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 specifies
364 the address space for which you want to return the dirty bitmap.
365 They must be less than the value that KVM_CHECK_EXTENSION returns for
366 the KVM_CAP_MULTI_ADDRESS_SPACE capability.
368 The bits in the dirty bitmap are cleared before the ioctl returns, unless
369 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is enabled. For more information,
370 see the description of the capability.
372 4.9 KVM_SET_MEMORY_ALIAS
373 ------------------------
378 :Parameters: struct kvm_memory_alias (in)
379 :Returns: 0 (success), -1 (error)
381 This ioctl is obsolete and has been removed.
391 :Returns: 0 on success, -1 on error
395 ======= ==============================================================
396 EINTR an unmasked signal is pending
397 ENOEXEC the vcpu hasn't been initialized or the guest tried to execute
398 instructions from device memory (arm64)
399 ENOSYS data abort outside memslots with no syndrome info and
400 KVM_CAP_ARM_NISV_TO_USER not enabled (arm64)
401 EPERM SVE feature set but not finalized (arm64)
402 ======= ==============================================================
404 This ioctl is used to run a guest virtual cpu. While there are no
405 explicit parameters, there is an implicit parameter block that can be
406 obtained by mmap()ing the vcpu fd at offset 0, with the size given by
407 KVM_GET_VCPU_MMAP_SIZE. The parameter block is formatted as a 'struct
408 kvm_run' (see below).
415 :Architectures: all except ARM, arm64
417 :Parameters: struct kvm_regs (out)
418 :Returns: 0 on success, -1 on error
420 Reads the general purpose registers from the vcpu.
426 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
427 __u64 rax, rbx, rcx, rdx;
428 __u64 rsi, rdi, rsp, rbp;
429 __u64 r8, r9, r10, r11;
430 __u64 r12, r13, r14, r15;
436 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
448 :Architectures: all except ARM, arm64
450 :Parameters: struct kvm_regs (in)
451 :Returns: 0 on success, -1 on error
453 Writes the general purpose registers into the vcpu.
455 See KVM_GET_REGS for the data structure.
462 :Architectures: x86, ppc
464 :Parameters: struct kvm_sregs (out)
465 :Returns: 0 on success, -1 on error
467 Reads special registers from the vcpu.
473 struct kvm_segment cs, ds, es, fs, gs, ss;
474 struct kvm_segment tr, ldt;
475 struct kvm_dtable gdt, idt;
476 __u64 cr0, cr2, cr3, cr4, cr8;
479 __u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64];
482 /* ppc -- see arch/powerpc/include/uapi/asm/kvm.h */
484 interrupt_bitmap is a bitmap of pending external interrupts. At most
485 one bit may be set. This interrupt has been acknowledged by the APIC
486 but not yet injected into the cpu core.
493 :Architectures: x86, ppc
495 :Parameters: struct kvm_sregs (in)
496 :Returns: 0 on success, -1 on error
498 Writes special registers into the vcpu. See KVM_GET_SREGS for the
508 :Parameters: struct kvm_translation (in/out)
509 :Returns: 0 on success, -1 on error
511 Translates a virtual address according to the vcpu's current address
516 struct kvm_translation {
518 __u64 linear_address;
521 __u64 physical_address;
533 :Architectures: x86, ppc, mips
535 :Parameters: struct kvm_interrupt (in)
536 :Returns: 0 on success, negative on failure.
538 Queues a hardware interrupt vector to be injected.
542 /* for KVM_INTERRUPT */
543 struct kvm_interrupt {
553 ========= ===================================
555 -EEXIST if an interrupt is already enqueued
556 -EINVAL the irq number is invalid
557 -ENXIO if the PIC is in the kernel
558 -EFAULT if the pointer is invalid
559 ========= ===================================
561 Note 'irq' is an interrupt vector, not an interrupt pin or line. This
562 ioctl is useful if the in-kernel PIC is not used.
567 Queues an external interrupt to be injected. This ioctl is overleaded
568 with 3 different irq values:
572 This injects an edge type external interrupt into the guest once it's ready
573 to receive interrupts. When injected, the interrupt is done.
575 b) KVM_INTERRUPT_UNSET
577 This unsets any pending interrupt.
579 Only available with KVM_CAP_PPC_UNSET_IRQ.
581 c) KVM_INTERRUPT_SET_LEVEL
583 This injects a level type external interrupt into the guest context. The
584 interrupt stays pending until a specific ioctl with KVM_INTERRUPT_UNSET
587 Only available with KVM_CAP_PPC_IRQ_LEVEL.
589 Note that any value for 'irq' other than the ones stated above is invalid
590 and incurs unexpected behavior.
592 This is an asynchronous vcpu ioctl and can be invoked from any thread.
597 Queues an external interrupt to be injected into the virtual CPU. A negative
598 interrupt number dequeues the interrupt.
600 This is an asynchronous vcpu ioctl and can be invoked from any thread.
610 :Returns: -1 on error
612 Support for this has been removed. Use KVM_SET_GUEST_DEBUG instead.
618 :Capability: basic (vcpu), KVM_CAP_GET_MSR_FEATURES (system)
620 :Type: system ioctl, vcpu ioctl
621 :Parameters: struct kvm_msrs (in/out)
622 :Returns: number of msrs successfully returned;
625 When used as a system ioctl:
626 Reads the values of MSR-based features that are available for the VM. This
627 is similar to KVM_GET_SUPPORTED_CPUID, but it returns MSR indices and values.
628 The list of msr-based features can be obtained using KVM_GET_MSR_FEATURE_INDEX_LIST
631 When used as a vcpu ioctl:
632 Reads model-specific registers from the vcpu. Supported msr indices can
633 be obtained using KVM_GET_MSR_INDEX_LIST in a system ioctl.
638 __u32 nmsrs; /* number of msrs in entries */
641 struct kvm_msr_entry entries[0];
644 struct kvm_msr_entry {
650 Application code should set the 'nmsrs' member (which indicates the
651 size of the entries array) and the 'index' member of each array entry.
652 kvm will fill in the 'data' member.
661 :Parameters: struct kvm_msrs (in)
662 :Returns: number of msrs successfully set (see below), -1 on error
664 Writes model-specific registers to the vcpu. See KVM_GET_MSRS for the
667 Application code should set the 'nmsrs' member (which indicates the
668 size of the entries array), and the 'index' and 'data' members of each
671 It tries to set the MSRs in array entries[] one by one. If setting an MSR
672 fails, e.g., due to setting reserved bits, the MSR isn't supported/emulated
673 by KVM, etc..., it stops processing the MSR list and returns the number of
674 MSRs that have been set successfully.
683 :Parameters: struct kvm_cpuid (in)
684 :Returns: 0 on success, -1 on error
686 Defines the vcpu responses to the cpuid instruction. Applications
687 should use the KVM_SET_CPUID2 ioctl if available.
689 Note, when this IOCTL fails, KVM gives no guarantees that previous valid CPUID
690 configuration (if there is) is not corrupted. Userspace can get a copy of the
691 resulting CPUID configuration through KVM_GET_CPUID2 in case.
695 struct kvm_cpuid_entry {
704 /* for KVM_SET_CPUID */
708 struct kvm_cpuid_entry entries[0];
712 4.21 KVM_SET_SIGNAL_MASK
713 ------------------------
718 :Parameters: struct kvm_signal_mask (in)
719 :Returns: 0 on success, -1 on error
721 Defines which signals are blocked during execution of KVM_RUN. This
722 signal mask temporarily overrides the threads signal mask. Any
723 unblocked signal received (except SIGKILL and SIGSTOP, which retain
724 their traditional behaviour) will cause KVM_RUN to return with -EINTR.
726 Note the signal will only be delivered if not blocked by the original
731 /* for KVM_SET_SIGNAL_MASK */
732 struct kvm_signal_mask {
744 :Parameters: struct kvm_fpu (out)
745 :Returns: 0 on success, -1 on error
747 Reads the floating point state from the vcpu.
751 /* for KVM_GET_FPU and KVM_SET_FPU */
756 __u8 ftwx; /* in fxsave format */
773 :Parameters: struct kvm_fpu (in)
774 :Returns: 0 on success, -1 on error
776 Writes the floating point state to the vcpu.
780 /* for KVM_GET_FPU and KVM_SET_FPU */
785 __u8 ftwx; /* in fxsave format */
796 4.24 KVM_CREATE_IRQCHIP
797 -----------------------
799 :Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQCHIP (s390)
800 :Architectures: x86, ARM, arm64, s390
803 :Returns: 0 on success, -1 on error
805 Creates an interrupt controller model in the kernel.
806 On x86, creates a virtual ioapic, a virtual PIC (two PICs, nested), and sets up
807 future vcpus to have a local APIC. IRQ routing for GSIs 0-15 is set to both
808 PIC and IOAPIC; GSI 16-23 only go to the IOAPIC.
809 On ARM/arm64, a GICv2 is created. Any other GIC versions require the usage of
810 KVM_CREATE_DEVICE, which also supports creating a GICv2. Using
811 KVM_CREATE_DEVICE is preferred over KVM_CREATE_IRQCHIP for GICv2.
812 On s390, a dummy irq routing table is created.
814 Note that on s390 the KVM_CAP_S390_IRQCHIP vm capability needs to be enabled
815 before KVM_CREATE_IRQCHIP can be used.
821 :Capability: KVM_CAP_IRQCHIP
822 :Architectures: x86, arm, arm64
824 :Parameters: struct kvm_irq_level
825 :Returns: 0 on success, -1 on error
827 Sets the level of a GSI input to the interrupt controller model in the kernel.
828 On some architectures it is required that an interrupt controller model has
829 been previously created with KVM_CREATE_IRQCHIP. Note that edge-triggered
830 interrupts require the level to be set to 1 and then back to 0.
832 On real hardware, interrupt pins can be active-low or active-high. This
833 does not matter for the level field of struct kvm_irq_level: 1 always
834 means active (asserted), 0 means inactive (deasserted).
836 x86 allows the operating system to program the interrupt polarity
837 (active-low/active-high) for level-triggered interrupts, and KVM used
838 to consider the polarity. However, due to bitrot in the handling of
839 active-low interrupts, the above convention is now valid on x86 too.
840 This is signaled by KVM_CAP_X86_IOAPIC_POLARITY_IGNORED. Userspace
841 should not present interrupts to the guest as active-low unless this
842 capability is present (or unless it is not using the in-kernel irqchip,
846 ARM/arm64 can signal an interrupt either at the CPU level, or at the
847 in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to
848 use PPIs designated for specific cpus. The irq field is interpreted
851 Â bits: | 31 ... 28 | 27 ... 24 | 23 ... 16 | 15 ... 0 |
852 field: | vcpu2_index | irq_type | vcpu_index | irq_id |
854 The irq_type field has the following values:
857 out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ
859 in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.)
860 (the vcpu_index field is ignored)
862 in-kernel GIC: PPI, irq_id between 16 and 31 (incl.)
864 (The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs)
866 In both cases, level is used to assert/deassert the line.
868 When KVM_CAP_ARM_IRQ_LINE_LAYOUT_2 is supported, the target vcpu is
869 identified as (256 * vcpu2_index + vcpu_index). Otherwise, vcpu2_index
872 Note that on arm/arm64, the KVM_CAP_IRQCHIP capability only conditions
873 injection of interrupts for the in-kernel irqchip. KVM_IRQ_LINE can always
874 be used for a userspace interrupt controller.
878 struct kvm_irq_level {
881 __s32 status; /* not used for KVM_IRQ_LEVEL */
883 __u32 level; /* 0 or 1 */
890 :Capability: KVM_CAP_IRQCHIP
893 :Parameters: struct kvm_irqchip (in/out)
894 :Returns: 0 on success, -1 on error
896 Reads the state of a kernel interrupt controller created with
897 KVM_CREATE_IRQCHIP into a buffer provided by the caller.
902 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
905 char dummy[512]; /* reserving space */
906 struct kvm_pic_state pic;
907 struct kvm_ioapic_state ioapic;
915 :Capability: KVM_CAP_IRQCHIP
918 :Parameters: struct kvm_irqchip (in)
919 :Returns: 0 on success, -1 on error
921 Sets the state of a kernel interrupt controller created with
922 KVM_CREATE_IRQCHIP from a buffer provided by the caller.
927 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
930 char dummy[512]; /* reserving space */
931 struct kvm_pic_state pic;
932 struct kvm_ioapic_state ioapic;
937 4.28 KVM_XEN_HVM_CONFIG
938 -----------------------
940 :Capability: KVM_CAP_XEN_HVM
943 :Parameters: struct kvm_xen_hvm_config (in)
944 :Returns: 0 on success, -1 on error
946 Sets the MSR that the Xen HVM guest uses to initialize its hypercall
947 page, and provides the starting address and size of the hypercall
948 blobs in userspace. When the guest writes the MSR, kvm copies one
949 page of a blob (32- or 64-bit, depending on the vcpu mode) to guest
954 struct kvm_xen_hvm_config {
968 :Capability: KVM_CAP_ADJUST_CLOCK
971 :Parameters: struct kvm_clock_data (out)
972 :Returns: 0 on success, -1 on error
974 Gets the current timestamp of kvmclock as seen by the current guest. In
975 conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios
978 When KVM_CAP_ADJUST_CLOCK is passed to KVM_CHECK_EXTENSION, it returns the
979 set of bits that KVM can return in struct kvm_clock_data's flag member.
981 The only flag defined now is KVM_CLOCK_TSC_STABLE. If set, the returned
982 value is the exact kvmclock value seen by all VCPUs at the instant
983 when KVM_GET_CLOCK was called. If clear, the returned value is simply
984 CLOCK_MONOTONIC plus a constant offset; the offset can be modified
985 with KVM_SET_CLOCK. KVM will try to make all VCPUs follow this clock,
986 but the exact value read by each VCPU could differ, because the host
991 struct kvm_clock_data {
992 __u64 clock; /* kvmclock current value */
1001 :Capability: KVM_CAP_ADJUST_CLOCK
1004 :Parameters: struct kvm_clock_data (in)
1005 :Returns: 0 on success, -1 on error
1007 Sets the current timestamp of kvmclock to the value specified in its parameter.
1008 In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios
1013 struct kvm_clock_data {
1014 __u64 clock; /* kvmclock current value */
1020 4.31 KVM_GET_VCPU_EVENTS
1021 ------------------------
1023 :Capability: KVM_CAP_VCPU_EVENTS
1024 :Extended by: KVM_CAP_INTR_SHADOW
1025 :Architectures: x86, arm, arm64
1027 :Parameters: struct kvm_vcpu_event (out)
1028 :Returns: 0 on success, -1 on error
1033 Gets currently pending exceptions, interrupts, and NMIs as well as related
1038 struct kvm_vcpu_events {
1042 __u8 has_error_code;
1063 __u8 smm_inside_nmi;
1067 __u8 exception_has_payload;
1068 __u64 exception_payload;
1071 The following bits are defined in the flags field:
1073 - KVM_VCPUEVENT_VALID_SHADOW may be set to signal that
1074 interrupt.shadow contains a valid state.
1076 - KVM_VCPUEVENT_VALID_SMM may be set to signal that smi contains a
1079 - KVM_VCPUEVENT_VALID_PAYLOAD may be set to signal that the
1080 exception_has_payload, exception_payload, and exception.pending
1081 fields contain a valid state. This bit will be set whenever
1082 KVM_CAP_EXCEPTION_PAYLOAD is enabled.
1087 If the guest accesses a device that is being emulated by the host kernel in
1088 such a way that a real device would generate a physical SError, KVM may make
1089 a virtual SError pending for that VCPU. This system error interrupt remains
1090 pending until the guest takes the exception by unmasking PSTATE.A.
1092 Running the VCPU may cause it to take a pending SError, or make an access that
1093 causes an SError to become pending. The event's description is only valid while
1094 the VPCU is not running.
1096 This API provides a way to read and write the pending 'event' state that is not
1097 visible to the guest. To save, restore or migrate a VCPU the struct representing
1098 the state can be read then written using this GET/SET API, along with the other
1099 guest-visible registers. It is not possible to 'cancel' an SError that has been
1102 A device being emulated in user-space may also wish to generate an SError. To do
1103 this the events structure can be populated by user-space. The current state
1104 should be read first, to ensure no existing SError is pending. If an existing
1105 SError is pending, the architecture's 'Multiple SError interrupts' rules should
1106 be followed. (2.5.3 of DDI0587.a "ARM Reliability, Availability, and
1107 Serviceability (RAS) Specification").
1109 SError exceptions always have an ESR value. Some CPUs have the ability to
1110 specify what the virtual SError's ESR value should be. These systems will
1111 advertise KVM_CAP_ARM_INJECT_SERROR_ESR. In this case exception.has_esr will
1112 always have a non-zero value when read, and the agent making an SError pending
1113 should specify the ISS field in the lower 24 bits of exception.serror_esr. If
1114 the system supports KVM_CAP_ARM_INJECT_SERROR_ESR, but user-space sets the events
1115 with exception.has_esr as zero, KVM will choose an ESR.
1117 Specifying exception.has_esr on a system that does not support it will return
1118 -EINVAL. Setting anything other than the lower 24bits of exception.serror_esr
1119 will return -EINVAL.
1121 It is not possible to read back a pending external abort (injected via
1122 KVM_SET_VCPU_EVENTS or otherwise) because such an exception is always delivered
1123 directly to the virtual CPU).
1127 struct kvm_vcpu_events {
1129 __u8 serror_pending;
1130 __u8 serror_has_esr;
1131 __u8 ext_dabt_pending;
1132 /* Align it to 8 bytes */
1139 4.32 KVM_SET_VCPU_EVENTS
1140 ------------------------
1142 :Capability: KVM_CAP_VCPU_EVENTS
1143 :Extended by: KVM_CAP_INTR_SHADOW
1144 :Architectures: x86, arm, arm64
1146 :Parameters: struct kvm_vcpu_event (in)
1147 :Returns: 0 on success, -1 on error
1152 Set pending exceptions, interrupts, and NMIs as well as related states of the
1155 See KVM_GET_VCPU_EVENTS for the data structure.
1157 Fields that may be modified asynchronously by running VCPUs can be excluded
1158 from the update. These fields are nmi.pending, sipi_vector, smi.smm,
1159 smi.pending. Keep the corresponding bits in the flags field cleared to
1160 suppress overwriting the current in-kernel state. The bits are:
1162 =============================== ==================================
1163 KVM_VCPUEVENT_VALID_NMI_PENDING transfer nmi.pending to the kernel
1164 KVM_VCPUEVENT_VALID_SIPI_VECTOR transfer sipi_vector
1165 KVM_VCPUEVENT_VALID_SMM transfer the smi sub-struct.
1166 =============================== ==================================
1168 If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in
1169 the flags field to signal that interrupt.shadow contains a valid state and
1170 shall be written into the VCPU.
1172 KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available.
1174 If KVM_CAP_EXCEPTION_PAYLOAD is enabled, KVM_VCPUEVENT_VALID_PAYLOAD
1175 can be set in the flags field to signal that the
1176 exception_has_payload, exception_payload, and exception.pending fields
1177 contain a valid state and shall be written into the VCPU.
1182 User space may need to inject several types of events to the guest.
1184 Set the pending SError exception state for this VCPU. It is not possible to
1185 'cancel' an Serror that has been made pending.
1187 If the guest performed an access to I/O memory which could not be handled by
1188 userspace, for example because of missing instruction syndrome decode
1189 information or because there is no device mapped at the accessed IPA, then
1190 userspace can ask the kernel to inject an external abort using the address
1191 from the exiting fault on the VCPU. It is a programming error to set
1192 ext_dabt_pending after an exit which was not either KVM_EXIT_MMIO or
1193 KVM_EXIT_ARM_NISV. This feature is only available if the system supports
1194 KVM_CAP_ARM_INJECT_EXT_DABT. This is a helper which provides commonality in
1195 how userspace reports accesses for the above cases to guests, across different
1196 userspace implementations. Nevertheless, userspace can still emulate all Arm
1197 exceptions by manipulating individual registers using the KVM_SET_ONE_REG API.
1199 See KVM_GET_VCPU_EVENTS for the data structure.
1202 4.33 KVM_GET_DEBUGREGS
1203 ----------------------
1205 :Capability: KVM_CAP_DEBUGREGS
1208 :Parameters: struct kvm_debugregs (out)
1209 :Returns: 0 on success, -1 on error
1211 Reads debug registers from the vcpu.
1215 struct kvm_debugregs {
1224 4.34 KVM_SET_DEBUGREGS
1225 ----------------------
1227 :Capability: KVM_CAP_DEBUGREGS
1230 :Parameters: struct kvm_debugregs (in)
1231 :Returns: 0 on success, -1 on error
1233 Writes debug registers into the vcpu.
1235 See KVM_GET_DEBUGREGS for the data structure. The flags field is unused
1236 yet and must be cleared on entry.
1239 4.35 KVM_SET_USER_MEMORY_REGION
1240 -------------------------------
1242 :Capability: KVM_CAP_USER_MEMORY
1245 :Parameters: struct kvm_userspace_memory_region (in)
1246 :Returns: 0 on success, -1 on error
1250 struct kvm_userspace_memory_region {
1253 __u64 guest_phys_addr;
1254 __u64 memory_size; /* bytes */
1255 __u64 userspace_addr; /* start of the userspace allocated memory */
1258 /* for kvm_memory_region::flags */
1259 #define KVM_MEM_LOG_DIRTY_PAGES (1UL << 0)
1260 #define KVM_MEM_READONLY (1UL << 1)
1262 This ioctl allows the user to create, modify or delete a guest physical
1263 memory slot. Bits 0-15 of "slot" specify the slot id and this value
1264 should be less than the maximum number of user memory slots supported per
1265 VM. The maximum allowed slots can be queried using KVM_CAP_NR_MEMSLOTS.
1266 Slots may not overlap in guest physical address space.
1268 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot"
1269 specifies the address space which is being modified. They must be
1270 less than the value that KVM_CHECK_EXTENSION returns for the
1271 KVM_CAP_MULTI_ADDRESS_SPACE capability. Slots in separate address spaces
1272 are unrelated; the restriction on overlapping slots only applies within
1275 Deleting a slot is done by passing zero for memory_size. When changing
1276 an existing slot, it may be moved in the guest physical memory space,
1277 or its flags may be modified, but it may not be resized.
1279 Memory for the region is taken starting at the address denoted by the
1280 field userspace_addr, which must point at user addressable memory for
1281 the entire memory slot size. Any object may back this memory, including
1282 anonymous memory, ordinary files, and hugetlbfs.
1284 On architectures that support a form of address tagging, userspace_addr must
1285 be an untagged address.
1287 It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr
1288 be identical. This allows large pages in the guest to be backed by large
1291 The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and
1292 KVM_MEM_READONLY. The former can be set to instruct KVM to keep track of
1293 writes to memory within the slot. See KVM_GET_DIRTY_LOG ioctl to know how to
1294 use it. The latter can be set, if KVM_CAP_READONLY_MEM capability allows it,
1295 to make a new slot read-only. In this case, writes to this memory will be
1296 posted to userspace as KVM_EXIT_MMIO exits.
1298 When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of
1299 the memory region are automatically reflected into the guest. For example, an
1300 mmap() that affects the region will be made visible immediately. Another
1301 example is madvise(MADV_DROP).
1303 It is recommended to use this API instead of the KVM_SET_MEMORY_REGION ioctl.
1304 The KVM_SET_MEMORY_REGION does not allow fine grained control over memory
1305 allocation and is deprecated.
1308 4.36 KVM_SET_TSS_ADDR
1309 ---------------------
1311 :Capability: KVM_CAP_SET_TSS_ADDR
1314 :Parameters: unsigned long tss_address (in)
1315 :Returns: 0 on success, -1 on error
1317 This ioctl defines the physical address of a three-page region in the guest
1318 physical address space. The region must be within the first 4GB of the
1319 guest physical address space and must not conflict with any memory slot
1320 or any mmio address. The guest may malfunction if it accesses this memory
1323 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1324 because of a quirk in the virtualization implementation (see the internals
1325 documentation when it pops into existence).
1331 :Capability: KVM_CAP_ENABLE_CAP
1332 :Architectures: mips, ppc, s390
1334 :Parameters: struct kvm_enable_cap (in)
1335 :Returns: 0 on success; -1 on error
1337 :Capability: KVM_CAP_ENABLE_CAP_VM
1340 :Parameters: struct kvm_enable_cap (in)
1341 :Returns: 0 on success; -1 on error
1345 Not all extensions are enabled by default. Using this ioctl the application
1346 can enable an extension, making it available to the guest.
1348 On systems that do not support this ioctl, it always fails. On systems that
1349 do support it, it only works for extensions that are supported for enablement.
