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 Creation of the VM will fail if the requested IPA size (whether it is
186 implicit or explicit) is unsupported on the host.
188 Please note that configuring the IPA size does not affect the capability
189 exposed by the guest CPUs in ID_AA64MMFR0_EL1[PARange]. It only affects
190 size of the address translated by the stage2 level (guest physical to
191 host physical address translations).
194 4.3 KVM_GET_MSR_INDEX_LIST, KVM_GET_MSR_FEATURE_INDEX_LIST
195 ----------------------------------------------------------
197 :Capability: basic, KVM_CAP_GET_MSR_FEATURES for KVM_GET_MSR_FEATURE_INDEX_LIST
200 :Parameters: struct kvm_msr_list (in/out)
201 :Returns: 0 on success; -1 on error
205 ====== ============================================================
206 EFAULT the msr index list cannot be read from or written to
207 E2BIG the msr index list is to be to fit in the array specified by
209 ====== ============================================================
213 struct kvm_msr_list {
214 __u32 nmsrs; /* number of msrs in entries */
218 The user fills in the size of the indices array in nmsrs, and in return
219 kvm adjusts nmsrs to reflect the actual number of msrs and fills in the
220 indices array with their numbers.
222 KVM_GET_MSR_INDEX_LIST returns the guest msrs that are supported. The list
223 varies by kvm version and host processor, but does not change otherwise.
225 Note: if kvm indicates supports MCE (KVM_CAP_MCE), then the MCE bank MSRs are
226 not returned in the MSR list, as different vcpus can have a different number
227 of banks, as set via the KVM_X86_SETUP_MCE ioctl.
229 KVM_GET_MSR_FEATURE_INDEX_LIST returns the list of MSRs that can be passed
230 to the KVM_GET_MSRS system ioctl. This lets userspace probe host capabilities
231 and processor features that are exposed via MSRs (e.g., VMX capabilities).
232 This list also varies by kvm version and host processor, but does not change
236 4.4 KVM_CHECK_EXTENSION
237 -----------------------
239 :Capability: basic, KVM_CAP_CHECK_EXTENSION_VM for vm ioctl
241 :Type: system ioctl, vm ioctl
242 :Parameters: extension identifier (KVM_CAP_*)
243 :Returns: 0 if unsupported; 1 (or some other positive integer) if supported
245 The API allows the application to query about extensions to the core
246 kvm API. Userspace passes an extension identifier (an integer) and
247 receives an integer that describes the extension availability.
248 Generally 0 means no and 1 means yes, but some extensions may report
249 additional information in the integer return value.
251 Based on their initialization different VMs may have different capabilities.
252 It is thus encouraged to use the vm ioctl to query for capabilities (available
253 with KVM_CAP_CHECK_EXTENSION_VM on the vm fd)
255 4.5 KVM_GET_VCPU_MMAP_SIZE
256 --------------------------
262 :Returns: size of vcpu mmap area, in bytes
264 The KVM_RUN ioctl (cf.) communicates with userspace via a shared
265 memory region. This ioctl returns the size of that region. See the
266 KVM_RUN documentation for details.
268 Besides the size of the KVM_RUN communication region, other areas of
269 the VCPU file descriptor can be mmap-ed, including:
271 - if KVM_CAP_COALESCED_MMIO is available, a page at
272 KVM_COALESCED_MMIO_PAGE_OFFSET * PAGE_SIZE; for historical reasons,
273 this page is included in the result of KVM_GET_VCPU_MMAP_SIZE.
274 KVM_CAP_COALESCED_MMIO is not documented yet.
276 - if KVM_CAP_DIRTY_LOG_RING is available, a number of pages at
277 KVM_DIRTY_LOG_PAGE_OFFSET * PAGE_SIZE. For more information on
278 KVM_CAP_DIRTY_LOG_RING, see section 8.3.
281 4.6 KVM_SET_MEMORY_REGION
282 -------------------------
287 :Parameters: struct kvm_memory_region (in)
288 :Returns: 0 on success, -1 on error
290 This ioctl is obsolete and has been removed.
299 :Parameters: vcpu id (apic id on x86)
300 :Returns: vcpu fd on success, -1 on error
302 This API adds a vcpu to a virtual machine. No more than max_vcpus may be added.
303 The vcpu id is an integer in the range [0, max_vcpu_id).
305 The recommended max_vcpus value can be retrieved using the KVM_CAP_NR_VCPUS of
306 the KVM_CHECK_EXTENSION ioctl() at run-time.
307 The maximum possible value for max_vcpus can be retrieved using the
308 KVM_CAP_MAX_VCPUS of the KVM_CHECK_EXTENSION ioctl() at run-time.
310 If the KVM_CAP_NR_VCPUS does not exist, you should assume that max_vcpus is 4
312 If the KVM_CAP_MAX_VCPUS does not exist, you should assume that max_vcpus is
313 same as the value returned from KVM_CAP_NR_VCPUS.
315 The maximum possible value for max_vcpu_id can be retrieved using the
316 KVM_CAP_MAX_VCPU_ID of the KVM_CHECK_EXTENSION ioctl() at run-time.
318 If the KVM_CAP_MAX_VCPU_ID does not exist, you should assume that max_vcpu_id
319 is the same as the value returned from KVM_CAP_MAX_VCPUS.
321 On powerpc using book3s_hv mode, the vcpus are mapped onto virtual
322 threads in one or more virtual CPU cores. (This is because the
323 hardware requires all the hardware threads in a CPU core to be in the
324 same partition.) The KVM_CAP_PPC_SMT capability indicates the number
325 of vcpus per virtual core (vcore). The vcore id is obtained by
326 dividing the vcpu id by the number of vcpus per vcore. The vcpus in a
327 given vcore will always be in the same physical core as each other
328 (though that might be a different physical core from time to time).
329 Userspace can control the threading (SMT) mode of the guest by its
330 allocation of vcpu ids. For example, if userspace wants
331 single-threaded guest vcpus, it should make all vcpu ids be a multiple
332 of the number of vcpus per vcore.
334 For virtual cpus that have been created with S390 user controlled virtual
335 machines, the resulting vcpu fd can be memory mapped at page offset
336 KVM_S390_SIE_PAGE_OFFSET in order to obtain a memory map of the virtual
337 cpu's hardware control block.
340 4.8 KVM_GET_DIRTY_LOG (vm ioctl)
341 --------------------------------
346 :Parameters: struct kvm_dirty_log (in/out)
347 :Returns: 0 on success, -1 on error
351 /* for KVM_GET_DIRTY_LOG */
352 struct kvm_dirty_log {
356 void __user *dirty_bitmap; /* one bit per page */
361 Given a memory slot, return a bitmap containing any pages dirtied
362 since the last call to this ioctl. Bit 0 is the first page in the
363 memory slot. Ensure the entire structure is cleared to avoid padding
366 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies
367 the address space for which you want to return the dirty bitmap. See
368 KVM_SET_USER_MEMORY_REGION for details on the usage of slot field.
370 The bits in the dirty bitmap are cleared before the ioctl returns, unless
371 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is enabled. For more information,
372 see the description of the capability.
374 4.9 KVM_SET_MEMORY_ALIAS
375 ------------------------
380 :Parameters: struct kvm_memory_alias (in)
381 :Returns: 0 (success), -1 (error)
383 This ioctl is obsolete and has been removed.
393 :Returns: 0 on success, -1 on error
397 ======= ==============================================================
398 EINTR an unmasked signal is pending
399 ENOEXEC the vcpu hasn't been initialized or the guest tried to execute
400 instructions from device memory (arm64)
401 ENOSYS data abort outside memslots with no syndrome info and
402 KVM_CAP_ARM_NISV_TO_USER not enabled (arm64)
403 EPERM SVE feature set but not finalized (arm64)
404 ======= ==============================================================
406 This ioctl is used to run a guest virtual cpu. While there are no
407 explicit parameters, there is an implicit parameter block that can be
408 obtained by mmap()ing the vcpu fd at offset 0, with the size given by
409 KVM_GET_VCPU_MMAP_SIZE. The parameter block is formatted as a 'struct
410 kvm_run' (see below).
417 :Architectures: all except ARM, arm64
419 :Parameters: struct kvm_regs (out)
420 :Returns: 0 on success, -1 on error
422 Reads the general purpose registers from the vcpu.
428 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
429 __u64 rax, rbx, rcx, rdx;
430 __u64 rsi, rdi, rsp, rbp;
431 __u64 r8, r9, r10, r11;
432 __u64 r12, r13, r14, r15;
438 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
450 :Architectures: all except ARM, arm64
452 :Parameters: struct kvm_regs (in)
453 :Returns: 0 on success, -1 on error
455 Writes the general purpose registers into the vcpu.
457 See KVM_GET_REGS for the data structure.
464 :Architectures: x86, ppc
466 :Parameters: struct kvm_sregs (out)
467 :Returns: 0 on success, -1 on error
469 Reads special registers from the vcpu.
475 struct kvm_segment cs, ds, es, fs, gs, ss;
476 struct kvm_segment tr, ldt;
477 struct kvm_dtable gdt, idt;
478 __u64 cr0, cr2, cr3, cr4, cr8;
481 __u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64];
484 /* ppc -- see arch/powerpc/include/uapi/asm/kvm.h */
486 interrupt_bitmap is a bitmap of pending external interrupts. At most
487 one bit may be set. This interrupt has been acknowledged by the APIC
488 but not yet injected into the cpu core.
495 :Architectures: x86, ppc
497 :Parameters: struct kvm_sregs (in)
498 :Returns: 0 on success, -1 on error
500 Writes special registers into the vcpu. See KVM_GET_SREGS for the
510 :Parameters: struct kvm_translation (in/out)
511 :Returns: 0 on success, -1 on error
513 Translates a virtual address according to the vcpu's current address
518 struct kvm_translation {
520 __u64 linear_address;
523 __u64 physical_address;
535 :Architectures: x86, ppc, mips
537 :Parameters: struct kvm_interrupt (in)
538 :Returns: 0 on success, negative on failure.
540 Queues a hardware interrupt vector to be injected.
544 /* for KVM_INTERRUPT */
545 struct kvm_interrupt {
555 ========= ===================================
557 -EEXIST if an interrupt is already enqueued
558 -EINVAL the irq number is invalid
559 -ENXIO if the PIC is in the kernel
560 -EFAULT if the pointer is invalid
561 ========= ===================================
563 Note 'irq' is an interrupt vector, not an interrupt pin or line. This
564 ioctl is useful if the in-kernel PIC is not used.
569 Queues an external interrupt to be injected. This ioctl is overleaded
570 with 3 different irq values:
574 This injects an edge type external interrupt into the guest once it's ready
575 to receive interrupts. When injected, the interrupt is done.
577 b) KVM_INTERRUPT_UNSET
579 This unsets any pending interrupt.
581 Only available with KVM_CAP_PPC_UNSET_IRQ.
583 c) KVM_INTERRUPT_SET_LEVEL
585 This injects a level type external interrupt into the guest context. The
586 interrupt stays pending until a specific ioctl with KVM_INTERRUPT_UNSET
589 Only available with KVM_CAP_PPC_IRQ_LEVEL.
591 Note that any value for 'irq' other than the ones stated above is invalid
592 and incurs unexpected behavior.
594 This is an asynchronous vcpu ioctl and can be invoked from any thread.
599 Queues an external interrupt to be injected into the virtual CPU. A negative
600 interrupt number dequeues the interrupt.
602 This is an asynchronous vcpu ioctl and can be invoked from any thread.
612 :Returns: -1 on error
614 Support for this has been removed. Use KVM_SET_GUEST_DEBUG instead.
620 :Capability: basic (vcpu), KVM_CAP_GET_MSR_FEATURES (system)
622 :Type: system ioctl, vcpu ioctl
623 :Parameters: struct kvm_msrs (in/out)
624 :Returns: number of msrs successfully returned;
627 When used as a system ioctl:
628 Reads the values of MSR-based features that are available for the VM. This
629 is similar to KVM_GET_SUPPORTED_CPUID, but it returns MSR indices and values.
630 The list of msr-based features can be obtained using KVM_GET_MSR_FEATURE_INDEX_LIST
633 When used as a vcpu ioctl:
634 Reads model-specific registers from the vcpu. Supported msr indices can
635 be obtained using KVM_GET_MSR_INDEX_LIST in a system ioctl.
640 __u32 nmsrs; /* number of msrs in entries */
643 struct kvm_msr_entry entries[0];
646 struct kvm_msr_entry {
652 Application code should set the 'nmsrs' member (which indicates the
653 size of the entries array) and the 'index' member of each array entry.
654 kvm will fill in the 'data' member.
663 :Parameters: struct kvm_msrs (in)
664 :Returns: number of msrs successfully set (see below), -1 on error
666 Writes model-specific registers to the vcpu. See KVM_GET_MSRS for the
669 Application code should set the 'nmsrs' member (which indicates the
670 size of the entries array), and the 'index' and 'data' members of each
673 It tries to set the MSRs in array entries[] one by one. If setting an MSR
674 fails, e.g., due to setting reserved bits, the MSR isn't supported/emulated
675 by KVM, etc..., it stops processing the MSR list and returns the number of
676 MSRs that have been set successfully.
685 :Parameters: struct kvm_cpuid (in)
686 :Returns: 0 on success, -1 on error
688 Defines the vcpu responses to the cpuid instruction. Applications
689 should use the KVM_SET_CPUID2 ioctl if available.
691 Note, when this IOCTL fails, KVM gives no guarantees that previous valid CPUID
692 configuration (if there is) is not corrupted. Userspace can get a copy of the
693 resulting CPUID configuration through KVM_GET_CPUID2 in case.
697 struct kvm_cpuid_entry {
706 /* for KVM_SET_CPUID */
710 struct kvm_cpuid_entry entries[0];
714 4.21 KVM_SET_SIGNAL_MASK
715 ------------------------
720 :Parameters: struct kvm_signal_mask (in)
721 :Returns: 0 on success, -1 on error
723 Defines which signals are blocked during execution of KVM_RUN. This
724 signal mask temporarily overrides the threads signal mask. Any
725 unblocked signal received (except SIGKILL and SIGSTOP, which retain
726 their traditional behaviour) will cause KVM_RUN to return with -EINTR.
728 Note the signal will only be delivered if not blocked by the original
733 /* for KVM_SET_SIGNAL_MASK */
734 struct kvm_signal_mask {
746 :Parameters: struct kvm_fpu (out)
747 :Returns: 0 on success, -1 on error
749 Reads the floating point state from the vcpu.
753 /* for KVM_GET_FPU and KVM_SET_FPU */
758 __u8 ftwx; /* in fxsave format */
775 :Parameters: struct kvm_fpu (in)
776 :Returns: 0 on success, -1 on error
778 Writes the floating point state to the vcpu.
782 /* for KVM_GET_FPU and KVM_SET_FPU */
787 __u8 ftwx; /* in fxsave format */
798 4.24 KVM_CREATE_IRQCHIP
799 -----------------------
801 :Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQCHIP (s390)
802 :Architectures: x86, ARM, arm64, s390
805 :Returns: 0 on success, -1 on error
807 Creates an interrupt controller model in the kernel.
808 On x86, creates a virtual ioapic, a virtual PIC (two PICs, nested), and sets up
809 future vcpus to have a local APIC. IRQ routing for GSIs 0-15 is set to both
810 PIC and IOAPIC; GSI 16-23 only go to the IOAPIC.
811 On ARM/arm64, a GICv2 is created. Any other GIC versions require the usage of
812 KVM_CREATE_DEVICE, which also supports creating a GICv2. Using
813 KVM_CREATE_DEVICE is preferred over KVM_CREATE_IRQCHIP for GICv2.
814 On s390, a dummy irq routing table is created.
816 Note that on s390 the KVM_CAP_S390_IRQCHIP vm capability needs to be enabled
817 before KVM_CREATE_IRQCHIP can be used.
823 :Capability: KVM_CAP_IRQCHIP
824 :Architectures: x86, arm, arm64
826 :Parameters: struct kvm_irq_level
827 :Returns: 0 on success, -1 on error
829 Sets the level of a GSI input to the interrupt controller model in the kernel.
830 On some architectures it is required that an interrupt controller model has
831 been previously created with KVM_CREATE_IRQCHIP. Note that edge-triggered
832 interrupts require the level to be set to 1 and then back to 0.
834 On real hardware, interrupt pins can be active-low or active-high. This
835 does not matter for the level field of struct kvm_irq_level: 1 always
836 means active (asserted), 0 means inactive (deasserted).
838 x86 allows the operating system to program the interrupt polarity
839 (active-low/active-high) for level-triggered interrupts, and KVM used
840 to consider the polarity. However, due to bitrot in the handling of
841 active-low interrupts, the above convention is now valid on x86 too.
842 This is signaled by KVM_CAP_X86_IOAPIC_POLARITY_IGNORED. Userspace
843 should not present interrupts to the guest as active-low unless this
844 capability is present (or unless it is not using the in-kernel irqchip,
848 ARM/arm64 can signal an interrupt either at the CPU level, or at the
849 in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to
850 use PPIs designated for specific cpus. The irq field is interpreted
853 Â bits: | 31 ... 28 | 27 ... 24 | 23 ... 16 | 15 ... 0 |
854 field: | vcpu2_index | irq_type | vcpu_index | irq_id |
856 The irq_type field has the following values:
859 out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ
861 in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.)
862 (the vcpu_index field is ignored)
864 in-kernel GIC: PPI, irq_id between 16 and 31 (incl.)
866 (The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs)
868 In both cases, level is used to assert/deassert the line.
870 When KVM_CAP_ARM_IRQ_LINE_LAYOUT_2 is supported, the target vcpu is
871 identified as (256 * vcpu2_index + vcpu_index). Otherwise, vcpu2_index
874 Note that on arm/arm64, the KVM_CAP_IRQCHIP capability only conditions
875 injection of interrupts for the in-kernel irqchip. KVM_IRQ_LINE can always
876 be used for a userspace interrupt controller.
880 struct kvm_irq_level {
883 __s32 status; /* not used for KVM_IRQ_LEVEL */
885 __u32 level; /* 0 or 1 */
892 :Capability: KVM_CAP_IRQCHIP
895 :Parameters: struct kvm_irqchip (in/out)
896 :Returns: 0 on success, -1 on error
898 Reads the state of a kernel interrupt controller created with
899 KVM_CREATE_IRQCHIP into a buffer provided by the caller.
904 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
907 char dummy[512]; /* reserving space */
908 struct kvm_pic_state pic;
909 struct kvm_ioapic_state ioapic;
917 :Capability: KVM_CAP_IRQCHIP
920 :Parameters: struct kvm_irqchip (in)
921 :Returns: 0 on success, -1 on error
923 Sets the state of a kernel interrupt controller created with
924 KVM_CREATE_IRQCHIP from a buffer provided by the caller.
929 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
932 char dummy[512]; /* reserving space */
933 struct kvm_pic_state pic;
934 struct kvm_ioapic_state ioapic;
939 4.28 KVM_XEN_HVM_CONFIG
940 -----------------------
942 :Capability: KVM_CAP_XEN_HVM
945 :Parameters: struct kvm_xen_hvm_config (in)
946 :Returns: 0 on success, -1 on error
948 Sets the MSR that the Xen HVM guest uses to initialize its hypercall
949 page, and provides the starting address and size of the hypercall
950 blobs in userspace. When the guest writes the MSR, kvm copies one
951 page of a blob (32- or 64-bit, depending on the vcpu mode) to guest
956 struct kvm_xen_hvm_config {
966 If the KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL flag is returned from the
967 KVM_CAP_XEN_HVM check, it may be set in the flags field of this ioctl.
968 This requests KVM to generate the contents of the hypercall page
969 automatically; hypercalls will be intercepted and passed to userspace
970 through KVM_EXIT_XEN. In this case, all of the blob size and address
973 No other flags are currently valid in the struct kvm_xen_hvm_config.
978 :Capability: KVM_CAP_ADJUST_CLOCK
981 :Parameters: struct kvm_clock_data (out)
982 :Returns: 0 on success, -1 on error
984 Gets the current timestamp of kvmclock as seen by the current guest. In
985 conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios
988 When KVM_CAP_ADJUST_CLOCK is passed to KVM_CHECK_EXTENSION, it returns the
989 set of bits that KVM can return in struct kvm_clock_data's flag member.
991 The only flag defined now is KVM_CLOCK_TSC_STABLE. If set, the returned
992 value is the exact kvmclock value seen by all VCPUs at the instant
993 when KVM_GET_CLOCK was called. If clear, the returned value is simply
994 CLOCK_MONOTONIC plus a constant offset; the offset can be modified
995 with KVM_SET_CLOCK. KVM will try to make all VCPUs follow this clock,
996 but the exact value read by each VCPU could differ, because the host
1001 struct kvm_clock_data {
1002 __u64 clock; /* kvmclock current value */
1011 :Capability: KVM_CAP_ADJUST_CLOCK
1014 :Parameters: struct kvm_clock_data (in)
1015 :Returns: 0 on success, -1 on error
1017 Sets the current timestamp of kvmclock to the value specified in its parameter.
1018 In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios
1023 struct kvm_clock_data {
1024 __u64 clock; /* kvmclock current value */
1030 4.31 KVM_GET_VCPU_EVENTS
1031 ------------------------
1033 :Capability: KVM_CAP_VCPU_EVENTS
1034 :Extended by: KVM_CAP_INTR_SHADOW
1035 :Architectures: x86, arm, arm64
1037 :Parameters: struct kvm_vcpu_event (out)
1038 :Returns: 0 on success, -1 on error
1043 Gets currently pending exceptions, interrupts, and NMIs as well as related
1048 struct kvm_vcpu_events {
1052 __u8 has_error_code;
1073 __u8 smm_inside_nmi;
1077 __u8 exception_has_payload;
1078 __u64 exception_payload;
1081 The following bits are defined in the flags field:
1083 - KVM_VCPUEVENT_VALID_SHADOW may be set to signal that
1084 interrupt.shadow contains a valid state.
1086 - KVM_VCPUEVENT_VALID_SMM may be set to signal that smi contains a
1089 - KVM_VCPUEVENT_VALID_PAYLOAD may be set to signal that the
1090 exception_has_payload, exception_payload, and exception.pending
1091 fields contain a valid state. This bit will be set whenever
1092 KVM_CAP_EXCEPTION_PAYLOAD is enabled.
1097 If the guest accesses a device that is being emulated by the host kernel in
1098 such a way that a real device would generate a physical SError, KVM may make
1099 a virtual SError pending for that VCPU. This system error interrupt remains
1100 pending until the guest takes the exception by unmasking PSTATE.A.
1102 Running the VCPU may cause it to take a pending SError, or make an access that
1103 causes an SError to become pending. The event's description is only valid while
1104 the VPCU is not running.
1106 This API provides a way to read and write the pending 'event' state that is not
1107 visible to the guest. To save, restore or migrate a VCPU the struct representing
1108 the state can be read then written using this GET/SET API, along with the other
1109 guest-visible registers. It is not possible to 'cancel' an SError that has been
1112 A device being emulated in user-space may also wish to generate an SError. To do
1113 this the events structure can be populated by user-space. The current state
1114 should be read first, to ensure no existing SError is pending. If an existing
1115 SError is pending, the architecture's 'Multiple SError interrupts' rules should
1116 be followed. (2.5.3 of DDI0587.a "ARM Reliability, Availability, and
1117 Serviceability (RAS) Specification").
1119 SError exceptions always have an ESR value. Some CPUs have the ability to
1120 specify what the virtual SError's ESR value should be. These systems will
1121 advertise KVM_CAP_ARM_INJECT_SERROR_ESR. In this case exception.has_esr will
1122 always have a non-zero value when read, and the agent making an SError pending
1123 should specify the ISS field in the lower 24 bits of exception.serror_esr. If
1124 the system supports KVM_CAP_ARM_INJECT_SERROR_ESR, but user-space sets the events
1125 with exception.has_esr as zero, KVM will choose an ESR.
1127 Specifying exception.has_esr on a system that does not support it will return
1128 -EINVAL. Setting anything other than the lower 24bits of exception.serror_esr
1129 will return -EINVAL.
1131 It is not possible to read back a pending external abort (injected via
1132 KVM_SET_VCPU_EVENTS or otherwise) because such an exception is always delivered
1133 directly to the virtual CPU).
1137 struct kvm_vcpu_events {
1139 __u8 serror_pending;
1140 __u8 serror_has_esr;
1141 __u8 ext_dabt_pending;
1142 /* Align it to 8 bytes */
1149 4.32 KVM_SET_VCPU_EVENTS
1150 ------------------------
1152 :Capability: KVM_CAP_VCPU_EVENTS
1153 :Extended by: KVM_CAP_INTR_SHADOW
1154 :Architectures: x86, arm, arm64
1156 :Parameters: struct kvm_vcpu_event (in)
1157 :Returns: 0 on success, -1 on error
1162 Set pending exceptions, interrupts, and NMIs as well as related states of the
1165 See KVM_GET_VCPU_EVENTS for the data structure.
1167 Fields that may be modified asynchronously by running VCPUs can be excluded
1168 from the update. These fields are nmi.pending, sipi_vector, smi.smm,
1169 smi.pending. Keep the corresponding bits in the flags field cleared to
1170 suppress overwriting the current in-kernel state. The bits are:
1172 =============================== ==================================
1173 KVM_VCPUEVENT_VALID_NMI_PENDING transfer nmi.pending to the kernel
1174 KVM_VCPUEVENT_VALID_SIPI_VECTOR transfer sipi_vector
1175 KVM_VCPUEVENT_VALID_SMM transfer the smi sub-struct.
1176 =============================== ==================================
1178 If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in
1179 the flags field to signal that interrupt.shadow contains a valid state and
1180 shall be written into the VCPU.
1182 KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available.
1184 If KVM_CAP_EXCEPTION_PAYLOAD is enabled, KVM_VCPUEVENT_VALID_PAYLOAD
1185 can be set in the flags field to signal that the
1186 exception_has_payload, exception_payload, and exception.pending fields
1187 contain a valid state and shall be written into the VCPU.
1192 User space may need to inject several types of events to the guest.
1194 Set the pending SError exception state for this VCPU. It is not possible to
1195 'cancel' an Serror that has been made pending.
