greater than KVM_S390_SKEYS_MAX, then it is considered to be
4064 KVM_S390_SKEYS_MAX. KVM_S390_SKEYS_MAX is re-used for consistency with
4067 The result is written in the buffer pointed to by the field values, and
4068 the values of the input parameter are updated as follows.
4070 Depending on the flags, different actions are performed. The only
4071 supported flag so far is KVM_S390_CMMA_PEEK.
4073 The default behaviour if KVM_S390_CMMA_PEEK is not set is:
4074 start_gfn will indicate the first page frame whose CMMA bits were dirty.
4075 It is not necessarily the same as the one passed as input, as clean pages
4078 count will indicate the number of bytes actually written in the buffer.
4079 It can (and very often will) be smaller than the input value, since the
4080 buffer is only filled until 16 bytes of clean values are found (which
4081 are then not copied in the buffer). Since a CMMA migration block needs
4082 the base address and the length, for a total of 16 bytes, we will send
4083 back some clean data if there is some dirty data afterwards, as long as
4084 the size of the clean data does not exceed the size of the header. This
4085 allows to minimize the amount of data to be saved or transferred over
4086 the network at the expense of more roundtrips to userspace. The next
4087 invocation of the ioctl will skip over all the clean values, saving
4088 potentially more than just the 16 bytes we found.
4090 If KVM_S390_CMMA_PEEK is set:
4091 the existing storage attributes are read even when not in migration
4092 mode, and no other action is performed;
4094 the output start_gfn will be equal to the input start_gfn,
4096 the output count will be equal to the input count, except if the end of
4097 memory has been reached.
4100 the field "remaining" will indicate the total number of dirty CMMA values
4101 still remaining, or 0 if KVM_S390_CMMA_PEEK is set and migration mode is
4106 values points to the userspace buffer where the result will be stored.
4108 This ioctl can fail with -ENOMEM if not enough memory can be allocated to
4109 complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
4110 KVM_S390_CMMA_PEEK is not set but migration mode was not enabled, with
4111 -EFAULT if the userspace address is invalid or if no page table is
4112 present for the addresses (e.g. when using hugepages).
4114 4.108 KVM_S390_SET_CMMA_BITS
4115 ----------------------------
4117 :Capability: KVM_CAP_S390_CMMA_MIGRATION
4118 :Architectures: s390
4120 :Parameters: struct kvm_s390_cmma_log (in)
4121 :Returns: 0 on success, a negative value on error
4123 This ioctl is used to set the values of the CMMA bits on the s390
4124 architecture. It is meant to be used during live migration to restore
4125 the CMMA values, but there are no restrictions on its use.
4126 The ioctl takes parameters via the kvm_s390_cmma_values struct.
4127 Each CMMA value takes up one byte.
4131 struct kvm_s390_cmma_log {
4142 start_gfn indicates the starting guest frame number,
4144 count indicates how many values are to be considered in the buffer,
4146 flags is not used and must be 0.
4148 mask indicates which PGSTE bits are to be considered.
4150 remaining is not used.
4152 values points to the buffer in userspace where to store the values.
4154 This ioctl can fail with -ENOMEM if not enough memory can be allocated to
4155 complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
4156 the count field is too large (e.g. more than KVM_S390_CMMA_SIZE_MAX) or
4157 if the flags field was not 0, with -EFAULT if the userspace address is
4158 invalid, if invalid pages are written to (e.g. after the end of memory)
4159 or if no page table is present for the addresses (e.g. when using
4162 4.109 KVM_PPC_GET_CPU_CHAR
4163 --------------------------
4165 :Capability: KVM_CAP_PPC_GET_CPU_CHAR
4166 :Architectures: powerpc
4168 :Parameters: struct kvm_ppc_cpu_char (out)
4169 :Returns: 0 on successful completion,
4170 -EFAULT if struct kvm_ppc_cpu_char cannot be written
4172 This ioctl gives userspace information about certain characteristics
4173 of the CPU relating to speculative execution of instructions and
4174 possible information leakage resulting from speculative execution (see
4175 CVE-2017-5715, CVE-2017-5753 and CVE-2017-5754). The information is
4176 returned in struct kvm_ppc_cpu_char, which looks like this::
4178 struct kvm_ppc_cpu_char {
4179 __u64 character; /* characteristics of the CPU */
4180 __u64 behaviour; /* recommended software behaviour */
4181 __u64 character_mask; /* valid bits in character */
4182 __u64 behaviour_mask; /* valid bits in behaviour */
4185 For extensibility, the character_mask and behaviour_mask fields
4186 indicate which bits of character and behaviour have been filled in by
4187 the kernel. If the set of defined bits is extended in future then
4188 userspace will be able to tell whether it is running on a kernel that
4189 knows about the new bits.
4191 The character field describes attributes of the CPU which can help
4192 with preventing inadvertent information disclosure - specifically,
4193 whether there is an instruction to flash-invalidate the L1 data cache
4194 (ori 30,30,0 or mtspr SPRN_TRIG2,rN), whether the L1 data cache is set
4195 to a mode where entries can only be used by the thread that created
4196 them, whether the bcctr[l] instruction prevents speculation, and
4197 whether a speculation barrier instruction (ori 31,31,0) is provided.
4199 The behaviour field describes actions that software should take to
4200 prevent inadvertent information disclosure, and thus describes which
4201 vulnerabilities the hardware is subject to; specifically whether the
4202 L1 data cache should be flushed when returning to user mode from the
4203 kernel, and whether a speculation barrier should be placed between an
4204 array bounds check and the array access.
4206 These fields use the same bit definitions as the new
4207 H_GET_CPU_CHARACTERISTICS hypercall.
4209 4.110 KVM_MEMORY_ENCRYPT_OP
4210 ---------------------------
4215 :Parameters: an opaque platform specific structure (in/out)
4216 :Returns: 0 on success; -1 on error
4218 If the platform supports creating encrypted VMs then this ioctl can be used
4219 for issuing platform-specific memory encryption commands to manage those
4222 Currently, this ioctl is used for issuing Secure Encrypted Virtualization
4223 (SEV) commands on AMD Processors. The SEV commands are defined in
4224 Documentation/virt/kvm/amd-memory-encryption.rst.
4226 4.111 KVM_MEMORY_ENCRYPT_REG_REGION
4227 -----------------------------------
4232 :Parameters: struct kvm_enc_region (in)
4233 :Returns: 0 on success; -1 on error
4235 This ioctl can be used to register a guest memory region which may
4236 contain encrypted data (e.g. guest RAM, SMRAM etc).
4238 It is used in the SEV-enabled guest. When encryption is enabled, a guest
4239 memory region may contain encrypted data. The SEV memory encryption
4240 engine uses a tweak such that two identical plaintext pages, each at
4241 different locations will have differing ciphertexts. So swapping or
4242 moving ciphertext of those pages will not result in plaintext being
4243 swapped. So relocating (or migrating) physical backing pages for the SEV
4244 guest will require some additional steps.
4246 Note: The current SEV key management spec does not provide commands to
4247 swap or migrate (move) ciphertext pages. Hence, for now we pin the guest
4248 memory region registered with the ioctl.
4250 4.112 KVM_MEMORY_ENCRYPT_UNREG_REGION
4251 -------------------------------------
4256 :Parameters: struct kvm_enc_region (in)
4257 :Returns: 0 on success; -1 on error
4259 This ioctl can be used to unregister the guest memory region registered
4260 with KVM_MEMORY_ENCRYPT_REG_REGION ioctl above.
4262 4.113 KVM_HYPERV_EVENTFD
4263 ------------------------
4265 :Capability: KVM_CAP_HYPERV_EVENTFD
4268 :Parameters: struct kvm_hyperv_eventfd (in)
4270 This ioctl (un)registers an eventfd to receive notifications from the guest on
4271 the specified Hyper-V connection id through the SIGNAL_EVENT hypercall, without
4272 causing a user exit. SIGNAL_EVENT hypercall with non-zero event flag number
4273 (bits 24-31) still triggers a KVM_EXIT_HYPERV_HCALL user exit.
4277 struct kvm_hyperv_eventfd {
4284 The conn_id field should fit within 24 bits::
4286 #define KVM_HYPERV_CONN_ID_MASK 0x00ffffff
4288 The acceptable values for the flags field are::
4290 #define KVM_HYPERV_EVENTFD_DEASSIGN (1 << 0)
4292 :Returns: 0 on success,
4293 -EINVAL if conn_id or flags is outside the allowed range,
4294 -ENOENT on deassign if the conn_id isn't registered,
4295 -EEXIST on assign if the conn_id is already registered
4297 4.114 KVM_GET_NESTED_STATE
4298 --------------------------
4300 :Capability: KVM_CAP_NESTED_STATE
4303 :Parameters: struct kvm_nested_state (in/out)
4304 :Returns: 0 on success, -1 on error
4308 ===== =============================================================
4309 E2BIG the total state size exceeds the value of 'size' specified by
4310 the user; the size required will be written into size.