1351 To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should
1356 struct kvm_enable_cap {
1360 The capability that is supposed to get enabled.
1366 A bitfield indicating future enhancements. Has to be 0 for now.
1372 Arguments for enabling a feature. If a feature needs initial values to
1373 function properly, this is the place to put them.
1380 The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl
1381 for vm-wide capabilities.
1383 4.38 KVM_GET_MP_STATE
1384 ---------------------
1386 :Capability: KVM_CAP_MP_STATE
1387 :Architectures: x86, s390, arm, arm64
1389 :Parameters: struct kvm_mp_state (out)
1390 :Returns: 0 on success; -1 on error
1394 struct kvm_mp_state {
1398 Returns the vcpu's current "multiprocessing state" (though also valid on
1399 uniprocessor guests).
1401 Possible values are:
1403 ========================== ===============================================
1404 KVM_MP_STATE_RUNNABLE the vcpu is currently running [x86,arm/arm64]
1405 KVM_MP_STATE_UNINITIALIZED the vcpu is an application processor (AP)
1406 which has not yet received an INIT signal [x86]
1407 KVM_MP_STATE_INIT_RECEIVED the vcpu has received an INIT signal, and is
1408 now ready for a SIPI [x86]
1409 KVM_MP_STATE_HALTED the vcpu has executed a HLT instruction and
1410 is waiting for an interrupt [x86]
1411 KVM_MP_STATE_SIPI_RECEIVED the vcpu has just received a SIPI (vector
1412 accessible via KVM_GET_VCPU_EVENTS) [x86]
1413 KVM_MP_STATE_STOPPED the vcpu is stopped [s390,arm/arm64]
1414 KVM_MP_STATE_CHECK_STOP the vcpu is in a special error state [s390]
1415 KVM_MP_STATE_OPERATING the vcpu is operating (running or halted)
1417 KVM_MP_STATE_LOAD the vcpu is in a special load/startup state
1419 ========================== ===============================================
1421 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1422 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1423 these architectures.
1428 The only states that are valid are KVM_MP_STATE_STOPPED and
1429 KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not.
1431 4.39 KVM_SET_MP_STATE
1432 ---------------------
1434 :Capability: KVM_CAP_MP_STATE
1435 :Architectures: x86, s390, arm, arm64
1437 :Parameters: struct kvm_mp_state (in)
1438 :Returns: 0 on success; -1 on error
1440 Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for
1443 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1444 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1445 these architectures.
1450 The only states that are valid are KVM_MP_STATE_STOPPED and
1451 KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not.
1453 4.40 KVM_SET_IDENTITY_MAP_ADDR
1454 ------------------------------
1456 :Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR
1459 :Parameters: unsigned long identity (in)
1460 :Returns: 0 on success, -1 on error
1462 This ioctl defines the physical address of a one-page region in the guest
1463 physical address space. The region must be within the first 4GB of the
1464 guest physical address space and must not conflict with any memory slot
1465 or any mmio address. The guest may malfunction if it accesses this memory
1468 Setting the address to 0 will result in resetting the address to its default
1471 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1472 because of a quirk in the virtualization implementation (see the internals
1473 documentation when it pops into existence).
1475 Fails if any VCPU has already been created.
1477 4.41 KVM_SET_BOOT_CPU_ID
1478 ------------------------
1480 :Capability: KVM_CAP_SET_BOOT_CPU_ID
1483 :Parameters: unsigned long vcpu_id
1484 :Returns: 0 on success, -1 on error
1486 Define which vcpu is the Bootstrap Processor (BSP). Values are the same
1487 as the vcpu id in KVM_CREATE_VCPU. If this ioctl is not called, the default
1494 :Capability: KVM_CAP_XSAVE
1497 :Parameters: struct kvm_xsave (out)
1498 :Returns: 0 on success, -1 on error
1507 This ioctl would copy current vcpu's xsave struct to the userspace.
1513 :Capability: KVM_CAP_XSAVE
1516 :Parameters: struct kvm_xsave (in)
1517 :Returns: 0 on success, -1 on error
1526 This ioctl would copy userspace's xsave struct to the kernel.
1532 :Capability: KVM_CAP_XCRS
1535 :Parameters: struct kvm_xcrs (out)
1536 :Returns: 0 on success, -1 on error
1549 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1553 This ioctl would copy current vcpu's xcrs to the userspace.
1559 :Capability: KVM_CAP_XCRS
1562 :Parameters: struct kvm_xcrs (in)
1563 :Returns: 0 on success, -1 on error
1576 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1580 This ioctl would set vcpu's xcr to the value userspace specified.
1583 4.46 KVM_GET_SUPPORTED_CPUID
1584 ----------------------------
1586 :Capability: KVM_CAP_EXT_CPUID
1589 :Parameters: struct kvm_cpuid2 (in/out)
1590 :Returns: 0 on success, -1 on error
1597 struct kvm_cpuid_entry2 entries[0];
1600 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
1601 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */
1602 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */
1604 struct kvm_cpuid_entry2 {
1615 This ioctl returns x86 cpuid features which are supported by both the
1616 hardware and kvm in its default configuration. Userspace can use the
1617 information returned by this ioctl to construct cpuid information (for
1618 KVM_SET_CPUID2) that is consistent with hardware, kernel, and
1619 userspace capabilities, and with user requirements (for example, the
1620 user may wish to constrain cpuid to emulate older hardware, or for
1621 feature consistency across a cluster).
1623 Note that certain capabilities, such as KVM_CAP_X86_DISABLE_EXITS, may
1624 expose cpuid features (e.g. MONITOR) which are not supported by kvm in
1625 its default configuration. If userspace enables such capabilities, it
1626 is responsible for modifying the results of this ioctl appropriately.
1628 Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure
1629 with the 'nent' field indicating the number of entries in the variable-size
1630 array 'entries'. If the number of entries is too low to describe the cpu
1631 capabilities, an error (E2BIG) is returned. If the number is too high,
1632 the 'nent' field is adjusted and an error (ENOMEM) is returned. If the
1633 number is just right, the 'nent' field is adjusted to the number of valid
1634 entries in the 'entries' array, which is then filled.
1636 The entries returned are the host cpuid as returned by the cpuid instruction,
1637 with unknown or unsupported features masked out. Some features (for example,
1638 x2apic), may not be present in the host cpu, but are exposed by kvm if it can
1639 emulate them efficiently. The fields in each entry are defined as follows:
1642 the eax value used to obtain the entry
1645 the ecx value used to obtain the entry (for entries that are
1649 an OR of zero or more of the following:
1651 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
1652 if the index field is valid
1655 the values returned by the cpuid instruction for
1656 this function/index combination
1658 The TSC deadline timer feature (CPUID leaf 1, ecx[24]) is always returned
1659 as false, since the feature depends on KVM_CREATE_IRQCHIP for local APIC
1660 support. Instead it is reported via::
1662 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER)
1664 if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the
1665 feature in userspace, then you can enable the feature for KVM_SET_CPUID2.
1668 4.47 KVM_PPC_GET_PVINFO
1669 -----------------------
1671 :Capability: KVM_CAP_PPC_GET_PVINFO
1674 :Parameters: struct kvm_ppc_pvinfo (out)
1675 :Returns: 0 on success, !0 on error
1679 struct kvm_ppc_pvinfo {
1685 This ioctl fetches PV specific information that need to be passed to the guest
1686 using the device tree or other means from vm context.
1688 The hcall array defines 4 instructions that make up a hypercall.
1690 If any additional field gets added to this structure later on, a bit for that
1691 additional piece of information will be set in the flags bitmap.
1693 The flags bitmap is defined as::
1695 /* the host supports the ePAPR idle hcall
1696 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1<<0)
1698 4.52 KVM_SET_GSI_ROUTING
1699 ------------------------
1701 :Capability: KVM_CAP_IRQ_ROUTING
1702 :Architectures: x86 s390 arm arm64
1704 :Parameters: struct kvm_irq_routing (in)
1705 :Returns: 0 on success, -1 on error
1707 Sets the GSI routing table entries, overwriting any previously set entries.
1709 On arm/arm64, GSI routing has the following limitation:
1711 - GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD.
1715 struct kvm_irq_routing {
1718 struct kvm_irq_routing_entry entries[0];
1721 No flags are specified so far, the corresponding field must be set to zero.
1725 struct kvm_irq_routing_entry {
1731 struct kvm_irq_routing_irqchip irqchip;
1732 struct kvm_irq_routing_msi msi;
1733 struct kvm_irq_routing_s390_adapter adapter;
1734 struct kvm_irq_routing_hv_sint hv_sint;
1739 /* gsi routing entry types */
1740 #define KVM_IRQ_ROUTING_IRQCHIP 1
1741 #define KVM_IRQ_ROUTING_MSI 2
1742 #define KVM_IRQ_ROUTING_S390_ADAPTER 3
1743 #define KVM_IRQ_ROUTING_HV_SINT 4
1747 - KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry
1748 type, specifies that the devid field contains a valid value. The per-VM
1749 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
1750 the device ID. If this capability is not available, userspace should
1751 never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
1756 struct kvm_irq_routing_irqchip {
1761 struct kvm_irq_routing_msi {
1771 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
1772 for the device that wrote the MSI message. For PCI, this is usually a
1773 BFD identifier in the lower 16 bits.
1775 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
1776 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
1777 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
1778 address_hi must be zero.
1782 struct kvm_irq_routing_s390_adapter {
1786 __u32 summary_offset;
1790 struct kvm_irq_routing_hv_sint {
1796 4.55 KVM_SET_TSC_KHZ
1797 --------------------
1799 :Capability: KVM_CAP_TSC_CONTROL
1802 :Parameters: virtual tsc_khz
1803 :Returns: 0 on success, -1 on error
1805 Specifies the tsc frequency for the virtual machine. The unit of the
1809 4.56 KVM_GET_TSC_KHZ
1810 --------------------
1812 :Capability: KVM_CAP_GET_TSC_KHZ
1816 :Returns: virtual tsc-khz on success, negative value on error
1818 Returns the tsc frequency of the guest. The unit of the return value is
1819 KHz. If the host has unstable tsc this ioctl returns -EIO instead as an
1826 :Capability: KVM_CAP_IRQCHIP
1829 :Parameters: struct kvm_lapic_state (out)
1830 :Returns: 0 on success, -1 on error
1834 #define KVM_APIC_REG_SIZE 0x400
1835 struct kvm_lapic_state {
1836 char regs[KVM_APIC_REG_SIZE];
1839 Reads the Local APIC registers and copies them into the input argument. The
1840 data format and layout are the same as documented in the architecture manual.
1842 If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is
1843 enabled, then the format of APIC_ID register depends on the APIC mode
1844 (reported by MSR_IA32_APICBASE) of its VCPU. x2APIC stores APIC ID in
1845 the APIC_ID register (bytes 32-35). xAPIC only allows an 8-bit APIC ID
1846 which is stored in bits 31-24 of the APIC register, or equivalently in
1847 byte 35 of struct kvm_lapic_state's regs field. KVM_GET_LAPIC must then
1848 be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR.
1850 If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state
1851 always uses xAPIC format.
1857 :Capability: KVM_CAP_IRQCHIP
1860 :Parameters: struct kvm_lapic_state (in)
1861 :Returns: 0 on success, -1 on error
1865 #define KVM_APIC_REG_SIZE 0x400
1866 struct kvm_lapic_state {
1867 char regs[KVM_APIC_REG_SIZE];
1870 Copies the input argument into the Local APIC registers. The data format
1871 and layout are the same as documented in the architecture manual.
1873 The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's
1874 regs field) depends on the state of the KVM_CAP_X2APIC_API capability.
1875 See the note in KVM_GET_LAPIC.
1881 :Capability: KVM_CAP_IOEVENTFD
1884 :Parameters: struct kvm_ioeventfd (in)
1885 :Returns: 0 on success, !0 on error
1887 This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address
1888 within the guest. A guest write in the registered address will signal the
1889 provided event instead of triggering an exit.
1893 struct kvm_ioeventfd {
1895 __u64 addr; /* legal pio/mmio address */
1896 __u32 len; /* 0, 1, 2, 4, or 8 bytes */
1902 For the special case of virtio-ccw devices on s390, the ioevent is matched
1903 to a subchannel/virtqueue tuple instead.
1905 The following flags are defined::
1907 #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch)
1908 #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio)
1909 #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign)
1910 #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \
1911 (1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify)
1913 If datamatch flag is set, the event will be signaled only if the written value
1914 to the registered address is equal to datamatch in struct kvm_ioeventfd.
1916 For virtio-ccw devices, addr contains the subchannel id and datamatch the
1919 With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and
1920 the kernel will ignore the length of guest write and may get a faster vmexit.
1921 The speedup may only apply to specific architectures, but the ioeventfd will
1927 :Capability: KVM_CAP_SW_TLB
1930 :Parameters: struct kvm_dirty_tlb (in)
1931 :Returns: 0 on success, -1 on error
1935 struct kvm_dirty_tlb {
1940 This must be called whenever userspace has changed an entry in the shared
1941 TLB, prior to calling KVM_RUN on the associated vcpu.
1943 The "bitmap" field is the userspace address of an array. This array
1944 consists of a number of bits, equal to the total number of TLB entries as
1945 determined by the last successful call to KVM_CONFIG_TLB, rounded up to the
1946 nearest multiple of 64.
1948 Each bit corresponds to one TLB entry, ordered the same as in the shared TLB
1951 The array is little-endian: the bit 0 is the least significant bit of the
1952 first byte, bit 8 is the least significant bit of the second byte, etc.
1953 This avoids any complications with differing word sizes.
1955 The "num_dirty" field is a performance hint for KVM to determine whether it
1956 should skip processing the bitmap and just invalidate everything. It must
1957 be set to the number of set bits in the bitmap.
1960 4.62 KVM_CREATE_SPAPR_TCE
1961 -------------------------
1963 :Capability: KVM_CAP_SPAPR_TCE
1964 :Architectures: powerpc
1966 :Parameters: struct kvm_create_spapr_tce (in)
1967 :Returns: file descriptor for manipulating the created TCE table
1969 This creates a virtual TCE (translation control entry) table, which
1970 is an IOMMU for PAPR-style virtual I/O. It is used to translate
1971 logical addresses used in virtual I/O into guest physical addresses,
1972 and provides a scatter/gather capability for PAPR virtual I/O.
1976 /* for KVM_CAP_SPAPR_TCE */
1977 struct kvm_create_spapr_tce {
1982 The liobn field gives the logical IO bus number for which to create a
1983 TCE table. The window_size field specifies the size of the DMA window
1984 which this TCE table will translate - the table will contain one 64
1985 bit TCE entry for every 4kiB of the DMA window.
1987 When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE
1988 table has been created using this ioctl(), the kernel will handle it
1989 in real mode, updating the TCE table. H_PUT_TCE calls for other
1990 liobns will cause a vm exit and must be handled by userspace.
1992 The return value is a file descriptor which can be passed to mmap(2)
1993 to map the created TCE table into userspace. This lets userspace read
1994 the entries written by kernel-handled H_PUT_TCE calls, and also lets
1995 userspace update the TCE table directly which is useful in some
1999 4.63 KVM_ALLOCATE_RMA
2000 ---------------------
2002 :Capability: KVM_CAP_PPC_RMA
2003 :Architectures: powerpc
2005 :Parameters: struct kvm_allocate_rma (out)
2006 :Returns: file descriptor for mapping the allocated RMA
2008 This allocates a Real Mode Area (RMA) from the pool allocated at boot
2009 time by the kernel. An RMA is a physically-contiguous, aligned region
2010 of memory used on older POWER processors to provide the memory which
2011 will be accessed by real-mode (MMU off) accesses in a KVM guest.
2012 POWER processors support a set of sizes for the RMA that usually
2013 includes 64MB, 128MB, 256MB and some larger powers of two.
2017 /* for KVM_ALLOCATE_RMA */
2018 struct kvm_allocate_rma {
2022 The return value is a file descriptor which can be passed to mmap(2)
2023 to map the allocated RMA into userspace. The mapped area can then be
2024 passed to the KVM_SET_USER_MEMORY_REGION ioctl to establish it as the
2025 RMA for a virtual machine. The size of the RMA in bytes (which is
2026 fixed at host kernel boot time) is returned in the rma_size field of
2027 the argument structure.
2029 The KVM_CAP_PPC_RMA capability is 1 or 2 if the KVM_ALLOCATE_RMA ioctl
2030 is supported; 2 if the processor requires all virtual machines to have
2031 an RMA, or 1 if the processor can use an RMA but doesn't require it,
2032 because it supports the Virtual RMA (VRMA) facility.
2038 :Capability: KVM_CAP_USER_NMI
2042 :Returns: 0 on success, -1 on error
2044 Queues an NMI on the thread's vcpu. Note this is well defined only
2045 when KVM_CREATE_IRQCHIP has not been called, since this is an interface
2046 between the virtual cpu core and virtual local APIC. After KVM_CREATE_IRQCHIP
2047 has been called, this interface is completely emulated within the kernel.
2049 To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the
2050 following algorithm:
2053 - read the local APIC's state (KVM_GET_LAPIC)
2054 - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1)
2055 - if so, issue KVM_NMI
2058 Some guests configure the LINT1 NMI input to cause a panic, aiding in
2062 4.65 KVM_S390_UCAS_MAP
2063 ----------------------
2065 :Capability: KVM_CAP_S390_UCONTROL
2066 :Architectures: s390
2068 :Parameters: struct kvm_s390_ucas_mapping (in)
2069 :Returns: 0 in case of success
2071 The parameter is defined like this::
2073 struct kvm_s390_ucas_mapping {
2079 This ioctl maps the memory at "user_addr" with the length "length" to
2080 the vcpu's address space starting at "vcpu_addr". All parameters need to
2081 be aligned by 1 megabyte.
2084 4.66 KVM_S390_UCAS_UNMAP
2085 ------------------------
2087 :Capability: KVM_CAP_S390_UCONTROL
2088 :Architectures: s390
2090 :Parameters: struct kvm_s390_ucas_mapping (in)
2091 :Returns: 0 in case of success
2093 The parameter is defined like this::
2095 struct kvm_s390_ucas_mapping {
2101 This ioctl unmaps the memory in the vcpu's address space starting at
2102 "vcpu_addr" with the length "length". The field "user_addr" is ignored.
2103 All parameters need to be aligned by 1 megabyte.
2106 4.67 KVM_S390_VCPU_FAULT
2107 ------------------------
2109 :Capability: KVM_CAP_S390_UCONTROL
2110 :Architectures: s390
2112 :Parameters: vcpu absolute address (in)
2113 :Returns: 0 in case of success
2115 This call creates a page table entry on the virtual cpu's address space
2116 (for user controlled virtual machines) or the virtual machine's address
2117 space (for regular virtual machines). This only works for minor faults,
2118 thus it's recommended to access subject memory page via the user page
2119 table upfront. This is useful to handle validity intercepts for user
2120 controlled virtual machines to fault in the virtual cpu's lowcore pages
2121 prior to calling the KVM_RUN ioctl.
2124 4.68 KVM_SET_ONE_REG
2125 --------------------
2127 :Capability: KVM_CAP_ONE_REG
2130 :Parameters: struct kvm_one_reg (in)
2131 :Returns: 0 on success, negative value on failure
2135 ====== ============================================================
2136 Â ENOENT Â Â no such register
2137 Â EINVAL Â Â invalid register ID, or no such register or used with VMs in
2138 protected virtualization mode on s390
2139 Â EPERM Â Â Â (arm64) register access not allowed before vcpu finalization
2140 ====== ============================================================
2142 (These error codes are indicative only: do not rely on a specific error
2143 code being returned in a specific situation.)