1197 If the guest performed an access to I/O memory which could not be handled by
1198 userspace, for example because of missing instruction syndrome decode
1199 information or because there is no device mapped at the accessed IPA, then
1200 userspace can ask the kernel to inject an external abort using the address
1201 from the exiting fault on the VCPU. It is a programming error to set
1202 ext_dabt_pending after an exit which was not either KVM_EXIT_MMIO or
1203 KVM_EXIT_ARM_NISV. This feature is only available if the system supports
1204 KVM_CAP_ARM_INJECT_EXT_DABT. This is a helper which provides commonality in
1205 how userspace reports accesses for the above cases to guests, across different
1206 userspace implementations. Nevertheless, userspace can still emulate all Arm
1207 exceptions by manipulating individual registers using the KVM_SET_ONE_REG API.
1209 See KVM_GET_VCPU_EVENTS for the data structure.
1212 4.33 KVM_GET_DEBUGREGS
1213 ----------------------
1215 :Capability: KVM_CAP_DEBUGREGS
1218 :Parameters: struct kvm_debugregs (out)
1219 :Returns: 0 on success, -1 on error
1221 Reads debug registers from the vcpu.
1225 struct kvm_debugregs {
1234 4.34 KVM_SET_DEBUGREGS
1235 ----------------------
1237 :Capability: KVM_CAP_DEBUGREGS
1240 :Parameters: struct kvm_debugregs (in)
1241 :Returns: 0 on success, -1 on error
1243 Writes debug registers into the vcpu.
1245 See KVM_GET_DEBUGREGS for the data structure. The flags field is unused
1246 yet and must be cleared on entry.
1249 4.35 KVM_SET_USER_MEMORY_REGION
1250 -------------------------------
1252 :Capability: KVM_CAP_USER_MEMORY
1255 :Parameters: struct kvm_userspace_memory_region (in)
1256 :Returns: 0 on success, -1 on error
1260 struct kvm_userspace_memory_region {
1263 __u64 guest_phys_addr;
1264 __u64 memory_size; /* bytes */
1265 __u64 userspace_addr; /* start of the userspace allocated memory */
1268 /* for kvm_memory_region::flags */
1269 #define KVM_MEM_LOG_DIRTY_PAGES (1UL << 0)
1270 #define KVM_MEM_READONLY (1UL << 1)
1272 This ioctl allows the user to create, modify or delete a guest physical
1273 memory slot. Bits 0-15 of "slot" specify the slot id and this value
1274 should be less than the maximum number of user memory slots supported per
1275 VM. The maximum allowed slots can be queried using KVM_CAP_NR_MEMSLOTS.
1276 Slots may not overlap in guest physical address space.
1278 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot"
1279 specifies the address space which is being modified. They must be
1280 less than the value that KVM_CHECK_EXTENSION returns for the
1281 KVM_CAP_MULTI_ADDRESS_SPACE capability. Slots in separate address spaces
1282 are unrelated; the restriction on overlapping slots only applies within
1285 Deleting a slot is done by passing zero for memory_size. When changing
1286 an existing slot, it may be moved in the guest physical memory space,
1287 or its flags may be modified, but it may not be resized.
1289 Memory for the region is taken starting at the address denoted by the
1290 field userspace_addr, which must point at user addressable memory for
1291 the entire memory slot size. Any object may back this memory, including
1292 anonymous memory, ordinary files, and hugetlbfs.
1294 On architectures that support a form of address tagging, userspace_addr must
1295 be an untagged address.
1297 It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr
1298 be identical. This allows large pages in the guest to be backed by large
1301 The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and
1302 KVM_MEM_READONLY. The former can be set to instruct KVM to keep track of
1303 writes to memory within the slot. See KVM_GET_DIRTY_LOG ioctl to know how to
1304 use it. The latter can be set, if KVM_CAP_READONLY_MEM capability allows it,
1305 to make a new slot read-only. In this case, writes to this memory will be
1306 posted to userspace as KVM_EXIT_MMIO exits.
1308 When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of
1309 the memory region are automatically reflected into the guest. For example, an
1310 mmap() that affects the region will be made visible immediately. Another
1311 example is madvise(MADV_DROP).
1313 It is recommended to use this API instead of the KVM_SET_MEMORY_REGION ioctl.
1314 The KVM_SET_MEMORY_REGION does not allow fine grained control over memory
1315 allocation and is deprecated.
1318 4.36 KVM_SET_TSS_ADDR
1319 ---------------------
1321 :Capability: KVM_CAP_SET_TSS_ADDR
1324 :Parameters: unsigned long tss_address (in)
1325 :Returns: 0 on success, -1 on error
1327 This ioctl defines the physical address of a three-page region in the guest
1328 physical address space. The region must be within the first 4GB of the
1329 guest physical address space and must not conflict with any memory slot
1330 or any mmio address. The guest may malfunction if it accesses this memory
1333 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1334 because of a quirk in the virtualization implementation (see the internals
1335 documentation when it pops into existence).
1341 :Capability: KVM_CAP_ENABLE_CAP
1342 :Architectures: mips, ppc, s390
1344 :Parameters: struct kvm_enable_cap (in)
1345 :Returns: 0 on success; -1 on error
1347 :Capability: KVM_CAP_ENABLE_CAP_VM
1350 :Parameters: struct kvm_enable_cap (in)
1351 :Returns: 0 on success; -1 on error
1355 Not all extensions are enabled by default. Using this ioctl the application
1356 can enable an extension, making it available to the guest.
1358 On systems that do not support this ioctl, it always fails. On systems that
1359 do support it, it only works for extensions that are supported for enablement.
1361 To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should
1366 struct kvm_enable_cap {
1370 The capability that is supposed to get enabled.
1376 A bitfield indicating future enhancements. Has to be 0 for now.
1382 Arguments for enabling a feature. If a feature needs initial values to
1383 function properly, this is the place to put them.
1390 The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl
1391 for vm-wide capabilities.
1393 4.38 KVM_GET_MP_STATE
1394 ---------------------
1396 :Capability: KVM_CAP_MP_STATE
1397 :Architectures: x86, s390, arm, arm64
1399 :Parameters: struct kvm_mp_state (out)
1400 :Returns: 0 on success; -1 on error
1404 struct kvm_mp_state {
1408 Returns the vcpu's current "multiprocessing state" (though also valid on
1409 uniprocessor guests).
1411 Possible values are:
1413 ========================== ===============================================
1414 KVM_MP_STATE_RUNNABLE the vcpu is currently running [x86,arm/arm64]
1415 KVM_MP_STATE_UNINITIALIZED the vcpu is an application processor (AP)
1416 which has not yet received an INIT signal [x86]
1417 KVM_MP_STATE_INIT_RECEIVED the vcpu has received an INIT signal, and is
1418 now ready for a SIPI [x86]
1419 KVM_MP_STATE_HALTED the vcpu has executed a HLT instruction and
1420 is waiting for an interrupt [x86]
1421 KVM_MP_STATE_SIPI_RECEIVED the vcpu has just received a SIPI (vector
1422 accessible via KVM_GET_VCPU_EVENTS) [x86]
1423 KVM_MP_STATE_STOPPED the vcpu is stopped [s390,arm/arm64]
1424 KVM_MP_STATE_CHECK_STOP the vcpu is in a special error state [s390]
1425 KVM_MP_STATE_OPERATING the vcpu is operating (running or halted)
1427 KVM_MP_STATE_LOAD the vcpu is in a special load/startup state
1429 ========================== ===============================================
1431 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1432 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1433 these architectures.
1438 The only states that are valid are KVM_MP_STATE_STOPPED and
1439 KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not.
1441 4.39 KVM_SET_MP_STATE
1442 ---------------------
1444 :Capability: KVM_CAP_MP_STATE
1445 :Architectures: x86, s390, arm, arm64
1447 :Parameters: struct kvm_mp_state (in)
1448 :Returns: 0 on success; -1 on error
1450 Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for
1453 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1454 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1455 these architectures.
1460 The only states that are valid are KVM_MP_STATE_STOPPED and
1461 KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not.
1463 4.40 KVM_SET_IDENTITY_MAP_ADDR
1464 ------------------------------
1466 :Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR
1469 :Parameters: unsigned long identity (in)
1470 :Returns: 0 on success, -1 on error
1472 This ioctl defines the physical address of a one-page region in the guest
1473 physical address space. The region must be within the first 4GB of the
1474 guest physical address space and must not conflict with any memory slot
1475 or any mmio address. The guest may malfunction if it accesses this memory
1478 Setting the address to 0 will result in resetting the address to its default
1481 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1482 because of a quirk in the virtualization implementation (see the internals
1483 documentation when it pops into existence).
1485 Fails if any VCPU has already been created.
1487 4.41 KVM_SET_BOOT_CPU_ID
1488 ------------------------
1490 :Capability: KVM_CAP_SET_BOOT_CPU_ID
1493 :Parameters: unsigned long vcpu_id
1494 :Returns: 0 on success, -1 on error
1496 Define which vcpu is the Bootstrap Processor (BSP). Values are the same
1497 as the vcpu id in KVM_CREATE_VCPU. If this ioctl is not called, the default
1504 :Capability: KVM_CAP_XSAVE
1507 :Parameters: struct kvm_xsave (out)
1508 :Returns: 0 on success, -1 on error
1517 This ioctl would copy current vcpu's xsave struct to the userspace.
1523 :Capability: KVM_CAP_XSAVE
1526 :Parameters: struct kvm_xsave (in)
1527 :Returns: 0 on success, -1 on error
1536 This ioctl would copy userspace's xsave struct to the kernel.
1542 :Capability: KVM_CAP_XCRS
1545 :Parameters: struct kvm_xcrs (out)
1546 :Returns: 0 on success, -1 on error
1559 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1563 This ioctl would copy current vcpu's xcrs to the userspace.
1569 :Capability: KVM_CAP_XCRS
1572 :Parameters: struct kvm_xcrs (in)
1573 :Returns: 0 on success, -1 on error
1586 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1590 This ioctl would set vcpu's xcr to the value userspace specified.
1593 4.46 KVM_GET_SUPPORTED_CPUID
1594 ----------------------------
1596 :Capability: KVM_CAP_EXT_CPUID
1599 :Parameters: struct kvm_cpuid2 (in/out)
1600 :Returns: 0 on success, -1 on error
1607 struct kvm_cpuid_entry2 entries[0];
1610 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
1611 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */
1612 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */
1614 struct kvm_cpuid_entry2 {
1625 This ioctl returns x86 cpuid features which are supported by both the
1626 hardware and kvm in its default configuration. Userspace can use the
1627 information returned by this ioctl to construct cpuid information (for
1628 KVM_SET_CPUID2) that is consistent with hardware, kernel, and
1629 userspace capabilities, and with user requirements (for example, the
1630 user may wish to constrain cpuid to emulate older hardware, or for
1631 feature consistency across a cluster).
1633 Note that certain capabilities, such as KVM_CAP_X86_DISABLE_EXITS, may
1634 expose cpuid features (e.g. MONITOR) which are not supported by kvm in
1635 its default configuration. If userspace enables such capabilities, it
1636 is responsible for modifying the results of this ioctl appropriately.
1638 Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure
1639 with the 'nent' field indicating the number of entries in the variable-size
1640 array 'entries'. If the number of entries is too low to describe the cpu
1641 capabilities, an error (E2BIG) is returned. If the number is too high,
1642 the 'nent' field is adjusted and an error (ENOMEM) is returned. If the
1643 number is just right, the 'nent' field is adjusted to the number of valid
1644 entries in the 'entries' array, which is then filled.
1646 The entries returned are the host cpuid as returned by the cpuid instruction,
1647 with unknown or unsupported features masked out. Some features (for example,
1648 x2apic), may not be present in the host cpu, but are exposed by kvm if it can
1649 emulate them efficiently. The fields in each entry are defined as follows:
1652 the eax value used to obtain the entry
1655 the ecx value used to obtain the entry (for entries that are
1659 an OR of zero or more of the following:
1661 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
1662 if the index field is valid
1665 the values returned by the cpuid instruction for
1666 this function/index combination
1668 The TSC deadline timer feature (CPUID leaf 1, ecx[24]) is always returned
1669 as false, since the feature depends on KVM_CREATE_IRQCHIP for local APIC
1670 support. Instead it is reported via::
1672 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER)
1674 if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the
1675 feature in userspace, then you can enable the feature for KVM_SET_CPUID2.
1678 4.47 KVM_PPC_GET_PVINFO
1679 -----------------------
1681 :Capability: KVM_CAP_PPC_GET_PVINFO
1684 :Parameters: struct kvm_ppc_pvinfo (out)
1685 :Returns: 0 on success, !0 on error
1689 struct kvm_ppc_pvinfo {
1695 This ioctl fetches PV specific information that need to be passed to the guest
1696 using the device tree or other means from vm context.
1698 The hcall array defines 4 instructions that make up a hypercall.
1700 If any additional field gets added to this structure later on, a bit for that
1701 additional piece of information will be set in the flags bitmap.
1703 The flags bitmap is defined as::
1705 /* the host supports the ePAPR idle hcall
1706 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1<<0)
1708 4.52 KVM_SET_GSI_ROUTING
1709 ------------------------
1711 :Capability: KVM_CAP_IRQ_ROUTING
1712 :Architectures: x86 s390 arm arm64
1714 :Parameters: struct kvm_irq_routing (in)
1715 :Returns: 0 on success, -1 on error
1717 Sets the GSI routing table entries, overwriting any previously set entries.
1719 On arm/arm64, GSI routing has the following limitation:
1721 - GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD.
1725 struct kvm_irq_routing {
1728 struct kvm_irq_routing_entry entries[0];
1731 No flags are specified so far, the corresponding field must be set to zero.
1735 struct kvm_irq_routing_entry {
1741 struct kvm_irq_routing_irqchip irqchip;
1742 struct kvm_irq_routing_msi msi;
1743 struct kvm_irq_routing_s390_adapter adapter;
1744 struct kvm_irq_routing_hv_sint hv_sint;
1749 /* gsi routing entry types */
1750 #define KVM_IRQ_ROUTING_IRQCHIP 1
1751 #define KVM_IRQ_ROUTING_MSI 2
1752 #define KVM_IRQ_ROUTING_S390_ADAPTER 3
1753 #define KVM_IRQ_ROUTING_HV_SINT 4
1757 - KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry
1758 type, specifies that the devid field contains a valid value. The per-VM
1759 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
1760 the device ID. If this capability is not available, userspace should
1761 never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
1766 struct kvm_irq_routing_irqchip {
1771 struct kvm_irq_routing_msi {
1781 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
1782 for the device that wrote the MSI message. For PCI, this is usually a
1783 BFD identifier in the lower 16 bits.
1785 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
1786 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
1787 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
1788 address_hi must be zero.
1792 struct kvm_irq_routing_s390_adapter {
1796 __u32 summary_offset;
1800 struct kvm_irq_routing_hv_sint {
1806 4.55 KVM_SET_TSC_KHZ
1807 --------------------
1809 :Capability: KVM_CAP_TSC_CONTROL
1812 :Parameters: virtual tsc_khz
1813 :Returns: 0 on success, -1 on error
1815 Specifies the tsc frequency for the virtual machine. The unit of the
1819 4.56 KVM_GET_TSC_KHZ
1820 --------------------
1822 :Capability: KVM_CAP_GET_TSC_KHZ
1826 :Returns: virtual tsc-khz on success, negative value on error
1828 Returns the tsc frequency of the guest. The unit of the return value is
1829 KHz. If the host has unstable tsc this ioctl returns -EIO instead as an
1836 :Capability: KVM_CAP_IRQCHIP
1839 :Parameters: struct kvm_lapic_state (out)
1840 :Returns: 0 on success, -1 on error
1844 #define KVM_APIC_REG_SIZE 0x400
1845 struct kvm_lapic_state {
1846 char regs[KVM_APIC_REG_SIZE];
1849 Reads the Local APIC registers and copies them into the input argument. The
1850 data format and layout are the same as documented in the architecture manual.
1852 If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is
1853 enabled, then the format of APIC_ID register depends on the APIC mode
1854 (reported by MSR_IA32_APICBASE) of its VCPU. x2APIC stores APIC ID in
1855 the APIC_ID register (bytes 32-35). xAPIC only allows an 8-bit APIC ID
1856 which is stored in bits 31-24 of the APIC register, or equivalently in
1857 byte 35 of struct kvm_lapic_state's regs field. KVM_GET_LAPIC must then
1858 be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR.
1860 If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state
1861 always uses xAPIC format.
1867 :Capability: KVM_CAP_IRQCHIP
1870 :Parameters: struct kvm_lapic_state (in)
1871 :Returns: 0 on success, -1 on error
1875 #define KVM_APIC_REG_SIZE 0x400
1876 struct kvm_lapic_state {
1877 char regs[KVM_APIC_REG_SIZE];
1880 Copies the input argument into the Local APIC registers. The data format
1881 and layout are the same as documented in the architecture manual.
1883 The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's
1884 regs field) depends on the state of the KVM_CAP_X2APIC_API capability.
1885 See the note in KVM_GET_LAPIC.
1891 :Capability: KVM_CAP_IOEVENTFD
1894 :Parameters: struct kvm_ioeventfd (in)
1895 :Returns: 0 on success, !0 on error
1897 This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address
1898 within the guest. A guest write in the registered address will signal the
1899 provided event instead of triggering an exit.
1903 struct kvm_ioeventfd {
1905 __u64 addr; /* legal pio/mmio address */
1906 __u32 len; /* 0, 1, 2, 4, or 8 bytes */
1912 For the special case of virtio-ccw devices on s390, the ioevent is matched
1913 to a subchannel/virtqueue tuple instead.
1915 The following flags are defined::
1917 #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch)
1918 #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio)
1919 #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign)
1920 #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \
1921 (1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify)
1923 If datamatch flag is set, the event will be signaled only if the written value
1924 to the registered address is equal to datamatch in struct kvm_ioeventfd.
1926 For virtio-ccw devices, addr contains the subchannel id and datamatch the
1929 With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and
1930 the kernel will ignore the length of guest write and may get a faster vmexit.
1931 The speedup may only apply to specific architectures, but the ioeventfd will
1937 :Capability: KVM_CAP_SW_TLB
1940 :Parameters: struct kvm_dirty_tlb (in)
1941 :Returns: 0 on success, -1 on error
1945 struct kvm_dirty_tlb {
1950 This must be called whenever userspace has changed an entry in the shared
1951 TLB, prior to calling KVM_RUN on the associated vcpu.
1953 The "bitmap" field is the userspace address of an array. This array
1954 consists of a number of bits, equal to the total number of TLB entries as
1955 determined by the last successful call to KVM_CONFIG_TLB, rounded up to the
1956 nearest multiple of 64.
1958 Each bit corresponds to one TLB entry, ordered the same as in the shared TLB
1961 The array is little-endian: the bit 0 is the least significant bit of the
1962 first byte, bit 8 is the least significant bit of the second byte, etc.
1963 This avoids any complications with differing word sizes.
1965 The "num_dirty" field is a performance hint for KVM to determine whether it
1966 should skip processing the bitmap and just invalidate everything. It must
1967 be set to the number of set bits in the bitmap.
1970 4.62 KVM_CREATE_SPAPR_TCE
1971 -------------------------
1973 :Capability: KVM_CAP_SPAPR_TCE
1974 :Architectures: powerpc
1976 :Parameters: struct kvm_create_spapr_tce (in)
1977 :Returns: file descriptor for manipulating the created TCE table
1979 This creates a virtual TCE (translation control entry) table, which
1980 is an IOMMU for PAPR-style virtual I/O. It is used to translate
1981 logical addresses used in virtual I/O into guest physical addresses,
1982 and provides a scatter/gather capability for PAPR virtual I/O.
1986 /* for KVM_CAP_SPAPR_TCE */
1987 struct kvm_create_spapr_tce {
1992 The liobn field gives the logical IO bus number for which to create a
1993 TCE table. The window_size field specifies the size of the DMA window
1994 which this TCE table will translate - the table will contain one 64
1995 bit TCE entry for every 4kiB of the DMA window.
1997 When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE
1998 table has been created using this ioctl(), the kernel will handle it
1999 in real mode, updating the TCE table. H_PUT_TCE calls for other
2000 liobns will cause a vm exit and must be handled by userspace.
2002 The return value is a file descriptor which can be passed to mmap(2)
2003 to map the created TCE table into userspace. This lets userspace read
2004 the entries written by kernel-handled H_PUT_TCE calls, and also lets
2005 userspace update the TCE table directly which is useful in some
2009 4.63 KVM_ALLOCATE_RMA
2010 ---------------------
2012 :Capability: KVM_CAP_PPC_RMA
2013 :Architectures: powerpc
2015 :Parameters: struct kvm_allocate_rma (out)
2016 :Returns: file descriptor for mapping the allocated RMA
2018 This allocates a Real Mode Area (RMA) from the pool allocated at boot
2019 time by the kernel. An RMA is a physically-contiguous, aligned region
2020 of memory used on older POWER processors to provide the memory which
2021 will be accessed by real-mode (MMU off) accesses in a KVM guest.
2022 POWER processors support a set of sizes for the RMA that usually
2023 includes 64MB, 128MB, 256MB and some larger powers of two.
2027 /* for KVM_ALLOCATE_RMA */
2028 struct kvm_allocate_rma {
2032 The return value is a file descriptor which can be passed to mmap(2)
2033 to map the allocated RMA into userspace. The mapped area can then be
2034 passed to the KVM_SET_USER_MEMORY_REGION ioctl to establish it as the
2035 RMA for a virtual machine. The size of the RMA in bytes (which is
2036 fixed at host kernel boot time) is returned in the rma_size field of
2037 the argument structure.
2039 The KVM_CAP_PPC_RMA capability is 1 or 2 if the KVM_ALLOCATE_RMA ioctl
2040 is supported; 2 if the processor requires all virtual machines to have
2041 an RMA, or 1 if the processor can use an RMA but doesn't require it,
2042 because it supports the Virtual RMA (VRMA) facility.
2048 :Capability: KVM_CAP_USER_NMI
2052 :Returns: 0 on success, -1 on error
2054 Queues an NMI on the thread's vcpu. Note this is well defined only
2055 when KVM_CREATE_IRQCHIP has not been called, since this is an interface
2056 between the virtual cpu core and virtual local APIC. After KVM_CREATE_IRQCHIP
2057 has been called, this interface is completely emulated within the kernel.
2059 To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the
2060 following algorithm:
2063 - read the local APIC's state (KVM_GET_LAPIC)
2064 - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1)
2065 - if so, issue KVM_NMI
2068 Some guests configure the LINT1 NMI input to cause a panic, aiding in
2072 4.65 KVM_S390_UCAS_MAP
2073 ----------------------
2075 :Capability: KVM_CAP_S390_UCONTROL
2076 :Architectures: s390
2078 :Parameters: struct kvm_s390_ucas_mapping (in)
2079 :Returns: 0 in case of success
2081 The parameter is defined like this::
2083 struct kvm_s390_ucas_mapping {
2089 This ioctl maps the memory at "user_addr" with the length "length" to
2090 the vcpu's address space starting at "vcpu_addr". All parameters need to
2091 be aligned by 1 megabyte.
2094 4.66 KVM_S390_UCAS_UNMAP
2095 ------------------------
2097 :Capability: KVM_CAP_S390_UCONTROL
2098 :Architectures: s390
2100 :Parameters: struct kvm_s390_ucas_mapping (in)
2101 :Returns: 0 in case of success
2103 The parameter is defined like this::
2105 struct kvm_s390_ucas_mapping {
2111 This ioctl unmaps the memory in the vcpu's address space starting at
2112 "vcpu_addr" with the length "length". The field "user_addr" is ignored.
2113 All parameters need to be aligned by 1 megabyte.
2116 4.67 KVM_S390_VCPU_FAULT
2117 ------------------------
2119 :Capability: KVM_CAP_S390_UCONTROL
2120 :Architectures: s390
2122 :Parameters: vcpu absolute address (in)
2123 :Returns: 0 in case of success
2125 This call creates a page table entry on the virtual cpu's address space
2126 (for user controlled virtual machines) or the virtual machine's address
2127 space (for regular virtual machines). This only works for minor faults,
2128 thus it's recommended to access subject memory page via the user page
2129 table upfront. This is useful to handle validity intercepts for user
2130 controlled virtual machines to fault in the virtual cpu's lowcore pages
2131 prior to calling the KVM_RUN ioctl.
2134 4.68 KVM_SET_ONE_REG
2135 --------------------
2137 :Capability: KVM_CAP_ONE_REG
2140 :Parameters: struct kvm_one_reg (in)
2141 :Returns: 0 on success, negative value on failure
2145 ====== ============================================================
2146 Â ENOENT Â Â no such register
2147 Â EINVAL Â Â invalid register ID, or no such register or used with VMs in
2148 protected virtualization mode on s390
2149 Â EPERM Â Â Â (arm64) register access not allowed before vcpu finalization
2150 ====== ============================================================
2152 (These error codes are indicative only: do not rely on a specific error
2153 code being returned in a specific situation.)