4311 ===== =============================================================
4315 struct kvm_nested_state {
4321 struct kvm_vmx_nested_state_hdr vmx;
4322 struct kvm_svm_nested_state_hdr svm;
4324 /* Pad the header to 128 bytes. */
4329 struct kvm_vmx_nested_state_data vmx[0];
4330 struct kvm_svm_nested_state_data svm[0];
4334 #define KVM_STATE_NESTED_GUEST_MODE 0x00000001
4335 #define KVM_STATE_NESTED_RUN_PENDING 0x00000002
4336 #define KVM_STATE_NESTED_EVMCS 0x00000004
4338 #define KVM_STATE_NESTED_FORMAT_VMX 0
4339 #define KVM_STATE_NESTED_FORMAT_SVM 1
4341 #define KVM_STATE_NESTED_VMX_VMCS_SIZE 0x1000
4343 #define KVM_STATE_NESTED_VMX_SMM_GUEST_MODE 0x00000001
4344 #define KVM_STATE_NESTED_VMX_SMM_VMXON 0x00000002
4346 #define KVM_STATE_VMX_PREEMPTION_TIMER_DEADLINE 0x00000001
4348 struct kvm_vmx_nested_state_hdr {
4357 __u64 preemption_timer_deadline;
4360 struct kvm_vmx_nested_state_data {
4361 __u8 vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE];
4362 __u8 shadow_vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE];
4365 This ioctl copies the vcpu's nested virtualization state from the kernel to
4368 The maximum size of the state can be retrieved by passing KVM_CAP_NESTED_STATE
4369 to the KVM_CHECK_EXTENSION ioctl().
4371 4.115 KVM_SET_NESTED_STATE
4372 --------------------------
4374 :Capability: KVM_CAP_NESTED_STATE
4377 :Parameters: struct kvm_nested_state (in)
4378 :Returns: 0 on success, -1 on error
4380 This copies the vcpu's kvm_nested_state struct from userspace to the kernel.
4381 For the definition of struct kvm_nested_state, see KVM_GET_NESTED_STATE.
4383 4.116 KVM_(UN)REGISTER_COALESCED_MMIO
4384 -------------------------------------
4386 :Capability: KVM_CAP_COALESCED_MMIO (for coalesced mmio)
4387 KVM_CAP_COALESCED_PIO (for coalesced pio)
4390 :Parameters: struct kvm_coalesced_mmio_zone
4391 :Returns: 0 on success, < 0 on error
4393 Coalesced I/O is a performance optimization that defers hardware
4394 register write emulation so that userspace exits are avoided. It is
4395 typically used to reduce the overhead of emulating frequently accessed
4398 When a hardware register is configured for coalesced I/O, write accesses
4399 do not exit to userspace and their value is recorded in a ring buffer
4400 that is shared between kernel and userspace.
4402 Coalesced I/O is used if one or more write accesses to a hardware
4403 register can be deferred until a read or a write to another hardware
4404 register on the same device. This last access will cause a vmexit and
4405 userspace will process accesses from the ring buffer before emulating
4406 it. That will avoid exiting to userspace on repeated writes.
4408 Coalesced pio is based on coalesced mmio. There is little difference
4409 between coalesced mmio and pio except that coalesced pio records accesses
4412 4.117 KVM_CLEAR_DIRTY_LOG (vm ioctl)
4413 ------------------------------------
4415 :Capability: KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
4416 :Architectures: x86, arm, arm64, mips
4418 :Parameters: struct kvm_dirty_log (in)
4419 :Returns: 0 on success, -1 on error
4423 /* for KVM_CLEAR_DIRTY_LOG */
4424 struct kvm_clear_dirty_log {
4429 void __user *dirty_bitmap; /* one bit per page */
4434 The ioctl clears the dirty status of pages in a memory slot, according to
4435 the bitmap that is passed in struct kvm_clear_dirty_log's dirty_bitmap
4436 field. Bit 0 of the bitmap corresponds to page "first_page" in the
4437 memory slot, and num_pages is the size in bits of the input bitmap.
4438 first_page must be a multiple of 64; num_pages must also be a multiple of
4439 64 unless first_page + num_pages is the size of the memory slot. For each
4440 bit that is set in the input bitmap, the corresponding page is marked "clean"
4441 in KVM's dirty bitmap, and dirty tracking is re-enabled for that page
4442 (for example via write-protection, or by clearing the dirty bit in
4443 a page table entry).
4445 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 specifies
4446 the address space for which you want to return the dirty bitmap.
4447 They must be less than the value that KVM_CHECK_EXTENSION returns for
4448 the KVM_CAP_MULTI_ADDRESS_SPACE capability.
4450 This ioctl is mostly useful when KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
4451 is enabled; for more information, see the description of the capability.
4452 However, it can always be used as long as KVM_CHECK_EXTENSION confirms
4453 that KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is present.
4455 4.118 KVM_GET_SUPPORTED_HV_CPUID
4456 --------------------------------
4458 :Capability: KVM_CAP_HYPERV_CPUID
4461 :Parameters: struct kvm_cpuid2 (in/out)
4462 :Returns: 0 on success, -1 on error
4469 struct kvm_cpuid_entry2 entries[0];
4472 struct kvm_cpuid_entry2 {
4483 This ioctl returns x86 cpuid features leaves related to Hyper-V emulation in
4484 KVM. Userspace can use the information returned by this ioctl to construct
4485 cpuid information presented to guests consuming Hyper-V enlightenments (e.g.
4486 Windows or Hyper-V guests).
4488 CPUID feature leaves returned by this ioctl are defined by Hyper-V Top Level
4489 Functional Specification (TLFS). These leaves can't be obtained with
4490 KVM_GET_SUPPORTED_CPUID ioctl because some of them intersect with KVM feature
4491 leaves (0x40000000, 0x40000001).
4493 Currently, the following list of CPUID leaves are returned:
4494 - HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS
4495 - HYPERV_CPUID_INTERFACE
4496 - HYPERV_CPUID_VERSION
4497 - HYPERV_CPUID_FEATURES
4498 - HYPERV_CPUID_ENLIGHTMENT_INFO
4499 - HYPERV_CPUID_IMPLEMENT_LIMITS
4500 - HYPERV_CPUID_NESTED_FEATURES
4502 HYPERV_CPUID_NESTED_FEATURES leaf is only exposed when Enlightened VMCS was
4503 enabled on the corresponding vCPU (KVM_CAP_HYPERV_ENLIGHTENED_VMCS).
4505 Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure
4506 with the 'nent' field indicating the number of entries in the variable-size
4507 array 'entries'. If the number of entries is too low to describe all Hyper-V
4508 feature leaves, an error (E2BIG) is returned. If the number is more or equal
4509 to the number of Hyper-V feature leaves, the 'nent' field is adjusted to the
4510 number of valid entries in the 'entries' array, which is then filled.
4512 'index' and 'flags' fields in 'struct kvm_cpuid_entry2' are currently reserved,
4513 userspace should not expect to get any particular value there.
4515 4.119 KVM_ARM_VCPU_FINALIZE
4516 ---------------------------
4518 :Architectures: arm, arm64
4520 :Parameters: int feature (in)
4521 :Returns: 0 on success, -1 on error
4525 ====== ==============================================================
4526 EPERM feature not enabled, needs configuration, or already finalized
4527 EINVAL feature unknown or not present
4528 ====== ==============================================================
4530 Recognised values for feature:
4532 ===== ===========================================
4533 arm64 KVM_ARM_VCPU_SVE (requires KVM_CAP_ARM_SVE)
4534 ===== ===========================================
4536 Finalizes the configuration of the specified vcpu feature.
4538 The vcpu must already have been initialised, enabling the affected feature, by
4539 means of a successful KVM_ARM_VCPU_INIT call with the appropriate flag set in
4542 For affected vcpu features, this is a mandatory step that must be performed
4543 before the vcpu is fully usable.
4545 Between KVM_ARM_VCPU_INIT and KVM_ARM_VCPU_FINALIZE, the feature may be
4546 configured by use of ioctls such as KVM_SET_ONE_REG. The exact configuration
4547 that should be performaned and how to do it are feature-dependent.
