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.
265 Besides the size of the KVM_RUN communication region, other areas of
266 the VCPU file descriptor can be mmap-ed, including:
268 - if KVM_CAP_COALESCED_MMIO is available, a page at
269 KVM_COALESCED_MMIO_PAGE_OFFSET * PAGE_SIZE; for historical reasons,
270 this page is included in the result of KVM_GET_VCPU_MMAP_SIZE.
271 KVM_CAP_COALESCED_MMIO is not documented yet.
273 - if KVM_CAP_DIRTY_LOG_RING is available, a number of pages at
274 KVM_DIRTY_LOG_PAGE_OFFSET * PAGE_SIZE. For more information on
275 KVM_CAP_DIRTY_LOG_RING, see section 8.3.
278 4.6 KVM_SET_MEMORY_REGION
279 -------------------------
284 :Parameters: struct kvm_memory_region (in)
285 :Returns: 0 on success, -1 on error
287 This ioctl is obsolete and has been removed.
296 :Parameters: vcpu id (apic id on x86)
297 :Returns: vcpu fd on success, -1 on error
299 This API adds a vcpu to a virtual machine. No more than max_vcpus may be added.
300 The vcpu id is an integer in the range [0, max_vcpu_id).
302 The recommended max_vcpus value can be retrieved using the KVM_CAP_NR_VCPUS of
303 the KVM_CHECK_EXTENSION ioctl() at run-time.
304 The maximum possible value for max_vcpus can be retrieved using the
305 KVM_CAP_MAX_VCPUS of the KVM_CHECK_EXTENSION ioctl() at run-time.
307 If the KVM_CAP_NR_VCPUS does not exist, you should assume that max_vcpus is 4
309 If the KVM_CAP_MAX_VCPUS does not exist, you should assume that max_vcpus is
310 same as the value returned from KVM_CAP_NR_VCPUS.
312 The maximum possible value for max_vcpu_id can be retrieved using the
313 KVM_CAP_MAX_VCPU_ID of the KVM_CHECK_EXTENSION ioctl() at run-time.
315 If the KVM_CAP_MAX_VCPU_ID does not exist, you should assume that max_vcpu_id
316 is the same as the value returned from KVM_CAP_MAX_VCPUS.
318 On powerpc using book3s_hv mode, the vcpus are mapped onto virtual
319 threads in one or more virtual CPU cores. (This is because the
320 hardware requires all the hardware threads in a CPU core to be in the
321 same partition.) The KVM_CAP_PPC_SMT capability indicates the number
322 of vcpus per virtual core (vcore). The vcore id is obtained by
323 dividing the vcpu id by the number of vcpus per vcore. The vcpus in a
324 given vcore will always be in the same physical core as each other
325 (though that might be a different physical core from time to time).
326 Userspace can control the threading (SMT) mode of the guest by its
327 allocation of vcpu ids. For example, if userspace wants
328 single-threaded guest vcpus, it should make all vcpu ids be a multiple
329 of the number of vcpus per vcore.
331 For virtual cpus that have been created with S390 user controlled virtual
332 machines, the resulting vcpu fd can be memory mapped at page offset
333 KVM_S390_SIE_PAGE_OFFSET in order to obtain a memory map of the virtual
334 cpu's hardware control block.
337 4.8 KVM_GET_DIRTY_LOG (vm ioctl)
338 --------------------------------
343 :Parameters: struct kvm_dirty_log (in/out)
344 :Returns: 0 on success, -1 on error
348 /* for KVM_GET_DIRTY_LOG */
349 struct kvm_dirty_log {
353 void __user *dirty_bitmap; /* one bit per page */
358 Given a memory slot, return a bitmap containing any pages dirtied
359 since the last call to this ioctl. Bit 0 is the first page in the
360 memory slot. Ensure the entire structure is cleared to avoid padding
363 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies
364 the address space for which you want to return the dirty bitmap. See
365 KVM_SET_USER_MEMORY_REGION for details on the usage of slot field.
367 The bits in the dirty bitmap are cleared before the ioctl returns, unless
368 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is enabled. For more information,
369 see the description of the capability.
371 4.9 KVM_SET_MEMORY_ALIAS
372 ------------------------
377 :Parameters: struct kvm_memory_alias (in)
378 :Returns: 0 (success), -1 (error)
380 This ioctl is obsolete and has been removed.
390 :Returns: 0 on success, -1 on error
394 ======= ==============================================================
395 EINTR an unmasked signal is pending
396 ENOEXEC the vcpu hasn't been initialized or the guest tried to execute
397 instructions from device memory (arm64)
398 ENOSYS data abort outside memslots with no syndrome info and
399 KVM_CAP_ARM_NISV_TO_USER not enabled (arm64)
400 EPERM SVE feature set but not finalized (arm64)
401 ======= ==============================================================
403 This ioctl is used to run a guest virtual cpu. While there are no
404 explicit parameters, there is an implicit parameter block that can be
405 obtained by mmap()ing the vcpu fd at offset 0, with the size given by
406 KVM_GET_VCPU_MMAP_SIZE. The parameter block is formatted as a 'struct
407 kvm_run' (see below).
414 :Architectures: all except ARM, arm64
416 :Parameters: struct kvm_regs (out)
417 :Returns: 0 on success, -1 on error
419 Reads the general purpose registers from the vcpu.
425 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
426 __u64 rax, rbx, rcx, rdx;
427 __u64 rsi, rdi, rsp, rbp;
428 __u64 r8, r9, r10, r11;
429 __u64 r12, r13, r14, r15;
435 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
447 :Architectures: all except ARM, arm64
449 :Parameters: struct kvm_regs (in)
450 :Returns: 0 on success, -1 on error
452 Writes the general purpose registers into the vcpu.
454 See KVM_GET_REGS for the data structure.
461 :Architectures: x86, ppc
463 :Parameters: struct kvm_sregs (out)
464 :Returns: 0 on success, -1 on error
466 Reads special registers from the vcpu.
472 struct kvm_segment cs, ds, es, fs, gs, ss;
473 struct kvm_segment tr, ldt;
474 struct kvm_dtable gdt, idt;
475 __u64 cr0, cr2, cr3, cr4, cr8;
478 __u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64];
481 /* ppc -- see arch/powerpc/include/uapi/asm/kvm.h */
483 interrupt_bitmap is a bitmap of pending external interrupts. At most
484 one bit may be set. This interrupt has been acknowledged by the APIC
485 but not yet injected into the cpu core.
492 :Architectures: x86, ppc
494 :Parameters: struct kvm_sregs (in)
495 :Returns: 0 on success, -1 on error
497 Writes special registers into the vcpu. See KVM_GET_SREGS for the
507 :Parameters: struct kvm_translation (in/out)
508 :Returns: 0 on success, -1 on error
510 Translates a virtual address according to the vcpu's current address
515 struct kvm_translation {
517 __u64 linear_address;
520 __u64 physical_address;
532 :Architectures: x86, ppc, mips
534 :Parameters: struct kvm_interrupt (in)
535 :Returns: 0 on success, negative on failure.
537 Queues a hardware interrupt vector to be injected.
541 /* for KVM_INTERRUPT */
542 struct kvm_interrupt {
552 ========= ===================================
554 -EEXIST if an interrupt is already enqueued
555 -EINVAL the irq number is invalid
556 -ENXIO if the PIC is in the kernel
557 -EFAULT if the pointer is invalid
558 ========= ===================================
560 Note 'irq' is an interrupt vector, not an interrupt pin or line. This
561 ioctl is useful if the in-kernel PIC is not used.
566 Queues an external interrupt to be injected. This ioctl is overleaded
567 with 3 different irq values:
571 This injects an edge type external interrupt into the guest once it's ready
572 to receive interrupts. When injected, the interrupt is done.
574 b) KVM_INTERRUPT_UNSET
576 This unsets any pending interrupt.
578 Only available with KVM_CAP_PPC_UNSET_IRQ.
580 c) KVM_INTERRUPT_SET_LEVEL
582 This injects a level type external interrupt into the guest context. The
583 interrupt stays pending until a specific ioctl with KVM_INTERRUPT_UNSET
586 Only available with KVM_CAP_PPC_IRQ_LEVEL.
588 Note that any value for 'irq' other than the ones stated above is invalid
589 and incurs unexpected behavior.
591 This is an asynchronous vcpu ioctl and can be invoked from any thread.
596 Queues an external interrupt to be injected into the virtual CPU. A negative
597 interrupt number dequeues the interrupt.
599 This is an asynchronous vcpu ioctl and can be invoked from any thread.
609 :Returns: -1 on error
611 Support for this has been removed. Use KVM_SET_GUEST_DEBUG instead.
617 :Capability: basic (vcpu), KVM_CAP_GET_MSR_FEATURES (system)
619 :Type: system ioctl, vcpu ioctl
620 :Parameters: struct kvm_msrs (in/out)
621 :Returns: number of msrs successfully returned;
624 When used as a system ioctl:
625 Reads the values of MSR-based features that are available for the VM. This
626 is similar to KVM_GET_SUPPORTED_CPUID, but it returns MSR indices and values.
627 The list of msr-based features can be obtained using KVM_GET_MSR_FEATURE_INDEX_LIST
630 When used as a vcpu ioctl:
631 Reads model-specific registers from the vcpu. Supported msr indices can
632 be obtained using KVM_GET_MSR_INDEX_LIST in a system ioctl.
637 __u32 nmsrs; /* number of msrs in entries */
640 struct kvm_msr_entry entries[0];
643 struct kvm_msr_entry {
649 Application code should set the 'nmsrs' member (which indicates the
650 size of the entries array) and the 'index' member of each array entry.
651 kvm will fill in the 'data' member.
660 :Parameters: struct kvm_msrs (in)
661 :Returns: number of msrs successfully set (see below), -1 on error
663 Writes model-specific registers to the vcpu. See KVM_GET_MSRS for the
666 Application code should set the 'nmsrs' member (which indicates the
667 size of the entries array), and the 'index' and 'data' members of each
670 It tries to set the MSRs in array entries[] one by one. If setting an MSR
671 fails, e.g., due to setting reserved bits, the MSR isn't supported/emulated
672 by KVM, etc..., it stops processing the MSR list and returns the number of
673 MSRs that have been set successfully.
682 :Parameters: struct kvm_cpuid (in)
683 :Returns: 0 on success, -1 on error
685 Defines the vcpu responses to the cpuid instruction. Applications
686 should use the KVM_SET_CPUID2 ioctl if available.
688 Note, when this IOCTL fails, KVM gives no guarantees that previous valid CPUID
689 configuration (if there is) is not corrupted. Userspace can get a copy of the
690 resulting CPUID configuration through KVM_GET_CPUID2 in case.
694 struct kvm_cpuid_entry {
703 /* for KVM_SET_CPUID */
707 struct kvm_cpuid_entry entries[0];
711 4.21 KVM_SET_SIGNAL_MASK
712 ------------------------
717 :Parameters: struct kvm_signal_mask (in)
718 :Returns: 0 on success, -1 on error
720 Defines which signals are blocked during execution of KVM_RUN. This
721 signal mask temporarily overrides the threads signal mask. Any
722 unblocked signal received (except SIGKILL and SIGSTOP, which retain
723 their traditional behaviour) will cause KVM_RUN to return with -EINTR.
725 Note the signal will only be delivered if not blocked by the original
730 /* for KVM_SET_SIGNAL_MASK */
731 struct kvm_signal_mask {
743 :Parameters: struct kvm_fpu (out)
744 :Returns: 0 on success, -1 on error
746 Reads the floating point state from the vcpu.
750 /* for KVM_GET_FPU and KVM_SET_FPU */
755 __u8 ftwx; /* in fxsave format */
772 :Parameters: struct kvm_fpu (in)
773 :Returns: 0 on success, -1 on error
775 Writes the floating point state to the vcpu.
779 /* for KVM_GET_FPU and KVM_SET_FPU */
784 __u8 ftwx; /* in fxsave format */
795 4.24 KVM_CREATE_IRQCHIP
796 -----------------------
798 :Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQCHIP (s390)
799 :Architectures: x86, ARM, arm64, s390
802 :Returns: 0 on success, -1 on error
804 Creates an interrupt controller model in the kernel.
805 On x86, creates a virtual ioapic, a virtual PIC (two PICs, nested), and sets up
806 future vcpus to have a local APIC. IRQ routing for GSIs 0-15 is set to both
807 PIC and IOAPIC; GSI 16-23 only go to the IOAPIC.
808 On ARM/arm64, a GICv2 is created. Any other GIC versions require the usage of
809 KVM_CREATE_DEVICE, which also supports creating a GICv2. Using
810 KVM_CREATE_DEVICE is preferred over KVM_CREATE_IRQCHIP for GICv2.
811 On s390, a dummy irq routing table is created.
813 Note that on s390 the KVM_CAP_S390_IRQCHIP vm capability needs to be enabled
814 before KVM_CREATE_IRQCHIP can be used.
820 :Capability: KVM_CAP_IRQCHIP
821 :Architectures: x86, arm, arm64
823 :Parameters: struct kvm_irq_level
824 :Returns: 0 on success, -1 on error
826 Sets the level of a GSI input to the interrupt controller model in the kernel.
827 On some architectures it is required that an interrupt controller model has
828 been previously created with KVM_CREATE_IRQCHIP. Note that edge-triggered
829 interrupts require the level to be set to 1 and then back to 0.
831 On real hardware, interrupt pins can be active-low or active-high. This
832 does not matter for the level field of struct kvm_irq_level: 1 always
833 means active (asserted), 0 means inactive (deasserted).
835 x86 allows the operating system to program the interrupt polarity
836 (active-low/active-high) for level-triggered interrupts, and KVM used
837 to consider the polarity. However, due to bitrot in the handling of
838 active-low interrupts, the above convention is now valid on x86 too.
839 This is signaled by KVM_CAP_X86_IOAPIC_POLARITY_IGNORED. Userspace
840 should not present interrupts to the guest as active-low unless this
841 capability is present (or unless it is not using the in-kernel irqchip,
845 ARM/arm64 can signal an interrupt either at the CPU level, or at the
846 in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to
847 use PPIs designated for specific cpus. The irq field is interpreted
850 Â bits: | 31 ... 28 | 27 ... 24 | 23 ... 16 | 15 ... 0 |
851 field: | vcpu2_index | irq_type | vcpu_index | irq_id |
853 The irq_type field has the following values:
856 out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ
858 in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.)
859 (the vcpu_index field is ignored)
861 in-kernel GIC: PPI, irq_id between 16 and 31 (incl.)
863 (The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs)
865 In both cases, level is used to assert/deassert the line.
867 When KVM_CAP_ARM_IRQ_LINE_LAYOUT_2 is supported, the target vcpu is
868 identified as (256 * vcpu2_index + vcpu_index). Otherwise, vcpu2_index
871 Note that on arm/arm64, the KVM_CAP_IRQCHIP capability only conditions
872 injection of interrupts for the in-kernel irqchip. KVM_IRQ_LINE can always
873 be used for a userspace interrupt controller.
877 struct kvm_irq_level {
880 __s32 status; /* not used for KVM_IRQ_LEVEL */
882 __u32 level; /* 0 or 1 */
889 :Capability: KVM_CAP_IRQCHIP
892 :Parameters: struct kvm_irqchip (in/out)
893 :Returns: 0 on success, -1 on error
895 Reads the state of a kernel interrupt controller created with
896 KVM_CREATE_IRQCHIP into a buffer provided by the caller.
901 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
904 char dummy[512]; /* reserving space */
905 struct kvm_pic_state pic;
906 struct kvm_ioapic_state ioapic;
914 :Capability: KVM_CAP_IRQCHIP
917 :Parameters: struct kvm_irqchip (in)
918 :Returns: 0 on success, -1 on error
920 Sets the state of a kernel interrupt controller created with
921 KVM_CREATE_IRQCHIP from a buffer provided by the caller.
926 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
929 char dummy[512]; /* reserving space */
930 struct kvm_pic_state pic;
931 struct kvm_ioapic_state ioapic;
936 4.28 KVM_XEN_HVM_CONFIG
937 -----------------------
939 :Capability: KVM_CAP_XEN_HVM
942 :Parameters: struct kvm_xen_hvm_config (in)
943 :Returns: 0 on success, -1 on error
945 Sets the MSR that the Xen HVM guest uses to initialize its hypercall
946 page, and provides the starting address and size of the hypercall
947 blobs in userspace. When the guest writes the MSR, kvm copies one
948 page of a blob (32- or 64-bit, depending on the vcpu mode) to guest
953 struct kvm_xen_hvm_config {
963 If the KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL flag is returned from the
964 KVM_CAP_XEN_HVM check, it may be set in the flags field of this ioctl.
965 This requests KVM to generate the contents of the hypercall page
966 automatically; hypercalls will be intercepted and passed to userspace
967 through KVM_EXIT_XEN. In this case, all of the blob size and address
970 No other flags are currently valid in the struct kvm_xen_hvm_config.
975 :Capability: KVM_CAP_ADJUST_CLOCK
978 :Parameters: struct kvm_clock_data (out)
979 :Returns: 0 on success, -1 on error
981 Gets the current timestamp of kvmclock as seen by the current guest. In
982 conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios
985 When KVM_CAP_ADJUST_CLOCK is passed to KVM_CHECK_EXTENSION, it returns the
986 set of bits that KVM can return in struct kvm_clock_data's flag member.
988 The only flag defined now is KVM_CLOCK_TSC_STABLE. If set, the returned
989 value is the exact kvmclock value seen by all VCPUs at the instant
990 when KVM_GET_CLOCK was called. If clear, the returned value is simply
991 CLOCK_MONOTONIC plus a constant offset; the offset can be modified
992 with KVM_SET_CLOCK. KVM will try to make all VCPUs follow this clock,
993 but the exact value read by each VCPU could differ, because the host
998 struct kvm_clock_data {
999 __u64 clock; /* kvmclock current value */
1008 :Capability: KVM_CAP_ADJUST_CLOCK
1011 :Parameters: struct kvm_clock_data (in)
1012 :Returns: 0 on success, -1 on error
1014 Sets the current timestamp of kvmclock to the value specified in its parameter.
1015 In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios
1020 struct kvm_clock_data {
1021 __u64 clock; /* kvmclock current value */
1027 4.31 KVM_GET_VCPU_EVENTS
1028 ------------------------
1030 :Capability: KVM_CAP_VCPU_EVENTS
1031 :Extended by: KVM_CAP_INTR_SHADOW
1032 :Architectures: x86, arm, arm64
1034 :Parameters: struct kvm_vcpu_event (out)
1035 :Returns: 0 on success, -1 on error
1040 Gets currently pending exceptions, interrupts, and NMIs as well as related
1045 struct kvm_vcpu_events {
1049 __u8 has_error_code;
1070 __u8 smm_inside_nmi;
1074 __u8 exception_has_payload;
1075 __u64 exception_payload;
1078 The following bits are defined in the flags field:
1080 - KVM_VCPUEVENT_VALID_SHADOW may be set to signal that
1081 interrupt.shadow contains a valid state.
1083 - KVM_VCPUEVENT_VALID_SMM may be set to signal that smi contains a
1086 - KVM_VCPUEVENT_VALID_PAYLOAD may be set to signal that the
1087 exception_has_payload, exception_payload, and exception.pending
1088 fields contain a valid state. This bit will be set whenever
1089 KVM_CAP_EXCEPTION_PAYLOAD is enabled.
1094 If the guest accesses a device that is being emulated by the host kernel in
1095 such a way that a real device would generate a physical SError, KVM may make
1096 a virtual SError pending for that VCPU. This system error interrupt remains
1097 pending until the guest takes the exception by unmasking PSTATE.A.
1099 Running the VCPU may cause it to take a pending SError, or make an access that
1100 causes an SError to become pending. The event's description is only valid while
1101 the VPCU is not running.
1103 This API provides a way to read and write the pending 'event' state that is not
1104 visible to the guest. To save, restore or migrate a VCPU the struct representing
1105 the state can be read then written using this GET/SET API, along with the other
1106 guest-visible registers. It is not possible to 'cancel' an SError that has been
1109 A device being emulated in user-space may also wish to generate an SError. To do
1110 this the events structure can be populated by user-space. The current state
1111 should be read first, to ensure no existing SError is pending. If an existing
1112 SError is pending, the architecture's 'Multiple SError interrupts' rules should
1113 be followed. (2.5.3 of DDI0587.a "ARM Reliability, Availability, and
1114 Serviceability (RAS) Specification").
1116 SError exceptions always have an ESR value. Some CPUs have the ability to
1117 specify what the virtual SError's ESR value should be. These systems will
1118 advertise KVM_CAP_ARM_INJECT_SERROR_ESR. In this case exception.has_esr will
1119 always have a non-zero value when read, and the agent making an SError pending
1120 should specify the ISS field in the lower 24 bits of exception.serror_esr. If
1121 the system supports KVM_CAP_ARM_INJECT_SERROR_ESR, but user-space sets the events
1122 with exception.has_esr as zero, KVM will choose an ESR.
1124 Specifying exception.has_esr on a system that does not support it will return
1125 -EINVAL. Setting anything other than the lower 24bits of exception.serror_esr
1126 will return -EINVAL.
1128 It is not possible to read back a pending external abort (injected via
1129 KVM_SET_VCPU_EVENTS or otherwise) because such an exception is always delivered
1130 directly to the virtual CPU).
1134 struct kvm_vcpu_events {
1136 __u8 serror_pending;
1137 __u8 serror_has_esr;
1138 __u8 ext_dabt_pending;
1139 /* Align it to 8 bytes */
1146 4.32 KVM_SET_VCPU_EVENTS
1147 ------------------------
1149 :Capability: KVM_CAP_VCPU_EVENTS
1150 :Extended by: KVM_CAP_INTR_SHADOW
1151 :Architectures: x86, arm, arm64
1153 :Parameters: struct kvm_vcpu_event (in)
1154 :Returns: 0 on success, -1 on error
1159 Set pending exceptions, interrupts, and NMIs as well as related states of the
1162 See KVM_GET_VCPU_EVENTS for the data structure.
1164 Fields that may be modified asynchronously by running VCPUs can be excluded
1165 from the update. These fields are nmi.pending, sipi_vector, smi.smm,
1166 smi.pending. Keep the corresponding bits in the flags field cleared to
1167 suppress overwriting the current in-kernel state. The bits are:
1169 =============================== ==================================
1170 KVM_VCPUEVENT_VALID_NMI_PENDING transfer nmi.pending to the kernel
1171 KVM_VCPUEVENT_VALID_SIPI_VECTOR transfer sipi_vector
1172 KVM_VCPUEVENT_VALID_SMM transfer the smi sub-struct.
1173 =============================== ==================================
1175 If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in
1176 the flags field to signal that interrupt.shadow contains a valid state and
1177 shall be written into the VCPU.
1179 KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available.
1181 If KVM_CAP_EXCEPTION_PAYLOAD is enabled, KVM_VCPUEVENT_VALID_PAYLOAD
1182 can be set in the flags field to signal that the
1183 exception_has_payload, exception_payload, and exception.pending fields
1184 contain a valid state and shall be written into the VCPU.
1189 User space may need to inject several types of events to the guest.
1191 Set the pending SError exception state for this VCPU. It is not possible to
1192 'cancel' an Serror that has been made pending.
1194 If the guest performed an access to I/O memory which could not be handled by
1195 userspace, for example because of missing instruction syndrome decode
1196 information or because there is no device mapped at the accessed IPA, then
1197 userspace can ask the kernel to inject an external abort using the address
1198 from the exiting fault on the VCPU. It is a programming error to set
1199 ext_dabt_pending after an exit which was not either KVM_EXIT_MMIO or
1200 KVM_EXIT_ARM_NISV. This feature is only available if the system supports
1201 KVM_CAP_ARM_INJECT_EXT_DABT. This is a helper which provides commonality in
1202 how userspace reports accesses for the above cases to guests, across different
1203 userspace implementations. Nevertheless, userspace can still emulate all Arm
1204 exceptions by manipulating individual registers using the KVM_SET_ONE_REG API.
1206 See KVM_GET_VCPU_EVENTS for the data structure.
1209 4.33 KVM_GET_DEBUGREGS
1210 ----------------------
1212 :Capability: KVM_CAP_DEBUGREGS
1215 :Parameters: struct kvm_debugregs (out)
1216 :Returns: 0 on success, -1 on error
1218 Reads debug registers from the vcpu.
1222 struct kvm_debugregs {
1231 4.34 KVM_SET_DEBUGREGS
1232 ----------------------
1234 :Capability: KVM_CAP_DEBUGREGS
1237 :Parameters: struct kvm_debugregs (in)
1238 :Returns: 0 on success, -1 on error
1240 Writes debug registers into the vcpu.
1242 See KVM_GET_DEBUGREGS for the data structure. The flags field is unused
1243 yet and must be cleared on entry.
1246 4.35 KVM_SET_USER_MEMORY_REGION
1247 -------------------------------
1249 :Capability: KVM_CAP_USER_MEMORY
1252 :Parameters: struct kvm_userspace_memory_region (in)
1253 :Returns: 0 on success, -1 on error
1257 struct kvm_userspace_memory_region {
1260 __u64 guest_phys_addr;
1261 __u64 memory_size; /* bytes */
1262 __u64 userspace_addr; /* start of the userspace allocated memory */
1265 /* for kvm_memory_region::flags */
1266 #define KVM_MEM_LOG_DIRTY_PAGES (1UL << 0)
1267 #define KVM_MEM_READONLY (1UL << 1)
1269 This ioctl allows the user to create, modify or delete a guest physical
1270 memory slot. Bits 0-15 of "slot" specify the slot id and this value
1271 should be less than the maximum number of user memory slots supported per
1272 VM. The maximum allowed slots can be queried using KVM_CAP_NR_MEMSLOTS.
1273 Slots may not overlap in guest physical address space.
1275 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot"
1276 specifies the address space which is being modified. They must be
1277 less than the value that KVM_CHECK_EXTENSION returns for the
1278 KVM_CAP_MULTI_ADDRESS_SPACE capability. Slots in separate address spaces
1279 are unrelated; the restriction on overlapping slots only applies within
1282 Deleting a slot is done by passing zero for memory_size. When changing
1283 an existing slot, it may be moved in the guest physical memory space,
1284 or its flags may be modified, but it may not be resized.