2157 struct kvm_one_reg {
2162 Using this ioctl, a single vcpu register can be set to a specific value
2163 defined by user space with the passed in struct kvm_one_reg, where id
2164 refers to the register identifier as described below and addr is a pointer
2165 to a variable with the respective size. There can be architecture agnostic
2166 and architecture specific registers. Each have their own range of operation
2167 and their own constants and width. To keep track of the implemented
2168 registers, find a list below:
2170 ======= =============================== ============
2171 Arch Register Width (bits)
2172 ======= =============================== ============
2173 PPC KVM_REG_PPC_HIOR 64
2174 PPC KVM_REG_PPC_IAC1 64
2175 PPC KVM_REG_PPC_IAC2 64
2176 PPC KVM_REG_PPC_IAC3 64
2177 PPC KVM_REG_PPC_IAC4 64
2178 PPC KVM_REG_PPC_DAC1 64
2179 PPC KVM_REG_PPC_DAC2 64
2180 PPC KVM_REG_PPC_DABR 64
2181 PPC KVM_REG_PPC_DSCR 64
2182 PPC KVM_REG_PPC_PURR 64
2183 PPC KVM_REG_PPC_SPURR 64
2184 PPC KVM_REG_PPC_DAR 64
2185 PPC KVM_REG_PPC_DSISR 32
2186 PPC KVM_REG_PPC_AMR 64
2187 PPC KVM_REG_PPC_UAMOR 64
2188 PPC KVM_REG_PPC_MMCR0 64
2189 PPC KVM_REG_PPC_MMCR1 64
2190 PPC KVM_REG_PPC_MMCRA 64
2191 PPC KVM_REG_PPC_MMCR2 64
2192 PPC KVM_REG_PPC_MMCRS 64
2193 PPC KVM_REG_PPC_MMCR3 64
2194 PPC KVM_REG_PPC_SIAR 64
2195 PPC KVM_REG_PPC_SDAR 64
2196 PPC KVM_REG_PPC_SIER 64
2197 PPC KVM_REG_PPC_SIER2 64
2198 PPC KVM_REG_PPC_SIER3 64
2199 PPC KVM_REG_PPC_PMC1 32
2200 PPC KVM_REG_PPC_PMC2 32
2201 PPC KVM_REG_PPC_PMC3 32
2202 PPC KVM_REG_PPC_PMC4 32
2203 PPC KVM_REG_PPC_PMC5 32
2204 PPC KVM_REG_PPC_PMC6 32
2205 PPC KVM_REG_PPC_PMC7 32
2206 PPC KVM_REG_PPC_PMC8 32
2207 PPC KVM_REG_PPC_FPR0 64
2209 PPC KVM_REG_PPC_FPR31 64
2210 PPC KVM_REG_PPC_VR0 128
2212 PPC KVM_REG_PPC_VR31 128
2213 PPC KVM_REG_PPC_VSR0 128
2215 PPC KVM_REG_PPC_VSR31 128
2216 PPC KVM_REG_PPC_FPSCR 64
2217 PPC KVM_REG_PPC_VSCR 32
2218 PPC KVM_REG_PPC_VPA_ADDR 64
2219 PPC KVM_REG_PPC_VPA_SLB 128
2220 PPC KVM_REG_PPC_VPA_DTL 128
2221 PPC KVM_REG_PPC_EPCR 32
2222 PPC KVM_REG_PPC_EPR 32
2223 PPC KVM_REG_PPC_TCR 32
2224 PPC KVM_REG_PPC_TSR 32
2225 PPC KVM_REG_PPC_OR_TSR 32
2226 PPC KVM_REG_PPC_CLEAR_TSR 32
2227 PPC KVM_REG_PPC_MAS0 32
2228 PPC KVM_REG_PPC_MAS1 32
2229 PPC KVM_REG_PPC_MAS2 64
2230 PPC KVM_REG_PPC_MAS7_3 64
2231 PPC KVM_REG_PPC_MAS4 32
2232 PPC KVM_REG_PPC_MAS6 32
2233 PPC KVM_REG_PPC_MMUCFG 32
2234 PPC KVM_REG_PPC_TLB0CFG 32
2235 PPC KVM_REG_PPC_TLB1CFG 32
2236 PPC KVM_REG_PPC_TLB2CFG 32
2237 PPC KVM_REG_PPC_TLB3CFG 32
2238 PPC KVM_REG_PPC_TLB0PS 32
2239 PPC KVM_REG_PPC_TLB1PS 32
2240 PPC KVM_REG_PPC_TLB2PS 32
2241 PPC KVM_REG_PPC_TLB3PS 32
2242 PPC KVM_REG_PPC_EPTCFG 32
2243 PPC KVM_REG_PPC_ICP_STATE 64
2244 PPC KVM_REG_PPC_VP_STATE 128
2245 PPC KVM_REG_PPC_TB_OFFSET 64
2246 PPC KVM_REG_PPC_SPMC1 32
2247 PPC KVM_REG_PPC_SPMC2 32
2248 PPC KVM_REG_PPC_IAMR 64
2249 PPC KVM_REG_PPC_TFHAR 64
2250 PPC KVM_REG_PPC_TFIAR 64
2251 PPC KVM_REG_PPC_TEXASR 64
2252 PPC KVM_REG_PPC_FSCR 64
2253 PPC KVM_REG_PPC_PSPB 32
2254 PPC KVM_REG_PPC_EBBHR 64
2255 PPC KVM_REG_PPC_EBBRR 64
2256 PPC KVM_REG_PPC_BESCR 64
2257 PPC KVM_REG_PPC_TAR 64
2258 PPC KVM_REG_PPC_DPDES 64
2259 PPC KVM_REG_PPC_DAWR 64
2260 PPC KVM_REG_PPC_DAWRX 64
2261 PPC KVM_REG_PPC_CIABR 64
2262 PPC KVM_REG_PPC_IC 64
2263 PPC KVM_REG_PPC_VTB 64
2264 PPC KVM_REG_PPC_CSIGR 64
2265 PPC KVM_REG_PPC_TACR 64
2266 PPC KVM_REG_PPC_TCSCR 64
2267 PPC KVM_REG_PPC_PID 64
2268 PPC KVM_REG_PPC_ACOP 64
2269 PPC KVM_REG_PPC_VRSAVE 32
2270 PPC KVM_REG_PPC_LPCR 32
2271 PPC KVM_REG_PPC_LPCR_64 64
2272 PPC KVM_REG_PPC_PPR 64
2273 PPC KVM_REG_PPC_ARCH_COMPAT 32
2274 PPC KVM_REG_PPC_DABRX 32
2275 PPC KVM_REG_PPC_WORT 64
2276 PPC KVM_REG_PPC_SPRG9 64
2277 PPC KVM_REG_PPC_DBSR 32
2278 PPC KVM_REG_PPC_TIDR 64
2279 PPC KVM_REG_PPC_PSSCR 64
2280 PPC KVM_REG_PPC_DEC_EXPIRY 64
2281 PPC KVM_REG_PPC_PTCR 64
2282 PPC KVM_REG_PPC_DAWR1 64
2283 PPC KVM_REG_PPC_DAWRX1 64
2284 PPC KVM_REG_PPC_TM_GPR0 64
2286 PPC KVM_REG_PPC_TM_GPR31 64
2287 PPC KVM_REG_PPC_TM_VSR0 128
2289 PPC KVM_REG_PPC_TM_VSR63 128
2290 PPC KVM_REG_PPC_TM_CR 64
2291 PPC KVM_REG_PPC_TM_LR 64
2292 PPC KVM_REG_PPC_TM_CTR 64
2293 PPC KVM_REG_PPC_TM_FPSCR 64
2294 PPC KVM_REG_PPC_TM_AMR 64
2295 PPC KVM_REG_PPC_TM_PPR 64
2296 PPC KVM_REG_PPC_TM_VRSAVE 64
2297 PPC KVM_REG_PPC_TM_VSCR 32
2298 PPC KVM_REG_PPC_TM_DSCR 64
2299 PPC KVM_REG_PPC_TM_TAR 64
2300 PPC KVM_REG_PPC_TM_XER 64
2302 MIPS KVM_REG_MIPS_R0 64
2304 MIPS KVM_REG_MIPS_R31 64
2305 MIPS KVM_REG_MIPS_HI 64
2306 MIPS KVM_REG_MIPS_LO 64
2307 MIPS KVM_REG_MIPS_PC 64
2308 MIPS KVM_REG_MIPS_CP0_INDEX 32
2309 MIPS KVM_REG_MIPS_CP0_ENTRYLO0 64
2310 MIPS KVM_REG_MIPS_CP0_ENTRYLO1 64
2311 MIPS KVM_REG_MIPS_CP0_CONTEXT 64
2312 MIPS KVM_REG_MIPS_CP0_CONTEXTCONFIG 32
2313 MIPS KVM_REG_MIPS_CP0_USERLOCAL 64
2314 MIPS KVM_REG_MIPS_CP0_XCONTEXTCONFIG 64
2315 MIPS KVM_REG_MIPS_CP0_PAGEMASK 32
2316 MIPS KVM_REG_MIPS_CP0_PAGEGRAIN 32
2317 MIPS KVM_REG_MIPS_CP0_SEGCTL0 64
2318 MIPS KVM_REG_MIPS_CP0_SEGCTL1 64
2319 MIPS KVM_REG_MIPS_CP0_SEGCTL2 64
2320 MIPS KVM_REG_MIPS_CP0_PWBASE 64
2321 MIPS KVM_REG_MIPS_CP0_PWFIELD 64
2322 MIPS KVM_REG_MIPS_CP0_PWSIZE 64
2323 MIPS KVM_REG_MIPS_CP0_WIRED 32
2324 MIPS KVM_REG_MIPS_CP0_PWCTL 32
2325 MIPS KVM_REG_MIPS_CP0_HWRENA 32
2326 MIPS KVM_REG_MIPS_CP0_BADVADDR 64
2327 MIPS KVM_REG_MIPS_CP0_BADINSTR 32
2328 MIPS KVM_REG_MIPS_CP0_BADINSTRP 32
2329 MIPS KVM_REG_MIPS_CP0_COUNT 32
2330 MIPS KVM_REG_MIPS_CP0_ENTRYHI 64
2331 MIPS KVM_REG_MIPS_CP0_COMPARE 32
2332 MIPS KVM_REG_MIPS_CP0_STATUS 32
2333 MIPS KVM_REG_MIPS_CP0_INTCTL 32
2334 MIPS KVM_REG_MIPS_CP0_CAUSE 32
2335 MIPS KVM_REG_MIPS_CP0_EPC 64
2336 MIPS KVM_REG_MIPS_CP0_PRID 32
2337 MIPS KVM_REG_MIPS_CP0_EBASE 64
2338 MIPS KVM_REG_MIPS_CP0_CONFIG 32
2339 MIPS KVM_REG_MIPS_CP0_CONFIG1 32
2340 MIPS KVM_REG_MIPS_CP0_CONFIG2 32
2341 MIPS KVM_REG_MIPS_CP0_CONFIG3 32
2342 MIPS KVM_REG_MIPS_CP0_CONFIG4 32
2343 MIPS KVM_REG_MIPS_CP0_CONFIG5 32
2344 MIPS KVM_REG_MIPS_CP0_CONFIG7 32
2345 MIPS KVM_REG_MIPS_CP0_XCONTEXT 64
2346 MIPS KVM_REG_MIPS_CP0_ERROREPC 64
2347 MIPS KVM_REG_MIPS_CP0_KSCRATCH1 64
2348 MIPS KVM_REG_MIPS_CP0_KSCRATCH2 64
2349 MIPS KVM_REG_MIPS_CP0_KSCRATCH3 64
2350 MIPS KVM_REG_MIPS_CP0_KSCRATCH4 64
2351 MIPS KVM_REG_MIPS_CP0_KSCRATCH5 64
2352 MIPS KVM_REG_MIPS_CP0_KSCRATCH6 64
2353 MIPS KVM_REG_MIPS_CP0_MAAR(0..63) 64
2354 MIPS KVM_REG_MIPS_COUNT_CTL 64
2355 MIPS KVM_REG_MIPS_COUNT_RESUME 64
2356 MIPS KVM_REG_MIPS_COUNT_HZ 64
2357 MIPS KVM_REG_MIPS_FPR_32(0..31) 32
2358 MIPS KVM_REG_MIPS_FPR_64(0..31) 64
2359 MIPS KVM_REG_MIPS_VEC_128(0..31) 128
2360 MIPS KVM_REG_MIPS_FCR_IR 32
2361 MIPS KVM_REG_MIPS_FCR_CSR 32
2362 MIPS KVM_REG_MIPS_MSA_IR 32
2363 MIPS KVM_REG_MIPS_MSA_CSR 32
2364 ======= =============================== ============
2366 ARM registers are mapped using the lower 32 bits. The upper 16 of that
2367 is the register group type, or coprocessor number:
2369 ARM core registers have the following id bit patterns::
2371 0x4020 0000 0010 <index into the kvm_regs struct:16>
2373 ARM 32-bit CP15 registers have the following id bit patterns::
2375 0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3>
2377 ARM 64-bit CP15 registers have the following id bit patterns::
2379 0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3>
2381 ARM CCSIDR registers are demultiplexed by CSSELR value::
2383 0x4020 0000 0011 00 <csselr:8>
2385 ARM 32-bit VFP control registers have the following id bit patterns::
2387 0x4020 0000 0012 1 <regno:12>
2389 ARM 64-bit FP registers have the following id bit patterns::
2391 0x4030 0000 0012 0 <regno:12>
2393 ARM firmware pseudo-registers have the following bit pattern::
2395 0x4030 0000 0014 <regno:16>
2398 arm64 registers are mapped using the lower 32 bits. The upper 16 of
2399 that is the register group type, or coprocessor number:
2401 arm64 core/FP-SIMD registers have the following id bit patterns. Note
2402 that the size of the access is variable, as the kvm_regs structure
2403 contains elements ranging from 32 to 128 bits. The index is a 32bit
2404 value in the kvm_regs structure seen as a 32bit array::
2406 0x60x0 0000 0010 <index into the kvm_regs struct:16>
2410 ======================= ========= ===== =======================================
2411 Encoding Register Bits kvm_regs member
2412 ======================= ========= ===== =======================================
2413 0x6030 0000 0010 0000 X0 64 regs.regs[0]
2414 0x6030 0000 0010 0002 X1 64 regs.regs[1]
2416 0x6030 0000 0010 003c X30 64 regs.regs[30]
2417 0x6030 0000 0010 003e SP 64 regs.sp
2418 0x6030 0000 0010 0040 PC 64 regs.pc
2419 0x6030 0000 0010 0042 PSTATE 64 regs.pstate
2420 0x6030 0000 0010 0044 SP_EL1 64 sp_el1
2421 0x6030 0000 0010 0046 ELR_EL1 64 elr_el1
2422 0x6030 0000 0010 0048 SPSR_EL1 64 spsr[KVM_SPSR_EL1] (alias SPSR_SVC)
2423 0x6030 0000 0010 004a SPSR_ABT 64 spsr[KVM_SPSR_ABT]
2424 0x6030 0000 0010 004c SPSR_UND 64 spsr[KVM_SPSR_UND]
2425 0x6030 0000 0010 004e SPSR_IRQ 64 spsr[KVM_SPSR_IRQ]
2426 0x6060 0000 0010 0050 SPSR_FIQ 64 spsr[KVM_SPSR_FIQ]
2427 0x6040 0000 0010 0054 V0 128 fp_regs.vregs[0] [1]_
2428 0x6040 0000 0010 0058 V1 128 fp_regs.vregs[1] [1]_
2430 0x6040 0000 0010 00d0 V31 128 fp_regs.vregs[31] [1]_
2431 0x6020 0000 0010 00d4 FPSR 32 fp_regs.fpsr
2432 0x6020 0000 0010 00d5 FPCR 32 fp_regs.fpcr
2433 ======================= ========= ===== =======================================
2435 .. [1] These encodings are not accepted for SVE-enabled vcpus. See
2438 The equivalent register content can be accessed via bits [127:0] of
2439 the corresponding SVE Zn registers instead for vcpus that have SVE
2440 enabled (see below).
2442 arm64 CCSIDR registers are demultiplexed by CSSELR value::
2444 0x6020 0000 0011 00 <csselr:8>
2446 arm64 system registers have the following id bit patterns::
2448 0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3>
2452 Two system register IDs do not follow the specified pattern. These
2453 are KVM_REG_ARM_TIMER_CVAL and KVM_REG_ARM_TIMER_CNT, which map to
2454 system registers CNTV_CVAL_EL0 and CNTVCT_EL0 respectively. These
2455 two had their values accidentally swapped, which means TIMER_CVAL is
2456 derived from the register encoding for CNTVCT_EL0 and TIMER_CNT is
2457 derived from the register encoding for CNTV_CVAL_EL0. As this is
2458 API, it must remain this way.
2460 arm64 firmware pseudo-registers have the following bit pattern::
2462 0x6030 0000 0014 <regno:16>
2464 arm64 SVE registers have the following bit patterns::
2466 0x6080 0000 0015 00 <n:5> <slice:5> Zn bits[2048*slice + 2047 : 2048*slice]
2467 0x6050 0000 0015 04 <n:4> <slice:5> Pn bits[256*slice + 255 : 256*slice]
2468 0x6050 0000 0015 060 <slice:5> FFR bits[256*slice + 255 : 256*slice]
2469 0x6060 0000 0015 ffff KVM_REG_ARM64_SVE_VLS pseudo-register
2471 Access to register IDs where 2048 * slice >= 128 * max_vq will fail with
2472 ENOENT. max_vq is the vcpu's maximum supported vector length in 128-bit
2473 quadwords: see [2]_ below.
2475 These registers are only accessible on vcpus for which SVE is enabled.
2476 See KVM_ARM_VCPU_INIT for details.
2478 In addition, except for KVM_REG_ARM64_SVE_VLS, these registers are not
2479 accessible until the vcpu's SVE configuration has been finalized
2480 using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). See KVM_ARM_VCPU_INIT
2481 and KVM_ARM_VCPU_FINALIZE for more information about this procedure.
2483 KVM_REG_ARM64_SVE_VLS is a pseudo-register that allows the set of vector
2484 lengths supported by the vcpu to be discovered and configured by
2485 userspace. When transferred to or from user memory via KVM_GET_ONE_REG
2486 or KVM_SET_ONE_REG, the value of this register is of type
2487 __u64[KVM_ARM64_SVE_VLS_WORDS], and encodes the set of vector lengths as
2490 __u64 vector_lengths[KVM_ARM64_SVE_VLS_WORDS];
2492 if (vq >= SVE_VQ_MIN && vq <= SVE_VQ_MAX &&
2493 ((vector_lengths[(vq - KVM_ARM64_SVE_VQ_MIN) / 64] >>
2494 ((vq - KVM_ARM64_SVE_VQ_MIN) % 64)) & 1))
2495 /* Vector length vq * 16 bytes supported */
2497 /* Vector length vq * 16 bytes not supported */
2499 .. [2] The maximum value vq for which the above condition is true is
2500 max_vq. This is the maximum vector length available to the guest on
2501 this vcpu, and determines which register slices are visible through
2502 this ioctl interface.
2504 (See Documentation/arm64/sve.rst for an explanation of the "vq"
2507 KVM_REG_ARM64_SVE_VLS is only accessible after KVM_ARM_VCPU_INIT.
2508 KVM_ARM_VCPU_INIT initialises it to the best set of vector lengths that
2511 Userspace may subsequently modify it if desired until the vcpu's SVE
2512 configuration is finalized using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE).
2514 Apart from simply removing all vector lengths from the host set that
2515 exceed some value, support for arbitrarily chosen sets of vector lengths
2516 is hardware-dependent and may not be available. Attempting to configure
2517 an invalid set of vector lengths via KVM_SET_ONE_REG will fail with
2520 After the vcpu's SVE configuration is finalized, further attempts to
2521 write this register will fail with EPERM.
2524 MIPS registers are mapped using the lower 32 bits. The upper 16 of that is
2525 the register group type:
2527 MIPS core registers (see above) have the following id bit patterns::
2529 0x7030 0000 0000 <reg:16>
2531 MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit
2532 patterns depending on whether they're 32-bit or 64-bit registers::
2534 0x7020 0000 0001 00 <reg:5> <sel:3> (32-bit)
2535 0x7030 0000 0001 00 <reg:5> <sel:3> (64-bit)
2537 Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_MIPS_CP0_ENTRYLO1 are the MIPS64
2538 versions of the EntryLo registers regardless of the word size of the host
2539 hardware, host kernel, guest, and whether XPA is present in the guest, i.e.
2540 with the RI and XI bits (if they exist) in bits 63 and 62 respectively, and
2541 the PFNX field starting at bit 30.
2543 MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) above) have the following id bit
2546 0x7030 0000 0001 01 <reg:8>
2548 MIPS KVM control registers (see above) have the following id bit patterns::
2550 0x7030 0000 0002 <reg:16>
2552 MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following
2553 id bit patterns depending on the size of the register being accessed. They are
2554 always accessed according to the current guest FPU mode (Status.FR and
2555 Config5.FRE), i.e. as the guest would see them, and they become unpredictable
2556 if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector
2557 registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they
2558 overlap the FPU registers::
2560 0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers)
2561 0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers)
2562 0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers)
2564 MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the
2565 following id bit patterns::
2567 0x7020 0000 0003 01 <0:3> <reg:5>
2569 MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the
2570 following id bit patterns::
2572 0x7020 0000 0003 02 <0:3> <reg:5>
2575 4.69 KVM_GET_ONE_REG
2576 --------------------
2578 :Capability: KVM_CAP_ONE_REG
2581 :Parameters: struct kvm_one_reg (in and out)
2582 :Returns: 0 on success, negative value on failure
2586 ======== ============================================================
2587 Â ENOENT Â Â no such register
2588 Â EINVAL Â Â invalid register ID, or no such register or used with VMs in
2589 protected virtualization mode on s390
2590 Â EPERM Â Â Â (arm64) register access not allowed before vcpu finalization
2591 ======== ============================================================
2593 (These error codes are indicative only: do not rely on a specific error
2594 code being returned in a specific situation.)
2596 This ioctl allows to receive the value of a single register implemented
2597 in a vcpu. The register to read is indicated by the "id" field of the
2598 kvm_one_reg struct passed in. On success, the register value can be found
2599 at the memory location pointed to by "addr".
2601 The list of registers accessible using this interface is identical to the
2605 4.70 KVM_KVMCLOCK_CTRL
2606 ----------------------
2608 :Capability: KVM_CAP_KVMCLOCK_CTRL
2609 :Architectures: Any that implement pvclocks (currently x86 only)
2612 :Returns: 0 on success, -1 on error
2614 This ioctl sets a flag accessible to the guest indicating that the specified
2615 vCPU has been paused by the host userspace.
2617 The host will set a flag in the pvclock structure that is checked from the
2618 soft lockup watchdog. The flag is part of the pvclock structure that is
2619 shared between guest and host, specifically the second bit of the flags
2620 field of the pvclock_vcpu_time_info structure. It will be set exclusively by
2621 the host and read/cleared exclusively by the guest. The guest operation of
2622 checking and clearing the flag must be an atomic operation so
2623 load-link/store-conditional, or equivalent must be used. There are two cases
2624 where the guest will clear the flag: when the soft lockup watchdog timer resets
2625 itself or when a soft lockup is detected. This ioctl can be called any time
2626 after pausing the vcpu, but before it is resumed.
2632 :Capability: KVM_CAP_SIGNAL_MSI
2633 :Architectures: x86 arm arm64
2635 :Parameters: struct kvm_msi (in)
2636 :Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error
2638 Directly inject a MSI message. Only valid with in-kernel irqchip that handles
2653 KVM_MSI_VALID_DEVID: devid contains a valid value. The per-VM
2654 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
2655 the device ID. If this capability is not available, userspace
2656 should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
2658 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
2659 for the device that wrote the MSI message. For PCI, this is usually a
2660 BFD identifier in the lower 16 bits.
2662 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
2663 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
2664 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
2665 address_hi must be zero.
2668 4.71 KVM_CREATE_PIT2
2669 --------------------
2671 :Capability: KVM_CAP_PIT2
2674 :Parameters: struct kvm_pit_config (in)
2675 :Returns: 0 on success, -1 on error
2677 Creates an in-kernel device model for the i8254 PIT. This call is only valid
2678 after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following
2679 parameters have to be passed::
2681 struct kvm_pit_config {
2688 #define KVM_PIT_SPEAKER_DUMMY 1 /* emulate speaker port stub */
2690 PIT timer interrupts may use a per-VM kernel thread for injection. If it
2691 exists, this thread will have a name of the following pattern::
2693 kvm-pit/<owner-process-pid>
2695 When running a guest with elevated priorities, the scheduling parameters of
2696 this thread may have to be adjusted accordingly.
2698 This IOCTL replaces the obsolete KVM_CREATE_PIT.
2704 :Capability: KVM_CAP_PIT_STATE2
2707 :Parameters: struct kvm_pit_state2 (out)
2708 :Returns: 0 on success, -1 on error
2710 Retrieves the state of the in-kernel PIT model. Only valid after
2711 KVM_CREATE_PIT2. The state is returned in the following structure::
2713 struct kvm_pit_state2 {
2714 struct kvm_pit_channel_state channels[3];
2721 /* disable PIT in HPET legacy mode */
2722 #define KVM_PIT_FLAGS_HPET_LEGACY 0x00000001
2724 This IOCTL replaces the obsolete KVM_GET_PIT.
2730 :Capability: KVM_CAP_PIT_STATE2
2733 :Parameters: struct kvm_pit_state2 (in)
2734 :Returns: 0 on success, -1 on error
2736 Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2.
2737 See KVM_GET_PIT2 for details on struct kvm_pit_state2.
2739 This IOCTL replaces the obsolete KVM_SET_PIT.
2742 4.74 KVM_PPC_GET_SMMU_INFO
2743 --------------------------
2745 :Capability: KVM_CAP_PPC_GET_SMMU_INFO
2746 :Architectures: powerpc
2749 :Returns: 0 on success, -1 on error
2751 This populates and returns a structure describing the features of
2752 the "Server" class MMU emulation supported by KVM.
2753 This can in turn be used by userspace to generate the appropriate
2754 device-tree properties for the guest operating system.
2756 The structure contains some global information, followed by an
2757 array of supported segment page sizes::
2759 struct kvm_ppc_smmu_info {
2763 struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ];
2766 The supported flags are:
2768 - KVM_PPC_PAGE_SIZES_REAL:
2769 When that flag is set, guest page sizes must "fit" the backing
2770 store page sizes. When not set, any page size in the list can
2771 be used regardless of how they are backed by userspace.
2773 - KVM_PPC_1T_SEGMENTS
2774 The emulated MMU supports 1T segments in addition to the
2778 This flag indicates that HPT guests are not supported by KVM,
2779 thus all guests must use radix MMU mode.
2781 The "slb_size" field indicates how many SLB entries are supported
2783 The "sps" array contains 8 entries indicating the supported base
2784 page sizes for a segment in increasing order. Each entry is defined
2787 struct kvm_ppc_one_seg_page_size {
2788 __u32 page_shift; /* Base page shift of segment (or 0) */
2789 __u32 slb_enc; /* SLB encoding for BookS */
2790 struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ];
2793 An entry with a "page_shift" of 0 is unused. Because the array is
2794 organized in increasing order, a lookup can stop when encoutering
2797 The "slb_enc" field provides the encoding to use in the SLB for the
2798 page size. The bits are in positions such as the value can directly
2799 be OR'ed into the "vsid" argument of the slbmte instruction.
2801 The "enc" array is a list which for each of those segment base page
2802 size provides the list of supported actual page sizes (which can be
2803 only larger or equal to the base page size), along with the
2804 corresponding encoding in the hash PTE. Similarly, the array is
2805 8 entries sorted by increasing sizes and an entry with a "0" shift
2806 is an empty entry and a terminator::
2808 struct kvm_ppc_one_page_size {
2809 __u32 page_shift; /* Page shift (or 0) */
2810 __u32 pte_enc; /* Encoding in the HPTE (>>12) */
2813 The "pte_enc" field provides a value that can OR'ed into the hash
2814 PTE's RPN field (ie, it needs to be shifted left by 12 to OR it
2815 into the hash PTE second double word).