2147 struct kvm_one_reg {
2152 Using this ioctl, a single vcpu register can be set to a specific value
2153 defined by user space with the passed in struct kvm_one_reg, where id
2154 refers to the register identifier as described below and addr is a pointer
2155 to a variable with the respective size. There can be architecture agnostic
2156 and architecture specific registers. Each have their own range of operation
2157 and their own constants and width. To keep track of the implemented
2158 registers, find a list below:
2160 ======= =============================== ============
2161 Arch Register Width (bits)
2162 ======= =============================== ============
2163 PPC KVM_REG_PPC_HIOR 64
2164 PPC KVM_REG_PPC_IAC1 64
2165 PPC KVM_REG_PPC_IAC2 64
2166 PPC KVM_REG_PPC_IAC3 64
2167 PPC KVM_REG_PPC_IAC4 64
2168 PPC KVM_REG_PPC_DAC1 64
2169 PPC KVM_REG_PPC_DAC2 64
2170 PPC KVM_REG_PPC_DABR 64
2171 PPC KVM_REG_PPC_DSCR 64
2172 PPC KVM_REG_PPC_PURR 64
2173 PPC KVM_REG_PPC_SPURR 64
2174 PPC KVM_REG_PPC_DAR 64
2175 PPC KVM_REG_PPC_DSISR 32
2176 PPC KVM_REG_PPC_AMR 64
2177 PPC KVM_REG_PPC_UAMOR 64
2178 PPC KVM_REG_PPC_MMCR0 64
2179 PPC KVM_REG_PPC_MMCR1 64
2180 PPC KVM_REG_PPC_MMCRA 64
2181 PPC KVM_REG_PPC_MMCR2 64
2182 PPC KVM_REG_PPC_MMCRS 64
2183 PPC KVM_REG_PPC_MMCR3 64
2184 PPC KVM_REG_PPC_SIAR 64
2185 PPC KVM_REG_PPC_SDAR 64
2186 PPC KVM_REG_PPC_SIER 64
2187 PPC KVM_REG_PPC_SIER2 64
2188 PPC KVM_REG_PPC_SIER3 64
2189 PPC KVM_REG_PPC_PMC1 32
2190 PPC KVM_REG_PPC_PMC2 32
2191 PPC KVM_REG_PPC_PMC3 32
2192 PPC KVM_REG_PPC_PMC4 32
2193 PPC KVM_REG_PPC_PMC5 32
2194 PPC KVM_REG_PPC_PMC6 32
2195 PPC KVM_REG_PPC_PMC7 32
2196 PPC KVM_REG_PPC_PMC8 32
2197 PPC KVM_REG_PPC_FPR0 64
2199 PPC KVM_REG_PPC_FPR31 64
2200 PPC KVM_REG_PPC_VR0 128
2202 PPC KVM_REG_PPC_VR31 128
2203 PPC KVM_REG_PPC_VSR0 128
2205 PPC KVM_REG_PPC_VSR31 128
2206 PPC KVM_REG_PPC_FPSCR 64
2207 PPC KVM_REG_PPC_VSCR 32
2208 PPC KVM_REG_PPC_VPA_ADDR 64
2209 PPC KVM_REG_PPC_VPA_SLB 128
2210 PPC KVM_REG_PPC_VPA_DTL 128
2211 PPC KVM_REG_PPC_EPCR 32
2212 PPC KVM_REG_PPC_EPR 32
2213 PPC KVM_REG_PPC_TCR 32
2214 PPC KVM_REG_PPC_TSR 32
2215 PPC KVM_REG_PPC_OR_TSR 32
2216 PPC KVM_REG_PPC_CLEAR_TSR 32
2217 PPC KVM_REG_PPC_MAS0 32
2218 PPC KVM_REG_PPC_MAS1 32
2219 PPC KVM_REG_PPC_MAS2 64
2220 PPC KVM_REG_PPC_MAS7_3 64
2221 PPC KVM_REG_PPC_MAS4 32
2222 PPC KVM_REG_PPC_MAS6 32
2223 PPC KVM_REG_PPC_MMUCFG 32
2224 PPC KVM_REG_PPC_TLB0CFG 32
2225 PPC KVM_REG_PPC_TLB1CFG 32
2226 PPC KVM_REG_PPC_TLB2CFG 32
2227 PPC KVM_REG_PPC_TLB3CFG 32
2228 PPC KVM_REG_PPC_TLB0PS 32
2229 PPC KVM_REG_PPC_TLB1PS 32
2230 PPC KVM_REG_PPC_TLB2PS 32
2231 PPC KVM_REG_PPC_TLB3PS 32
2232 PPC KVM_REG_PPC_EPTCFG 32
2233 PPC KVM_REG_PPC_ICP_STATE 64
2234 PPC KVM_REG_PPC_VP_STATE 128
2235 PPC KVM_REG_PPC_TB_OFFSET 64
2236 PPC KVM_REG_PPC_SPMC1 32
2237 PPC KVM_REG_PPC_SPMC2 32
2238 PPC KVM_REG_PPC_IAMR 64
2239 PPC KVM_REG_PPC_TFHAR 64
2240 PPC KVM_REG_PPC_TFIAR 64
2241 PPC KVM_REG_PPC_TEXASR 64
2242 PPC KVM_REG_PPC_FSCR 64
2243 PPC KVM_REG_PPC_PSPB 32
2244 PPC KVM_REG_PPC_EBBHR 64
2245 PPC KVM_REG_PPC_EBBRR 64
2246 PPC KVM_REG_PPC_BESCR 64
2247 PPC KVM_REG_PPC_TAR 64
2248 PPC KVM_REG_PPC_DPDES 64
2249 PPC KVM_REG_PPC_DAWR 64
2250 PPC KVM_REG_PPC_DAWRX 64
2251 PPC KVM_REG_PPC_CIABR 64
2252 PPC KVM_REG_PPC_IC 64
2253 PPC KVM_REG_PPC_VTB 64
2254 PPC KVM_REG_PPC_CSIGR 64
2255 PPC KVM_REG_PPC_TACR 64
2256 PPC KVM_REG_PPC_TCSCR 64
2257 PPC KVM_REG_PPC_PID 64
2258 PPC KVM_REG_PPC_ACOP 64
2259 PPC KVM_REG_PPC_VRSAVE 32
2260 PPC KVM_REG_PPC_LPCR 32
2261 PPC KVM_REG_PPC_LPCR_64 64
2262 PPC KVM_REG_PPC_PPR 64
2263 PPC KVM_REG_PPC_ARCH_COMPAT 32
2264 PPC KVM_REG_PPC_DABRX 32
2265 PPC KVM_REG_PPC_WORT 64
2266 PPC KVM_REG_PPC_SPRG9 64
2267 PPC KVM_REG_PPC_DBSR 32
2268 PPC KVM_REG_PPC_TIDR 64
2269 PPC KVM_REG_PPC_PSSCR 64
2270 PPC KVM_REG_PPC_DEC_EXPIRY 64
2271 PPC KVM_REG_PPC_PTCR 64
2272 PPC KVM_REG_PPC_TM_GPR0 64
2274 PPC KVM_REG_PPC_TM_GPR31 64
2275 PPC KVM_REG_PPC_TM_VSR0 128
2277 PPC KVM_REG_PPC_TM_VSR63 128
2278 PPC KVM_REG_PPC_TM_CR 64
2279 PPC KVM_REG_PPC_TM_LR 64
2280 PPC KVM_REG_PPC_TM_CTR 64
2281 PPC KVM_REG_PPC_TM_FPSCR 64
2282 PPC KVM_REG_PPC_TM_AMR 64
2283 PPC KVM_REG_PPC_TM_PPR 64
2284 PPC KVM_REG_PPC_TM_VRSAVE 64
2285 PPC KVM_REG_PPC_TM_VSCR 32
2286 PPC KVM_REG_PPC_TM_DSCR 64
2287 PPC KVM_REG_PPC_TM_TAR 64
2288 PPC KVM_REG_PPC_TM_XER 64
2290 MIPS KVM_REG_MIPS_R0 64
2292 MIPS KVM_REG_MIPS_R31 64
2293 MIPS KVM_REG_MIPS_HI 64
2294 MIPS KVM_REG_MIPS_LO 64
2295 MIPS KVM_REG_MIPS_PC 64
2296 MIPS KVM_REG_MIPS_CP0_INDEX 32
2297 MIPS KVM_REG_MIPS_CP0_ENTRYLO0 64
2298 MIPS KVM_REG_MIPS_CP0_ENTRYLO1 64
2299 MIPS KVM_REG_MIPS_CP0_CONTEXT 64
2300 MIPS KVM_REG_MIPS_CP0_CONTEXTCONFIG 32
2301 MIPS KVM_REG_MIPS_CP0_USERLOCAL 64
2302 MIPS KVM_REG_MIPS_CP0_XCONTEXTCONFIG 64
2303 MIPS KVM_REG_MIPS_CP0_PAGEMASK 32
2304 MIPS KVM_REG_MIPS_CP0_PAGEGRAIN 32
2305 MIPS KVM_REG_MIPS_CP0_SEGCTL0 64
2306 MIPS KVM_REG_MIPS_CP0_SEGCTL1 64
2307 MIPS KVM_REG_MIPS_CP0_SEGCTL2 64
2308 MIPS KVM_REG_MIPS_CP0_PWBASE 64
2309 MIPS KVM_REG_MIPS_CP0_PWFIELD 64
2310 MIPS KVM_REG_MIPS_CP0_PWSIZE 64
2311 MIPS KVM_REG_MIPS_CP0_WIRED 32
2312 MIPS KVM_REG_MIPS_CP0_PWCTL 32
2313 MIPS KVM_REG_MIPS_CP0_HWRENA 32
2314 MIPS KVM_REG_MIPS_CP0_BADVADDR 64
2315 MIPS KVM_REG_MIPS_CP0_BADINSTR 32
2316 MIPS KVM_REG_MIPS_CP0_BADINSTRP 32
2317 MIPS KVM_REG_MIPS_CP0_COUNT 32
2318 MIPS KVM_REG_MIPS_CP0_ENTRYHI 64
2319 MIPS KVM_REG_MIPS_CP0_COMPARE 32
2320 MIPS KVM_REG_MIPS_CP0_STATUS 32
2321 MIPS KVM_REG_MIPS_CP0_INTCTL 32
2322 MIPS KVM_REG_MIPS_CP0_CAUSE 32
2323 MIPS KVM_REG_MIPS_CP0_EPC 64
2324 MIPS KVM_REG_MIPS_CP0_PRID 32
2325 MIPS KVM_REG_MIPS_CP0_EBASE 64
2326 MIPS KVM_REG_MIPS_CP0_CONFIG 32
2327 MIPS KVM_REG_MIPS_CP0_CONFIG1 32
2328 MIPS KVM_REG_MIPS_CP0_CONFIG2 32
2329 MIPS KVM_REG_MIPS_CP0_CONFIG3 32
2330 MIPS KVM_REG_MIPS_CP0_CONFIG4 32
2331 MIPS KVM_REG_MIPS_CP0_CONFIG5 32
2332 MIPS KVM_REG_MIPS_CP0_CONFIG7 32
2333 MIPS KVM_REG_MIPS_CP0_XCONTEXT 64
2334 MIPS KVM_REG_MIPS_CP0_ERROREPC 64
2335 MIPS KVM_REG_MIPS_CP0_KSCRATCH1 64
2336 MIPS KVM_REG_MIPS_CP0_KSCRATCH2 64
2337 MIPS KVM_REG_MIPS_CP0_KSCRATCH3 64
2338 MIPS KVM_REG_MIPS_CP0_KSCRATCH4 64
2339 MIPS KVM_REG_MIPS_CP0_KSCRATCH5 64
2340 MIPS KVM_REG_MIPS_CP0_KSCRATCH6 64
2341 MIPS KVM_REG_MIPS_CP0_MAAR(0..63) 64
2342 MIPS KVM_REG_MIPS_COUNT_CTL 64
2343 MIPS KVM_REG_MIPS_COUNT_RESUME 64
2344 MIPS KVM_REG_MIPS_COUNT_HZ 64
2345 MIPS KVM_REG_MIPS_FPR_32(0..31) 32
2346 MIPS KVM_REG_MIPS_FPR_64(0..31) 64
2347 MIPS KVM_REG_MIPS_VEC_128(0..31) 128
2348 MIPS KVM_REG_MIPS_FCR_IR 32
2349 MIPS KVM_REG_MIPS_FCR_CSR 32
2350 MIPS KVM_REG_MIPS_MSA_IR 32
2351 MIPS KVM_REG_MIPS_MSA_CSR 32
2352 ======= =============================== ============
2354 ARM registers are mapped using the lower 32 bits. The upper 16 of that
2355 is the register group type, or coprocessor number:
2357 ARM core registers have the following id bit patterns::
2359 0x4020 0000 0010 <index into the kvm_regs struct:16>
2361 ARM 32-bit CP15 registers have the following id bit patterns::
2363 0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3>
2365 ARM 64-bit CP15 registers have the following id bit patterns::
2367 0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3>
2369 ARM CCSIDR registers are demultiplexed by CSSELR value::
2371 0x4020 0000 0011 00 <csselr:8>
2373 ARM 32-bit VFP control registers have the following id bit patterns::
2375 0x4020 0000 0012 1 <regno:12>
2377 ARM 64-bit FP registers have the following id bit patterns::
2379 0x4030 0000 0012 0 <regno:12>
2381 ARM firmware pseudo-registers have the following bit pattern::
2383 0x4030 0000 0014 <regno:16>
2386 arm64 registers are mapped using the lower 32 bits. The upper 16 of
2387 that is the register group type, or coprocessor number:
2389 arm64 core/FP-SIMD registers have the following id bit patterns. Note
2390 that the size of the access is variable, as the kvm_regs structure
2391 contains elements ranging from 32 to 128 bits. The index is a 32bit
2392 value in the kvm_regs structure seen as a 32bit array::
2394 0x60x0 0000 0010 <index into the kvm_regs struct:16>
2398 ======================= ========= ===== =======================================
2399 Encoding Register Bits kvm_regs member
2400 ======================= ========= ===== =======================================
2401 0x6030 0000 0010 0000 X0 64 regs.regs[0]
2402 0x6030 0000 0010 0002 X1 64 regs.regs[1]
2404 0x6030 0000 0010 003c X30 64 regs.regs[30]
2405 0x6030 0000 0010 003e SP 64 regs.sp
2406 0x6030 0000 0010 0040 PC 64 regs.pc
2407 0x6030 0000 0010 0042 PSTATE 64 regs.pstate
2408 0x6030 0000 0010 0044 SP_EL1 64 sp_el1
2409 0x6030 0000 0010 0046 ELR_EL1 64 elr_el1
2410 0x6030 0000 0010 0048 SPSR_EL1 64 spsr[KVM_SPSR_EL1] (alias SPSR_SVC)
2411 0x6030 0000 0010 004a SPSR_ABT 64 spsr[KVM_SPSR_ABT]
2412 0x6030 0000 0010 004c SPSR_UND 64 spsr[KVM_SPSR_UND]
2413 0x6030 0000 0010 004e SPSR_IRQ 64 spsr[KVM_SPSR_IRQ]
2414 0x6060 0000 0010 0050 SPSR_FIQ 64 spsr[KVM_SPSR_FIQ]
2415 0x6040 0000 0010 0054 V0 128 fp_regs.vregs[0] [1]_
2416 0x6040 0000 0010 0058 V1 128 fp_regs.vregs[1] [1]_
2418 0x6040 0000 0010 00d0 V31 128 fp_regs.vregs[31] [1]_
2419 0x6020 0000 0010 00d4 FPSR 32 fp_regs.fpsr
2420 0x6020 0000 0010 00d5 FPCR 32 fp_regs.fpcr
2421 ======================= ========= ===== =======================================
2423 .. [1] These encodings are not accepted for SVE-enabled vcpus. See
2426 The equivalent register content can be accessed via bits [127:0] of
2427 the corresponding SVE Zn registers instead for vcpus that have SVE
2428 enabled (see below).
2430 arm64 CCSIDR registers are demultiplexed by CSSELR value::
2432 0x6020 0000 0011 00 <csselr:8>
2434 arm64 system registers have the following id bit patterns::
2436 0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3>
2440 Two system register IDs do not follow the specified pattern. These
2441 are KVM_REG_ARM_TIMER_CVAL and KVM_REG_ARM_TIMER_CNT, which map to
2442 system registers CNTV_CVAL_EL0 and CNTVCT_EL0 respectively. These
2443 two had their values accidentally swapped, which means TIMER_CVAL is
2444 derived from the register encoding for CNTVCT_EL0 and TIMER_CNT is
2445 derived from the register encoding for CNTV_CVAL_EL0. As this is
2446 API, it must remain this way.
2448 arm64 firmware pseudo-registers have the following bit pattern::
2450 0x6030 0000 0014 <regno:16>
2452 arm64 SVE registers have the following bit patterns::
2454 0x6080 0000 0015 00 <n:5> <slice:5> Zn bits[2048*slice + 2047 : 2048*slice]
2455 0x6050 0000 0015 04 <n:4> <slice:5> Pn bits[256*slice + 255 : 256*slice]
2456 0x6050 0000 0015 060 <slice:5> FFR bits[256*slice + 255 : 256*slice]
2457 0x6060 0000 0015 ffff KVM_REG_ARM64_SVE_VLS pseudo-register
2459 Access to register IDs where 2048 * slice >= 128 * max_vq will fail with
2460 ENOENT. max_vq is the vcpu's maximum supported vector length in 128-bit
2461 quadwords: see [2]_ below.
2463 These registers are only accessible on vcpus for which SVE is enabled.
2464 See KVM_ARM_VCPU_INIT for details.
2466 In addition, except for KVM_REG_ARM64_SVE_VLS, these registers are not
2467 accessible until the vcpu's SVE configuration has been finalized
2468 using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). See KVM_ARM_VCPU_INIT
2469 and KVM_ARM_VCPU_FINALIZE for more information about this procedure.
2471 KVM_REG_ARM64_SVE_VLS is a pseudo-register that allows the set of vector
2472 lengths supported by the vcpu to be discovered and configured by
2473 userspace. When transferred to or from user memory via KVM_GET_ONE_REG
2474 or KVM_SET_ONE_REG, the value of this register is of type
2475 __u64[KVM_ARM64_SVE_VLS_WORDS], and encodes the set of vector lengths as
2478 __u64 vector_lengths[KVM_ARM64_SVE_VLS_WORDS];
2480 if (vq >= SVE_VQ_MIN && vq <= SVE_VQ_MAX &&
2481 ((vector_lengths[(vq - KVM_ARM64_SVE_VQ_MIN) / 64] >>
2482 ((vq - KVM_ARM64_SVE_VQ_MIN) % 64)) & 1))
2483 /* Vector length vq * 16 bytes supported */
2485 /* Vector length vq * 16 bytes not supported */
2487 .. [2] The maximum value vq for which the above condition is true is
2488 max_vq. This is the maximum vector length available to the guest on
2489 this vcpu, and determines which register slices are visible through
2490 this ioctl interface.
2492 (See Documentation/arm64/sve.rst for an explanation of the "vq"
2495 KVM_REG_ARM64_SVE_VLS is only accessible after KVM_ARM_VCPU_INIT.
2496 KVM_ARM_VCPU_INIT initialises it to the best set of vector lengths that
2499 Userspace may subsequently modify it if desired until the vcpu's SVE
2500 configuration is finalized using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE).
2502 Apart from simply removing all vector lengths from the host set that
2503 exceed some value, support for arbitrarily chosen sets of vector lengths
2504 is hardware-dependent and may not be available. Attempting to configure
2505 an invalid set of vector lengths via KVM_SET_ONE_REG will fail with
2508 After the vcpu's SVE configuration is finalized, further attempts to
2509 write this register will fail with EPERM.
2512 MIPS registers are mapped using the lower 32 bits. The upper 16 of that is
2513 the register group type:
2515 MIPS core registers (see above) have the following id bit patterns::
2517 0x7030 0000 0000 <reg:16>
2519 MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit
2520 patterns depending on whether they're 32-bit or 64-bit registers::
2522 0x7020 0000 0001 00 <reg:5> <sel:3> (32-bit)
2523 0x7030 0000 0001 00 <reg:5> <sel:3> (64-bit)
2525 Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_MIPS_CP0_ENTRYLO1 are the MIPS64
2526 versions of the EntryLo registers regardless of the word size of the host
2527 hardware, host kernel, guest, and whether XPA is present in the guest, i.e.
2528 with the RI and XI bits (if they exist) in bits 63 and 62 respectively, and
2529 the PFNX field starting at bit 30.
2531 MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) above) have the following id bit
2534 0x7030 0000 0001 01 <reg:8>
2536 MIPS KVM control registers (see above) have the following id bit patterns::
2538 0x7030 0000 0002 <reg:16>
2540 MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following
2541 id bit patterns depending on the size of the register being accessed. They are
2542 always accessed according to the current guest FPU mode (Status.FR and
2543 Config5.FRE), i.e. as the guest would see them, and they become unpredictable
2544 if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector
2545 registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they
2546 overlap the FPU registers::
2548 0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers)
2549 0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers)
2550 0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers)
2552 MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the
2553 following id bit patterns::
2555 0x7020 0000 0003 01 <0:3> <reg:5>
2557 MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the
2558 following id bit patterns::
2560 0x7020 0000 0003 02 <0:3> <reg:5>
2563 4.69 KVM_GET_ONE_REG
2564 --------------------
2566 :Capability: KVM_CAP_ONE_REG
2569 :Parameters: struct kvm_one_reg (in and out)
2570 :Returns: 0 on success, negative value on failure
2574 ======== ============================================================
2575 Â ENOENT Â Â no such register
2576 Â EINVAL Â Â invalid register ID, or no such register or used with VMs in
2577 protected virtualization mode on s390
2578 Â EPERM Â Â Â (arm64) register access not allowed before vcpu finalization
2579 ======== ============================================================
2581 (These error codes are indicative only: do not rely on a specific error
2582 code being returned in a specific situation.)
2584 This ioctl allows to receive the value of a single register implemented
2585 in a vcpu. The register to read is indicated by the "id" field of the
2586 kvm_one_reg struct passed in. On success, the register value can be found
2587 at the memory location pointed to by "addr".
2589 The list of registers accessible using this interface is identical to the
2593 4.70 KVM_KVMCLOCK_CTRL
2594 ----------------------
2596 :Capability: KVM_CAP_KVMCLOCK_CTRL
2597 :Architectures: Any that implement pvclocks (currently x86 only)
2600 :Returns: 0 on success, -1 on error
2602 This ioctl sets a flag accessible to the guest indicating that the specified
2603 vCPU has been paused by the host userspace.
2605 The host will set a flag in the pvclock structure that is checked from the
2606 soft lockup watchdog. The flag is part of the pvclock structure that is
2607 shared between guest and host, specifically the second bit of the flags
2608 field of the pvclock_vcpu_time_info structure. It will be set exclusively by
2609 the host and read/cleared exclusively by the guest. The guest operation of
2610 checking and clearing the flag must be an atomic operation so
2611 load-link/store-conditional, or equivalent must be used. There are two cases
2612 where the guest will clear the flag: when the soft lockup watchdog timer resets
2613 itself or when a soft lockup is detected. This ioctl can be called any time
2614 after pausing the vcpu, but before it is resumed.
2620 :Capability: KVM_CAP_SIGNAL_MSI
2621 :Architectures: x86 arm arm64
2623 :Parameters: struct kvm_msi (in)
2624 :Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error
2626 Directly inject a MSI message. Only valid with in-kernel irqchip that handles
2641 KVM_MSI_VALID_DEVID: devid contains a valid value. The per-VM
2642 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
2643 the device ID. If this capability is not available, userspace
2644 should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
2646 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
2647 for the device that wrote the MSI message. For PCI, this is usually a
2648 BFD identifier in the lower 16 bits.
2650 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
2651 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
2652 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
2653 address_hi must be zero.
2656 4.71 KVM_CREATE_PIT2
2657 --------------------
2659 :Capability: KVM_CAP_PIT2
2662 :Parameters: struct kvm_pit_config (in)
2663 :Returns: 0 on success, -1 on error
2665 Creates an in-kernel device model for the i8254 PIT. This call is only valid
2666 after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following
2667 parameters have to be passed::
2669 struct kvm_pit_config {
2676 #define KVM_PIT_SPEAKER_DUMMY 1 /* emulate speaker port stub */
2678 PIT timer interrupts may use a per-VM kernel thread for injection. If it
2679 exists, this thread will have a name of the following pattern::
2681 kvm-pit/<owner-process-pid>
2683 When running a guest with elevated priorities, the scheduling parameters of
2684 this thread may have to be adjusted accordingly.
2686 This IOCTL replaces the obsolete KVM_CREATE_PIT.
2692 :Capability: KVM_CAP_PIT_STATE2
2695 :Parameters: struct kvm_pit_state2 (out)
2696 :Returns: 0 on success, -1 on error
2698 Retrieves the state of the in-kernel PIT model. Only valid after
2699 KVM_CREATE_PIT2. The state is returned in the following structure::
2701 struct kvm_pit_state2 {
2702 struct kvm_pit_channel_state channels[3];
2709 /* disable PIT in HPET legacy mode */
2710 #define KVM_PIT_FLAGS_HPET_LEGACY 0x00000001
2712 This IOCTL replaces the obsolete KVM_GET_PIT.
2718 :Capability: KVM_CAP_PIT_STATE2
2721 :Parameters: struct kvm_pit_state2 (in)
2722 :Returns: 0 on success, -1 on error
2724 Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2.
2725 See KVM_GET_PIT2 for details on struct kvm_pit_state2.
2727 This IOCTL replaces the obsolete KVM_SET_PIT.
2730 4.74 KVM_PPC_GET_SMMU_INFO
2731 --------------------------
2733 :Capability: KVM_CAP_PPC_GET_SMMU_INFO
2734 :Architectures: powerpc
2737 :Returns: 0 on success, -1 on error
2739 This populates and returns a structure describing the features of
2740 the "Server" class MMU emulation supported by KVM.
2741 This can in turn be used by userspace to generate the appropriate
2742 device-tree properties for the guest operating system.
2744 The structure contains some global information, followed by an
2745 array of supported segment page sizes::
2747 struct kvm_ppc_smmu_info {
2751 struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ];
2754 The supported flags are:
2756 - KVM_PPC_PAGE_SIZES_REAL:
2757 When that flag is set, guest page sizes must "fit" the backing
2758 store page sizes. When not set, any page size in the list can
2759 be used regardless of how they are backed by userspace.
2761 - KVM_PPC_1T_SEGMENTS
2762 The emulated MMU supports 1T segments in addition to the
2766 This flag indicates that HPT guests are not supported by KVM,
2767 thus all guests must use radix MMU mode.
2769 The "slb_size" field indicates how many SLB entries are supported
2771 The "sps" array contains 8 entries indicating the supported base
2772 page sizes for a segment in increasing order. Each entry is defined
2775 struct kvm_ppc_one_seg_page_size {
2776 __u32 page_shift; /* Base page shift of segment (or 0) */
2777 __u32 slb_enc; /* SLB encoding for BookS */
2778 struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ];
2781 An entry with a "page_shift" of 0 is unused. Because the array is
2782 organized in increasing order, a lookup can stop when encoutering
2785 The "slb_enc" field provides the encoding to use in the SLB for the
2786 page size. The bits are in positions such as the value can directly
2787 be OR'ed into the "vsid" argument of the slbmte instruction.
2789 The "enc" array is a list which for each of those segment base page
2790 size provides the list of supported actual page sizes (which can be
2791 only larger or equal to the base page size), along with the
2792 corresponding encoding in the hash PTE. Similarly, the array is
2793 8 entries sorted by increasing sizes and an entry with a "0" shift
2794 is an empty entry and a terminator::
2796 struct kvm_ppc_one_page_size {
2797 __u32 page_shift; /* Page shift (or 0) */
2798 __u32 pte_enc; /* Encoding in the HPTE (>>12) */
2801 The "pte_enc" field provides a value that can OR'ed into the hash
2802 PTE's RPN field (ie, it needs to be shifted left by 12 to OR it
2803 into the hash PTE second double word).
2808 :Capability: KVM_CAP_IRQFD
2809 :Architectures: x86 s390 arm arm64
2811 :Parameters: struct kvm_irqfd (in)
2812 :Returns: 0 on success, -1 on error
2814 Allows setting an eventfd to directly trigger a guest interrupt.
2815 kvm_irqfd.fd specifies the file descriptor to use as the eventfd and
2816 kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When
2817 an event is triggered on the eventfd, an interrupt is injected into
2818 the guest using the specified gsi pin. The irqfd is removed using
2819 the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd
2822 With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify
2823 mechanism allowing emulation of level-triggered, irqfd-based
2824 interrupts. When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an
2825 additional eventfd in the kvm_irqfd.resamplefd field. When operating
2826 in resample mode, posting of an interrupt through kvm_irq.fd asserts
2827 the specified gsi in the irqchip. When the irqchip is resampled, such
2828 as from an EOI, the gsi is de-asserted and the user is notified via
2829 kvm_irqfd.resamplefd. It is the user's responsibility to re-queue
2830 the interrupt if the device making use of it still requires service.
2831 Note that closing the resamplefd is not sufficient to disable the
2832 irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment
2833 and need not be specified with KVM_IRQFD_FLAG_DEASSIGN.
2835 On arm/arm64, gsi routing being supported, the following can happen:
2837 - in case no routing entry is associated to this gsi, injection fails
2838 - in case the gsi is associated to an irqchip routing entry,
2839 irqchip.pin + 32 corresponds to the injected SPI ID.
2840 - in case the gsi is associated to an MSI routing entry, the MSI
2841 message and device ID are translated into an LPI (support restricted
2842 to GICv3 ITS in-kernel emulation).