1286 Memory for the region is taken starting at the address denoted by the
1287 field userspace_addr, which must point at user addressable memory for
1288 the entire memory slot size. Any object may back this memory, including
1289 anonymous memory, ordinary files, and hugetlbfs.
1291 On architectures that support a form of address tagging, userspace_addr must
1292 be an untagged address.
1294 It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr
1295 be identical. This allows large pages in the guest to be backed by large
1298 The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and
1299 KVM_MEM_READONLY. The former can be set to instruct KVM to keep track of
1300 writes to memory within the slot. See KVM_GET_DIRTY_LOG ioctl to know how to
1301 use it. The latter can be set, if KVM_CAP_READONLY_MEM capability allows it,
1302 to make a new slot read-only. In this case, writes to this memory will be
1303 posted to userspace as KVM_EXIT_MMIO exits.
1305 When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of
1306 the memory region are automatically reflected into the guest. For example, an
1307 mmap() that affects the region will be made visible immediately. Another
1308 example is madvise(MADV_DROP).
1310 It is recommended to use this API instead of the KVM_SET_MEMORY_REGION ioctl.
1311 The KVM_SET_MEMORY_REGION does not allow fine grained control over memory
1312 allocation and is deprecated.
1315 4.36 KVM_SET_TSS_ADDR
1316 ---------------------
1318 :Capability: KVM_CAP_SET_TSS_ADDR
1321 :Parameters: unsigned long tss_address (in)
1322 :Returns: 0 on success, -1 on error
1324 This ioctl defines the physical address of a three-page region in the guest
1325 physical address space. The region must be within the first 4GB of the
1326 guest physical address space and must not conflict with any memory slot
1327 or any mmio address. The guest may malfunction if it accesses this memory
1330 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1331 because of a quirk in the virtualization implementation (see the internals
1332 documentation when it pops into existence).
1338 :Capability: KVM_CAP_ENABLE_CAP
1339 :Architectures: mips, ppc, s390
1341 :Parameters: struct kvm_enable_cap (in)
1342 :Returns: 0 on success; -1 on error
1344 :Capability: KVM_CAP_ENABLE_CAP_VM
1347 :Parameters: struct kvm_enable_cap (in)
1348 :Returns: 0 on success; -1 on error
1352 Not all extensions are enabled by default. Using this ioctl the application
1353 can enable an extension, making it available to the guest.
1355 On systems that do not support this ioctl, it always fails. On systems that
1356 do support it, it only works for extensions that are supported for enablement.
1358 To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should
1363 struct kvm_enable_cap {
1367 The capability that is supposed to get enabled.
1373 A bitfield indicating future enhancements. Has to be 0 for now.
1379 Arguments for enabling a feature. If a feature needs initial values to
1380 function properly, this is the place to put them.
1387 The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl
1388 for vm-wide capabilities.
1390 4.38 KVM_GET_MP_STATE
1391 ---------------------
1393 :Capability: KVM_CAP_MP_STATE
1394 :Architectures: x86, s390, arm, arm64
1396 :Parameters: struct kvm_mp_state (out)
1397 :Returns: 0 on success; -1 on error
1401 struct kvm_mp_state {
1405 Returns the vcpu's current "multiprocessing state" (though also valid on
1406 uniprocessor guests).
1408 Possible values are:
1410 ========================== ===============================================
1411 KVM_MP_STATE_RUNNABLE the vcpu is currently running [x86,arm/arm64]
1412 KVM_MP_STATE_UNINITIALIZED the vcpu is an application processor (AP)
1413 which has not yet received an INIT signal [x86]
1414 KVM_MP_STATE_INIT_RECEIVED the vcpu has received an INIT signal, and is
1415 now ready for a SIPI [x86]
1416 KVM_MP_STATE_HALTED the vcpu has executed a HLT instruction and
1417 is waiting for an interrupt [x86]
1418 KVM_MP_STATE_SIPI_RECEIVED the vcpu has just received a SIPI (vector
1419 accessible via KVM_GET_VCPU_EVENTS) [x86]
1420 KVM_MP_STATE_STOPPED the vcpu is stopped [s390,arm/arm64]
1421 KVM_MP_STATE_CHECK_STOP the vcpu is in a special error state [s390]
1422 KVM_MP_STATE_OPERATING the vcpu is operating (running or halted)
1424 KVM_MP_STATE_LOAD the vcpu is in a special load/startup state
1426 ========================== ===============================================
1428 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1429 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1430 these architectures.
1435 The only states that are valid are KVM_MP_STATE_STOPPED and
1436 KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not.
1438 4.39 KVM_SET_MP_STATE
1439 ---------------------
1441 :Capability: KVM_CAP_MP_STATE
1442 :Architectures: x86, s390, arm, arm64
1444 :Parameters: struct kvm_mp_state (in)
1445 :Returns: 0 on success; -1 on error
1447 Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for
1450 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1451 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1452 these architectures.
1457 The only states that are valid are KVM_MP_STATE_STOPPED and
1458 KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not.
1460 4.40 KVM_SET_IDENTITY_MAP_ADDR
1461 ------------------------------
1463 :Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR
1466 :Parameters: unsigned long identity (in)
1467 :Returns: 0 on success, -1 on error
1469 This ioctl defines the physical address of a one-page region in the guest
1470 physical address space. The region must be within the first 4GB of the
1471 guest physical address space and must not conflict with any memory slot
1472 or any mmio address. The guest may malfunction if it accesses this memory
1475 Setting the address to 0 will result in resetting the address to its default
1478 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1479 because of a quirk in the virtualization implementation (see the internals
1480 documentation when it pops into existence).
1482 Fails if any VCPU has already been created.
1484 4.41 KVM_SET_BOOT_CPU_ID
1485 ------------------------
1487 :Capability: KVM_CAP_SET_BOOT_CPU_ID
1490 :Parameters: unsigned long vcpu_id
1491 :Returns: 0 on success, -1 on error
1493 Define which vcpu is the Bootstrap Processor (BSP). Values are the same
1494 as the vcpu id in KVM_CREATE_VCPU. If this ioctl is not called, the default
1501 :Capability: KVM_CAP_XSAVE
1504 :Parameters: struct kvm_xsave (out)
1505 :Returns: 0 on success, -1 on error
1514 This ioctl would copy current vcpu's xsave struct to the userspace.
1520 :Capability: KVM_CAP_XSAVE
1523 :Parameters: struct kvm_xsave (in)
1524 :Returns: 0 on success, -1 on error
1533 This ioctl would copy userspace's xsave struct to the kernel.
1539 :Capability: KVM_CAP_XCRS
1542 :Parameters: struct kvm_xcrs (out)
1543 :Returns: 0 on success, -1 on error
1556 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1560 This ioctl would copy current vcpu's xcrs to the userspace.
1566 :Capability: KVM_CAP_XCRS
1569 :Parameters: struct kvm_xcrs (in)
1570 :Returns: 0 on success, -1 on error
1583 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1587 This ioctl would set vcpu's xcr to the value userspace specified.
1590 4.46 KVM_GET_SUPPORTED_CPUID
1591 ----------------------------
1593 :Capability: KVM_CAP_EXT_CPUID
1596 :Parameters: struct kvm_cpuid2 (in/out)
1597 :Returns: 0 on success, -1 on error
1604 struct kvm_cpuid_entry2 entries[0];
1607 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
1608 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */
1609 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */
1611 struct kvm_cpuid_entry2 {
1622 This ioctl returns x86 cpuid features which are supported by both the
1623 hardware and kvm in its default configuration. Userspace can use the
1624 information returned by this ioctl to construct cpuid information (for
1625 KVM_SET_CPUID2) that is consistent with hardware, kernel, and
1626 userspace capabilities, and with user requirements (for example, the
1627 user may wish to constrain cpuid to emulate older hardware, or for
1628 feature consistency across a cluster).
1630 Note that certain capabilities, such as KVM_CAP_X86_DISABLE_EXITS, may
1631 expose cpuid features (e.g. MONITOR) which are not supported by kvm in
1632 its default configuration. If userspace enables such capabilities, it
1633 is responsible for modifying the results of this ioctl appropriately.
1635 Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure
1636 with the 'nent' field indicating the number of entries in the variable-size
1637 array 'entries'. If the number of entries is too low to describe the cpu
1638 capabilities, an error (E2BIG) is returned. If the number is too high,
1639 the 'nent' field is adjusted and an error (ENOMEM) is returned. If the
1640 number is just right, the 'nent' field is adjusted to the number of valid
1641 entries in the 'entries' array, which is then filled.
1643 The entries returned are the host cpuid as returned by the cpuid instruction,
1644 with unknown or unsupported features masked out. Some features (for example,
1645 x2apic), may not be present in the host cpu, but are exposed by kvm if it can
1646 emulate them efficiently. The fields in each entry are defined as follows:
1649 the eax value used to obtain the entry
1652 the ecx value used to obtain the entry (for entries that are
1656 an OR of zero or more of the following:
1658 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
1659 if the index field is valid
1662 the values returned by the cpuid instruction for
1663 this function/index combination
1665 The TSC deadline timer feature (CPUID leaf 1, ecx[24]) is always returned
1666 as false, since the feature depends on KVM_CREATE_IRQCHIP for local APIC
1667 support. Instead it is reported via::
1669 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER)
1671 if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the
1672 feature in userspace, then you can enable the feature for KVM_SET_CPUID2.
1675 4.47 KVM_PPC_GET_PVINFO
1676 -----------------------
1678 :Capability: KVM_CAP_PPC_GET_PVINFO
1681 :Parameters: struct kvm_ppc_pvinfo (out)
1682 :Returns: 0 on success, !0 on error
1686 struct kvm_ppc_pvinfo {
1692 This ioctl fetches PV specific information that need to be passed to the guest
1693 using the device tree or other means from vm context.
1695 The hcall array defines 4 instructions that make up a hypercall.
1697 If any additional field gets added to this structure later on, a bit for that
1698 additional piece of information will be set in the flags bitmap.
1700 The flags bitmap is defined as::
1702 /* the host supports the ePAPR idle hcall
1703 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1<<0)
1705 4.52 KVM_SET_GSI_ROUTING
1706 ------------------------
1708 :Capability: KVM_CAP_IRQ_ROUTING
1709 :Architectures: x86 s390 arm arm64
1711 :Parameters: struct kvm_irq_routing (in)
1712 :Returns: 0 on success, -1 on error
1714 Sets the GSI routing table entries, overwriting any previously set entries.
1716 On arm/arm64, GSI routing has the following limitation:
1718 - GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD.
1722 struct kvm_irq_routing {
1725 struct kvm_irq_routing_entry entries[0];
1728 No flags are specified so far, the corresponding field must be set to zero.
1732 struct kvm_irq_routing_entry {
1738 struct kvm_irq_routing_irqchip irqchip;
1739 struct kvm_irq_routing_msi msi;
1740 struct kvm_irq_routing_s390_adapter adapter;
1741 struct kvm_irq_routing_hv_sint hv_sint;
1746 /* gsi routing entry types */
1747 #define KVM_IRQ_ROUTING_IRQCHIP 1
1748 #define KVM_IRQ_ROUTING_MSI 2
1749 #define KVM_IRQ_ROUTING_S390_ADAPTER 3
1750 #define KVM_IRQ_ROUTING_HV_SINT 4
1754 - KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry
1755 type, specifies that the devid field contains a valid value. The per-VM
1756 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
1757 the device ID. If this capability is not available, userspace should
1758 never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
1763 struct kvm_irq_routing_irqchip {
1768 struct kvm_irq_routing_msi {
1778 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
1779 for the device that wrote the MSI message. For PCI, this is usually a
1780 BFD identifier in the lower 16 bits.
1782 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
1783 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
1784 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
1785 address_hi must be zero.
1789 struct kvm_irq_routing_s390_adapter {
1793 __u32 summary_offset;
1797 struct kvm_irq_routing_hv_sint {
1803 4.55 KVM_SET_TSC_KHZ
1804 --------------------
1806 :Capability: KVM_CAP_TSC_CONTROL
1809 :Parameters: virtual tsc_khz
1810 :Returns: 0 on success, -1 on error
1812 Specifies the tsc frequency for the virtual machine. The unit of the
1816 4.56 KVM_GET_TSC_KHZ
1817 --------------------
1819 :Capability: KVM_CAP_GET_TSC_KHZ
1823 :Returns: virtual tsc-khz on success, negative value on error
1825 Returns the tsc frequency of the guest. The unit of the return value is
1826 KHz. If the host has unstable tsc this ioctl returns -EIO instead as an
1833 :Capability: KVM_CAP_IRQCHIP
1836 :Parameters: struct kvm_lapic_state (out)
1837 :Returns: 0 on success, -1 on error
1841 #define KVM_APIC_REG_SIZE 0x400
1842 struct kvm_lapic_state {
1843 char regs[KVM_APIC_REG_SIZE];
1846 Reads the Local APIC registers and copies them into the input argument. The
1847 data format and layout are the same as documented in the architecture manual.
1849 If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is
1850 enabled, then the format of APIC_ID register depends on the APIC mode
1851 (reported by MSR_IA32_APICBASE) of its VCPU. x2APIC stores APIC ID in
1852 the APIC_ID register (bytes 32-35). xAPIC only allows an 8-bit APIC ID
1853 which is stored in bits 31-24 of the APIC register, or equivalently in
1854 byte 35 of struct kvm_lapic_state's regs field. KVM_GET_LAPIC must then
1855 be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR.
1857 If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state
1858 always uses xAPIC format.
1864 :Capability: KVM_CAP_IRQCHIP
1867 :Parameters: struct kvm_lapic_state (in)
1868 :Returns: 0 on success, -1 on error
1872 #define KVM_APIC_REG_SIZE 0x400
1873 struct kvm_lapic_state {
1874 char regs[KVM_APIC_REG_SIZE];
1877 Copies the input argument into the Local APIC registers. The data format
1878 and layout are the same as documented in the architecture manual.
1880 The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's
1881 regs field) depends on the state of the KVM_CAP_X2APIC_API capability.
1882 See the note in KVM_GET_LAPIC.
1888 :Capability: KVM_CAP_IOEVENTFD
1891 :Parameters: struct kvm_ioeventfd (in)
1892 :Returns: 0 on success, !0 on error
1894 This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address
1895 within the guest. A guest write in the registered address will signal the
1896 provided event instead of triggering an exit.
1900 struct kvm_ioeventfd {
1902 __u64 addr; /* legal pio/mmio address */
1903 __u32 len; /* 0, 1, 2, 4, or 8 bytes */
1909 For the special case of virtio-ccw devices on s390, the ioevent is matched
1910 to a subchannel/virtqueue tuple instead.
1912 The following flags are defined::
1914 #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch)
1915 #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio)
1916 #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign)
1917 #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \
1918 (1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify)
1920 If datamatch flag is set, the event will be signaled only if the written value
1921 to the registered address is equal to datamatch in struct kvm_ioeventfd.
1923 For virtio-ccw devices, addr contains the subchannel id and datamatch the
1926 With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and
1927 the kernel will ignore the length of guest write and may get a faster vmexit.
1928 The speedup may only apply to specific architectures, but the ioeventfd will
1934 :Capability: KVM_CAP_SW_TLB
1937 :Parameters: struct kvm_dirty_tlb (in)
1938 :Returns: 0 on success, -1 on error
1942 struct kvm_dirty_tlb {
1947 This must be called whenever userspace has changed an entry in the shared
1948 TLB, prior to calling KVM_RUN on the associated vcpu.
1950 The "bitmap" field is the userspace address of an array. This array
1951 consists of a number of bits, equal to the total number of TLB entries as
1952 determined by the last successful call to KVM_CONFIG_TLB, rounded up to the
1953 nearest multiple of 64.
1955 Each bit corresponds to one TLB entry, ordered the same as in the shared TLB
1958 The array is little-endian: the bit 0 is the least significant bit of the
1959 first byte, bit 8 is the least significant bit of the second byte, etc.
1960 This avoids any complications with differing word sizes.
1962 The "num_dirty" field is a performance hint for KVM to determine whether it
1963 should skip processing the bitmap and just invalidate everything. It must
1964 be set to the number of set bits in the bitmap.
1967 4.62 KVM_CREATE_SPAPR_TCE
1968 -------------------------
1970 :Capability: KVM_CAP_SPAPR_TCE
1971 :Architectures: powerpc
1973 :Parameters: struct kvm_create_spapr_tce (in)
1974 :Returns: file descriptor for manipulating the created TCE table
1976 This creates a virtual TCE (translation control entry) table, which
1977 is an IOMMU for PAPR-style virtual I/O. It is used to translate
1978 logical addresses used in virtual I/O into guest physical addresses,
1979 and provides a scatter/gather capability for PAPR virtual I/O.
1983 /* for KVM_CAP_SPAPR_TCE */
1984 struct kvm_create_spapr_tce {
1989 The liobn field gives the logical IO bus number for which to create a
1990 TCE table. The window_size field specifies the size of the DMA window
1991 which this TCE table will translate - the table will contain one 64
1992 bit TCE entry for every 4kiB of the DMA window.
1994 When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE
1995 table has been created using this ioctl(), the kernel will handle it
1996 in real mode, updating the TCE table. H_PUT_TCE calls for other
1997 liobns will cause a vm exit and must be handled by userspace.
1999 The return value is a file descriptor which can be passed to mmap(2)
2000 to map the created TCE table into userspace. This lets userspace read
2001 the entries written by kernel-handled H_PUT_TCE calls, and also lets
2002 userspace update the TCE table directly which is useful in some
2006 4.63 KVM_ALLOCATE_RMA
2007 ---------------------
2009 :Capability: KVM_CAP_PPC_RMA
2010 :Architectures: powerpc
2012 :Parameters: struct kvm_allocate_rma (out)
2013 :Returns: file descriptor for mapping the allocated RMA
2015 This allocates a Real Mode Area (RMA) from the pool allocated at boot
2016 time by the kernel. An RMA is a physically-contiguous, aligned region
2017 of memory used on older POWER processors to provide the memory which
2018 will be accessed by real-mode (MMU off) accesses in a KVM guest.
2019 POWER processors support a set of sizes for the RMA that usually
2020 includes 64MB, 128MB, 256MB and some larger powers of two.
2024 /* for KVM_ALLOCATE_RMA */
2025 struct kvm_allocate_rma {
2029 The return value is a file descriptor which can be passed to mmap(2)
2030 to map the allocated RMA into userspace. The mapped area can then be
2031 passed to the KVM_SET_USER_MEMORY_REGION ioctl to establish it as the
2032 RMA for a virtual machine. The size of the RMA in bytes (which is
2033 fixed at host kernel boot time) is returned in the rma_size field of
2034 the argument structure.
2036 The KVM_CAP_PPC_RMA capability is 1 or 2 if the KVM_ALLOCATE_RMA ioctl
2037 is supported; 2 if the processor requires all virtual machines to have
2038 an RMA, or 1 if the processor can use an RMA but doesn't require it,
2039 because it supports the Virtual RMA (VRMA) facility.
2045 :Capability: KVM_CAP_USER_NMI
2049 :Returns: 0 on success, -1 on error
2051 Queues an NMI on the thread's vcpu. Note this is well defined only
2052 when KVM_CREATE_IRQCHIP has not been called, since this is an interface
2053 between the virtual cpu core and virtual local APIC. After KVM_CREATE_IRQCHIP
2054 has been called, this interface is completely emulated within the kernel.
2056 To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the
2057 following algorithm:
2060 - read the local APIC's state (KVM_GET_LAPIC)
2061 - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1)
2062 - if so, issue KVM_NMI
2065 Some guests configure the LINT1 NMI input to cause a panic, aiding in
2069 4.65 KVM_S390_UCAS_MAP
2070 ----------------------
2072 :Capability: KVM_CAP_S390_UCONTROL
2073 :Architectures: s390
2075 :Parameters: struct kvm_s390_ucas_mapping (in)
2076 :Returns: 0 in case of success
2078 The parameter is defined like this::
2080 struct kvm_s390_ucas_mapping {
2086 This ioctl maps the memory at "user_addr" with the length "length" to
2087 the vcpu's address space starting at "vcpu_addr". All parameters need to
2088 be aligned by 1 megabyte.
2091 4.66 KVM_S390_UCAS_UNMAP
2092 ------------------------
2094 :Capability: KVM_CAP_S390_UCONTROL
2095 :Architectures: s390
2097 :Parameters: struct kvm_s390_ucas_mapping (in)
2098 :Returns: 0 in case of success
2100 The parameter is defined like this::
2102 struct kvm_s390_ucas_mapping {
2108 This ioctl unmaps the memory in the vcpu's address space starting at
2109 "vcpu_addr" with the length "length". The field "user_addr" is ignored.
2110 All parameters need to be aligned by 1 megabyte.
2113 4.67 KVM_S390_VCPU_FAULT
2114 ------------------------
2116 :Capability: KVM_CAP_S390_UCONTROL
2117 :Architectures: s390
2119 :Parameters: vcpu absolute address (in)
2120 :Returns: 0 in case of success
2122 This call creates a page table entry on the virtual cpu's address space
2123 (for user controlled virtual machines) or the virtual machine's address
2124 space (for regular virtual machines). This only works for minor faults,
2125 thus it's recommended to access subject memory page via the user page
2126 table upfront. This is useful to handle validity intercepts for user
2127 controlled virtual machines to fault in the virtual cpu's lowcore pages
2128 prior to calling the KVM_RUN ioctl.
2131 4.68 KVM_SET_ONE_REG
2132 --------------------
2134 :Capability: KVM_CAP_ONE_REG
2137 :Parameters: struct kvm_one_reg (in)
2138 :Returns: 0 on success, negative value on failure
2142 ====== ============================================================
2143 Â ENOENT Â Â no such register
2144 Â EINVAL Â Â invalid register ID, or no such register or used with VMs in
2145 protected virtualization mode on s390
2146 Â EPERM Â Â Â (arm64) register access not allowed before vcpu finalization
2147 ====== ============================================================
2149 (These error codes are indicative only: do not rely on a specific error
2150 code being returned in a specific situation.)