4549 Other calls that depend on a particular feature being finalized, such as
4550 KVM_RUN, KVM_GET_REG_LIST, KVM_GET_ONE_REG and KVM_SET_ONE_REG, will fail with
4551 -EPERM unless the feature has already been finalized by means of a
4552 KVM_ARM_VCPU_FINALIZE call.
4554 See KVM_ARM_VCPU_INIT for details of vcpu features that require finalization
4557 4.120 KVM_SET_PMU_EVENT_FILTER
4558 ------------------------------
4560 :Capability: KVM_CAP_PMU_EVENT_FILTER
4563 :Parameters: struct kvm_pmu_event_filter (in)
4564 :Returns: 0 on success, -1 on error
4568 struct kvm_pmu_event_filter {
4571 __u32 fixed_counter_bitmap;
4577 This ioctl restricts the set of PMU events that the guest can program.
4578 The argument holds a list of events which will be allowed or denied.
4579 The eventsel+umask of each event the guest attempts to program is compared
4580 against the events field to determine whether the guest should have access.
4581 The events field only controls general purpose counters; fixed purpose
4582 counters are controlled by the fixed_counter_bitmap.
4584 No flags are defined yet, the field must be zero.
4586 Valid values for 'action'::
4588 #define KVM_PMU_EVENT_ALLOW 0
4589 #define KVM_PMU_EVENT_DENY 1
4591 4.121 KVM_PPC_SVM_OFF
4592 ---------------------
4595 :Architectures: powerpc
4598 :Returns: 0 on successful completion,
4602 ====== ================================================================
4603 EINVAL if ultravisor failed to terminate the secure guest
4604 ENOMEM if hypervisor failed to allocate new radix page tables for guest
4605 ====== ================================================================
4607 This ioctl is used to turn off the secure mode of the guest or transition
4608 the guest from secure mode to normal mode. This is invoked when the guest
4609 is reset. This has no effect if called for a normal guest.
4611 This ioctl issues an ultravisor call to terminate the secure guest,
4612 unpins the VPA pages and releases all the device pages that are used to
4613 track the secure pages by hypervisor.
4615 4.122 KVM_S390_NORMAL_RESET
4616 ---------------------------
4618 :Capability: KVM_CAP_S390_VCPU_RESETS
4619 :Architectures: s390
4624 This ioctl resets VCPU registers and control structures according to
4625 the cpu reset definition in the POP (Principles Of Operation).
4627 4.123 KVM_S390_INITIAL_RESET
4628 ----------------------------
4631 :Architectures: s390
4636 This ioctl resets VCPU registers and control structures according to
4637 the initial cpu reset definition in the POP. However, the cpu is not
4638 put into ESA mode. This reset is a superset of the normal reset.
4640 4.124 KVM_S390_CLEAR_RESET
4641 --------------------------
4643 :Capability: KVM_CAP_S390_VCPU_RESETS
4644 :Architectures: s390
4649 This ioctl resets VCPU registers and control structures according to
4650 the clear cpu reset definition in the POP. However, the cpu is not put
4651 into ESA mode. This reset is a superset of the initial reset.
4654 4.125 KVM_S390_PV_COMMAND
4655 -------------------------
4657 :Capability: KVM_CAP_S390_PROTECTED
4658 :Architectures: s390
4660 :Parameters: struct kvm_pv_cmd
4661 :Returns: 0 on success, < 0 on error
4666 __u32 cmd; /* Command to be executed */
4667 __u16 rc; /* Ultravisor return code */
4668 __u16 rrc; /* Ultravisor return reason code */
4669 __u64 data; /* Data or address */
4670 __u32 flags; /* flags for future extensions. Must be 0 for now */
4677 Allocate memory and register the VM with the Ultravisor, thereby
4678 donating memory to the Ultravisor that will become inaccessible to
4679 KVM. All existing CPUs are converted to protected ones. After this
4680 command has succeeded, any CPU added via hotplug will become
4681 protected during its creation as well.
4685 ===== =============================
4686 EINTR an unmasked signal is pending
4687 ===== =============================
4691 Deregister the VM from the Ultravisor and reclaim the memory that
4692 had been donated to the Ultravisor, making it usable by the kernel
4693 again. All registered VCPUs are converted back to non-protected
4696 KVM_PV_VM_SET_SEC_PARMS
4697 Pass the image header from VM memory to the Ultravisor in
4698 preparation of image unpacking and verification.
4701 Unpack (protect and decrypt) a page of the encrypted boot image.
4704 Verify the integrity of the unpacked image. Only if this succeeds,
4705 KVM is allowed to start protected VCPUs.
4708 5. The kvm_run structure
4709 ========================
4711 Application code obtains a pointer to the kvm_run structure by
4712 mmap()ing a vcpu fd. From that point, application code can control
4713 execution by changing fields in kvm_run prior to calling the KVM_RUN
4714 ioctl, and obtain information about the reason KVM_RUN returned by
4715 looking up structure members.
4721 __u8 request_interrupt_window;
4723 Request that KVM_RUN return when it becomes possible to inject external
4724 interrupts into the guest. Useful in conjunction with KVM_INTERRUPT.
4728 __u8 immediate_exit;
4730 This field is polled once when KVM_RUN starts; if non-zero, KVM_RUN
4731 exits immediately, returning -EINTR. In the common scenario where a
4732 signal is used to "kick" a VCPU out of KVM_RUN, this field can be used
4733 to avoid usage of KVM_SET_SIGNAL_MASK, which has worse scalability.
4734 Rather than blocking the signal outside KVM_RUN, userspace can set up
4735 a signal handler that sets run->immediate_exit to a non-zero value.
4737 This field is ignored if KVM_CAP_IMMEDIATE_EXIT is not available.
4746 When KVM_RUN has returned successfully (return value 0), this informs
4747 application code why KVM_RUN has returned. Allowable values for this
4748 field are detailed below.
4752 __u8 ready_for_interrupt_injection;
4754 If request_interrupt_window has been specified, this field indicates
4755 an interrupt can be injected now with KVM_INTERRUPT.
4761 The value of the current interrupt flag. Only valid if in-kernel
4762 local APIC is not used.
4768 More architecture-specific flags detailing state of the VCPU that may
4769 affect the device's behavior. The only currently defined flag is
4770 KVM_RUN_X86_SMM, which is valid on x86 machines and is set if the
4771 VCPU is in system management mode.
4775 /* in (pre_kvm_run), out (post_kvm_run) */
4778 The value of the cr8 register. Only valid if in-kernel local APIC is
4779 not used. Both input and output.
4785 The value of the APIC BASE msr. Only valid if in-kernel local
4786 APIC is not used. Both input and output.
4791 /* KVM_EXIT_UNKNOWN */
4793 __u64 hardware_exit_reason;
4796 If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown
4797 reasons. Further architecture-specific information is available in
4798 hardware_exit_reason.
4802 /* KVM_EXIT_FAIL_ENTRY */
4804 __u64 hardware_entry_failure_reason;
4805 __u32 cpu; /* if KVM_LAST_CPU */
4808 If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due
4809 to unknown reasons. Further architecture-specific information is
4810 available in hardware_entry_failure_reason.
4814 /* KVM_EXIT_EXCEPTION */
4826 #define KVM_EXIT_IO_IN 0
4827 #define KVM_EXIT_IO_OUT 1
4829 __u8 size; /* bytes */
4832 __u64 data_offset; /* relative to kvm_run start */
4835 If exit_reason is KVM_EXIT_IO, then the vcpu has
4836 executed a port I/O instruction which could not be satisfied by kvm.
4837 data_offset describes where the data is located (KVM_EXIT_IO_OUT) or
4838 where kvm expects application code to place the data for the next
4839 KVM_RUN invocation (KVM_EXIT_IO_IN). Data format is a packed array.
4843 /* KVM_EXIT_DEBUG */
4845 struct kvm_debug_exit_arch arch;
4848 If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event
4849 for which architecture specific information is returned.
4861 If exit_reason is KVM_EXIT_MMIO, then the vcpu has
4862 executed a memory-mapped I/O instruction which could not be satisfied
4863 by kvm. The 'data' member contains the written data if 'is_write' is
4864 true, and should be filled by application code otherwise.
4866 The 'data' member contains, in its first 'len' bytes, the value as it would
4867 appear if the VCPU performed a load or store of the appropriate width directly
4872 For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR and
4873 KVM_EXIT_EPR the corresponding
4875 operations are complete (and guest state is consistent) only after userspace
4876 has re-entered the kernel with KVM_RUN. The kernel side will first finish
4877 incomplete operations and then check for pending signals. Userspace
4878 can re-enter the guest with an unmasked signal pending to complete
4883 /* KVM_EXIT_HYPERCALL */
4892 Unused. This was once used for 'hypercall to userspace'. To implement
4893 such functionality, use KVM_EXIT_IO (x86) or KVM_EXIT_MMIO (all except s390).
4895 .. note:: KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO.