2820 :Capability: KVM_CAP_IRQFD
2821 :Architectures: x86 s390 arm arm64
2823 :Parameters: struct kvm_irqfd (in)
2824 :Returns: 0 on success, -1 on error
2826 Allows setting an eventfd to directly trigger a guest interrupt.
2827 kvm_irqfd.fd specifies the file descriptor to use as the eventfd and
2828 kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When
2829 an event is triggered on the eventfd, an interrupt is injected into
2830 the guest using the specified gsi pin. The irqfd is removed using
2831 the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd
2834 With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify
2835 mechanism allowing emulation of level-triggered, irqfd-based
2836 interrupts. When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an
2837 additional eventfd in the kvm_irqfd.resamplefd field. When operating
2838 in resample mode, posting of an interrupt through kvm_irq.fd asserts
2839 the specified gsi in the irqchip. When the irqchip is resampled, such
2840 as from an EOI, the gsi is de-asserted and the user is notified via
2841 kvm_irqfd.resamplefd. It is the user's responsibility to re-queue
2842 the interrupt if the device making use of it still requires service.
2843 Note that closing the resamplefd is not sufficient to disable the
2844 irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment
2845 and need not be specified with KVM_IRQFD_FLAG_DEASSIGN.
2847 On arm/arm64, gsi routing being supported, the following can happen:
2849 - in case no routing entry is associated to this gsi, injection fails
2850 - in case the gsi is associated to an irqchip routing entry,
2851 irqchip.pin + 32 corresponds to the injected SPI ID.
2852 - in case the gsi is associated to an MSI routing entry, the MSI
2853 message and device ID are translated into an LPI (support restricted
2854 to GICv3 ITS in-kernel emulation).
2856 4.76 KVM_PPC_ALLOCATE_HTAB
2857 --------------------------
2859 :Capability: KVM_CAP_PPC_ALLOC_HTAB
2860 :Architectures: powerpc
2862 :Parameters: Pointer to u32 containing hash table order (in/out)
2863 :Returns: 0 on success, -1 on error
2865 This requests the host kernel to allocate an MMU hash table for a
2866 guest using the PAPR paravirtualization interface. This only does
2867 anything if the kernel is configured to use the Book 3S HV style of
2868 virtualization. Otherwise the capability doesn't exist and the ioctl
2869 returns an ENOTTY error. The rest of this description assumes Book 3S
2872 There must be no vcpus running when this ioctl is called; if there
2873 are, it will do nothing and return an EBUSY error.
2875 The parameter is a pointer to a 32-bit unsigned integer variable
2876 containing the order (log base 2) of the desired size of the hash
2877 table, which must be between 18 and 46. On successful return from the
2878 ioctl, the value will not be changed by the kernel.
2880 If no hash table has been allocated when any vcpu is asked to run
2881 (with the KVM_RUN ioctl), the host kernel will allocate a
2882 default-sized hash table (16 MB).
2884 If this ioctl is called when a hash table has already been allocated,
2885 with a different order from the existing hash table, the existing hash
2886 table will be freed and a new one allocated. If this is ioctl is
2887 called when a hash table has already been allocated of the same order
2888 as specified, the kernel will clear out the existing hash table (zero
2889 all HPTEs). In either case, if the guest is using the virtualized
2890 real-mode area (VRMA) facility, the kernel will re-create the VMRA
2891 HPTEs on the next KVM_RUN of any vcpu.
2893 4.77 KVM_S390_INTERRUPT
2894 -----------------------
2897 :Architectures: s390
2898 :Type: vm ioctl, vcpu ioctl
2899 :Parameters: struct kvm_s390_interrupt (in)
2900 :Returns: 0 on success, -1 on error
2902 Allows to inject an interrupt to the guest. Interrupts can be floating
2903 (vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type.
2905 Interrupt parameters are passed via kvm_s390_interrupt::
2907 struct kvm_s390_interrupt {
2913 type can be one of the following:
2915 KVM_S390_SIGP_STOP (vcpu)
2916 - sigp stop; optional flags in parm
2917 KVM_S390_PROGRAM_INT (vcpu)
2918 - program check; code in parm
2919 KVM_S390_SIGP_SET_PREFIX (vcpu)
2920 - sigp set prefix; prefix address in parm
2921 KVM_S390_RESTART (vcpu)
2923 KVM_S390_INT_CLOCK_COMP (vcpu)
2924 - clock comparator interrupt
2925 KVM_S390_INT_CPU_TIMER (vcpu)
2926 - CPU timer interrupt
2927 KVM_S390_INT_VIRTIO (vm)
2928 - virtio external interrupt; external interrupt
2929 parameters in parm and parm64
2930 KVM_S390_INT_SERVICE (vm)
2931 - sclp external interrupt; sclp parameter in parm
2932 KVM_S390_INT_EMERGENCY (vcpu)
2933 - sigp emergency; source cpu in parm
2934 KVM_S390_INT_EXTERNAL_CALL (vcpu)
2935 - sigp external call; source cpu in parm
2936 KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm)
2937 - compound value to indicate an
2938 I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel);
2939 I/O interruption parameters in parm (subchannel) and parm64 (intparm,
2940 interruption subclass)
2941 KVM_S390_MCHK (vm, vcpu)
2942 - machine check interrupt; cr 14 bits in parm, machine check interrupt
2943 code in parm64 (note that machine checks needing further payload are not
2944 supported by this ioctl)
2946 This is an asynchronous vcpu ioctl and can be invoked from any thread.
2948 4.78 KVM_PPC_GET_HTAB_FD
2949 ------------------------
2951 :Capability: KVM_CAP_PPC_HTAB_FD
2952 :Architectures: powerpc
2954 :Parameters: Pointer to struct kvm_get_htab_fd (in)
2955 :Returns: file descriptor number (>= 0) on success, -1 on error
2957 This returns a file descriptor that can be used either to read out the
2958 entries in the guest's hashed page table (HPT), or to write entries to
2959 initialize the HPT. The returned fd can only be written to if the
2960 KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and
2961 can only be read if that bit is clear. The argument struct looks like
2964 /* For KVM_PPC_GET_HTAB_FD */
2965 struct kvm_get_htab_fd {
2971 /* Values for kvm_get_htab_fd.flags */
2972 #define KVM_GET_HTAB_BOLTED_ONLY ((__u64)0x1)
2973 #define KVM_GET_HTAB_WRITE ((__u64)0x2)
2975 The 'start_index' field gives the index in the HPT of the entry at
2976 which to start reading. It is ignored when writing.
2978 Reads on the fd will initially supply information about all
2979 "interesting" HPT entries. Interesting entries are those with the
2980 bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise
2981 all entries. When the end of the HPT is reached, the read() will
2982 return. If read() is called again on the fd, it will start again from
2983 the beginning of the HPT, but will only return HPT entries that have
2984 changed since they were last read.
2986 Data read or written is structured as a header (8 bytes) followed by a
2987 series of valid HPT entries (16 bytes) each. The header indicates how
2988 many valid HPT entries there are and how many invalid entries follow
2989 the valid entries. The invalid entries are not represented explicitly
2990 in the stream. The header format is::
2992 struct kvm_get_htab_header {
2998 Writes to the fd create HPT entries starting at the index given in the
2999 header; first 'n_valid' valid entries with contents from the data
3000 written, then 'n_invalid' invalid entries, invalidating any previously
3001 valid entries found.
3003 4.79 KVM_CREATE_DEVICE
3004 ----------------------
3006 :Capability: KVM_CAP_DEVICE_CTRL
3008 :Parameters: struct kvm_create_device (in/out)
3009 :Returns: 0 on success, -1 on error
3013 ====== =======================================================
3014 ENODEV The device type is unknown or unsupported
3015 EEXIST Device already created, and this type of device may not
3016 be instantiated multiple times
3017 ====== =======================================================
3019 Other error conditions may be defined by individual device types or
3020 have their standard meanings.
3022 Creates an emulated device in the kernel. The file descriptor returned
3023 in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR.
3025 If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the
3026 device type is supported (not necessarily whether it can be created
3029 Individual devices should not define flags. Attributes should be used
3030 for specifying any behavior that is not implied by the device type
3035 struct kvm_create_device {
3036 __u32 type; /* in: KVM_DEV_TYPE_xxx */
3037 __u32 fd; /* out: device handle */
3038 __u32 flags; /* in: KVM_CREATE_DEVICE_xxx */
3041 4.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR
3042 --------------------------------------------
3044 :Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
3045 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
3046 :Type: device ioctl, vm ioctl, vcpu ioctl
3047 :Parameters: struct kvm_device_attr
3048 :Returns: 0 on success, -1 on error
3052 ===== =============================================================
3053 ENXIO The group or attribute is unknown/unsupported for this device
3054 or hardware support is missing.
3055 EPERM The attribute cannot (currently) be accessed this way
3056 (e.g. read-only attribute, or attribute that only makes
3057 sense when the device is in a different state)
3058 ===== =============================================================
3060 Other error conditions may be defined by individual device types.
3062 Gets/sets a specified piece of device configuration and/or state. The
3063 semantics are device-specific. See individual device documentation in
3064 the "devices" directory. As with ONE_REG, the size of the data
3065 transferred is defined by the particular attribute.
3069 struct kvm_device_attr {
3070 __u32 flags; /* no flags currently defined */
3071 __u32 group; /* device-defined */
3072 __u64 attr; /* group-defined */
3073 __u64 addr; /* userspace address of attr data */
3076 4.81 KVM_HAS_DEVICE_ATTR
3077 ------------------------
3079 :Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
3080 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
3081 :Type: device ioctl, vm ioctl, vcpu ioctl
3082 :Parameters: struct kvm_device_attr
3083 :Returns: 0 on success, -1 on error
3087 ===== =============================================================
3088 ENXIO The group or attribute is unknown/unsupported for this device
3089 or hardware support is missing.
3090 ===== =============================================================
3092 Tests whether a device supports a particular attribute. A successful
3093 return indicates the attribute is implemented. It does not necessarily
3094 indicate that the attribute can be read or written in the device's
3095 current state. "addr" is ignored.
3097 4.82 KVM_ARM_VCPU_INIT
3098 ----------------------
3101 :Architectures: arm, arm64
3103 :Parameters: struct kvm_vcpu_init (in)
3104 :Returns: 0 on success; -1 on error
3108 ====== =================================================================
3109 Â EINVAL Â Â Â the target is unknown, or the combination of features is invalid.
3110 Â ENOENT Â Â Â a features bit specified is unknown.
3111 ====== =================================================================
3113 This tells KVM what type of CPU to present to the guest, and what
3114 optional features it should have. Â This will cause a reset of the cpu
3115 registers to their initial values. Â If this is not called, KVM_RUN will
3116 return ENOEXEC for that vcpu.
3118 Note that because some registers reflect machine topology, all vcpus
3119 should be created before this ioctl is invoked.
3121 Userspace can call this function multiple times for a given vcpu, including
3122 after the vcpu has been run. This will reset the vcpu to its initial
3123 state. All calls to this function after the initial call must use the same
3124 target and same set of feature flags, otherwise EINVAL will be returned.
3128 - KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state.
3129 Depends on KVM_CAP_ARM_PSCI. If not set, the CPU will be powered on
3130 and execute guest code when KVM_RUN is called.
3131 - KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode.
3132 Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only).
3133 - KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 (or a future revision
3134 backward compatible with v0.2) for the CPU.
3135 Depends on KVM_CAP_ARM_PSCI_0_2.
3136 - KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU.
3137 Depends on KVM_CAP_ARM_PMU_V3.
3139 - KVM_ARM_VCPU_PTRAUTH_ADDRESS: Enables Address Pointer authentication
3141 Depends on KVM_CAP_ARM_PTRAUTH_ADDRESS.
3142 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are
3143 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and
3144 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be
3147 - KVM_ARM_VCPU_PTRAUTH_GENERIC: Enables Generic Pointer authentication
3149 Depends on KVM_CAP_ARM_PTRAUTH_GENERIC.
3150 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are
3151 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and
3152 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be
3155 - KVM_ARM_VCPU_SVE: Enables SVE for the CPU (arm64 only).
3156 Depends on KVM_CAP_ARM_SVE.
3157 Requires KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3159 * After KVM_ARM_VCPU_INIT:
3161 - KVM_REG_ARM64_SVE_VLS may be read using KVM_GET_ONE_REG: the
3162 initial value of this pseudo-register indicates the best set of
3163 vector lengths possible for a vcpu on this host.
3165 * Before KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3167 - KVM_RUN and KVM_GET_REG_LIST are not available;
3169 - KVM_GET_ONE_REG and KVM_SET_ONE_REG cannot be used to access
3170 the scalable archietctural SVE registers
3171 KVM_REG_ARM64_SVE_ZREG(), KVM_REG_ARM64_SVE_PREG() or
3172 KVM_REG_ARM64_SVE_FFR;
3174 - KVM_REG_ARM64_SVE_VLS may optionally be written using
3175 KVM_SET_ONE_REG, to modify the set of vector lengths available
3178 * After KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3180 - the KVM_REG_ARM64_SVE_VLS pseudo-register is immutable, and can
3181 no longer be written using KVM_SET_ONE_REG.
3183 4.83 KVM_ARM_PREFERRED_TARGET
3184 -----------------------------
3187 :Architectures: arm, arm64
3189 :Parameters: struct kvm_vcpu_init (out)
3190 :Returns: 0 on success; -1 on error
3194 ====== ==========================================
3195 ENODEV no preferred target available for the host
3196 ====== ==========================================
3198 This queries KVM for preferred CPU target type which can be emulated
3199 by KVM on underlying host.
3201 The ioctl returns struct kvm_vcpu_init instance containing information
3202 about preferred CPU target type and recommended features for it. The
3203 kvm_vcpu_init->features bitmap returned will have feature bits set if
3204 the preferred target recommends setting these features, but this is
3207 The information returned by this ioctl can be used to prepare an instance
3208 of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in
3209 VCPU matching underlying host.
3212 4.84 KVM_GET_REG_LIST
3213 ---------------------
3216 :Architectures: arm, arm64, mips
3218 :Parameters: struct kvm_reg_list (in/out)
3219 :Returns: 0 on success; -1 on error
3223 ===== ==============================================================
3224 Â E2BIG Â Â Â Â the reg index list is too big to fit in the array specified by
3225 Â Â Â Â Â Â Â Â Â Â Â Â the user (the number required will be written into n).
3226 ===== ==============================================================
3230 struct kvm_reg_list {
3231 __u64 n; /* number of registers in reg[] */
3235 This ioctl returns the guest registers that are supported for the
3236 KVM_GET_ONE_REG/KVM_SET_ONE_REG calls.
3239 4.85 KVM_ARM_SET_DEVICE_ADDR (deprecated)
3240 -----------------------------------------
3242 :Capability: KVM_CAP_ARM_SET_DEVICE_ADDR
3243 :Architectures: arm, arm64
3245 :Parameters: struct kvm_arm_device_address (in)
3246 :Returns: 0 on success, -1 on error
3250 ====== ============================================
3251 ENODEV The device id is unknown
3252 ENXIO Device not supported on current system
3253 EEXIST Address already set
3254 E2BIG Address outside guest physical address space
3255 EBUSY Address overlaps with other device range
3256 ====== ============================================
3260 struct kvm_arm_device_addr {
3265 Specify a device address in the guest's physical address space where guests
3266 can access emulated or directly exposed devices, which the host kernel needs
3267 to know about. The id field is an architecture specific identifier for a
3270 ARM/arm64 divides the id field into two parts, a device id and an
3271 address type id specific to the individual device::
3273 Â bits: | 63 ... 32 | 31 ... 16 | 15 ... 0 |
3274 field: | 0x00000000 | device id | addr type id |
3276 ARM/arm64 currently only require this when using the in-kernel GIC
3277 support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2
3278 as the device id. When setting the base address for the guest's
3279 mapping of the VGIC virtual CPU and distributor interface, the ioctl
3280 must be called after calling KVM_CREATE_IRQCHIP, but before calling
3281 KVM_RUN on any of the VCPUs. Calling this ioctl twice for any of the
3282 base addresses will return -EEXIST.
3284 Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API
3285 should be used instead.
3288 4.86 KVM_PPC_RTAS_DEFINE_TOKEN
3289 ------------------------------
3291 :Capability: KVM_CAP_PPC_RTAS
3294 :Parameters: struct kvm_rtas_token_args
3295 :Returns: 0 on success, -1 on error
3297 Defines a token value for a RTAS (Run Time Abstraction Services)
3298 service in order to allow it to be handled in the kernel. The
3299 argument struct gives the name of the service, which must be the name
3300 of a service that has a kernel-side implementation. If the token
3301 value is non-zero, it will be associated with that service, and
3302 subsequent RTAS calls by the guest specifying that token will be
3303 handled by the kernel. If the token value is 0, then any token
3304 associated with the service will be forgotten, and subsequent RTAS
3305 calls by the guest for that service will be passed to userspace to be
3308 4.87 KVM_SET_GUEST_DEBUG
3309 ------------------------
3311 :Capability: KVM_CAP_SET_GUEST_DEBUG
3312 :Architectures: x86, s390, ppc, arm64
3314 :Parameters: struct kvm_guest_debug (in)
3315 :Returns: 0 on success; -1 on error
3319 struct kvm_guest_debug {
3322 struct kvm_guest_debug_arch arch;
3325 Set up the processor specific debug registers and configure vcpu for
3326 handling guest debug events. There are two parts to the structure, the
3327 first a control bitfield indicates the type of debug events to handle
3328 when running. Common control bits are:
3330 - KVM_GUESTDBG_ENABLE: guest debugging is enabled
3331 - KVM_GUESTDBG_SINGLESTEP: the next run should single-step
3333 The top 16 bits of the control field are architecture specific control
3334 flags which can include the following:
3336 - KVM_GUESTDBG_USE_SW_BP: using software breakpoints [x86, arm64]
3337 - KVM_GUESTDBG_USE_HW_BP: using hardware breakpoints [x86, s390, arm64]
3338 - KVM_GUESTDBG_INJECT_DB: inject DB type exception [x86]
3339 - KVM_GUESTDBG_INJECT_BP: inject BP type exception [x86]
3340 - KVM_GUESTDBG_EXIT_PENDING: trigger an immediate guest exit [s390]
3342 For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints
3343 are enabled in memory so we need to ensure breakpoint exceptions are
3344 correctly trapped and the KVM run loop exits at the breakpoint and not
3345 running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP
3346 we need to ensure the guest vCPUs architecture specific registers are
3347 updated to the correct (supplied) values.
3349 The second part of the structure is architecture specific and
3350 typically contains a set of debug registers.
3352 For arm64 the number of debug registers is implementation defined and
3353 can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and
3354 KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number
3355 indicating the number of supported registers.
3357 For ppc, the KVM_CAP_PPC_GUEST_DEBUG_SSTEP capability indicates whether
3358 the single-step debug event (KVM_GUESTDBG_SINGLESTEP) is supported.
3360 When debug events exit the main run loop with the reason
3361 KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run
3362 structure containing architecture specific debug information.
3364 4.88 KVM_GET_EMULATED_CPUID
3365 ---------------------------
3367 :Capability: KVM_CAP_EXT_EMUL_CPUID
3370 :Parameters: struct kvm_cpuid2 (in/out)
3371 :Returns: 0 on success, -1 on error
3378 struct kvm_cpuid_entry2 entries[0];
3381 The member 'flags' is used for passing flags from userspace.
3385 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
3386 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */
3387 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */
3389 struct kvm_cpuid_entry2 {
3400 This ioctl returns x86 cpuid features which are emulated by
3401 kvm.Userspace can use the information returned by this ioctl to query
3402 which features are emulated by kvm instead of being present natively.
3404 Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2
3405 structure with the 'nent' field indicating the number of entries in
3406 the variable-size array 'entries'. If the number of entries is too low
3407 to describe the cpu capabilities, an error (E2BIG) is returned. If the
3408 number is too high, the 'nent' field is adjusted and an error (ENOMEM)
3409 is returned. If the number is just right, the 'nent' field is adjusted
3410 to the number of valid entries in the 'entries' array, which is then
3413 The entries returned are the set CPUID bits of the respective features
3414 which kvm emulates, as returned by the CPUID instruction, with unknown
3415 or unsupported feature bits cleared.
3417 Features like x2apic, for example, may not be present in the host cpu
3418 but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be
3419 emulated efficiently and thus not included here.
3421 The fields in each entry are defined as follows:
3424 the eax value used to obtain the entry
3426 the ecx value used to obtain the entry (for entries that are
3429 an OR of zero or more of the following:
3431 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
3432 if the index field is valid
3436 the values returned by the cpuid instruction for
3437 this function/index combination
3439 4.89 KVM_S390_MEM_OP
3440 --------------------
3442 :Capability: KVM_CAP_S390_MEM_OP
3443 :Architectures: s390
3445 :Parameters: struct kvm_s390_mem_op (in)
3446 :Returns: = 0 on success,
3447 < 0 on generic error (e.g. -EFAULT or -ENOMEM),
3448 > 0 if an exception occurred while walking the page tables
3450 Read or write data from/to the logical (virtual) memory of a VCPU.
3452 Parameters are specified via the following structure::
3454 struct kvm_s390_mem_op {
3455 __u64 gaddr; /* the guest address */
3456 __u64 flags; /* flags */
3457 __u32 size; /* amount of bytes */
3458 __u32 op; /* type of operation */
3459 __u64 buf; /* buffer in userspace */
3460 __u8 ar; /* the access register number */
3461 __u8 reserved[31]; /* should be set to 0 */
3464 The type of operation is specified in the "op" field. It is either
3465 KVM_S390_MEMOP_LOGICAL_READ for reading from logical memory space or
3466 KVM_S390_MEMOP_LOGICAL_WRITE for writing to logical memory space. The
3467 KVM_S390_MEMOP_F_CHECK_ONLY flag can be set in the "flags" field to check
3468 whether the corresponding memory access would create an access exception
3469 (without touching the data in the memory at the destination). In case an
3470 access exception occurred while walking the MMU tables of the guest, the
3471 ioctl returns a positive error number to indicate the type of exception.
3472 This exception is also raised directly at the corresponding VCPU if the
3473 flag KVM_S390_MEMOP_F_INJECT_EXCEPTION is set in the "flags" field.
3475 The start address of the memory region has to be specified in the "gaddr"
3476 field, and the length of the region in the "size" field (which must not
3477 be 0). The maximum value for "size" can be obtained by checking the
3478 KVM_CAP_S390_MEM_OP capability. "buf" is the buffer supplied by the
3479 userspace application where the read data should be written to for
3480 KVM_S390_MEMOP_LOGICAL_READ, or where the data that should be written is
3481 stored for a KVM_S390_MEMOP_LOGICAL_WRITE. When KVM_S390_MEMOP_F_CHECK_ONLY
3482 is specified, "buf" is unused and can be NULL. "ar" designates the access
3483 register number to be used; the valid range is 0..15.
3485 The "reserved" field is meant for future extensions. It is not used by
3486 KVM with the currently defined set of flags.
3488 4.90 KVM_S390_GET_SKEYS
3489 -----------------------
3491 :Capability: KVM_CAP_S390_SKEYS
3492 :Architectures: s390
3494 :Parameters: struct kvm_s390_skeys
3495 :Returns: 0 on success, KVM_S390_GET_KEYS_NONE if guest is not using storage
3496 keys, negative value on error
3498 This ioctl is used to get guest storage key values on the s390
3499 architecture. The ioctl takes parameters via the kvm_s390_skeys struct::
3501 struct kvm_s390_skeys {
3504 __u64 skeydata_addr;
3509 The start_gfn field is the number of the first guest frame whose storage keys
3512 The count field is the number of consecutive frames (starting from start_gfn)
3513 whose storage keys to get. The count field must be at least 1 and the maximum
3514 allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range
3515 will cause the ioctl to return -EINVAL.
3517 The skeydata_addr field is the address to a buffer large enough to hold count
3518 bytes. This buffer will be filled with storage key data by the ioctl.
3520 4.91 KVM_S390_SET_SKEYS
3521 -----------------------
3523 :Capability: KVM_CAP_S390_SKEYS
3524 :Architectures: s390
3526 :Parameters: struct kvm_s390_skeys
3527 :Returns: 0 on success, negative value on error
3529 This ioctl is used to set guest storage key values on the s390
3530 architecture. The ioctl takes parameters via the kvm_s390_skeys struct.
3531 See section on KVM_S390_GET_SKEYS for struct definition.
3533 The start_gfn field is the number of the first guest frame whose storage keys
3536 The count field is the number of consecutive frames (starting from start_gfn)
3537 whose storage keys to get. The count field must be at least 1 and the maximum
3538 allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range
3539 will cause the ioctl to return -EINVAL.
3541 The skeydata_addr field is the address to a buffer containing count bytes of
3542 storage keys. Each byte in the buffer will be set as the storage key for a
3543 single frame starting at start_gfn for count frames.
3545 Note: If any architecturally invalid key value is found in the given data then
3546 the ioctl will return -EINVAL.
3551 :Capability: KVM_CAP_S390_INJECT_IRQ
3552 :Architectures: s390
3554 :Parameters: struct kvm_s390_irq (in)
3555 :Returns: 0 on success, -1 on error
3560 ====== =================================================================
3561 EINVAL interrupt type is invalid
3562 type is KVM_S390_SIGP_STOP and flag parameter is invalid value,
3563 type is KVM_S390_INT_EXTERNAL_CALL and code is bigger
3564 than the maximum of VCPUs
3565 EBUSY type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped,
3566 type is KVM_S390_SIGP_STOP and a stop irq is already pending,
3567 type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt
3569 ====== =================================================================
3571 Allows to inject an interrupt to the guest.
3573 Using struct kvm_s390_irq as a parameter allows
3574 to inject additional payload which is not
3575 possible via KVM_S390_INTERRUPT.