2844 4.76 KVM_PPC_ALLOCATE_HTAB
2845 --------------------------
2847 :Capability: KVM_CAP_PPC_ALLOC_HTAB
2848 :Architectures: powerpc
2850 :Parameters: Pointer to u32 containing hash table order (in/out)
2851 :Returns: 0 on success, -1 on error
2853 This requests the host kernel to allocate an MMU hash table for a
2854 guest using the PAPR paravirtualization interface. This only does
2855 anything if the kernel is configured to use the Book 3S HV style of
2856 virtualization. Otherwise the capability doesn't exist and the ioctl
2857 returns an ENOTTY error. The rest of this description assumes Book 3S
2860 There must be no vcpus running when this ioctl is called; if there
2861 are, it will do nothing and return an EBUSY error.
2863 The parameter is a pointer to a 32-bit unsigned integer variable
2864 containing the order (log base 2) of the desired size of the hash
2865 table, which must be between 18 and 46. On successful return from the
2866 ioctl, the value will not be changed by the kernel.
2868 If no hash table has been allocated when any vcpu is asked to run
2869 (with the KVM_RUN ioctl), the host kernel will allocate a
2870 default-sized hash table (16 MB).
2872 If this ioctl is called when a hash table has already been allocated,
2873 with a different order from the existing hash table, the existing hash
2874 table will be freed and a new one allocated. If this is ioctl is
2875 called when a hash table has already been allocated of the same order
2876 as specified, the kernel will clear out the existing hash table (zero
2877 all HPTEs). In either case, if the guest is using the virtualized
2878 real-mode area (VRMA) facility, the kernel will re-create the VMRA
2879 HPTEs on the next KVM_RUN of any vcpu.
2881 4.77 KVM_S390_INTERRUPT
2882 -----------------------
2885 :Architectures: s390
2886 :Type: vm ioctl, vcpu ioctl
2887 :Parameters: struct kvm_s390_interrupt (in)
2888 :Returns: 0 on success, -1 on error
2890 Allows to inject an interrupt to the guest. Interrupts can be floating
2891 (vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type.
2893 Interrupt parameters are passed via kvm_s390_interrupt::
2895 struct kvm_s390_interrupt {
2901 type can be one of the following:
2903 KVM_S390_SIGP_STOP (vcpu)
2904 - sigp stop; optional flags in parm
2905 KVM_S390_PROGRAM_INT (vcpu)
2906 - program check; code in parm
2907 KVM_S390_SIGP_SET_PREFIX (vcpu)
2908 - sigp set prefix; prefix address in parm
2909 KVM_S390_RESTART (vcpu)
2911 KVM_S390_INT_CLOCK_COMP (vcpu)
2912 - clock comparator interrupt
2913 KVM_S390_INT_CPU_TIMER (vcpu)
2914 - CPU timer interrupt
2915 KVM_S390_INT_VIRTIO (vm)
2916 - virtio external interrupt; external interrupt
2917 parameters in parm and parm64
2918 KVM_S390_INT_SERVICE (vm)
2919 - sclp external interrupt; sclp parameter in parm
2920 KVM_S390_INT_EMERGENCY (vcpu)
2921 - sigp emergency; source cpu in parm
2922 KVM_S390_INT_EXTERNAL_CALL (vcpu)
2923 - sigp external call; source cpu in parm
2924 KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm)
2925 - compound value to indicate an
2926 I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel);
2927 I/O interruption parameters in parm (subchannel) and parm64 (intparm,
2928 interruption subclass)
2929 KVM_S390_MCHK (vm, vcpu)
2930 - machine check interrupt; cr 14 bits in parm, machine check interrupt
2931 code in parm64 (note that machine checks needing further payload are not
2932 supported by this ioctl)
2934 This is an asynchronous vcpu ioctl and can be invoked from any thread.
2936 4.78 KVM_PPC_GET_HTAB_FD
2937 ------------------------
2939 :Capability: KVM_CAP_PPC_HTAB_FD
2940 :Architectures: powerpc
2942 :Parameters: Pointer to struct kvm_get_htab_fd (in)
2943 :Returns: file descriptor number (>= 0) on success, -1 on error
2945 This returns a file descriptor that can be used either to read out the
2946 entries in the guest's hashed page table (HPT), or to write entries to
2947 initialize the HPT. The returned fd can only be written to if the
2948 KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and
2949 can only be read if that bit is clear. The argument struct looks like
2952 /* For KVM_PPC_GET_HTAB_FD */
2953 struct kvm_get_htab_fd {
2959 /* Values for kvm_get_htab_fd.flags */
2960 #define KVM_GET_HTAB_BOLTED_ONLY ((__u64)0x1)
2961 #define KVM_GET_HTAB_WRITE ((__u64)0x2)
2963 The 'start_index' field gives the index in the HPT of the entry at
2964 which to start reading. It is ignored when writing.
2966 Reads on the fd will initially supply information about all
2967 "interesting" HPT entries. Interesting entries are those with the
2968 bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise
2969 all entries. When the end of the HPT is reached, the read() will
2970 return. If read() is called again on the fd, it will start again from
2971 the beginning of the HPT, but will only return HPT entries that have
2972 changed since they were last read.
2974 Data read or written is structured as a header (8 bytes) followed by a
2975 series of valid HPT entries (16 bytes) each. The header indicates how
2976 many valid HPT entries there are and how many invalid entries follow
2977 the valid entries. The invalid entries are not represented explicitly
2978 in the stream. The header format is::
2980 struct kvm_get_htab_header {
2986 Writes to the fd create HPT entries starting at the index given in the
2987 header; first 'n_valid' valid entries with contents from the data
2988 written, then 'n_invalid' invalid entries, invalidating any previously
2989 valid entries found.
2991 4.79 KVM_CREATE_DEVICE
2992 ----------------------
2994 :Capability: KVM_CAP_DEVICE_CTRL
2996 :Parameters: struct kvm_create_device (in/out)
2997 :Returns: 0 on success, -1 on error
3001 ====== =======================================================
3002 ENODEV The device type is unknown or unsupported
3003 EEXIST Device already created, and this type of device may not
3004 be instantiated multiple times
3005 ====== =======================================================
3007 Other error conditions may be defined by individual device types or
3008 have their standard meanings.
3010 Creates an emulated device in the kernel. The file descriptor returned
3011 in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR.
3013 If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the
3014 device type is supported (not necessarily whether it can be created
3017 Individual devices should not define flags. Attributes should be used
3018 for specifying any behavior that is not implied by the device type
3023 struct kvm_create_device {
3024 __u32 type; /* in: KVM_DEV_TYPE_xxx */
3025 __u32 fd; /* out: device handle */
3026 __u32 flags; /* in: KVM_CREATE_DEVICE_xxx */
3029 4.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR
3030 --------------------------------------------
3032 :Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
3033 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
3034 :Type: device ioctl, vm ioctl, vcpu ioctl
3035 :Parameters: struct kvm_device_attr
3036 :Returns: 0 on success, -1 on error
3040 ===== =============================================================
3041 ENXIO The group or attribute is unknown/unsupported for this device
3042 or hardware support is missing.
3043 EPERM The attribute cannot (currently) be accessed this way
3044 (e.g. read-only attribute, or attribute that only makes
3045 sense when the device is in a different state)
3046 ===== =============================================================
3048 Other error conditions may be defined by individual device types.
3050 Gets/sets a specified piece of device configuration and/or state. The
3051 semantics are device-specific. See individual device documentation in
3052 the "devices" directory. As with ONE_REG, the size of the data
3053 transferred is defined by the particular attribute.
3057 struct kvm_device_attr {
3058 __u32 flags; /* no flags currently defined */
3059 __u32 group; /* device-defined */
3060 __u64 attr; /* group-defined */
3061 __u64 addr; /* userspace address of attr data */
3064 4.81 KVM_HAS_DEVICE_ATTR
3065 ------------------------
3067 :Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
3068 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
3069 :Type: device ioctl, vm ioctl, vcpu ioctl
3070 :Parameters: struct kvm_device_attr
3071 :Returns: 0 on success, -1 on error
3075 ===== =============================================================
3076 ENXIO The group or attribute is unknown/unsupported for this device
3077 or hardware support is missing.
3078 ===== =============================================================
3080 Tests whether a device supports a particular attribute. A successful
3081 return indicates the attribute is implemented. It does not necessarily
3082 indicate that the attribute can be read or written in the device's
3083 current state. "addr" is ignored.
3085 4.82 KVM_ARM_VCPU_INIT
3086 ----------------------
3089 :Architectures: arm, arm64
3091 :Parameters: struct kvm_vcpu_init (in)
3092 :Returns: 0 on success; -1 on error
3096 ====== =================================================================
3097 Â EINVAL Â Â Â the target is unknown, or the combination of features is invalid.
3098 Â ENOENT Â Â Â a features bit specified is unknown.
3099 ====== =================================================================
3101 This tells KVM what type of CPU to present to the guest, and what
3102 optional features it should have. Â This will cause a reset of the cpu
3103 registers to their initial values. Â If this is not called, KVM_RUN will
3104 return ENOEXEC for that vcpu.
3106 Note that because some registers reflect machine topology, all vcpus
3107 should be created before this ioctl is invoked.
3109 Userspace can call this function multiple times for a given vcpu, including
3110 after the vcpu has been run. This will reset the vcpu to its initial
3111 state. All calls to this function after the initial call must use the same
3112 target and same set of feature flags, otherwise EINVAL will be returned.
3116 - KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state.
3117 Depends on KVM_CAP_ARM_PSCI. If not set, the CPU will be powered on
3118 and execute guest code when KVM_RUN is called.
3119 - KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode.
3120 Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only).
3121 - KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 (or a future revision
3122 backward compatible with v0.2) for the CPU.
3123 Depends on KVM_CAP_ARM_PSCI_0_2.
3124 - KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU.
3125 Depends on KVM_CAP_ARM_PMU_V3.
3127 - KVM_ARM_VCPU_PTRAUTH_ADDRESS: Enables Address Pointer authentication
3129 Depends on KVM_CAP_ARM_PTRAUTH_ADDRESS.
3130 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are
3131 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and
3132 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be
3135 - KVM_ARM_VCPU_PTRAUTH_GENERIC: Enables Generic Pointer authentication
3137 Depends on KVM_CAP_ARM_PTRAUTH_GENERIC.
3138 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are
3139 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and
3140 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be
3143 - KVM_ARM_VCPU_SVE: Enables SVE for the CPU (arm64 only).
3144 Depends on KVM_CAP_ARM_SVE.
3145 Requires KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3147 * After KVM_ARM_VCPU_INIT:
3149 - KVM_REG_ARM64_SVE_VLS may be read using KVM_GET_ONE_REG: the
3150 initial value of this pseudo-register indicates the best set of
3151 vector lengths possible for a vcpu on this host.
3153 * Before KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3155 - KVM_RUN and KVM_GET_REG_LIST are not available;
3157 - KVM_GET_ONE_REG and KVM_SET_ONE_REG cannot be used to access
3158 the scalable archietctural SVE registers
3159 KVM_REG_ARM64_SVE_ZREG(), KVM_REG_ARM64_SVE_PREG() or
3160 KVM_REG_ARM64_SVE_FFR;
3162 - KVM_REG_ARM64_SVE_VLS may optionally be written using
3163 KVM_SET_ONE_REG, to modify the set of vector lengths available
3166 * After KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3168 - the KVM_REG_ARM64_SVE_VLS pseudo-register is immutable, and can
3169 no longer be written using KVM_SET_ONE_REG.
3171 4.83 KVM_ARM_PREFERRED_TARGET
3172 -----------------------------
3175 :Architectures: arm, arm64
3177 :Parameters: struct kvm_vcpu_init (out)
3178 :Returns: 0 on success; -1 on error
3182 ====== ==========================================
3183 ENODEV no preferred target available for the host
3184 ====== ==========================================
3186 This queries KVM for preferred CPU target type which can be emulated
3187 by KVM on underlying host.
3189 The ioctl returns struct kvm_vcpu_init instance containing information
3190 about preferred CPU target type and recommended features for it. The
3191 kvm_vcpu_init->features bitmap returned will have feature bits set if
3192 the preferred target recommends setting these features, but this is
3195 The information returned by this ioctl can be used to prepare an instance
3196 of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in
3197 VCPU matching underlying host.
3200 4.84 KVM_GET_REG_LIST
3201 ---------------------
3204 :Architectures: arm, arm64, mips
3206 :Parameters: struct kvm_reg_list (in/out)
3207 :Returns: 0 on success; -1 on error
3211 ===== ==============================================================
3212 Â E2BIG Â Â Â Â the reg index list is too big to fit in the array specified by
3213 Â Â Â Â Â Â Â Â Â Â Â Â the user (the number required will be written into n).
3214 ===== ==============================================================
3218 struct kvm_reg_list {
3219 __u64 n; /* number of registers in reg[] */
3223 This ioctl returns the guest registers that are supported for the
3224 KVM_GET_ONE_REG/KVM_SET_ONE_REG calls.
3227 4.85 KVM_ARM_SET_DEVICE_ADDR (deprecated)
3228 -----------------------------------------
3230 :Capability: KVM_CAP_ARM_SET_DEVICE_ADDR
3231 :Architectures: arm, arm64
3233 :Parameters: struct kvm_arm_device_address (in)
3234 :Returns: 0 on success, -1 on error
3238 ====== ============================================
3239 ENODEV The device id is unknown
3240 ENXIO Device not supported on current system
3241 EEXIST Address already set
3242 E2BIG Address outside guest physical address space
3243 EBUSY Address overlaps with other device range
3244 ====== ============================================
3248 struct kvm_arm_device_addr {
3253 Specify a device address in the guest's physical address space where guests
3254 can access emulated or directly exposed devices, which the host kernel needs
3255 to know about. The id field is an architecture specific identifier for a
3258 ARM/arm64 divides the id field into two parts, a device id and an
3259 address type id specific to the individual device::
3261 Â bits: | 63 ... 32 | 31 ... 16 | 15 ... 0 |
3262 field: | 0x00000000 | device id | addr type id |
3264 ARM/arm64 currently only require this when using the in-kernel GIC
3265 support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2
3266 as the device id. When setting the base address for the guest's
3267 mapping of the VGIC virtual CPU and distributor interface, the ioctl
3268 must be called after calling KVM_CREATE_IRQCHIP, but before calling
3269 KVM_RUN on any of the VCPUs. Calling this ioctl twice for any of the
3270 base addresses will return -EEXIST.
3272 Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API
3273 should be used instead.
3276 4.86 KVM_PPC_RTAS_DEFINE_TOKEN
3277 ------------------------------
3279 :Capability: KVM_CAP_PPC_RTAS
3282 :Parameters: struct kvm_rtas_token_args
3283 :Returns: 0 on success, -1 on error
3285 Defines a token value for a RTAS (Run Time Abstraction Services)
3286 service in order to allow it to be handled in the kernel. The
3287 argument struct gives the name of the service, which must be the name
3288 of a service that has a kernel-side implementation. If the token
3289 value is non-zero, it will be associated with that service, and
3290 subsequent RTAS calls by the guest specifying that token will be
3291 handled by the kernel. If the token value is 0, then any token
3292 associated with the service will be forgotten, and subsequent RTAS
3293 calls by the guest for that service will be passed to userspace to be
3296 4.87 KVM_SET_GUEST_DEBUG
3297 ------------------------
3299 :Capability: KVM_CAP_SET_GUEST_DEBUG
3300 :Architectures: x86, s390, ppc, arm64
3302 :Parameters: struct kvm_guest_debug (in)
3303 :Returns: 0 on success; -1 on error
3307 struct kvm_guest_debug {
3310 struct kvm_guest_debug_arch arch;
3313 Set up the processor specific debug registers and configure vcpu for
3314 handling guest debug events. There are two parts to the structure, the
3315 first a control bitfield indicates the type of debug events to handle
3316 when running. Common control bits are:
3318 - KVM_GUESTDBG_ENABLE: guest debugging is enabled
3319 - KVM_GUESTDBG_SINGLESTEP: the next run should single-step
3321 The top 16 bits of the control field are architecture specific control
3322 flags which can include the following:
3324 - KVM_GUESTDBG_USE_SW_BP: using software breakpoints [x86, arm64]
3325 - KVM_GUESTDBG_USE_HW_BP: using hardware breakpoints [x86, s390, arm64]
3326 - KVM_GUESTDBG_INJECT_DB: inject DB type exception [x86]
3327 - KVM_GUESTDBG_INJECT_BP: inject BP type exception [x86]
3328 - KVM_GUESTDBG_EXIT_PENDING: trigger an immediate guest exit [s390]
3330 For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints
3331 are enabled in memory so we need to ensure breakpoint exceptions are
3332 correctly trapped and the KVM run loop exits at the breakpoint and not
3333 running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP
3334 we need to ensure the guest vCPUs architecture specific registers are
3335 updated to the correct (supplied) values.
3337 The second part of the structure is architecture specific and
3338 typically contains a set of debug registers.
3340 For arm64 the number of debug registers is implementation defined and
3341 can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and
3342 KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number
3343 indicating the number of supported registers.
3345 For ppc, the KVM_CAP_PPC_GUEST_DEBUG_SSTEP capability indicates whether
3346 the single-step debug event (KVM_GUESTDBG_SINGLESTEP) is supported.
3348 When debug events exit the main run loop with the reason
3349 KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run
3350 structure containing architecture specific debug information.
3352 4.88 KVM_GET_EMULATED_CPUID
3353 ---------------------------
3355 :Capability: KVM_CAP_EXT_EMUL_CPUID
3358 :Parameters: struct kvm_cpuid2 (in/out)
3359 :Returns: 0 on success, -1 on error
3366 struct kvm_cpuid_entry2 entries[0];
3369 The member 'flags' is used for passing flags from userspace.
3373 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
3374 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */
3375 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */
3377 struct kvm_cpuid_entry2 {
3388 This ioctl returns x86 cpuid features which are emulated by
3389 kvm.Userspace can use the information returned by this ioctl to query
3390 which features are emulated by kvm instead of being present natively.
3392 Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2
3393 structure with the 'nent' field indicating the number of entries in
3394 the variable-size array 'entries'. If the number of entries is too low
3395 to describe the cpu capabilities, an error (E2BIG) is returned. If the
3396 number is too high, the 'nent' field is adjusted and an error (ENOMEM)
3397 is returned. If the number is just right, the 'nent' field is adjusted
3398 to the number of valid entries in the 'entries' array, which is then
3401 The entries returned are the set CPUID bits of the respective features
3402 which kvm emulates, as returned by the CPUID instruction, with unknown
3403 or unsupported feature bits cleared.
3405 Features like x2apic, for example, may not be present in the host cpu
3406 but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be
3407 emulated efficiently and thus not included here.
3409 The fields in each entry are defined as follows:
3412 the eax value used to obtain the entry
3414 the ecx value used to obtain the entry (for entries that are
3417 an OR of zero or more of the following:
3419 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
3420 if the index field is valid
3424 the values returned by the cpuid instruction for
3425 this function/index combination
3427 4.89 KVM_S390_MEM_OP
3428 --------------------
3430 :Capability: KVM_CAP_S390_MEM_OP
3431 :Architectures: s390
3433 :Parameters: struct kvm_s390_mem_op (in)
3434 :Returns: = 0 on success,
3435 < 0 on generic error (e.g. -EFAULT or -ENOMEM),
3436 > 0 if an exception occurred while walking the page tables
3438 Read or write data from/to the logical (virtual) memory of a VCPU.
3440 Parameters are specified via the following structure::
3442 struct kvm_s390_mem_op {
3443 __u64 gaddr; /* the guest address */
3444 __u64 flags; /* flags */
3445 __u32 size; /* amount of bytes */
3446 __u32 op; /* type of operation */
3447 __u64 buf; /* buffer in userspace */
3448 __u8 ar; /* the access register number */
3449 __u8 reserved[31]; /* should be set to 0 */
3452 The type of operation is specified in the "op" field. It is either
3453 KVM_S390_MEMOP_LOGICAL_READ for reading from logical memory space or
3454 KVM_S390_MEMOP_LOGICAL_WRITE for writing to logical memory space. The
3455 KVM_S390_MEMOP_F_CHECK_ONLY flag can be set in the "flags" field to check
3456 whether the corresponding memory access would create an access exception
3457 (without touching the data in the memory at the destination). In case an
3458 access exception occurred while walking the MMU tables of the guest, the
3459 ioctl returns a positive error number to indicate the type of exception.
3460 This exception is also raised directly at the corresponding VCPU if the
3461 flag KVM_S390_MEMOP_F_INJECT_EXCEPTION is set in the "flags" field.
3463 The start address of the memory region has to be specified in the "gaddr"
3464 field, and the length of the region in the "size" field (which must not
3465 be 0). The maximum value for "size" can be obtained by checking the
3466 KVM_CAP_S390_MEM_OP capability. "buf" is the buffer supplied by the
3467 userspace application where the read data should be written to for
3468 KVM_S390_MEMOP_LOGICAL_READ, or where the data that should be written is
3469 stored for a KVM_S390_MEMOP_LOGICAL_WRITE. When KVM_S390_MEMOP_F_CHECK_ONLY
3470 is specified, "buf" is unused and can be NULL. "ar" designates the access
3471 register number to be used; the valid range is 0..15.
3473 The "reserved" field is meant for future extensions. It is not used by
3474 KVM with the currently defined set of flags.
3476 4.90 KVM_S390_GET_SKEYS
3477 -----------------------
3479 :Capability: KVM_CAP_S390_SKEYS
3480 :Architectures: s390
3482 :Parameters: struct kvm_s390_skeys
3483 :Returns: 0 on success, KVM_S390_GET_KEYS_NONE if guest is not using storage
3484 keys, negative value on error
3486 This ioctl is used to get guest storage key values on the s390
3487 architecture. The ioctl takes parameters via the kvm_s390_skeys struct::
3489 struct kvm_s390_skeys {
3492 __u64 skeydata_addr;
3497 The start_gfn field is the number of the first guest frame whose storage keys
3500 The count field is the number of consecutive frames (starting from start_gfn)
3501 whose storage keys to get. The count field must be at least 1 and the maximum
3502 allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range
3503 will cause the ioctl to return -EINVAL.
3505 The skeydata_addr field is the address to a buffer large enough to hold count
3506 bytes. This buffer will be filled with storage key data by the ioctl.
3508 4.91 KVM_S390_SET_SKEYS
3509 -----------------------
3511 :Capability: KVM_CAP_S390_SKEYS
3512 :Architectures: s390
3514 :Parameters: struct kvm_s390_skeys
3515 :Returns: 0 on success, negative value on error
3517 This ioctl is used to set guest storage key values on the s390
3518 architecture. The ioctl takes parameters via the kvm_s390_skeys struct.
3519 See section on KVM_S390_GET_SKEYS for struct definition.
3521 The start_gfn field is the number of the first guest frame whose storage keys
3524 The count field is the number of consecutive frames (starting from start_gfn)
3525 whose storage keys to get. The count field must be at least 1 and the maximum
3526 allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range
3527 will cause the ioctl to return -EINVAL.
3529 The skeydata_addr field is the address to a buffer containing count bytes of
3530 storage keys. Each byte in the buffer will be set as the storage key for a
3531 single frame starting at start_gfn for count frames.
3533 Note: If any architecturally invalid key value is found in the given data then
3534 the ioctl will return -EINVAL.
3539 :Capability: KVM_CAP_S390_INJECT_IRQ
3540 :Architectures: s390
3542 :Parameters: struct kvm_s390_irq (in)
3543 :Returns: 0 on success, -1 on error
3548 ====== =================================================================
3549 EINVAL interrupt type is invalid
3550 type is KVM_S390_SIGP_STOP and flag parameter is invalid value,
3551 type is KVM_S390_INT_EXTERNAL_CALL and code is bigger
3552 than the maximum of VCPUs
3553 EBUSY type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped,
3554 type is KVM_S390_SIGP_STOP and a stop irq is already pending,
3555 type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt
3557 ====== =================================================================
3559 Allows to inject an interrupt to the guest.
3561 Using struct kvm_s390_irq as a parameter allows
3562 to inject additional payload which is not
3563 possible via KVM_S390_INTERRUPT.