4899 /* KVM_EXIT_TPR_ACCESS */
4906 To be documented (KVM_TPR_ACCESS_REPORTING).
4910 /* KVM_EXIT_S390_SIEIC */
4913 __u64 mask; /* psw upper half */
4914 __u64 addr; /* psw lower half */
4923 /* KVM_EXIT_S390_RESET */
4924 #define KVM_S390_RESET_POR 1
4925 #define KVM_S390_RESET_CLEAR 2
4926 #define KVM_S390_RESET_SUBSYSTEM 4
4927 #define KVM_S390_RESET_CPU_INIT 8
4928 #define KVM_S390_RESET_IPL 16
4929 __u64 s390_reset_flags;
4935 /* KVM_EXIT_S390_UCONTROL */
4937 __u64 trans_exc_code;
4941 s390 specific. A page fault has occurred for a user controlled virtual
4942 machine (KVM_VM_S390_UNCONTROL) on it's host page table that cannot be
4943 resolved by the kernel.
4944 The program code and the translation exception code that were placed
4945 in the cpu's lowcore are presented here as defined by the z Architecture
4946 Principles of Operation Book in the Chapter for Dynamic Address Translation
4958 Deprecated - was used for 440 KVM.
4967 MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch
4968 hypercalls and exit with this exit struct that contains all the guest gprs.
4970 If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall.
4971 Userspace can now handle the hypercall and when it's done modify the gprs as
4972 necessary. Upon guest entry all guest GPRs will then be replaced by the values
4977 /* KVM_EXIT_PAPR_HCALL */
4984 This is used on 64-bit PowerPC when emulating a pSeries partition,
4985 e.g. with the 'pseries' machine type in qemu. It occurs when the
4986 guest does a hypercall using the 'sc 1' instruction. The 'nr' field
4987 contains the hypercall number (from the guest R3), and 'args' contains
4988 the arguments (from the guest R4 - R12). Userspace should put the
4989 return code in 'ret' and any extra returned values in args[].
4990 The possible hypercalls are defined in the Power Architecture Platform
4991 Requirements (PAPR) document available from www.power.org (free
4992 developer registration required to access it).
4996 /* KVM_EXIT_S390_TSCH */
4998 __u16 subchannel_id;
4999 __u16 subchannel_nr;
5006 s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled
5007 and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O
5008 interrupt for the target subchannel has been dequeued and subchannel_id,
5009 subchannel_nr, io_int_parm and io_int_word contain the parameters for that
5010 interrupt. ipb is needed for instruction parameter decoding.
5019 On FSL BookE PowerPC chips, the interrupt controller has a fast patch
5020 interrupt acknowledge path to the core. When the core successfully
5021 delivers an interrupt, it automatically populates the EPR register with
5022 the interrupt vector number and acknowledges the interrupt inside
5023 the interrupt controller.
5025 In case the interrupt controller lives in user space, we need to do
5026 the interrupt acknowledge cycle through it to fetch the next to be
5027 delivered interrupt vector using this exit.
5029 It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an
5030 external interrupt has just been delivered into the guest. User space
5031 should put the acknowledged interrupt vector into the 'epr' field.
5035 /* KVM_EXIT_SYSTEM_EVENT */
5037 #define KVM_SYSTEM_EVENT_SHUTDOWN 1
5038 #define KVM_SYSTEM_EVENT_RESET 2
5039 #define KVM_SYSTEM_EVENT_CRASH 3
5044 If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered
5045 a system-level event using some architecture specific mechanism (hypercall
5046 or some special instruction). In case of ARM/ARM64, this is triggered using
5047 HVC instruction based PSCI call from the vcpu. The 'type' field describes
5048 the system-level event type. The 'flags' field describes architecture
5049 specific flags for the system-level event.
5051 Valid values for 'type' are:
5053 - KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the
5054 VM. Userspace is not obliged to honour this, and if it does honour
5055 this does not need to destroy the VM synchronously (ie it may call
5056 KVM_RUN again before shutdown finally occurs).
5057 - KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM.
5058 As with SHUTDOWN, userspace can choose to ignore the request, or
5059 to schedule the reset to occur in the future and may call KVM_RUN again.
5060 - KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest
5061 has requested a crash condition maintenance. Userspace can choose
5062 to ignore the request, or to gather VM memory core dump and/or
5063 reset/shutdown of the VM.
5067 /* KVM_EXIT_IOAPIC_EOI */
5072 Indicates that the VCPU's in-kernel local APIC received an EOI for a
5073 level-triggered IOAPIC interrupt. This exit only triggers when the
5074 IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled);
5075 the userspace IOAPIC should process the EOI and retrigger the interrupt if
5076 it is still asserted. Vector is the LAPIC interrupt vector for which the
5081 struct kvm_hyperv_exit {
5082 #define KVM_EXIT_HYPERV_SYNIC 1
5083 #define KVM_EXIT_HYPERV_HCALL 2
5084 #define KVM_EXIT_HYPERV_SYNDBG 3
5111 /* KVM_EXIT_HYPERV */
5112 struct kvm_hyperv_exit hyperv;
5114 Indicates that the VCPU exits into userspace to process some tasks
5115 related to Hyper-V emulation.
5117 Valid values for 'type' are:
5119 - KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about
5121 Hyper-V SynIC state change. Notification is used to remap SynIC
5122 event/message pages and to enable/disable SynIC messages/events processing
5125 - KVM_EXIT_HYPERV_SYNDBG -- synchronously notify user-space about
5127 Hyper-V Synthetic debugger state change. Notification is used to either update
5128 the pending_page location or to send a control command (send the buffer located
5129 in send_page or recv a buffer to recv_page).
5133 /* KVM_EXIT_ARM_NISV */
5139 Used on arm and arm64 systems. If a guest accesses memory not in a memslot,
5140 KVM will typically return to userspace and ask it to do MMIO emulation on its
5141 behalf. However, for certain classes of instructions, no instruction decode
5142 (direction, length of memory access) is provided, and fetching and decoding
5143 the instruction from the VM is overly complicated to live in the kernel.
5145 Historically, when this situation occurred, KVM would print a warning and kill
5146 the VM. KVM assumed that if the guest accessed non-memslot memory, it was
5147 trying to do I/O, which just couldn't be emulated, and the warning message was
5148 phrased accordingly. However, what happened more often was that a guest bug
5149 caused access outside the guest memory areas which should lead to a more
5150 meaningful warning message and an external abort in the guest, if the access
5151 did not fall within an I/O window.
5153 Userspace implementations can query for KVM_CAP_ARM_NISV_TO_USER, and enable
5154 this capability at VM creation. Once this is done, these types of errors will
5155 instead return to userspace with KVM_EXIT_ARM_NISV, with the valid bits from
5156 the HSR (arm) and ESR_EL2 (arm64) in the esr_iss field, and the faulting IPA
5157 in the fault_ipa field. Userspace can either fix up the access if it's
5158 actually an I/O access by decoding the instruction from guest memory (if it's
5159 very brave) and continue executing the guest, or it can decide to suspend,
5160 dump, or restart the guest.
5162 Note that KVM does not skip the faulting instruction as it does for
5163 KVM_EXIT_MMIO, but userspace has to emulate any change to the processing state
5164 if it decides to decode and emulate the instruction.
5168 /* Fix the size of the union. */
5173 * shared registers between kvm and userspace.
5174 * kvm_valid_regs specifies the register classes set by the host
5175 * kvm_dirty_regs specified the register classes dirtied by userspace
5176 * struct kvm_sync_regs is architecture specific, as well as the
5177 * bits for kvm_valid_regs and kvm_dirty_regs
5179 __u64 kvm_valid_regs;
5180 __u64 kvm_dirty_regs;
5182 struct kvm_sync_regs regs;
5183 char padding[SYNC_REGS_SIZE_BYTES];
5186 If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access
5187 certain guest registers without having to call SET/GET_*REGS. Thus we can
5188 avoid some system call overhead if userspace has to handle the exit.
5189 Userspace can query the validity of the structure by checking
5190 kvm_valid_regs for specific bits. These bits are architecture specific
5191 and usually define the validity of a groups of registers. (e.g. one bit
5192 for general purpose registers)
5194 Please note that the kernel is allowed to use the kvm_run structure as the
5195 primary storage for certain register types. Therefore, the kernel may use the
5196 values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set.
5204 6. Capabilities that can be enabled on vCPUs
5205 ============================================
5207 There are certain capabilities that change the behavior of the virtual CPU or
5208 the virtual machine when enabled. To enable them, please see section 4.37.
5209 Below you can find a list of capabilities and what their effect on the vCPU or
5210 the virtual machine is when enabling them.
5212 The following information is provided along with the description:
5215 which instruction set architectures provide this ioctl.
5216 x86 includes both i386 and x86_64.
5219 whether this is a per-vcpu or per-vm capability.
5222 what parameters are accepted by the capability.