2154 struct kvm_one_reg {
2159 Using this ioctl, a single vcpu register can be set to a specific value
2160 defined by user space with the passed in struct kvm_one_reg, where id
2161 refers to the register identifier as described below and addr is a pointer
2162 to a variable with the respective size. There can be architecture agnostic
2163 and architecture specific registers. Each have their own range of operation
2164 and their own constants and width. To keep track of the implemented
2165 registers, find a list below:
2167 ======= =============================== ============
2168 Arch Register Width (bits)
2169 ======= =============================== ============
2170 PPC KVM_REG_PPC_HIOR 64
2171 PPC KVM_REG_PPC_IAC1 64
2172 PPC KVM_REG_PPC_IAC2 64
2173 PPC KVM_REG_PPC_IAC3 64
2174 PPC KVM_REG_PPC_IAC4 64
2175 PPC KVM_REG_PPC_DAC1 64
2176 PPC KVM_REG_PPC_DAC2 64
2177 PPC KVM_REG_PPC_DABR 64
2178 PPC KVM_REG_PPC_DSCR 64
2179 PPC KVM_REG_PPC_PURR 64
2180 PPC KVM_REG_PPC_SPURR 64
2181 PPC KVM_REG_PPC_DAR 64
2182 PPC KVM_REG_PPC_DSISR 32
2183 PPC KVM_REG_PPC_AMR 64
2184 PPC KVM_REG_PPC_UAMOR 64
2185 PPC KVM_REG_PPC_MMCR0 64
2186 PPC KVM_REG_PPC_MMCR1 64
2187 PPC KVM_REG_PPC_MMCRA 64
2188 PPC KVM_REG_PPC_MMCR2 64
2189 PPC KVM_REG_PPC_MMCRS 64
2190 PPC KVM_REG_PPC_MMCR3 64
2191 PPC KVM_REG_PPC_SIAR 64
2192 PPC KVM_REG_PPC_SDAR 64
2193 PPC KVM_REG_PPC_SIER 64
2194 PPC KVM_REG_PPC_SIER2 64
2195 PPC KVM_REG_PPC_SIER3 64
2196 PPC KVM_REG_PPC_PMC1 32
2197 PPC KVM_REG_PPC_PMC2 32
2198 PPC KVM_REG_PPC_PMC3 32
2199 PPC KVM_REG_PPC_PMC4 32
2200 PPC KVM_REG_PPC_PMC5 32
2201 PPC KVM_REG_PPC_PMC6 32
2202 PPC KVM_REG_PPC_PMC7 32
2203 PPC KVM_REG_PPC_PMC8 32
2204 PPC KVM_REG_PPC_FPR0 64
2206 PPC KVM_REG_PPC_FPR31 64
2207 PPC KVM_REG_PPC_VR0 128
2209 PPC KVM_REG_PPC_VR31 128
2210 PPC KVM_REG_PPC_VSR0 128
2212 PPC KVM_REG_PPC_VSR31 128
2213 PPC KVM_REG_PPC_FPSCR 64
2214 PPC KVM_REG_PPC_VSCR 32
2215 PPC KVM_REG_PPC_VPA_ADDR 64
2216 PPC KVM_REG_PPC_VPA_SLB 128
2217 PPC KVM_REG_PPC_VPA_DTL 128
2218 PPC KVM_REG_PPC_EPCR 32
2219 PPC KVM_REG_PPC_EPR 32
2220 PPC KVM_REG_PPC_TCR 32
2221 PPC KVM_REG_PPC_TSR 32
2222 PPC KVM_REG_PPC_OR_TSR 32
2223 PPC KVM_REG_PPC_CLEAR_TSR 32
2224 PPC KVM_REG_PPC_MAS0 32
2225 PPC KVM_REG_PPC_MAS1 32
2226 PPC KVM_REG_PPC_MAS2 64
2227 PPC KVM_REG_PPC_MAS7_3 64
2228 PPC KVM_REG_PPC_MAS4 32
2229 PPC KVM_REG_PPC_MAS6 32
2230 PPC KVM_REG_PPC_MMUCFG 32
2231 PPC KVM_REG_PPC_TLB0CFG 32
2232 PPC KVM_REG_PPC_TLB1CFG 32
2233 PPC KVM_REG_PPC_TLB2CFG 32
2234 PPC KVM_REG_PPC_TLB3CFG 32
2235 PPC KVM_REG_PPC_TLB0PS 32
2236 PPC KVM_REG_PPC_TLB1PS 32
2237 PPC KVM_REG_PPC_TLB2PS 32
2238 PPC KVM_REG_PPC_TLB3PS 32
2239 PPC KVM_REG_PPC_EPTCFG 32
2240 PPC KVM_REG_PPC_ICP_STATE 64
2241 PPC KVM_REG_PPC_VP_STATE 128
2242 PPC KVM_REG_PPC_TB_OFFSET 64
2243 PPC KVM_REG_PPC_SPMC1 32
2244 PPC KVM_REG_PPC_SPMC2 32
2245 PPC KVM_REG_PPC_IAMR 64
2246 PPC KVM_REG_PPC_TFHAR 64
2247 PPC KVM_REG_PPC_TFIAR 64
2248 PPC KVM_REG_PPC_TEXASR 64
2249 PPC KVM_REG_PPC_FSCR 64
2250 PPC KVM_REG_PPC_PSPB 32
2251 PPC KVM_REG_PPC_EBBHR 64
2252 PPC KVM_REG_PPC_EBBRR 64
2253 PPC KVM_REG_PPC_BESCR 64
2254 PPC KVM_REG_PPC_TAR 64
2255 PPC KVM_REG_PPC_DPDES 64
2256 PPC KVM_REG_PPC_DAWR 64
2257 PPC KVM_REG_PPC_DAWRX 64
2258 PPC KVM_REG_PPC_CIABR 64
2259 PPC KVM_REG_PPC_IC 64
2260 PPC KVM_REG_PPC_VTB 64
2261 PPC KVM_REG_PPC_CSIGR 64
2262 PPC KVM_REG_PPC_TACR 64
2263 PPC KVM_REG_PPC_TCSCR 64
2264 PPC KVM_REG_PPC_PID 64
2265 PPC KVM_REG_PPC_ACOP 64
2266 PPC KVM_REG_PPC_VRSAVE 32
2267 PPC KVM_REG_PPC_LPCR 32
2268 PPC KVM_REG_PPC_LPCR_64 64
2269 PPC KVM_REG_PPC_PPR 64
2270 PPC KVM_REG_PPC_ARCH_COMPAT 32
2271 PPC KVM_REG_PPC_DABRX 32
2272 PPC KVM_REG_PPC_WORT 64
2273 PPC KVM_REG_PPC_SPRG9 64
2274 PPC KVM_REG_PPC_DBSR 32
2275 PPC KVM_REG_PPC_TIDR 64
2276 PPC KVM_REG_PPC_PSSCR 64
2277 PPC KVM_REG_PPC_DEC_EXPIRY 64
2278 PPC KVM_REG_PPC_PTCR 64
2279 PPC KVM_REG_PPC_DAWR1 64
2280 PPC KVM_REG_PPC_DAWRX1 64
2281 PPC KVM_REG_PPC_TM_GPR0 64
2283 PPC KVM_REG_PPC_TM_GPR31 64
2284 PPC KVM_REG_PPC_TM_VSR0 128
2286 PPC KVM_REG_PPC_TM_VSR63 128
2287 PPC KVM_REG_PPC_TM_CR 64
2288 PPC KVM_REG_PPC_TM_LR 64
2289 PPC KVM_REG_PPC_TM_CTR 64
2290 PPC KVM_REG_PPC_TM_FPSCR 64
2291 PPC KVM_REG_PPC_TM_AMR 64
2292 PPC KVM_REG_PPC_TM_PPR 64
2293 PPC KVM_REG_PPC_TM_VRSAVE 64
2294 PPC KVM_REG_PPC_TM_VSCR 32
2295 PPC KVM_REG_PPC_TM_DSCR 64
2296 PPC KVM_REG_PPC_TM_TAR 64
2297 PPC KVM_REG_PPC_TM_XER 64
2299 MIPS KVM_REG_MIPS_R0 64
2301 MIPS KVM_REG_MIPS_R31 64
2302 MIPS KVM_REG_MIPS_HI 64
2303 MIPS KVM_REG_MIPS_LO 64
2304 MIPS KVM_REG_MIPS_PC 64
2305 MIPS KVM_REG_MIPS_CP0_INDEX 32
2306 MIPS KVM_REG_MIPS_CP0_ENTRYLO0 64
2307 MIPS KVM_REG_MIPS_CP0_ENTRYLO1 64
2308 MIPS KVM_REG_MIPS_CP0_CONTEXT 64
2309 MIPS KVM_REG_MIPS_CP0_CONTEXTCONFIG 32
2310 MIPS KVM_REG_MIPS_CP0_USERLOCAL 64
2311 MIPS KVM_REG_MIPS_CP0_XCONTEXTCONFIG 64
2312 MIPS KVM_REG_MIPS_CP0_PAGEMASK 32
2313 MIPS KVM_REG_MIPS_CP0_PAGEGRAIN 32
2314 MIPS KVM_REG_MIPS_CP0_SEGCTL0 64
2315 MIPS KVM_REG_MIPS_CP0_SEGCTL1 64
2316 MIPS KVM_REG_MIPS_CP0_SEGCTL2 64
2317 MIPS KVM_REG_MIPS_CP0_PWBASE 64
2318 MIPS KVM_REG_MIPS_CP0_PWFIELD 64
2319 MIPS KVM_REG_MIPS_CP0_PWSIZE 64
2320 MIPS KVM_REG_MIPS_CP0_WIRED 32
2321 MIPS KVM_REG_MIPS_CP0_PWCTL 32
2322 MIPS KVM_REG_MIPS_CP0_HWRENA 32
2323 MIPS KVM_REG_MIPS_CP0_BADVADDR 64
2324 MIPS KVM_REG_MIPS_CP0_BADINSTR 32
2325 MIPS KVM_REG_MIPS_CP0_BADINSTRP 32
2326 MIPS KVM_REG_MIPS_CP0_COUNT 32
2327 MIPS KVM_REG_MIPS_CP0_ENTRYHI 64
2328 MIPS KVM_REG_MIPS_CP0_COMPARE 32
2329 MIPS KVM_REG_MIPS_CP0_STATUS 32
2330 MIPS KVM_REG_MIPS_CP0_INTCTL 32
2331 MIPS KVM_REG_MIPS_CP0_CAUSE 32
2332 MIPS KVM_REG_MIPS_CP0_EPC 64
2333 MIPS KVM_REG_MIPS_CP0_PRID 32
2334 MIPS KVM_REG_MIPS_CP0_EBASE 64
2335 MIPS KVM_REG_MIPS_CP0_CONFIG 32
2336 MIPS KVM_REG_MIPS_CP0_CONFIG1 32
2337 MIPS KVM_REG_MIPS_CP0_CONFIG2 32
2338 MIPS KVM_REG_MIPS_CP0_CONFIG3 32
2339 MIPS KVM_REG_MIPS_CP0_CONFIG4 32
2340 MIPS KVM_REG_MIPS_CP0_CONFIG5 32
2341 MIPS KVM_REG_MIPS_CP0_CONFIG7 32
2342 MIPS KVM_REG_MIPS_CP0_XCONTEXT 64
2343 MIPS KVM_REG_MIPS_CP0_ERROREPC 64
2344 MIPS KVM_REG_MIPS_CP0_KSCRATCH1 64
2345 MIPS KVM_REG_MIPS_CP0_KSCRATCH2 64
2346 MIPS KVM_REG_MIPS_CP0_KSCRATCH3 64
2347 MIPS KVM_REG_MIPS_CP0_KSCRATCH4 64
2348 MIPS KVM_REG_MIPS_CP0_KSCRATCH5 64
2349 MIPS KVM_REG_MIPS_CP0_KSCRATCH6 64
2350 MIPS KVM_REG_MIPS_CP0_MAAR(0..63) 64
2351 MIPS KVM_REG_MIPS_COUNT_CTL 64
2352 MIPS KVM_REG_MIPS_COUNT_RESUME 64
2353 MIPS KVM_REG_MIPS_COUNT_HZ 64
2354 MIPS KVM_REG_MIPS_FPR_32(0..31) 32
2355 MIPS KVM_REG_MIPS_FPR_64(0..31) 64
2356 MIPS KVM_REG_MIPS_VEC_128(0..31) 128
2357 MIPS KVM_REG_MIPS_FCR_IR 32
2358 MIPS KVM_REG_MIPS_FCR_CSR 32
2359 MIPS KVM_REG_MIPS_MSA_IR 32
2360 MIPS KVM_REG_MIPS_MSA_CSR 32
2361 ======= =============================== ============
2363 ARM registers are mapped using the lower 32 bits. The upper 16 of that
2364 is the register group type, or coprocessor number:
2366 ARM core registers have the following id bit patterns::
2368 0x4020 0000 0010 <index into the kvm_regs struct:16>
2370 ARM 32-bit CP15 registers have the following id bit patterns::
2372 0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3>
2374 ARM 64-bit CP15 registers have the following id bit patterns::
2376 0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3>
2378 ARM CCSIDR registers are demultiplexed by CSSELR value::
2380 0x4020 0000 0011 00 <csselr:8>
2382 ARM 32-bit VFP control registers have the following id bit patterns::
2384 0x4020 0000 0012 1 <regno:12>
2386 ARM 64-bit FP registers have the following id bit patterns::
2388 0x4030 0000 0012 0 <regno:12>
2390 ARM firmware pseudo-registers have the following bit pattern::
2392 0x4030 0000 0014 <regno:16>
2395 arm64 registers are mapped using the lower 32 bits. The upper 16 of
2396 that is the register group type, or coprocessor number:
2398 arm64 core/FP-SIMD registers have the following id bit patterns. Note
2399 that the size of the access is variable, as the kvm_regs structure
2400 contains elements ranging from 32 to 128 bits. The index is a 32bit
2401 value in the kvm_regs structure seen as a 32bit array::
2403 0x60x0 0000 0010 <index into the kvm_regs struct:16>
2407 ======================= ========= ===== =======================================
2408 Encoding Register Bits kvm_regs member
2409 ======================= ========= ===== =======================================
2410 0x6030 0000 0010 0000 X0 64 regs.regs[0]
2411 0x6030 0000 0010 0002 X1 64 regs.regs[1]
2413 0x6030 0000 0010 003c X30 64 regs.regs[30]
2414 0x6030 0000 0010 003e SP 64 regs.sp
2415 0x6030 0000 0010 0040 PC 64 regs.pc
2416 0x6030 0000 0010 0042 PSTATE 64 regs.pstate
2417 0x6030 0000 0010 0044 SP_EL1 64 sp_el1
2418 0x6030 0000 0010 0046 ELR_EL1 64 elr_el1
2419 0x6030 0000 0010 0048 SPSR_EL1 64 spsr[KVM_SPSR_EL1] (alias SPSR_SVC)
2420 0x6030 0000 0010 004a SPSR_ABT 64 spsr[KVM_SPSR_ABT]
2421 0x6030 0000 0010 004c SPSR_UND 64 spsr[KVM_SPSR_UND]
2422 0x6030 0000 0010 004e SPSR_IRQ 64 spsr[KVM_SPSR_IRQ]
2423 0x6060 0000 0010 0050 SPSR_FIQ 64 spsr[KVM_SPSR_FIQ]
2424 0x6040 0000 0010 0054 V0 128 fp_regs.vregs[0] [1]_
2425 0x6040 0000 0010 0058 V1 128 fp_regs.vregs[1] [1]_
2427 0x6040 0000 0010 00d0 V31 128 fp_regs.vregs[31] [1]_
2428 0x6020 0000 0010 00d4 FPSR 32 fp_regs.fpsr
2429 0x6020 0000 0010 00d5 FPCR 32 fp_regs.fpcr
2430 ======================= ========= ===== =======================================
2432 .. [1] These encodings are not accepted for SVE-enabled vcpus. See
2435 The equivalent register content can be accessed via bits [127:0] of
2436 the corresponding SVE Zn registers instead for vcpus that have SVE
2437 enabled (see below).
2439 arm64 CCSIDR registers are demultiplexed by CSSELR value::
2441 0x6020 0000 0011 00 <csselr:8>
2443 arm64 system registers have the following id bit patterns::
2445 0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3>
2449 Two system register IDs do not follow the specified pattern. These
2450 are KVM_REG_ARM_TIMER_CVAL and KVM_REG_ARM_TIMER_CNT, which map to
2451 system registers CNTV_CVAL_EL0 and CNTVCT_EL0 respectively. These
2452 two had their values accidentally swapped, which means TIMER_CVAL is
2453 derived from the register encoding for CNTVCT_EL0 and TIMER_CNT is
2454 derived from the register encoding for CNTV_CVAL_EL0. As this is
2455 API, it must remain this way.
2457 arm64 firmware pseudo-registers have the following bit pattern::
2459 0x6030 0000 0014 <regno:16>
2461 arm64 SVE registers have the following bit patterns::
2463 0x6080 0000 0015 00 <n:5> <slice:5> Zn bits[2048*slice + 2047 : 2048*slice]
2464 0x6050 0000 0015 04 <n:4> <slice:5> Pn bits[256*slice + 255 : 256*slice]
2465 0x6050 0000 0015 060 <slice:5> FFR bits[256*slice + 255 : 256*slice]
2466 0x6060 0000 0015 ffff KVM_REG_ARM64_SVE_VLS pseudo-register
2468 Access to register IDs where 2048 * slice >= 128 * max_vq will fail with
2469 ENOENT. max_vq is the vcpu's maximum supported vector length in 128-bit
2470 quadwords: see [2]_ below.
2472 These registers are only accessible on vcpus for which SVE is enabled.
2473 See KVM_ARM_VCPU_INIT for details.
2475 In addition, except for KVM_REG_ARM64_SVE_VLS, these registers are not
2476 accessible until the vcpu's SVE configuration has been finalized
2477 using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). See KVM_ARM_VCPU_INIT
2478 and KVM_ARM_VCPU_FINALIZE for more information about this procedure.
2480 KVM_REG_ARM64_SVE_VLS is a pseudo-register that allows the set of vector
2481 lengths supported by the vcpu to be discovered and configured by
2482 userspace. When transferred to or from user memory via KVM_GET_ONE_REG
2483 or KVM_SET_ONE_REG, the value of this register is of type
2484 __u64[KVM_ARM64_SVE_VLS_WORDS], and encodes the set of vector lengths as
2487 __u64 vector_lengths[KVM_ARM64_SVE_VLS_WORDS];
2489 if (vq >= SVE_VQ_MIN && vq <= SVE_VQ_MAX &&
2490 ((vector_lengths[(vq - KVM_ARM64_SVE_VQ_MIN) / 64] >>
2491 ((vq - KVM_ARM64_SVE_VQ_MIN) % 64)) & 1))
2492 /* Vector length vq * 16 bytes supported */
2494 /* Vector length vq * 16 bytes not supported */
2496 .. [2] The maximum value vq for which the above condition is true is
2497 max_vq. This is the maximum vector length available to the guest on
2498 this vcpu, and determines which register slices are visible through
2499 this ioctl interface.
2501 (See Documentation/arm64/sve.rst for an explanation of the "vq"
2504 KVM_REG_ARM64_SVE_VLS is only accessible after KVM_ARM_VCPU_INIT.
2505 KVM_ARM_VCPU_INIT initialises it to the best set of vector lengths that
2508 Userspace may subsequently modify it if desired until the vcpu's SVE
2509 configuration is finalized using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE).
2511 Apart from simply removing all vector lengths from the host set that
2512 exceed some value, support for arbitrarily chosen sets of vector lengths
2513 is hardware-dependent and may not be available. Attempting to configure
2514 an invalid set of vector lengths via KVM_SET_ONE_REG will fail with
2517 After the vcpu's SVE configuration is finalized, further attempts to
2518 write this register will fail with EPERM.
2521 MIPS registers are mapped using the lower 32 bits. The upper 16 of that is
2522 the register group type:
2524 MIPS core registers (see above) have the following id bit patterns::
2526 0x7030 0000 0000 <reg:16>
2528 MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit
2529 patterns depending on whether they're 32-bit or 64-bit registers::
2531 0x7020 0000 0001 00 <reg:5> <sel:3> (32-bit)
2532 0x7030 0000 0001 00 <reg:5> <sel:3> (64-bit)
2534 Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_MIPS_CP0_ENTRYLO1 are the MIPS64
2535 versions of the EntryLo registers regardless of the word size of the host
2536 hardware, host kernel, guest, and whether XPA is present in the guest, i.e.
2537 with the RI and XI bits (if they exist) in bits 63 and 62 respectively, and
2538 the PFNX field starting at bit 30.
2540 MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) above) have the following id bit
2543 0x7030 0000 0001 01 <reg:8>
2545 MIPS KVM control registers (see above) have the following id bit patterns::
2547 0x7030 0000 0002 <reg:16>
2549 MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following
2550 id bit patterns depending on the size of the register being accessed. They are
2551 always accessed according to the current guest FPU mode (Status.FR and
2552 Config5.FRE), i.e. as the guest would see them, and they become unpredictable
2553 if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector
2554 registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they
2555 overlap the FPU registers::
2557 0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers)
2558 0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers)
2559 0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers)
2561 MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the
2562 following id bit patterns::
2564 0x7020 0000 0003 01 <0:3> <reg:5>
2566 MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the
2567 following id bit patterns::
2569 0x7020 0000 0003 02 <0:3> <reg:5>
2572 4.69 KVM_GET_ONE_REG
2573 --------------------
2575 :Capability: KVM_CAP_ONE_REG
2578 :Parameters: struct kvm_one_reg (in and out)
2579 :Returns: 0 on success, negative value on failure
2583 ======== ============================================================
2584 Â ENOENT Â Â no such register
2585 Â EINVAL Â Â invalid register ID, or no such register or used with VMs in
2586 protected virtualization mode on s390
2587 Â EPERM Â Â Â (arm64) register access not allowed before vcpu finalization
2588 ======== ============================================================
2590 (These error codes are indicative only: do not rely on a specific error
2591 code being returned in a specific situation.)
2593 This ioctl allows to receive the value of a single register implemented
2594 in a vcpu. The register to read is indicated by the "id" field of the
2595 kvm_one_reg struct passed in. On success, the register value can be found
2596 at the memory location pointed to by "addr".
2598 The list of registers accessible using this interface is identical to the
2602 4.70 KVM_KVMCLOCK_CTRL
2603 ----------------------
2605 :Capability: KVM_CAP_KVMCLOCK_CTRL
2606 :Architectures: Any that implement pvclocks (currently x86 only)
2609 :Returns: 0 on success, -1 on error
2611 This ioctl sets a flag accessible to the guest indicating that the specified
2612 vCPU has been paused by the host userspace.
2614 The host will set a flag in the pvclock structure that is checked from the
2615 soft lockup watchdog. The flag is part of the pvclock structure that is
2616 shared between guest and host, specifically the second bit of the flags
2617 field of the pvclock_vcpu_time_info structure. It will be set exclusively by
2618 the host and read/cleared exclusively by the guest. The guest operation of
2619 checking and clearing the flag must be an atomic operation so
2620 load-link/store-conditional, or equivalent must be used. There are two cases
2621 where the guest will clear the flag: when the soft lockup watchdog timer resets
2622 itself or when a soft lockup is detected. This ioctl can be called any time
2623 after pausing the vcpu, but before it is resumed.
2629 :Capability: KVM_CAP_SIGNAL_MSI
2630 :Architectures: x86 arm arm64
2632 :Parameters: struct kvm_msi (in)
2633 :Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error
2635 Directly inject a MSI message. Only valid with in-kernel irqchip that handles
2650 KVM_MSI_VALID_DEVID: devid contains a valid value. The per-VM
2651 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
2652 the device ID. If this capability is not available, userspace
2653 should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
2655 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
2656 for the device that wrote the MSI message. For PCI, this is usually a
2657 BFD identifier in the lower 16 bits.
2659 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
2660 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
2661 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
2662 address_hi must be zero.
2665 4.71 KVM_CREATE_PIT2
2666 --------------------
2668 :Capability: KVM_CAP_PIT2
2671 :Parameters: struct kvm_pit_config (in)
2672 :Returns: 0 on success, -1 on error
2674 Creates an in-kernel device model for the i8254 PIT. This call is only valid
2675 after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following
2676 parameters have to be passed::
2678 struct kvm_pit_config {
2685 #define KVM_PIT_SPEAKER_DUMMY 1 /* emulate speaker port stub */
2687 PIT timer interrupts may use a per-VM kernel thread for injection. If it
2688 exists, this thread will have a name of the following pattern::
2690 kvm-pit/<owner-process-pid>
2692 When running a guest with elevated priorities, the scheduling parameters of
2693 this thread may have to be adjusted accordingly.
2695 This IOCTL replaces the obsolete KVM_CREATE_PIT.
2701 :Capability: KVM_CAP_PIT_STATE2
2704 :Parameters: struct kvm_pit_state2 (out)
2705 :Returns: 0 on success, -1 on error
2707 Retrieves the state of the in-kernel PIT model. Only valid after
2708 KVM_CREATE_PIT2. The state is returned in the following structure::
2710 struct kvm_pit_state2 {
2711 struct kvm_pit_channel_state channels[3];
2718 /* disable PIT in HPET legacy mode */
2719 #define KVM_PIT_FLAGS_HPET_LEGACY 0x00000001
2721 This IOCTL replaces the obsolete KVM_GET_PIT.
2727 :Capability: KVM_CAP_PIT_STATE2
2730 :Parameters: struct kvm_pit_state2 (in)
2731 :Returns: 0 on success, -1 on error
2733 Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2.
2734 See KVM_GET_PIT2 for details on struct kvm_pit_state2.
2736 This IOCTL replaces the obsolete KVM_SET_PIT.
2739 4.74 KVM_PPC_GET_SMMU_INFO
2740 --------------------------
2742 :Capability: KVM_CAP_PPC_GET_SMMU_INFO
2743 :Architectures: powerpc
2746 :Returns: 0 on success, -1 on error
2748 This populates and returns a structure describing the features of
2749 the "Server" class MMU emulation supported by KVM.
2750 This can in turn be used by userspace to generate the appropriate
2751 device-tree properties for the guest operating system.
2753 The structure contains some global information, followed by an
2754 array of supported segment page sizes::
2756 struct kvm_ppc_smmu_info {
2760 struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ];
2763 The supported flags are:
2765 - KVM_PPC_PAGE_SIZES_REAL:
2766 When that flag is set, guest page sizes must "fit" the backing
2767 store page sizes. When not set, any page size in the list can
2768 be used regardless of how they are backed by userspace.
2770 - KVM_PPC_1T_SEGMENTS
2771 The emulated MMU supports 1T segments in addition to the
2775 This flag indicates that HPT guests are not supported by KVM,
2776 thus all guests must use radix MMU mode.
2778 The "slb_size" field indicates how many SLB entries are supported
2780 The "sps" array contains 8 entries indicating the supported base
2781 page sizes for a segment in increasing order. Each entry is defined
2784 struct kvm_ppc_one_seg_page_size {
2785 __u32 page_shift; /* Base page shift of segment (or 0) */
2786 __u32 slb_enc; /* SLB encoding for BookS */
2787 struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ];
2790 An entry with a "page_shift" of 0 is unused. Because the array is
2791 organized in increasing order, a lookup can stop when encoutering
2794 The "slb_enc" field provides the encoding to use in the SLB for the
2795 page size. The bits are in positions such as the value can directly
2796 be OR'ed into the "vsid" argument of the slbmte instruction.
2798 The "enc" array is a list which for each of those segment base page
2799 size provides the list of supported actual page sizes (which can be
2800 only larger or equal to the base page size), along with the
2801 corresponding encoding in the hash PTE. Similarly, the array is
2802 8 entries sorted by increasing sizes and an entry with a "0" shift
2803 is an empty entry and a terminator::
2805 struct kvm_ppc_one_page_size {
2806 __u32 page_shift; /* Page shift (or 0) */
2807 __u32 pte_enc; /* Encoding in the HPTE (>>12) */
2810 The "pte_enc" field provides a value that can OR'ed into the hash
2811 PTE's RPN field (ie, it needs to be shifted left by 12 to OR it
2812 into the hash PTE second double word).
2817 :Capability: KVM_CAP_IRQFD
2818 :Architectures: x86 s390 arm arm64
2820 :Parameters: struct kvm_irqfd (in)
2821 :Returns: 0 on success, -1 on error
2823 Allows setting an eventfd to directly trigger a guest interrupt.
2824 kvm_irqfd.fd specifies the file descriptor to use as the eventfd and
2825 kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When
2826 an event is triggered on the eventfd, an interrupt is injected into
2827 the guest using the specified gsi pin. The irqfd is removed using
2828 the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd
2831 With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify
2832 mechanism allowing emulation of level-triggered, irqfd-based
2833 interrupts. When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an
2834 additional eventfd in the kvm_irqfd.resamplefd field. When operating
2835 in resample mode, posting of an interrupt through kvm_irq.fd asserts
2836 the specified gsi in the irqchip. When the irqchip is resampled, such
2837 as from an EOI, the gsi is de-asserted and the user is notified via
2838 kvm_irqfd.resamplefd. It is the user's responsibility to re-queue
2839 the interrupt if the device making use of it still requires service.
2840 Note that closing the resamplefd is not sufficient to disable the
2841 irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment
2842 and need not be specified with KVM_IRQFD_FLAG_DEASSIGN.