3565 Interrupt parameters are passed via kvm_s390_irq::
3567 struct kvm_s390_irq {
3570 struct kvm_s390_io_info io;
3571 struct kvm_s390_ext_info ext;
3572 struct kvm_s390_pgm_info pgm;
3573 struct kvm_s390_emerg_info emerg;
3574 struct kvm_s390_extcall_info extcall;
3575 struct kvm_s390_prefix_info prefix;
3576 struct kvm_s390_stop_info stop;
3577 struct kvm_s390_mchk_info mchk;
3582 type can be one of the following:
3584 - KVM_S390_SIGP_STOP - sigp stop; parameter in .stop
3585 - KVM_S390_PROGRAM_INT - program check; parameters in .pgm
3586 - KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix
3587 - KVM_S390_RESTART - restart; no parameters
3588 - KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters
3589 - KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters
3590 - KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg
3591 - KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall
3592 - KVM_S390_MCHK - machine check interrupt; parameters in .mchk
3594 This is an asynchronous vcpu ioctl and can be invoked from any thread.
3596 4.94 KVM_S390_GET_IRQ_STATE
3597 ---------------------------
3599 :Capability: KVM_CAP_S390_IRQ_STATE
3600 :Architectures: s390
3602 :Parameters: struct kvm_s390_irq_state (out)
3603 :Returns: >= number of bytes copied into buffer,
3604 -EINVAL if buffer size is 0,
3605 -ENOBUFS if buffer size is too small to fit all pending interrupts,
3606 -EFAULT if the buffer address was invalid
3608 This ioctl allows userspace to retrieve the complete state of all currently
3609 pending interrupts in a single buffer. Use cases include migration
3610 and introspection. The parameter structure contains the address of a
3611 userspace buffer and its length::
3613 struct kvm_s390_irq_state {
3615 __u32 flags; /* will stay unused for compatibility reasons */
3617 __u32 reserved[4]; /* will stay unused for compatibility reasons */
3620 Userspace passes in the above struct and for each pending interrupt a
3621 struct kvm_s390_irq is copied to the provided buffer.
3623 The structure contains a flags and a reserved field for future extensions. As
3624 the kernel never checked for flags == 0 and QEMU never pre-zeroed flags and
3625 reserved, these fields can not be used in the future without breaking
3628 If -ENOBUFS is returned the buffer provided was too small and userspace
3629 may retry with a bigger buffer.
3631 4.95 KVM_S390_SET_IRQ_STATE
3632 ---------------------------
3634 :Capability: KVM_CAP_S390_IRQ_STATE
3635 :Architectures: s390
3637 :Parameters: struct kvm_s390_irq_state (in)
3638 :Returns: 0 on success,
3639 -EFAULT if the buffer address was invalid,
3640 -EINVAL for an invalid buffer length (see below),
3641 -EBUSY if there were already interrupts pending,
3642 errors occurring when actually injecting the
3643 interrupt. See KVM_S390_IRQ.
3645 This ioctl allows userspace to set the complete state of all cpu-local
3646 interrupts currently pending for the vcpu. It is intended for restoring
3647 interrupt state after a migration. The input parameter is a userspace buffer
3648 containing a struct kvm_s390_irq_state::
3650 struct kvm_s390_irq_state {
3652 __u32 flags; /* will stay unused for compatibility reasons */
3654 __u32 reserved[4]; /* will stay unused for compatibility reasons */
3657 The restrictions for flags and reserved apply as well.
3658 (see KVM_S390_GET_IRQ_STATE)
3660 The userspace memory referenced by buf contains a struct kvm_s390_irq
3661 for each interrupt to be injected into the guest.
3662 If one of the interrupts could not be injected for some reason the
3665 len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0
3666 and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq),
3667 which is the maximum number of possibly pending cpu-local interrupts.
3672 :Capability: KVM_CAP_X86_SMM
3676 :Returns: 0 on success, -1 on error
3678 Queues an SMI on the thread's vcpu.
3680 4.97 KVM_CAP_PPC_MULTITCE
3681 -------------------------
3683 :Capability: KVM_CAP_PPC_MULTITCE
3687 This capability means the kernel is capable of handling hypercalls
3688 H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user
3689 space. This significantly accelerates DMA operations for PPC KVM guests.
3690 User space should expect that its handlers for these hypercalls
3691 are not going to be called if user space previously registered LIOBN
3692 in KVM (via KVM_CREATE_SPAPR_TCE or similar calls).
3694 In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest,
3695 user space might have to advertise it for the guest. For example,
3696 IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is
3697 present in the "ibm,hypertas-functions" device-tree property.
3699 The hypercalls mentioned above may or may not be processed successfully
3700 in the kernel based fast path. If they can not be handled by the kernel,
3701 they will get passed on to user space. So user space still has to have
3702 an implementation for these despite the in kernel acceleration.
3704 This capability is always enabled.
3706 4.98 KVM_CREATE_SPAPR_TCE_64
3707 ----------------------------
3709 :Capability: KVM_CAP_SPAPR_TCE_64
3710 :Architectures: powerpc
3712 :Parameters: struct kvm_create_spapr_tce_64 (in)
3713 :Returns: file descriptor for manipulating the created TCE table
3715 This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit
3716 windows, described in 4.62 KVM_CREATE_SPAPR_TCE
3718 This capability uses extended struct in ioctl interface::
3720 /* for KVM_CAP_SPAPR_TCE_64 */
3721 struct kvm_create_spapr_tce_64 {
3725 __u64 offset; /* in pages */
3726 __u64 size; /* in pages */
3729 The aim of extension is to support an additional bigger DMA window with
3730 a variable page size.
3731 KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and
3732 a bus offset of the corresponding DMA window, @size and @offset are numbers
3735 @flags are not used at the moment.
3737 The rest of functionality is identical to KVM_CREATE_SPAPR_TCE.
3739 4.99 KVM_REINJECT_CONTROL
3740 -------------------------
3742 :Capability: KVM_CAP_REINJECT_CONTROL
3745 :Parameters: struct kvm_reinject_control (in)
3746 :Returns: 0 on success,
3747 -EFAULT if struct kvm_reinject_control cannot be read,
3748 -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier.
3750 i8254 (PIT) has two modes, reinject and !reinject. The default is reinject,
3751 where KVM queues elapsed i8254 ticks and monitors completion of interrupt from
3752 vector(s) that i8254 injects. Reinject mode dequeues a tick and injects its
3753 interrupt whenever there isn't a pending interrupt from i8254.
3754 !reinject mode injects an interrupt as soon as a tick arrives.
3758 struct kvm_reinject_control {
3763 pit_reinject = 0 (!reinject mode) is recommended, unless running an old
3764 operating system that uses the PIT for timing (e.g. Linux 2.4.x).
3766 4.100 KVM_PPC_CONFIGURE_V3_MMU
3767 ------------------------------
3769 :Capability: KVM_CAP_PPC_RADIX_MMU or KVM_CAP_PPC_HASH_MMU_V3
3772 :Parameters: struct kvm_ppc_mmuv3_cfg (in)
3773 :Returns: 0 on success,
3774 -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read,
3775 -EINVAL if the configuration is invalid
3777 This ioctl controls whether the guest will use radix or HPT (hashed
3778 page table) translation, and sets the pointer to the process table for
3783 struct kvm_ppc_mmuv3_cfg {
3785 __u64 process_table;
3788 There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and
3789 KVM_PPC_MMUV3_GTSE. KVM_PPC_MMUV3_RADIX, if set, configures the guest
3790 to use radix tree translation, and if clear, to use HPT translation.
3791 KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest
3792 to be able to use the global TLB and SLB invalidation instructions;
3793 if clear, the guest may not use these instructions.
3795 The process_table field specifies the address and size of the guest
3796 process table, which is in the guest's space. This field is formatted
3797 as the second doubleword of the partition table entry, as defined in
3798 the Power ISA V3.00, Book III section 5.7.6.1.
3800 4.101 KVM_PPC_GET_RMMU_INFO
3801 ---------------------------
3803 :Capability: KVM_CAP_PPC_RADIX_MMU
3806 :Parameters: struct kvm_ppc_rmmu_info (out)
3807 :Returns: 0 on success,
3808 -EFAULT if struct kvm_ppc_rmmu_info cannot be written,
3809 -EINVAL if no useful information can be returned
3811 This ioctl returns a structure containing two things: (a) a list
3812 containing supported radix tree geometries, and (b) a list that maps
3813 page sizes to put in the "AP" (actual page size) field for the tlbie
3814 (TLB invalidate entry) instruction.
3818 struct kvm_ppc_rmmu_info {
3819 struct kvm_ppc_radix_geom {
3824 __u32 ap_encodings[8];
3827 The geometries[] field gives up to 8 supported geometries for the
3828 radix page table, in terms of the log base 2 of the smallest page
3829 size, and the number of bits indexed at each level of the tree, from
3830 the PTE level up to the PGD level in that order. Any unused entries
3831 will have 0 in the page_shift field.
3833 The ap_encodings gives the supported page sizes and their AP field
3834 encodings, encoded with the AP value in the top 3 bits and the log
3835 base 2 of the page size in the bottom 6 bits.
3837 4.102 KVM_PPC_RESIZE_HPT_PREPARE
3838 --------------------------------
3840 :Capability: KVM_CAP_SPAPR_RESIZE_HPT
3841 :Architectures: powerpc
3843 :Parameters: struct kvm_ppc_resize_hpt (in)
3844 :Returns: 0 on successful completion,
3845 >0 if a new HPT is being prepared, the value is an estimated
3846 number of milliseconds until preparation is complete,
3847 -EFAULT if struct kvm_reinject_control cannot be read,
3848 -EINVAL if the supplied shift or flags are invalid,
3849 -ENOMEM if unable to allocate the new HPT,
3850 -ENOSPC if there was a hash collision
3854 struct kvm_ppc_rmmu_info {
3855 struct kvm_ppc_radix_geom {
3860 __u32 ap_encodings[8];
3863 The geometries[] field gives up to 8 supported geometries for the
3864 radix page table, in terms of the log base 2 of the smallest page
3865 size, and the number of bits indexed at each level of the tree, from
3866 the PTE level up to the PGD level in that order. Any unused entries
3867 will have 0 in the page_shift field.
3869 The ap_encodings gives the supported page sizes and their AP field
3870 encodings, encoded with the AP value in the top 3 bits and the log
3871 base 2 of the page size in the bottom 6 bits.
3873 4.102 KVM_PPC_RESIZE_HPT_PREPARE
3874 --------------------------------
3876 :Capability: KVM_CAP_SPAPR_RESIZE_HPT
3877 :Architectures: powerpc
3879 :Parameters: struct kvm_ppc_resize_hpt (in)
3880 :Returns: 0 on successful completion,
3881 >0 if a new HPT is being prepared, the value is an estimated
3882 number of milliseconds until preparation is complete,
3883 -EFAULT if struct kvm_reinject_control cannot be read,
3884 -EINVAL if the supplied shift or flags are invalid,when moving existing
3885 HPT entries to the new HPT,
3886 -EIO on other error conditions
3888 Used to implement the PAPR extension for runtime resizing of a guest's
3889 Hashed Page Table (HPT). Specifically this starts, stops or monitors
3890 the preparation of a new potential HPT for the guest, essentially
3891 implementing the H_RESIZE_HPT_PREPARE hypercall.
3893 If called with shift > 0 when there is no pending HPT for the guest,
3894 this begins preparation of a new pending HPT of size 2^(shift) bytes.
3895 It then returns a positive integer with the estimated number of
3896 milliseconds until preparation is complete.
3898 If called when there is a pending HPT whose size does not match that
3899 requested in the parameters, discards the existing pending HPT and
3900 creates a new one as above.
3902 If called when there is a pending HPT of the size requested, will:
3904 * If preparation of the pending HPT is already complete, return 0
3905 * If preparation of the pending HPT has failed, return an error
3906 code, then discard the pending HPT.
3907 * If preparation of the pending HPT is still in progress, return an
3908 estimated number of milliseconds until preparation is complete.
3910 If called with shift == 0, discards any currently pending HPT and
3911 returns 0 (i.e. cancels any in-progress preparation).
3913 flags is reserved for future expansion, currently setting any bits in
3914 flags will result in an -EINVAL.
3916 Normally this will be called repeatedly with the same parameters until
3917 it returns <= 0. The first call will initiate preparation, subsequent
3918 ones will monitor preparation until it completes or fails.
3922 struct kvm_ppc_resize_hpt {
3928 4.103 KVM_PPC_RESIZE_HPT_COMMIT
3929 -------------------------------
3931 :Capability: KVM_CAP_SPAPR_RESIZE_HPT
3932 :Architectures: powerpc
3934 :Parameters: struct kvm_ppc_resize_hpt (in)
3935 :Returns: 0 on successful completion,
3936 -EFAULT if struct kvm_reinject_control cannot be read,
3937 -EINVAL if the supplied shift or flags are invalid,
3938 -ENXIO is there is no pending HPT, or the pending HPT doesn't
3939 have the requested size,
3940 -EBUSY if the pending HPT is not fully prepared,
3941 -ENOSPC if there was a hash collision when moving existing
3942 HPT entries to the new HPT,
3943 -EIO on other error conditions
3945 Used to implement the PAPR extension for runtime resizing of a guest's
3946 Hashed Page Table (HPT). Specifically this requests that the guest be
3947 transferred to working with the new HPT, essentially implementing the
3948 H_RESIZE_HPT_COMMIT hypercall.
3950 This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has
3951 returned 0 with the same parameters. In other cases
3952 KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or
3953 -EBUSY, though others may be possible if the preparation was started,
3956 This will have undefined effects on the guest if it has not already
3957 placed itself in a quiescent state where no vcpu will make MMU enabled
3960 On succsful completion, the pending HPT will become the guest's active
3961 HPT and the previous HPT will be discarded.
3963 On failure, the guest will still be operating on its previous HPT.
3967 struct kvm_ppc_resize_hpt {
3973 4.104 KVM_X86_GET_MCE_CAP_SUPPORTED
3974 -----------------------------------
3976 :Capability: KVM_CAP_MCE
3979 :Parameters: u64 mce_cap (out)
3980 :Returns: 0 on success, -1 on error
3982 Returns supported MCE capabilities. The u64 mce_cap parameter
3983 has the same format as the MSR_IA32_MCG_CAP register. Supported
3984 capabilities will have the corresponding bits set.
3986 4.105 KVM_X86_SETUP_MCE
3987 -----------------------
3989 :Capability: KVM_CAP_MCE
3992 :Parameters: u64 mcg_cap (in)
3993 :Returns: 0 on success,
3994 -EFAULT if u64 mcg_cap cannot be read,
3995 -EINVAL if the requested number of banks is invalid,
3996 -EINVAL if requested MCE capability is not supported.
3998 Initializes MCE support for use. The u64 mcg_cap parameter
3999 has the same format as the MSR_IA32_MCG_CAP register and
4000 specifies which capabilities should be enabled. The maximum
4001 supported number of error-reporting banks can be retrieved when
4002 checking for KVM_CAP_MCE. The supported capabilities can be
4003 retrieved with KVM_X86_GET_MCE_CAP_SUPPORTED.
4005 4.106 KVM_X86_SET_MCE
4006 ---------------------
4008 :Capability: KVM_CAP_MCE
4011 :Parameters: struct kvm_x86_mce (in)
4012 :Returns: 0 on success,
4013 -EFAULT if struct kvm_x86_mce cannot be read,
4014 -EINVAL if the bank number is invalid,
4015 -EINVAL if VAL bit is not set in status field.
4017 Inject a machine check error (MCE) into the guest. The input
4020 struct kvm_x86_mce {
4030 If the MCE being reported is an uncorrected error, KVM will
4031 inject it as an MCE exception into the guest. If the guest
4032 MCG_STATUS register reports that an MCE is in progress, KVM
4033 causes an KVM_EXIT_SHUTDOWN vmexit.
4035 Otherwise, if the MCE is a corrected error, KVM will just
4036 store it in the corresponding bank (provided this bank is
4037 not holding a previously reported uncorrected error).
4039 4.107 KVM_S390_GET_CMMA_BITS
4040 ----------------------------
4042 :Capability: KVM_CAP_S390_CMMA_MIGRATION
4043 :Architectures: s390
4045 :Parameters: struct kvm_s390_cmma_log (in, out)
4046 :Returns: 0 on success, a negative value on error
4048 This ioctl is used to get the values of the CMMA bits on the s390
4049 architecture. It is meant to be used in two scenarios:
4051 - During live migration to save the CMMA values. Live migration needs
4052 to be enabled via the KVM_REQ_START_MIGRATION VM property.
4053 - To non-destructively peek at the CMMA values, with the flag
4054 KVM_S390_CMMA_PEEK set.
4056 The ioctl takes parameters via the kvm_s390_cmma_log struct. The desired
4057 values are written to a buffer whose location is indicated via the "values"
4058 member in the kvm_s390_cmma_log struct. The values in the input struct are
4059 also updated as needed.
4061 Each CMMA value takes up one byte.
4065 struct kvm_s390_cmma_log {
4076 start_gfn is the number of the first guest frame whose CMMA values are
4079 count is the length of the buffer in bytes,
4081 values points to the buffer where the result will be written to.
4083 If count is greater than KVM_S390_SKEYS_MAX, then it is considered to be
4084 KVM_S390_SKEYS_MAX. KVM_S390_SKEYS_MAX is re-used for consistency with
4087 The result is written in the buffer pointed to by the field values, and
4088 the values of the input parameter are updated as follows.
4090 Depending on the flags, different actions are performed. The only
4091 supported flag so far is KVM_S390_CMMA_PEEK.
4093 The default behaviour if KVM_S390_CMMA_PEEK is not set is:
4094 start_gfn will indicate the first page frame whose CMMA bits were dirty.
4095 It is not necessarily the same as the one passed as input, as clean pages
4098 count will indicate the number of bytes actually written in the buffer.
4099 It can (and very often will) be smaller than the input value, since the
4100 buffer is only filled until 16 bytes of clean values are found (which
4101 are then not copied in the buffer). Since a CMMA migration block needs
4102 the base address and the length, for a total of 16 bytes, we will send
4103 back some clean data if there is some dirty data afterwards, as long as
4104 the size of the clean data does not exceed the size of the header. This
4105 allows to minimize the amount of data to be saved or transferred over
4106 the network at the expense of more roundtrips to userspace. The next
4107 invocation of the ioctl will skip over all the clean values, saving
4108 potentially more than just the 16 bytes we found.
4110 If KVM_S390_CMMA_PEEK is set:
4111 the existing storage attributes are read even when not in migration
4112 mode, and no other action is performed;
4114 the output start_gfn will be equal to the input start_gfn,
4116 the output count will be equal to the input count, except if the end of
4117 memory has been reached.
4120 the field "remaining" will indicate the total number of dirty CMMA values
4121 still remaining, or 0 if KVM_S390_CMMA_PEEK is set and migration mode is
4126 values points to the userspace buffer where the result will be stored.
4128 This ioctl can fail with -ENOMEM if not enough memory can be allocated to
4129 complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
4130 KVM_S390_CMMA_PEEK is not set but migration mode was not enabled, with
4131 -EFAULT if the userspace address is invalid or if no page table is
4132 present for the addresses (e.g. when using hugepages).
4134 4.108 KVM_S390_SET_CMMA_BITS
4135 ----------------------------
4137 :Capability: KVM_CAP_S390_CMMA_MIGRATION
4138 :Architectures: s390
4140 :Parameters: struct kvm_s390_cmma_log (in)
4141 :Returns: 0 on success, a negative value on error
4143 This ioctl is used to set the values of the CMMA bits on the s390
4144 architecture. It is meant to be used during live migration to restore
4145 the CMMA values, but there are no restrictions on its use.
4146 The ioctl takes parameters via the kvm_s390_cmma_values struct.
4147 Each CMMA value takes up one byte.
4151 struct kvm_s390_cmma_log {
4162 start_gfn indicates the starting guest frame number,
4164 count indicates how many values are to be considered in the buffer,
4166 flags is not used and must be 0.
4168 mask indicates which PGSTE bits are to be considered.
4170 remaining is not used.
4172 values points to the buffer in userspace where to store the values.
4174 This ioctl can fail with -ENOMEM if not enough memory can be allocated to
4175 complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
4176 the count field is too large (e.g. more than KVM_S390_CMMA_SIZE_MAX) or
4177 if the flags field was not 0, with -EFAULT if the userspace address is
4178 invalid, if invalid pages are written to (e.g. after the end of memory)
4179 or if no page table is present for the addresses (e.g. when using
4182 4.109 KVM_PPC_GET_CPU_CHAR
4183 --------------------------
4185 :Capability: KVM_CAP_PPC_GET_CPU_CHAR
4186 :Architectures: powerpc
4188 :Parameters: struct kvm_ppc_cpu_char (out)
4189 :Returns: 0 on successful completion,
4190 -EFAULT if struct kvm_ppc_cpu_char cannot be written
4192 This ioctl gives userspace information about certain characteristics
4193 of the CPU relating to speculative execution of instructions and
4194 possible information leakage resulting from speculative execution (see
4195 CVE-2017-5715, CVE-2017-5753 and CVE-2017-5754). The information is
4196 returned in struct kvm_ppc_cpu_char, which looks like this::
4198 struct kvm_ppc_cpu_char {
4199 __u64 character; /* characteristics of the CPU */
4200 __u64 behaviour; /* recommended software behaviour */
4201 __u64 character_mask; /* valid bits in character */
4202 __u64 behaviour_mask; /* valid bits in behaviour */
4205 For extensibility, the character_mask and behaviour_mask fields
4206 indicate which bits of character and behaviour have been filled in by
4207 the kernel. If the set of defined bits is extended in future then
4208 userspace will be able to tell whether it is running on a kernel that
4209 knows about the new bits.
4211 The character field describes attributes of the CPU which can help
4212 with preventing inadvertent information disclosure - specifically,
4213 whether there is an instruction to flash-invalidate the L1 data cache
4214 (ori 30,30,0 or mtspr SPRN_TRIG2,rN), whether the L1 data cache is set
4215 to a mode where entries can only be used by the thread that created
4216 them, whether the bcctr[l] instruction prevents speculation, and
4217 whether a speculation barrier instruction (ori 31,31,0) is provided.
4219 The behaviour field describes actions that software should take to
4220 prevent inadvertent information disclosure, and thus describes which
4221 vulnerabilities the hardware is subject to; specifically whether the
4222 L1 data cache should be flushed when returning to user mode from the
4223 kernel, and whether a speculation barrier should be placed between an
4224 array bounds check and the array access.
4226 These fields use the same bit definitions as the new
4227 H_GET_CPU_CHARACTERISTICS hypercall.
4229 4.110 KVM_MEMORY_ENCRYPT_OP
4230 ---------------------------
4235 :Parameters: an opaque platform specific structure (in/out)
4236 :Returns: 0 on success; -1 on error
4238 If the platform supports creating encrypted VMs then this ioctl can be used
4239 for issuing platform-specific memory encryption commands to manage those
4242 Currently, this ioctl is used for issuing Secure Encrypted Virtualization
4243 (SEV) commands on AMD Processors. The SEV commands are defined in
4244 Documentation/virt/kvm/amd-memory-encryption.rst.
4246 4.111 KVM_MEMORY_ENCRYPT_REG_REGION
4247 -----------------------------------
4252 :Parameters: struct kvm_enc_region (in)
4253 :Returns: 0 on success; -1 on error
4255 This ioctl can be used to register a guest memory region which may
4256 contain encrypted data (e.g. guest RAM, SMRAM etc).
4258 It is used in the SEV-enabled guest. When encryption is enabled, a guest
4259 memory region may contain encrypted data. The SEV memory encryption
4260 engine uses a tweak such that two identical plaintext pages, each at
4261 different locations will have differing ciphertexts. So swapping or
4262 moving ciphertext of those pages will not result in plaintext being
4263 swapped. So relocating (or migrating) physical backing pages for the SEV
4264 guest will require some additional steps.
4266 Note: The current SEV key management spec does not provide commands to
4267 swap or migrate (move) ciphertext pages. Hence, for now we pin the guest
4268 memory region registered with the ioctl.
4270 4.112 KVM_MEMORY_ENCRYPT_UNREG_REGION
4271 -------------------------------------
4276 :Parameters: struct kvm_enc_region (in)
4277 :Returns: 0 on success; -1 on error
4279 This ioctl can be used to unregister the guest memory region registered
4280 with KVM_MEMORY_ENCRYPT_REG_REGION ioctl above.
4282 4.113 KVM_HYPERV_EVENTFD
4283 ------------------------
4285 :Capability: KVM_CAP_HYPERV_EVENTFD
4288 :Parameters: struct kvm_hyperv_eventfd (in)
4290 This ioctl (un)registers an eventfd to receive notifications from the guest on
4291 the specified Hyper-V connection id through the SIGNAL_EVENT hypercall, without
4292 causing a user exit. SIGNAL_EVENT hypercall with non-zero event flag number
4293 (bits 24-31) still triggers a KVM_EXIT_HYPERV_HCALL user exit.