5225 the return value. General error numbers (EBADF, ENOMEM, EINVAL)
5226 are not detailed, but errors with specific meanings are.
5235 :Returns: 0 on success; -1 on error
5237 This capability enables interception of OSI hypercalls that otherwise would
5238 be treated as normal system calls to be injected into the guest. OSI hypercalls
5239 were invented by Mac-on-Linux to have a standardized communication mechanism
5240 between the guest and the host.
5242 When this capability is enabled, KVM_EXIT_OSI can occur.
5245 6.2 KVM_CAP_PPC_PAPR
5246 --------------------
5251 :Returns: 0 on success; -1 on error
5253 This capability enables interception of PAPR hypercalls. PAPR hypercalls are
5254 done using the hypercall instruction "sc 1".
5256 It also sets the guest privilege level to "supervisor" mode. Usually the guest
5257 runs in "hypervisor" privilege mode with a few missing features.
5259 In addition to the above, it changes the semantics of SDR1. In this mode, the
5260 HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the
5261 HTAB invisible to the guest.
5263 When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur.
5271 :Parameters: args[0] is the address of a struct kvm_config_tlb
5272 :Returns: 0 on success; -1 on error
5276 struct kvm_config_tlb {
5283 Configures the virtual CPU's TLB array, establishing a shared memory area
5284 between userspace and KVM. The "params" and "array" fields are userspace
5285 addresses of mmu-type-specific data structures. The "array_len" field is an
5286 safety mechanism, and should be set to the size in bytes of the memory that
5287 userspace has reserved for the array. It must be at least the size dictated
5288 by "mmu_type" and "params".
5290 While KVM_RUN is active, the shared region is under control of KVM. Its
5291 contents are undefined, and any modification by userspace results in
5292 boundedly undefined behavior.
5294 On return from KVM_RUN, the shared region will reflect the current state of
5295 the guest's TLB. If userspace makes any changes, it must call KVM_DIRTY_TLB
5296 to tell KVM which entries have been changed, prior to calling KVM_RUN again
5299 For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV:
5301 - The "params" field is of type "struct kvm_book3e_206_tlb_params".
5302 - The "array" field points to an array of type "struct
5303 kvm_book3e_206_tlb_entry".
5304 - The array consists of all entries in the first TLB, followed by all
5305 entries in the second TLB.
5306 - Within a TLB, entries are ordered first by increasing set number. Within a
5307 set, entries are ordered by way (increasing ESEL).
5308 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1)
5309 where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value.
5310 - The tsize field of mas1 shall be set to 4K on TLB0, even though the
5311 hardware ignores this value for TLB0.
5313 6.4 KVM_CAP_S390_CSS_SUPPORT
5314 ----------------------------
5316 :Architectures: s390
5319 :Returns: 0 on success; -1 on error
5321 This capability enables support for handling of channel I/O instructions.
5323 TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are
5324 handled in-kernel, while the other I/O instructions are passed to userspace.
5326 When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST
5327 SUBCHANNEL intercepts.
5329 Note that even though this capability is enabled per-vcpu, the complete
5330 virtual machine is affected.
5337 :Parameters: args[0] defines whether the proxy facility is active
5338 :Returns: 0 on success; -1 on error
5340 This capability enables or disables the delivery of interrupts through the
5341 external proxy facility.
5343 When enabled (args[0] != 0), every time the guest gets an external interrupt
5344 delivered, it automatically exits into user space with a KVM_EXIT_EPR exit
5345 to receive the topmost interrupt vector.
5347 When disabled (args[0] == 0), behavior is as if this facility is unsupported.
5349 When this capability is enabled, KVM_EXIT_EPR can occur.
5351 6.6 KVM_CAP_IRQ_MPIC
5352 --------------------
5355 :Parameters: args[0] is the MPIC device fd;
5356 args[1] is the MPIC CPU number for this vcpu
5358 This capability connects the vcpu to an in-kernel MPIC device.
5360 6.7 KVM_CAP_IRQ_XICS
5361 --------------------
5365 :Parameters: args[0] is the XICS device fd;
5366 args[1] is the XICS CPU number (server ID) for this vcpu
5368 This capability connects the vcpu to an in-kernel XICS device.
5370 6.8 KVM_CAP_S390_IRQCHIP
5371 ------------------------
5373 :Architectures: s390
5377 This capability enables the in-kernel irqchip for s390. Please refer to
5378 "4.24 KVM_CREATE_IRQCHIP" for details.
5380 6.9 KVM_CAP_MIPS_FPU
5381 --------------------
5383 :Architectures: mips
5385 :Parameters: args[0] is reserved for future use (should be 0).
5387 This capability allows the use of the host Floating Point Unit by the guest. It
5388 allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is
5389 done the ``KVM_REG_MIPS_FPR_*`` and ``KVM_REG_MIPS_FCR_*`` registers can be
5390 accessed (depending on the current guest FPU register mode), and the Status.FR,
5391 Config5.FRE bits are accessible via the KVM API and also from the guest,
5392 depending on them being supported by the FPU.
5394 6.10 KVM_CAP_MIPS_MSA
5395 ---------------------
5397 :Architectures: mips
5399 :Parameters: args[0] is reserved for future use (should be 0).
5401 This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest.
5402 It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest.
5403 Once this is done the ``KVM_REG_MIPS_VEC_*`` and ``KVM_REG_MIPS_MSA_*``
5404 registers can be accessed, and the Config5.MSAEn bit is accessible via the
5405 KVM API and also from the guest.
5407 6.74 KVM_CAP_SYNC_REGS
5408 ----------------------
5410 :Architectures: s390, x86
5411 :Target: s390: always enabled, x86: vcpu
5413 :Returns: x86: KVM_CHECK_EXTENSION returns a bit-array indicating which register
5415 (bitfields defined in arch/x86/include/uapi/asm/kvm.h).
5417 As described above in the kvm_sync_regs struct info in section 5 (kvm_run):
5418 KVM_CAP_SYNC_REGS "allow[s] userspace to access certain guest registers
5419 without having to call SET/GET_*REGS". This reduces overhead by eliminating
5420 repeated ioctl calls for setting and/or getting register values. This is
5421 particularly important when userspace is making synchronous guest state
5422 modifications, e.g. when emulating and/or intercepting instructions in
5425 For s390 specifics, please refer to the source code.
5429 - the register sets to be copied out to kvm_run are selectable
5430 by userspace (rather that all sets being copied out for every exit).
5431 - vcpu_events are available in addition to regs and sregs.
5433 For x86, the 'kvm_valid_regs' field of struct kvm_run is overloaded to
5434 function as an input bit-array field set by userspace to indicate the
5435 specific register sets to be copied out on the next exit.
5437 To indicate when userspace has modified values that should be copied into
5438 the vCPU, the all architecture bitarray field, 'kvm_dirty_regs' must be set.
5439 This is done using the same bitflags as for the 'kvm_valid_regs' field.
5440 If the dirty bit is not set, then the register set values will not be copied
5441 into the vCPU even if they've been modified.
5443 Unused bitfields in the bitarrays must be set to zero.
5447 struct kvm_sync_regs {
5448 struct kvm_regs regs;
5449 struct kvm_sregs sregs;
5450 struct kvm_vcpu_events events;
5453 6.75 KVM_CAP_PPC_IRQ_XIVE
5454 -------------------------
5458 :Parameters: args[0] is the XIVE device fd;
5459 args[1] is the XIVE CPU number (server ID) for this vcpu
5461 This capability connects the vcpu to an in-kernel XIVE device.
5463 7. Capabilities that can be enabled on VMs
5464 ==========================================
5466 There are certain capabilities that change the behavior of the virtual
5467 machine when enabled. To enable them, please see section 4.37. Below
5468 you can find a list of capabilities and what their effect on the VM
5469 is when enabling them.
5471 The following information is provided along with the description:
5474 which instruction set architectures provide this ioctl.
5475 x86 includes both i386 and x86_64.
5478 what parameters are accepted by the capability.
5481 the return value. General error numbers (EBADF, ENOMEM, EINVAL)
5482 are not detailed, but errors with specific meanings are.
5485 7.1 KVM_CAP_PPC_ENABLE_HCALL
5486 ----------------------------
5489 :Parameters: args[0] is the sPAPR hcall number;
5490 args[1] is 0 to disable, 1 to enable in-kernel handling
5492 This capability controls whether individual sPAPR hypercalls (hcalls)
5493 get handled by the kernel or not. Enabling or disabling in-kernel
5494 handling of an hcall is effective across the VM. On creation, an
5495 initial set of hcalls are enabled for in-kernel handling, which
5496 consists of those hcalls for which in-kernel handlers were implemented
5497 before this capability was implemented. If disabled, the kernel will
5498 not to attempt to handle the hcall, but will always exit to userspace
5499 to handle it. Note that it may not make sense to enable some and
5500 disable others of a group of related hcalls, but KVM does not prevent
5501 userspace from doing that.