3577 Interrupt parameters are passed via kvm_s390_irq::
3579 struct kvm_s390_irq {
3582 struct kvm_s390_io_info io;
3583 struct kvm_s390_ext_info ext;
3584 struct kvm_s390_pgm_info pgm;
3585 struct kvm_s390_emerg_info emerg;
3586 struct kvm_s390_extcall_info extcall;
3587 struct kvm_s390_prefix_info prefix;
3588 struct kvm_s390_stop_info stop;
3589 struct kvm_s390_mchk_info mchk;
3594 type can be one of the following:
3596 - KVM_S390_SIGP_STOP - sigp stop; parameter in .stop
3597 - KVM_S390_PROGRAM_INT - program check; parameters in .pgm
3598 - KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix
3599 - KVM_S390_RESTART - restart; no parameters
3600 - KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters
3601 - KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters
3602 - KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg
3603 - KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall
3604 - KVM_S390_MCHK - machine check interrupt; parameters in .mchk
3606 This is an asynchronous vcpu ioctl and can be invoked from any thread.
3608 4.94 KVM_S390_GET_IRQ_STATE
3609 ---------------------------
3611 :Capability: KVM_CAP_S390_IRQ_STATE
3612 :Architectures: s390
3614 :Parameters: struct kvm_s390_irq_state (out)
3615 :Returns: >= number of bytes copied into buffer,
3616 -EINVAL if buffer size is 0,
3617 -ENOBUFS if buffer size is too small to fit all pending interrupts,
3618 -EFAULT if the buffer address was invalid
3620 This ioctl allows userspace to retrieve the complete state of all currently
3621 pending interrupts in a single buffer. Use cases include migration
3622 and introspection. The parameter structure contains the address of a
3623 userspace buffer and its length::
3625 struct kvm_s390_irq_state {
3627 __u32 flags; /* will stay unused for compatibility reasons */
3629 __u32 reserved[4]; /* will stay unused for compatibility reasons */
3632 Userspace passes in the above struct and for each pending interrupt a
3633 struct kvm_s390_irq is copied to the provided buffer.
3635 The structure contains a flags and a reserved field for future extensions. As
3636 the kernel never checked for flags == 0 and QEMU never pre-zeroed flags and
3637 reserved, these fields can not be used in the future without breaking
3640 If -ENOBUFS is returned the buffer provided was too small and userspace
3641 may retry with a bigger buffer.
3643 4.95 KVM_S390_SET_IRQ_STATE
3644 ---------------------------
3646 :Capability: KVM_CAP_S390_IRQ_STATE
3647 :Architectures: s390
3649 :Parameters: struct kvm_s390_irq_state (in)
3650 :Returns: 0 on success,
3651 -EFAULT if the buffer address was invalid,
3652 -EINVAL for an invalid buffer length (see below),
3653 -EBUSY if there were already interrupts pending,
3654 errors occurring when actually injecting the
3655 interrupt. See KVM_S390_IRQ.
3657 This ioctl allows userspace to set the complete state of all cpu-local
3658 interrupts currently pending for the vcpu. It is intended for restoring
3659 interrupt state after a migration. The input parameter is a userspace buffer
3660 containing a struct kvm_s390_irq_state::
3662 struct kvm_s390_irq_state {
3664 __u32 flags; /* will stay unused for compatibility reasons */
3666 __u32 reserved[4]; /* will stay unused for compatibility reasons */
3669 The restrictions for flags and reserved apply as well.
3670 (see KVM_S390_GET_IRQ_STATE)
3672 The userspace memory referenced by buf contains a struct kvm_s390_irq
3673 for each interrupt to be injected into the guest.
3674 If one of the interrupts could not be injected for some reason the
3677 len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0
3678 and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq),
3679 which is the maximum number of possibly pending cpu-local interrupts.
3684 :Capability: KVM_CAP_X86_SMM
3688 :Returns: 0 on success, -1 on error
3690 Queues an SMI on the thread's vcpu.
3692 4.97 KVM_CAP_PPC_MULTITCE
3693 -------------------------
3695 :Capability: KVM_CAP_PPC_MULTITCE
3699 This capability means the kernel is capable of handling hypercalls
3700 H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user
3701 space. This significantly accelerates DMA operations for PPC KVM guests.
3702 User space should expect that its handlers for these hypercalls
3703 are not going to be called if user space previously registered LIOBN
3704 in KVM (via KVM_CREATE_SPAPR_TCE or similar calls).
3706 In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest,
3707 user space might have to advertise it for the guest. For example,
3708 IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is
3709 present in the "ibm,hypertas-functions" device-tree property.
3711 The hypercalls mentioned above may or may not be processed successfully
3712 in the kernel based fast path. If they can not be handled by the kernel,
3713 they will get passed on to user space. So user space still has to have
3714 an implementation for these despite the in kernel acceleration.
3716 This capability is always enabled.
3718 4.98 KVM_CREATE_SPAPR_TCE_64
3719 ----------------------------
3721 :Capability: KVM_CAP_SPAPR_TCE_64
3722 :Architectures: powerpc
3724 :Parameters: struct kvm_create_spapr_tce_64 (in)
3725 :Returns: file descriptor for manipulating the created TCE table
3727 This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit
3728 windows, described in 4.62 KVM_CREATE_SPAPR_TCE
3730 This capability uses extended struct in ioctl interface::
3732 /* for KVM_CAP_SPAPR_TCE_64 */
3733 struct kvm_create_spapr_tce_64 {
3737 __u64 offset; /* in pages */
3738 __u64 size; /* in pages */
3741 The aim of extension is to support an additional bigger DMA window with
3742 a variable page size.
3743 KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and
3744 a bus offset of the corresponding DMA window, @size and @offset are numbers
3747 @flags are not used at the moment.
3749 The rest of functionality is identical to KVM_CREATE_SPAPR_TCE.
3751 4.99 KVM_REINJECT_CONTROL
3752 -------------------------
3754 :Capability: KVM_CAP_REINJECT_CONTROL
3757 :Parameters: struct kvm_reinject_control (in)
3758 :Returns: 0 on success,
3759 -EFAULT if struct kvm_reinject_control cannot be read,
3760 -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier.
3762 i8254 (PIT) has two modes, reinject and !reinject. The default is reinject,
3763 where KVM queues elapsed i8254 ticks and monitors completion of interrupt from
3764 vector(s) that i8254 injects. Reinject mode dequeues a tick and injects its
3765 interrupt whenever there isn't a pending interrupt from i8254.
3766 !reinject mode injects an interrupt as soon as a tick arrives.
3770 struct kvm_reinject_control {
3775 pit_reinject = 0 (!reinject mode) is recommended, unless running an old
3776 operating system that uses the PIT for timing (e.g. Linux 2.4.x).
3778 4.100 KVM_PPC_CONFIGURE_V3_MMU
3779 ------------------------------
3781 :Capability: KVM_CAP_PPC_RADIX_MMU or KVM_CAP_PPC_HASH_MMU_V3
3784 :Parameters: struct kvm_ppc_mmuv3_cfg (in)
3785 :Returns: 0 on success,
3786 -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read,
3787 -EINVAL if the configuration is invalid
3789 This ioctl controls whether the guest will use radix or HPT (hashed
3790 page table) translation, and sets the pointer to the process table for
3795 struct kvm_ppc_mmuv3_cfg {
3797 __u64 process_table;
3800 There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and
3801 KVM_PPC_MMUV3_GTSE. KVM_PPC_MMUV3_RADIX, if set, configures the guest
3802 to use radix tree translation, and if clear, to use HPT translation.
3803 KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest
3804 to be able to use the global TLB and SLB invalidation instructions;
3805 if clear, the guest may not use these instructions.
3807 The process_table field specifies the address and size of the guest
3808 process table, which is in the guest's space. This field is formatted
3809 as the second doubleword of the partition table entry, as defined in
3810 the Power ISA V3.00, Book III section 5.7.6.1.
3812 4.101 KVM_PPC_GET_RMMU_INFO
3813 ---------------------------
3815 :Capability: KVM_CAP_PPC_RADIX_MMU
3818 :Parameters: struct kvm_ppc_rmmu_info (out)
3819 :Returns: 0 on success,
3820 -EFAULT if struct kvm_ppc_rmmu_info cannot be written,
3821 -EINVAL if no useful information can be returned
3823 This ioctl returns a structure containing two things: (a) a list
3824 containing supported radix tree geometries, and (b) a list that maps
3825 page sizes to put in the "AP" (actual page size) field for the tlbie
3826 (TLB invalidate entry) instruction.
3830 struct kvm_ppc_rmmu_info {
3831 struct kvm_ppc_radix_geom {
3836 __u32 ap_encodings[8];
3839 The geometries[] field gives up to 8 supported geometries for the
3840 radix page table, in terms of the log base 2 of the smallest page
3841 size, and the number of bits indexed at each level of the tree, from
3842 the PTE level up to the PGD level in that order. Any unused entries
3843 will have 0 in the page_shift field.
3845 The ap_encodings gives the supported page sizes and their AP field
3846 encodings, encoded with the AP value in the top 3 bits and the log
3847 base 2 of the page size in the bottom 6 bits.
3849 4.102 KVM_PPC_RESIZE_HPT_PREPARE
3850 --------------------------------
3852 :Capability: KVM_CAP_SPAPR_RESIZE_HPT
3853 :Architectures: powerpc
3855 :Parameters: struct kvm_ppc_resize_hpt (in)
3856 :Returns: 0 on successful completion,
3857 >0 if a new HPT is being prepared, the value is an estimated
3858 number of milliseconds until preparation is complete,
3859 -EFAULT if struct kvm_reinject_control cannot be read,
3860 -EINVAL if the supplied shift or flags are invalid,
3861 -ENOMEM if unable to allocate the new HPT,
3863 Used to implement the PAPR extension for runtime resizing of a guest's
3864 Hashed Page Table (HPT). Specifically this starts, stops or monitors
3865 the preparation of a new potential HPT for the guest, essentially
3866 implementing the H_RESIZE_HPT_PREPARE hypercall.
3870 struct kvm_ppc_resize_hpt {
3876 If called with shift > 0 when there is no pending HPT for the guest,
3877 this begins preparation of a new pending HPT of size 2^(shift) bytes.
3878 It then returns a positive integer with the estimated number of
3879 milliseconds until preparation is complete.
3881 If called when there is a pending HPT whose size does not match that
3882 requested in the parameters, discards the existing pending HPT and
3883 creates a new one as above.
3885 If called when there is a pending HPT of the size requested, will:
3887 * If preparation of the pending HPT is already complete, return 0
3888 * If preparation of the pending HPT has failed, return an error
3889 code, then discard the pending HPT.
3890 * If preparation of the pending HPT is still in progress, return an
3891 estimated number of milliseconds until preparation is complete.
3893 If called with shift == 0, discards any currently pending HPT and
3894 returns 0 (i.e. cancels any in-progress preparation).
3896 flags is reserved for future expansion, currently setting any bits in
3897 flags will result in an -EINVAL.
3899 Normally this will be called repeatedly with the same parameters until
3900 it returns <= 0. The first call will initiate preparation, subsequent
3901 ones will monitor preparation until it completes or fails.
3903 4.103 KVM_PPC_RESIZE_HPT_COMMIT
3904 -------------------------------
3906 :Capability: KVM_CAP_SPAPR_RESIZE_HPT
3907 :Architectures: powerpc
3909 :Parameters: struct kvm_ppc_resize_hpt (in)
3910 :Returns: 0 on successful completion,
3911 -EFAULT if struct kvm_reinject_control cannot be read,
3912 -EINVAL if the supplied shift or flags are invalid,
3913 -ENXIO is there is no pending HPT, or the pending HPT doesn't
3914 have the requested size,
3915 -EBUSY if the pending HPT is not fully prepared,
3916 -ENOSPC if there was a hash collision when moving existing
3917 HPT entries to the new HPT,
3918 -EIO on other error conditions
3920 Used to implement the PAPR extension for runtime resizing of a guest's
3921 Hashed Page Table (HPT). Specifically this requests that the guest be
3922 transferred to working with the new HPT, essentially implementing the
3923 H_RESIZE_HPT_COMMIT hypercall.
3927 struct kvm_ppc_resize_hpt {
3933 This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has
3934 returned 0 with the same parameters. In other cases
3935 KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or
3936 -EBUSY, though others may be possible if the preparation was started,
3939 This will have undefined effects on the guest if it has not already
3940 placed itself in a quiescent state where no vcpu will make MMU enabled
3943 On succsful completion, the pending HPT will become the guest's active
3944 HPT and the previous HPT will be discarded.
3946 On failure, the guest will still be operating on its previous HPT.
3948 4.104 KVM_X86_GET_MCE_CAP_SUPPORTED
3949 -----------------------------------
3951 :Capability: KVM_CAP_MCE
3954 :Parameters: u64 mce_cap (out)
3955 :Returns: 0 on success, -1 on error
3957 Returns supported MCE capabilities. The u64 mce_cap parameter
3958 has the same format as the MSR_IA32_MCG_CAP register. Supported
3959 capabilities will have the corresponding bits set.
3961 4.105 KVM_X86_SETUP_MCE
3962 -----------------------
3964 :Capability: KVM_CAP_MCE
3967 :Parameters: u64 mcg_cap (in)
3968 :Returns: 0 on success,
3969 -EFAULT if u64 mcg_cap cannot be read,
3970 -EINVAL if the requested number of banks is invalid,
3971 -EINVAL if requested MCE capability is not supported.
3973 Initializes MCE support for use. The u64 mcg_cap parameter
3974 has the same format as the MSR_IA32_MCG_CAP register and
3975 specifies which capabilities should be enabled. The maximum
3976 supported number of error-reporting banks can be retrieved when
3977 checking for KVM_CAP_MCE. The supported capabilities can be
3978 retrieved with KVM_X86_GET_MCE_CAP_SUPPORTED.
3980 4.106 KVM_X86_SET_MCE
3981 ---------------------
3983 :Capability: KVM_CAP_MCE
3986 :Parameters: struct kvm_x86_mce (in)
3987 :Returns: 0 on success,
3988 -EFAULT if struct kvm_x86_mce cannot be read,
3989 -EINVAL if the bank number is invalid,
3990 -EINVAL if VAL bit is not set in status field.
3992 Inject a machine check error (MCE) into the guest. The input
3995 struct kvm_x86_mce {
4005 If the MCE being reported is an uncorrected error, KVM will
4006 inject it as an MCE exception into the guest. If the guest
4007 MCG_STATUS register reports that an MCE is in progress, KVM
4008 causes an KVM_EXIT_SHUTDOWN vmexit.
4010 Otherwise, if the MCE is a corrected error, KVM will just
4011 store it in the corresponding bank (provided this bank is
4012 not holding a previously reported uncorrected error).
4014 4.107 KVM_S390_GET_CMMA_BITS
4015 ----------------------------
4017 :Capability: KVM_CAP_S390_CMMA_MIGRATION
4018 :Architectures: s390
4020 :Parameters: struct kvm_s390_cmma_log (in, out)
4021 :Returns: 0 on success, a negative value on error
4023 This ioctl is used to get the values of the CMMA bits on the s390
4024 architecture. It is meant to be used in two scenarios:
4026 - During live migration to save the CMMA values. Live migration needs
4027 to be enabled via the KVM_REQ_START_MIGRATION VM property.
4028 - To non-destructively peek at the CMMA values, with the flag
4029 KVM_S390_CMMA_PEEK set.
4031 The ioctl takes parameters via the kvm_s390_cmma_log struct. The desired
4032 values are written to a buffer whose location is indicated via the "values"
4033 member in the kvm_s390_cmma_log struct. The values in the input struct are
4034 also updated as needed.
4036 Each CMMA value takes up one byte.
4040 struct kvm_s390_cmma_log {
4051 start_gfn is the number of the first guest frame whose CMMA values are
4054 count is the length of the buffer in bytes,
4056 values points to the buffer where the result will be written to.
4058 If count is greater than KVM_S390_SKEYS_MAX, then it is considered to be
4059 KVM_S390_SKEYS_MAX. KVM_S390_SKEYS_MAX is re-used for consistency with
4062 The result is written in the buffer pointed to by the field values, and
4063 the values of the input parameter are updated as follows.
4065 Depending on the flags, different actions are performed. The only
4066 supported flag so far is KVM_S390_CMMA_PEEK.
4068 The default behaviour if KVM_S390_CMMA_PEEK is not set is:
4069 start_gfn will indicate the first page frame whose CMMA bits were dirty.
4070 It is not necessarily the same as the one passed as input, as clean pages
4073 count will indicate the number of bytes actually written in the buffer.
4074 It can (and very often will) be smaller than the input value, since the
4075 buffer is only filled until 16 bytes of clean values are found (which
4076 are then not copied in the buffer). Since a CMMA migration block needs
4077 the base address and the length, for a total of 16 bytes, we will send
4078 back some clean data if there is some dirty data afterwards, as long as
4079 the size of the clean data does not exceed the size of the header. This
4080 allows to minimize the amount of data to be saved or transferred over
4081 the network at the expense of more roundtrips to userspace. The next
4082 invocation of the ioctl will skip over all the clean values, saving
4083 potentially more than just the 16 bytes we found.
4085 If KVM_S390_CMMA_PEEK is set:
4086 the existing storage attributes are read even when not in migration
4087 mode, and no other action is performed;
4089 the output start_gfn will be equal to the input start_gfn,
4091 the output count will be equal to the input count, except if the end of
4092 memory has been reached.
4095 the field "remaining" will indicate the total number of dirty CMMA values
4096 still remaining, or 0 if KVM_S390_CMMA_PEEK is set and migration mode is
4101 values points to the userspace buffer where the result will be stored.
4103 This ioctl can fail with -ENOMEM if not enough memory can be allocated to
4104 complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
4105 KVM_S390_CMMA_PEEK is not set but migration mode was not enabled, with
4106 -EFAULT if the userspace address is invalid or if no page table is
4107 present for the addresses (e.g. when using hugepages).
4109 4.108 KVM_S390_SET_CMMA_BITS
4110 ----------------------------
4112 :Capability: KVM_CAP_S390_CMMA_MIGRATION
4113 :Architectures: s390
4115 :Parameters: struct kvm_s390_cmma_log (in)
4116 :Returns: 0 on success, a negative value on error
4118 This ioctl is used to set the values of the CMMA bits on the s390
4119 architecture. It is meant to be used during live migration to restore
4120 the CMMA values, but there are no restrictions on its use.
4121 The ioctl takes parameters via the kvm_s390_cmma_values struct.
4122 Each CMMA value takes up one byte.
4126 struct kvm_s390_cmma_log {
4137 start_gfn indicates the starting guest frame number,
4139 count indicates how many values are to be considered in the buffer,
4141 flags is not used and must be 0.
4143 mask indicates which PGSTE bits are to be considered.
4145 remaining is not used.
4147 values points to the buffer in userspace where to store the values.
4149 This ioctl can fail with -ENOMEM if not enough memory can be allocated to
4150 complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
4151 the count field is too large (e.g. more than KVM_S390_CMMA_SIZE_MAX) or
4152 if the flags field was not 0, with -EFAULT if the userspace address is
4153 invalid, if invalid pages are written to (e.g. after the end of memory)
4154 or if no page table is present for the addresses (e.g. when using
4157 4.109 KVM_PPC_GET_CPU_CHAR
4158 --------------------------
4160 :Capability: KVM_CAP_PPC_GET_CPU_CHAR
4161 :Architectures: powerpc
4163 :Parameters: struct kvm_ppc_cpu_char (out)
4164 :Returns: 0 on successful completion,
4165 -EFAULT if struct kvm_ppc_cpu_char cannot be written
4167 This ioctl gives userspace information about certain characteristics
4168 of the CPU relating to speculative execution of instructions and
4169 possible information leakage resulting from speculative execution (see
4170 CVE-2017-5715, CVE-2017-5753 and CVE-2017-5754). The information is
4171 returned in struct kvm_ppc_cpu_char, which looks like this::
4173 struct kvm_ppc_cpu_char {
4174 __u64 character; /* characteristics of the CPU */
4175 __u64 behaviour; /* recommended software behaviour */
4176 __u64 character_mask; /* valid bits in character */
4177 __u64 behaviour_mask; /* valid bits in behaviour */
4180 For extensibility, the character_mask and behaviour_mask fields
4181 indicate which bits of character and behaviour have been filled in by
4182 the kernel. If the set of defined bits is extended in future then
4183 userspace will be able to tell whether it is running on a kernel that
4184 knows about the new bits.
4186 The character field describes attributes of the CPU which can help
4187 with preventing inadvertent information disclosure - specifically,
4188 whether there is an instruction to flash-invalidate the L1 data cache
4189 (ori 30,30,0 or mtspr SPRN_TRIG2,rN), whether the L1 data cache is set
4190 to a mode where entries can only be used by the thread that created
4191 them, whether the bcctr[l] instruction prevents speculation, and
4192 whether a speculation barrier instruction (ori 31,31,0) is provided.
4194 The behaviour field describes actions that software should take to
4195 prevent inadvertent information disclosure, and thus describes which
4196 vulnerabilities the hardware is subject to; specifically whether the
4197 L1 data cache should be flushed when returning to user mode from the
4198 kernel, and whether a speculation barrier should be placed between an
4199 array bounds check and the array access.
4201 These fields use the same bit definitions as the new
4202 H_GET_CPU_CHARACTERISTICS hypercall.
4204 4.110 KVM_MEMORY_ENCRYPT_OP
4205 ---------------------------
4210 :Parameters: an opaque platform specific structure (in/out)
4211 :Returns: 0 on success; -1 on error
4213 If the platform supports creating encrypted VMs then this ioctl can be used
4214 for issuing platform-specific memory encryption commands to manage those
4217 Currently, this ioctl is used for issuing Secure Encrypted Virtualization
4218 (SEV) commands on AMD Processors. The SEV commands are defined in
4219 Documentation/virt/kvm/amd-memory-encryption.rst.
4221 4.111 KVM_MEMORY_ENCRYPT_REG_REGION
4222 -----------------------------------
4227 :Parameters: struct kvm_enc_region (in)
4228 :Returns: 0 on success; -1 on error
4230 This ioctl can be used to register a guest memory region which may
4231 contain encrypted data (e.g. guest RAM, SMRAM etc).
4233 It is used in the SEV-enabled guest. When encryption is enabled, a guest
4234 memory region may contain encrypted data. The SEV memory encryption
4235 engine uses a tweak such that two identical plaintext pages, each at
4236 different locations will have differing ciphertexts. So swapping or
4237 moving ciphertext of those pages will not result in plaintext being
4238 swapped. So relocating (or migrating) physical backing pages for the SEV
4239 guest will require some additional steps.
4241 Note: The current SEV key management spec does not provide commands to
4242 swap or migrate (move) ciphertext pages. Hence, for now we pin the guest
4243 memory region registered with the ioctl.
4245 4.112 KVM_MEMORY_ENCRYPT_UNREG_REGION
4246 -------------------------------------
4251 :Parameters: struct kvm_enc_region (in)
4252 :Returns: 0 on success; -1 on error
4254 This ioctl can be used to unregister the guest memory region registered
4255 with KVM_MEMORY_ENCRYPT_REG_REGION ioctl above.
4257 4.113 KVM_HYPERV_EVENTFD
4258 ------------------------
4260 :Capability: KVM_CAP_HYPERV_EVENTFD
4263 :Parameters: struct kvm_hyperv_eventfd (in)
4265 This ioctl (un)registers an eventfd to receive notifications from the guest on
4266 the specified Hyper-V connection id through the SIGNAL_EVENT hypercall, without
4267 causing a user exit. SIGNAL_EVENT hypercall with non-zero event flag number
4268 (bits 24-31) still triggers a KVM_EXIT_HYPERV_HCALL user exit.
4272 struct kvm_hyperv_eventfd {
4279 The conn_id field should fit within 24 bits::
4281 #define KVM_HYPERV_CONN_ID_MASK 0x00ffffff
4283 The acceptable values for the flags field are::
4285 #define KVM_HYPERV_EVENTFD_DEASSIGN (1 << 0)
4287 :Returns: 0 on success,
4288 -EINVAL if conn_id or flags is outside the allowed range,
4289 -ENOENT on deassign if the conn_id isn't registered,
4290 -EEXIST on assign if the conn_id is already registered
4292 4.114 KVM_GET_NESTED_STATE
4293 --------------------------
4295 :Capability: KVM_CAP_NESTED_STATE
4298 :Parameters: struct kvm_nested_state (in/out)
4299 :Returns: 0 on success, -1 on error
4303 ===== =============================================================
4304 E2BIG the total state size exceeds the value of 'size' specified by
4305 the user; the size required will be written into size.
4306 ===== =============================================================
4310 struct kvm_nested_state {
4316 struct kvm_vmx_nested_state_hdr vmx;
4317 struct kvm_svm_nested_state_hdr svm;
4319 /* Pad the header to 128 bytes. */
4324 struct kvm_vmx_nested_state_data vmx[0];
4325 struct kvm_svm_nested_state_data svm[0];
4329 #define KVM_STATE_NESTED_GUEST_MODE 0x00000001
4330 #define KVM_STATE_NESTED_RUN_PENDING 0x00000002
4331 #define KVM_STATE_NESTED_EVMCS 0x00000004
4333 #define KVM_STATE_NESTED_FORMAT_VMX 0
4334 #define KVM_STATE_NESTED_FORMAT_SVM 1
4336 #define KVM_STATE_NESTED_VMX_VMCS_SIZE 0x1000
4338 #define KVM_STATE_NESTED_VMX_SMM_GUEST_MODE 0x00000001
4339 #define KVM_STATE_NESTED_VMX_SMM_VMXON 0x00000002
4341 #define KVM_STATE_VMX_PREEMPTION_TIMER_DEADLINE 0x00000001
4343 struct kvm_vmx_nested_state_hdr {
4352 __u64 preemption_timer_deadline;
4355 struct kvm_vmx_nested_state_data {
4356 __u8 vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE];
4357 __u8 shadow_vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE];
4360 This ioctl copies the vcpu's nested virtualization state from the kernel to
4363 The maximum size of the state can be retrieved by passing KVM_CAP_NESTED_STATE
4364 to the KVM_CHECK_EXTENSION ioctl().
4366 4.115 KVM_SET_NESTED_STATE
4367 --------------------------
4369 :Capability: KVM_CAP_NESTED_STATE
4372 :Parameters: struct kvm_nested_state (in)
4373 :Returns: 0 on success, -1 on error
4375 This copies the vcpu's kvm_nested_state struct from userspace to the kernel.
4376 For the definition of struct kvm_nested_state, see KVM_GET_NESTED_STATE.
4378 4.116 KVM_(UN)REGISTER_COALESCED_MMIO
4379 -------------------------------------
4381 :Capability: KVM_CAP_COALESCED_MMIO (for coalesced mmio)
4382 KVM_CAP_COALESCED_PIO (for coalesced pio)
4385 :Parameters: struct kvm_coalesced_mmio_zone
4386 :Returns: 0 on success, < 0 on error
4388 Coalesced I/O is a performance optimization that defers hardware
4389 register write emulation so that userspace exits are avoided. It is
4390 typically used to reduce the overhead of emulating frequently accessed
4393 When a hardware register is configured for coalesced I/O, write accesses
4394 do not exit to userspace and their value is recorded in a ring buffer
4395 that is shared between kernel and userspace.