4297 struct kvm_hyperv_eventfd {
4304 The conn_id field should fit within 24 bits::
4306 #define KVM_HYPERV_CONN_ID_MASK 0x00ffffff
4308 The acceptable values for the flags field are::
4310 #define KVM_HYPERV_EVENTFD_DEASSIGN (1 << 0)
4312 :Returns: 0 on success,
4313 -EINVAL if conn_id or flags is outside the allowed range,
4314 -ENOENT on deassign if the conn_id isn't registered,
4315 -EEXIST on assign if the conn_id is already registered
4317 4.114 KVM_GET_NESTED_STATE
4318 --------------------------
4320 :Capability: KVM_CAP_NESTED_STATE
4323 :Parameters: struct kvm_nested_state (in/out)
4324 :Returns: 0 on success, -1 on error
4328 ===== =============================================================
4329 E2BIG the total state size exceeds the value of 'size' specified by
4330 the user; the size required will be written into size.
4331 ===== =============================================================
4335 struct kvm_nested_state {
4341 struct kvm_vmx_nested_state_hdr vmx;
4342 struct kvm_svm_nested_state_hdr svm;
4344 /* Pad the header to 128 bytes. */
4349 struct kvm_vmx_nested_state_data vmx[0];
4350 struct kvm_svm_nested_state_data svm[0];
4354 #define KVM_STATE_NESTED_GUEST_MODE 0x00000001
4355 #define KVM_STATE_NESTED_RUN_PENDING 0x00000002
4356 #define KVM_STATE_NESTED_EVMCS 0x00000004
4358 #define KVM_STATE_NESTED_FORMAT_VMX 0
4359 #define KVM_STATE_NESTED_FORMAT_SVM 1
4361 #define KVM_STATE_NESTED_VMX_VMCS_SIZE 0x1000
4363 #define KVM_STATE_NESTED_VMX_SMM_GUEST_MODE 0x00000001
4364 #define KVM_STATE_NESTED_VMX_SMM_VMXON 0x00000002
4366 #define KVM_STATE_VMX_PREEMPTION_TIMER_DEADLINE 0x00000001
4368 struct kvm_vmx_nested_state_hdr {
4377 __u64 preemption_timer_deadline;
4380 struct kvm_vmx_nested_state_data {
4381 __u8 vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE];
4382 __u8 shadow_vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE];
4385 This ioctl copies the vcpu's nested virtualization state from the kernel to
4388 The maximum size of the state can be retrieved by passing KVM_CAP_NESTED_STATE
4389 to the KVM_CHECK_EXTENSION ioctl().
4391 4.115 KVM_SET_NESTED_STATE
4392 --------------------------
4394 :Capability: KVM_CAP_NESTED_STATE
4397 :Parameters: struct kvm_nested_state (in)
4398 :Returns: 0 on success, -1 on error
4400 This copies the vcpu's kvm_nested_state struct from userspace to the kernel.
4401 For the definition of struct kvm_nested_state, see KVM_GET_NESTED_STATE.
4403 4.116 KVM_(UN)REGISTER_COALESCED_MMIO
4404 -------------------------------------
4406 :Capability: KVM_CAP_COALESCED_MMIO (for coalesced mmio)
4407 KVM_CAP_COALESCED_PIO (for coalesced pio)
4410 :Parameters: struct kvm_coalesced_mmio_zone
4411 :Returns: 0 on success, < 0 on error
4413 Coalesced I/O is a performance optimization that defers hardware
4414 register write emulation so that userspace exits are avoided. It is
4415 typically used to reduce the overhead of emulating frequently accessed
4418 When a hardware register is configured for coalesced I/O, write accesses
4419 do not exit to userspace and their value is recorded in a ring buffer
4420 that is shared between kernel and userspace.
4422 Coalesced I/O is used if one or more write accesses to a hardware
4423 register can be deferred until a read or a write to another hardware
4424 register on the same device. This last access will cause a vmexit and
4425 userspace will process accesses from the ring buffer before emulating
4426 it. That will avoid exiting to userspace on repeated writes.
4428 Coalesced pio is based on coalesced mmio. There is little difference
4429 between coalesced mmio and pio except that coalesced pio records accesses
4432 4.117 KVM_CLEAR_DIRTY_LOG (vm ioctl)
4433 ------------------------------------
4435 :Capability: KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
4436 :Architectures: x86, arm, arm64, mips
4438 :Parameters: struct kvm_dirty_log (in)
4439 :Returns: 0 on success, -1 on error
4443 /* for KVM_CLEAR_DIRTY_LOG */
4444 struct kvm_clear_dirty_log {
4449 void __user *dirty_bitmap; /* one bit per page */
4454 The ioctl clears the dirty status of pages in a memory slot, according to
4455 the bitmap that is passed in struct kvm_clear_dirty_log's dirty_bitmap
4456 field. Bit 0 of the bitmap corresponds to page "first_page" in the
4457 memory slot, and num_pages is the size in bits of the input bitmap.
4458 first_page must be a multiple of 64; num_pages must also be a multiple of
4459 64 unless first_page + num_pages is the size of the memory slot. For each
4460 bit that is set in the input bitmap, the corresponding page is marked "clean"
4461 in KVM's dirty bitmap, and dirty tracking is re-enabled for that page
4462 (for example via write-protection, or by clearing the dirty bit in
4463 a page table entry).
4465 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 specifies
4466 the address space for which you want to return the dirty bitmap.
4467 They must be less than the value that KVM_CHECK_EXTENSION returns for
4468 the KVM_CAP_MULTI_ADDRESS_SPACE capability.
4470 This ioctl is mostly useful when KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
4471 is enabled; for more information, see the description of the capability.
4472 However, it can always be used as long as KVM_CHECK_EXTENSION confirms
4473 that KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is present.
4475 4.118 KVM_GET_SUPPORTED_HV_CPUID
4476 --------------------------------
4478 :Capability: KVM_CAP_HYPERV_CPUID (vcpu), KVM_CAP_SYS_HYPERV_CPUID (system)
4480 :Type: system ioctl, vcpu ioctl
4481 :Parameters: struct kvm_cpuid2 (in/out)
4482 :Returns: 0 on success, -1 on error
4489 struct kvm_cpuid_entry2 entries[0];
4492 struct kvm_cpuid_entry2 {
4503 This ioctl returns x86 cpuid features leaves related to Hyper-V emulation in
4504 KVM. Userspace can use the information returned by this ioctl to construct
4505 cpuid information presented to guests consuming Hyper-V enlightenments (e.g.
4506 Windows or Hyper-V guests).
4508 CPUID feature leaves returned by this ioctl are defined by Hyper-V Top Level
4509 Functional Specification (TLFS). These leaves can't be obtained with
4510 KVM_GET_SUPPORTED_CPUID ioctl because some of them intersect with KVM feature
4511 leaves (0x40000000, 0x40000001).
4513 Currently, the following list of CPUID leaves are returned:
4514 - HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS
4515 - HYPERV_CPUID_INTERFACE
4516 - HYPERV_CPUID_VERSION
4517 - HYPERV_CPUID_FEATURES
4518 - HYPERV_CPUID_ENLIGHTMENT_INFO
4519 - HYPERV_CPUID_IMPLEMENT_LIMITS
4520 - HYPERV_CPUID_NESTED_FEATURES
4521 - HYPERV_CPUID_SYNDBG_VENDOR_AND_MAX_FUNCTIONS
4522 - HYPERV_CPUID_SYNDBG_INTERFACE
4523 - HYPERV_CPUID_SYNDBG_PLATFORM_CAPABILITIES
4525 Userspace invokes KVM_GET_SUPPORTED_HV_CPUID by passing a kvm_cpuid2 structure
4526 with the 'nent' field indicating the number of entries in the variable-size
4527 array 'entries'. If the number of entries is too low to describe all Hyper-V
4528 feature leaves, an error (E2BIG) is returned. If the number is more or equal
4529 to the number of Hyper-V feature leaves, the 'nent' field is adjusted to the
4530 number of valid entries in the 'entries' array, which is then filled.
4532 'index' and 'flags' fields in 'struct kvm_cpuid_entry2' are currently reserved,
4533 userspace should not expect to get any particular value there.
4535 Note, vcpu version of KVM_GET_SUPPORTED_HV_CPUID is currently deprecated. Unlike
4536 system ioctl which exposes all supported feature bits unconditionally, vcpu
4537 version has the following quirks:
4538 - HYPERV_CPUID_NESTED_FEATURES leaf and HV_X64_ENLIGHTENED_VMCS_RECOMMENDED
4539 feature bit are only exposed when Enlightened VMCS was previously enabled
4540 on the corresponding vCPU (KVM_CAP_HYPERV_ENLIGHTENED_VMCS).
4541 - HV_STIMER_DIRECT_MODE_AVAILABLE bit is only exposed with in-kernel LAPIC.
4542 (presumes KVM_CREATE_IRQCHIP has already been called).
4544 4.119 KVM_ARM_VCPU_FINALIZE
4545 ---------------------------
4547 :Architectures: arm, arm64
4549 :Parameters: int feature (in)
4550 :Returns: 0 on success, -1 on error
4554 ====== ==============================================================
4555 EPERM feature not enabled, needs configuration, or already finalized
4556 EINVAL feature unknown or not present
4557 ====== ==============================================================
4559 Recognised values for feature:
4561 ===== ===========================================
4562 arm64 KVM_ARM_VCPU_SVE (requires KVM_CAP_ARM_SVE)
4563 ===== ===========================================
4565 Finalizes the configuration of the specified vcpu feature.
4567 The vcpu must already have been initialised, enabling the affected feature, by
4568 means of a successful KVM_ARM_VCPU_INIT call with the appropriate flag set in
4571 For affected vcpu features, this is a mandatory step that must be performed
4572 before the vcpu is fully usable.
4574 Between KVM_ARM_VCPU_INIT and KVM_ARM_VCPU_FINALIZE, the feature may be
4575 configured by use of ioctls such as KVM_SET_ONE_REG. The exact configuration
4576 that should be performaned and how to do it are feature-dependent.
4578 Other calls that depend on a particular feature being finalized, such as
4579 KVM_RUN, KVM_GET_REG_LIST, KVM_GET_ONE_REG and KVM_SET_ONE_REG, will fail with
4580 -EPERM unless the feature has already been finalized by means of a
4581 KVM_ARM_VCPU_FINALIZE call.
4583 See KVM_ARM_VCPU_INIT for details of vcpu features that require finalization
4586 4.120 KVM_SET_PMU_EVENT_FILTER
4587 ------------------------------
4589 :Capability: KVM_CAP_PMU_EVENT_FILTER
4592 :Parameters: struct kvm_pmu_event_filter (in)
4593 :Returns: 0 on success, -1 on error
4597 struct kvm_pmu_event_filter {
4600 __u32 fixed_counter_bitmap;
4606 This ioctl restricts the set of PMU events that the guest can program.
4607 The argument holds a list of events which will be allowed or denied.
4608 The eventsel+umask of each event the guest attempts to program is compared
4609 against the events field to determine whether the guest should have access.
4610 The events field only controls general purpose counters; fixed purpose
4611 counters are controlled by the fixed_counter_bitmap.
4613 No flags are defined yet, the field must be zero.
4615 Valid values for 'action'::
4617 #define KVM_PMU_EVENT_ALLOW 0
4618 #define KVM_PMU_EVENT_DENY 1
4620 4.121 KVM_PPC_SVM_OFF
4621 ---------------------
4624 :Architectures: powerpc
4627 :Returns: 0 on successful completion,
4631 ====== ================================================================
4632 EINVAL if ultravisor failed to terminate the secure guest
4633 ENOMEM if hypervisor failed to allocate new radix page tables for guest
4634 ====== ================================================================
4636 This ioctl is used to turn off the secure mode of the guest or transition
4637 the guest from secure mode to normal mode. This is invoked when the guest
4638 is reset. This has no effect if called for a normal guest.
4640 This ioctl issues an ultravisor call to terminate the secure guest,
4641 unpins the VPA pages and releases all the device pages that are used to
4642 track the secure pages by hypervisor.
4644 4.122 KVM_S390_NORMAL_RESET
4645 ---------------------------
4647 :Capability: KVM_CAP_S390_VCPU_RESETS
4648 :Architectures: s390
4653 This ioctl resets VCPU registers and control structures according to
4654 the cpu reset definition in the POP (Principles Of Operation).
4656 4.123 KVM_S390_INITIAL_RESET
4657 ----------------------------
4660 :Architectures: s390
4665 This ioctl resets VCPU registers and control structures according to
4666 the initial cpu reset definition in the POP. However, the cpu is not
4667 put into ESA mode. This reset is a superset of the normal reset.
4669 4.124 KVM_S390_CLEAR_RESET
4670 --------------------------
4672 :Capability: KVM_CAP_S390_VCPU_RESETS
4673 :Architectures: s390
4678 This ioctl resets VCPU registers and control structures according to
4679 the clear cpu reset definition in the POP. However, the cpu is not put
4680 into ESA mode. This reset is a superset of the initial reset.
4683 4.125 KVM_S390_PV_COMMAND
4684 -------------------------
4686 :Capability: KVM_CAP_S390_PROTECTED
4687 :Architectures: s390
4689 :Parameters: struct kvm_pv_cmd
4690 :Returns: 0 on success, < 0 on error
4695 __u32 cmd; /* Command to be executed */
4696 __u16 rc; /* Ultravisor return code */
4697 __u16 rrc; /* Ultravisor return reason code */
4698 __u64 data; /* Data or address */
4699 __u32 flags; /* flags for future extensions. Must be 0 for now */
4706 Allocate memory and register the VM with the Ultravisor, thereby
4707 donating memory to the Ultravisor that will become inaccessible to
4708 KVM. All existing CPUs are converted to protected ones. After this
4709 command has succeeded, any CPU added via hotplug will become
4710 protected during its creation as well.
4714 ===== =============================
4715 EINTR an unmasked signal is pending
4716 ===== =============================
4720 Deregister the VM from the Ultravisor and reclaim the memory that
4721 had been donated to the Ultravisor, making it usable by the kernel
4722 again. All registered VCPUs are converted back to non-protected
4725 KVM_PV_VM_SET_SEC_PARMS
4726 Pass the image header from VM memory to the Ultravisor in
4727 preparation of image unpacking and verification.
4730 Unpack (protect and decrypt) a page of the encrypted boot image.
4733 Verify the integrity of the unpacked image. Only if this succeeds,
4734 KVM is allowed to start protected VCPUs.
4736 4.126 KVM_X86_SET_MSR_FILTER
4737 ----------------------------
4739 :Capability: KVM_X86_SET_MSR_FILTER
4742 :Parameters: struct kvm_msr_filter
4743 :Returns: 0 on success, < 0 on error
4747 struct kvm_msr_filter_range {
4748 #define KVM_MSR_FILTER_READ (1 << 0)
4749 #define KVM_MSR_FILTER_WRITE (1 << 1)
4751 __u32 nmsrs; /* number of msrs in bitmap */
4752 __u32 base; /* MSR index the bitmap starts at */
4753 __u8 *bitmap; /* a 1 bit allows the operations in flags, 0 denies */
4756 #define KVM_MSR_FILTER_MAX_RANGES 16
4757 struct kvm_msr_filter {
4758 #define KVM_MSR_FILTER_DEFAULT_ALLOW (0 << 0)
4759 #define KVM_MSR_FILTER_DEFAULT_DENY (1 << 0)
4761 struct kvm_msr_filter_range ranges[KVM_MSR_FILTER_MAX_RANGES];
4764 flags values for ``struct kvm_msr_filter_range``:
4766 ``KVM_MSR_FILTER_READ``
4768 Filter read accesses to MSRs using the given bitmap. A 0 in the bitmap
4769 indicates that a read should immediately fail, while a 1 indicates that
4770 a read for a particular MSR should be handled regardless of the default
4773 ``KVM_MSR_FILTER_WRITE``
4775 Filter write accesses to MSRs using the given bitmap. A 0 in the bitmap
4776 indicates that a write should immediately fail, while a 1 indicates that
4777 a write for a particular MSR should be handled regardless of the default
4780 ``KVM_MSR_FILTER_READ | KVM_MSR_FILTER_WRITE``
4782 Filter both read and write accesses to MSRs using the given bitmap. A 0
4783 in the bitmap indicates that both reads and writes should immediately fail,
4784 while a 1 indicates that reads and writes for a particular MSR are not
4785 filtered by this range.
4787 flags values for ``struct kvm_msr_filter``:
4789 ``KVM_MSR_FILTER_DEFAULT_ALLOW``
4791 If no filter range matches an MSR index that is getting accessed, KVM will
4792 fall back to allowing access to the MSR.
4794 ``KVM_MSR_FILTER_DEFAULT_DENY``
4796 If no filter range matches an MSR index that is getting accessed, KVM will
4797 fall back to rejecting access to the MSR. In this mode, all MSRs that should
4798 be processed by KVM need to explicitly be marked as allowed in the bitmaps.
4800 This ioctl allows user space to define up to 16 bitmaps of MSR ranges to
4801 specify whether a certain MSR access should be explicitly filtered for or not.
4803 If this ioctl has never been invoked, MSR accesses are not guarded and the
4804 default KVM in-kernel emulation behavior is fully preserved.
4806 Calling this ioctl with an empty set of ranges (all nmsrs == 0) disables MSR
4807 filtering. In that mode, ``KVM_MSR_FILTER_DEFAULT_DENY`` is invalid and causes
4810 As soon as the filtering is in place, every MSR access is processed through
4811 the filtering except for accesses to the x2APIC MSRs (from 0x800 to 0x8ff);
4812 x2APIC MSRs are always allowed, independent of the ``default_allow`` setting,
4813 and their behavior depends on the ``X2APIC_ENABLE`` bit of the APIC base
4816 If a bit is within one of the defined ranges, read and write accesses are
4817 guarded by the bitmap's value for the MSR index if the kind of access
4818 is included in the ``struct kvm_msr_filter_range`` flags. If no range
4819 cover this particular access, the behavior is determined by the flags
4820 field in the kvm_msr_filter struct: ``KVM_MSR_FILTER_DEFAULT_ALLOW``
4821 and ``KVM_MSR_FILTER_DEFAULT_DENY``.
4823 Each bitmap range specifies a range of MSRs to potentially allow access on.
4824 The range goes from MSR index [base .. base+nmsrs]. The flags field
4825 indicates whether reads, writes or both reads and writes are filtered
4826 by setting a 1 bit in the bitmap for the corresponding MSR index.
4828 If an MSR access is not permitted through the filtering, it generates a
4829 #GP inside the guest. When combined with KVM_CAP_X86_USER_SPACE_MSR, that
4830 allows user space to deflect and potentially handle various MSR accesses
4833 If a vCPU is in running state while this ioctl is invoked, the vCPU may
4834 experience inconsistent filtering behavior on MSR accesses.
4837 5. The kvm_run structure
4838 ========================
4840 Application code obtains a pointer to the kvm_run structure by
4841 mmap()ing a vcpu fd. From that point, application code can control
4842 execution by changing fields in kvm_run prior to calling the KVM_RUN
4843 ioctl, and obtain information about the reason KVM_RUN returned by
4844 looking up structure members.
4850 __u8 request_interrupt_window;
4852 Request that KVM_RUN return when it becomes possible to inject external
4853 interrupts into the guest. Useful in conjunction with KVM_INTERRUPT.
4857 __u8 immediate_exit;
4859 This field is polled once when KVM_RUN starts; if non-zero, KVM_RUN
4860 exits immediately, returning -EINTR. In the common scenario where a
4861 signal is used to "kick" a VCPU out of KVM_RUN, this field can be used
4862 to avoid usage of KVM_SET_SIGNAL_MASK, which has worse scalability.
4863 Rather than blocking the signal outside KVM_RUN, userspace can set up
4864 a signal handler that sets run->immediate_exit to a non-zero value.
4866 This field is ignored if KVM_CAP_IMMEDIATE_EXIT is not available.
4875 When KVM_RUN has returned successfully (return value 0), this informs
4876 application code why KVM_RUN has returned. Allowable values for this
4877 field are detailed below.
4881 __u8 ready_for_interrupt_injection;
4883 If request_interrupt_window has been specified, this field indicates
4884 an interrupt can be injected now with KVM_INTERRUPT.
4890 The value of the current interrupt flag. Only valid if in-kernel
4891 local APIC is not used.
4897 More architecture-specific flags detailing state of the VCPU that may
4898 affect the device's behavior. The only currently defined flag is
4899 KVM_RUN_X86_SMM, which is valid on x86 machines and is set if the
4900 VCPU is in system management mode.
4904 /* in (pre_kvm_run), out (post_kvm_run) */
4907 The value of the cr8 register. Only valid if in-kernel local APIC is
4908 not used. Both input and output.
4914 The value of the APIC BASE msr. Only valid if in-kernel local
4915 APIC is not used. Both input and output.
4920 /* KVM_EXIT_UNKNOWN */
4922 __u64 hardware_exit_reason;
4925 If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown
4926 reasons. Further architecture-specific information is available in
4927 hardware_exit_reason.
4931 /* KVM_EXIT_FAIL_ENTRY */
4933 __u64 hardware_entry_failure_reason;
4934 __u32 cpu; /* if KVM_LAST_CPU */
4937 If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due
4938 to unknown reasons. Further architecture-specific information is
4939 available in hardware_entry_failure_reason.
4943 /* KVM_EXIT_EXCEPTION */
4955 #define KVM_EXIT_IO_IN 0
4956 #define KVM_EXIT_IO_OUT 1
4958 __u8 size; /* bytes */
4961 __u64 data_offset; /* relative to kvm_run start */
4964 If exit_reason is KVM_EXIT_IO, then the vcpu has
4965 executed a port I/O instruction which could not be satisfied by kvm.
4966 data_offset describes where the data is located (KVM_EXIT_IO_OUT) or
4967 where kvm expects application code to place the data for the next
4968 KVM_RUN invocation (KVM_EXIT_IO_IN). Data format is a packed array.
4972 /* KVM_EXIT_DEBUG */
4974 struct kvm_debug_exit_arch arch;
4977 If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event
4978 for which architecture specific information is returned.
4990 If exit_reason is KVM_EXIT_MMIO, then the vcpu has
4991 executed a memory-mapped I/O instruction which could not be satisfied
4992 by kvm. The 'data' member contains the written data if 'is_write' is
4993 true, and should be filled by application code otherwise.
4995 The 'data' member contains, in its first 'len' bytes, the value as it would
4996 appear if the VCPU performed a load or store of the appropriate width directly
5001 For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR,
5002 KVM_EXIT_EPR, KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR the corresponding
5003 operations are complete (and guest state is consistent) only after userspace
5004 has re-entered the kernel with KVM_RUN. The kernel side will first finish
5005 incomplete operations and then check for pending signals. Userspace
5006 can re-enter the guest with an unmasked signal pending to complete
5011 /* KVM_EXIT_HYPERCALL */
5020 Unused. This was once used for 'hypercall to userspace'. To implement
5021 such functionality, use KVM_EXIT_IO (x86) or KVM_EXIT_MMIO (all except s390).
5023 .. note:: KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO.
5027 /* KVM_EXIT_TPR_ACCESS */
5034 To be documented (KVM_TPR_ACCESS_REPORTING).
5038 /* KVM_EXIT_S390_SIEIC */
5041 __u64 mask; /* psw upper half */
5042 __u64 addr; /* psw lower half */
5051 /* KVM_EXIT_S390_RESET */
5052 #define KVM_S390_RESET_POR 1
5053 #define KVM_S390_RESET_CLEAR 2
5054 #define KVM_S390_RESET_SUBSYSTEM 4
5055 #define KVM_S390_RESET_CPU_INIT 8
5056 #define KVM_S390_RESET_IPL 16
5057 __u64 s390_reset_flags;
5063 /* KVM_EXIT_S390_UCONTROL */
5065 __u64 trans_exc_code;
5069 s390 specific. A page fault has occurred for a user controlled virtual
5070 machine (KVM_VM_S390_UNCONTROL) on it's host page table that cannot be
5071 resolved by the kernel.
5072 The program code and the translation exception code that were placed
5073 in the cpu's lowcore are presented here as defined by the z Architecture
5074 Principles of Operation Book in the Chapter for Dynamic Address Translation
5086 Deprecated - was used for 440 KVM.
5095 MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch
5096 hypercalls and exit with this exit struct that contains all the guest gprs.
5098 If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall.
5099 Userspace can now handle the hypercall and when it's done modify the gprs as
5100 necessary. Upon guest entry all guest GPRs will then be replaced by the values
5105 /* KVM_EXIT_PAPR_HCALL */
5112 This is used on 64-bit PowerPC when emulating a pSeries partition,
5113 e.g. with the 'pseries' machine type in qemu. It occurs when the
5114 guest does a hypercall using the 'sc 1' instruction. The 'nr' field
5115 contains the hypercall number (from the guest R3), and 'args' contains
5116 the arguments (from the guest R4 - R12). Userspace should put the
5117 return code in 'ret' and any extra returned values in args[].