5503 If the hcall number specified is not one that has an in-kernel
5504 implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL
5507 7.2 KVM_CAP_S390_USER_SIGP
5508 --------------------------
5510 :Architectures: s390
5513 This capability controls which SIGP orders will be handled completely in user
5514 space. With this capability enabled, all fast orders will be handled completely
5521 - CONDITIONAL EMERGENCY SIGNAL
5523 All other orders will be handled completely in user space.
5525 Only privileged operation exceptions will be checked for in the kernel (or even
5526 in the hardware prior to interception). If this capability is not enabled, the
5527 old way of handling SIGP orders is used (partially in kernel and user space).
5529 7.3 KVM_CAP_S390_VECTOR_REGISTERS
5530 ---------------------------------
5532 :Architectures: s390
5534 :Returns: 0 on success, negative value on error
5536 Allows use of the vector registers introduced with z13 processor, and
5537 provides for the synchronization between host and user space. Will
5538 return -EINVAL if the machine does not support vectors.
5540 7.4 KVM_CAP_S390_USER_STSI
5541 --------------------------
5543 :Architectures: s390
5546 This capability allows post-handlers for the STSI instruction. After
5547 initial handling in the kernel, KVM exits to user space with
5548 KVM_EXIT_S390_STSI to allow user space to insert further data.
5550 Before exiting to userspace, kvm handlers should fill in s390_stsi field of
5562 @addr - guest address of STSI SYSIB
5566 @ar - access register number
5568 KVM handlers should exit to userspace with rc = -EREMOTE.
5570 7.5 KVM_CAP_SPLIT_IRQCHIP
5571 -------------------------
5574 :Parameters: args[0] - number of routes reserved for userspace IOAPICs
5575 :Returns: 0 on success, -1 on error
5577 Create a local apic for each processor in the kernel. This can be used
5578 instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the
5579 IOAPIC and PIC (and also the PIT, even though this has to be enabled
5582 This capability also enables in kernel routing of interrupt requests;
5583 when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are
5584 used in the IRQ routing table. The first args[0] MSI routes are reserved
5585 for the IOAPIC pins. Whenever the LAPIC receives an EOI for these routes,
5586 a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace.
5588 Fails if VCPU has already been created, or if the irqchip is already in the
5589 kernel (i.e. KVM_CREATE_IRQCHIP has already been called).
5594 :Architectures: s390
5597 Allows use of runtime-instrumentation introduced with zEC12 processor.
5598 Will return -EINVAL if the machine does not support runtime-instrumentation.
5599 Will return -EBUSY if a VCPU has already been created.
5601 7.7 KVM_CAP_X2APIC_API
5602 ----------------------
5605 :Parameters: args[0] - features that should be enabled
5606 :Returns: 0 on success, -EINVAL when args[0] contains invalid features
5608 Valid feature flags in args[0] are::
5610 #define KVM_X2APIC_API_USE_32BIT_IDS (1ULL << 0)
5611 #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK (1ULL << 1)
5613 Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of
5614 KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC,
5615 allowing the use of 32-bit APIC IDs. See KVM_CAP_X2APIC_API in their
5616 respective sections.
5618 KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work
5619 in logical mode or with more than 255 VCPUs. Otherwise, KVM treats 0xff
5620 as a broadcast even in x2APIC mode in order to support physical x2APIC
5621 without interrupt remapping. This is undesirable in logical mode,
5622 where 0xff represents CPUs 0-7 in cluster 0.
5624 7.8 KVM_CAP_S390_USER_INSTR0
5625 ----------------------------
5627 :Architectures: s390
5630 With this capability enabled, all illegal instructions 0x0000 (2 bytes) will
5631 be intercepted and forwarded to user space. User space can use this
5632 mechanism e.g. to realize 2-byte software breakpoints. The kernel will
5633 not inject an operating exception for these instructions, user space has
5634 to take care of that.
5636 This capability can be enabled dynamically even if VCPUs were already
5637 created and are running.
5642 :Architectures: s390
5644 :Returns: 0 on success; -EINVAL if the machine does not support
5645 guarded storage; -EBUSY if a VCPU has already been created.
5647 Allows use of guarded storage for the KVM guest.
5649 7.10 KVM_CAP_S390_AIS
5650 ---------------------
5652 :Architectures: s390
5655 Allow use of adapter-interruption suppression.
5656 :Returns: 0 on success; -EBUSY if a VCPU has already been created.
5658 7.11 KVM_CAP_PPC_SMT
5659 --------------------
5662 :Parameters: vsmt_mode, flags
5664 Enabling this capability on a VM provides userspace with a way to set
5665 the desired virtual SMT mode (i.e. the number of virtual CPUs per
5666 virtual core). The virtual SMT mode, vsmt_mode, must be a power of 2
5667 between 1 and 8. On POWER8, vsmt_mode must also be no greater than
5668 the number of threads per subcore for the host. Currently flags must
5669 be 0. A successful call to enable this capability will result in
5670 vsmt_mode being returned when the KVM_CAP_PPC_SMT capability is
5671 subsequently queried for the VM. This capability is only supported by
5672 HV KVM, and can only be set before any VCPUs have been created.
5673 The KVM_CAP_PPC_SMT_POSSIBLE capability indicates which virtual SMT
5674 modes are available.
5676 7.12 KVM_CAP_PPC_FWNMI
5677 ----------------------
5682 With this capability a machine check exception in the guest address
5683 space will cause KVM to exit the guest with NMI exit reason. This
5684 enables QEMU to build error log and branch to guest kernel registered
5685 machine check handling routine. Without this capability KVM will
5686 branch to guests' 0x200 interrupt vector.
5688 7.13 KVM_CAP_X86_DISABLE_EXITS
5689 ------------------------------
5692 :Parameters: args[0] defines which exits are disabled
5693 :Returns: 0 on success, -EINVAL when args[0] contains invalid exits
5695 Valid bits in args[0] are::
5697 #define KVM_X86_DISABLE_EXITS_MWAIT (1 << 0)
5698 #define KVM_X86_DISABLE_EXITS_HLT (1 << 1)
5699 #define KVM_X86_DISABLE_EXITS_PAUSE (1 << 2)
5700 #define KVM_X86_DISABLE_EXITS_CSTATE (1 << 3)
5702 Enabling this capability on a VM provides userspace with a way to no
5703 longer intercept some instructions for improved latency in some
5704 workloads, and is suggested when vCPUs are associated to dedicated
5705 physical CPUs. More bits can be added in the future; userspace can
5706 just pass the KVM_CHECK_EXTENSION result to KVM_ENABLE_CAP to disable
5709 Do not enable KVM_FEATURE_PV_UNHALT if you disable HLT exits.
5711 7.14 KVM_CAP_S390_HPAGE_1M
5712 --------------------------
5714 :Architectures: s390
5716 :Returns: 0 on success, -EINVAL if hpage module parameter was not set
5717 or cmma is enabled, or the VM has the KVM_VM_S390_UCONTROL
5720 With this capability the KVM support for memory backing with 1m pages
5721 through hugetlbfs can be enabled for a VM. After the capability is
5722 enabled, cmma can't be enabled anymore and pfmfi and the storage key
5723 interpretation are disabled. If cmma has already been enabled or the
5724 hpage module parameter is not set to 1, -EINVAL is returned.
5726 While it is generally possible to create a huge page backed VM without
5727 this capability, the VM will not be able to run.
5729 7.15 KVM_CAP_MSR_PLATFORM_INFO
5730 ------------------------------
5733 :Parameters: args[0] whether feature should be enabled or not
5735 With this capability, a guest may read the MSR_PLATFORM_INFO MSR. Otherwise,
5736 a #GP would be raised when the guest tries to access. Currently, this
5737 capability does not enable write permissions of this MSR for the guest.
5739 7.16 KVM_CAP_PPC_NESTED_HV
5740 --------------------------
5744 :Returns: 0 on success, -EINVAL when the implementation doesn't support
5745 nested-HV virtualization.
5747 HV-KVM on POWER9 and later systems allows for "nested-HV"
5748 virtualization, which provides a way for a guest VM to run guests that
5749 can run using the CPU's supervisor mode (privileged non-hypervisor
5750 state). Enabling this capability on a VM depends on the CPU having
5751 the necessary functionality and on the facility being enabled with a
5752 kvm-hv module parameter.