2844 On arm/arm64, gsi routing being supported, the following can happen:
2846 - in case no routing entry is associated to this gsi, injection fails
2847 - in case the gsi is associated to an irqchip routing entry,
2848 irqchip.pin + 32 corresponds to the injected SPI ID.
2849 - in case the gsi is associated to an MSI routing entry, the MSI
2850 message and device ID are translated into an LPI (support restricted
2851 to GICv3 ITS in-kernel emulation).
2853 4.76 KVM_PPC_ALLOCATE_HTAB
2854 --------------------------
2856 :Capability: KVM_CAP_PPC_ALLOC_HTAB
2857 :Architectures: powerpc
2859 :Parameters: Pointer to u32 containing hash table order (in/out)
2860 :Returns: 0 on success, -1 on error
2862 This requests the host kernel to allocate an MMU hash table for a
2863 guest using the PAPR paravirtualization interface. This only does
2864 anything if the kernel is configured to use the Book 3S HV style of
2865 virtualization. Otherwise the capability doesn't exist and the ioctl
2866 returns an ENOTTY error. The rest of this description assumes Book 3S
2869 There must be no vcpus running when this ioctl is called; if there
2870 are, it will do nothing and return an EBUSY error.
2872 The parameter is a pointer to a 32-bit unsigned integer variable
2873 containing the order (log base 2) of the desired size of the hash
2874 table, which must be between 18 and 46. On successful return from the
2875 ioctl, the value will not be changed by the kernel.
2877 If no hash table has been allocated when any vcpu is asked to run
2878 (with the KVM_RUN ioctl), the host kernel will allocate a
2879 default-sized hash table (16 MB).
2881 If this ioctl is called when a hash table has already been allocated,
2882 with a different order from the existing hash table, the existing hash
2883 table will be freed and a new one allocated. If this is ioctl is
2884 called when a hash table has already been allocated of the same order
2885 as specified, the kernel will clear out the existing hash table (zero
2886 all HPTEs). In either case, if the guest is using the virtualized
2887 real-mode area (VRMA) facility, the kernel will re-create the VMRA
2888 HPTEs on the next KVM_RUN of any vcpu.
2890 4.77 KVM_S390_INTERRUPT
2891 -----------------------
2894 :Architectures: s390
2895 :Type: vm ioctl, vcpu ioctl
2896 :Parameters: struct kvm_s390_interrupt (in)
2897 :Returns: 0 on success, -1 on error
2899 Allows to inject an interrupt to the guest. Interrupts can be floating
2900 (vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type.
2902 Interrupt parameters are passed via kvm_s390_interrupt::
2904 struct kvm_s390_interrupt {
2910 type can be one of the following:
2912 KVM_S390_SIGP_STOP (vcpu)
2913 - sigp stop; optional flags in parm
2914 KVM_S390_PROGRAM_INT (vcpu)
2915 - program check; code in parm
2916 KVM_S390_SIGP_SET_PREFIX (vcpu)
2917 - sigp set prefix; prefix address in parm
2918 KVM_S390_RESTART (vcpu)
2920 KVM_S390_INT_CLOCK_COMP (vcpu)
2921 - clock comparator interrupt
2922 KVM_S390_INT_CPU_TIMER (vcpu)
2923 - CPU timer interrupt
2924 KVM_S390_INT_VIRTIO (vm)
2925 - virtio external interrupt; external interrupt
2926 parameters in parm and parm64
2927 KVM_S390_INT_SERVICE (vm)
2928 - sclp external interrupt; sclp parameter in parm
2929 KVM_S390_INT_EMERGENCY (vcpu)
2930 - sigp emergency; source cpu in parm
2931 KVM_S390_INT_EXTERNAL_CALL (vcpu)
2932 - sigp external call; source cpu in parm
2933 KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm)
2934 - compound value to indicate an
2935 I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel);
2936 I/O interruption parameters in parm (subchannel) and parm64 (intparm,
2937 interruption subclass)
2938 KVM_S390_MCHK (vm, vcpu)
2939 - machine check interrupt; cr 14 bits in parm, machine check interrupt
2940 code in parm64 (note that machine checks needing further payload are not
2941 supported by this ioctl)
2943 This is an asynchronous vcpu ioctl and can be invoked from any thread.
2945 4.78 KVM_PPC_GET_HTAB_FD
2946 ------------------------
2948 :Capability: KVM_CAP_PPC_HTAB_FD
2949 :Architectures: powerpc
2951 :Parameters: Pointer to struct kvm_get_htab_fd (in)
2952 :Returns: file descriptor number (>= 0) on success, -1 on error
2954 This returns a file descriptor that can be used either to read out the
2955 entries in the guest's hashed page table (HPT), or to write entries to
2956 initialize the HPT. The returned fd can only be written to if the
2957 KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and
2958 can only be read if that bit is clear. The argument struct looks like
2961 /* For KVM_PPC_GET_HTAB_FD */
2962 struct kvm_get_htab_fd {
2968 /* Values for kvm_get_htab_fd.flags */
2969 #define KVM_GET_HTAB_BOLTED_ONLY ((__u64)0x1)
2970 #define KVM_GET_HTAB_WRITE ((__u64)0x2)
2972 The 'start_index' field gives the index in the HPT of the entry at
2973 which to start reading. It is ignored when writing.
2975 Reads on the fd will initially supply information about all
2976 "interesting" HPT entries. Interesting entries are those with the
2977 bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise
2978 all entries. When the end of the HPT is reached, the read() will
2979 return. If read() is called again on the fd, it will start again from
2980 the beginning of the HPT, but will only return HPT entries that have
2981 changed since they were last read.
2983 Data read or written is structured as a header (8 bytes) followed by a
2984 series of valid HPT entries (16 bytes) each. The header indicates how
2985 many valid HPT entries there are and how many invalid entries follow
2986 the valid entries. The invalid entries are not represented explicitly
2987 in the stream. The header format is::
2989 struct kvm_get_htab_header {
2995 Writes to the fd create HPT entries starting at the index given in the
2996 header; first 'n_valid' valid entries with contents from the data
2997 written, then 'n_invalid' invalid entries, invalidating any previously
2998 valid entries found.
3000 4.79 KVM_CREATE_DEVICE
3001 ----------------------
3003 :Capability: KVM_CAP_DEVICE_CTRL
3005 :Parameters: struct kvm_create_device (in/out)
3006 :Returns: 0 on success, -1 on error
3010 ====== =======================================================
3011 ENODEV The device type is unknown or unsupported
3012 EEXIST Device already created, and this type of device may not
3013 be instantiated multiple times
3014 ====== =======================================================
3016 Other error conditions may be defined by individual device types or
3017 have their standard meanings.
3019 Creates an emulated device in the kernel. The file descriptor returned
3020 in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR.
3022 If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the
3023 device type is supported (not necessarily whether it can be created
3026 Individual devices should not define flags. Attributes should be used
3027 for specifying any behavior that is not implied by the device type
3032 struct kvm_create_device {
3033 __u32 type; /* in: KVM_DEV_TYPE_xxx */
3034 __u32 fd; /* out: device handle */
3035 __u32 flags; /* in: KVM_CREATE_DEVICE_xxx */
3038 4.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR
3039 --------------------------------------------
3041 :Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
3042 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
3043 :Type: device ioctl, vm ioctl, vcpu ioctl
3044 :Parameters: struct kvm_device_attr
3045 :Returns: 0 on success, -1 on error
3049 ===== =============================================================
3050 ENXIO The group or attribute is unknown/unsupported for this device
3051 or hardware support is missing.
3052 EPERM The attribute cannot (currently) be accessed this way
3053 (e.g. read-only attribute, or attribute that only makes
3054 sense when the device is in a different state)
3055 ===== =============================================================
3057 Other error conditions may be defined by individual device types.
3059 Gets/sets a specified piece of device configuration and/or state. The
3060 semantics are device-specific. See individual device documentation in
3061 the "devices" directory. As with ONE_REG, the size of the data
3062 transferred is defined by the particular attribute.
3066 struct kvm_device_attr {
3067 __u32 flags; /* no flags currently defined */
3068 __u32 group; /* device-defined */
3069 __u64 attr; /* group-defined */
3070 __u64 addr; /* userspace address of attr data */
3073 4.81 KVM_HAS_DEVICE_ATTR
3074 ------------------------
3076 :Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
3077 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
3078 :Type: device ioctl, vm ioctl, vcpu ioctl
3079 :Parameters: struct kvm_device_attr
3080 :Returns: 0 on success, -1 on error
3084 ===== =============================================================
3085 ENXIO The group or attribute is unknown/unsupported for this device
3086 or hardware support is missing.
3087 ===== =============================================================
3089 Tests whether a device supports a particular attribute. A successful
3090 return indicates the attribute is implemented. It does not necessarily
3091 indicate that the attribute can be read or written in the device's
3092 current state. "addr" is ignored.
3094 4.82 KVM_ARM_VCPU_INIT
3095 ----------------------
3098 :Architectures: arm, arm64
3100 :Parameters: struct kvm_vcpu_init (in)
3101 :Returns: 0 on success; -1 on error
3105 ====== =================================================================
3106 Â EINVAL Â Â Â the target is unknown, or the combination of features is invalid.
3107 Â ENOENT Â Â Â a features bit specified is unknown.
3108 ====== =================================================================
3110 This tells KVM what type of CPU to present to the guest, and what
3111 optional features it should have. Â This will cause a reset of the cpu
3112 registers to their initial values. Â If this is not called, KVM_RUN will
3113 return ENOEXEC for that vcpu.
3115 Note that because some registers reflect machine topology, all vcpus
3116 should be created before this ioctl is invoked.
3118 Userspace can call this function multiple times for a given vcpu, including
3119 after the vcpu has been run. This will reset the vcpu to its initial
3120 state. All calls to this function after the initial call must use the same
3121 target and same set of feature flags, otherwise EINVAL will be returned.
3125 - KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state.
3126 Depends on KVM_CAP_ARM_PSCI. If not set, the CPU will be powered on
3127 and execute guest code when KVM_RUN is called.
3128 - KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode.
3129 Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only).
3130 - KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 (or a future revision
3131 backward compatible with v0.2) for the CPU.
3132 Depends on KVM_CAP_ARM_PSCI_0_2.
3133 - KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU.
3134 Depends on KVM_CAP_ARM_PMU_V3.
3136 - KVM_ARM_VCPU_PTRAUTH_ADDRESS: Enables Address Pointer authentication
3138 Depends on KVM_CAP_ARM_PTRAUTH_ADDRESS.
3139 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are
3140 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and
3141 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be
3144 - KVM_ARM_VCPU_PTRAUTH_GENERIC: Enables Generic Pointer authentication
3146 Depends on KVM_CAP_ARM_PTRAUTH_GENERIC.
3147 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are
3148 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and
3149 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be
3152 - KVM_ARM_VCPU_SVE: Enables SVE for the CPU (arm64 only).
3153 Depends on KVM_CAP_ARM_SVE.
3154 Requires KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3156 * After KVM_ARM_VCPU_INIT:
3158 - KVM_REG_ARM64_SVE_VLS may be read using KVM_GET_ONE_REG: the
3159 initial value of this pseudo-register indicates the best set of
3160 vector lengths possible for a vcpu on this host.
3162 * Before KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3164 - KVM_RUN and KVM_GET_REG_LIST are not available;
3166 - KVM_GET_ONE_REG and KVM_SET_ONE_REG cannot be used to access
3167 the scalable archietctural SVE registers
3168 KVM_REG_ARM64_SVE_ZREG(), KVM_REG_ARM64_SVE_PREG() or
3169 KVM_REG_ARM64_SVE_FFR;
3171 - KVM_REG_ARM64_SVE_VLS may optionally be written using
3172 KVM_SET_ONE_REG, to modify the set of vector lengths available
3175 * After KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3177 - the KVM_REG_ARM64_SVE_VLS pseudo-register is immutable, and can
3178 no longer be written using KVM_SET_ONE_REG.
3180 4.83 KVM_ARM_PREFERRED_TARGET
3181 -----------------------------
3184 :Architectures: arm, arm64
3186 :Parameters: struct kvm_vcpu_init (out)
3187 :Returns: 0 on success; -1 on error
3191 ====== ==========================================
3192 ENODEV no preferred target available for the host
3193 ====== ==========================================
3195 This queries KVM for preferred CPU target type which can be emulated
3196 by KVM on underlying host.
3198 The ioctl returns struct kvm_vcpu_init instance containing information
3199 about preferred CPU target type and recommended features for it. The
3200 kvm_vcpu_init->features bitmap returned will have feature bits set if
3201 the preferred target recommends setting these features, but this is
3204 The information returned by this ioctl can be used to prepare an instance
3205 of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in
3206 VCPU matching underlying host.
3209 4.84 KVM_GET_REG_LIST
3210 ---------------------
3213 :Architectures: arm, arm64, mips
3215 :Parameters: struct kvm_reg_list (in/out)
3216 :Returns: 0 on success; -1 on error
3220 ===== ==============================================================
3221 Â E2BIG Â Â Â Â the reg index list is too big to fit in the array specified by
3222 Â Â Â Â Â Â Â Â Â Â Â Â the user (the number required will be written into n).
3223 ===== ==============================================================
3227 struct kvm_reg_list {
3228 __u64 n; /* number of registers in reg[] */
3232 This ioctl returns the guest registers that are supported for the
3233 KVM_GET_ONE_REG/KVM_SET_ONE_REG calls.
3236 4.85 KVM_ARM_SET_DEVICE_ADDR (deprecated)
3237 -----------------------------------------
3239 :Capability: KVM_CAP_ARM_SET_DEVICE_ADDR
3240 :Architectures: arm, arm64
3242 :Parameters: struct kvm_arm_device_address (in)
3243 :Returns: 0 on success, -1 on error
3247 ====== ============================================
3248 ENODEV The device id is unknown
3249 ENXIO Device not supported on current system
3250 EEXIST Address already set
3251 E2BIG Address outside guest physical address space
3252 EBUSY Address overlaps with other device range
3253 ====== ============================================
3257 struct kvm_arm_device_addr {
3262 Specify a device address in the guest's physical address space where guests
3263 can access emulated or directly exposed devices, which the host kernel needs
3264 to know about. The id field is an architecture specific identifier for a
3267 ARM/arm64 divides the id field into two parts, a device id and an
3268 address type id specific to the individual device::
3270 Â bits: | 63 ... 32 | 31 ... 16 | 15 ... 0 |
3271 field: | 0x00000000 | device id | addr type id |
3273 ARM/arm64 currently only require this when using the in-kernel GIC
3274 support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2
3275 as the device id. When setting the base address for the guest's
3276 mapping of the VGIC virtual CPU and distributor interface, the ioctl
3277 must be called after calling KVM_CREATE_IRQCHIP, but before calling
3278 KVM_RUN on any of the VCPUs. Calling this ioctl twice for any of the
3279 base addresses will return -EEXIST.
3281 Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API
3282 should be used instead.
3285 4.86 KVM_PPC_RTAS_DEFINE_TOKEN
3286 ------------------------------
3288 :Capability: KVM_CAP_PPC_RTAS
3291 :Parameters: struct kvm_rtas_token_args
3292 :Returns: 0 on success, -1 on error
3294 Defines a token value for a RTAS (Run Time Abstraction Services)
3295 service in order to allow it to be handled in the kernel. The
3296 argument struct gives the name of the service, which must be the name
3297 of a service that has a kernel-side implementation. If the token
3298 value is non-zero, it will be associated with that service, and
3299 subsequent RTAS calls by the guest specifying that token will be
3300 handled by the kernel. If the token value is 0, then any token
3301 associated with the service will be forgotten, and subsequent RTAS
3302 calls by the guest for that service will be passed to userspace to be
3305 4.87 KVM_SET_GUEST_DEBUG
3306 ------------------------
3308 :Capability: KVM_CAP_SET_GUEST_DEBUG
3309 :Architectures: x86, s390, ppc, arm64
3311 :Parameters: struct kvm_guest_debug (in)
3312 :Returns: 0 on success; -1 on error
3316 struct kvm_guest_debug {
3319 struct kvm_guest_debug_arch arch;
3322 Set up the processor specific debug registers and configure vcpu for
3323 handling guest debug events. There are two parts to the structure, the
3324 first a control bitfield indicates the type of debug events to handle
3325 when running. Common control bits are:
3327 - KVM_GUESTDBG_ENABLE: guest debugging is enabled
3328 - KVM_GUESTDBG_SINGLESTEP: the next run should single-step
3330 The top 16 bits of the control field are architecture specific control
3331 flags which can include the following:
3333 - KVM_GUESTDBG_USE_SW_BP: using software breakpoints [x86, arm64]
3334 - KVM_GUESTDBG_USE_HW_BP: using hardware breakpoints [x86, s390, arm64]
3335 - KVM_GUESTDBG_INJECT_DB: inject DB type exception [x86]
3336 - KVM_GUESTDBG_INJECT_BP: inject BP type exception [x86]
3337 - KVM_GUESTDBG_EXIT_PENDING: trigger an immediate guest exit [s390]
3339 For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints
3340 are enabled in memory so we need to ensure breakpoint exceptions are
3341 correctly trapped and the KVM run loop exits at the breakpoint and not
3342 running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP
3343 we need to ensure the guest vCPUs architecture specific registers are
3344 updated to the correct (supplied) values.
3346 The second part of the structure is architecture specific and
3347 typically contains a set of debug registers.
3349 For arm64 the number of debug registers is implementation defined and
3350 can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and
3351 KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number
3352 indicating the number of supported registers.
3354 For ppc, the KVM_CAP_PPC_GUEST_DEBUG_SSTEP capability indicates whether
3355 the single-step debug event (KVM_GUESTDBG_SINGLESTEP) is supported.
3357 When debug events exit the main run loop with the reason
3358 KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run
3359 structure containing architecture specific debug information.
3361 4.88 KVM_GET_EMULATED_CPUID
3362 ---------------------------
3364 :Capability: KVM_CAP_EXT_EMUL_CPUID
3367 :Parameters: struct kvm_cpuid2 (in/out)
3368 :Returns: 0 on success, -1 on error
3375 struct kvm_cpuid_entry2 entries[0];
3378 The member 'flags' is used for passing flags from userspace.
3382 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
3383 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */
3384 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */
3386 struct kvm_cpuid_entry2 {
3397 This ioctl returns x86 cpuid features which are emulated by
3398 kvm.Userspace can use the information returned by this ioctl to query
3399 which features are emulated by kvm instead of being present natively.
3401 Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2
3402 structure with the 'nent' field indicating the number of entries in
3403 the variable-size array 'entries'. If the number of entries is too low
3404 to describe the cpu capabilities, an error (E2BIG) is returned. If the
3405 number is too high, the 'nent' field is adjusted and an error (ENOMEM)
3406 is returned. If the number is just right, the 'nent' field is adjusted
3407 to the number of valid entries in the 'entries' array, which is then
3410 The entries returned are the set CPUID bits of the respective features
3411 which kvm emulates, as returned by the CPUID instruction, with unknown
3412 or unsupported feature bits cleared.
3414 Features like x2apic, for example, may not be present in the host cpu
3415 but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be
3416 emulated efficiently and thus not included here.
3418 The fields in each entry are defined as follows:
3421 the eax value used to obtain the entry
3423 the ecx value used to obtain the entry (for entries that are
3426 an OR of zero or more of the following:
3428 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
3429 if the index field is valid
3433 the values returned by the cpuid instruction for
3434 this function/index combination
3436 4.89 KVM_S390_MEM_OP
3437 --------------------
3439 :Capability: KVM_CAP_S390_MEM_OP
3440 :Architectures: s390
3442 :Parameters: struct kvm_s390_mem_op (in)
3443 :Returns: = 0 on success,
3444 < 0 on generic error (e.g. -EFAULT or -ENOMEM),
3445 > 0 if an exception occurred while walking the page tables
3447 Read or write data from/to the logical (virtual) memory of a VCPU.
3449 Parameters are specified via the following structure::
3451 struct kvm_s390_mem_op {
3452 __u64 gaddr; /* the guest address */
3453 __u64 flags; /* flags */
3454 __u32 size; /* amount of bytes */
3455 __u32 op; /* type of operation */
3456 __u64 buf; /* buffer in userspace */
3457 __u8 ar; /* the access register number */
3458 __u8 reserved[31]; /* should be set to 0 */
3461 The type of operation is specified in the "op" field. It is either
3462 KVM_S390_MEMOP_LOGICAL_READ for reading from logical memory space or
3463 KVM_S390_MEMOP_LOGICAL_WRITE for writing to logical memory space. The
3464 KVM_S390_MEMOP_F_CHECK_ONLY flag can be set in the "flags" field to check
3465 whether the corresponding memory access would create an access exception
3466 (without touching the data in the memory at the destination). In case an
3467 access exception occurred while walking the MMU tables of the guest, the
3468 ioctl returns a positive error number to indicate the type of exception.
3469 This exception is also raised directly at the corresponding VCPU if the
3470 flag KVM_S390_MEMOP_F_INJECT_EXCEPTION is set in the "flags" field.
3472 The start address of the memory region has to be specified in the "gaddr"
3473 field, and the length of the region in the "size" field (which must not
3474 be 0). The maximum value for "size" can be obtained by checking the
3475 KVM_CAP_S390_MEM_OP capability. "buf" is the buffer supplied by the
3476 userspace application where the read data should be written to for
3477 KVM_S390_MEMOP_LOGICAL_READ, or where the data that should be written is
3478 stored for a KVM_S390_MEMOP_LOGICAL_WRITE. When KVM_S390_MEMOP_F_CHECK_ONLY
3479 is specified, "buf" is unused and can be NULL. "ar" designates the access
3480 register number to be used; the valid range is 0..15.
3482 The "reserved" field is meant for future extensions. It is not used by
3483 KVM with the currently defined set of flags.
3485 4.90 KVM_S390_GET_SKEYS
3486 -----------------------
3488 :Capability: KVM_CAP_S390_SKEYS
3489 :Architectures: s390
3491 :Parameters: struct kvm_s390_skeys
3492 :Returns: 0 on success, KVM_S390_GET_KEYS_NONE if guest is not using storage
3493 keys, negative value on error
3495 This ioctl is used to get guest storage key values on the s390
3496 architecture. The ioctl takes parameters via the kvm_s390_skeys struct::
3498 struct kvm_s390_skeys {
3501 __u64 skeydata_addr;
3506 The start_gfn field is the number of the first guest frame whose storage keys
3509 The count field is the number of consecutive frames (starting from start_gfn)
3510 whose storage keys to get. The count field must be at least 1 and the maximum
3511 allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range
3512 will cause the ioctl to return -EINVAL.
3514 The skeydata_addr field is the address to a buffer large enough to hold count
3515 bytes. This buffer will be filled with storage key data by the ioctl.
3517 4.91 KVM_S390_SET_SKEYS
3518 -----------------------
3520 :Capability: KVM_CAP_S390_SKEYS
3521 :Architectures: s390
3523 :Parameters: struct kvm_s390_skeys
3524 :Returns: 0 on success, negative value on error
3526 This ioctl is used to set guest storage key values on the s390
3527 architecture. The ioctl takes parameters via the kvm_s390_skeys struct.
3528 See section on KVM_S390_GET_SKEYS for struct definition.
3530 The start_gfn field is the number of the first guest frame whose storage keys
3533 The count field is the number of consecutive frames (starting from start_gfn)
3534 whose storage keys to get. The count field must be at least 1 and the maximum
3535 allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range
3536 will cause the ioctl to return -EINVAL.
3538 The skeydata_addr field is the address to a buffer containing count bytes of
3539 storage keys. Each byte in the buffer will be set as the storage key for a
3540 single frame starting at start_gfn for count frames.
3542 Note: If any architecturally invalid key value is found in the given data then
3543 the ioctl will return -EINVAL.
3548 :Capability: KVM_CAP_S390_INJECT_IRQ
3549 :Architectures: s390
3551 :Parameters: struct kvm_s390_irq (in)
3552 :Returns: 0 on success, -1 on error
3557 ====== =================================================================
3558 EINVAL interrupt type is invalid
3559 type is KVM_S390_SIGP_STOP and flag parameter is invalid value,
3560 type is KVM_S390_INT_EXTERNAL_CALL and code is bigger
3561 than the maximum of VCPUs
3562 EBUSY type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped,
3563 type is KVM_S390_SIGP_STOP and a stop irq is already pending,
3564 type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt
3566 ====== =================================================================
3568 Allows to inject an interrupt to the guest.
3570 Using struct kvm_s390_irq as a parameter allows
3571 to inject additional payload which is not
3572 possible via KVM_S390_INTERRUPT.
3574 Interrupt parameters are passed via kvm_s390_irq::
3576 struct kvm_s390_irq {
3579 struct kvm_s390_io_info io;
3580 struct kvm_s390_ext_info ext;
3581 struct kvm_s390_pgm_info pgm;
3582 struct kvm_s390_emerg_info emerg;
3583 struct kvm_s390_extcall_info extcall;
3584 struct kvm_s390_prefix_info prefix;
3585 struct kvm_s390_stop_info stop;
3586 struct kvm_s390_mchk_info mchk;
3591 type can be one of the following:
3593 - KVM_S390_SIGP_STOP - sigp stop; parameter in .stop
3594 - KVM_S390_PROGRAM_INT - program check; parameters in .pgm
3595 - KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix
3596 - KVM_S390_RESTART - restart; no parameters
3597 - KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters
3598 - KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters
3599 - KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg
3600 - KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall
3601 - KVM_S390_MCHK - machine check interrupt; parameters in .mchk
3603 This is an asynchronous vcpu ioctl and can be invoked from any thread.
3605 4.94 KVM_S390_GET_IRQ_STATE
3606 ---------------------------
3608 :Capability: KVM_CAP_S390_IRQ_STATE
3609 :Architectures: s390
3611 :Parameters: struct kvm_s390_irq_state (out)
3612 :Returns: >= number of bytes copied into buffer,
3613 -EINVAL if buffer size is 0,
3614 -ENOBUFS if buffer size is too small to fit all pending interrupts,
3615 -EFAULT if the buffer address was invalid
3617 This ioctl allows userspace to retrieve the complete state of all currently
3618 pending interrupts in a single buffer. Use cases include migration
3619 and introspection. The parameter structure contains the address of a
3620 userspace buffer and its length::
3622 struct kvm_s390_irq_state {
3624 __u32 flags; /* will stay unused for compatibility reasons */
3626 __u32 reserved[4]; /* will stay unused for compatibility reasons */
3629 Userspace passes in the above struct and for each pending interrupt a
3630 struct kvm_s390_irq is copied to the provided buffer.