5118 The possible hypercalls are defined in the Power Architecture Platform
5119 Requirements (PAPR) document available from www.power.org (free
5120 developer registration required to access it).
5124 /* KVM_EXIT_S390_TSCH */
5126 __u16 subchannel_id;
5127 __u16 subchannel_nr;
5134 s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled
5135 and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O
5136 interrupt for the target subchannel has been dequeued and subchannel_id,
5137 subchannel_nr, io_int_parm and io_int_word contain the parameters for that
5138 interrupt. ipb is needed for instruction parameter decoding.
5147 On FSL BookE PowerPC chips, the interrupt controller has a fast patch
5148 interrupt acknowledge path to the core. When the core successfully
5149 delivers an interrupt, it automatically populates the EPR register with
5150 the interrupt vector number and acknowledges the interrupt inside
5151 the interrupt controller.
5153 In case the interrupt controller lives in user space, we need to do
5154 the interrupt acknowledge cycle through it to fetch the next to be
5155 delivered interrupt vector using this exit.
5157 It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an
5158 external interrupt has just been delivered into the guest. User space
5159 should put the acknowledged interrupt vector into the 'epr' field.
5163 /* KVM_EXIT_SYSTEM_EVENT */
5165 #define KVM_SYSTEM_EVENT_SHUTDOWN 1
5166 #define KVM_SYSTEM_EVENT_RESET 2
5167 #define KVM_SYSTEM_EVENT_CRASH 3
5172 If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered
5173 a system-level event using some architecture specific mechanism (hypercall
5174 or some special instruction). In case of ARM/ARM64, this is triggered using
5175 HVC instruction based PSCI call from the vcpu. The 'type' field describes
5176 the system-level event type. The 'flags' field describes architecture
5177 specific flags for the system-level event.
5179 Valid values for 'type' are:
5181 - KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the
5182 VM. Userspace is not obliged to honour this, and if it does honour
5183 this does not need to destroy the VM synchronously (ie it may call
5184 KVM_RUN again before shutdown finally occurs).
5185 - KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM.
5186 As with SHUTDOWN, userspace can choose to ignore the request, or
5187 to schedule the reset to occur in the future and may call KVM_RUN again.
5188 - KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest
5189 has requested a crash condition maintenance. Userspace can choose
5190 to ignore the request, or to gather VM memory core dump and/or
5191 reset/shutdown of the VM.
5195 /* KVM_EXIT_IOAPIC_EOI */
5200 Indicates that the VCPU's in-kernel local APIC received an EOI for a
5201 level-triggered IOAPIC interrupt. This exit only triggers when the
5202 IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled);
5203 the userspace IOAPIC should process the EOI and retrigger the interrupt if
5204 it is still asserted. Vector is the LAPIC interrupt vector for which the
5209 struct kvm_hyperv_exit {
5210 #define KVM_EXIT_HYPERV_SYNIC 1
5211 #define KVM_EXIT_HYPERV_HCALL 2
5212 #define KVM_EXIT_HYPERV_SYNDBG 3
5239 /* KVM_EXIT_HYPERV */
5240 struct kvm_hyperv_exit hyperv;
5242 Indicates that the VCPU exits into userspace to process some tasks
5243 related to Hyper-V emulation.
5245 Valid values for 'type' are:
5247 - KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about
5249 Hyper-V SynIC state change. Notification is used to remap SynIC
5250 event/message pages and to enable/disable SynIC messages/events processing
5253 - KVM_EXIT_HYPERV_SYNDBG -- synchronously notify user-space about
5255 Hyper-V Synthetic debugger state change. Notification is used to either update
5256 the pending_page location or to send a control command (send the buffer located
5257 in send_page or recv a buffer to recv_page).
5261 /* KVM_EXIT_ARM_NISV */
5267 Used on arm and arm64 systems. If a guest accesses memory not in a memslot,
5268 KVM will typically return to userspace and ask it to do MMIO emulation on its
5269 behalf. However, for certain classes of instructions, no instruction decode
5270 (direction, length of memory access) is provided, and fetching and decoding
5271 the instruction from the VM is overly complicated to live in the kernel.
5273 Historically, when this situation occurred, KVM would print a warning and kill
5274 the VM. KVM assumed that if the guest accessed non-memslot memory, it was
5275 trying to do I/O, which just couldn't be emulated, and the warning message was
5276 phrased accordingly. However, what happened more often was that a guest bug
5277 caused access outside the guest memory areas which should lead to a more
5278 meaningful warning message and an external abort in the guest, if the access
5279 did not fall within an I/O window.
5281 Userspace implementations can query for KVM_CAP_ARM_NISV_TO_USER, and enable
5282 this capability at VM creation. Once this is done, these types of errors will
5283 instead return to userspace with KVM_EXIT_ARM_NISV, with the valid bits from
5284 the HSR (arm) and ESR_EL2 (arm64) in the esr_iss field, and the faulting IPA
5285 in the fault_ipa field. Userspace can either fix up the access if it's
5286 actually an I/O access by decoding the instruction from guest memory (if it's
5287 very brave) and continue executing the guest, or it can decide to suspend,
5288 dump, or restart the guest.
5290 Note that KVM does not skip the faulting instruction as it does for
5291 KVM_EXIT_MMIO, but userspace has to emulate any change to the processing state
5292 if it decides to decode and emulate the instruction.
5296 /* KVM_EXIT_X86_RDMSR / KVM_EXIT_X86_WRMSR */
5298 __u8 error; /* user -> kernel */
5300 __u32 reason; /* kernel -> user */
5301 __u32 index; /* kernel -> user */
5302 __u64 data; /* kernel <-> user */
5305 Used on x86 systems. When the VM capability KVM_CAP_X86_USER_SPACE_MSR is
5306 enabled, MSR accesses to registers that would invoke a #GP by KVM kernel code
5307 will instead trigger a KVM_EXIT_X86_RDMSR exit for reads and KVM_EXIT_X86_WRMSR
5310 The "reason" field specifies why the MSR trap occurred. User space will only
5311 receive MSR exit traps when a particular reason was requested during through
5312 ENABLE_CAP. Currently valid exit reasons are:
5314 KVM_MSR_EXIT_REASON_UNKNOWN - access to MSR that is unknown to KVM
5315 KVM_MSR_EXIT_REASON_INVAL - access to invalid MSRs or reserved bits
5316 KVM_MSR_EXIT_REASON_FILTER - access blocked by KVM_X86_SET_MSR_FILTER
5318 For KVM_EXIT_X86_RDMSR, the "index" field tells user space which MSR the guest
5319 wants to read. To respond to this request with a successful read, user space
5320 writes the respective data into the "data" field and must continue guest
5321 execution to ensure the read data is transferred into guest register state.
5323 If the RDMSR request was unsuccessful, user space indicates that with a "1" in
5324 the "error" field. This will inject a #GP into the guest when the VCPU is
5327 For KVM_EXIT_X86_WRMSR, the "index" field tells user space which MSR the guest
5328 wants to write. Once finished processing the event, user space must continue
5329 vCPU execution. If the MSR write was unsuccessful, user space also sets the
5330 "error" field to "1".
5334 /* Fix the size of the union. */
5339 * shared registers between kvm and userspace.
5340 * kvm_valid_regs specifies the register classes set by the host
5341 * kvm_dirty_regs specified the register classes dirtied by userspace
5342 * struct kvm_sync_regs is architecture specific, as well as the
5343 * bits for kvm_valid_regs and kvm_dirty_regs
5345 __u64 kvm_valid_regs;
5346 __u64 kvm_dirty_regs;
5348 struct kvm_sync_regs regs;
5349 char padding[SYNC_REGS_SIZE_BYTES];
5352 If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access
5353 certain guest registers without having to call SET/GET_*REGS. Thus we can
5354 avoid some system call overhead if userspace has to handle the exit.
5355 Userspace can query the validity of the structure by checking
5356 kvm_valid_regs for specific bits. These bits are architecture specific
5357 and usually define the validity of a groups of registers. (e.g. one bit
5358 for general purpose registers)
5360 Please note that the kernel is allowed to use the kvm_run structure as the
5361 primary storage for certain register types. Therefore, the kernel may use the
5362 values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set.
5370 6. Capabilities that can be enabled on vCPUs
5371 ============================================
5373 There are certain capabilities that change the behavior of the virtual CPU or
5374 the virtual machine when enabled. To enable them, please see section 4.37.
5375 Below you can find a list of capabilities and what their effect on the vCPU or
5376 the virtual machine is when enabling them.
5378 The following information is provided along with the description:
5381 which instruction set architectures provide this ioctl.
5382 x86 includes both i386 and x86_64.
5385 whether this is a per-vcpu or per-vm capability.
5388 what parameters are accepted by the capability.
5391 the return value. General error numbers (EBADF, ENOMEM, EINVAL)
5392 are not detailed, but errors with specific meanings are.
5401 :Returns: 0 on success; -1 on error
5403 This capability enables interception of OSI hypercalls that otherwise would
5404 be treated as normal system calls to be injected into the guest. OSI hypercalls
5405 were invented by Mac-on-Linux to have a standardized communication mechanism
5406 between the guest and the host.
5408 When this capability is enabled, KVM_EXIT_OSI can occur.
5411 6.2 KVM_CAP_PPC_PAPR
5412 --------------------
5417 :Returns: 0 on success; -1 on error
5419 This capability enables interception of PAPR hypercalls. PAPR hypercalls are
5420 done using the hypercall instruction "sc 1".
5422 It also sets the guest privilege level to "supervisor" mode. Usually the guest
5423 runs in "hypervisor" privilege mode with a few missing features.
5425 In addition to the above, it changes the semantics of SDR1. In this mode, the
5426 HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the
5427 HTAB invisible to the guest.
5429 When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur.
5437 :Parameters: args[0] is the address of a struct kvm_config_tlb
5438 :Returns: 0 on success; -1 on error
5442 struct kvm_config_tlb {
5449 Configures the virtual CPU's TLB array, establishing a shared memory area
5450 between userspace and KVM. The "params" and "array" fields are userspace
5451 addresses of mmu-type-specific data structures. The "array_len" field is an
5452 safety mechanism, and should be set to the size in bytes of the memory that
5453 userspace has reserved for the array. It must be at least the size dictated
5454 by "mmu_type" and "params".
5456 While KVM_RUN is active, the shared region is under control of KVM. Its
5457 contents are undefined, and any modification by userspace results in
5458 boundedly undefined behavior.
5460 On return from KVM_RUN, the shared region will reflect the current state of
5461 the guest's TLB. If userspace makes any changes, it must call KVM_DIRTY_TLB
5462 to tell KVM which entries have been changed, prior to calling KVM_RUN again
5465 For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV:
5467 - The "params" field is of type "struct kvm_book3e_206_tlb_params".
5468 - The "array" field points to an array of type "struct
5469 kvm_book3e_206_tlb_entry".
5470 - The array consists of all entries in the first TLB, followed by all
5471 entries in the second TLB.
5472 - Within a TLB, entries are ordered first by increasing set number. Within a
5473 set, entries are ordered by way (increasing ESEL).
5474 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1)
5475 where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value.
5476 - The tsize field of mas1 shall be set to 4K on TLB0, even though the
5477 hardware ignores this value for TLB0.
5479 6.4 KVM_CAP_S390_CSS_SUPPORT
5480 ----------------------------
5482 :Architectures: s390
5485 :Returns: 0 on success; -1 on error
5487 This capability enables support for handling of channel I/O instructions.
5489 TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are
5490 handled in-kernel, while the other I/O instructions are passed to userspace.
5492 When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST
5493 SUBCHANNEL intercepts.
5495 Note that even though this capability is enabled per-vcpu, the complete
5496 virtual machine is affected.
5503 :Parameters: args[0] defines whether the proxy facility is active
5504 :Returns: 0 on success; -1 on error
5506 This capability enables or disables the delivery of interrupts through the
5507 external proxy facility.
5509 When enabled (args[0] != 0), every time the guest gets an external interrupt
5510 delivered, it automatically exits into user space with a KVM_EXIT_EPR exit
5511 to receive the topmost interrupt vector.
5513 When disabled (args[0] == 0), behavior is as if this facility is unsupported.
5515 When this capability is enabled, KVM_EXIT_EPR can occur.
5517 6.6 KVM_CAP_IRQ_MPIC
5518 --------------------
5521 :Parameters: args[0] is the MPIC device fd;
5522 args[1] is the MPIC CPU number for this vcpu
5524 This capability connects the vcpu to an in-kernel MPIC device.
5526 6.7 KVM_CAP_IRQ_XICS
5527 --------------------
5531 :Parameters: args[0] is the XICS device fd;
5532 args[1] is the XICS CPU number (server ID) for this vcpu
5534 This capability connects the vcpu to an in-kernel XICS device.
5536 6.8 KVM_CAP_S390_IRQCHIP
5537 ------------------------
5539 :Architectures: s390
5543 This capability enables the in-kernel irqchip for s390. Please refer to
5544 "4.24 KVM_CREATE_IRQCHIP" for details.
5546 6.9 KVM_CAP_MIPS_FPU
5547 --------------------
5549 :Architectures: mips
5551 :Parameters: args[0] is reserved for future use (should be 0).
5553 This capability allows the use of the host Floating Point Unit by the guest. It
5554 allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is
5555 done the ``KVM_REG_MIPS_FPR_*`` and ``KVM_REG_MIPS_FCR_*`` registers can be
5556 accessed (depending on the current guest FPU register mode), and the Status.FR,
5557 Config5.FRE bits are accessible via the KVM API and also from the guest,
5558 depending on them being supported by the FPU.
5560 6.10 KVM_CAP_MIPS_MSA
5561 ---------------------
5563 :Architectures: mips
5565 :Parameters: args[0] is reserved for future use (should be 0).
5567 This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest.
5568 It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest.
5569 Once this is done the ``KVM_REG_MIPS_VEC_*`` and ``KVM_REG_MIPS_MSA_*``
5570 registers can be accessed, and the Config5.MSAEn bit is accessible via the
5571 KVM API and also from the guest.
5573 6.74 KVM_CAP_SYNC_REGS
5574 ----------------------
5576 :Architectures: s390, x86
5577 :Target: s390: always enabled, x86: vcpu
5579 :Returns: x86: KVM_CHECK_EXTENSION returns a bit-array indicating which register
5581 (bitfields defined in arch/x86/include/uapi/asm/kvm.h).
5583 As described above in the kvm_sync_regs struct info in section 5 (kvm_run):
5584 KVM_CAP_SYNC_REGS "allow[s] userspace to access certain guest registers
5585 without having to call SET/GET_*REGS". This reduces overhead by eliminating
5586 repeated ioctl calls for setting and/or getting register values. This is
5587 particularly important when userspace is making synchronous guest state
5588 modifications, e.g. when emulating and/or intercepting instructions in
5591 For s390 specifics, please refer to the source code.
5595 - the register sets to be copied out to kvm_run are selectable
5596 by userspace (rather that all sets being copied out for every exit).
5597 - vcpu_events are available in addition to regs and sregs.
5599 For x86, the 'kvm_valid_regs' field of struct kvm_run is overloaded to
5600 function as an input bit-array field set by userspace to indicate the
5601 specific register sets to be copied out on the next exit.
5603 To indicate when userspace has modified values that should be copied into
5604 the vCPU, the all architecture bitarray field, 'kvm_dirty_regs' must be set.
5605 This is done using the same bitflags as for the 'kvm_valid_regs' field.
5606 If the dirty bit is not set, then the register set values will not be copied
5607 into the vCPU even if they've been modified.
5609 Unused bitfields in the bitarrays must be set to zero.
5613 struct kvm_sync_regs {
5614 struct kvm_regs regs;
5615 struct kvm_sregs sregs;
5616 struct kvm_vcpu_events events;
5619 6.75 KVM_CAP_PPC_IRQ_XIVE
5620 -------------------------
5624 :Parameters: args[0] is the XIVE device fd;
5625 args[1] is the XIVE CPU number (server ID) for this vcpu
5627 This capability connects the vcpu to an in-kernel XIVE device.
5629 7. Capabilities that can be enabled on VMs
5630 ==========================================
5632 There are certain capabilities that change the behavior of the virtual
5633 machine when enabled. To enable them, please see section 4.37. Below
5634 you can find a list of capabilities and what their effect on the VM
5635 is when enabling them.
5637 The following information is provided along with the description:
5640 which instruction set architectures provide this ioctl.
5641 x86 includes both i386 and x86_64.
5644 what parameters are accepted by the capability.
5647 the return value. General error numbers (EBADF, ENOMEM, EINVAL)
5648 are not detailed, but errors with specific meanings are.
5651 7.1 KVM_CAP_PPC_ENABLE_HCALL
5652 ----------------------------
5655 :Parameters: args[0] is the sPAPR hcall number;
5656 args[1] is 0 to disable, 1 to enable in-kernel handling
5658 This capability controls whether individual sPAPR hypercalls (hcalls)
5659 get handled by the kernel or not. Enabling or disabling in-kernel
5660 handling of an hcall is effective across the VM. On creation, an
5661 initial set of hcalls are enabled for in-kernel handling, which
5662 consists of those hcalls for which in-kernel handlers were implemented
5663 before this capability was implemented. If disabled, the kernel will
5664 not to attempt to handle the hcall, but will always exit to userspace
5665 to handle it. Note that it may not make sense to enable some and
5666 disable others of a group of related hcalls, but KVM does not prevent
5667 userspace from doing that.
5669 If the hcall number specified is not one that has an in-kernel
5670 implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL
5673 7.2 KVM_CAP_S390_USER_SIGP
5674 --------------------------
5676 :Architectures: s390
5679 This capability controls which SIGP orders will be handled completely in user
5680 space. With this capability enabled, all fast orders will be handled completely
5687 - CONDITIONAL EMERGENCY SIGNAL
5689 All other orders will be handled completely in user space.
5691 Only privileged operation exceptions will be checked for in the kernel (or even
5692 in the hardware prior to interception). If this capability is not enabled, the
5693 old way of handling SIGP orders is used (partially in kernel and user space).
5695 7.3 KVM_CAP_S390_VECTOR_REGISTERS
5696 ---------------------------------
5698 :Architectures: s390
5700 :Returns: 0 on success, negative value on error
5702 Allows use of the vector registers introduced with z13 processor, and
5703 provides for the synchronization between host and user space. Will
5704 return -EINVAL if the machine does not support vectors.
5706 7.4 KVM_CAP_S390_USER_STSI
5707 --------------------------
5709 :Architectures: s390
5712 This capability allows post-handlers for the STSI instruction. After
5713 initial handling in the kernel, KVM exits to user space with
5714 KVM_EXIT_S390_STSI to allow user space to insert further data.
5716 Before exiting to userspace, kvm handlers should fill in s390_stsi field of
5728 @addr - guest address of STSI SYSIB
5732 @ar - access register number
5734 KVM handlers should exit to userspace with rc = -EREMOTE.
5736 7.5 KVM_CAP_SPLIT_IRQCHIP
5737 -------------------------
5740 :Parameters: args[0] - number of routes reserved for userspace IOAPICs
5741 :Returns: 0 on success, -1 on error
5743 Create a local apic for each processor in the kernel. This can be used
5744 instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the
5745 IOAPIC and PIC (and also the PIT, even though this has to be enabled
5748 This capability also enables in kernel routing of interrupt requests;
5749 when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are
5750 used in the IRQ routing table. The first args[0] MSI routes are reserved
5751 for the IOAPIC pins. Whenever the LAPIC receives an EOI for these routes,
5752 a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace.
5754 Fails if VCPU has already been created, or if the irqchip is already in the
5755 kernel (i.e. KVM_CREATE_IRQCHIP has already been called).
5760 :Architectures: s390
5763 Allows use of runtime-instrumentation introduced with zEC12 processor.
5764 Will return -EINVAL if the machine does not support runtime-instrumentation.
5765 Will return -EBUSY if a VCPU has already been created.
5767 7.7 KVM_CAP_X2APIC_API
5768 ----------------------
5771 :Parameters: args[0] - features that should be enabled
5772 :Returns: 0 on success, -EINVAL when args[0] contains invalid features
5774 Valid feature flags in args[0] are::
5776 #define KVM_X2APIC_API_USE_32BIT_IDS (1ULL << 0)
5777 #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK (1ULL << 1)
5779 Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of
5780 KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC,
5781 allowing the use of 32-bit APIC IDs. See KVM_CAP_X2APIC_API in their
5782 respective sections.
5784 KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work
5785 in logical mode or with more than 255 VCPUs. Otherwise, KVM treats 0xff
5786 as a broadcast even in x2APIC mode in order to support physical x2APIC
5787 without interrupt remapping. This is undesirable in logical mode,
5788 where 0xff represents CPUs 0-7 in cluster 0.
5790 7.8 KVM_CAP_S390_USER_INSTR0
5791 ----------------------------
5793 :Architectures: s390
5796 With this capability enabled, all illegal instructions 0x0000 (2 bytes) will
5797 be intercepted and forwarded to user space. User space can use this
5798 mechanism e.g. to realize 2-byte software breakpoints. The kernel will
5799 not inject an operating exception for these instructions, user space has
5800 to take care of that.
5802 This capability can be enabled dynamically even if VCPUs were already
5803 created and are running.
5808 :Architectures: s390
5810 :Returns: 0 on success; -EINVAL if the machine does not support
5811 guarded storage; -EBUSY if a VCPU has already been created.
5813 Allows use of guarded storage for the KVM guest.
5815 7.10 KVM_CAP_S390_AIS
5816 ---------------------
5818 :Architectures: s390
5821 Allow use of adapter-interruption suppression.
5822 :Returns: 0 on success; -EBUSY if a VCPU has already been created.
5824 7.11 KVM_CAP_PPC_SMT
5825 --------------------
5828 :Parameters: vsmt_mode, flags
5830 Enabling this capability on a VM provides userspace with a way to set
5831 the desired virtual SMT mode (i.e. the number of virtual CPUs per
5832 virtual core). The virtual SMT mode, vsmt_mode, must be a power of 2
5833 between 1 and 8. On POWER8, vsmt_mode must also be no greater than
5834 the number of threads per subcore for the host. Currently flags must
5835 be 0. A successful call to enable this capability will result in
5836 vsmt_mode being returned when the KVM_CAP_PPC_SMT capability is
5837 subsequently queried for the VM. This capability is only supported by
5838 HV KVM, and can only be set before any VCPUs have been created.
5839 The KVM_CAP_PPC_SMT_POSSIBLE capability indicates which virtual SMT
5840 modes are available.
5842 7.12 KVM_CAP_PPC_FWNMI
5843 ----------------------
5848 With this capability a machine check exception in the guest address
5849 space will cause KVM to exit the guest with NMI exit reason. This
5850 enables QEMU to build error log and branch to guest kernel registered
5851 machine check handling routine. Without this capability KVM will
5852 branch to guests' 0x200 interrupt vector.
5854 7.13 KVM_CAP_X86_DISABLE_EXITS
5855 ------------------------------
5858 :Parameters: args[0] defines which exits are disabled
5859 :Returns: 0 on success, -EINVAL when args[0] contains invalid exits
5861 Valid bits in args[0] are::
5863 #define KVM_X86_DISABLE_EXITS_MWAIT (1 << 0)
5864 #define KVM_X86_DISABLE_EXITS_HLT (1 << 1)
5865 #define KVM_X86_DISABLE_EXITS_PAUSE (1 << 2)
5866 #define KVM_X86_DISABLE_EXITS_CSTATE (1 << 3)
5868 Enabling this capability on a VM provides userspace with a way to no
5869 longer intercept some instructions for improved latency in some
5870 workloads, and is suggested when vCPUs are associated to dedicated
5871 physical CPUs. More bits can be added in the future; userspace can
5872 just pass the KVM_CHECK_EXTENSION result to KVM_ENABLE_CAP to disable
5875 Do not enable KVM_FEATURE_PV_UNHALT if you disable HLT exits.
5877 7.14 KVM_CAP_S390_HPAGE_1M
5878 --------------------------
5880 :Architectures: s390
5882 :Returns: 0 on success, -EINVAL if hpage module parameter was not set
5883 or cmma is enabled, or the VM has the KVM_VM_S390_UCONTROL
5886 With this capability the KVM support for memory backing with 1m pages
5887 through hugetlbfs can be enabled for a VM. After the capability is
5888 enabled, cmma can't be enabled anymore and pfmfi and the storage key
5889 interpretation are disabled. If cmma has already been enabled or the
5890 hpage module parameter is not set to 1, -EINVAL is returned.
5892 While it is generally possible to create a huge page backed VM without
5893 this capability, the VM will not be able to run.