5754 7.17 KVM_CAP_EXCEPTION_PAYLOAD
5755 ------------------------------
5758 :Parameters: args[0] whether feature should be enabled or not
5760 With this capability enabled, CR2 will not be modified prior to the
5761 emulated VM-exit when L1 intercepts a #PF exception that occurs in
5762 L2. Similarly, for kvm-intel only, DR6 will not be modified prior to
5763 the emulated VM-exit when L1 intercepts a #DB exception that occurs in
5764 L2. As a result, when KVM_GET_VCPU_EVENTS reports a pending #PF (or
5765 #DB) exception for L2, exception.has_payload will be set and the
5766 faulting address (or the new DR6 bits*) will be reported in the
5767 exception_payload field. Similarly, when userspace injects a #PF (or
5768 #DB) into L2 using KVM_SET_VCPU_EVENTS, it is expected to set
5769 exception.has_payload and to put the faulting address - or the new DR6
5770 bits\ [#]_ - in the exception_payload field.
5772 This capability also enables exception.pending in struct
5773 kvm_vcpu_events, which allows userspace to distinguish between pending
5774 and injected exceptions.
5777 .. [#] For the new DR6 bits, note that bit 16 is set iff the #DB exception
5780 7.18 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
5782 :Architectures: x86, arm, arm64, mips
5783 :Parameters: args[0] whether feature should be enabled or not
5787 #define KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (1 << 0)
5788 #define KVM_DIRTY_LOG_INITIALLY_SET (1 << 1)
5790 With KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE is set, KVM_GET_DIRTY_LOG will not
5791 automatically clear and write-protect all pages that are returned as dirty.
5792 Rather, userspace will have to do this operation separately using
5793 KVM_CLEAR_DIRTY_LOG.
5795 At the cost of a slightly more complicated operation, this provides better
5796 scalability and responsiveness for two reasons. First,
5797 KVM_CLEAR_DIRTY_LOG ioctl can operate on a 64-page granularity rather
5798 than requiring to sync a full memslot; this ensures that KVM does not
5799 take spinlocks for an extended period of time. Second, in some cases a
5800 large amount of time can pass between a call to KVM_GET_DIRTY_LOG and
5801 userspace actually using the data in the page. Pages can be modified
5802 during this time, which is inefficient for both the guest and userspace:
5803 the guest will incur a higher penalty due to write protection faults,
5804 while userspace can see false reports of dirty pages. Manual reprotection
5805 helps reducing this time, improving guest performance and reducing the
5806 number of dirty log false positives.
5808 With KVM_DIRTY_LOG_INITIALLY_SET set, all the bits of the dirty bitmap
5809 will be initialized to 1 when created. This also improves performance because
5810 dirty logging can be enabled gradually in small chunks on the first call
5811 to KVM_CLEAR_DIRTY_LOG. KVM_DIRTY_LOG_INITIALLY_SET depends on
5812 KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (it is also only available on
5813 x86 and arm64 for now).
5815 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 was previously available under the name
5816 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT, but the implementation had bugs that make
5817 it hard or impossible to use it correctly. The availability of
5818 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 signals that those bugs are fixed.
5819 Userspace should not try to use KVM_CAP_MANUAL_DIRTY_LOG_PROTECT.
5821 7.19 KVM_CAP_PPC_SECURE_GUEST
5822 ------------------------------
5826 This capability indicates that KVM is running on a host that has
5827 ultravisor firmware and thus can support a secure guest. On such a
5828 system, a guest can ask the ultravisor to make it a secure guest,
5829 one whose memory is inaccessible to the host except for pages which
5830 are explicitly requested to be shared with the host. The ultravisor
5831 notifies KVM when a guest requests to become a secure guest, and KVM
5832 has the opportunity to veto the transition.
5834 If present, this capability can be enabled for a VM, meaning that KVM
5835 will allow the transition to secure guest mode. Otherwise KVM will
5836 veto the transition.
5838 7.20 KVM_CAP_HALT_POLL
5839 ----------------------
5843 :Parameters: args[0] is the maximum poll time in nanoseconds
5844 :Returns: 0 on success; -1 on error
5846 This capability overrides the kvm module parameter halt_poll_ns for the
5849 VCPU polling allows a VCPU to poll for wakeup events instead of immediately
5850 scheduling during guest halts. The maximum time a VCPU can spend polling is
5851 controlled by the kvm module parameter halt_poll_ns. This capability allows
5852 the maximum halt time to specified on a per-VM basis, effectively overriding
5853 the module parameter for the target VM.
5855 8. Other capabilities.
5856 ======================
5858 This section lists capabilities that give information about other
5859 features of the KVM implementation.
5861 8.1 KVM_CAP_PPC_HWRNG
5862 ---------------------
5866 This capability, if KVM_CHECK_EXTENSION indicates that it is
5867 available, means that the kernel has an implementation of the
5868 H_RANDOM hypercall backed by a hardware random-number generator.
5869 If present, the kernel H_RANDOM handler can be enabled for guest use
5870 with the KVM_CAP_PPC_ENABLE_HCALL capability.
5872 8.2 KVM_CAP_HYPERV_SYNIC
5873 ------------------------
5877 This capability, if KVM_CHECK_EXTENSION indicates that it is
5878 available, means that the kernel has an implementation of the
5879 Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is
5880 used to support Windows Hyper-V based guest paravirt drivers(VMBus).
5882 In order to use SynIC, it has to be activated by setting this
5883 capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this
5884 will disable the use of APIC hardware virtualization even if supported
5885 by the CPU, as it's incompatible with SynIC auto-EOI behavior.
5887 8.3 KVM_CAP_PPC_RADIX_MMU
5888 -------------------------
5892 This capability, if KVM_CHECK_EXTENSION indicates that it is
5893 available, means that the kernel can support guests using the
5894 radix MMU defined in Power ISA V3.00 (as implemented in the POWER9
5897 8.4 KVM_CAP_PPC_HASH_MMU_V3
5898 ---------------------------
5902 This capability, if KVM_CHECK_EXTENSION indicates that it is
5903 available, means that the kernel can support guests using the
5904 hashed page table MMU defined in Power ISA V3.00 (as implemented in
5905 the POWER9 processor), including in-memory segment tables.
5910 :Architectures: mips
5912 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
5913 it is available, means that full hardware assisted virtualization capabilities
5914 of the hardware are available for use through KVM. An appropriate
5915 KVM_VM_MIPS_* type must be passed to KVM_CREATE_VM to create a VM which
5918 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
5919 available, it means that the VM is using full hardware assisted virtualization
5920 capabilities of the hardware. This is useful to check after creating a VM with
5921 KVM_VM_MIPS_DEFAULT.
5923 The value returned by KVM_CHECK_EXTENSION should be compared against known
5924 values (see below). All other values are reserved. This is to allow for the
5925 possibility of other hardware assisted virtualization implementations which
5926 may be incompatible with the MIPS VZ ASE.
5928 == ==========================================================================
5929 0 The trap & emulate implementation is in use to run guest code in user
5930 mode. Guest virtual memory segments are rearranged to fit the guest in the
5931 user mode address space.
5933 1 The MIPS VZ ASE is in use, providing full hardware assisted
5934 virtualization, including standard guest virtual memory segments.
5935 == ==========================================================================
5940 :Architectures: mips
5942 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
5943 it is available, means that the trap & emulate implementation is available to
5944 run guest code in user mode, even if KVM_CAP_MIPS_VZ indicates that hardware
5945 assisted virtualisation is also available. KVM_VM_MIPS_TE (0) must be passed
5946 to KVM_CREATE_VM to create a VM which utilises it.
5948 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
5949 available, it means that the VM is using trap & emulate.
5951 8.7 KVM_CAP_MIPS_64BIT
5952 ----------------------
5954 :Architectures: mips
5956 This capability indicates the supported architecture type of the guest, i.e. the
5957 supported register and address width.
5959 The values returned when this capability is checked by KVM_CHECK_EXTENSION on a
5960 kvm VM handle correspond roughly to the CP0_Config.AT register field, and should
5961 be checked specifically against known values (see below). All other values are
5964 == ========================================================================
5965 0 MIPS32 or microMIPS32.
5966 Both registers and addresses are 32-bits wide.
5967 It will only be possible to run 32-bit guest code.
5969 1 MIPS64 or microMIPS64 with access only to 32-bit compatibility segments.
5970 Registers are 64-bits wide, but addresses are 32-bits wide.
5971 64-bit guest code may run but cannot access MIPS64 memory segments.
5972 It will also be possible to run 32-bit guest code.
5974 2 MIPS64 or microMIPS64 with access to all address segments.
5975 Both registers and addresses are 64-bits wide.
5976 It will be possible to run 64-bit or 32-bit guest code.
5977 == ========================================================================
5979 8.9 KVM_CAP_ARM_USER_IRQ
5980 ------------------------
5982 :Architectures: arm, arm64
5984 This capability, if KVM_CHECK_EXTENSION indicates that it is available, means
5985 that if userspace creates a VM without an in-kernel interrupt controller, it
5986 will be notified of changes to the output level of in-kernel emulated devices,
5987 which can generate virtual interrupts, presented to the VM.