4397 Coalesced I/O is used if one or more write accesses to a hardware
4398 register can be deferred until a read or a write to another hardware
4399 register on the same device. This last access will cause a vmexit and
4400 userspace will process accesses from the ring buffer before emulating
4401 it. That will avoid exiting to userspace on repeated writes.
4403 Coalesced pio is based on coalesced mmio. There is little difference
4404 between coalesced mmio and pio except that coalesced pio records accesses
4407 4.117 KVM_CLEAR_DIRTY_LOG (vm ioctl)
4408 ------------------------------------
4410 :Capability: KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
4411 :Architectures: x86, arm, arm64, mips
4413 :Parameters: struct kvm_clear_dirty_log (in)
4414 :Returns: 0 on success, -1 on error
4418 /* for KVM_CLEAR_DIRTY_LOG */
4419 struct kvm_clear_dirty_log {
4424 void __user *dirty_bitmap; /* one bit per page */
4429 The ioctl clears the dirty status of pages in a memory slot, according to
4430 the bitmap that is passed in struct kvm_clear_dirty_log's dirty_bitmap
4431 field. Bit 0 of the bitmap corresponds to page "first_page" in the
4432 memory slot, and num_pages is the size in bits of the input bitmap.
4433 first_page must be a multiple of 64; num_pages must also be a multiple of
4434 64 unless first_page + num_pages is the size of the memory slot. For each
4435 bit that is set in the input bitmap, the corresponding page is marked "clean"
4436 in KVM's dirty bitmap, and dirty tracking is re-enabled for that page
4437 (for example via write-protection, or by clearing the dirty bit in
4438 a page table entry).
4440 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies
4441 the address space for which you want to clear the dirty status. See
4442 KVM_SET_USER_MEMORY_REGION for details on the usage of slot field.
4444 This ioctl is mostly useful when KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
4445 is enabled; for more information, see the description of the capability.
4446 However, it can always be used as long as KVM_CHECK_EXTENSION confirms
4447 that KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is present.
4449 4.118 KVM_GET_SUPPORTED_HV_CPUID
4450 --------------------------------
4452 :Capability: KVM_CAP_HYPERV_CPUID (vcpu), KVM_CAP_SYS_HYPERV_CPUID (system)
4454 :Type: system ioctl, vcpu ioctl
4455 :Parameters: struct kvm_cpuid2 (in/out)
4456 :Returns: 0 on success, -1 on error
4463 struct kvm_cpuid_entry2 entries[0];
4466 struct kvm_cpuid_entry2 {
4477 This ioctl returns x86 cpuid features leaves related to Hyper-V emulation in
4478 KVM. Userspace can use the information returned by this ioctl to construct
4479 cpuid information presented to guests consuming Hyper-V enlightenments (e.g.
4480 Windows or Hyper-V guests).
4482 CPUID feature leaves returned by this ioctl are defined by Hyper-V Top Level
4483 Functional Specification (TLFS). These leaves can't be obtained with
4484 KVM_GET_SUPPORTED_CPUID ioctl because some of them intersect with KVM feature
4485 leaves (0x40000000, 0x40000001).
4487 Currently, the following list of CPUID leaves are returned:
4489 - HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS
4490 - HYPERV_CPUID_INTERFACE
4491 - HYPERV_CPUID_VERSION
4492 - HYPERV_CPUID_FEATURES
4493 - HYPERV_CPUID_ENLIGHTMENT_INFO
4494 - HYPERV_CPUID_IMPLEMENT_LIMITS
4495 - HYPERV_CPUID_NESTED_FEATURES
4496 - HYPERV_CPUID_SYNDBG_VENDOR_AND_MAX_FUNCTIONS
4497 - HYPERV_CPUID_SYNDBG_INTERFACE
4498 - HYPERV_CPUID_SYNDBG_PLATFORM_CAPABILITIES
4500 Userspace invokes KVM_GET_SUPPORTED_HV_CPUID by passing a kvm_cpuid2 structure
4501 with the 'nent' field indicating the number of entries in the variable-size
4502 array 'entries'. If the number of entries is too low to describe all Hyper-V
4503 feature leaves, an error (E2BIG) is returned. If the number is more or equal
4504 to the number of Hyper-V feature leaves, the 'nent' field is adjusted to the
4505 number of valid entries in the 'entries' array, which is then filled.
4507 'index' and 'flags' fields in 'struct kvm_cpuid_entry2' are currently reserved,
4508 userspace should not expect to get any particular value there.
4510 Note, vcpu version of KVM_GET_SUPPORTED_HV_CPUID is currently deprecated. Unlike
4511 system ioctl which exposes all supported feature bits unconditionally, vcpu
4512 version has the following quirks:
4514 - HYPERV_CPUID_NESTED_FEATURES leaf and HV_X64_ENLIGHTENED_VMCS_RECOMMENDED
4515 feature bit are only exposed when Enlightened VMCS was previously enabled
4516 on the corresponding vCPU (KVM_CAP_HYPERV_ENLIGHTENED_VMCS).
4517 - HV_STIMER_DIRECT_MODE_AVAILABLE bit is only exposed with in-kernel LAPIC.
4518 (presumes KVM_CREATE_IRQCHIP has already been called).
4520 4.119 KVM_ARM_VCPU_FINALIZE
4521 ---------------------------
4523 :Architectures: arm, arm64
4525 :Parameters: int feature (in)
4526 :Returns: 0 on success, -1 on error
4530 ====== ==============================================================
4531 EPERM feature not enabled, needs configuration, or already finalized
4532 EINVAL feature unknown or not present
4533 ====== ==============================================================
4535 Recognised values for feature:
4537 ===== ===========================================
4538 arm64 KVM_ARM_VCPU_SVE (requires KVM_CAP_ARM_SVE)
4539 ===== ===========================================
4541 Finalizes the configuration of the specified vcpu feature.
4543 The vcpu must already have been initialised, enabling the affected feature, by
4544 means of a successful KVM_ARM_VCPU_INIT call with the appropriate flag set in
4547 For affected vcpu features, this is a mandatory step that must be performed
4548 before the vcpu is fully usable.
4550 Between KVM_ARM_VCPU_INIT and KVM_ARM_VCPU_FINALIZE, the feature may be
4551 configured by use of ioctls such as KVM_SET_ONE_REG. The exact configuration
4552 that should be performaned and how to do it are feature-dependent.
4554 Other calls that depend on a particular feature being finalized, such as
4555 KVM_RUN, KVM_GET_REG_LIST, KVM_GET_ONE_REG and KVM_SET_ONE_REG, will fail with
4556 -EPERM unless the feature has already been finalized by means of a
4557 KVM_ARM_VCPU_FINALIZE call.
4559 See KVM_ARM_VCPU_INIT for details of vcpu features that require finalization
4562 4.120 KVM_SET_PMU_EVENT_FILTER
4563 ------------------------------
4565 :Capability: KVM_CAP_PMU_EVENT_FILTER
4568 :Parameters: struct kvm_pmu_event_filter (in)
4569 :Returns: 0 on success, -1 on error
4573 struct kvm_pmu_event_filter {
4576 __u32 fixed_counter_bitmap;
4582 This ioctl restricts the set of PMU events that the guest can program.
4583 The argument holds a list of events which will be allowed or denied.
4584 The eventsel+umask of each event the guest attempts to program is compared
4585 against the events field to determine whether the guest should have access.
4586 The events field only controls general purpose counters; fixed purpose
4587 counters are controlled by the fixed_counter_bitmap.
4589 No flags are defined yet, the field must be zero.
4591 Valid values for 'action'::
4593 #define KVM_PMU_EVENT_ALLOW 0
4594 #define KVM_PMU_EVENT_DENY 1
4596 4.121 KVM_PPC_SVM_OFF
4597 ---------------------
4600 :Architectures: powerpc
4603 :Returns: 0 on successful completion,
4607 ====== ================================================================
4608 EINVAL if ultravisor failed to terminate the secure guest
4609 ENOMEM if hypervisor failed to allocate new radix page tables for guest
4610 ====== ================================================================
4612 This ioctl is used to turn off the secure mode of the guest or transition
4613 the guest from secure mode to normal mode. This is invoked when the guest
4614 is reset. This has no effect if called for a normal guest.
4616 This ioctl issues an ultravisor call to terminate the secure guest,
4617 unpins the VPA pages and releases all the device pages that are used to
4618 track the secure pages by hypervisor.
4620 4.122 KVM_S390_NORMAL_RESET
4621 ---------------------------
4623 :Capability: KVM_CAP_S390_VCPU_RESETS
4624 :Architectures: s390
4629 This ioctl resets VCPU registers and control structures according to
4630 the cpu reset definition in the POP (Principles Of Operation).
4632 4.123 KVM_S390_INITIAL_RESET
4633 ----------------------------
4636 :Architectures: s390
4641 This ioctl resets VCPU registers and control structures according to
4642 the initial cpu reset definition in the POP. However, the cpu is not
4643 put into ESA mode. This reset is a superset of the normal reset.
4645 4.124 KVM_S390_CLEAR_RESET
4646 --------------------------
4648 :Capability: KVM_CAP_S390_VCPU_RESETS
4649 :Architectures: s390
4654 This ioctl resets VCPU registers and control structures according to
4655 the clear cpu reset definition in the POP. However, the cpu is not put
4656 into ESA mode. This reset is a superset of the initial reset.
4659 4.125 KVM_S390_PV_COMMAND
4660 -------------------------
4662 :Capability: KVM_CAP_S390_PROTECTED
4663 :Architectures: s390
4665 :Parameters: struct kvm_pv_cmd
4666 :Returns: 0 on success, < 0 on error
4671 __u32 cmd; /* Command to be executed */
4672 __u16 rc; /* Ultravisor return code */
4673 __u16 rrc; /* Ultravisor return reason code */
4674 __u64 data; /* Data or address */
4675 __u32 flags; /* flags for future extensions. Must be 0 for now */
4682 Allocate memory and register the VM with the Ultravisor, thereby
4683 donating memory to the Ultravisor that will become inaccessible to
4684 KVM. All existing CPUs are converted to protected ones. After this
4685 command has succeeded, any CPU added via hotplug will become
4686 protected during its creation as well.
4690 ===== =============================
4691 EINTR an unmasked signal is pending
4692 ===== =============================
4696 Deregister the VM from the Ultravisor and reclaim the memory that
4697 had been donated to the Ultravisor, making it usable by the kernel
4698 again. All registered VCPUs are converted back to non-protected
4701 KVM_PV_VM_SET_SEC_PARMS
4702 Pass the image header from VM memory to the Ultravisor in
4703 preparation of image unpacking and verification.
4706 Unpack (protect and decrypt) a page of the encrypted boot image.
4709 Verify the integrity of the unpacked image. Only if this succeeds,
4710 KVM is allowed to start protected VCPUs.
4712 4.126 KVM_X86_SET_MSR_FILTER
4713 ----------------------------
4715 :Capability: KVM_X86_SET_MSR_FILTER
4718 :Parameters: struct kvm_msr_filter
4719 :Returns: 0 on success, < 0 on error
4723 struct kvm_msr_filter_range {
4724 #define KVM_MSR_FILTER_READ (1 << 0)
4725 #define KVM_MSR_FILTER_WRITE (1 << 1)
4727 __u32 nmsrs; /* number of msrs in bitmap */
4728 __u32 base; /* MSR index the bitmap starts at */
4729 __u8 *bitmap; /* a 1 bit allows the operations in flags, 0 denies */
4732 #define KVM_MSR_FILTER_MAX_RANGES 16
4733 struct kvm_msr_filter {
4734 #define KVM_MSR_FILTER_DEFAULT_ALLOW (0 << 0)
4735 #define KVM_MSR_FILTER_DEFAULT_DENY (1 << 0)
4737 struct kvm_msr_filter_range ranges[KVM_MSR_FILTER_MAX_RANGES];
4740 flags values for ``struct kvm_msr_filter_range``:
4742 ``KVM_MSR_FILTER_READ``
4744 Filter read accesses to MSRs using the given bitmap. A 0 in the bitmap
4745 indicates that a read should immediately fail, while a 1 indicates that
4746 a read for a particular MSR should be handled regardless of the default
4749 ``KVM_MSR_FILTER_WRITE``
4751 Filter write accesses to MSRs using the given bitmap. A 0 in the bitmap
4752 indicates that a write should immediately fail, while a 1 indicates that
4753 a write for a particular MSR should be handled regardless of the default
4756 ``KVM_MSR_FILTER_READ | KVM_MSR_FILTER_WRITE``
4758 Filter both read and write accesses to MSRs using the given bitmap. A 0
4759 in the bitmap indicates that both reads and writes should immediately fail,
4760 while a 1 indicates that reads and writes for a particular MSR are not
4761 filtered by this range.
4763 flags values for ``struct kvm_msr_filter``:
4765 ``KVM_MSR_FILTER_DEFAULT_ALLOW``
4767 If no filter range matches an MSR index that is getting accessed, KVM will
4768 fall back to allowing access to the MSR.
4770 ``KVM_MSR_FILTER_DEFAULT_DENY``
4772 If no filter range matches an MSR index that is getting accessed, KVM will
4773 fall back to rejecting access to the MSR. In this mode, all MSRs that should
4774 be processed by KVM need to explicitly be marked as allowed in the bitmaps.
4776 This ioctl allows user space to define up to 16 bitmaps of MSR ranges to
4777 specify whether a certain MSR access should be explicitly filtered for or not.
4779 If this ioctl has never been invoked, MSR accesses are not guarded and the
4780 default KVM in-kernel emulation behavior is fully preserved.
4782 Calling this ioctl with an empty set of ranges (all nmsrs == 0) disables MSR
4783 filtering. In that mode, ``KVM_MSR_FILTER_DEFAULT_DENY`` is invalid and causes
4786 As soon as the filtering is in place, every MSR access is processed through
4787 the filtering except for accesses to the x2APIC MSRs (from 0x800 to 0x8ff);
4788 x2APIC MSRs are always allowed, independent of the ``default_allow`` setting,
4789 and their behavior depends on the ``X2APIC_ENABLE`` bit of the APIC base
4792 If a bit is within one of the defined ranges, read and write accesses are
4793 guarded by the bitmap's value for the MSR index if the kind of access
4794 is included in the ``struct kvm_msr_filter_range`` flags. If no range
4795 cover this particular access, the behavior is determined by the flags
4796 field in the kvm_msr_filter struct: ``KVM_MSR_FILTER_DEFAULT_ALLOW``
4797 and ``KVM_MSR_FILTER_DEFAULT_DENY``.
4799 Each bitmap range specifies a range of MSRs to potentially allow access on.
4800 The range goes from MSR index [base .. base+nmsrs]. The flags field
4801 indicates whether reads, writes or both reads and writes are filtered
4802 by setting a 1 bit in the bitmap for the corresponding MSR index.
4804 If an MSR access is not permitted through the filtering, it generates a
4805 #GP inside the guest. When combined with KVM_CAP_X86_USER_SPACE_MSR, that
4806 allows user space to deflect and potentially handle various MSR accesses
4809 If a vCPU is in running state while this ioctl is invoked, the vCPU may
4810 experience inconsistent filtering behavior on MSR accesses.
4812 4.127 KVM_XEN_HVM_SET_ATTR
4813 --------------------------
4815 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
4818 :Parameters: struct kvm_xen_hvm_attr
4819 :Returns: 0 on success, < 0 on error
4823 struct kvm_xen_hvm_attr {
4838 KVM_XEN_ATTR_TYPE_LONG_MODE
4839 Sets the ABI mode of the VM to 32-bit or 64-bit (long mode). This
4840 determines the layout of the shared info pages exposed to the VM.
4842 KVM_XEN_ATTR_TYPE_SHARED_INFO
4843 Sets the guest physical frame number at which the Xen "shared info"
4844 page resides. Note that although Xen places vcpu_info for the first
4845 32 vCPUs in the shared_info page, KVM does not automatically do so
4846 and instead requires that KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO be used
4847 explicitly even when the vcpu_info for a given vCPU resides at the
4848 "default" location in the shared_info page. This is because KVM is
4849 not aware of the Xen CPU id which is used as the index into the
4850 vcpu_info[] array, so cannot know the correct default location.
4852 KVM_XEN_ATTR_TYPE_UPCALL_VECTOR
4853 Sets the exception vector used to deliver Xen event channel upcalls.
4855 4.128 KVM_XEN_HVM_GET_ATTR
4856 --------------------------
4858 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
4861 :Parameters: struct kvm_xen_hvm_attr
4862 :Returns: 0 on success, < 0 on error
4864 Allows Xen VM attributes to be read. For the structure and types,
4865 see KVM_XEN_HVM_SET_ATTR above.
4867 4.129 KVM_XEN_VCPU_SET_ATTR
4868 ---------------------------
4870 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
4873 :Parameters: struct kvm_xen_vcpu_attr
4874 :Returns: 0 on success, < 0 on error
4878 struct kvm_xen_vcpu_attr {
4886 __u64 state_entry_time;
4888 __u64 time_runnable;
4897 KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO
4898 Sets the guest physical address of the vcpu_info for a given vCPU.
4900 KVM_XEN_VCPU_ATTR_TYPE_VCPU_TIME_INFO
4901 Sets the guest physical address of an additional pvclock structure
4902 for a given vCPU. This is typically used for guest vsyscall support.
4904 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR
4905 Sets the guest physical address of the vcpu_runstate_info for a given
4906 vCPU. This is how a Xen guest tracks CPU state such as steal time.
4908 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_CURRENT
4909 Sets the runstate (RUNSTATE_running/_runnable/_blocked/_offline) of
4910 the given vCPU from the .u.runstate.state member of the structure.
4911 KVM automatically accounts running and runnable time but blocked
4912 and offline states are only entered explicitly.
4914 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_DATA
4915 Sets all fields of the vCPU runstate data from the .u.runstate member
4916 of the structure, including the current runstate. The state_entry_time
4917 must equal the sum of the other four times.
4919 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST
4920 This *adds* the contents of the .u.runstate members of the structure
4921 to the corresponding members of the given vCPU's runstate data, thus
4922 permitting atomic adjustments to the runstate times. The adjustment
4923 to the state_entry_time must equal the sum of the adjustments to the
4924 other four times. The state field must be set to -1, or to a valid
4925 runstate value (RUNSTATE_running, RUNSTATE_runnable, RUNSTATE_blocked
4926 or RUNSTATE_offline) to set the current accounted state as of the
4927 adjusted state_entry_time.
4929 4.130 KVM_XEN_VCPU_GET_ATTR
4930 ---------------------------
4932 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
4935 :Parameters: struct kvm_xen_vcpu_attr
4936 :Returns: 0 on success, < 0 on error
4938 Allows Xen vCPU attributes to be read. For the structure and types,
4939 see KVM_XEN_VCPU_SET_ATTR above.
4941 The KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST type may not be used
4942 with the KVM_XEN_VCPU_GET_ATTR ioctl.
4944 5. The kvm_run structure
4945 ========================
4947 Application code obtains a pointer to the kvm_run structure by
4948 mmap()ing a vcpu fd. From that point, application code can control
4949 execution by changing fields in kvm_run prior to calling the KVM_RUN
4950 ioctl, and obtain information about the reason KVM_RUN returned by
4951 looking up structure members.
4957 __u8 request_interrupt_window;
4959 Request that KVM_RUN return when it becomes possible to inject external
4960 interrupts into the guest. Useful in conjunction with KVM_INTERRUPT.
4964 __u8 immediate_exit;
4966 This field is polled once when KVM_RUN starts; if non-zero, KVM_RUN
4967 exits immediately, returning -EINTR. In the common scenario where a
4968 signal is used to "kick" a VCPU out of KVM_RUN, this field can be used
4969 to avoid usage of KVM_SET_SIGNAL_MASK, which has worse scalability.
4970 Rather than blocking the signal outside KVM_RUN, userspace can set up
4971 a signal handler that sets run->immediate_exit to a non-zero value.
4973 This field is ignored if KVM_CAP_IMMEDIATE_EXIT is not available.
4982 When KVM_RUN has returned successfully (return value 0), this informs
4983 application code why KVM_RUN has returned. Allowable values for this
4984 field are detailed below.
4988 __u8 ready_for_interrupt_injection;
4990 If request_interrupt_window has been specified, this field indicates
4991 an interrupt can be injected now with KVM_INTERRUPT.
4997 The value of the current interrupt flag. Only valid if in-kernel
4998 local APIC is not used.
5004 More architecture-specific flags detailing state of the VCPU that may
5005 affect the device's behavior. Current defined flags::
5007 /* x86, set if the VCPU is in system management mode */
5008 #define KVM_RUN_X86_SMM (1 << 0)
5009 /* x86, set if bus lock detected in VM */
5010 #define KVM_RUN_BUS_LOCK (1 << 1)
5014 /* in (pre_kvm_run), out (post_kvm_run) */
5017 The value of the cr8 register. Only valid if in-kernel local APIC is
5018 not used. Both input and output.
5024 The value of the APIC BASE msr. Only valid if in-kernel local
5025 APIC is not used. Both input and output.
5030 /* KVM_EXIT_UNKNOWN */
5032 __u64 hardware_exit_reason;
5035 If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown
5036 reasons. Further architecture-specific information is available in
5037 hardware_exit_reason.
5041 /* KVM_EXIT_FAIL_ENTRY */
5043 __u64 hardware_entry_failure_reason;
5044 __u32 cpu; /* if KVM_LAST_CPU */
5047 If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due
5048 to unknown reasons. Further architecture-specific information is
5049 available in hardware_entry_failure_reason.
5053 /* KVM_EXIT_EXCEPTION */
5065 #define KVM_EXIT_IO_IN 0
5066 #define KVM_EXIT_IO_OUT 1
5068 __u8 size; /* bytes */
5071 __u64 data_offset; /* relative to kvm_run start */
5074 If exit_reason is KVM_EXIT_IO, then the vcpu has
5075 executed a port I/O instruction which could not be satisfied by kvm.
5076 data_offset describes where the data is located (KVM_EXIT_IO_OUT) or
5077 where kvm expects application code to place the data for the next
5078 KVM_RUN invocation (KVM_EXIT_IO_IN). Data format is a packed array.
5082 /* KVM_EXIT_DEBUG */
5084 struct kvm_debug_exit_arch arch;
5087 If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event
5088 for which architecture specific information is returned.
5100 If exit_reason is KVM_EXIT_MMIO, then the vcpu has
5101 executed a memory-mapped I/O instruction which could not be satisfied
5102 by kvm. The 'data' member contains the written data if 'is_write' is
5103 true, and should be filled by application code otherwise.
5105 The 'data' member contains, in its first 'len' bytes, the value as it would
5106 appear if the VCPU performed a load or store of the appropriate width directly
5111 For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR, KVM_EXIT_XEN,
5112 KVM_EXIT_EPR, KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR the corresponding
5113 operations are complete (and guest state is consistent) only after userspace
5114 has re-entered the kernel with KVM_RUN. The kernel side will first finish
5115 incomplete operations and then check for pending signals.
5117 The pending state of the operation is not preserved in state which is
5118 visible to userspace, thus userspace should ensure that the operation is
5119 completed before performing a live migration. Userspace can re-enter the
5120 guest with an unmasked signal pending or with the immediate_exit field set
5121 to complete pending operations without allowing any further instructions
5126 /* KVM_EXIT_HYPERCALL */
5135 Unused. This was once used for 'hypercall to userspace'. To implement
5136 such functionality, use KVM_EXIT_IO (x86) or KVM_EXIT_MMIO (all except s390).
5138 .. note:: KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO.
5142 /* KVM_EXIT_TPR_ACCESS */
5149 To be documented (KVM_TPR_ACCESS_REPORTING).
5153 /* KVM_EXIT_S390_SIEIC */
5156 __u64 mask; /* psw upper half */
5157 __u64 addr; /* psw lower half */
5166 /* KVM_EXIT_S390_RESET */
5167 #define KVM_S390_RESET_POR 1
5168 #define KVM_S390_RESET_CLEAR 2
5169 #define KVM_S390_RESET_SUBSYSTEM 4
5170 #define KVM_S390_RESET_CPU_INIT 8
5171 #define KVM_S390_RESET_IPL 16
5172 __u64 s390_reset_flags;
5178 /* KVM_EXIT_S390_UCONTROL */
5180 __u64 trans_exc_code;
5184 s390 specific. A page fault has occurred for a user controlled virtual
5185 machine (KVM_VM_S390_UNCONTROL) on it's host page table that cannot be
5186 resolved by the kernel.
5187 The program code and the translation exception code that were placed
5188 in the cpu's lowcore are presented here as defined by the z Architecture
5189 Principles of Operation Book in the Chapter for Dynamic Address Translation
5201 Deprecated - was used for 440 KVM.
5210 MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch
5211 hypercalls and exit with this exit struct that contains all the guest gprs.
5213 If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall.
5214 Userspace can now handle the hypercall and when it's done modify the gprs as
5215 necessary. Upon guest entry all guest GPRs will then be replaced by the values
5220 /* KVM_EXIT_PAPR_HCALL */
5227 This is used on 64-bit PowerPC when emulating a pSeries partition,
5228 e.g. with the 'pseries' machine type in qemu. It occurs when the
5229 guest does a hypercall using the 'sc 1' instruction. The 'nr' field
5230 contains the hypercall number (from the guest R3), and 'args' contains
5231 the arguments (from the guest R4 - R12). Userspace should put the
5232 return code in 'ret' and any extra returned values in args[].
5233 The possible hypercalls are defined in the Power Architecture Platform
5234 Requirements (PAPR) document available from www.power.org (free
5235 developer registration required to access it).
5239 /* KVM_EXIT_S390_TSCH */
5241 __u16 subchannel_id;
5242 __u16 subchannel_nr;
5249 s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled
5250 and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O
5251 interrupt for the target subchannel has been dequeued and subchannel_id,
5252 subchannel_nr, io_int_parm and io_int_word contain the parameters for that
5253 interrupt. ipb is needed for instruction parameter decoding.
5262 On FSL BookE PowerPC chips, the interrupt controller has a fast patch
5263 interrupt acknowledge path to the core. When the core successfully
5264 delivers an interrupt, it automatically populates the EPR register with
5265 the interrupt vector number and acknowledges the interrupt inside
5266 the interrupt controller.