5895 7.15 KVM_CAP_MSR_PLATFORM_INFO
5896 ------------------------------
5899 :Parameters: args[0] whether feature should be enabled or not
5901 With this capability, a guest may read the MSR_PLATFORM_INFO MSR. Otherwise,
5902 a #GP would be raised when the guest tries to access. Currently, this
5903 capability does not enable write permissions of this MSR for the guest.
5905 7.16 KVM_CAP_PPC_NESTED_HV
5906 --------------------------
5910 :Returns: 0 on success, -EINVAL when the implementation doesn't support
5911 nested-HV virtualization.
5913 HV-KVM on POWER9 and later systems allows for "nested-HV"
5914 virtualization, which provides a way for a guest VM to run guests that
5915 can run using the CPU's supervisor mode (privileged non-hypervisor
5916 state). Enabling this capability on a VM depends on the CPU having
5917 the necessary functionality and on the facility being enabled with a
5918 kvm-hv module parameter.
5920 7.17 KVM_CAP_EXCEPTION_PAYLOAD
5921 ------------------------------
5924 :Parameters: args[0] whether feature should be enabled or not
5926 With this capability enabled, CR2 will not be modified prior to the
5927 emulated VM-exit when L1 intercepts a #PF exception that occurs in
5928 L2. Similarly, for kvm-intel only, DR6 will not be modified prior to
5929 the emulated VM-exit when L1 intercepts a #DB exception that occurs in
5930 L2. As a result, when KVM_GET_VCPU_EVENTS reports a pending #PF (or
5931 #DB) exception for L2, exception.has_payload will be set and the
5932 faulting address (or the new DR6 bits*) will be reported in the
5933 exception_payload field. Similarly, when userspace injects a #PF (or
5934 #DB) into L2 using KVM_SET_VCPU_EVENTS, it is expected to set
5935 exception.has_payload and to put the faulting address - or the new DR6
5936 bits\ [#]_ - in the exception_payload field.
5938 This capability also enables exception.pending in struct
5939 kvm_vcpu_events, which allows userspace to distinguish between pending
5940 and injected exceptions.
5943 .. [#] For the new DR6 bits, note that bit 16 is set iff the #DB exception
5946 7.18 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
5948 :Architectures: x86, arm, arm64, mips
5949 :Parameters: args[0] whether feature should be enabled or not
5953 #define KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (1 << 0)
5954 #define KVM_DIRTY_LOG_INITIALLY_SET (1 << 1)
5956 With KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE is set, KVM_GET_DIRTY_LOG will not
5957 automatically clear and write-protect all pages that are returned as dirty.
5958 Rather, userspace will have to do this operation separately using
5959 KVM_CLEAR_DIRTY_LOG.
5961 At the cost of a slightly more complicated operation, this provides better
5962 scalability and responsiveness for two reasons. First,
5963 KVM_CLEAR_DIRTY_LOG ioctl can operate on a 64-page granularity rather
5964 than requiring to sync a full memslot; this ensures that KVM does not
5965 take spinlocks for an extended period of time. Second, in some cases a
5966 large amount of time can pass between a call to KVM_GET_DIRTY_LOG and
5967 userspace actually using the data in the page. Pages can be modified
5968 during this time, which is inefficient for both the guest and userspace:
5969 the guest will incur a higher penalty due to write protection faults,
5970 while userspace can see false reports of dirty pages. Manual reprotection
5971 helps reducing this time, improving guest performance and reducing the
5972 number of dirty log false positives.
5974 With KVM_DIRTY_LOG_INITIALLY_SET set, all the bits of the dirty bitmap
5975 will be initialized to 1 when created. This also improves performance because
5976 dirty logging can be enabled gradually in small chunks on the first call
5977 to KVM_CLEAR_DIRTY_LOG. KVM_DIRTY_LOG_INITIALLY_SET depends on
5978 KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (it is also only available on
5979 x86 and arm64 for now).
5981 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 was previously available under the name
5982 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT, but the implementation had bugs that make
5983 it hard or impossible to use it correctly. The availability of
5984 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 signals that those bugs are fixed.
5985 Userspace should not try to use KVM_CAP_MANUAL_DIRTY_LOG_PROTECT.
5987 7.19 KVM_CAP_PPC_SECURE_GUEST
5988 ------------------------------
5992 This capability indicates that KVM is running on a host that has
5993 ultravisor firmware and thus can support a secure guest. On such a
5994 system, a guest can ask the ultravisor to make it a secure guest,
5995 one whose memory is inaccessible to the host except for pages which
5996 are explicitly requested to be shared with the host. The ultravisor
5997 notifies KVM when a guest requests to become a secure guest, and KVM
5998 has the opportunity to veto the transition.
6000 If present, this capability can be enabled for a VM, meaning that KVM
6001 will allow the transition to secure guest mode. Otherwise KVM will
6002 veto the transition.
6004 7.20 KVM_CAP_HALT_POLL
6005 ----------------------
6009 :Parameters: args[0] is the maximum poll time in nanoseconds
6010 :Returns: 0 on success; -1 on error
6012 This capability overrides the kvm module parameter halt_poll_ns for the
6015 VCPU polling allows a VCPU to poll for wakeup events instead of immediately
6016 scheduling during guest halts. The maximum time a VCPU can spend polling is
6017 controlled by the kvm module parameter halt_poll_ns. This capability allows
6018 the maximum halt time to specified on a per-VM basis, effectively overriding
6019 the module parameter for the target VM.
6021 7.21 KVM_CAP_X86_USER_SPACE_MSR
6022 -------------------------------
6026 :Parameters: args[0] contains the mask of KVM_MSR_EXIT_REASON_* events to report
6027 :Returns: 0 on success; -1 on error
6029 This capability enables trapping of #GP invoking RDMSR and WRMSR instructions
6032 When a guest requests to read or write an MSR, KVM may not implement all MSRs
6033 that are relevant to a respective system. It also does not differentiate by
6036 To allow more fine grained control over MSR handling, user space may enable
6037 this capability. With it enabled, MSR accesses that match the mask specified in
6038 args[0] and trigger a #GP event inside the guest by KVM will instead trigger
6039 KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR exit notifications which user space
6040 can then handle to implement model specific MSR handling and/or user notifications
6041 to inform a user that an MSR was not handled.
6043 8. Other capabilities.
6044 ======================
6046 This section lists capabilities that give information about other
6047 features of the KVM implementation.
6049 8.1 KVM_CAP_PPC_HWRNG
6050 ---------------------
6054 This capability, if KVM_CHECK_EXTENSION indicates that it is
6055 available, means that the kernel has an implementation of the
6056 H_RANDOM hypercall backed by a hardware random-number generator.
6057 If present, the kernel H_RANDOM handler can be enabled for guest use
6058 with the KVM_CAP_PPC_ENABLE_HCALL capability.
6060 8.2 KVM_CAP_HYPERV_SYNIC
6061 ------------------------
6065 This capability, if KVM_CHECK_EXTENSION indicates that it is
6066 available, means that the kernel has an implementation of the
6067 Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is
6068 used to support Windows Hyper-V based guest paravirt drivers(VMBus).
6070 In order to use SynIC, it has to be activated by setting this
6071 capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this
6072 will disable the use of APIC hardware virtualization even if supported
6073 by the CPU, as it's incompatible with SynIC auto-EOI behavior.
6075 8.3 KVM_CAP_PPC_RADIX_MMU
6076 -------------------------
6080 This capability, if KVM_CHECK_EXTENSION indicates that it is
6081 available, means that the kernel can support guests using the
6082 radix MMU defined in Power ISA V3.00 (as implemented in the POWER9
6085 8.4 KVM_CAP_PPC_HASH_MMU_V3
6086 ---------------------------
6090 This capability, if KVM_CHECK_EXTENSION indicates that it is
6091 available, means that the kernel can support guests using the
6092 hashed page table MMU defined in Power ISA V3.00 (as implemented in
6093 the POWER9 processor), including in-memory segment tables.
6098 :Architectures: mips
6100 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
6101 it is available, means that full hardware assisted virtualization capabilities
6102 of the hardware are available for use through KVM. An appropriate
6103 KVM_VM_MIPS_* type must be passed to KVM_CREATE_VM to create a VM which
6106 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
6107 available, it means that the VM is using full hardware assisted virtualization
6108 capabilities of the hardware. This is useful to check after creating a VM with
6109 KVM_VM_MIPS_DEFAULT.
6111 The value returned by KVM_CHECK_EXTENSION should be compared against known
6112 values (see below). All other values are reserved. This is to allow for the
6113 possibility of other hardware assisted virtualization implementations which
6114 may be incompatible with the MIPS VZ ASE.
6116 == ==========================================================================
6117 0 The trap & emulate implementation is in use to run guest code in user
6118 mode. Guest virtual memory segments are rearranged to fit the guest in the
6119 user mode address space.
6121 1 The MIPS VZ ASE is in use, providing full hardware assisted
6122 virtualization, including standard guest virtual memory segments.
6123 == ==========================================================================
6128 :Architectures: mips
6130 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
6131 it is available, means that the trap & emulate implementation is available to
6132 run guest code in user mode, even if KVM_CAP_MIPS_VZ indicates that hardware
6133 assisted virtualisation is also available. KVM_VM_MIPS_TE (0) must be passed
6134 to KVM_CREATE_VM to create a VM which utilises it.
6136 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
6137 available, it means that the VM is using trap & emulate.
6139 8.7 KVM_CAP_MIPS_64BIT
6140 ----------------------
6142 :Architectures: mips
6144 This capability indicates the supported architecture type of the guest, i.e. the
6145 supported register and address width.
6147 The values returned when this capability is checked by KVM_CHECK_EXTENSION on a
6148 kvm VM handle correspond roughly to the CP0_Config.AT register field, and should
6149 be checked specifically against known values (see below). All other values are
6152 == ========================================================================
6153 0 MIPS32 or microMIPS32.
6154 Both registers and addresses are 32-bits wide.
6155 It will only be possible to run 32-bit guest code.
6157 1 MIPS64 or microMIPS64 with access only to 32-bit compatibility segments.
6158 Registers are 64-bits wide, but addresses are 32-bits wide.
6159 64-bit guest code may run but cannot access MIPS64 memory segments.
6160 It will also be possible to run 32-bit guest code.
6162 2 MIPS64 or microMIPS64 with access to all address segments.
6163 Both registers and addresses are 64-bits wide.
6164 It will be possible to run 64-bit or 32-bit guest code.
6165 == ========================================================================
6167 8.9 KVM_CAP_ARM_USER_IRQ
6168 ------------------------
6170 :Architectures: arm, arm64
6172 This capability, if KVM_CHECK_EXTENSION indicates that it is available, means
6173 that if userspace creates a VM without an in-kernel interrupt controller, it
6174 will be notified of changes to the output level of in-kernel emulated devices,
6175 which can generate virtual interrupts, presented to the VM.
6176 For such VMs, on every return to userspace, the kernel
6177 updates the vcpu's run->s.regs.device_irq_level field to represent the actual
6178 output level of the device.
6180 Whenever kvm detects a change in the device output level, kvm guarantees at
6181 least one return to userspace before running the VM. This exit could either
6182 be a KVM_EXIT_INTR or any other exit event, like KVM_EXIT_MMIO. This way,
6183 userspace can always sample the device output level and re-compute the state of
6184 the userspace interrupt controller. Userspace should always check the state
6185 of run->s.regs.device_irq_level on every kvm exit.
6186 The value in run->s.regs.device_irq_level can represent both level and edge
6187 triggered interrupt signals, depending on the device. Edge triggered interrupt
6188 signals will exit to userspace with the bit in run->s.regs.device_irq_level
6189 set exactly once per edge signal.
6191 The field run->s.regs.device_irq_level is available independent of
6192 run->kvm_valid_regs or run->kvm_dirty_regs bits.
6194 If KVM_CAP_ARM_USER_IRQ is supported, the KVM_CHECK_EXTENSION ioctl returns a
6195 number larger than 0 indicating the version of this capability is implemented
6196 and thereby which bits in run->s.regs.device_irq_level can signal values.
6198 Currently the following bits are defined for the device_irq_level bitmap::
6200 KVM_CAP_ARM_USER_IRQ >= 1:
6202 KVM_ARM_DEV_EL1_VTIMER - EL1 virtual timer
6203 KVM_ARM_DEV_EL1_PTIMER - EL1 physical timer
6204 KVM_ARM_DEV_PMU - ARM PMU overflow interrupt signal
6206 Future versions of kvm may implement additional events. These will get
6207 indicated by returning a higher number from KVM_CHECK_EXTENSION and will be
6210 8.10 KVM_CAP_PPC_SMT_POSSIBLE
6211 -----------------------------
6215 Querying this capability returns a bitmap indicating the possible
6216 virtual SMT modes that can be set using KVM_CAP_PPC_SMT. If bit N
6217 (counting from the right) is set, then a virtual SMT mode of 2^N is
6220 8.11 KVM_CAP_HYPERV_SYNIC2
6221 --------------------------
6225 This capability enables a newer version of Hyper-V Synthetic interrupt
6226 controller (SynIC). The only difference with KVM_CAP_HYPERV_SYNIC is that KVM
6227 doesn't clear SynIC message and event flags pages when they are enabled by
6228 writing to the respective MSRs.
6230 8.12 KVM_CAP_HYPERV_VP_INDEX
6231 ----------------------------
6235 This capability indicates that userspace can load HV_X64_MSR_VP_INDEX msr. Its
6236 value is used to denote the target vcpu for a SynIC interrupt. For
6237 compatibilty, KVM initializes this msr to KVM's internal vcpu index. When this
6238 capability is absent, userspace can still query this msr's value.
6240 8.13 KVM_CAP_S390_AIS_MIGRATION
6241 -------------------------------
6243 :Architectures: s390
6246 This capability indicates if the flic device will be able to get/set the
6247 AIS states for migration via the KVM_DEV_FLIC_AISM_ALL attribute and allows
6248 to discover this without having to create a flic device.
6250 8.14 KVM_CAP_S390_PSW
6251 ---------------------
6253 :Architectures: s390
6255 This capability indicates that the PSW is exposed via the kvm_run structure.
6257 8.15 KVM_CAP_S390_GMAP
6258 ----------------------
6260 :Architectures: s390
6262 This capability indicates that the user space memory used as guest mapping can
6263 be anywhere in the user memory address space, as long as the memory slots are
6264 aligned and sized to a segment (1MB) boundary.
6266 8.16 KVM_CAP_S390_COW
6267 ---------------------
6269 :Architectures: s390
6271 This capability indicates that the user space memory used as guest mapping can
6272 use copy-on-write semantics as well as dirty pages tracking via read-only page
6275 8.17 KVM_CAP_S390_BPB
6276 ---------------------
6278 :Architectures: s390
6280 This capability indicates that kvm will implement the interfaces to handle
6281 reset, migration and nested KVM for branch prediction blocking. The stfle
6282 facility 82 should not be provided to the guest without this capability.
6284 8.18 KVM_CAP_HYPERV_TLBFLUSH
6285 ----------------------------
6289 This capability indicates that KVM supports paravirtualized Hyper-V TLB Flush
6291 HvFlushVirtualAddressSpace, HvFlushVirtualAddressSpaceEx,
6292 HvFlushVirtualAddressList, HvFlushVirtualAddressListEx.
6294 8.19 KVM_CAP_ARM_INJECT_SERROR_ESR
6295 ----------------------------------
6297 :Architectures: arm, arm64
6299 This capability indicates that userspace can specify (via the
6300 KVM_SET_VCPU_EVENTS ioctl) the syndrome value reported to the guest when it
6301 takes a virtual SError interrupt exception.
6302 If KVM advertises this capability, userspace can only specify the ISS field for
6303 the ESR syndrome. Other parts of the ESR, such as the EC are generated by the
6304 CPU when the exception is taken. If this virtual SError is taken to EL1 using
6305 AArch64, this value will be reported in the ISS field of ESR_ELx.
6307 See KVM_CAP_VCPU_EVENTS for more details.
6309 8.20 KVM_CAP_HYPERV_SEND_IPI
6310 ----------------------------
6314 This capability indicates that KVM supports paravirtualized Hyper-V IPI send
6316 HvCallSendSyntheticClusterIpi, HvCallSendSyntheticClusterIpiEx.
6318 8.21 KVM_CAP_HYPERV_DIRECT_TLBFLUSH
6319 -----------------------------------
6323 This capability indicates that KVM running on top of Hyper-V hypervisor
6324 enables Direct TLB flush for its guests meaning that TLB flush
6325 hypercalls are handled by Level 0 hypervisor (Hyper-V) bypassing KVM.
6326 Due to the different ABI for hypercall parameters between Hyper-V and
6327 KVM, enabling this capability effectively disables all hypercall
6328 handling by KVM (as some KVM hypercall may be mistakenly treated as TLB
6329 flush hypercalls by Hyper-V) so userspace should disable KVM identification
6330 in CPUID and only exposes Hyper-V identification. In this case, guest
6331 thinks it's running on Hyper-V and only use Hyper-V hypercalls.
6333 8.22 KVM_CAP_S390_VCPU_RESETS
6334 -----------------------------
6336 :Architectures: s390
6338 This capability indicates that the KVM_S390_NORMAL_RESET and
6339 KVM_S390_CLEAR_RESET ioctls are available.
6341 8.23 KVM_CAP_S390_PROTECTED
6342 ---------------------------
6344 :Architectures: s390
6346 This capability indicates that the Ultravisor has been initialized and
6347 KVM can therefore start protected VMs.
6348 This capability governs the KVM_S390_PV_COMMAND ioctl and the
6349 KVM_MP_STATE_LOAD MP_STATE. KVM_SET_MP_STATE can fail for protected
6350 guests when the state change is invalid.
6352 8.24 KVM_CAP_STEAL_TIME
6353 -----------------------
6355 :Architectures: arm64, x86
6357 This capability indicates that KVM supports steal time accounting.
6358 When steal time accounting is supported it may be enabled with
6359 architecture-specific interfaces. This capability and the architecture-
6360 specific interfaces must be consistent, i.e. if one says the feature
6361 is supported, than the other should as well and vice versa. For arm64
6362 see Documentation/virt/kvm/devices/vcpu.rst "KVM_ARM_VCPU_PVTIME_CTRL".
6363 For x86 see Documentation/virt/kvm/msr.rst "MSR_KVM_STEAL_TIME".
6365 8.25 KVM_CAP_S390_DIAG318
6366 -------------------------
6368 :Architectures: s390
6370 This capability enables a guest to set information about its control program
6371 (i.e. guest kernel type and version). The information is helpful during
6372 system/firmware service events, providing additional data about the guest
6373 environments running on the machine.
6375 The information is associated with the DIAGNOSE 0x318 instruction, which sets
6376 an 8-byte value consisting of a one-byte Control Program Name Code (CPNC) and
6377 a 7-byte Control Program Version Code (CPVC). The CPNC determines what
6378 environment the control program is running in (e.g. Linux, z/VM...), and the
6379 CPVC is used for information specific to OS (e.g. Linux version, Linux
6382 If this capability is available, then the CPNC and CPVC can be synchronized
6383 between KVM and userspace via the sync regs mechanism (KVM_SYNC_DIAG318).
6385 8.26 KVM_CAP_X86_USER_SPACE_MSR
6386 -------------------------------
6390 This capability indicates that KVM supports deflection of MSR reads and
6391 writes to user space. It can be enabled on a VM level. If enabled, MSR
6392 accesses that would usually trigger a #GP by KVM into the guest will
6393 instead get bounced to user space through the KVM_EXIT_X86_RDMSR and
6394 KVM_EXIT_X86_WRMSR exit notifications.
6396 8.27 KVM_X86_SET_MSR_FILTER
6397 ---------------------------
6401 This capability indicates that KVM supports that accesses to user defined MSRs
6402 may be rejected. With this capability exposed, KVM exports new VM ioctl
6403 KVM_X86_SET_MSR_FILTER which user space can call to specify bitmaps of MSR
6404 ranges that KVM should reject access to.
6406 In combination with KVM_CAP_X86_USER_SPACE_MSR, this allows user space to
6407 trap and emulate MSRs that are outside of the scope of KVM as well as
6408 limit the attack surface on KVM's MSR emulation code.
6410 8.28 KVM_CAP_ENFORCE_PV_CPUID
6411 -----------------------------
6415 When enabled, KVM will disable paravirtual features provided to the
6416 guest according to the bits in the KVM_CPUID_FEATURES CPUID leaf
6417 (0x40000001). Otherwise, a guest may use the paravirtual features
6418 regardless of what has actually been exposed through the CPUID leaf.
6421 8.29 KVM_CAP_DIRTY_LOG_RING
6422 ---------------------------
6425 :Parameters: args[0] - size of the dirty log ring
6427 KVM is capable of tracking dirty memory using ring buffers that are
6428 mmaped into userspace; there is one dirty ring per vcpu.
6430 The dirty ring is available to userspace as an array of
6431 ``struct kvm_dirty_gfn``. Each dirty entry it's defined as::
6433 struct kvm_dirty_gfn {
6435 __u32 slot; /* as_id | slot_id */
6439 The following values are defined for the flags field to define the
6440 current state of the entry::
6442 #define KVM_DIRTY_GFN_F_DIRTY BIT(0)
6443 #define KVM_DIRTY_GFN_F_RESET BIT(1)
6444 #define KVM_DIRTY_GFN_F_MASK 0x3
6446 Userspace should call KVM_ENABLE_CAP ioctl right after KVM_CREATE_VM
6447 ioctl to enable this capability for the new guest and set the size of
6448 the rings. Enabling the capability is only allowed before creating any
6449 vCPU, and the size of the ring must be a power of two. The larger the
6450 ring buffer, the less likely the ring is full and the VM is forced to
6451 exit to userspace. The optimal size depends on the workload, but it is
6452 recommended that it be at least 64 KiB (4096 entries).
6454 Just like for dirty page bitmaps, the buffer tracks writes to
6455 all user memory regions for which the KVM_MEM_LOG_DIRTY_PAGES flag was
6456 set in KVM_SET_USER_MEMORY_REGION. Once a memory region is registered
6457 with the flag set, userspace can start harvesting dirty pages from the
6460 An entry in the ring buffer can be unused (flag bits ``00``),
6461 dirty (flag bits ``01``) or harvested (flag bits ``1X``). The
6462 state machine for the entry is as follows::
6464 dirtied harvested reset
6465 00 -----------> 01 -------------> 1X -------+
6468 +------------------------------------------+
6470 To harvest the dirty pages, userspace accesses the mmaped ring buffer
6471 to read the dirty GFNs. If the flags has the DIRTY bit set (at this stage
6472 the RESET bit must be cleared), then it means this GFN is a dirty GFN.
6473 The userspace should harvest this GFN and mark the flags from state
6474 ``01b`` to ``1Xb`` (bit 0 will be ignored by KVM, but bit 1 must be set
6475 to show that this GFN is harvested and waiting for a reset), and move
6476 on to the next GFN. The userspace should continue to do this until the
6477 flags of a GFN have the DIRTY bit cleared, meaning that it has harvested
6478 all the dirty GFNs that were available.
6480 It's not necessary for userspace to harvest the all dirty GFNs at once.
6481 However it must collect the dirty GFNs in sequence, i.e., the userspace
6482 program cannot skip one dirty GFN to collect the one next to it.
6484 After processing one or more entries in the ring buffer, userspace
6485 calls the VM ioctl KVM_RESET_DIRTY_RINGS to notify the kernel about
6486 it, so that the kernel will reprotect those collected GFNs.
6487 Therefore, the ioctl must be called *before* reading the content of
6490 The dirty ring can get full. When it happens, the KVM_RUN of the
6491 vcpu will return with exit reason KVM_EXIT_DIRTY_LOG_FULL.
6493 The dirty ring interface has a major difference comparing to the
6494 KVM_GET_DIRTY_LOG interface in that, when reading the dirty ring from
6495 userspace, it's still possible that the kernel has not yet flushed the
6496 processor's dirty page buffers into the kernel buffer (with dirty bitmaps, the
6497 flushing is done by the KVM_GET_DIRTY_LOG ioctl). To achieve that, one
6498 needs to kick the vcpu out of KVM_RUN using a signal. The resulting
6499 vmexit ensures that all dirty GFNs are flushed to the dirty rings.
6501 NOTE: the capability KVM_CAP_DIRTY_LOG_RING and the corresponding
6502 ioctl KVM_RESET_DIRTY_RINGS are mutual exclusive to the existing ioctls
6503 KVM_GET_DIRTY_LOG and KVM_CLEAR_DIRTY_LOG. After enabling
6504 KVM_CAP_DIRTY_LOG_RING with an acceptable dirty ring size, the virtual
6505 machine will switch to ring-buffer dirty page tracking and further
6506 KVM_GET_DIRTY_LOG or KVM_CLEAR_DIRTY_LOG ioctls will fail.