3632 The structure contains a flags and a reserved field for future extensions. As
3633 the kernel never checked for flags == 0 and QEMU never pre-zeroed flags and
3634 reserved, these fields can not be used in the future without breaking
3637 If -ENOBUFS is returned the buffer provided was too small and userspace
3638 may retry with a bigger buffer.
3640 4.95 KVM_S390_SET_IRQ_STATE
3641 ---------------------------
3643 :Capability: KVM_CAP_S390_IRQ_STATE
3644 :Architectures: s390
3646 :Parameters: struct kvm_s390_irq_state (in)
3647 :Returns: 0 on success,
3648 -EFAULT if the buffer address was invalid,
3649 -EINVAL for an invalid buffer length (see below),
3650 -EBUSY if there were already interrupts pending,
3651 errors occurring when actually injecting the
3652 interrupt. See KVM_S390_IRQ.
3654 This ioctl allows userspace to set the complete state of all cpu-local
3655 interrupts currently pending for the vcpu. It is intended for restoring
3656 interrupt state after a migration. The input parameter is a userspace buffer
3657 containing a struct kvm_s390_irq_state::
3659 struct kvm_s390_irq_state {
3661 __u32 flags; /* will stay unused for compatibility reasons */
3663 __u32 reserved[4]; /* will stay unused for compatibility reasons */
3666 The restrictions for flags and reserved apply as well.
3667 (see KVM_S390_GET_IRQ_STATE)
3669 The userspace memory referenced by buf contains a struct kvm_s390_irq
3670 for each interrupt to be injected into the guest.
3671 If one of the interrupts could not be injected for some reason the
3674 len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0
3675 and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq),
3676 which is the maximum number of possibly pending cpu-local interrupts.
3681 :Capability: KVM_CAP_X86_SMM
3685 :Returns: 0 on success, -1 on error
3687 Queues an SMI on the thread's vcpu.
3689 4.97 KVM_CAP_PPC_MULTITCE
3690 -------------------------
3692 :Capability: KVM_CAP_PPC_MULTITCE
3696 This capability means the kernel is capable of handling hypercalls
3697 H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user
3698 space. This significantly accelerates DMA operations for PPC KVM guests.
3699 User space should expect that its handlers for these hypercalls
3700 are not going to be called if user space previously registered LIOBN
3701 in KVM (via KVM_CREATE_SPAPR_TCE or similar calls).
3703 In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest,
3704 user space might have to advertise it for the guest. For example,
3705 IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is
3706 present in the "ibm,hypertas-functions" device-tree property.
3708 The hypercalls mentioned above may or may not be processed successfully
3709 in the kernel based fast path. If they can not be handled by the kernel,
3710 they will get passed on to user space. So user space still has to have
3711 an implementation for these despite the in kernel acceleration.
3713 This capability is always enabled.
3715 4.98 KVM_CREATE_SPAPR_TCE_64
3716 ----------------------------
3718 :Capability: KVM_CAP_SPAPR_TCE_64
3719 :Architectures: powerpc
3721 :Parameters: struct kvm_create_spapr_tce_64 (in)
3722 :Returns: file descriptor for manipulating the created TCE table
3724 This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit
3725 windows, described in 4.62 KVM_CREATE_SPAPR_TCE
3727 This capability uses extended struct in ioctl interface::
3729 /* for KVM_CAP_SPAPR_TCE_64 */
3730 struct kvm_create_spapr_tce_64 {
3734 __u64 offset; /* in pages */
3735 __u64 size; /* in pages */
3738 The aim of extension is to support an additional bigger DMA window with
3739 a variable page size.
3740 KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and
3741 a bus offset of the corresponding DMA window, @size and @offset are numbers
3744 @flags are not used at the moment.
3746 The rest of functionality is identical to KVM_CREATE_SPAPR_TCE.
3748 4.99 KVM_REINJECT_CONTROL
3749 -------------------------
3751 :Capability: KVM_CAP_REINJECT_CONTROL
3754 :Parameters: struct kvm_reinject_control (in)
3755 :Returns: 0 on success,
3756 -EFAULT if struct kvm_reinject_control cannot be read,
3757 -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier.
3759 i8254 (PIT) has two modes, reinject and !reinject. The default is reinject,
3760 where KVM queues elapsed i8254 ticks and monitors completion of interrupt from
3761 vector(s) that i8254 injects. Reinject mode dequeues a tick and injects its
3762 interrupt whenever there isn't a pending interrupt from i8254.
3763 !reinject mode injects an interrupt as soon as a tick arrives.
3767 struct kvm_reinject_control {
3772 pit_reinject = 0 (!reinject mode) is recommended, unless running an old
3773 operating system that uses the PIT for timing (e.g. Linux 2.4.x).
3775 4.100 KVM_PPC_CONFIGURE_V3_MMU
3776 ------------------------------
3778 :Capability: KVM_CAP_PPC_RADIX_MMU or KVM_CAP_PPC_HASH_MMU_V3
3781 :Parameters: struct kvm_ppc_mmuv3_cfg (in)
3782 :Returns: 0 on success,
3783 -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read,
3784 -EINVAL if the configuration is invalid
3786 This ioctl controls whether the guest will use radix or HPT (hashed
3787 page table) translation, and sets the pointer to the process table for
3792 struct kvm_ppc_mmuv3_cfg {
3794 __u64 process_table;
3797 There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and
3798 KVM_PPC_MMUV3_GTSE. KVM_PPC_MMUV3_RADIX, if set, configures the guest
3799 to use radix tree translation, and if clear, to use HPT translation.
3800 KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest
3801 to be able to use the global TLB and SLB invalidation instructions;
3802 if clear, the guest may not use these instructions.
3804 The process_table field specifies the address and size of the guest
3805 process table, which is in the guest's space. This field is formatted
3806 as the second doubleword of the partition table entry, as defined in
3807 the Power ISA V3.00, Book III section 5.7.6.1.
3809 4.101 KVM_PPC_GET_RMMU_INFO
3810 ---------------------------
3812 :Capability: KVM_CAP_PPC_RADIX_MMU
3815 :Parameters: struct kvm_ppc_rmmu_info (out)
3816 :Returns: 0 on success,
3817 -EFAULT if struct kvm_ppc_rmmu_info cannot be written,
3818 -EINVAL if no useful information can be returned
3820 This ioctl returns a structure containing two things: (a) a list
3821 containing supported radix tree geometries, and (b) a list that maps
3822 page sizes to put in the "AP" (actual page size) field for the tlbie
3823 (TLB invalidate entry) instruction.
3827 struct kvm_ppc_rmmu_info {
3828 struct kvm_ppc_radix_geom {
3833 __u32 ap_encodings[8];
3836 The geometries[] field gives up to 8 supported geometries for the
3837 radix page table, in terms of the log base 2 of the smallest page
3838 size, and the number of bits indexed at each level of the tree, from
3839 the PTE level up to the PGD level in that order. Any unused entries
3840 will have 0 in the page_shift field.
3842 The ap_encodings gives the supported page sizes and their AP field
3843 encodings, encoded with the AP value in the top 3 bits and the log
3844 base 2 of the page size in the bottom 6 bits.
3846 4.102 KVM_PPC_RESIZE_HPT_PREPARE
3847 --------------------------------
3849 :Capability: KVM_CAP_SPAPR_RESIZE_HPT
3850 :Architectures: powerpc
3852 :Parameters: struct kvm_ppc_resize_hpt (in)
3853 :Returns: 0 on successful completion,
3854 >0 if a new HPT is being prepared, the value is an estimated
3855 number of milliseconds until preparation is complete,
3856 -EFAULT if struct kvm_reinject_control cannot be read,
3857 -EINVAL if the supplied shift or flags are invalid,
3858 -ENOMEM if unable to allocate the new HPT,
3860 Used to implement the PAPR extension for runtime resizing of a guest's
3861 Hashed Page Table (HPT). Specifically this starts, stops or monitors
3862 the preparation of a new potential HPT for the guest, essentially
3863 implementing the H_RESIZE_HPT_PREPARE hypercall.
3867 struct kvm_ppc_resize_hpt {
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.
3900 4.103 KVM_PPC_RESIZE_HPT_COMMIT
3901 -------------------------------
3903 :Capability: KVM_CAP_SPAPR_RESIZE_HPT
3904 :Architectures: powerpc
3906 :Parameters: struct kvm_ppc_resize_hpt (in)
3907 :Returns: 0 on successful completion,
3908 -EFAULT if struct kvm_reinject_control cannot be read,
3909 -EINVAL if the supplied shift or flags are invalid,
3910 -ENXIO is there is no pending HPT, or the pending HPT doesn't
3911 have the requested size,
3912 -EBUSY if the pending HPT is not fully prepared,
3913 -ENOSPC if there was a hash collision when moving existing
3914 HPT entries to the new HPT,
3915 -EIO on other error conditions
3917 Used to implement the PAPR extension for runtime resizing of a guest's
3918 Hashed Page Table (HPT). Specifically this requests that the guest be
3919 transferred to working with the new HPT, essentially implementing the
3920 H_RESIZE_HPT_COMMIT hypercall.
3924 struct kvm_ppc_resize_hpt {
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.
3945 4.104 KVM_X86_GET_MCE_CAP_SUPPORTED
3946 -----------------------------------
3948 :Capability: KVM_CAP_MCE
3951 :Parameters: u64 mce_cap (out)
3952 :Returns: 0 on success, -1 on error
3954 Returns supported MCE capabilities. The u64 mce_cap parameter
3955 has the same format as the MSR_IA32_MCG_CAP register. Supported
3956 capabilities will have the corresponding bits set.
3958 4.105 KVM_X86_SETUP_MCE
3959 -----------------------
3961 :Capability: KVM_CAP_MCE
3964 :Parameters: u64 mcg_cap (in)
3965 :Returns: 0 on success,
3966 -EFAULT if u64 mcg_cap cannot be read,
3967 -EINVAL if the requested number of banks is invalid,
3968 -EINVAL if requested MCE capability is not supported.
3970 Initializes MCE support for use. The u64 mcg_cap parameter
3971 has the same format as the MSR_IA32_MCG_CAP register and
3972 specifies which capabilities should be enabled. The maximum
3973 supported number of error-reporting banks can be retrieved when
3974 checking for KVM_CAP_MCE. The supported capabilities can be
3975 retrieved with KVM_X86_GET_MCE_CAP_SUPPORTED.
3977 4.106 KVM_X86_SET_MCE
3978 ---------------------
3980 :Capability: KVM_CAP_MCE
3983 :Parameters: struct kvm_x86_mce (in)
3984 :Returns: 0 on success,
3985 -EFAULT if struct kvm_x86_mce cannot be read,
3986 -EINVAL if the bank number is invalid,
3987 -EINVAL if VAL bit is not set in status field.
3989 Inject a machine check error (MCE) into the guest. The input
3992 struct kvm_x86_mce {
4002 If the MCE being reported is an uncorrected error, KVM will
4003 inject it as an MCE exception into the guest. If the guest
4004 MCG_STATUS register reports that an MCE is in progress, KVM
4005 causes an KVM_EXIT_SHUTDOWN vmexit.
4007 Otherwise, if the MCE is a corrected error, KVM will just
4008 store it in the corresponding bank (provided this bank is
4009 not holding a previously reported uncorrected error).
4011 4.107 KVM_S390_GET_CMMA_BITS
4012 ----------------------------
4014 :Capability: KVM_CAP_S390_CMMA_MIGRATION
4015 :Architectures: s390
4017 :Parameters: struct kvm_s390_cmma_log (in, out)
4018 :Returns: 0 on success, a negative value on error
4020 This ioctl is used to get the values of the CMMA bits on the s390
4021 architecture. It is meant to be used in two scenarios:
4023 - During live migration to save the CMMA values. Live migration needs
4024 to be enabled via the KVM_REQ_START_MIGRATION VM property.
4025 - To non-destructively peek at the CMMA values, with the flag
4026 KVM_S390_CMMA_PEEK set.
4028 The ioctl takes parameters via the kvm_s390_cmma_log struct. The desired
4029 values are written to a buffer whose location is indicated via the "values"
4030 member in the kvm_s390_cmma_log struct. The values in the input struct are
4031 also updated as needed.
4033 Each CMMA value takes up one byte.
4037 struct kvm_s390_cmma_log {
4048 start_gfn is the number of the first guest frame whose CMMA values are
4051 count is the length of the buffer in bytes,
4053 values points to the buffer where the result will be written to.
4055 If count is greater than KVM_S390_SKEYS_MAX, then it is considered to be
4056 KVM_S390_SKEYS_MAX. KVM_S390_SKEYS_MAX is re-used for consistency with
4059 The result is written in the buffer pointed to by the field values, and
4060 the values of the input parameter are updated as follows.
4062 Depending on the flags, different actions are performed. The only
4063 supported flag so far is KVM_S390_CMMA_PEEK.
4065 The default behaviour if KVM_S390_CMMA_PEEK is not set is:
4066 start_gfn will indicate the first page frame whose CMMA bits were dirty.
4067 It is not necessarily the same as the one passed as input, as clean pages
4070 count will indicate the number of bytes actually written in the buffer.
4071 It can (and very often will) be smaller than the input value, since the
4072 buffer is only filled until 16 bytes of clean values are found (which
4073 are then not copied in the buffer). Since a CMMA migration block needs
4074 the base address and the length, for a total of 16 bytes, we will send
4075 back some clean data if there is some dirty data afterwards, as long as
4076 the size of the clean data does not exceed the size of the header. This
4077 allows to minimize the amount of data to be saved or transferred over
4078 the network at the expense of more roundtrips to userspace. The next
4079 invocation of the ioctl will skip over all the clean values, saving
4080 potentially more than just the 16 bytes we found.
4082 If KVM_S390_CMMA_PEEK is set:
4083 the existing storage attributes are read even when not in migration
4084 mode, and no other action is performed;
4086 the output start_gfn will be equal to the input start_gfn,
4088 the output count will be equal to the input count, except if the end of
4089 memory has been reached.
4092 the field "remaining" will indicate the total number of dirty CMMA values
4093 still remaining, or 0 if KVM_S390_CMMA_PEEK is set and migration mode is
4098 values points to the userspace buffer where the result will be stored.
4100 This ioctl can fail with -ENOMEM if not enough memory can be allocated to
4101 complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
4102 KVM_S390_CMMA_PEEK is not set but migration mode was not enabled, with
4103 -EFAULT if the userspace address is invalid or if no page table is
4104 present for the addresses (e.g. when using hugepages).
4106 4.108 KVM_S390_SET_CMMA_BITS
4107 ----------------------------
4109 :Capability: KVM_CAP_S390_CMMA_MIGRATION
4110 :Architectures: s390
4112 :Parameters: struct kvm_s390_cmma_log (in)
4113 :Returns: 0 on success, a negative value on error
4115 This ioctl is used to set the values of the CMMA bits on the s390
4116 architecture. It is meant to be used during live migration to restore
4117 the CMMA values, but there are no restrictions on its use.
4118 The ioctl takes parameters via the kvm_s390_cmma_values struct.
4119 Each CMMA value takes up one byte.
4123 struct kvm_s390_cmma_log {
4134 start_gfn indicates the starting guest frame number,
4136 count indicates how many values are to be considered in the buffer,
4138 flags is not used and must be 0.
4140 mask indicates which PGSTE bits are to be considered.
4142 remaining is not used.
4144 values points to the buffer in userspace where to store the values.
4146 This ioctl can fail with -ENOMEM if not enough memory can be allocated to
4147 complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
4148 the count field is too large (e.g. more than KVM_S390_CMMA_SIZE_MAX) or
4149 if the flags field was not 0, with -EFAULT if the userspace address is
4150 invalid, if invalid pages are written to (e.g. after the end of memory)
4151 or if no page table is present for the addresses (e.g. when using
4154 4.109 KVM_PPC_GET_CPU_CHAR
4155 --------------------------
4157 :Capability: KVM_CAP_PPC_GET_CPU_CHAR
4158 :Architectures: powerpc
4160 :Parameters: struct kvm_ppc_cpu_char (out)
4161 :Returns: 0 on successful completion,
4162 -EFAULT if struct kvm_ppc_cpu_char cannot be written
4164 This ioctl gives userspace information about certain characteristics
4165 of the CPU relating to speculative execution of instructions and
4166 possible information leakage resulting from speculative execution (see
4167 CVE-2017-5715, CVE-2017-5753 and CVE-2017-5754). The information is
4168 returned in struct kvm_ppc_cpu_char, which looks like this::
4170 struct kvm_ppc_cpu_char {
4171 __u64 character; /* characteristics of the CPU */
4172 __u64 behaviour; /* recommended software behaviour */
4173 __u64 character_mask; /* valid bits in character */
4174 __u64 behaviour_mask; /* valid bits in behaviour */
4177 For extensibility, the character_mask and behaviour_mask fields
4178 indicate which bits of character and behaviour have been filled in by
4179 the kernel. If the set of defined bits is extended in future then
4180 userspace will be able to tell whether it is running on a kernel that
4181 knows about the new bits.
4183 The character field describes attributes of the CPU which can help
4184 with preventing inadvertent information disclosure - specifically,
4185 whether there is an instruction to flash-invalidate the L1 data cache
4186 (ori 30,30,0 or mtspr SPRN_TRIG2,rN), whether the L1 data cache is set
4187 to a mode where entries can only be used by the thread that created
4188 them, whether the bcctr[l] instruction prevents speculation, and
4189 whether a speculation barrier instruction (ori 31,31,0) is provided.
4191 The behaviour field describes actions that software should take to
4192 prevent inadvertent information disclosure, and thus describes which
4193 vulnerabilities the hardware is subject to; specifically whether the
4194 L1 data cache should be flushed when returning to user mode from the
4195 kernel, and whether a speculation barrier should be placed between an
4196 array bounds check and the array access.
4198 These fields use the same bit definitions as the new
4199 H_GET_CPU_CHARACTERISTICS hypercall.
4201 4.110 KVM_MEMORY_ENCRYPT_OP
4202 ---------------------------
4207 :Parameters: an opaque platform specific structure (in/out)
4208 :Returns: 0 on success; -1 on error
4210 If the platform supports creating encrypted VMs then this ioctl can be used
4211 for issuing platform-specific memory encryption commands to manage those
4214 Currently, this ioctl is used for issuing Secure Encrypted Virtualization
4215 (SEV) commands on AMD Processors. The SEV commands are defined in
4216 Documentation/virt/kvm/amd-memory-encryption.rst.
4218 4.111 KVM_MEMORY_ENCRYPT_REG_REGION
4219 -----------------------------------
4224 :Parameters: struct kvm_enc_region (in)
4225 :Returns: 0 on success; -1 on error
4227 This ioctl can be used to register a guest memory region which may
4228 contain encrypted data (e.g. guest RAM, SMRAM etc).
4230 It is used in the SEV-enabled guest. When encryption is enabled, a guest
4231 memory region may contain encrypted data. The SEV memory encryption
4232 engine uses a tweak such that two identical plaintext pages, each at
4233 different locations will have differing ciphertexts. So swapping or
4234 moving ciphertext of those pages will not result in plaintext being
4235 swapped. So relocating (or migrating) physical backing pages for the SEV
4236 guest will require some additional steps.
4238 Note: The current SEV key management spec does not provide commands to
4239 swap or migrate (move) ciphertext pages. Hence, for now we pin the guest
4240 memory region registered with the ioctl.
4242 4.112 KVM_MEMORY_ENCRYPT_UNREG_REGION
4243 -------------------------------------
4248 :Parameters: struct kvm_enc_region (in)
4249 :Returns: 0 on success; -1 on error
4251 This ioctl can be used to unregister the guest memory region registered
4252 with KVM_MEMORY_ENCRYPT_REG_REGION ioctl above.
4254 4.113 KVM_HYPERV_EVENTFD
4255 ------------------------
4257 :Capability: KVM_CAP_HYPERV_EVENTFD
4260 :Parameters: struct kvm_hyperv_eventfd (in)
4262 This ioctl (un)registers an eventfd to receive notifications from the guest on
4263 the specified Hyper-V connection id through the SIGNAL_EVENT hypercall, without
4264 causing a user exit. SIGNAL_EVENT hypercall with non-zero event flag number
4265 (bits 24-31) still triggers a KVM_EXIT_HYPERV_HCALL user exit.
4269 struct kvm_hyperv_eventfd {
4276 The conn_id field should fit within 24 bits::
4278 #define KVM_HYPERV_CONN_ID_MASK 0x00ffffff
4280 The acceptable values for the flags field are::
4282 #define KVM_HYPERV_EVENTFD_DEASSIGN (1 << 0)
4284 :Returns: 0 on success,
4285 -EINVAL if conn_id or flags is outside the allowed range,
4286 -ENOENT on deassign if the conn_id isn't registered,
4287 -EEXIST on assign if the conn_id is already registered
4289 4.114 KVM_GET_NESTED_STATE
4290 --------------------------
4292 :Capability: KVM_CAP_NESTED_STATE
4295 :Parameters: struct kvm_nested_state (in/out)
4296 :Returns: 0 on success, -1 on error
4300 ===== =============================================================
4301 E2BIG the total state size exceeds the value of 'size' specified by
4302 the user; the size required will be written into size.
4303 ===== =============================================================
4307 struct kvm_nested_state {
4313 struct kvm_vmx_nested_state_hdr vmx;
4314 struct kvm_svm_nested_state_hdr svm;
4316 /* Pad the header to 128 bytes. */
4321 struct kvm_vmx_nested_state_data vmx[0];
4322 struct kvm_svm_nested_state_data svm[0];
4326 #define KVM_STATE_NESTED_GUEST_MODE 0x00000001
4327 #define KVM_STATE_NESTED_RUN_PENDING 0x00000002
4328 #define KVM_STATE_NESTED_EVMCS 0x00000004
4330 #define KVM_STATE_NESTED_FORMAT_VMX 0
4331 #define KVM_STATE_NESTED_FORMAT_SVM 1
4333 #define KVM_STATE_NESTED_VMX_VMCS_SIZE 0x1000
4335 #define KVM_STATE_NESTED_VMX_SMM_GUEST_MODE 0x00000001
4336 #define KVM_STATE_NESTED_VMX_SMM_VMXON 0x00000002
4338 #define KVM_STATE_VMX_PREEMPTION_TIMER_DEADLINE 0x00000001
4340 struct kvm_vmx_nested_state_hdr {
4349 __u64 preemption_timer_deadline;
4352 struct kvm_vmx_nested_state_data {
4353 __u8 vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE];
4354 __u8 shadow_vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE];
4357 This ioctl copies the vcpu's nested virtualization state from the kernel to
4360 The maximum size of the state can be retrieved by passing KVM_CAP_NESTED_STATE
4361 to the KVM_CHECK_EXTENSION ioctl().
4363 4.115 KVM_SET_NESTED_STATE
4364 --------------------------
4366 :Capability: KVM_CAP_NESTED_STATE
4369 :Parameters: struct kvm_nested_state (in)
4370 :Returns: 0 on success, -1 on error
4372 This copies the vcpu's kvm_nested_state struct from userspace to the kernel.
4373 For the definition of struct kvm_nested_state, see KVM_GET_NESTED_STATE.
4375 4.116 KVM_(UN)REGISTER_COALESCED_MMIO
4376 -------------------------------------
4378 :Capability: KVM_CAP_COALESCED_MMIO (for coalesced mmio)
4379 KVM_CAP_COALESCED_PIO (for coalesced pio)
4382 :Parameters: struct kvm_coalesced_mmio_zone
4383 :Returns: 0 on success, < 0 on error
4385 Coalesced I/O is a performance optimization that defers hardware
4386 register write emulation so that userspace exits are avoided. It is
4387 typically used to reduce the overhead of emulating frequently accessed
4390 When a hardware register is configured for coalesced I/O, write accesses
4391 do not exit to userspace and their value is recorded in a ring buffer
4392 that is shared between kernel and userspace.
4394 Coalesced I/O is used if one or more write accesses to a hardware
4395 register can be deferred until a read or a write to another hardware
4396 register on the same device. This last access will cause a vmexit and
4397 userspace will process accesses from the ring buffer before emulating
4398 it. That will avoid exiting to userspace on repeated writes.
4400 Coalesced pio is based on coalesced mmio. There is little difference
4401 between coalesced mmio and pio except that coalesced pio records accesses
4404 4.117 KVM_CLEAR_DIRTY_LOG (vm ioctl)
4405 ------------------------------------
4407 :Capability: KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
4408 :Architectures: x86, arm, arm64, mips
4410 :Parameters: struct kvm_clear_dirty_log (in)
4411 :Returns: 0 on success, -1 on error
4415 /* for KVM_CLEAR_DIRTY_LOG */
4416 struct kvm_clear_dirty_log {
4421 void __user *dirty_bitmap; /* one bit per page */
4426 The ioctl clears the dirty status of pages in a memory slot, according to
4427 the bitmap that is passed in struct kvm_clear_dirty_log's dirty_bitmap
4428 field. Bit 0 of the bitmap corresponds to page "first_page" in the
4429 memory slot, and num_pages is the size in bits of the input bitmap.