5268 In case the interrupt controller lives in user space, we need to do
5269 the interrupt acknowledge cycle through it to fetch the next to be
5270 delivered interrupt vector using this exit.
5272 It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an
5273 external interrupt has just been delivered into the guest. User space
5274 should put the acknowledged interrupt vector into the 'epr' field.
5278 /* KVM_EXIT_SYSTEM_EVENT */
5280 #define KVM_SYSTEM_EVENT_SHUTDOWN 1
5281 #define KVM_SYSTEM_EVENT_RESET 2
5282 #define KVM_SYSTEM_EVENT_CRASH 3
5287 If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered
5288 a system-level event using some architecture specific mechanism (hypercall
5289 or some special instruction). In case of ARM/ARM64, this is triggered using
5290 HVC instruction based PSCI call from the vcpu. The 'type' field describes
5291 the system-level event type. The 'flags' field describes architecture
5292 specific flags for the system-level event.
5294 Valid values for 'type' are:
5296 - KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the
5297 VM. Userspace is not obliged to honour this, and if it does honour
5298 this does not need to destroy the VM synchronously (ie it may call
5299 KVM_RUN again before shutdown finally occurs).
5300 - KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM.
5301 As with SHUTDOWN, userspace can choose to ignore the request, or
5302 to schedule the reset to occur in the future and may call KVM_RUN again.
5303 - KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest
5304 has requested a crash condition maintenance. Userspace can choose
5305 to ignore the request, or to gather VM memory core dump and/or
5306 reset/shutdown of the VM.
5310 /* KVM_EXIT_IOAPIC_EOI */
5315 Indicates that the VCPU's in-kernel local APIC received an EOI for a
5316 level-triggered IOAPIC interrupt. This exit only triggers when the
5317 IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled);
5318 the userspace IOAPIC should process the EOI and retrigger the interrupt if
5319 it is still asserted. Vector is the LAPIC interrupt vector for which the
5324 struct kvm_hyperv_exit {
5325 #define KVM_EXIT_HYPERV_SYNIC 1
5326 #define KVM_EXIT_HYPERV_HCALL 2
5327 #define KVM_EXIT_HYPERV_SYNDBG 3
5354 /* KVM_EXIT_HYPERV */
5355 struct kvm_hyperv_exit hyperv;
5357 Indicates that the VCPU exits into userspace to process some tasks
5358 related to Hyper-V emulation.
5360 Valid values for 'type' are:
5362 - KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about
5364 Hyper-V SynIC state change. Notification is used to remap SynIC
5365 event/message pages and to enable/disable SynIC messages/events processing
5368 - KVM_EXIT_HYPERV_SYNDBG -- synchronously notify user-space about
5370 Hyper-V Synthetic debugger state change. Notification is used to either update
5371 the pending_page location or to send a control command (send the buffer located
5372 in send_page or recv a buffer to recv_page).
5376 /* KVM_EXIT_ARM_NISV */
5382 Used on arm and arm64 systems. If a guest accesses memory not in a memslot,
5383 KVM will typically return to userspace and ask it to do MMIO emulation on its
5384 behalf. However, for certain classes of instructions, no instruction decode
5385 (direction, length of memory access) is provided, and fetching and decoding
5386 the instruction from the VM is overly complicated to live in the kernel.
5388 Historically, when this situation occurred, KVM would print a warning and kill
5389 the VM. KVM assumed that if the guest accessed non-memslot memory, it was
5390 trying to do I/O, which just couldn't be emulated, and the warning message was
5391 phrased accordingly. However, what happened more often was that a guest bug
5392 caused access outside the guest memory areas which should lead to a more
5393 meaningful warning message and an external abort in the guest, if the access
5394 did not fall within an I/O window.
5396 Userspace implementations can query for KVM_CAP_ARM_NISV_TO_USER, and enable
5397 this capability at VM creation. Once this is done, these types of errors will
5398 instead return to userspace with KVM_EXIT_ARM_NISV, with the valid bits from
5399 the HSR (arm) and ESR_EL2 (arm64) in the esr_iss field, and the faulting IPA
5400 in the fault_ipa field. Userspace can either fix up the access if it's
5401 actually an I/O access by decoding the instruction from guest memory (if it's
5402 very brave) and continue executing the guest, or it can decide to suspend,
5403 dump, or restart the guest.
5405 Note that KVM does not skip the faulting instruction as it does for
5406 KVM_EXIT_MMIO, but userspace has to emulate any change to the processing state
5407 if it decides to decode and emulate the instruction.
5411 /* KVM_EXIT_X86_RDMSR / KVM_EXIT_X86_WRMSR */
5413 __u8 error; /* user -> kernel */
5415 __u32 reason; /* kernel -> user */
5416 __u32 index; /* kernel -> user */
5417 __u64 data; /* kernel <-> user */
5420 Used on x86 systems. When the VM capability KVM_CAP_X86_USER_SPACE_MSR is
5421 enabled, MSR accesses to registers that would invoke a #GP by KVM kernel code
5422 will instead trigger a KVM_EXIT_X86_RDMSR exit for reads and KVM_EXIT_X86_WRMSR
5425 The "reason" field specifies why the MSR trap occurred. User space will only
5426 receive MSR exit traps when a particular reason was requested during through
5427 ENABLE_CAP. Currently valid exit reasons are:
5429 KVM_MSR_EXIT_REASON_UNKNOWN - access to MSR that is unknown to KVM
5430 KVM_MSR_EXIT_REASON_INVAL - access to invalid MSRs or reserved bits
5431 KVM_MSR_EXIT_REASON_FILTER - access blocked by KVM_X86_SET_MSR_FILTER
5433 For KVM_EXIT_X86_RDMSR, the "index" field tells user space which MSR the guest
5434 wants to read. To respond to this request with a successful read, user space
5435 writes the respective data into the "data" field and must continue guest
5436 execution to ensure the read data is transferred into guest register state.
5438 If the RDMSR request was unsuccessful, user space indicates that with a "1" in
5439 the "error" field. This will inject a #GP into the guest when the VCPU is
5442 For KVM_EXIT_X86_WRMSR, the "index" field tells user space which MSR the guest
5443 wants to write. Once finished processing the event, user space must continue
5444 vCPU execution. If the MSR write was unsuccessful, user space also sets the
5445 "error" field to "1".
5450 struct kvm_xen_exit {
5451 #define KVM_EXIT_XEN_HCALL 1
5464 struct kvm_hyperv_exit xen;
5466 Indicates that the VCPU exits into userspace to process some tasks
5467 related to Xen emulation.
5469 Valid values for 'type' are:
5471 - KVM_EXIT_XEN_HCALL -- synchronously notify user-space about Xen hypercall.
5472 Userspace is expected to place the hypercall result into the appropriate
5473 field before invoking KVM_RUN again.
5477 /* Fix the size of the union. */
5482 * shared registers between kvm and userspace.
5483 * kvm_valid_regs specifies the register classes set by the host
5484 * kvm_dirty_regs specified the register classes dirtied by userspace
5485 * struct kvm_sync_regs is architecture specific, as well as the
5486 * bits for kvm_valid_regs and kvm_dirty_regs
5488 __u64 kvm_valid_regs;
5489 __u64 kvm_dirty_regs;
5491 struct kvm_sync_regs regs;
5492 char padding[SYNC_REGS_SIZE_BYTES];
5495 If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access
5496 certain guest registers without having to call SET/GET_*REGS. Thus we can
5497 avoid some system call overhead if userspace has to handle the exit.
5498 Userspace can query the validity of the structure by checking
5499 kvm_valid_regs for specific bits. These bits are architecture specific
5500 and usually define the validity of a groups of registers. (e.g. one bit
5501 for general purpose registers)
5503 Please note that the kernel is allowed to use the kvm_run structure as the
5504 primary storage for certain register types. Therefore, the kernel may use the
5505 values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set.
5513 6. Capabilities that can be enabled on vCPUs
5514 ============================================
5516 There are certain capabilities that change the behavior of the virtual CPU or
5517 the virtual machine when enabled. To enable them, please see section 4.37.
5518 Below you can find a list of capabilities and what their effect on the vCPU or
5519 the virtual machine is when enabling them.
5521 The following information is provided along with the description:
5524 which instruction set architectures provide this ioctl.
5525 x86 includes both i386 and x86_64.
5528 whether this is a per-vcpu or per-vm capability.
5531 what parameters are accepted by the capability.
5534 the return value. General error numbers (EBADF, ENOMEM, EINVAL)
5535 are not detailed, but errors with specific meanings are.
5544 :Returns: 0 on success; -1 on error
5546 This capability enables interception of OSI hypercalls that otherwise would
5547 be treated as normal system calls to be injected into the guest. OSI hypercalls
5548 were invented by Mac-on-Linux to have a standardized communication mechanism
5549 between the guest and the host.
5551 When this capability is enabled, KVM_EXIT_OSI can occur.
5554 6.2 KVM_CAP_PPC_PAPR
5555 --------------------
5560 :Returns: 0 on success; -1 on error
5562 This capability enables interception of PAPR hypercalls. PAPR hypercalls are
5563 done using the hypercall instruction "sc 1".
5565 It also sets the guest privilege level to "supervisor" mode. Usually the guest
5566 runs in "hypervisor" privilege mode with a few missing features.
5568 In addition to the above, it changes the semantics of SDR1. In this mode, the
5569 HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the
5570 HTAB invisible to the guest.
5572 When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur.
5580 :Parameters: args[0] is the address of a struct kvm_config_tlb
5581 :Returns: 0 on success; -1 on error
5585 struct kvm_config_tlb {
5592 Configures the virtual CPU's TLB array, establishing a shared memory area
5593 between userspace and KVM. The "params" and "array" fields are userspace
5594 addresses of mmu-type-specific data structures. The "array_len" field is an
5595 safety mechanism, and should be set to the size in bytes of the memory that
5596 userspace has reserved for the array. It must be at least the size dictated
5597 by "mmu_type" and "params".
5599 While KVM_RUN is active, the shared region is under control of KVM. Its
5600 contents are undefined, and any modification by userspace results in
5601 boundedly undefined behavior.
5603 On return from KVM_RUN, the shared region will reflect the current state of
5604 the guest's TLB. If userspace makes any changes, it must call KVM_DIRTY_TLB
5605 to tell KVM which entries have been changed, prior to calling KVM_RUN again
5608 For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV:
5610 - The "params" field is of type "struct kvm_book3e_206_tlb_params".
5611 - The "array" field points to an array of type "struct
5612 kvm_book3e_206_tlb_entry".
5613 - The array consists of all entries in the first TLB, followed by all
5614 entries in the second TLB.
5615 - Within a TLB, entries are ordered first by increasing set number. Within a
5616 set, entries are ordered by way (increasing ESEL).
5617 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1)
5618 where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value.
5619 - The tsize field of mas1 shall be set to 4K on TLB0, even though the
5620 hardware ignores this value for TLB0.
5622 6.4 KVM_CAP_S390_CSS_SUPPORT
5623 ----------------------------
5625 :Architectures: s390
5628 :Returns: 0 on success; -1 on error
5630 This capability enables support for handling of channel I/O instructions.
5632 TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are
5633 handled in-kernel, while the other I/O instructions are passed to userspace.
5635 When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST
5636 SUBCHANNEL intercepts.
5638 Note that even though this capability is enabled per-vcpu, the complete
5639 virtual machine is affected.
5646 :Parameters: args[0] defines whether the proxy facility is active
5647 :Returns: 0 on success; -1 on error
5649 This capability enables or disables the delivery of interrupts through the
5650 external proxy facility.
5652 When enabled (args[0] != 0), every time the guest gets an external interrupt
5653 delivered, it automatically exits into user space with a KVM_EXIT_EPR exit
5654 to receive the topmost interrupt vector.
5656 When disabled (args[0] == 0), behavior is as if this facility is unsupported.
5658 When this capability is enabled, KVM_EXIT_EPR can occur.
5660 6.6 KVM_CAP_IRQ_MPIC
5661 --------------------
5664 :Parameters: args[0] is the MPIC device fd;
5665 args[1] is the MPIC CPU number for this vcpu
5667 This capability connects the vcpu to an in-kernel MPIC device.
5669 6.7 KVM_CAP_IRQ_XICS
5670 --------------------
5674 :Parameters: args[0] is the XICS device fd;
5675 args[1] is the XICS CPU number (server ID) for this vcpu
5677 This capability connects the vcpu to an in-kernel XICS device.
5679 6.8 KVM_CAP_S390_IRQCHIP
5680 ------------------------
5682 :Architectures: s390
5686 This capability enables the in-kernel irqchip for s390. Please refer to
5687 "4.24 KVM_CREATE_IRQCHIP" for details.
5689 6.9 KVM_CAP_MIPS_FPU
5690 --------------------
5692 :Architectures: mips
5694 :Parameters: args[0] is reserved for future use (should be 0).
5696 This capability allows the use of the host Floating Point Unit by the guest. It
5697 allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is
5698 done the ``KVM_REG_MIPS_FPR_*`` and ``KVM_REG_MIPS_FCR_*`` registers can be
5699 accessed (depending on the current guest FPU register mode), and the Status.FR,
5700 Config5.FRE bits are accessible via the KVM API and also from the guest,
5701 depending on them being supported by the FPU.
5703 6.10 KVM_CAP_MIPS_MSA
5704 ---------------------
5706 :Architectures: mips
5708 :Parameters: args[0] is reserved for future use (should be 0).
5710 This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest.
5711 It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest.
5712 Once this is done the ``KVM_REG_MIPS_VEC_*`` and ``KVM_REG_MIPS_MSA_*``
5713 registers can be accessed, and the Config5.MSAEn bit is accessible via the
5714 KVM API and also from the guest.
5716 6.74 KVM_CAP_SYNC_REGS
5717 ----------------------
5719 :Architectures: s390, x86
5720 :Target: s390: always enabled, x86: vcpu
5722 :Returns: x86: KVM_CHECK_EXTENSION returns a bit-array indicating which register
5724 (bitfields defined in arch/x86/include/uapi/asm/kvm.h).
5726 As described above in the kvm_sync_regs struct info in section 5 (kvm_run):
5727 KVM_CAP_SYNC_REGS "allow[s] userspace to access certain guest registers
5728 without having to call SET/GET_*REGS". This reduces overhead by eliminating
5729 repeated ioctl calls for setting and/or getting register values. This is
5730 particularly important when userspace is making synchronous guest state
5731 modifications, e.g. when emulating and/or intercepting instructions in
5734 For s390 specifics, please refer to the source code.
5738 - the register sets to be copied out to kvm_run are selectable
5739 by userspace (rather that all sets being copied out for every exit).
5740 - vcpu_events are available in addition to regs and sregs.
5742 For x86, the 'kvm_valid_regs' field of struct kvm_run is overloaded to
5743 function as an input bit-array field set by userspace to indicate the
5744 specific register sets to be copied out on the next exit.
5746 To indicate when userspace has modified values that should be copied into
5747 the vCPU, the all architecture bitarray field, 'kvm_dirty_regs' must be set.
5748 This is done using the same bitflags as for the 'kvm_valid_regs' field.
5749 If the dirty bit is not set, then the register set values will not be copied
5750 into the vCPU even if they've been modified.
5752 Unused bitfields in the bitarrays must be set to zero.
5756 struct kvm_sync_regs {
5757 struct kvm_regs regs;
5758 struct kvm_sregs sregs;
5759 struct kvm_vcpu_events events;
5762 6.75 KVM_CAP_PPC_IRQ_XIVE
5763 -------------------------
5767 :Parameters: args[0] is the XIVE device fd;
5768 args[1] is the XIVE CPU number (server ID) for this vcpu
5770 This capability connects the vcpu to an in-kernel XIVE device.
5772 7. Capabilities that can be enabled on VMs
5773 ==========================================
5775 There are certain capabilities that change the behavior of the virtual
5776 machine when enabled. To enable them, please see section 4.37. Below
5777 you can find a list of capabilities and what their effect on the VM
5778 is when enabling them.
5780 The following information is provided along with the description:
5783 which instruction set architectures provide this ioctl.
5784 x86 includes both i386 and x86_64.
5787 what parameters are accepted by the capability.
5790 the return value. General error numbers (EBADF, ENOMEM, EINVAL)
5791 are not detailed, but errors with specific meanings are.
5794 7.1 KVM_CAP_PPC_ENABLE_HCALL
5795 ----------------------------
5798 :Parameters: args[0] is the sPAPR hcall number;
5799 args[1] is 0 to disable, 1 to enable in-kernel handling
5801 This capability controls whether individual sPAPR hypercalls (hcalls)
5802 get handled by the kernel or not. Enabling or disabling in-kernel
5803 handling of an hcall is effective across the VM. On creation, an
5804 initial set of hcalls are enabled for in-kernel handling, which
5805 consists of those hcalls for which in-kernel handlers were implemented
5806 before this capability was implemented. If disabled, the kernel will
5807 not to attempt to handle the hcall, but will always exit to userspace
5808 to handle it. Note that it may not make sense to enable some and
5809 disable others of a group of related hcalls, but KVM does not prevent
5810 userspace from doing that.
5812 If the hcall number specified is not one that has an in-kernel
5813 implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL
5816 7.2 KVM_CAP_S390_USER_SIGP
5817 --------------------------
5819 :Architectures: s390
5822 This capability controls which SIGP orders will be handled completely in user
5823 space. With this capability enabled, all fast orders will be handled completely
5830 - CONDITIONAL EMERGENCY SIGNAL
5832 All other orders will be handled completely in user space.
5834 Only privileged operation exceptions will be checked for in the kernel (or even
5835 in the hardware prior to interception). If this capability is not enabled, the
5836 old way of handling SIGP orders is used (partially in kernel and user space).
5838 7.3 KVM_CAP_S390_VECTOR_REGISTERS
5839 ---------------------------------
5841 :Architectures: s390
5843 :Returns: 0 on success, negative value on error
5845 Allows use of the vector registers introduced with z13 processor, and
5846 provides for the synchronization between host and user space. Will
5847 return -EINVAL if the machine does not support vectors.
5849 7.4 KVM_CAP_S390_USER_STSI
5850 --------------------------
5852 :Architectures: s390
5855 This capability allows post-handlers for the STSI instruction. After
5856 initial handling in the kernel, KVM exits to user space with
5857 KVM_EXIT_S390_STSI to allow user space to insert further data.
5859 Before exiting to userspace, kvm handlers should fill in s390_stsi field of
5871 @addr - guest address of STSI SYSIB
5875 @ar - access register number
5877 KVM handlers should exit to userspace with rc = -EREMOTE.
5879 7.5 KVM_CAP_SPLIT_IRQCHIP
5880 -------------------------
5883 :Parameters: args[0] - number of routes reserved for userspace IOAPICs
5884 :Returns: 0 on success, -1 on error
5886 Create a local apic for each processor in the kernel. This can be used
5887 instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the
5888 IOAPIC and PIC (and also the PIT, even though this has to be enabled
5891 This capability also enables in kernel routing of interrupt requests;
5892 when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are
5893 used in the IRQ routing table. The first args[0] MSI routes are reserved
5894 for the IOAPIC pins. Whenever the LAPIC receives an EOI for these routes,
5895 a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace.
5897 Fails if VCPU has already been created, or if the irqchip is already in the
5898 kernel (i.e. KVM_CREATE_IRQCHIP has already been called).
5903 :Architectures: s390
5906 Allows use of runtime-instrumentation introduced with zEC12 processor.
5907 Will return -EINVAL if the machine does not support runtime-instrumentation.
5908 Will return -EBUSY if a VCPU has already been created.
5910 7.7 KVM_CAP_X2APIC_API
5911 ----------------------
5914 :Parameters: args[0] - features that should be enabled
5915 :Returns: 0 on success, -EINVAL when args[0] contains invalid features
5917 Valid feature flags in args[0] are::
5919 #define KVM_X2APIC_API_USE_32BIT_IDS (1ULL << 0)
5920 #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK (1ULL << 1)
5922 Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of
5923 KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC,
5924 allowing the use of 32-bit APIC IDs. See KVM_CAP_X2APIC_API in their
5925 respective sections.
5927 KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work
5928 in logical mode or with more than 255 VCPUs. Otherwise, KVM treats 0xff
5929 as a broadcast even in x2APIC mode in order to support physical x2APIC
5930 without interrupt remapping. This is undesirable in logical mode,
5931 where 0xff represents CPUs 0-7 in cluster 0.
5933 7.8 KVM_CAP_S390_USER_INSTR0
5934 ----------------------------
5936 :Architectures: s390
5939 With this capability enabled, all illegal instructions 0x0000 (2 bytes) will
5940 be intercepted and forwarded to user space. User space can use this
5941 mechanism e.g. to realize 2-byte software breakpoints. The kernel will
5942 not inject an operating exception for these instructions, user space has
5943 to take care of that.
5945 This capability can be enabled dynamically even if VCPUs were already
5946 created and are running.
5951 :Architectures: s390
5953 :Returns: 0 on success; -EINVAL if the machine does not support
5954 guarded storage; -EBUSY if a VCPU has already been created.
5956 Allows use of guarded storage for the KVM guest.
5958 7.10 KVM_CAP_S390_AIS
5959 ---------------------
5961 :Architectures: s390
5964 Allow use of adapter-interruption suppression.
5965 :Returns: 0 on success; -EBUSY if a VCPU has already been created.
5967 7.11 KVM_CAP_PPC_SMT
5968 --------------------
5971 :Parameters: vsmt_mode, flags
5973 Enabling this capability on a VM provides userspace with a way to set
5974 the desired virtual SMT mode (i.e. the number of virtual CPUs per
5975 virtual core). The virtual SMT mode, vsmt_mode, must be a power of 2
5976 between 1 and 8. On POWER8, vsmt_mode must also be no greater than
5977 the number of threads per subcore for the host. Currently flags must
5978 be 0. A successful call to enable this capability will result in
5979 vsmt_mode being returned when the KVM_CAP_PPC_SMT capability is
5980 subsequently queried for the VM. This capability is only supported by
5981 HV KVM, and can only be set before any VCPUs have been created.
5982 The KVM_CAP_PPC_SMT_POSSIBLE capability indicates which virtual SMT
5983 modes are available.
5985 7.12 KVM_CAP_PPC_FWNMI
5986 ----------------------
5991 With this capability a machine check exception in the guest address
5992 space will cause KVM to exit the guest with NMI exit reason. This
5993 enables QEMU to build error log and branch to guest kernel registered
5994 machine check handling routine. Without this capability KVM will
5995 branch to guests' 0x200 interrupt vector.
5997 7.13 KVM_CAP_X86_DISABLE_EXITS
5998 ------------------------------
6001 :Parameters: args[0] defines which exits are disabled
6002 :Returns: 0 on success, -EINVAL when args[0] contains invalid exits
6004 Valid bits in args[0] are::
6006 #define KVM_X86_DISABLE_EXITS_MWAIT (1 << 0)
6007 #define KVM_X86_DISABLE_EXITS_HLT (1 << 1)
6008 #define KVM_X86_DISABLE_EXITS_PAUSE (1 << 2)
6009 #define KVM_X86_DISABLE_EXITS_CSTATE (1 << 3)
6011 Enabling this capability on a VM provides userspace with a way to no
6012 longer intercept some instructions for improved latency in some
6013 workloads, and is suggested when vCPUs are associated to dedicated
6014 physical CPUs. More bits can be added in the future; userspace can
6015 just pass the KVM_CHECK_EXTENSION result to KVM_ENABLE_CAP to disable
6018 Do not enable KVM_FEATURE_PV_UNHALT if you disable HLT exits.
6020 7.14 KVM_CAP_S390_HPAGE_1M
6021 --------------------------
6023 :Architectures: s390
6025 :Returns: 0 on success, -EINVAL if hpage module parameter was not set
6026 or cmma is enabled, or the VM has the KVM_VM_S390_UCONTROL
6029 With this capability the KVM support for memory backing with 1m pages
6030 through hugetlbfs can be enabled for a VM. After the capability is
6031 enabled, cmma can't be enabled anymore and pfmfi and the storage key
6032 interpretation are disabled. If cmma has already been enabled or the
6033 hpage module parameter is not set to 1, -EINVAL is returned.
6035 While it is generally possible to create a huge page backed VM without
6036 this capability, the VM will not be able to run.
6038 7.15 KVM_CAP_MSR_PLATFORM_INFO
6039 ------------------------------
6042 :Parameters: args[0] whether feature should be enabled or not
6044 With this capability, a guest may read the MSR_PLATFORM_INFO MSR. Otherwise,
6045 a #GP would be raised when the guest tries to access. Currently, this
6046 capability does not enable write permissions of this MSR for the guest.
6048 7.16 KVM_CAP_PPC_NESTED_HV
6049 --------------------------
6053 :Returns: 0 on success, -EINVAL when the implementation doesn't support
6054 nested-HV virtualization.
6056 HV-KVM on POWER9 and later systems allows for "nested-HV"
6057 virtualization, which provides a way for a guest VM to run guests that
6058 can run using the CPU's supervisor mode (privileged non-hypervisor
6059 state). Enabling this capability on a VM depends on the CPU having
6060 the necessary functionality and on the facility being enabled with a
6061 kvm-hv module parameter.
6063 7.17 KVM_CAP_EXCEPTION_PAYLOAD
6064 ------------------------------
6067 :Parameters: args[0] whether feature should be enabled or not
6069 With this capability enabled, CR2 will not be modified prior to the
6070 emulated VM-exit when L1 intercepts a #PF exception that occurs in
6071 L2. Similarly, for kvm-intel only, DR6 will not be modified prior to
6072 the emulated VM-exit when L1 intercepts a #DB exception that occurs in
6073 L2. As a result, when KVM_GET_VCPU_EVENTS reports a pending #PF (or
6074 #DB) exception for L2, exception.has_payload will be set and the
6075 faulting address (or the new DR6 bits*) will be reported in the
6076 exception_payload field. Similarly, when userspace injects a #PF (or
6077 #DB) into L2 using KVM_SET_VCPU_EVENTS, it is expected to set
6078 exception.has_payload and to put the faulting address - or the new DR6
6079 bits\ [#]_ - in the exception_payload field.
6081 This capability also enables exception.pending in struct
6082 kvm_vcpu_events, which allows userspace to distinguish between pending
6083 and injected exceptions.