5988 For such VMs, on every return to userspace, the kernel
5989 updates the vcpu's run->s.regs.device_irq_level field to represent the actual
5990 output level of the device.
5992 Whenever kvm detects a change in the device output level, kvm guarantees at
5993 least one return to userspace before running the VM. This exit could either
5994 be a KVM_EXIT_INTR or any other exit event, like KVM_EXIT_MMIO. This way,
5995 userspace can always sample the device output level and re-compute the state of
5996 the userspace interrupt controller. Userspace should always check the state
5997 of run->s.regs.device_irq_level on every kvm exit.
5998 The value in run->s.regs.device_irq_level can represent both level and edge
5999 triggered interrupt signals, depending on the device. Edge triggered interrupt
6000 signals will exit to userspace with the bit in run->s.regs.device_irq_level
6001 set exactly once per edge signal.
6003 The field run->s.regs.device_irq_level is available independent of
6004 run->kvm_valid_regs or run->kvm_dirty_regs bits.
6006 If KVM_CAP_ARM_USER_IRQ is supported, the KVM_CHECK_EXTENSION ioctl returns a
6007 number larger than 0 indicating the version of this capability is implemented
6008 and thereby which bits in run->s.regs.device_irq_level can signal values.
6010 Currently the following bits are defined for the device_irq_level bitmap::
6012 KVM_CAP_ARM_USER_IRQ >= 1:
6014 KVM_ARM_DEV_EL1_VTIMER - EL1 virtual timer
6015 KVM_ARM_DEV_EL1_PTIMER - EL1 physical timer
6016 KVM_ARM_DEV_PMU - ARM PMU overflow interrupt signal
6018 Future versions of kvm may implement additional events. These will get
6019 indicated by returning a higher number from KVM_CHECK_EXTENSION and will be
6022 8.10 KVM_CAP_PPC_SMT_POSSIBLE
6023 -----------------------------
6027 Querying this capability returns a bitmap indicating the possible
6028 virtual SMT modes that can be set using KVM_CAP_PPC_SMT. If bit N
6029 (counting from the right) is set, then a virtual SMT mode of 2^N is
6032 8.11 KVM_CAP_HYPERV_SYNIC2
6033 --------------------------
6037 This capability enables a newer version of Hyper-V Synthetic interrupt
6038 controller (SynIC). The only difference with KVM_CAP_HYPERV_SYNIC is that KVM
6039 doesn't clear SynIC message and event flags pages when they are enabled by
6040 writing to the respective MSRs.
6042 8.12 KVM_CAP_HYPERV_VP_INDEX
6043 ----------------------------
6047 This capability indicates that userspace can load HV_X64_MSR_VP_INDEX msr. Its
6048 value is used to denote the target vcpu for a SynIC interrupt. For
6049 compatibilty, KVM initializes this msr to KVM's internal vcpu index. When this
6050 capability is absent, userspace can still query this msr's value.
6052 8.13 KVM_CAP_S390_AIS_MIGRATION
6053 -------------------------------
6055 :Architectures: s390
6058 This capability indicates if the flic device will be able to get/set the
6059 AIS states for migration via the KVM_DEV_FLIC_AISM_ALL attribute and allows
6060 to discover this without having to create a flic device.
6062 8.14 KVM_CAP_S390_PSW
6063 ---------------------
6065 :Architectures: s390
6067 This capability indicates that the PSW is exposed via the kvm_run structure.
6069 8.15 KVM_CAP_S390_GMAP
6070 ----------------------
6072 :Architectures: s390
6074 This capability indicates that the user space memory used as guest mapping can
6075 be anywhere in the user memory address space, as long as the memory slots are
6076 aligned and sized to a segment (1MB) boundary.
6078 8.16 KVM_CAP_S390_COW
6079 ---------------------
6081 :Architectures: s390
6083 This capability indicates that the user space memory used as guest mapping can
6084 use copy-on-write semantics as well as dirty pages tracking via read-only page
6087 8.17 KVM_CAP_S390_BPB
6088 ---------------------
6090 :Architectures: s390
6092 This capability indicates that kvm will implement the interfaces to handle
6093 reset, migration and nested KVM for branch prediction blocking. The stfle
6094 facility 82 should not be provided to the guest without this capability.
6096 8.18 KVM_CAP_HYPERV_TLBFLUSH
6097 ----------------------------
6101 This capability indicates that KVM supports paravirtualized Hyper-V TLB Flush
6103 HvFlushVirtualAddressSpace, HvFlushVirtualAddressSpaceEx,
6104 HvFlushVirtualAddressList, HvFlushVirtualAddressListEx.
6106 8.19 KVM_CAP_ARM_INJECT_SERROR_ESR
6107 ----------------------------------
6109 :Architectures: arm, arm64
6111 This capability indicates that userspace can specify (via the
6112 KVM_SET_VCPU_EVENTS ioctl) the syndrome value reported to the guest when it
6113 takes a virtual SError interrupt exception.
6114 If KVM advertises this capability, userspace can only specify the ISS field for
6115 the ESR syndrome. Other parts of the ESR, such as the EC are generated by the
6116 CPU when the exception is taken. If this virtual SError is taken to EL1 using
6117 AArch64, this value will be reported in the ISS field of ESR_ELx.
6119 See KVM_CAP_VCPU_EVENTS for more details.
6121 8.20 KVM_CAP_HYPERV_SEND_IPI
6122 ----------------------------
6126 This capability indicates that KVM supports paravirtualized Hyper-V IPI send
6128 HvCallSendSyntheticClusterIpi, HvCallSendSyntheticClusterIpiEx.
6130 8.21 KVM_CAP_HYPERV_DIRECT_TLBFLUSH
6131 -----------------------------------
6135 This capability indicates that KVM running on top of Hyper-V hypervisor
6136 enables Direct TLB flush for its guests meaning that TLB flush
6137 hypercalls are handled by Level 0 hypervisor (Hyper-V) bypassing KVM.
6138 Due to the different ABI for hypercall parameters between Hyper-V and
6139 KVM, enabling this capability effectively disables all hypercall
6140 handling by KVM (as some KVM hypercall may be mistakenly treated as TLB
6141 flush hypercalls by Hyper-V) so userspace should disable KVM identification
6142 in CPUID and only exposes Hyper-V identification. In this case, guest
6143 thinks it's running on Hyper-V and only use Hyper-V hypercalls.
6145 8.22 KVM_CAP_S390_VCPU_RESETS
6146 -----------------------------
6148 :Architectures: s390
6150 This capability indicates that the KVM_S390_NORMAL_RESET and
6151 KVM_S390_CLEAR_RESET ioctls are available.
6153 8.23 KVM_CAP_S390_PROTECTED
6154 ---------------------------
6156 :Architectures: s390
6158 This capability indicates that the Ultravisor has been initialized and
6159 KVM can therefore start protected VMs.
6160 This capability governs the KVM_S390_PV_COMMAND ioctl and the
6161 KVM_MP_STATE_LOAD MP_STATE. KVM_SET_MP_STATE can fail for protected
6162 guests when the state change is invalid.
6164 8.24 KVM_CAP_STEAL_TIME
6165 -----------------------
6167 :Architectures: arm64, x86
6169 This capability indicates that KVM supports steal time accounting.
6170 When steal time accounting is supported it may be enabled with
6171 architecture-specific interfaces. This capability and the architecture-
6172 specific interfaces must be consistent, i.e. if one says the feature
6173 is supported, than the other should as well and vice versa. For arm64
6174 see Documentation/virt/kvm/devices/vcpu.rst "KVM_ARM_VCPU_PVTIME_CTRL".
6175 For x86 see Documentation/virt/kvm/msr.rst "MSR_KVM_STEAL_TIME".
6177 8.25 KVM_CAP_S390_DIAG318
6178 -------------------------
6180 :Architectures: s390
6182 This capability enables a guest to set information about its control program
6183 (i.e. guest kernel type and version). The information is helpful during
6184 system/firmware service events, providing additional data about the guest
6185 environments running on the machine.
6187 The information is associated with the DIAGNOSE 0x318 instruction, which sets
6188 an 8-byte value consisting of a one-byte Control Program Name Code (CPNC) and
6189 a 7-byte Control Program Version Code (CPVC). The CPNC determines what
6190 environment the control program is running in (e.g. Linux, z/VM...), and the
6191 CPVC is used for information specific to OS (e.g. Linux version, Linux
6194 If this capability is available, then the CPNC and CPVC can be synchronized
6195 between KVM and userspace via the sync regs mechanism (KVM_SYNC_DIAG318).