4430 first_page must be a multiple of 64; num_pages must also be a multiple of
4431 64 unless first_page + num_pages is the size of the memory slot. For each
4432 bit that is set in the input bitmap, the corresponding page is marked "clean"
4433 in KVM's dirty bitmap, and dirty tracking is re-enabled for that page
4434 (for example via write-protection, or by clearing the dirty bit in
4435 a page table entry).
4437 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies
4438 the address space for which you want to clear the dirty status. See
4439 KVM_SET_USER_MEMORY_REGION for details on the usage of slot field.
4441 This ioctl is mostly useful when KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
4442 is enabled; for more information, see the description of the capability.
4443 However, it can always be used as long as KVM_CHECK_EXTENSION confirms
4444 that KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is present.
4446 4.118 KVM_GET_SUPPORTED_HV_CPUID
4447 --------------------------------
4449 :Capability: KVM_CAP_HYPERV_CPUID (vcpu), KVM_CAP_SYS_HYPERV_CPUID (system)
4451 :Type: system ioctl, vcpu ioctl
4452 :Parameters: struct kvm_cpuid2 (in/out)
4453 :Returns: 0 on success, -1 on error
4460 struct kvm_cpuid_entry2 entries[0];
4463 struct kvm_cpuid_entry2 {
4474 This ioctl returns x86 cpuid features leaves related to Hyper-V emulation in
4475 KVM. Userspace can use the information returned by this ioctl to construct
4476 cpuid information presented to guests consuming Hyper-V enlightenments (e.g.
4477 Windows or Hyper-V guests).
4479 CPUID feature leaves returned by this ioctl are defined by Hyper-V Top Level
4480 Functional Specification (TLFS). These leaves can't be obtained with
4481 KVM_GET_SUPPORTED_CPUID ioctl because some of them intersect with KVM feature
4482 leaves (0x40000000, 0x40000001).
4484 Currently, the following list of CPUID leaves are returned:
4486 - HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS
4487 - HYPERV_CPUID_INTERFACE
4488 - HYPERV_CPUID_VERSION
4489 - HYPERV_CPUID_FEATURES
4490 - HYPERV_CPUID_ENLIGHTMENT_INFO
4491 - HYPERV_CPUID_IMPLEMENT_LIMITS
4492 - HYPERV_CPUID_NESTED_FEATURES
4493 - HYPERV_CPUID_SYNDBG_VENDOR_AND_MAX_FUNCTIONS
4494 - HYPERV_CPUID_SYNDBG_INTERFACE
4495 - HYPERV_CPUID_SYNDBG_PLATFORM_CAPABILITIES
4497 Userspace invokes KVM_GET_SUPPORTED_HV_CPUID by passing a kvm_cpuid2 structure
4498 with the 'nent' field indicating the number of entries in the variable-size
4499 array 'entries'. If the number of entries is too low to describe all Hyper-V
4500 feature leaves, an error (E2BIG) is returned. If the number is more or equal
4501 to the number of Hyper-V feature leaves, the 'nent' field is adjusted to the
4502 number of valid entries in the 'entries' array, which is then filled.
4504 'index' and 'flags' fields in 'struct kvm_cpuid_entry2' are currently reserved,
4505 userspace should not expect to get any particular value there.
4507 Note, vcpu version of KVM_GET_SUPPORTED_HV_CPUID is currently deprecated. Unlike
4508 system ioctl which exposes all supported feature bits unconditionally, vcpu
4509 version has the following quirks:
4511 - HYPERV_CPUID_NESTED_FEATURES leaf and HV_X64_ENLIGHTENED_VMCS_RECOMMENDED
4512 feature bit are only exposed when Enlightened VMCS was previously enabled
4513 on the corresponding vCPU (KVM_CAP_HYPERV_ENLIGHTENED_VMCS).
4514 - HV_STIMER_DIRECT_MODE_AVAILABLE bit is only exposed with in-kernel LAPIC.
4515 (presumes KVM_CREATE_IRQCHIP has already been called).
4517 4.119 KVM_ARM_VCPU_FINALIZE
4518 ---------------------------
4520 :Architectures: arm, arm64
4522 :Parameters: int feature (in)
4523 :Returns: 0 on success, -1 on error
4527 ====== ==============================================================
4528 EPERM feature not enabled, needs configuration, or already finalized
4529 EINVAL feature unknown or not present
4530 ====== ==============================================================
4532 Recognised values for feature:
4534 ===== ===========================================
4535 arm64 KVM_ARM_VCPU_SVE (requires KVM_CAP_ARM_SVE)
4536 ===== ===========================================
4538 Finalizes the configuration of the specified vcpu feature.
4540 The vcpu must already have been initialised, enabling the affected feature, by
4541 means of a successful KVM_ARM_VCPU_INIT call with the appropriate flag set in
4544 For affected vcpu features, this is a mandatory step that must be performed
4545 before the vcpu is fully usable.
4547 Between KVM_ARM_VCPU_INIT and KVM_ARM_VCPU_FINALIZE, the feature may be
4548 configured by use of ioctls such as KVM_SET_ONE_REG. The exact configuration
4549 that should be performaned and how to do it are feature-dependent.
4551 Other calls that depend on a particular feature being finalized, such as
4552 KVM_RUN, KVM_GET_REG_LIST, KVM_GET_ONE_REG and KVM_SET_ONE_REG, will fail with
4553 -EPERM unless the feature has already been finalized by means of a
4554 KVM_ARM_VCPU_FINALIZE call.
4556 See KVM_ARM_VCPU_INIT for details of vcpu features that require finalization
4559 4.120 KVM_SET_PMU_EVENT_FILTER
4560 ------------------------------
4562 :Capability: KVM_CAP_PMU_EVENT_FILTER
4565 :Parameters: struct kvm_pmu_event_filter (in)
4566 :Returns: 0 on success, -1 on error
4570 struct kvm_pmu_event_filter {
4573 __u32 fixed_counter_bitmap;
4579 This ioctl restricts the set of PMU events that the guest can program.
4580 The argument holds a list of events which will be allowed or denied.
4581 The eventsel+umask of each event the guest attempts to program is compared
4582 against the events field to determine whether the guest should have access.
4583 The events field only controls general purpose counters; fixed purpose
4584 counters are controlled by the fixed_counter_bitmap.
4586 No flags are defined yet, the field must be zero.
4588 Valid values for 'action'::
4590 #define KVM_PMU_EVENT_ALLOW 0
4591 #define KVM_PMU_EVENT_DENY 1
4593 4.121 KVM_PPC_SVM_OFF
4594 ---------------------
4597 :Architectures: powerpc
4600 :Returns: 0 on successful completion,
4604 ====== ================================================================
4605 EINVAL if ultravisor failed to terminate the secure guest
4606 ENOMEM if hypervisor failed to allocate new radix page tables for guest
4607 ====== ================================================================
4609 This ioctl is used to turn off the secure mode of the guest or transition
4610 the guest from secure mode to normal mode. This is invoked when the guest
4611 is reset. This has no effect if called for a normal guest.
4613 This ioctl issues an ultravisor call to terminate the secure guest,
4614 unpins the VPA pages and releases all the device pages that are used to
4615 track the secure pages by hypervisor.
4617 4.122 KVM_S390_NORMAL_RESET
4618 ---------------------------
4620 :Capability: KVM_CAP_S390_VCPU_RESETS
4621 :Architectures: s390
4626 This ioctl resets VCPU registers and control structures according to
4627 the cpu reset definition in the POP (Principles Of Operation).
4629 4.123 KVM_S390_INITIAL_RESET
4630 ----------------------------
4633 :Architectures: s390
4638 This ioctl resets VCPU registers and control structures according to
4639 the initial cpu reset definition in the POP. However, the cpu is not
4640 put into ESA mode. This reset is a superset of the normal reset.
4642 4.124 KVM_S390_CLEAR_RESET
4643 --------------------------
4645 :Capability: KVM_CAP_S390_VCPU_RESETS
4646 :Architectures: s390
4651 This ioctl resets VCPU registers and control structures according to
4652 the clear cpu reset definition in the POP. However, the cpu is not put
4653 into ESA mode. This reset is a superset of the initial reset.
4656 4.125 KVM_S390_PV_COMMAND
4657 -------------------------
4659 :Capability: KVM_CAP_S390_PROTECTED
4660 :Architectures: s390
4662 :Parameters: struct kvm_pv_cmd
4663 :Returns: 0 on success, < 0 on error
4668 __u32 cmd; /* Command to be executed */
4669 __u16 rc; /* Ultravisor return code */
4670 __u16 rrc; /* Ultravisor return reason code */
4671 __u64 data; /* Data or address */
4672 __u32 flags; /* flags for future extensions. Must be 0 for now */
4679 Allocate memory and register the VM with the Ultravisor, thereby
4680 donating memory to the Ultravisor that will become inaccessible to
4681 KVM. All existing CPUs are converted to protected ones. After this
4682 command has succeeded, any CPU added via hotplug will become
4683 protected during its creation as well.
4687 ===== =============================
4688 EINTR an unmasked signal is pending
4689 ===== =============================
4693 Deregister the VM from the Ultravisor and reclaim the memory that
4694 had been donated to the Ultravisor, making it usable by the kernel
4695 again. All registered VCPUs are converted back to non-protected
4698 KVM_PV_VM_SET_SEC_PARMS
4699 Pass the image header from VM memory to the Ultravisor in
4700 preparation of image unpacking and verification.
4703 Unpack (protect and decrypt) a page of the encrypted boot image.
4706 Verify the integrity of the unpacked image. Only if this succeeds,
4707 KVM is allowed to start protected VCPUs.
4709 4.126 KVM_X86_SET_MSR_FILTER
4710 ----------------------------
4712 :Capability: KVM_X86_SET_MSR_FILTER
4715 :Parameters: struct kvm_msr_filter
4716 :Returns: 0 on success, < 0 on error
4720 struct kvm_msr_filter_range {
4721 #define KVM_MSR_FILTER_READ (1 << 0)
4722 #define KVM_MSR_FILTER_WRITE (1 << 1)
4724 __u32 nmsrs; /* number of msrs in bitmap */
4725 __u32 base; /* MSR index the bitmap starts at */
4726 __u8 *bitmap; /* a 1 bit allows the operations in flags, 0 denies */
4729 #define KVM_MSR_FILTER_MAX_RANGES 16
4730 struct kvm_msr_filter {
4731 #define KVM_MSR_FILTER_DEFAULT_ALLOW (0 << 0)
4732 #define KVM_MSR_FILTER_DEFAULT_DENY (1 << 0)
4734 struct kvm_msr_filter_range ranges[KVM_MSR_FILTER_MAX_RANGES];
4737 flags values for ``struct kvm_msr_filter_range``:
4739 ``KVM_MSR_FILTER_READ``
4741 Filter read accesses to MSRs using the given bitmap. A 0 in the bitmap
4742 indicates that a read should immediately fail, while a 1 indicates that
4743 a read for a particular MSR should be handled regardless of the default
4746 ``KVM_MSR_FILTER_WRITE``
4748 Filter write accesses to MSRs using the given bitmap. A 0 in the bitmap
4749 indicates that a write should immediately fail, while a 1 indicates that
4750 a write for a particular MSR should be handled regardless of the default
4753 ``KVM_MSR_FILTER_READ | KVM_MSR_FILTER_WRITE``
4755 Filter both read and write accesses to MSRs using the given bitmap. A 0
4756 in the bitmap indicates that both reads and writes should immediately fail,
4757 while a 1 indicates that reads and writes for a particular MSR are not
4758 filtered by this range.
4760 flags values for ``struct kvm_msr_filter``:
4762 ``KVM_MSR_FILTER_DEFAULT_ALLOW``
4764 If no filter range matches an MSR index that is getting accessed, KVM will
4765 fall back to allowing access to the MSR.
4767 ``KVM_MSR_FILTER_DEFAULT_DENY``
4769 If no filter range matches an MSR index that is getting accessed, KVM will
4770 fall back to rejecting access to the MSR. In this mode, all MSRs that should
4771 be processed by KVM need to explicitly be marked as allowed in the bitmaps.
4773 This ioctl allows user space to define up to 16 bitmaps of MSR ranges to
4774 specify whether a certain MSR access should be explicitly filtered for or not.
4776 If this ioctl has never been invoked, MSR accesses are not guarded and the
4777 default KVM in-kernel emulation behavior is fully preserved.
4779 Calling this ioctl with an empty set of ranges (all nmsrs == 0) disables MSR
4780 filtering. In that mode, ``KVM_MSR_FILTER_DEFAULT_DENY`` is invalid and causes
4783 As soon as the filtering is in place, every MSR access is processed through
4784 the filtering except for accesses to the x2APIC MSRs (from 0x800 to 0x8ff);
4785 x2APIC MSRs are always allowed, independent of the ``default_allow`` setting,
4786 and their behavior depends on the ``X2APIC_ENABLE`` bit of the APIC base
4789 If a bit is within one of the defined ranges, read and write accesses are
4790 guarded by the bitmap's value for the MSR index if the kind of access
4791 is included in the ``struct kvm_msr_filter_range`` flags. If no range
4792 cover this particular access, the behavior is determined by the flags
4793 field in the kvm_msr_filter struct: ``KVM_MSR_FILTER_DEFAULT_ALLOW``
4794 and ``KVM_MSR_FILTER_DEFAULT_DENY``.
4796 Each bitmap range specifies a range of MSRs to potentially allow access on.
4797 The range goes from MSR index [base .. base+nmsrs]. The flags field
4798 indicates whether reads, writes or both reads and writes are filtered
4799 by setting a 1 bit in the bitmap for the corresponding MSR index.
4801 If an MSR access is not permitted through the filtering, it generates a
4802 #GP inside the guest. When combined with KVM_CAP_X86_USER_SPACE_MSR, that
4803 allows user space to deflect and potentially handle various MSR accesses
4806 If a vCPU is in running state while this ioctl is invoked, the vCPU may
4807 experience inconsistent filtering behavior on MSR accesses.
4809 4.127 KVM_XEN_HVM_SET_ATTR
4810 --------------------------
4812 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
4815 :Parameters: struct kvm_xen_hvm_attr
4816 :Returns: 0 on success, < 0 on error
4820 struct kvm_xen_hvm_attr {
4835 KVM_XEN_ATTR_TYPE_LONG_MODE
4836 Sets the ABI mode of the VM to 32-bit or 64-bit (long mode). This
4837 determines the layout of the shared info pages exposed to the VM.
4839 KVM_XEN_ATTR_TYPE_SHARED_INFO
4840 Sets the guest physical frame number at which the Xen "shared info"
4841 page resides. Note that although Xen places vcpu_info for the first
4842 32 vCPUs in the shared_info page, KVM does not automatically do so
4843 and instead requires that KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO be used
4844 explicitly even when the vcpu_info for a given vCPU resides at the
4845 "default" location in the shared_info page. This is because KVM is
4846 not aware of the Xen CPU id which is used as the index into the
4847 vcpu_info[] array, so cannot know the correct default location.
4849 KVM_XEN_ATTR_TYPE_UPCALL_VECTOR
4850 Sets the exception vector used to deliver Xen event channel upcalls.
4852 4.128 KVM_XEN_HVM_GET_ATTR
4853 --------------------------
4855 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
4858 :Parameters: struct kvm_xen_hvm_attr
4859 :Returns: 0 on success, < 0 on error
4861 Allows Xen VM attributes to be read. For the structure and types,
4862 see KVM_XEN_HVM_SET_ATTR above.
4864 4.129 KVM_XEN_VCPU_SET_ATTR
4865 ---------------------------
4867 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
4870 :Parameters: struct kvm_xen_vcpu_attr
4871 :Returns: 0 on success, < 0 on error
4875 struct kvm_xen_vcpu_attr {
4883 __u64 state_entry_time;
4885 __u64 time_runnable;
4894 KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO
4895 Sets the guest physical address of the vcpu_info for a given vCPU.
4897 KVM_XEN_VCPU_ATTR_TYPE_VCPU_TIME_INFO
4898 Sets the guest physical address of an additional pvclock structure
4899 for a given vCPU. This is typically used for guest vsyscall support.
4901 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR
4902 Sets the guest physical address of the vcpu_runstate_info for a given
4903 vCPU. This is how a Xen guest tracks CPU state such as steal time.
4905 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_CURRENT
4906 Sets the runstate (RUNSTATE_running/_runnable/_blocked/_offline) of
4907 the given vCPU from the .u.runstate.state member of the structure.
4908 KVM automatically accounts running and runnable time but blocked
4909 and offline states are only entered explicitly.
4911 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_DATA
4912 Sets all fields of the vCPU runstate data from the .u.runstate member
4913 of the structure, including the current runstate. The state_entry_time
4914 must equal the sum of the other four times.
4916 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST
4917 This *adds* the contents of the .u.runstate members of the structure
4918 to the corresponding members of the given vCPU's runstate data, thus
4919 permitting atomic adjustments to the runstate times. The adjustment
4920 to the state_entry_time must equal the sum of the adjustments to the
4921 other four times. The state field must be set to -1, or to a valid
4922 runstate value (RUNSTATE_running, RUNSTATE_runnable, RUNSTATE_blocked
4923 or RUNSTATE_offline) to set the current accounted state as of the
4924 adjusted state_entry_time.
4926 4.130 KVM_XEN_VCPU_GET_ATTR
4927 ---------------------------
4929 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
4932 :Parameters: struct kvm_xen_vcpu_attr
4933 :Returns: 0 on success, < 0 on error
4935 Allows Xen vCPU attributes to be read. For the structure and types,
4936 see KVM_XEN_VCPU_SET_ATTR above.
4938 The KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST type may not be used
4939 with the KVM_XEN_VCPU_GET_ATTR ioctl.
4941 5. The kvm_run structure
4942 ========================
4944 Application code obtains a pointer to the kvm_run structure by
4945 mmap()ing a vcpu fd. From that point, application code can control
4946 execution by changing fields in kvm_run prior to calling the KVM_RUN
4947 ioctl, and obtain information about the reason KVM_RUN returned by
4948 looking up structure members.
4954 __u8 request_interrupt_window;
4956 Request that KVM_RUN return when it becomes possible to inject external
4957 interrupts into the guest. Useful in conjunction with KVM_INTERRUPT.
4961 __u8 immediate_exit;
4963 This field is polled once when KVM_RUN starts; if non-zero, KVM_RUN
4964 exits immediately, returning -EINTR. In the common scenario where a
4965 signal is used to "kick" a VCPU out of KVM_RUN, this field can be used
4966 to avoid usage of KVM_SET_SIGNAL_MASK, which has worse scalability.
4967 Rather than blocking the signal outside KVM_RUN, userspace can set up
4968 a signal handler that sets run->immediate_exit to a non-zero value.
4970 This field is ignored if KVM_CAP_IMMEDIATE_EXIT is not available.
4979 When KVM_RUN has returned successfully (return value 0), this informs
4980 application code why KVM_RUN has returned. Allowable values for this
4981 field are detailed below.
4985 __u8 ready_for_interrupt_injection;
4987 If request_interrupt_window has been specified, this field indicates
4988 an interrupt can be injected now with KVM_INTERRUPT.
4994 The value of the current interrupt flag. Only valid if in-kernel
4995 local APIC is not used.
5001 More architecture-specific flags detailing state of the VCPU that may
5002 affect the device's behavior. Current defined flags::
5004 /* x86, set if the VCPU is in system management mode */
5005 #define KVM_RUN_X86_SMM (1 << 0)
5006 /* x86, set if bus lock detected in VM */
5007 #define KVM_RUN_BUS_LOCK (1 << 1)
5011 /* in (pre_kvm_run), out (post_kvm_run) */
5014 The value of the cr8 register. Only valid if in-kernel local APIC is
5015 not used. Both input and output.
5021 The value of the APIC BASE msr. Only valid if in-kernel local
5022 APIC is not used. Both input and output.
5027 /* KVM_EXIT_UNKNOWN */
5029 __u64 hardware_exit_reason;
5032 If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown
5033 reasons. Further architecture-specific information is available in
5034 hardware_exit_reason.
5038 /* KVM_EXIT_FAIL_ENTRY */
5040 __u64 hardware_entry_failure_reason;
5041 __u32 cpu; /* if KVM_LAST_CPU */
5044 If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due
5045 to unknown reasons. Further architecture-specific information is
5046 available in hardware_entry_failure_reason.
5050 /* KVM_EXIT_EXCEPTION */
5062 #define KVM_EXIT_IO_IN 0
5063 #define KVM_EXIT_IO_OUT 1
5065 __u8 size; /* bytes */
5068 __u64 data_offset; /* relative to kvm_run start */
5071 If exit_reason is KVM_EXIT_IO, then the vcpu has
5072 executed a port I/O instruction which could not be satisfied by kvm.
5073 data_offset describes where the data is located (KVM_EXIT_IO_OUT) or
5074 where kvm expects application code to place the data for the next
5075 KVM_RUN invocation (KVM_EXIT_IO_IN). Data format is a packed array.
5079 /* KVM_EXIT_DEBUG */
5081 struct kvm_debug_exit_arch arch;
5084 If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event
5085 for which architecture specific information is returned.
5097 If exit_reason is KVM_EXIT_MMIO, then the vcpu has
5098 executed a memory-mapped I/O instruction which could not be satisfied
5099 by kvm. The 'data' member contains the written data if 'is_write' is
5100 true, and should be filled by application code otherwise.
5102 The 'data' member contains, in its first 'len' bytes, the value as it would
5103 appear if the VCPU performed a load or store of the appropriate width directly
5108 For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR, KVM_EXIT_XEN,
5109 KVM_EXIT_EPR, KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR the corresponding
5110 operations are complete (and guest state is consistent) only after userspace
5111 has re-entered the kernel with KVM_RUN. The kernel side will first finish
5112 incomplete operations and then check for pending signals.
5114 The pending state of the operation is not preserved in state which is
5115 visible to userspace, thus userspace should ensure that the operation is
5116 completed before performing a live migration. Userspace can re-enter the
5117 guest with an unmasked signal pending or with the immediate_exit field set
5118 to complete pending operations without allowing any further instructions
5123 /* KVM_EXIT_HYPERCALL */
5132 Unused. This was once used for 'hypercall to userspace'. To implement
5133 such functionality, use KVM_EXIT_IO (x86) or KVM_EXIT_MMIO (all except s390).
5135 .. note:: KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO.
5139 /* KVM_EXIT_TPR_ACCESS */
5146 To be documented (KVM_TPR_ACCESS_REPORTING).
5150 /* KVM_EXIT_S390_SIEIC */
5153 __u64 mask; /* psw upper half */
5154 __u64 addr; /* psw lower half */
5163 /* KVM_EXIT_S390_RESET */
5164 #define KVM_S390_RESET_POR 1
5165 #define KVM_S390_RESET_CLEAR 2
5166 #define KVM_S390_RESET_SUBSYSTEM 4
5167 #define KVM_S390_RESET_CPU_INIT 8
5168 #define KVM_S390_RESET_IPL 16
5169 __u64 s390_reset_flags;
5175 /* KVM_EXIT_S390_UCONTROL */
5177 __u64 trans_exc_code;
5181 s390 specific. A page fault has occurred for a user controlled virtual
5182 machine (KVM_VM_S390_UNCONTROL) on it's host page table that cannot be
5183 resolved by the kernel.
5184 The program code and the translation exception code that were placed
5185 in the cpu's lowcore are presented here as defined by the z Architecture
5186 Principles of Operation Book in the Chapter for Dynamic Address Translation
5198 Deprecated - was used for 440 KVM.
5207 MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch
5208 hypercalls and exit with this exit struct that contains all the guest gprs.
5210 If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall.
5211 Userspace can now handle the hypercall and when it's done modify the gprs as
5212 necessary. Upon guest entry all guest GPRs will then be replaced by the values
5217 /* KVM_EXIT_PAPR_HCALL */
5224 This is used on 64-bit PowerPC when emulating a pSeries partition,
5225 e.g. with the 'pseries' machine type in qemu. It occurs when the
5226 guest does a hypercall using the 'sc 1' instruction. The 'nr' field
5227 contains the hypercall number (from the guest R3), and 'args' contains
5228 the arguments (from the guest R4 - R12). Userspace should put the
5229 return code in 'ret' and any extra returned values in args[].
5230 The possible hypercalls are defined in the Power Architecture Platform
5231 Requirements (PAPR) document available from www.power.org (free
5232 developer registration required to access it).
5236 /* KVM_EXIT_S390_TSCH */
5238 __u16 subchannel_id;
5239 __u16 subchannel_nr;
5246 s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled
5247 and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O
5248 interrupt for the target subchannel has been dequeued and subchannel_id,
5249 subchannel_nr, io_int_parm and io_int_word contain the parameters for that
5250 interrupt. ipb is needed for instruction parameter decoding.
5259 On FSL BookE PowerPC chips, the interrupt controller has a fast patch
5260 interrupt acknowledge path to the core. When the core successfully
5261 delivers an interrupt, it automatically populates the EPR register with
5262 the interrupt vector number and acknowledges the interrupt inside
5263 the interrupt controller.
5265 In case the interrupt controller lives in user space, we need to do
5266 the interrupt acknowledge cycle through it to fetch the next to be
5267 delivered interrupt vector using this exit.
5269 It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an
5270 external interrupt has just been delivered into the guest. User space
5271 should put the acknowledged interrupt vector into the 'epr' field.
5275 /* KVM_EXIT_SYSTEM_EVENT */
5277 #define KVM_SYSTEM_EVENT_SHUTDOWN 1
5278 #define KVM_SYSTEM_EVENT_RESET 2
5279 #define KVM_SYSTEM_EVENT_CRASH 3
5284 If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered
5285 a system-level event using some architecture specific mechanism (hypercall
5286 or some special instruction). In case of ARM/ARM64, this is triggered using
5287 HVC instruction based PSCI call from the vcpu. The 'type' field describes
5288 the system-level event type. The 'flags' field describes architecture
5289 specific flags for the system-level event.
5291 Valid values for 'type' are:
5293 - KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the
5294 VM. Userspace is not obliged to honour this, and if it does honour
5295 this does not need to destroy the VM synchronously (ie it may call
5296 KVM_RUN again before shutdown finally occurs).
5297 - KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM.