6086 .. [#] For the new DR6 bits, note that bit 16 is set iff the #DB exception
6089 7.18 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
6091 :Architectures: x86, arm, arm64, mips
6092 :Parameters: args[0] whether feature should be enabled or not
6096 #define KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (1 << 0)
6097 #define KVM_DIRTY_LOG_INITIALLY_SET (1 << 1)
6099 With KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE is set, KVM_GET_DIRTY_LOG will not
6100 automatically clear and write-protect all pages that are returned as dirty.
6101 Rather, userspace will have to do this operation separately using
6102 KVM_CLEAR_DIRTY_LOG.
6104 At the cost of a slightly more complicated operation, this provides better
6105 scalability and responsiveness for two reasons. First,
6106 KVM_CLEAR_DIRTY_LOG ioctl can operate on a 64-page granularity rather
6107 than requiring to sync a full memslot; this ensures that KVM does not
6108 take spinlocks for an extended period of time. Second, in some cases a
6109 large amount of time can pass between a call to KVM_GET_DIRTY_LOG and
6110 userspace actually using the data in the page. Pages can be modified
6111 during this time, which is inefficient for both the guest and userspace:
6112 the guest will incur a higher penalty due to write protection faults,
6113 while userspace can see false reports of dirty pages. Manual reprotection
6114 helps reducing this time, improving guest performance and reducing the
6115 number of dirty log false positives.
6117 With KVM_DIRTY_LOG_INITIALLY_SET set, all the bits of the dirty bitmap
6118 will be initialized to 1 when created. This also improves performance because
6119 dirty logging can be enabled gradually in small chunks on the first call
6120 to KVM_CLEAR_DIRTY_LOG. KVM_DIRTY_LOG_INITIALLY_SET depends on
6121 KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (it is also only available on
6122 x86 and arm64 for now).
6124 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 was previously available under the name
6125 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT, but the implementation had bugs that make
6126 it hard or impossible to use it correctly. The availability of
6127 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 signals that those bugs are fixed.
6128 Userspace should not try to use KVM_CAP_MANUAL_DIRTY_LOG_PROTECT.
6130 7.19 KVM_CAP_PPC_SECURE_GUEST
6131 ------------------------------
6135 This capability indicates that KVM is running on a host that has
6136 ultravisor firmware and thus can support a secure guest. On such a
6137 system, a guest can ask the ultravisor to make it a secure guest,
6138 one whose memory is inaccessible to the host except for pages which
6139 are explicitly requested to be shared with the host. The ultravisor
6140 notifies KVM when a guest requests to become a secure guest, and KVM
6141 has the opportunity to veto the transition.
6143 If present, this capability can be enabled for a VM, meaning that KVM
6144 will allow the transition to secure guest mode. Otherwise KVM will
6145 veto the transition.
6147 7.20 KVM_CAP_HALT_POLL
6148 ----------------------
6152 :Parameters: args[0] is the maximum poll time in nanoseconds
6153 :Returns: 0 on success; -1 on error
6155 This capability overrides the kvm module parameter halt_poll_ns for the
6158 VCPU polling allows a VCPU to poll for wakeup events instead of immediately
6159 scheduling during guest halts. The maximum time a VCPU can spend polling is
6160 controlled by the kvm module parameter halt_poll_ns. This capability allows
6161 the maximum halt time to specified on a per-VM basis, effectively overriding
6162 the module parameter for the target VM.
6164 7.21 KVM_CAP_X86_USER_SPACE_MSR
6165 -------------------------------
6169 :Parameters: args[0] contains the mask of KVM_MSR_EXIT_REASON_* events to report
6170 :Returns: 0 on success; -1 on error
6172 This capability enables trapping of #GP invoking RDMSR and WRMSR instructions
6175 When a guest requests to read or write an MSR, KVM may not implement all MSRs
6176 that are relevant to a respective system. It also does not differentiate by
6179 To allow more fine grained control over MSR handling, user space may enable
6180 this capability. With it enabled, MSR accesses that match the mask specified in
6181 args[0] and trigger a #GP event inside the guest by KVM will instead trigger
6182 KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR exit notifications which user space
6183 can then handle to implement model specific MSR handling and/or user notifications
6184 to inform a user that an MSR was not handled.
6186 7.22 KVM_CAP_X86_BUS_LOCK_EXIT
6187 -------------------------------
6191 :Parameters: args[0] defines the policy used when bus locks detected in guest
6192 :Returns: 0 on success, -EINVAL when args[0] contains invalid bits
6194 Valid bits in args[0] are::
6196 #define KVM_BUS_LOCK_DETECTION_OFF (1 << 0)
6197 #define KVM_BUS_LOCK_DETECTION_EXIT (1 << 1)
6199 Enabling this capability on a VM provides userspace with a way to select
6200 a policy to handle the bus locks detected in guest. Userspace can obtain
6201 the supported modes from the result of KVM_CHECK_EXTENSION and define it
6202 through the KVM_ENABLE_CAP.
6204 KVM_BUS_LOCK_DETECTION_OFF and KVM_BUS_LOCK_DETECTION_EXIT are supported
6205 currently and mutually exclusive with each other. More bits can be added in
6208 With KVM_BUS_LOCK_DETECTION_OFF set, bus locks in guest will not cause vm exits
6209 so that no additional actions are needed. This is the default mode.
6211 With KVM_BUS_LOCK_DETECTION_EXIT set, vm exits happen when bus lock detected
6212 in VM. KVM just exits to userspace when handling them. Userspace can enforce
6213 its own throttling or other policy based mitigations.
6215 This capability is aimed to address the thread that VM can exploit bus locks to
6216 degree the performance of the whole system. Once the userspace enable this
6217 capability and select the KVM_BUS_LOCK_DETECTION_EXIT mode, KVM will set the
6218 KVM_RUN_BUS_LOCK flag in vcpu-run->flags field and exit to userspace. Concerning
6219 the bus lock vm exit can be preempted by a higher priority VM exit, the exit
6220 notifications to userspace can be KVM_EXIT_BUS_LOCK or other reasons.
6221 KVM_RUN_BUS_LOCK flag is used to distinguish between them.
6223 7.23 KVM_CAP_PPC_DAWR1
6224 ----------------------
6228 :Returns: 0 on success, -EINVAL when CPU doesn't support 2nd DAWR
6230 This capability can be used to check / enable 2nd DAWR feature provided
6231 by POWER10 processor.
6233 8. Other capabilities.
6234 ======================
6236 This section lists capabilities that give information about other
6237 features of the KVM implementation.
6239 8.1 KVM_CAP_PPC_HWRNG
6240 ---------------------
6244 This capability, if KVM_CHECK_EXTENSION indicates that it is
6245 available, means that the kernel has an implementation of the
6246 H_RANDOM hypercall backed by a hardware random-number generator.
6247 If present, the kernel H_RANDOM handler can be enabled for guest use
6248 with the KVM_CAP_PPC_ENABLE_HCALL capability.
6250 8.2 KVM_CAP_HYPERV_SYNIC
6251 ------------------------
6255 This capability, if KVM_CHECK_EXTENSION indicates that it is
6256 available, means that the kernel has an implementation of the
6257 Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is
6258 used to support Windows Hyper-V based guest paravirt drivers(VMBus).
6260 In order to use SynIC, it has to be activated by setting this
6261 capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this
6262 will disable the use of APIC hardware virtualization even if supported
6263 by the CPU, as it's incompatible with SynIC auto-EOI behavior.
6265 8.3 KVM_CAP_PPC_RADIX_MMU
6266 -------------------------
6270 This capability, if KVM_CHECK_EXTENSION indicates that it is
6271 available, means that the kernel can support guests using the
6272 radix MMU defined in Power ISA V3.00 (as implemented in the POWER9
6275 8.4 KVM_CAP_PPC_HASH_MMU_V3
6276 ---------------------------
6280 This capability, if KVM_CHECK_EXTENSION indicates that it is
6281 available, means that the kernel can support guests using the
6282 hashed page table MMU defined in Power ISA V3.00 (as implemented in
6283 the POWER9 processor), including in-memory segment tables.
6288 :Architectures: mips
6290 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
6291 it is available, means that full hardware assisted virtualization capabilities
6292 of the hardware are available for use through KVM. An appropriate
6293 KVM_VM_MIPS_* type must be passed to KVM_CREATE_VM to create a VM which
6296 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
6297 available, it means that the VM is using full hardware assisted virtualization
6298 capabilities of the hardware. This is useful to check after creating a VM with
6299 KVM_VM_MIPS_DEFAULT.
6301 The value returned by KVM_CHECK_EXTENSION should be compared against known
6302 values (see below). All other values are reserved. This is to allow for the
6303 possibility of other hardware assisted virtualization implementations which
6304 may be incompatible with the MIPS VZ ASE.
6306 == ==========================================================================
6307 0 The trap & emulate implementation is in use to run guest code in user
6308 mode. Guest virtual memory segments are rearranged to fit the guest in the
6309 user mode address space.
6311 1 The MIPS VZ ASE is in use, providing full hardware assisted
6312 virtualization, including standard guest virtual memory segments.
6313 == ==========================================================================
6318 :Architectures: mips
6320 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
6321 it is available, means that the trap & emulate implementation is available to
6322 run guest code in user mode, even if KVM_CAP_MIPS_VZ indicates that hardware
6323 assisted virtualisation is also available. KVM_VM_MIPS_TE (0) must be passed
6324 to KVM_CREATE_VM to create a VM which utilises it.
6326 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
6327 available, it means that the VM is using trap & emulate.
6329 8.7 KVM_CAP_MIPS_64BIT
6330 ----------------------
6332 :Architectures: mips
6334 This capability indicates the supported architecture type of the guest, i.e. the
6335 supported register and address width.
6337 The values returned when this capability is checked by KVM_CHECK_EXTENSION on a
6338 kvm VM handle correspond roughly to the CP0_Config.AT register field, and should
6339 be checked specifically against known values (see below). All other values are
6342 == ========================================================================
6343 0 MIPS32 or microMIPS32.
6344 Both registers and addresses are 32-bits wide.
6345 It will only be possible to run 32-bit guest code.
6347 1 MIPS64 or microMIPS64 with access only to 32-bit compatibility segments.
6348 Registers are 64-bits wide, but addresses are 32-bits wide.
6349 64-bit guest code may run but cannot access MIPS64 memory segments.
6350 It will also be possible to run 32-bit guest code.
6352 2 MIPS64 or microMIPS64 with access to all address segments.
6353 Both registers and addresses are 64-bits wide.
6354 It will be possible to run 64-bit or 32-bit guest code.
6355 == ========================================================================
6357 8.9 KVM_CAP_ARM_USER_IRQ
6358 ------------------------
6360 :Architectures: arm, arm64
6362 This capability, if KVM_CHECK_EXTENSION indicates that it is available, means
6363 that if userspace creates a VM without an in-kernel interrupt controller, it
6364 will be notified of changes to the output level of in-kernel emulated devices,
6365 which can generate virtual interrupts, presented to the VM.
6366 For such VMs, on every return to userspace, the kernel
6367 updates the vcpu's run->s.regs.device_irq_level field to represent the actual
6368 output level of the device.
6370 Whenever kvm detects a change in the device output level, kvm guarantees at
6371 least one return to userspace before running the VM. This exit could either
6372 be a KVM_EXIT_INTR or any other exit event, like KVM_EXIT_MMIO. This way,
6373 userspace can always sample the device output level and re-compute the state of
6374 the userspace interrupt controller. Userspace should always check the state
6375 of run->s.regs.device_irq_level on every kvm exit.
6376 The value in run->s.regs.device_irq_level can represent both level and edge
6377 triggered interrupt signals, depending on the device. Edge triggered interrupt
6378 signals will exit to userspace with the bit in run->s.regs.device_irq_level
6379 set exactly once per edge signal.
6381 The field run->s.regs.device_irq_level is available independent of
6382 run->kvm_valid_regs or run->kvm_dirty_regs bits.
6384 If KVM_CAP_ARM_USER_IRQ is supported, the KVM_CHECK_EXTENSION ioctl returns a
6385 number larger than 0 indicating the version of this capability is implemented
6386 and thereby which bits in run->s.regs.device_irq_level can signal values.
6388 Currently the following bits are defined for the device_irq_level bitmap::
6390 KVM_CAP_ARM_USER_IRQ >= 1:
6392 KVM_ARM_DEV_EL1_VTIMER - EL1 virtual timer
6393 KVM_ARM_DEV_EL1_PTIMER - EL1 physical timer
6394 KVM_ARM_DEV_PMU - ARM PMU overflow interrupt signal
6396 Future versions of kvm may implement additional events. These will get
6397 indicated by returning a higher number from KVM_CHECK_EXTENSION and will be
6400 8.10 KVM_CAP_PPC_SMT_POSSIBLE
6401 -----------------------------
6405 Querying this capability returns a bitmap indicating the possible
6406 virtual SMT modes that can be set using KVM_CAP_PPC_SMT. If bit N
6407 (counting from the right) is set, then a virtual SMT mode of 2^N is
6410 8.11 KVM_CAP_HYPERV_SYNIC2
6411 --------------------------
6415 This capability enables a newer version of Hyper-V Synthetic interrupt
6416 controller (SynIC). The only difference with KVM_CAP_HYPERV_SYNIC is that KVM
6417 doesn't clear SynIC message and event flags pages when they are enabled by
6418 writing to the respective MSRs.
6420 8.12 KVM_CAP_HYPERV_VP_INDEX
6421 ----------------------------
6425 This capability indicates that userspace can load HV_X64_MSR_VP_INDEX msr. Its
6426 value is used to denote the target vcpu for a SynIC interrupt. For
6427 compatibilty, KVM initializes this msr to KVM's internal vcpu index. When this
6428 capability is absent, userspace can still query this msr's value.
6430 8.13 KVM_CAP_S390_AIS_MIGRATION
6431 -------------------------------
6433 :Architectures: s390
6436 This capability indicates if the flic device will be able to get/set the
6437 AIS states for migration via the KVM_DEV_FLIC_AISM_ALL attribute and allows
6438 to discover this without having to create a flic device.
6440 8.14 KVM_CAP_S390_PSW
6441 ---------------------
6443 :Architectures: s390
6445 This capability indicates that the PSW is exposed via the kvm_run structure.
6447 8.15 KVM_CAP_S390_GMAP
6448 ----------------------
6450 :Architectures: s390
6452 This capability indicates that the user space memory used as guest mapping can
6453 be anywhere in the user memory address space, as long as the memory slots are
6454 aligned and sized to a segment (1MB) boundary.
6456 8.16 KVM_CAP_S390_COW
6457 ---------------------
6459 :Architectures: s390
6461 This capability indicates that the user space memory used as guest mapping can
6462 use copy-on-write semantics as well as dirty pages tracking via read-only page
6465 8.17 KVM_CAP_S390_BPB
6466 ---------------------
6468 :Architectures: s390
6470 This capability indicates that kvm will implement the interfaces to handle
6471 reset, migration and nested KVM for branch prediction blocking. The stfle
6472 facility 82 should not be provided to the guest without this capability.
6474 8.18 KVM_CAP_HYPERV_TLBFLUSH
6475 ----------------------------
6479 This capability indicates that KVM supports paravirtualized Hyper-V TLB Flush
6481 HvFlushVirtualAddressSpace, HvFlushVirtualAddressSpaceEx,
6482 HvFlushVirtualAddressList, HvFlushVirtualAddressListEx.
6484 8.19 KVM_CAP_ARM_INJECT_SERROR_ESR
6485 ----------------------------------
6487 :Architectures: arm, arm64
6489 This capability indicates that userspace can specify (via the
6490 KVM_SET_VCPU_EVENTS ioctl) the syndrome value reported to the guest when it
6491 takes a virtual SError interrupt exception.
6492 If KVM advertises this capability, userspace can only specify the ISS field for
6493 the ESR syndrome. Other parts of the ESR, such as the EC are generated by the
6494 CPU when the exception is taken. If this virtual SError is taken to EL1 using
6495 AArch64, this value will be reported in the ISS field of ESR_ELx.
6497 See KVM_CAP_VCPU_EVENTS for more details.
6499 8.20 KVM_CAP_HYPERV_SEND_IPI
6500 ----------------------------
6504 This capability indicates that KVM supports paravirtualized Hyper-V IPI send
6506 HvCallSendSyntheticClusterIpi, HvCallSendSyntheticClusterIpiEx.
6508 8.21 KVM_CAP_HYPERV_DIRECT_TLBFLUSH
6509 -----------------------------------
6513 This capability indicates that KVM running on top of Hyper-V hypervisor
6514 enables Direct TLB flush for its guests meaning that TLB flush
6515 hypercalls are handled by Level 0 hypervisor (Hyper-V) bypassing KVM.
6516 Due to the different ABI for hypercall parameters between Hyper-V and
6517 KVM, enabling this capability effectively disables all hypercall
6518 handling by KVM (as some KVM hypercall may be mistakenly treated as TLB
6519 flush hypercalls by Hyper-V) so userspace should disable KVM identification
6520 in CPUID and only exposes Hyper-V identification. In this case, guest
6521 thinks it's running on Hyper-V and only use Hyper-V hypercalls.
6523 8.22 KVM_CAP_S390_VCPU_RESETS
6524 -----------------------------
6526 :Architectures: s390
6528 This capability indicates that the KVM_S390_NORMAL_RESET and
6529 KVM_S390_CLEAR_RESET ioctls are available.
6531 8.23 KVM_CAP_S390_PROTECTED
6532 ---------------------------
6534 :Architectures: s390
6536 This capability indicates that the Ultravisor has been initialized and
6537 KVM can therefore start protected VMs.
6538 This capability governs the KVM_S390_PV_COMMAND ioctl and the
6539 KVM_MP_STATE_LOAD MP_STATE. KVM_SET_MP_STATE can fail for protected
6540 guests when the state change is invalid.
6542 8.24 KVM_CAP_STEAL_TIME
6543 -----------------------
6545 :Architectures: arm64, x86
6547 This capability indicates that KVM supports steal time accounting.
6548 When steal time accounting is supported it may be enabled with
6549 architecture-specific interfaces. This capability and the architecture-
6550 specific interfaces must be consistent, i.e. if one says the feature
6551 is supported, than the other should as well and vice versa. For arm64
6552 see Documentation/virt/kvm/devices/vcpu.rst "KVM_ARM_VCPU_PVTIME_CTRL".
6553 For x86 see Documentation/virt/kvm/msr.rst "MSR_KVM_STEAL_TIME".
6555 8.25 KVM_CAP_S390_DIAG318
6556 -------------------------
6558 :Architectures: s390
6560 This capability enables a guest to set information about its control program
6561 (i.e. guest kernel type and version). The information is helpful during
6562 system/firmware service events, providing additional data about the guest
6563 environments running on the machine.
6565 The information is associated with the DIAGNOSE 0x318 instruction, which sets
6566 an 8-byte value consisting of a one-byte Control Program Name Code (CPNC) and
6567 a 7-byte Control Program Version Code (CPVC). The CPNC determines what
6568 environment the control program is running in (e.g. Linux, z/VM...), and the
6569 CPVC is used for information specific to OS (e.g. Linux version, Linux
6572 If this capability is available, then the CPNC and CPVC can be synchronized
6573 between KVM and userspace via the sync regs mechanism (KVM_SYNC_DIAG318).
6575 8.26 KVM_CAP_X86_USER_SPACE_MSR
6576 -------------------------------
6580 This capability indicates that KVM supports deflection of MSR reads and
6581 writes to user space. It can be enabled on a VM level. If enabled, MSR
6582 accesses that would usually trigger a #GP by KVM into the guest will
6583 instead get bounced to user space through the KVM_EXIT_X86_RDMSR and
6584 KVM_EXIT_X86_WRMSR exit notifications.
6586 8.27 KVM_X86_SET_MSR_FILTER
6587 ---------------------------
6591 This capability indicates that KVM supports that accesses to user defined MSRs
6592 may be rejected. With this capability exposed, KVM exports new VM ioctl
6593 KVM_X86_SET_MSR_FILTER which user space can call to specify bitmaps of MSR
6594 ranges that KVM should reject access to.
6596 In combination with KVM_CAP_X86_USER_SPACE_MSR, this allows user space to
6597 trap and emulate MSRs that are outside of the scope of KVM as well as
6598 limit the attack surface on KVM's MSR emulation code.
6600 8.28 KVM_CAP_ENFORCE_PV_CPUID
6601 -----------------------------
6605 When enabled, KVM will disable paravirtual features provided to the
6606 guest according to the bits in the KVM_CPUID_FEATURES CPUID leaf
6607 (0x40000001). Otherwise, a guest may use the paravirtual features
6608 regardless of what has actually been exposed through the CPUID leaf.
6610 8.29 KVM_CAP_DIRTY_LOG_RING
6611 ---------------------------
6614 :Parameters: args[0] - size of the dirty log ring
6616 KVM is capable of tracking dirty memory using ring buffers that are
6617 mmaped into userspace; there is one dirty ring per vcpu.
6619 The dirty ring is available to userspace as an array of
6620 ``struct kvm_dirty_gfn``. Each dirty entry it's defined as::
6622 struct kvm_dirty_gfn {
6624 __u32 slot; /* as_id | slot_id */
6628 The following values are defined for the flags field to define the
6629 current state of the entry::
6631 #define KVM_DIRTY_GFN_F_DIRTY BIT(0)
6632 #define KVM_DIRTY_GFN_F_RESET BIT(1)
6633 #define KVM_DIRTY_GFN_F_MASK 0x3
6635 Userspace should call KVM_ENABLE_CAP ioctl right after KVM_CREATE_VM
6636 ioctl to enable this capability for the new guest and set the size of
6637 the rings. Enabling the capability is only allowed before creating any
6638 vCPU, and the size of the ring must be a power of two. The larger the
6639 ring buffer, the less likely the ring is full and the VM is forced to
6640 exit to userspace. The optimal size depends on the workload, but it is
6641 recommended that it be at least 64 KiB (4096 entries).
6643 Just like for dirty page bitmaps, the buffer tracks writes to
6644 all user memory regions for which the KVM_MEM_LOG_DIRTY_PAGES flag was
6645 set in KVM_SET_USER_MEMORY_REGION. Once a memory region is registered
6646 with the flag set, userspace can start harvesting dirty pages from the
6649 An entry in the ring buffer can be unused (flag bits ``00``),
6650 dirty (flag bits ``01``) or harvested (flag bits ``1X``). The
6651 state machine for the entry is as follows::
6653 dirtied harvested reset
6654 00 -----------> 01 -------------> 1X -------+
6657 +------------------------------------------+
6659 To harvest the dirty pages, userspace accesses the mmaped ring buffer
6660 to read the dirty GFNs. If the flags has the DIRTY bit set (at this stage
6661 the RESET bit must be cleared), then it means this GFN is a dirty GFN.
6662 The userspace should harvest this GFN and mark the flags from state
6663 ``01b`` to ``1Xb`` (bit 0 will be ignored by KVM, but bit 1 must be set
6664 to show that this GFN is harvested and waiting for a reset), and move
6665 on to the next GFN. The userspace should continue to do this until the
6666 flags of a GFN have the DIRTY bit cleared, meaning that it has harvested
6667 all the dirty GFNs that were available.
6669 It's not necessary for userspace to harvest the all dirty GFNs at once.
6670 However it must collect the dirty GFNs in sequence, i.e., the userspace
6671 program cannot skip one dirty GFN to collect the one next to it.
6673 After processing one or more entries in the ring buffer, userspace
6674 calls the VM ioctl KVM_RESET_DIRTY_RINGS to notify the kernel about
6675 it, so that the kernel will reprotect those collected GFNs.
6676 Therefore, the ioctl must be called *before* reading the content of
6679 The dirty ring can get full. When it happens, the KVM_RUN of the
6680 vcpu will return with exit reason KVM_EXIT_DIRTY_LOG_FULL.
6682 The dirty ring interface has a major difference comparing to the
6683 KVM_GET_DIRTY_LOG interface in that, when reading the dirty ring from
6684 userspace, it's still possible that the kernel has not yet flushed the
6685 processor's dirty page buffers into the kernel buffer (with dirty bitmaps, the
6686 flushing is done by the KVM_GET_DIRTY_LOG ioctl). To achieve that, one
6687 needs to kick the vcpu out of KVM_RUN using a signal. The resulting
6688 vmexit ensures that all dirty GFNs are flushed to the dirty rings.
6690 NOTE: the capability KVM_CAP_DIRTY_LOG_RING and the corresponding
6691 ioctl KVM_RESET_DIRTY_RINGS are mutual exclusive to the existing ioctls
6692 KVM_GET_DIRTY_LOG and KVM_CLEAR_DIRTY_LOG. After enabling
6693 KVM_CAP_DIRTY_LOG_RING with an acceptable dirty ring size, the virtual
6694 machine will switch to ring-buffer dirty page tracking and further
6695 KVM_GET_DIRTY_LOG or KVM_CLEAR_DIRTY_LOG ioctls will fail.
6697 8.30 KVM_CAP_XEN_HVM
6698 --------------------
6702 This capability indicates the features that Xen supports for hosting Xen
6703 PVHVM guests. Valid flags are::
6705 #define KVM_XEN_HVM_CONFIG_HYPERCALL_MSR (1 << 0)
6706 #define KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL (1 << 1)
6707 #define KVM_XEN_HVM_CONFIG_SHARED_INFO (1 << 2)
6708 #define KVM_XEN_HVM_CONFIG_RUNSTATE (1 << 2)
6710 The KVM_XEN_HVM_CONFIG_HYPERCALL_MSR flag indicates that the KVM_XEN_HVM_CONFIG
6711 ioctl is available, for the guest to set its hypercall page.
6713 If KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL is also set, the same flag may also be
6714 provided in the flags to KVM_XEN_HVM_CONFIG, without providing hypercall page
6715 contents, to request that KVM generate hypercall page content automatically
6716 and also enable interception of guest hypercalls with KVM_EXIT_XEN.
6718 The KVM_XEN_HVM_CONFIG_SHARED_INFO flag indicates the availability of the
6719 KVM_XEN_HVM_SET_ATTR, KVM_XEN_HVM_GET_ATTR, KVM_XEN_VCPU_SET_ATTR and
6720 KVM_XEN_VCPU_GET_ATTR ioctls, as well as the delivery of exception vectors
6721 for event channel upcalls when the evtchn_upcall_pending field of a vcpu's
6724 The KVM_XEN_HVM_CONFIG_RUNSTATE flag indicates that the runstate-related
6725 features KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR/_CURRENT/_DATA/_ADJUST are
6726 supported by the KVM_XEN_VCPU_SET_ATTR/KVM_XEN_VCPU_GET_ATTR ioctls.