5298 As with SHUTDOWN, userspace can choose to ignore the request, or
5299 to schedule the reset to occur in the future and may call KVM_RUN again.
5300 - KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest
5301 has requested a crash condition maintenance. Userspace can choose
5302 to ignore the request, or to gather VM memory core dump and/or
5303 reset/shutdown of the VM.
5307 /* KVM_EXIT_IOAPIC_EOI */
5312 Indicates that the VCPU's in-kernel local APIC received an EOI for a
5313 level-triggered IOAPIC interrupt. This exit only triggers when the
5314 IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled);
5315 the userspace IOAPIC should process the EOI and retrigger the interrupt if
5316 it is still asserted. Vector is the LAPIC interrupt vector for which the
5321 struct kvm_hyperv_exit {
5322 #define KVM_EXIT_HYPERV_SYNIC 1
5323 #define KVM_EXIT_HYPERV_HCALL 2
5324 #define KVM_EXIT_HYPERV_SYNDBG 3
5351 /* KVM_EXIT_HYPERV */
5352 struct kvm_hyperv_exit hyperv;
5354 Indicates that the VCPU exits into userspace to process some tasks
5355 related to Hyper-V emulation.
5357 Valid values for 'type' are:
5359 - KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about
5361 Hyper-V SynIC state change. Notification is used to remap SynIC
5362 event/message pages and to enable/disable SynIC messages/events processing
5365 - KVM_EXIT_HYPERV_SYNDBG -- synchronously notify user-space about
5367 Hyper-V Synthetic debugger state change. Notification is used to either update
5368 the pending_page location or to send a control command (send the buffer located
5369 in send_page or recv a buffer to recv_page).
5373 /* KVM_EXIT_ARM_NISV */
5379 Used on arm and arm64 systems. If a guest accesses memory not in a memslot,
5380 KVM will typically return to userspace and ask it to do MMIO emulation on its
5381 behalf. However, for certain classes of instructions, no instruction decode
5382 (direction, length of memory access) is provided, and fetching and decoding
5383 the instruction from the VM is overly complicated to live in the kernel.
5385 Historically, when this situation occurred, KVM would print a warning and kill
5386 the VM. KVM assumed that if the guest accessed non-memslot memory, it was
5387 trying to do I/O, which just couldn't be emulated, and the warning message was
5388 phrased accordingly. However, what happened more often was that a guest bug
5389 caused access outside the guest memory areas which should lead to a more
5390 meaningful warning message and an external abort in the guest, if the access
5391 did not fall within an I/O window.
5393 Userspace implementations can query for KVM_CAP_ARM_NISV_TO_USER, and enable
5394 this capability at VM creation. Once this is done, these types of errors will
5395 instead return to userspace with KVM_EXIT_ARM_NISV, with the valid bits from
5396 the HSR (arm) and ESR_EL2 (arm64) in the esr_iss field, and the faulting IPA
5397 in the fault_ipa field. Userspace can either fix up the access if it's
5398 actually an I/O access by decoding the instruction from guest memory (if it's
5399 very brave) and continue executing the guest, or it can decide to suspend,
5400 dump, or restart the guest.
5402 Note that KVM does not skip the faulting instruction as it does for
5403 KVM_EXIT_MMIO, but userspace has to emulate any change to the processing state
5404 if it decides to decode and emulate the instruction.
5408 /* KVM_EXIT_X86_RDMSR / KVM_EXIT_X86_WRMSR */
5410 __u8 error; /* user -> kernel */
5412 __u32 reason; /* kernel -> user */
5413 __u32 index; /* kernel -> user */
5414 __u64 data; /* kernel <-> user */
5417 Used on x86 systems. When the VM capability KVM_CAP_X86_USER_SPACE_MSR is
5418 enabled, MSR accesses to registers that would invoke a #GP by KVM kernel code
5419 will instead trigger a KVM_EXIT_X86_RDMSR exit for reads and KVM_EXIT_X86_WRMSR
5422 The "reason" field specifies why the MSR trap occurred. User space will only
5423 receive MSR exit traps when a particular reason was requested during through
5424 ENABLE_CAP. Currently valid exit reasons are:
5426 KVM_MSR_EXIT_REASON_UNKNOWN - access to MSR that is unknown to KVM
5427 KVM_MSR_EXIT_REASON_INVAL - access to invalid MSRs or reserved bits
5428 KVM_MSR_EXIT_REASON_FILTER - access blocked by KVM_X86_SET_MSR_FILTER
5430 For KVM_EXIT_X86_RDMSR, the "index" field tells user space which MSR the guest
5431 wants to read. To respond to this request with a successful read, user space
5432 writes the respective data into the "data" field and must continue guest
5433 execution to ensure the read data is transferred into guest register state.
5435 If the RDMSR request was unsuccessful, user space indicates that with a "1" in
5436 the "error" field. This will inject a #GP into the guest when the VCPU is
5439 For KVM_EXIT_X86_WRMSR, the "index" field tells user space which MSR the guest
5440 wants to write. Once finished processing the event, user space must continue
5441 vCPU execution. If the MSR write was unsuccessful, user space also sets the
5442 "error" field to "1".
5447 struct kvm_xen_exit {
5448 #define KVM_EXIT_XEN_HCALL 1
5461 struct kvm_hyperv_exit xen;
5463 Indicates that the VCPU exits into userspace to process some tasks
5464 related to Xen emulation.
5466 Valid values for 'type' are:
5468 - KVM_EXIT_XEN_HCALL -- synchronously notify user-space about Xen hypercall.
5469 Userspace is expected to place the hypercall result into the appropriate
5470 field before invoking KVM_RUN again.
5474 /* Fix the size of the union. */
5479 * shared registers between kvm and userspace.
5480 * kvm_valid_regs specifies the register classes set by the host
5481 * kvm_dirty_regs specified the register classes dirtied by userspace
5482 * struct kvm_sync_regs is architecture specific, as well as the
5483 * bits for kvm_valid_regs and kvm_dirty_regs
5485 __u64 kvm_valid_regs;
5486 __u64 kvm_dirty_regs;
5488 struct kvm_sync_regs regs;
5489 char padding[SYNC_REGS_SIZE_BYTES];
5492 If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access
5493 certain guest registers without having to call SET/GET_*REGS. Thus we can
5494 avoid some system call overhead if userspace has to handle the exit.
5495 Userspace can query the validity of the structure by checking
5496 kvm_valid_regs for specific bits. These bits are architecture specific
5497 and usually define the validity of a groups of registers. (e.g. one bit
5498 for general purpose registers)
5500 Please note that the kernel is allowed to use the kvm_run structure as the
5501 primary storage for certain register types. Therefore, the kernel may use the
5502 values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set.
5510 6. Capabilities that can be enabled on vCPUs
5511 ============================================
5513 There are certain capabilities that change the behavior of the virtual CPU or
5514 the virtual machine when enabled. To enable them, please see section 4.37.
5515 Below you can find a list of capabilities and what their effect on the vCPU or
5516 the virtual machine is when enabling them.
5518 The following information is provided along with the description:
5521 which instruction set architectures provide this ioctl.
5522 x86 includes both i386 and x86_64.
5525 whether this is a per-vcpu or per-vm capability.
5528 what parameters are accepted by the capability.
5531 the return value. General error numbers (EBADF, ENOMEM, EINVAL)
5532 are not detailed, but errors with specific meanings are.
5541 :Returns: 0 on success; -1 on error
5543 This capability enables interception of OSI hypercalls that otherwise would
5544 be treated as normal system calls to be injected into the guest. OSI hypercalls
5545 were invented by Mac-on-Linux to have a standardized communication mechanism
5546 between the guest and the host.
5548 When this capability is enabled, KVM_EXIT_OSI can occur.
5551 6.2 KVM_CAP_PPC_PAPR
5552 --------------------
5557 :Returns: 0 on success; -1 on error
5559 This capability enables interception of PAPR hypercalls. PAPR hypercalls are
5560 done using the hypercall instruction "sc 1".
5562 It also sets the guest privilege level to "supervisor" mode. Usually the guest
5563 runs in "hypervisor" privilege mode with a few missing features.
5565 In addition to the above, it changes the semantics of SDR1. In this mode, the
5566 HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the
5567 HTAB invisible to the guest.
5569 When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur.
5577 :Parameters: args[0] is the address of a struct kvm_config_tlb
5578 :Returns: 0 on success; -1 on error
5582 struct kvm_config_tlb {
5589 Configures the virtual CPU's TLB array, establishing a shared memory area
5590 between userspace and KVM. The "params" and "array" fields are userspace
5591 addresses of mmu-type-specific data structures. The "array_len" field is an
5592 safety mechanism, and should be set to the size in bytes of the memory that
5593 userspace has reserved for the array. It must be at least the size dictated
5594 by "mmu_type" and "params".
5596 While KVM_RUN is active, the shared region is under control of KVM. Its
5597 contents are undefined, and any modification by userspace results in
5598 boundedly undefined behavior.
5600 On return from KVM_RUN, the shared region will reflect the current state of
5601 the guest's TLB. If userspace makes any changes, it must call KVM_DIRTY_TLB
5602 to tell KVM which entries have been changed, prior to calling KVM_RUN again
5605 For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV:
5607 - The "params" field is of type "struct kvm_book3e_206_tlb_params".
5608 - The "array" field points to an array of type "struct
5609 kvm_book3e_206_tlb_entry".
5610 - The array consists of all entries in the first TLB, followed by all
5611 entries in the second TLB.
5612 - Within a TLB, entries are ordered first by increasing set number. Within a
5613 set, entries are ordered by way (increasing ESEL).
5614 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1)
5615 where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value.
5616 - The tsize field of mas1 shall be set to 4K on TLB0, even though the
5617 hardware ignores this value for TLB0.
5619 6.4 KVM_CAP_S390_CSS_SUPPORT
5620 ----------------------------
5622 :Architectures: s390
5625 :Returns: 0 on success; -1 on error
5627 This capability enables support for handling of channel I/O instructions.
5629 TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are
5630 handled in-kernel, while the other I/O instructions are passed to userspace.
5632 When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST
5633 SUBCHANNEL intercepts.
5635 Note that even though this capability is enabled per-vcpu, the complete
5636 virtual machine is affected.
5643 :Parameters: args[0] defines whether the proxy facility is active
5644 :Returns: 0 on success; -1 on error
5646 This capability enables or disables the delivery of interrupts through the
5647 external proxy facility.
5649 When enabled (args[0] != 0), every time the guest gets an external interrupt
5650 delivered, it automatically exits into user space with a KVM_EXIT_EPR exit
5651 to receive the topmost interrupt vector.
5653 When disabled (args[0] == 0), behavior is as if this facility is unsupported.
5655 When this capability is enabled, KVM_EXIT_EPR can occur.
5657 6.6 KVM_CAP_IRQ_MPIC
5658 --------------------
5661 :Parameters: args[0] is the MPIC device fd;
5662 args[1] is the MPIC CPU number for this vcpu
5664 This capability connects the vcpu to an in-kernel MPIC device.
5666 6.7 KVM_CAP_IRQ_XICS
5667 --------------------
5671 :Parameters: args[0] is the XICS device fd;
5672 args[1] is the XICS CPU number (server ID) for this vcpu
5674 This capability connects the vcpu to an in-kernel XICS device.
5676 6.8 KVM_CAP_S390_IRQCHIP
5677 ------------------------
5679 :Architectures: s390
5683 This capability enables the in-kernel irqchip for s390. Please refer to
5684 "4.24 KVM_CREATE_IRQCHIP" for details.
5686 6.9 KVM_CAP_MIPS_FPU
5687 --------------------
5689 :Architectures: mips
5691 :Parameters: args[0] is reserved for future use (should be 0).
5693 This capability allows the use of the host Floating Point Unit by the guest. It
5694 allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is
5695 done the ``KVM_REG_MIPS_FPR_*`` and ``KVM_REG_MIPS_FCR_*`` registers can be
5696 accessed (depending on the current guest FPU register mode), and the Status.FR,
5697 Config5.FRE bits are accessible via the KVM API and also from the guest,
5698 depending on them being supported by the FPU.
5700 6.10 KVM_CAP_MIPS_MSA
5701 ---------------------
5703 :Architectures: mips
5705 :Parameters: args[0] is reserved for future use (should be 0).
5707 This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest.
5708 It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest.
5709 Once this is done the ``KVM_REG_MIPS_VEC_*`` and ``KVM_REG_MIPS_MSA_*``
5710 registers can be accessed, and the Config5.MSAEn bit is accessible via the
5711 KVM API and also from the guest.
5713 6.74 KVM_CAP_SYNC_REGS
5714 ----------------------
5716 :Architectures: s390, x86
5717 :Target: s390: always enabled, x86: vcpu
5719 :Returns: x86: KVM_CHECK_EXTENSION returns a bit-array indicating which register
5721 (bitfields defined in arch/x86/include/uapi/asm/kvm.h).
5723 As described above in the kvm_sync_regs struct info in section 5 (kvm_run):
5724 KVM_CAP_SYNC_REGS "allow[s] userspace to access certain guest registers
5725 without having to call SET/GET_*REGS". This reduces overhead by eliminating
5726 repeated ioctl calls for setting and/or getting register values. This is
5727 particularly important when userspace is making synchronous guest state
5728 modifications, e.g. when emulating and/or intercepting instructions in
5731 For s390 specifics, please refer to the source code.
5735 - the register sets to be copied out to kvm_run are selectable
5736 by userspace (rather that all sets being copied out for every exit).
5737 - vcpu_events are available in addition to regs and sregs.
5739 For x86, the 'kvm_valid_regs' field of struct kvm_run is overloaded to
5740 function as an input bit-array field set by userspace to indicate the
5741 specific register sets to be copied out on the next exit.
5743 To indicate when userspace has modified values that should be copied into
5744 the vCPU, the all architecture bitarray field, 'kvm_dirty_regs' must be set.
5745 This is done using the same bitflags as for the 'kvm_valid_regs' field.
5746 If the dirty bit is not set, then the register set values will not be copied
5747 into the vCPU even if they've been modified.
5749 Unused bitfields in the bitarrays must be set to zero.
5753 struct kvm_sync_regs {
5754 struct kvm_regs regs;
5755 struct kvm_sregs sregs;
5756 struct kvm_vcpu_events events;
5759 6.75 KVM_CAP_PPC_IRQ_XIVE
5760 -------------------------
5764 :Parameters: args[0] is the XIVE device fd;
5765 args[1] is the XIVE CPU number (server ID) for this vcpu
5767 This capability connects the vcpu to an in-kernel XIVE device.
5769 7. Capabilities that can be enabled on VMs
5770 ==========================================
5772 There are certain capabilities that change the behavior of the virtual
5773 machine when enabled. To enable them, please see section 4.37. Below
5774 you can find a list of capabilities and what their effect on the VM
5775 is when enabling them.
5777 The following information is provided along with the description:
5780 which instruction set architectures provide this ioctl.
5781 x86 includes both i386 and x86_64.
5784 what parameters are accepted by the capability.
5787 the return value. General error numbers (EBADF, ENOMEM, EINVAL)
5788 are not detailed, but errors with specific meanings are.
5791 7.1 KVM_CAP_PPC_ENABLE_HCALL
5792 ----------------------------
5795 :Parameters: args[0] is the sPAPR hcall number;
5796 args[1] is 0 to disable, 1 to enable in-kernel handling
5798 This capability controls whether individual sPAPR hypercalls (hcalls)
5799 get handled by the kernel or not. Enabling or disabling in-kernel
5800 handling of an hcall is effective across the VM. On creation, an
5801 initial set of hcalls are enabled for in-kernel handling, which
5802 consists of those hcalls for which in-kernel handlers were implemented
5803 before this capability was implemented. If disabled, the kernel will
5804 not to attempt to handle the hcall, but will always exit to userspace
5805 to handle it. Note that it may not make sense to enable some and
5806 disable others of a group of related hcalls, but KVM does not prevent
5807 userspace from doing that.
5809 If the hcall number specified is not one that has an in-kernel
5810 implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL
5813 7.2 KVM_CAP_S390_USER_SIGP
5814 --------------------------
5816 :Architectures: s390
5819 This capability controls which SIGP orders will be handled completely in user
5820 space. With this capability enabled, all fast orders will be handled completely
5827 - CONDITIONAL EMERGENCY SIGNAL
5829 All other orders will be handled completely in user space.
5831 Only privileged operation exceptions will be checked for in the kernel (or even
5832 in the hardware prior to interception). If this capability is not enabled, the
5833 old way of handling SIGP orders is used (partially in kernel and user space).
5835 7.3 KVM_CAP_S390_VECTOR_REGISTERS
5836 ---------------------------------
5838 :Architectures: s390
5840 :Returns: 0 on success, negative value on error
5842 Allows use of the vector registers introduced with z13 processor, and
5843 provides for the synchronization between host and user space. Will
5844 return -EINVAL if the machine does not support vectors.
5846 7.4 KVM_CAP_S390_USER_STSI
5847 --------------------------
5849 :Architectures: s390
5852 This capability allows post-handlers for the STSI instruction. After
5853 initial handling in the kernel, KVM exits to user space with
5854 KVM_EXIT_S390_STSI to allow user space to insert further data.
5856 Before exiting to userspace, kvm handlers should fill in s390_stsi field of
5868 @addr - guest address of STSI SYSIB
5872 @ar - access register number
5874 KVM handlers should exit to userspace with rc = -EREMOTE.
5876 7.5 KVM_CAP_SPLIT_IRQCHIP
5877 -------------------------
5880 :Parameters: args[0] - number of routes reserved for userspace IOAPICs
5881 :Returns: 0 on success, -1 on error
5883 Create a local apic for each processor in the kernel. This can be used
5884 instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the
5885 IOAPIC and PIC (and also the PIT, even though this has to be enabled
5888 This capability also enables in kernel routing of interrupt requests;
5889 when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are
5890 used in the IRQ routing table. The first args[0] MSI routes are reserved
5891 for the IOAPIC pins. Whenever the LAPIC receives an EOI for these routes,
5892 a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace.
5894 Fails if VCPU has already been created, or if the irqchip is already in the
5895 kernel (i.e. KVM_CREATE_IRQCHIP has already been called).
5900 :Architectures: s390
5903 Allows use of runtime-instrumentation introduced with zEC12 processor.
5904 Will return -EINVAL if the machine does not support runtime-instrumentation.
5905 Will return -EBUSY if a VCPU has already been created.
5907 7.7 KVM_CAP_X2APIC_API
5908 ----------------------
5911 :Parameters: args[0] - features that should be enabled
5912 :Returns: 0 on success, -EINVAL when args[0] contains invalid features
5914 Valid feature flags in args[0] are::
5916 #define KVM_X2APIC_API_USE_32BIT_IDS (1ULL << 0)
5917 #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK (1ULL << 1)
5919 Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of
5920 KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC,
5921 allowing the use of 32-bit APIC IDs. See KVM_CAP_X2APIC_API in their
5922 respective sections.
5924 KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work
5925 in logical mode or with more than 255 VCPUs. Otherwise, KVM treats 0xff
5926 as a broadcast even in x2APIC mode in order to support physical x2APIC
5927 without interrupt remapping. This is undesirable in logical mode,
5928 where 0xff represents CPUs 0-7 in cluster 0.
5930 7.8 KVM_CAP_S390_USER_INSTR0
5931 ----------------------------
5933 :Architectures: s390
5936 With this capability enabled, all illegal instructions 0x0000 (2 bytes) will
5937 be intercepted and forwarded to user space. User space can use this
5938 mechanism e.g. to realize 2-byte software breakpoints. The kernel will
5939 not inject an operating exception for these instructions, user space has
5940 to take care of that.
5942 This capability can be enabled dynamically even if VCPUs were already
5943 created and are running.
5948 :Architectures: s390
5950 :Returns: 0 on success; -EINVAL if the machine does not support
5951 guarded storage; -EBUSY if a VCPU has already been created.
5953 Allows use of guarded storage for the KVM guest.
5955 7.10 KVM_CAP_S390_AIS
5956 ---------------------
5958 :Architectures: s390
5961 Allow use of adapter-interruption suppression.
5962 :Returns: 0 on success; -EBUSY if a VCPU has already been created.
5964 7.11 KVM_CAP_PPC_SMT
5965 --------------------
5968 :Parameters: vsmt_mode, flags
5970 Enabling this capability on a VM provides userspace with a way to set
5971 the desired virtual SMT mode (i.e. the number of virtual CPUs per
5972 virtual core). The virtual SMT mode, vsmt_mode, must be a power of 2
5973 between 1 and 8. On POWER8, vsmt_mode must also be no greater than
5974 the number of threads per subcore for the host. Currently flags must
5975 be 0. A successful call to enable this capability will result in
5976 vsmt_mode being returned when the KVM_CAP_PPC_SMT capability is
5977 subsequently queried for the VM. This capability is only supported by
5978 HV KVM, and can only be set before any VCPUs have been created.
5979 The KVM_CAP_PPC_SMT_POSSIBLE capability indicates which virtual SMT
5980 modes are available.
5982 7.12 KVM_CAP_PPC_FWNMI
5983 ----------------------
5988 With this capability a machine check exception in the guest address
5989 space will cause KVM to exit the guest with NMI exit reason. This
5990 enables QEMU to build error log and branch to guest kernel registered
5991 machine check handling routine. Without this capability KVM will
5992 branch to guests' 0x200 interrupt vector.
5994 7.13 KVM_CAP_X86_DISABLE_EXITS
5995 ------------------------------
5998 :Parameters: args[0] defines which exits are disabled
5999 :Returns: 0 on success, -EINVAL when args[0] contains invalid exits
6001 Valid bits in args[0] are::
6003 #define KVM_X86_DISABLE_EXITS_MWAIT (1 << 0)
6004 #define KVM_X86_DISABLE_EXITS_HLT (1 << 1)
6005 #define KVM_X86_DISABLE_EXITS_PAUSE (1 << 2)
6006 #define KVM_X86_DISABLE_EXITS_CSTATE (1 << 3)
6008 Enabling this capability on a VM provides userspace with a way to no
6009 longer intercept some instructions for improved latency in some
6010 workloads, and is suggested when vCPUs are associated to dedicated
6011 physical CPUs. More bits can be added in the future; userspace can
6012 just pass the KVM_CHECK_EXTENSION result to KVM_ENABLE_CAP to disable
6015 Do not enable KVM_FEATURE_PV_UNHALT if you disable HLT exits.
6017 7.14 KVM_CAP_S390_HPAGE_1M
6018 --------------------------
6020 :Architectures: s390
6022 :Returns: 0 on success, -EINVAL if hpage module parameter was not set
6023 or cmma is enabled, or the VM has the KVM_VM_S390_UCONTROL
6026 With this capability the KVM support for memory backing with 1m pages
6027 through hugetlbfs can be enabled for a VM. After the capability is
6028 enabled, cmma can't be enabled anymore and pfmfi and the storage key
6029 interpretation are disabled. If cmma has already been enabled or the
6030 hpage module parameter is not set to 1, -EINVAL is returned.
6032 While it is generally possible to create a huge page backed VM without
6033 this capability, the VM will not be able to run.
6035 7.15 KVM_CAP_MSR_PLATFORM_INFO
6036 ------------------------------
6039 :Parameters: args[0] whether feature should be enabled or not
6041 With this capability, a guest may read the MSR_PLATFORM_INFO MSR. Otherwise,
6042 a #GP would be raised when the guest tries to access. Currently, this
6043 capability does not enable write permissions of this MSR for the guest.
6045 7.16 KVM_CAP_PPC_NESTED_HV
6046 --------------------------
6050 :Returns: 0 on success, -EINVAL when the implementation doesn't support
6051 nested-HV virtualization.
6053 HV-KVM on POWER9 and later systems allows for "nested-HV"
6054 virtualization, which provides a way for a guest VM to run guests that
6055 can run using the CPU's supervisor mode (privileged non-hypervisor
6056 state). Enabling this capability on a VM depends on the CPU having
6057 the necessary functionality and on the facility being enabled with a
6058 kvm-hv module parameter.
6060 7.17 KVM_CAP_EXCEPTION_PAYLOAD
6061 ------------------------------
6064 :Parameters: args[0] whether feature should be enabled or not
6066 With this capability enabled, CR2 will not be modified prior to the
6067 emulated VM-exit when L1 intercepts a #PF exception that occurs in
6068 L2. Similarly, for kvm-intel only, DR6 will not be modified prior to
6069 the emulated VM-exit when L1 intercepts a #DB exception that occurs in
6070 L2. As a result, when KVM_GET_VCPU_EVENTS reports a pending #PF (or
6071 #DB) exception for L2, exception.has_payload will be set and the
6072 faulting address (or the new DR6 bits*) will be reported in the
6073 exception_payload field. Similarly, when userspace injects a #PF (or
6074 #DB) into L2 using KVM_SET_VCPU_EVENTS, it is expected to set
6075 exception.has_payload and to put the faulting address - or the new DR6
6076 bits\ [#]_ - in the exception_payload field.
6078 This capability also enables exception.pending in struct
6079 kvm_vcpu_events, which allows userspace to distinguish between pending
6080 and injected exceptions.
6083 .. [#] For the new DR6 bits, note that bit 16 is set iff the #DB exception
6086 7.18 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
6088 :Architectures: x86, arm, arm64, mips
6089 :Parameters: args[0] whether feature should be enabled or not
6093 #define KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (1 << 0)
6094 #define KVM_DIRTY_LOG_INITIALLY_SET (1 << 1)
6096 With KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE is set, KVM_GET_DIRTY_LOG will not
6097 automatically clear and write-protect all pages that are returned as dirty.
6098 Rather, userspace will have to do this operation separately using
6099 KVM_CLEAR_DIRTY_LOG.
6101 At the cost of a slightly more complicated operation, this provides better
6102 scalability and responsiveness for two reasons. First,
6103 KVM_CLEAR_DIRTY_LOG ioctl can operate on a 64-page granularity rather
6104 than requiring to sync a full memslot; this ensures that KVM does not
6105 take spinlocks for an extended period of time. Second, in some cases a
6106 large amount of time can pass between a call to KVM_GET_DIRTY_LOG and
6107 userspace actually using the data in the page. Pages can be modified
6108 during this time, which is inefficient for both the guest and userspace:
6109 the guest will incur a higher penalty due to write protection faults,
6110 while userspace can see false reports of dirty pages. Manual reprotection
6111 helps reducing this time, improving guest performance and reducing the
6112 number of dirty log false positives.
6114 With KVM_DIRTY_LOG_INITIALLY_SET set, all the bits of the dirty bitmap
6115 will be initialized to 1 when created. This also improves performance because
6116 dirty logging can be enabled gradually in small chunks on the first call
6117 to KVM_CLEAR_DIRTY_LOG. KVM_DIRTY_LOG_INITIALLY_SET depends on
6118 KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (it is also only available on
6119 x86 and arm64 for now).
6121 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 was previously available under the name
6122 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT, but the implementation had bugs that make
6123 it hard or impossible to use it correctly. The availability of
6124 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 signals that those bugs are fixed.
6125 Userspace should not try to use KVM_CAP_MANUAL_DIRTY_LOG_PROTECT.
6127 7.19 KVM_CAP_PPC_SECURE_GUEST
6128 ------------------------------
6132 This capability indicates that KVM is running on a host that has
6133 ultravisor firmware and thus can support a secure guest. On such a
6134 system, a guest can ask the ultravisor to make it a secure guest,
6135 one whose memory is inaccessible to the host except for pages which
6136 are explicitly requested to be shared with the host. The ultravisor
6137 notifies KVM when a guest requests to become a secure guest, and KVM
6138 has the opportunity to veto the transition.
6140 If present, this capability can be enabled for a VM, meaning that KVM
6141 will allow the transition to secure guest mode. Otherwise KVM will
6142 veto the transition.
6144 7.20 KVM_CAP_HALT_POLL
6145 ----------------------
6149 :Parameters: args[0] is the maximum poll time in nanoseconds
6150 :Returns: 0 on success; -1 on error
6152 This capability overrides the kvm module parameter halt_poll_ns for the
6155 VCPU polling allows a VCPU to poll for wakeup events instead of immediately
6156 scheduling during guest halts. The maximum time a VCPU can spend polling is
6157 controlled by the kvm module parameter halt_poll_ns. This capability allows
6158 the maximum halt time to specified on a per-VM basis, effectively overriding
6159 the module parameter for the target VM.
6161 7.21 KVM_CAP_X86_USER_SPACE_MSR
6162 -------------------------------
6166 :Parameters: args[0] contains the mask of KVM_MSR_EXIT_REASON_* events to report
6167 :Returns: 0 on success; -1 on error
6169 This capability enables trapping of #GP invoking RDMSR and WRMSR instructions
6172 When a guest requests to read or write an MSR, KVM may not implement all MSRs
6173 that are relevant to a respective system. It also does not differentiate by
6176 To allow more fine grained control over MSR handling, user space may enable
6177 this capability. With it enabled, MSR accesses that match the mask specified in
6178 args[0] and trigger a #GP event inside the guest by KVM will instead trigger
6179 KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR exit notifications which user space
6180 can then handle to implement model specific MSR handling and/or user notifications
6181 to inform a user that an MSR was not handled.
6183 7.22 KVM_CAP_X86_BUS_LOCK_EXIT
6184 -------------------------------
6188 :Parameters: args[0] defines the policy used when bus locks detected in guest
6189 :Returns: 0 on success, -EINVAL when args[0] contains invalid bits
6191 Valid bits in args[0] are::
6193 #define KVM_BUS_LOCK_DETECTION_OFF (1 << 0)
6194 #define KVM_BUS_LOCK_DETECTION_EXIT (1 << 1)
6196 Enabling this capability on a VM provides userspace with a way to select
6197 a policy to handle the bus locks detected in guest. Userspace can obtain
6198 the supported modes from the result of KVM_CHECK_EXTENSION and define it
6199 through the KVM_ENABLE_CAP.
6201 KVM_BUS_LOCK_DETECTION_OFF and KVM_BUS_LOCK_DETECTION_EXIT are supported
6202 currently and mutually exclusive with each other. More bits can be added in
6205 With KVM_BUS_LOCK_DETECTION_OFF set, bus locks in guest will not cause vm exits
6206 so that no additional actions are needed. This is the default mode.
6208 With KVM_BUS_LOCK_DETECTION_EXIT set, vm exits happen when bus lock detected
6209 in VM. KVM just exits to userspace when handling them. Userspace can enforce
6210 its own throttling or other policy based mitigations.
6212 This capability is aimed to address the thread that VM can exploit bus locks to
6213 degree the performance of the whole system. Once the userspace enable this
6214 capability and select the KVM_BUS_LOCK_DETECTION_EXIT mode, KVM will set the
6215 KVM_RUN_BUS_LOCK flag in vcpu-run->flags field and exit to userspace. Concerning
6216 the bus lock vm exit can be preempted by a higher priority VM exit, the exit
6217 notifications to userspace can be KVM_EXIT_BUS_LOCK or other reasons.
6218 KVM_RUN_BUS_LOCK flag is used to distinguish between them.
6220 7.23 KVM_CAP_PPC_DAWR1
6221 ----------------------
6225 :Returns: 0 on success, -EINVAL when CPU doesn't support 2nd DAWR
6227 This capability can be used to check / enable 2nd DAWR feature provided
6228 by POWER10 processor.
6230 8. Other capabilities.
6231 ======================
6233 This section lists capabilities that give information about other
6234 features of the KVM implementation.
6236 8.1 KVM_CAP_PPC_HWRNG
6237 ---------------------
6241 This capability, if KVM_CHECK_EXTENSION indicates that it is
6242 available, means that the kernel has an implementation of the
6243 H_RANDOM hypercall backed by a hardware random-number generator.
6244 If present, the kernel H_RANDOM handler can be enabled for guest use
6245 with the KVM_CAP_PPC_ENABLE_HCALL capability.
6247 8.2 KVM_CAP_HYPERV_SYNIC
6248 ------------------------
6252 This capability, if KVM_CHECK_EXTENSION indicates that it is
6253 available, means that the kernel has an implementation of the
6254 Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is
6255 used to support Windows Hyper-V based guest paravirt drivers(VMBus).
6257 In order to use SynIC, it has to be activated by setting this
6258 capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this
6259 will disable the use of APIC hardware virtualization even if supported
6260 by the CPU, as it's incompatible with SynIC auto-EOI behavior.
6262 8.3 KVM_CAP_PPC_RADIX_MMU
6263 -------------------------
6267 This capability, if KVM_CHECK_EXTENSION indicates that it is
6268 available, means that the kernel can support guests using the
6269 radix MMU defined in Power ISA V3.00 (as implemented in the POWER9
6272 8.4 KVM_CAP_PPC_HASH_MMU_V3
6273 ---------------------------
6277 This capability, if KVM_CHECK_EXTENSION indicates that it is
6278 available, means that the kernel can support guests using the
6279 hashed page table MMU defined in Power ISA V3.00 (as implemented in
6280 the POWER9 processor), including in-memory segment tables.
6285 :Architectures: mips
6287 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
6288 it is available, means that full hardware assisted virtualization capabilities
6289 of the hardware are available for use through KVM. An appropriate
6290 KVM_VM_MIPS_* type must be passed to KVM_CREATE_VM to create a VM which
6293 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
6294 available, it means that the VM is using full hardware assisted virtualization
6295 capabilities of the hardware. This is useful to check after creating a VM with
6296 KVM_VM_MIPS_DEFAULT.
6298 The value returned by KVM_CHECK_EXTENSION should be compared against known
6299 values (see below). All other values are reserved. This is to allow for the
6300 possibility of other hardware assisted virtualization implementations which
6301 may be incompatible with the MIPS VZ ASE.
6303 == ==========================================================================
6304 0 The trap & emulate implementation is in use to run guest code in user
6305 mode. Guest virtual memory segments are rearranged to fit the guest in the
6306 user mode address space.
6308 1 The MIPS VZ ASE is in use, providing full hardware assisted
6309 virtualization, including standard guest virtual memory segments.
6310 == ==========================================================================
6315 :Architectures: mips
6317 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
6318 it is available, means that the trap & emulate implementation is available to
6319 run guest code in user mode, even if KVM_CAP_MIPS_VZ indicates that hardware
6320 assisted virtualisation is also available. KVM_VM_MIPS_TE (0) must be passed
6321 to KVM_CREATE_VM to create a VM which utilises it.
6323 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
6324 available, it means that the VM is using trap & emulate.
6326 8.7 KVM_CAP_MIPS_64BIT
6327 ----------------------
6329 :Architectures: mips
6331 This capability indicates the supported architecture type of the guest, i.e. the
6332 supported register and address width.
6334 The values returned when this capability is checked by KVM_CHECK_EXTENSION on a
6335 kvm VM handle correspond roughly to the CP0_Config.AT register field, and should
6336 be checked specifically against known values (see below). All other values are
6339 == ========================================================================
6340 0 MIPS32 or microMIPS32.
6341 Both registers and addresses are 32-bits wide.
6342 It will only be possible to run 32-bit guest code.
6344 1 MIPS64 or microMIPS64 with access only to 32-bit compatibility segments.
6345 Registers are 64-bits wide, but addresses are 32-bits wide.
6346 64-bit guest code may run but cannot access MIPS64 memory segments.
6347 It will also be possible to run 32-bit guest code.
6349 2 MIPS64 or microMIPS64 with access to all address segments.
6350 Both registers and addresses are 64-bits wide.
6351 It will be possible to run 64-bit or 32-bit guest code.
6352 == ========================================================================
6354 8.9 KVM_CAP_ARM_USER_IRQ
6355 ------------------------
6357 :Architectures: arm, arm64
6359 This capability, if KVM_CHECK_EXTENSION indicates that it is available, means
6360 that if userspace creates a VM without an in-kernel interrupt controller, it
6361 will be notified of changes to the output level of in-kernel emulated devices,
6362 which can generate virtual interrupts, presented to the VM.
6363 For such VMs, on every return to userspace, the kernel
6364 updates the vcpu's run->s.regs.device_irq_level field to represent the actual
6365 output level of the device.
6367 Whenever kvm detects a change in the device output level, kvm guarantees at
6368 least one return to userspace before running the VM. This exit could either
6369 be a KVM_EXIT_INTR or any other exit event, like KVM_EXIT_MMIO. This way,
6370 userspace can always sample the device output level and re-compute the state of
6371 the userspace interrupt controller. Userspace should always check the state
6372 of run->s.regs.device_irq_level on every kvm exit.
6373 The value in run->s.regs.device_irq_level can represent both level and edge
6374 triggered interrupt signals, depending on the device. Edge triggered interrupt
6375 signals will exit to userspace with the bit in run->s.regs.device_irq_level
6376 set exactly once per edge signal.
6378 The field run->s.regs.device_irq_level is available independent of
6379 run->kvm_valid_regs or run->kvm_dirty_regs bits.
6381 If KVM_CAP_ARM_USER_IRQ is supported, the KVM_CHECK_EXTENSION ioctl returns a
6382 number larger than 0 indicating the version of this capability is implemented
6383 and thereby which bits in run->s.regs.device_irq_level can signal values.
6385 Currently the following bits are defined for the device_irq_level bitmap::
6387 KVM_CAP_ARM_USER_IRQ >= 1:
6389 KVM_ARM_DEV_EL1_VTIMER - EL1 virtual timer
6390 KVM_ARM_DEV_EL1_PTIMER - EL1 physical timer
6391 KVM_ARM_DEV_PMU - ARM PMU overflow interrupt signal
6393 Future versions of kvm may implement additional events. These will get
6394 indicated by returning a higher number from KVM_CHECK_EXTENSION and will be
6397 8.10 KVM_CAP_PPC_SMT_POSSIBLE
6398 -----------------------------
6402 Querying this capability returns a bitmap indicating the possible
6403 virtual SMT modes that can be set using KVM_CAP_PPC_SMT. If bit N
6404 (counting from the right) is set, then a virtual SMT mode of 2^N is
6407 8.11 KVM_CAP_HYPERV_SYNIC2
6408 --------------------------
6412 This capability enables a newer version of Hyper-V Synthetic interrupt
6413 controller (SynIC). The only difference with KVM_CAP_HYPERV_SYNIC is that KVM
6414 doesn't clear SynIC message and event flags pages when they are enabled by
6415 writing to the respective MSRs.
6417 8.12 KVM_CAP_HYPERV_VP_INDEX
6418 ----------------------------
6422 This capability indicates that userspace can load HV_X64_MSR_VP_INDEX msr. Its
6423 value is used to denote the target vcpu for a SynIC interrupt. For
6424 compatibilty, KVM initializes this msr to KVM's internal vcpu index. When this
6425 capability is absent, userspace can still query this msr's value.
6427 8.13 KVM_CAP_S390_AIS_MIGRATION
6428 -------------------------------
6430 :Architectures: s390
6433 This capability indicates if the flic device will be able to get/set the
6434 AIS states for migration via the KVM_DEV_FLIC_AISM_ALL attribute and allows
6435 to discover this without having to create a flic device.
6437 8.14 KVM_CAP_S390_PSW
6438 ---------------------
6440 :Architectures: s390
6442 This capability indicates that the PSW is exposed via the kvm_run structure.
6444 8.15 KVM_CAP_S390_GMAP
6445 ----------------------
6447 :Architectures: s390
6449 This capability indicates that the user space memory used as guest mapping can
6450 be anywhere in the user memory address space, as long as the memory slots are
6451 aligned and sized to a segment (1MB) boundary.
6453 8.16 KVM_CAP_S390_COW
6454 ---------------------
6456 :Architectures: s390
6458 This capability indicates that the user space memory used as guest mapping can
6459 use copy-on-write semantics as well as dirty pages tracking via read-only page
6462 8.17 KVM_CAP_S390_BPB
6463 ---------------------
6465 :Architectures: s390
6467 This capability indicates that kvm will implement the interfaces to handle
6468 reset, migration and nested KVM for branch prediction blocking. The stfle
6469 facility 82 should not be provided to the guest without this capability.
6471 8.18 KVM_CAP_HYPERV_TLBFLUSH
6472 ----------------------------
6476 This capability indicates that KVM supports paravirtualized Hyper-V TLB Flush
6478 HvFlushVirtualAddressSpace, HvFlushVirtualAddressSpaceEx,
6479 HvFlushVirtualAddressList, HvFlushVirtualAddressListEx.
6481 8.19 KVM_CAP_ARM_INJECT_SERROR_ESR
6482 ----------------------------------
6484 :Architectures: arm, arm64
6486 This capability indicates that userspace can specify (via the
6487 KVM_SET_VCPU_EVENTS ioctl) the syndrome value reported to the guest when it
6488 takes a virtual SError interrupt exception.
6489 If KVM advertises this capability, userspace can only specify the ISS field for
6490 the ESR syndrome. Other parts of the ESR, such as the EC are generated by the
6491 CPU when the exception is taken. If this virtual SError is taken to EL1 using
6492 AArch64, this value will be reported in the ISS field of ESR_ELx.
6494 See KVM_CAP_VCPU_EVENTS for more details.
6496 8.20 KVM_CAP_HYPERV_SEND_IPI
6497 ----------------------------
6501 This capability indicates that KVM supports paravirtualized Hyper-V IPI send
6503 HvCallSendSyntheticClusterIpi, HvCallSendSyntheticClusterIpiEx.
6505 8.21 KVM_CAP_HYPERV_DIRECT_TLBFLUSH
6506 -----------------------------------
6510 This capability indicates that KVM running on top of Hyper-V hypervisor
6511 enables Direct TLB flush for its guests meaning that TLB flush
6512 hypercalls are handled by Level 0 hypervisor (Hyper-V) bypassing KVM.
6513 Due to the different ABI for hypercall parameters between Hyper-V and
6514 KVM, enabling this capability effectively disables all hypercall
6515 handling by KVM (as some KVM hypercall may be mistakenly treated as TLB
6516 flush hypercalls by Hyper-V) so userspace should disable KVM identification
6517 in CPUID and only exposes Hyper-V identification. In this case, guest
6518 thinks it's running on Hyper-V and only use Hyper-V hypercalls.
6520 8.22 KVM_CAP_S390_VCPU_RESETS
6521 -----------------------------
6523 :Architectures: s390
6525 This capability indicates that the KVM_S390_NORMAL_RESET and
6526 KVM_S390_CLEAR_RESET ioctls are available.
6528 8.23 KVM_CAP_S390_PROTECTED
6529 ---------------------------
6531 :Architectures: s390
6533 This capability indicates that the Ultravisor has been initialized and
6534 KVM can therefore start protected VMs.
6535 This capability governs the KVM_S390_PV_COMMAND ioctl and the
6536 KVM_MP_STATE_LOAD MP_STATE. KVM_SET_MP_STATE can fail for protected
6537 guests when the state change is invalid.
6539 8.24 KVM_CAP_STEAL_TIME
6540 -----------------------
6542 :Architectures: arm64, x86
6544 This capability indicates that KVM supports steal time accounting.
6545 When steal time accounting is supported it may be enabled with
6546 architecture-specific interfaces. This capability and the architecture-
6547 specific interfaces must be consistent, i.e. if one says the feature
6548 is supported, than the other should as well and vice versa. For arm64
6549 see Documentation/virt/kvm/devices/vcpu.rst "KVM_ARM_VCPU_PVTIME_CTRL".
6550 For x86 see Documentation/virt/kvm/msr.rst "MSR_KVM_STEAL_TIME".
6552 8.25 KVM_CAP_S390_DIAG318
6553 -------------------------
6555 :Architectures: s390
6557 This capability enables a guest to set information about its control program
6558 (i.e. guest kernel type and version). The information is helpful during
6559 system/firmware service events, providing additional data about the guest
6560 environments running on the machine.
6562 The information is associated with the DIAGNOSE 0x318 instruction, which sets
6563 an 8-byte value consisting of a one-byte Control Program Name Code (CPNC) and
6564 a 7-byte Control Program Version Code (CPVC). The CPNC determines what
6565 environment the control program is running in (e.g. Linux, z/VM...), and the
6566 CPVC is used for information specific to OS (e.g. Linux version, Linux
6569 If this capability is available, then the CPNC and CPVC can be synchronized
6570 between KVM and userspace via the sync regs mechanism (KVM_SYNC_DIAG318).
6572 8.26 KVM_CAP_X86_USER_SPACE_MSR
6573 -------------------------------
6577 This capability indicates that KVM supports deflection of MSR reads and
6578 writes to user space. It can be enabled on a VM level. If enabled, MSR
6579 accesses that would usually trigger a #GP by KVM into the guest will
6580 instead get bounced to user space through the KVM_EXIT_X86_RDMSR and
6581 KVM_EXIT_X86_WRMSR exit notifications.
6583 8.27 KVM_X86_SET_MSR_FILTER
6584 ---------------------------
6588 This capability indicates that KVM supports that accesses to user defined MSRs
6589 may be rejected. With this capability exposed, KVM exports new VM ioctl
6590 KVM_X86_SET_MSR_FILTER which user space can call to specify bitmaps of MSR
6591 ranges that KVM should reject access to.
6593 In combination with KVM_CAP_X86_USER_SPACE_MSR, this allows user space to
6594 trap and emulate MSRs that are outside of the scope of KVM as well as
6595 limit the attack surface on KVM's MSR emulation code.
6597 8.28 KVM_CAP_ENFORCE_PV_CPUID
6598 -----------------------------
6602 When enabled, KVM will disable paravirtual features provided to the
6603 guest according to the bits in the KVM_CPUID_FEATURES CPUID leaf
6604 (0x40000001). Otherwise, a guest may use the paravirtual features
6605 regardless of what has actually been exposed through the CPUID leaf.
6607 8.29 KVM_CAP_DIRTY_LOG_RING
6608 ---------------------------
6611 :Parameters: args[0] - size of the dirty log ring
6613 KVM is capable of tracking dirty memory using ring buffers that are
6614 mmaped into userspace; there is one dirty ring per vcpu.
6616 The dirty ring is available to userspace as an array of
6617 ``struct kvm_dirty_gfn``. Each dirty entry it's defined as::
6619 struct kvm_dirty_gfn {
6621 __u32 slot; /* as_id | slot_id */
6625 The following values are defined for the flags field to define the
6626 current state of the entry::
6628 #define KVM_DIRTY_GFN_F_DIRTY BIT(0)
6629 #define KVM_DIRTY_GFN_F_RESET BIT(1)
6630 #define KVM_DIRTY_GFN_F_MASK 0x3
6632 Userspace should call KVM_ENABLE_CAP ioctl right after KVM_CREATE_VM
6633 ioctl to enable this capability for the new guest and set the size of
6634 the rings. Enabling the capability is only allowed before creating any
6635 vCPU, and the size of the ring must be a power of two. The larger the
6636 ring buffer, the less likely the ring is full and the VM is forced to
6637 exit to userspace. The optimal size depends on the workload, but it is
6638 recommended that it be at least 64 KiB (4096 entries).
6640 Just like for dirty page bitmaps, the buffer tracks writes to
6641 all user memory regions for which the KVM_MEM_LOG_DIRTY_PAGES flag was
6642 set in KVM_SET_USER_MEMORY_REGION. Once a memory region is registered
6643 with the flag set, userspace can start harvesting dirty pages from the
6646 An entry in the ring buffer can be unused (flag bits ``00``),
6647 dirty (flag bits ``01``) or harvested (flag bits ``1X``). The
6648 state machine for the entry is as follows::
6650 dirtied harvested reset
6651 00 -----------> 01 -------------> 1X -------+
6654 +------------------------------------------+
6656 To harvest the dirty pages, userspace accesses the mmaped ring buffer
6657 to read the dirty GFNs. If the flags has the DIRTY bit set (at this stage
6658 the RESET bit must be cleared), then it means this GFN is a dirty GFN.
6659 The userspace should harvest this GFN and mark the flags from state
6660 ``01b`` to ``1Xb`` (bit 0 will be ignored by KVM, but bit 1 must be set
6661 to show that this GFN is harvested and waiting for a reset), and move
6662 on to the next GFN. The userspace should continue to do this until the
6663 flags of a GFN have the DIRTY bit cleared, meaning that it has harvested
6664 all the dirty GFNs that were available.
6666 It's not necessary for userspace to harvest the all dirty GFNs at once.
6667 However it must collect the dirty GFNs in sequence, i.e., the userspace
6668 program cannot skip one dirty GFN to collect the one next to it.
6670 After processing one or more entries in the ring buffer, userspace
6671 calls the VM ioctl KVM_RESET_DIRTY_RINGS to notify the kernel about
6672 it, so that the kernel will reprotect those collected GFNs.
6673 Therefore, the ioctl must be called *before* reading the content of
6676 The dirty ring can get full. When it happens, the KVM_RUN of the
6677 vcpu will return with exit reason KVM_EXIT_DIRTY_LOG_FULL.
6679 The dirty ring interface has a major difference comparing to the
6680 KVM_GET_DIRTY_LOG interface in that, when reading the dirty ring from
6681 userspace, it's still possible that the kernel has not yet flushed the
6682 processor's dirty page buffers into the kernel buffer (with dirty bitmaps, the
6683 flushing is done by the KVM_GET_DIRTY_LOG ioctl). To achieve that, one
6684 needs to kick the vcpu out of KVM_RUN using a signal. The resulting
6685 vmexit ensures that all dirty GFNs are flushed to the dirty rings.
6687 NOTE: the capability KVM_CAP_DIRTY_LOG_RING and the corresponding
6688 ioctl KVM_RESET_DIRTY_RINGS are mutual exclusive to the existing ioctls
6689 KVM_GET_DIRTY_LOG and KVM_CLEAR_DIRTY_LOG. After enabling
6690 KVM_CAP_DIRTY_LOG_RING with an acceptable dirty ring size, the virtual
6691 machine will switch to ring-buffer dirty page tracking and further
6692 KVM_GET_DIRTY_LOG or KVM_CLEAR_DIRTY_LOG ioctls will fail.
6694 8.30 KVM_CAP_XEN_HVM
6695 --------------------
6699 This capability indicates the features that Xen supports for hosting Xen
6700 PVHVM guests. Valid flags are::
6702 #define KVM_XEN_HVM_CONFIG_HYPERCALL_MSR (1 << 0)
6703 #define KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL (1 << 1)
6704 #define KVM_XEN_HVM_CONFIG_SHARED_INFO (1 << 2)
6705 #define KVM_XEN_HVM_CONFIG_RUNSTATE (1 << 2)
6707 The KVM_XEN_HVM_CONFIG_HYPERCALL_MSR flag indicates that the KVM_XEN_HVM_CONFIG
6708 ioctl is available, for the guest to set its hypercall page.
6710 If KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL is also set, the same flag may also be
6711 provided in the flags to KVM_XEN_HVM_CONFIG, without providing hypercall page
6712 contents, to request that KVM generate hypercall page content automatically
6713 and also enable interception of guest hypercalls with KVM_EXIT_XEN.
6715 The KVM_XEN_HVM_CONFIG_SHARED_INFO flag indicates the availability of the
6716 KVM_XEN_HVM_SET_ATTR, KVM_XEN_HVM_GET_ATTR, KVM_XEN_VCPU_SET_ATTR and
6717 KVM_XEN_VCPU_GET_ATTR ioctls, as well as the delivery of exception vectors
6718 for event channel upcalls when the evtchn_upcall_pending field of a vcpu's
6721 The KVM_XEN_HVM_CONFIG_RUNSTATE flag indicates that the runstate-related
6722 features KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR/_CURRENT/_DATA/_ADJUST are
6723 supported by the KVM_XEN_VCPU_SET_ATTR/KVM_XEN_VCPU_GET_ATTR ioctls.